Flask Documentation

Release 0.12
Jan 25, 2017


I User’s Guide 1
1 Foreword 3

2 Foreword for Experienced Programmers 5

3 Installation 7

4 Quickstart 11

5 Tutorial 29

6 Templates 45

7 Testing Flask Applications 49

8 Application Errors 57

9 Debugging Application Errors 63

10 Configuration Handling 65

11 Signals 73

12 Pluggable Views 77

13 The Application Context 83

14 The Request Context 87

15 Modular Applications with Blueprints 93

16 Flask Extensions 99

17 Command Line Interface 101

18 Development Server 107

19 Working with the Shell 109


20 Patterns for Flask 111

21 Deployment Options 169

22 Becoming Big 183

II API Reference 187
23 API 189

III Additional Notes 255
24 Design Decisions in Flask 257


26 Security Considerations 267

27 Unicode in Flask 269

28 Flask Extension Development 273

29 Pocoo Styleguide 281

30 Python 3 Support 285

31 Upgrading to Newer Releases 287

32 Flask Changelog 297

33 License 311


Part I

This part of the documentation, which is mostly prose, begins with some background
information about Flask, then focuses on step-by-step instructions for web develop-
ment with Flask.



such as what database to use. nor does it mean that Flask is lacking in functionality. with sensible defaults. and a few conventions when getting started. but it’s ready for production use on a variety of needs. are easy to change. templates and static files are stored in subdi- rectories within the application’s Python source tree. Flask won’t make many decisions for you. various open authentication technologies. CHAPTER ONE FOREWORD Read this before you get started with Flask. especially when getting started. By convention.1 What does “micro” mean? “Micro” does not mean that your whole web application has to fit into a single Python file (although it certainly can). form val- idation. Flask may be “micro”. Instead. 3 . By default. This hopefully answers some questions about the purpose and goals of the project. Everything else is up to you. Flask supports extensions to add such functionality to your application as if it was imple- mented in Flask itself. The “micro” in microframework means Flask aims to keep the core simple but exten- sible. such as what templating engine to use. upload handling. and more. form validation or any- thing else where different libraries already exist that can handle that.2 Configuration and Conventions Flask has many configuration values. you usually don’t have to. Numerous extensions provide database integration. with the names templates and static respectively. While this can be changed. Flask does not include a database abstraction layer. 1. so that Flask can be everything you need and nothing you don’t. Those decisions that it does make. and when you should or should not be using it. 1.

Should you need more cus- tomization. You can implement advanced patterns in SQLAlchemy or an- other database tool. the Python web interface. introduce non-relational data persistence as appropriate. Flask includes many hooks to customize its behavior. Flask will continue to provide a very simple glue layer to the best that Python has to offer.1. If you are curious about the Flask design principles. or the Foreword for Experienced Programmers. check out the Becoming Big chapter. you’ll find a variety of extensions available in the community to integrate your project for production. the Flask class is built for subclassing. head over to the section about Design Decisions in Flask. you are free to make the design decisions appropriate for your project. and take advantage of framework-agnostic tools built for WSGI. Continue to Installation. If you are interested in that.3 Growing with Flask Once you have Flask up and running. the Quickstart. The Flask core team reviews extensions and ensures approved extensions do not break with future releases. As your codebase grows. 4 .

2. they should not take a lot of code and yet they should not limit you. Flask uses thread-local objects internally so that you don’t have to pass ob- jects around from function to function within a request in order to stay threadsafe. Flask protects you against one of the most common security problems of modern web applications: cross-site scripting (XSS). The documentation will warn you about aspects of web development that require at- tention to security. For example.1 Thread-Locals in Flask One of the design decisions in Flask was that simple tasks should be simple. If you write a web application. And even if you are the only user that might leave data in your application. The Flask project is honest about thread-locals. Unfortunately. Unless you deliberately mark in- secure HTML as secure.until a clever attacker figures out a way to exploit our applications. The users are entrusting you with data. Because of that. 5 . but requires a valid request context for dependency in- jection or when attempting to reuse code which uses a value pegged to the request. does not hide them. And don’t think that your application is not important enough to attract an attacker. and we all sometimes underestimate the likelihood that a vulnerability will be exploited . CHAPTER TWO FOREWORD FOR EXPERIENCED PROGRAMMERS 2. and calls out in the code and documentation where they are used. Flask and the underlying Jinja2 template engine have you covered.2 Develop for the Web with Caution Always keep security in mind when building web applications. Some of these security concerns are far more complex than one might think. you are probably allowing users to register and leave their data on your server. there are many ways the security of a web application can be com- promised. This approach is convenient. Flask has a few design choices that some people might find surprising or unorthodox. But there are many more ways to cause security problems. you still want that data to be stored securely.

3 Python 3 Support in Flask Flask. its dependencies. 6 . chances are that automated bots are probing for ways to fill your database with spam. watching for exploits when building to your requirements. and the like. If you want to use Flask with Python 3 have a look at the Python 3 Support page. links to malicious software. and most Flask extensions all support Python 3. 2. Flask is no different from any other framework in that you the developer must build with caution.Depending on the kind of attack. Continue to Installation or the Quickstart.

or at least different versions of Python libraries. 3. the more likely it is that you will be working with different versions of Python itself. So what do you do if two or more of your projects have conflicting dependencies? Virtualenv to the rescue! Virtualenv enables multiple side-by-side installations of Python. one for each project. chances are you want to use it for other projects besides Flask-based web applications. too. Werkzeug is a toolkit for WSGI. You will need Python 2.6 or newer to get started.x installation. Maybe it’s even in your package manager. Jinja2 renders templates. What problem does virtualenv solve? If you like Python as much as I do. you’ll probably want to use it there. and if you have shell access to your production machines. try: 7 . But the more projects you have. but the most kick-ass method is virtualenv. For using Flask with Python 3 have a look at Python 3 Support. If you are on Mac OS X or Linux. Let’s face it: quite often libraries break backwards compatibility. so be sure to have an up-to-date Python 2. If you use Ubuntu. chances are that the following command will work for you: $ sudo pip install virtualenv It will probably install virtualenv on your system.1 virtualenv Virtualenv is probably what you want to use during development. It doesn’t actually install separate copies of Python. So how do you get all that on your computer quickly? There are many ways you could do that. so let’s have a look at that first. Let’s see how virtualenv works. like Werkzeug and Jinja2. CHAPTER THREE INSTALLATION Flask depends on some external libraries. the standard Python interface between web applications and a vari- ety of servers for both development and deployment. and it’s unlikely that any serious application will have zero dependencies. but it does provide a clever way to keep different project environments isolated.

Now..$ sudo apt-get install python-virtualenv If you are on Windows and don’t have the easy_install command. the prompt of your shell should be as familiar as before.2 System-Wide Installation This is possible as well. Now.. but without the sudo prefix. whenever you want to work on a project. venv/bin/activate If you are a Windows user. pip. the following command is for you: $ venv\Scripts\activate Either way.) 8 .. Just run pip with root privi- leges: $ sudo pip install Flask (On Windows systems. do the following: $ ..... let’s move on. use the following command: $ deactivate After doing this. Enter the following command to get Flask activated in your vir- tualenv: $ pip install Flask A few seconds later and you are good to go.. you only have to activate the corre- sponding environment. you must install it first. Once you have virtualenv installed. Once you have it installed. and leave out sudo. Check the pip and setuptools on Windows section for more information about how to do that. you should now be using your virtualenv (notice how the prompt of your shell has changed to show the active environment).. On OS X and Linux. run it in a command-prompt window with administrator priv- ileges. I usually create a project folder and a venv folder within: $ mkdir myproject $ cd myproject $ virtualenv venv New python executable in venv/bin/python Installing setuptools. 3.done. just fire up a shell and create your own environ- ment... though I do not recommend it. And if you want to go back to the real world. run the same commands as above.

pip.py develop . $ . because it doesn’t know where those executables are (give either a try!).. virtualenv is recommended. to install virtualenv. then Scripts. then find the win_add2path.3 Living on the Edge If you want to work with the latest version of Flask. 3. If you don’t currently have pip. once you pull up a command prompt you want to be able to type pip and python which will run those things.4 pip and setuptools on Windows Sometimes getting the standard “Python packaging tools” like pip. or you can tell it to operate on a git checkout. The crucial package you will need is pip .git/ $ cd flask $ virtualenv venv New python executable in venv/bin/python Installing setuptools. Finished processing dependencies for Flask This will pull in the dependencies and activate the git head as the current version inside the virtualenv. you can simply run: 9 . Get the git checkout in a new virtualenv and run in development mode: $ git clone http://github. venv/bin/activate $ python setup. Then all you have to do is run git pull origin to update to the latest version... you should be able to navigate to your Python install directory (e. Open a new Command Prompt and check that you can now just type python to bring up the interpreter. there are two ways: you can either let pip pull in the development version.py file and run that.git Initialized empty Git repository in ~/dev/flask/.py will install it for you.done.3.... then get-pip.py It should be double-clickable once you download it. get-pip..g C:Python27). If you already have pip. you can upgrade them by running: > pip install --upgrade pip setuptools Most often.. but nothing very hard. Either way... Fortunately there is a “bootstrap script” you can run to install. but this might not automatically happen on Windows.com/pallets/flask. To fix this.. then go to Tools... setuptools and virtualenv can be a little trickier.this will let you install anything else (like virtualenv). Finally.

> pip install virtualenv Then you can be off on your way following the installation instructions above. 10 .

The first argument is the name of the application’s module or package. and returns the message we want to display in the user’s browser. This is needed so that Flask knows where to look for templates. 4.py because this would conflict with Flask itself. If you are using a single module (as in this example). 3. 4. First we imported the Flask class. Next we create an instance of this class. and so on. CHAPTER FOUR QUICKSTART Eager to get started? This page gives a good introduction to Flask. 2. It assumes you already have Flask installed.1 A Minimal Application A minimal Flask application looks something like this: from flask import Flask app = Flask(__name__) @app. If you do not. static files. head over to the Installation section. 11 . We then use the route() decorator to tell Flask what URL should trigger our function.py or something similar. Just save it as hello. you should use __name__ because depending on if it’s started as ap- plication or imported as module the name will be different ('__main__' versus the actual import name). Make sure to not call your application flask. An instance of this class will be our WSGI application. For more information have a look at the Flask documentation.route('/') def hello_world(): return 'Hello. World!' So what did that code do? 1. The function is given a name which is also used to generate URLs for that partic- ular function.

This is the default because in debugging mode a user of the application can execute arbitrary Python code on your computer.2 What to do if the Server does not Start In case the python -m flask fails or flask does not exist. there are multiple reasons this might be the case.0.py $ python -m flask run * Running on http://127. Before you can do that you need to tell your terminal the application to work with by exporting the FLASK_APP environment variable: $ export FLASK_APP=hello. which is good enough for testing but prob- ably not what you want to use in production. Alternatively you can use python -m flask: $ export FLASK_APP=hello.py $ flask run * Running on http://127. If you have the debugger disabled or trust the users on your network.0.0.1:5000/ If you are on Windows you need to use set instead of export. Externally Visible Server If you run the server you will notice that the server is only accessible from your own computer. First of all you need to look at the error message.2. For deployment options see Deployment Options. 12 . In short.1:5000/ This launches a very simple builtin server.0. Now head over to http://127. you can make the server publicly available simply by adding --host=0.0.To run the application you can either use the flask command or python’s -m switch with Flask. 4. not from any other in the network.11 use to have different ways to start the application.0 to the command line: flask run --host=0. and neither did python -m flask.1 Old Version of Flask Versions of Flask older than 0.0 This tells your operating system to listen on all public IPs. In that case you have two options: either upgrade to newer Flask versions or have a look at the Development Server docs to see the alternative method for running a server.1:5000/.0. the flask command did not exist. and you should see your hello world greet- ing.0.

Attention Even though the interactive debugger does not work in forking environments (which makes it nearly impossible to use on production servers). but you would have to restart it manually after each change to your code. 4. In case that module is incorrectly named you will get an import error upon start (or if debug is enabled when you navigate to the application). To enable debug mode you can export the FLASK_DEBUG environment variable before running the server: $ export FLASK_DEBUG=1 $ flask run (On Windows you need to use set instead of export). it activates the debugger 2. Screenshot of the debugger in action: 13 .3 Debug Mode (Want to just log errors and stack traces? See Application Errors) The flask script is nice to start a local development server. That is not very nice and Flask can do better. There are more parameters that are explained in the Development Server docs.2 Invalid Import Name The FLASK_APP environment variable is the name of the module to import at flask run. If you enable debug support the server will reload itself on code changes. it still allows the execution of arbitrary code. and it will also provide you with a helpful debugger if things go wrong. This makes it a major security risk and therefore it must never be used on production machines.4. it enables the debug mode on the Flask application. The most common reason is a typo or because you did not actually create an app object.2. This does the following things: 1. it activates the automatic reloader 3. It will tell you what it tried to import and why it failed.

Have another debugger in mind? See Working with Debuggers. which is especially handy for applications that are used from mobile devices with slower network connections.route('/hello') def hello(): return 'Hello.route('/') def index(): return 'Index Page' @app. This helps people remember the URLs. If the user can directly go to the desired page with- out having to hit the index page it is more likely they will like the page and come back next time. the route() decorator is used to bind a function to a URL. Here are some basic examples: @app. As you have seen above. 4. World' 14 .4 Routing Modern web applications have beautiful URLs.

15 . The idea behind that module is to ensure beautiful and unique URLs based on precedents laid down by Apache and earlier HTTP servers. Take these two rules: @app. the canonical URL for the projects endpoint has a trailing slash.4. the id is an integer return 'Post %d' % post_id The following converters exist: string accepts any text without a slash (the default) int accepts integers float like int but for floating point values path like the default but also accepts slashes any matches one of the items provided uuid accepts UUID strings Unique URLs / Redirection Behavior Flask’s URL rules are based on Werkzeug’s routing module. 4. Such a part is then passed as a keyword argument to your function.But there is more to it! You can make certain parts of the URL dynamic and attach multiple rules to a function. In the first case. it is similar to a folder on a filesystem.route('/user/<username>') def show_user_profile(username): # show the user profile for that user return 'User %s' % username @app. they differ in their use of the trailing slash in the URL definition.1 Variable Rules To add variable parts to a URL you can mark these special sections as <variable_name>. Here are some nice examples: @app.route('/post/<int:post_id>') def show_post(post_id): # show the post with the given id.route('/about') def about(): return 'The about page' Though they look rather similar. Optionally a converter can be used by specifying a rule with <converter:variable_name>.route('/projects/') def projects(): return 'The project page' @app. Accessing it without a trailing slash will cause Flask to redirect to the canonical URL with the trailing slash. In that sense.

4. even though we are interacting with it through a Python shell. def profile(username): pass . Have a look at the explanation below... print url_for('login'.route('/user/<username>') . consistent with how Apache and other servers work.4. 16 .. It accepts the name of the function as first argument and a number of keyword arguments. Here are some examples: >>> from flask import Flask. url_for >>> app = Flask(__name__) >>> @app..route('/') . Unknown variable parts are appended to the URL as query parameters... print url_for('index') . More impor- tantly. >>> @app. Accessing the URL with a trailing slash will produce a 404 “Not Found” error.. / /login /login?next=/ /user/John%20Doe (This also uses the test_request_context() method. Also. Context Locals). This behavior allows relative URLs to continue working even if the trailing slash is omitted..test_request_context(): . each corresponding to the variable part of the URL rule. print url_for('login') . can Flask also generate them? Of course it can. explained below. next='/') ..route('/login') .. To build a URL to a specific function you can use the url_for() function.. without having to remember to change URLs all over the place.2 URL Building If it can match URLs.. print url_for('profile'.. It tells Flask to behave as though it is handling a request..In the second case. the URL is defined without a trailing slash.. >>> with app. the URLs will stay unique. username='John Doe') .. which helps search engines avoid indexing the same page twice.. however.. it allows you to change URLs in one go. Reversing is often more descriptive than hard-coding the URLs... rather like the pathname of a file on UNIX-like systems. Why would you want to build URLs using the URL reversing function url_for() in- stead of hard-coding them into your templates? There are three good reasons for this: 1.. def index(): pass .. >>> @app. def login(): pass .

You don’t have to deal with that. here is a quick introduction to HTTP methods and why they matter: The HTTP method (also often called “the verb”) tells the server what the client wants to do with the requested page. This is how HTML forms usually transmit data to the server. not the content of the page. but that can be changed by providing the methods argument to the route() decorator. a route only answers to GET requests. 2. but there are some good reasons to do it this way. but it is only interested in the headers. In Flask you don’t have to deal with that at all.url_for() will handle that properly for you. URL building will handle escaping of special characters and Unicode data trans- parently for you. 'POST']) def login(): if request. 4. HEAD The browser tells the server to get the information.4. PUT Similar to POST but the server might trigger the store procedure multiple times by overwriting the old values more than once. OPTIONS is implemented for you automatically as well. Likewise. the underlying Werkzeug library handles that for you. The following methods are very common: GET The browser tells the server to just get the information stored on that page and send it. 3.method == 'POST': do_the_login() else: show_the_login_form() If GET is present. as of Flask 0. It will also make sure that HEAD requests are handled as the HTTP RFC (the document describing the HTTP protocol) demands. so you can completely ignore that part of the HTTP specification. This is probably the most common method. You have no idea what an HTTP method is? Worry not. in /myapplication in- stead of / . Consider that the connection is lost during transmission: in this situation a system between the browser and the server might receive the request safely a second time without 17 . methods=['GET'. By default.6.route('/login'. Now you might be asking why this is useful. HEAD will be added automatically for you. so you don’t have to deal with them. If your application is placed outside the URL root . POST The browser tells the server that it wants to post some new information to that URL and that the server must ensure the data is stored and only stored once. An application is supposed to handle that as if a GET request was received but to not deliver the actual content.say.3 HTTP Methods HTTP (the protocol web applications are speaking) knows different methods for ac- cessing URLs. Here are some examples: from flask import request @app.

the only methods a form can submit to the server are GET and POST. That’s usually where the CSS and JavaScript files are coming from.route('/hello/') @app. breaking things. 4. use the special 'static' endpoint name: url_for('static'. To generate URLs for static files. But with JavaScript and future HTML standards you can use the other methods as well. but during development Flask can do that as well. With POST that would not be possible because it must only be triggered once. Because of that Flask configures the Jinja2 template engine for you automatically. Starting with Flask 0.css') The file has to be stored on the filesystem as static/style. Ideally your web server is configured to serve them for you. this folder is next to that module. Now the interesting part is that in HTML4 and XHTML1. many revision control systems use it. 4. Furthermore HTTP has become quite popular lately and browsers are no longer the only clients that are using HTTP. Here’s a simple example of how to render a template: from flask import render_template @app. For instance.html'.5 Static Files Dynamic web applications also need static files. this is implemented for you automatically. Just create a folder called static in your package or next to your module and it will be available at /static on the application. To render a template you can use the render_template() method.route('/hello/<name>') def hello(name=None): return render_template('hello. and actually pretty cumbersome be- cause you have to do the HTML escaping on your own to keep the application secure.6 Rendering Templates Generating HTML from within Python is not fun.css.6. filename='style. if it’s a package it’s actually inside your package: Case 1: a module: 18 . So if your application is a module. DELETE Remove the information at the given location. name=name) Flask will look for templates in the templates folder. OPTIONS Provides a quick way for a client to figure out which methods are supported by this URL. All you have to do is provide the name of the template and the variables you want to pass to the template engine as keyword arguments.

navigation and footer)./application. If you want to know how that works.striptags() u'Marked up \xbb HTML' Changed in version 0. so if name contains HTML it will be escaped automati- cally. Templates are especially useful if inheritance is used.html For templates you can use the full power of Jinja2 templates.!</strong>') >>> Markup.py /templates /hello.py /templates /hello. HTML'). 19 .blink&gt. World!</h1> {% endif %} Inside templates you also have access to the request. Here is an example template: <!doctype html> <title>Hello from Flask</title> {% if name %} <h1>Hello {{ name }}!</h1> {% else %} <h1>Hello.escape('<blink>hacker</blink>') Markup(u'&lt. The 1Unsure what that g object is? It’s something in which you can store information for your own needs. If you can trust a variable and you know that it will be safe HTML (for example because it came from a module that converts wiki markup to HTML) you can mark it as safe by using the Markup class or by using the |safe filter in the template./blink&gt.html Case 2: a package: /application /__init__. Basically tem- plate inheritance makes it possible to keep certain elements on each page (like header. head over to the Template Inheritance pattern documentation.blink&gt. Head over to the official Jinja2 Template Documentation for more information.hacker&lt. session and g1 objects as well as the get_flashed_messages() function. Head over to the Jinja 2 documentation for more examples.5: Autoescaping is no longer enabled for all templates. Automatic escaping is enabled. Here is a basic introduction to how the Markup class works: >>> from flask import Markup >>> Markup('<strong>Hello %s!</strong>') % '<blink>hacker</blink>' Markup(u'<strong>Hello &lt.hacker&lt. check the documentation of that object (g) and the Using SQLite 3 with Flask for more information./blink&gt.') >>> Markup('<em>Marked up</em> &raquo.

But that is actually quite easy to understand.1 Context Locals Insider Information If you want to understand how that works and how you can implement tests with context locals.xhtml.7 Accessing Request Data For web applications it’s crucial to react to the data a client sends to the server. such as basic assertions: assert request. These objects are actually proxies to objects that are local to a specific context. method='POST'): # now you can do something with the request until the # end of the with block. . The solution is creating a request object yourself and binding it to the context. Imagine the context being the handling thread. otherwise just skip it.htm. You will notice that code which depends on a request object will suddenly break because there is no request object.xml.method == 'POST' 20 .html. When Flask starts its inter- nal request handling it figures out that the current thread is the active context and binds the current application and the WSGI environments to that context (thread). It does that in an intelligent way so that one application can invoke another application without breaking.path == '/hello' assert request. If you have some experience with Python you might be wondering how that object can be global and how Flask manages to still be threadsafe. Here is an example: from flask import request with app. In combination with the with statement it will bind a test request so that you can interact with it. What a mouthful.7. . 4. the underlying object is capable of dealing with concurrency systems other than threads). In Flask this information is provided by the global request object. Templates loaded from a string will have autoescaping disabled.test_request_context('/hello'. read this section. but not of the usual kind. The easiest solution for unit testing is to use the test_request_context() context manager. The answer is context locals: 4. A request comes in and the web server decides to spawn a new thread (or something else. So what does this mean to you? Basically you can completely ignore that this is the case unless you are doing something like unit testing. Certain objects in Flask are global objects. .following extensions for templates trigger autoescaping: .

21 .get('key'.7. First of all you have to import it from the flask module: from flask import request The current request method is available by using the method attribute. methods=['POST'.form['password']): return log_the_user_in(request.method == 'POST' 4.request_context(environ): assert request.route('/login'. head over to the request documentation. To access form data (data transmitted in a POST or PUT request) you can use the form attribute.args. 'GET']) def login(): error = None if request. Here is a full example of the two attributes mentioned above: @app. error=error) What happens if the key does not exist in the form attribute? In that case a special KeyError is raised. So for many situations you don’t have to deal with that problem.html'. For a full list of methods and attributes of the request object.form['username']) else: error = 'Invalid username/password' # the code below is executed if the request method # was GET or the credentials were invalid return render_template('login.The other possibility is passing a whole WSGI environment to the request_context() method: from flask import request with app. a HTTP 400 Bad Request error page is shown instead. '') We recommend accessing URL parameters with get or by catching the KeyError be- cause users might change the URL and presenting them a 400 bad request page in that case is not user friendly.form['username']. You can catch it like a standard KeyError but if you don’t do that. Here is a broad overview of some of the most common operations. request.2 The Request Object The request object is documented in the API section and we will not cover it here in detail (see request).method == 'POST': if valid_login(request. To access parameters submitted in the URL (?key=value) you can use the args at- tribute: searchword = request.

txt') .save('/var/www/uploads/uploaded_file.7.4 Cookies To access cookies you can use the cookies attribute. 'POST']) def upload_file(): if request. methods=['GET'. It behaves just like a standard Python file object. Just make sure not to forget to set the enctype="multipart/form-data" attribute on your HTML form..3 File Uploads You can handle uploaded files with Flask easily. checkout the Uploading Files pattern.route('/upload'. methods=['GET'.4.route('/upload'. do not use the cookies directly but instead use the Sessions in Flask that add some security on top of cookies for you. 4. Uploaded files are stored in memory or at a temporary location on the filesystem.. To set cookies you can use the set_cookie method of response objects.method == 'POST': f = request. you can access the filename attribute. If you want to know how the file was named on the client before it was uploaded to your application.method == 'POST': f = request. You can access those files by looking at the files attribute on the request object. 'POST']) def upload_file(): if request... If you want to use the file- name of the client to store the file on the server. Here is a simple example showing how that works: from flask import request @app. However please keep in mind that this value can be forged so never ever trust that value. If you want to use sessions.files['the_file'] f. but it also has a save() method that allows you to store that file on the filesys- tem of the server. pass it through the secure_filename() function that Werkzeug provides for you: from flask import request from werkzeug. The cookies attribute of request objects is a dictionary with all the cookies the client transmits.utils import secure_filename @app. Reading cookies: 22 . Each uploaded file is stored in that dictionary. For some better examples.7. otherwise the browser will not transmit your files at all.files['the_file'] f.save('/var/www/uploads/' + secure_filename(f.filename)) .

route('/') def index(): resp = make_response(render_template(. This is possible by utilizing the Deferred Request Callbacks pattern.set_cookie('username'. If you explicitly want to do that you can use the make_response() function and then modify it. If you want to customize the error page. Since you normally just return strings from the view functions Flask will convert them into response objects for you.. 4.)) resp.get(key) instead of cookies[key] to not get a # KeyError if the cookie is missing. you can use the errorhandler() decorator: 23 . Storing cookies: from flask import make_response @app.route('/') def index(): return redirect(url_for('login')) @app.cookies.route('/') def index(): username = request. use the abort() function: from flask import abort.. By default a black and white error page is shown for each error code. 'the username') return resp Note that cookies are set on response objects.get('username') # use cookies. use the redirect() function. to abort a request early with an error code. redirect.from flask import request @app. url_for @app. For this also see About Responses.8 Redirects and Errors To redirect a user to another endpoint.route('/login') def login(): abort(401) this_is_never_executed() This is a rather pointless example because a user will be redirected from the index to a page they cannot access (401 means access denied) but it shows how that works. Sometimes you might want to set a cookie at a point where the response object does not exist yet.

If the return value is a string it’s converted into a response object with the string as response body. 2. If none of that works.errorhandler(404) def page_not_found(error): return render_template('page_not_found. headers) or (response. Such tuples have to be in the form (response. If a response object of the correct type is returned it’s directly returned from the view. a response object is created with that data and the default param- eters. headers) where at least one item has to be in the tuple. See Error handlers for more details.html').html'). The status value will override the status code and headers can be a list or dictionary of additional header values. 404 You just need to wrap the return expression with make_response() and get the re- sponse object to modify it. 404 Note the 404 after the render_template() call. 4. 3.9 About Responses The return value from a view function is automatically converted into a response ob- ject for you.html'). This tells Flask that the status code of that page should be 404 which means not found. By default 200 is assumed which translates to: all went well. If a tuple is returned the items in the tuple can provide extra information. a 200 OK status code and a text/html mimetype. If you want to get hold of the resulting response object inside the view you can use the make_response() function. Flask will assume the return value is a valid WSGI appli- cation and convert that into a response object. status.errorhandler(404) def not_found(error): return render_template('error.from flask import render_template @app. then return it: @app. 4. The logic that Flask applies to converting return values into response objects is as follows: 1. 404) 24 . Imagine you have a view like this: @app. If it’s a string.errorhandler(404) def not_found(error): resp = make_response(render_template('error.

Here is how sessions work: from flask import Flask.headers['X-Something'] = 'A value' return resp 4. This is implemented on top of cookies for you and signs the cookies cryptographically.secret_key = 'A0Zr98j/3yX R~XHH!jmN]LWX/.route('/') def index(): if 'username' in session: return 'Logged in as %s' % escape(session['username']) return 'You are not logged in' @app.route('/login'.?RT' The escape() mentioned here does escaping for you if you are not using the template engine (as in this example). methods=['GET'. session. request app = Flask(__name__) @app. unless they know the secret key used for signing. url_for. How to generate good secret keys 25 . 'POST']) def login(): if request.form['username'] return redirect(url_for('index')) return ''' <form method="post"> <p><input type=text name=username> <p><input type=submit value=Login> </form> ''' @app. In order to use sessions you have to set a secret key. redirect. resp.method == 'POST': session['username'] = request.pop('username'. escape.route('/logout') def logout(): # remove the username from the session if it's there session. What this means is that the user could look at the contents of your cookie but not modify it. keep this really secret: app. None) return redirect(url_for('index')) # set the secret key.10 Sessions In addition to the request object there is also a second object called session which allows you to store information specific to a user from one request to the next.

Check out the Mes- sage Flashing for a full example. but actually is not. there are several Flask extensions that support this. If you are finding some values do not per- sist across requests. but sometimes that won’t do and the code has to continue working. This is usually combined with a layout template to expose the message. 26 . For example you may have some client-side code that sends an HTTP request to the server but it’s obviously malformed.12 Logging New in version 0. To flash a message use the flash() method. This might be caused by a user tampering with the data. Your operating system has ways to generate pretty random stuff based on a cryptographic random generator which can be used to get such a key: >>> import os >>> os. If the user does not get enough feedback they will probably end up hating the application. cookies are indeed enabled. if you want to handle sessions on the server-side instead. and you are not getting a clear error message. or the client code failing. And a secret key should be as random as possible. A note on cookie-based sessions: Flask will take the values you put into the session object and serialize them into a cookie. check the size of the cookie in your page responses compared to the size supported by web browsers.urandom(24) '\xfd{H\xe5<\x95\xf9\xe3\x96. Besides the default client-side based sessions. Most of the time it’s okay to reply with 400 Bad Request in that situation. You may still want to log that something fishy happened. Sometimes you might be in a situation where you deal with data that should be correct. 4. The flashing system basically makes it possible to record a message at the end of a request and access it on the next (and only the next) request.3 a logger is preconfigured for you to use. This is where loggers come in handy. 4. Flask provides a really simple way to give feedback to a user with the flashing system. to get hold of the messages you can use get_flashed_messages() which is also available in the templates.The problem with random is that it’s hard to judge what is truly random. As of Flask 0.3.11 Message Flashing Good applications and user interfaces are all about feedback.5\xd1\x01O<!\xd5\xa2\xa0\x9fR"\xa1\xa8' Just take that thing and copy/paste it into your code and you're done.

4. 42) app.fixers import LighttpdCGIRootFix app.logger.wsgi_app) 4.wsgi_app = LighttpdCGIRootFix(app.14 Using Flask Extensions Extensions are packages that help you accomplish common tasks. For example. Read more on Application Errors. Flask- SQLAlchemy provides SQLAlchemy support that makes it simple and easy to use with Flask.15 Deploying to a Web Server Ready to deploy your new Flask app? Go to Deployment Options. have a look at Flask Extensions.debug('A value for debugging') app. For more on Flask extensions. 4.logger. you can do it like this: from werkzeug.13 Hooking in WSGI Middlewares If you want to add a WSGI middleware to your application you can wrap the internal WSGI application. 27 . so head over to the official logging documentation for more information.warning('A warning occurred (%d apples)'. For example if you want to use one of the middlewares from the Werkzeug package to work around bugs in lighttpd.Here are some example log calls: app.contrib.error('An error occurred') The attached logger is a standard logging Logger.logger.

28 .

When the user is logged in. It only supports one user that can create text-only entries and there are no feeds or comments. In this tutorial.1 Introducing Flaskr This tutorial will demonstrate a blogging application named Flaskr. CHAPTER FIVE TUTORIAL You want to develop an application with Python and Flask? Here you have the chance to learn by example. we will create a simple microblogging application. allowing you to target different relational databases at once and more. The index page shows all entries so far in reverse chronological order (newest on top) and the user can add new ones from there if logged in. it will do the following things: 1.0-ish name . 3. Here a screenshot of the final application: 29 . We will use Flask and SQLite as a database (which comes out of the box with Python) so there is nothing else you need. they can add new entries to the page consisting of a text-only title and some HTML for the text. Only one user is supported. Let the user sign in and out with credentials specified in the configuration. but feel free to choose your own less Web-2. however. 5. 2. You might also want to consider one of the popular NoSQL databases if your data is more suited for those. it makes a lot of sense to use SQLAlchemy. as it handles database connections in a more intelligent way. but it still features everything you need to get started. For larger applications.) Essentially. SQLite3 will be used directly for this application because it’s good enough for an ap- plication of this size. This HTML is not sanitized because we trust the user here. check out the example source. If you want the full source code in advance or for comparison.

you will need to create the folders needed for this application: /flaskr /flaskr /static /templates The application will be installed and run as Python package. You will see exactly how to run flaskr later on in this tutorial. You will see examples of this later on. This is the place where CSS and JavaScript files go. 5. For now go ahead and create the applications directory structure. 30 . the files inside of the static folder are available to users of the application via HTTP. As a quick side note. Inside the templates folder.Continue with Step 0: Creating The Folders.2 Step 0: Creating The Folders Before getting started. In the next few steps you will be creating the database schema as well as the main module. Flask will look for Jinja2 templates. This is the recommended way to install and run Flask applications.

it is possible to drop the configuration directly into the module.path. load that.join(app.py: app = Flask(__name__) # create the application instance :) app.py file.py # Load default config and override config from an environment variable app. session. redirect. 5.from_object(__name__) # load config from this file . and a text. flash The next couple lines will create the actual application instance and initialize it with the config from the same file in flaskr. you can create the application module.3 Step 1: Database Schema In this step.ini or . flaskr. Continue with Step 2: Application Setup Code. After that there will be a few lines of configuration code. flaskr. 31 . This file should be placed inside of the flaskr/flaskr folder. create table entries ( id integer primary key autoincrement. a cleaner solution is to create a separate . Here are the import statements (in flaskr. a title.4 Step 2: Application Setup Code Now that the schema is in place. However.py. 'text' text not null ). 'flaskr. request. All you need to do is put the following contents into a file named schema.config. Each row in this table has an id. title text not null. Only a single table is needed for this application and it will only support SQLite. \ render_template. abort.sql in the flaskr/flaskr folder: drop table if exists entries.db'). the other two are strings that must not be null. For small applications like flaskr. This schema consists of a single table called entries. and import the values from there.config. url_for. The first several lines of code in the application module are the needed import statements.root_path. g.For now you should continue with Step 1: Database Schema.py): # all the imports import os import sqlite3 from flask import Flask. you will create the database schema.update(dict( DATABASE=os. 5. The id is an automatically incrementing integer and a primary key.

only variable names that are uppercase are considered. The silent switch just tells Flask to not complain if no such environment key is set. Choose that key wisely and as hard to guess and complex as possible. you will add a method that allows for easy connections to the specified database.connect(app. You can create a simple database connection through SQLite and then tell it to use the sqlite3. In this example. it’s recommended to use Instance Folders instead.Row object to represent rows. For a real-world application. Flask will then initialize the variable from that module. you can use the from_object() method on the config object and provide it with an import name of a module. app. so it can be updated with new val- ues. files can then easily be found. This can be used to open a connection on request and also from the inter- active Python shell or a script. Usually. This allows the rows to be treated as if they were dictionaries instead of tuples. Database Path Operating systems know the concept of a current working directory for each process.config. USERNAME='admin'. Unfortunately. environment-specific configuration file. it is a good idea to load a separate.config['DATABASE']) 32 . SECRET_KEY='development key'. you cannot depend on this in web applications because you might have more than one application in the same process.from_envvar('FLASKR_SETTINGS'. This will come in handy later. Lastly. we place the database right next to it.path module. silent=True) The Config object works similarly to a dictionary. For this reason the app. Together with the os. def connect_db(): """Connects to the specific database.root_path attribute can be used to get the path to the applica- tion. The SECRET_KEY is needed to keep the client-side sessions secure.from_envvar('FLASKR_SETTINGS'.config.""" rv = sqlite3. PASSWORD='default' )) app. This enables robust configuration setups. silent=True) Simply define the environment variable FLASKR_SETTINGS that points to a config file to be loaded. Flask allows you to import multiple configurations and it will use the setting defined in the last import. from_envvar() can help achieve this. In addition to that. Note that in all cases.

py /static /templates flaskr.5 Step 3: Installing flaskr as a Package Flask is now shipped with built-in support for Click. Presently this involves creating two new files. install_requires=[ 'flask'. You also need to add an __init__. setup.in The content of the setup. Later in this tutorial you will see exactly how to extend the flask command line interface (CLI). A useful pattern to manage a Flask application is to install your app following the Python Packaging Guide.py file for flaskr is: from setuptools import setup setup( name='flaskr'. your code structure should be: /flaskr /flaskr __init__. Create the MANIFEST.in in the projects root directory.row_factory = sqlite3.py file to make the flaskr/flaskr directory a package. In this case.Row return rv In the next section you will see how to run the application. Continue with Step 3: Installing flaskr as a Package.in). ].in and add the following lines: 33 .py and MANIFEST. the static and templates directories need to be included. 5. as well as the schema. rv. Click provides Flask with en- hanced and extensible command line utilities. packages=['flaskr']. include_package_data=True.sql setup.py schema. After these changes. ) When using setuptools.py MANIFEST. it is also necessary to specify any special files that should be included in your package (in the MANIFEST.

py: from . you should be able to start up the application. it is recommended to install your Flask application within a virtualenv. Externally Visible Server Want your server to be publicly available? Check out the externally visible server section for more information. you should get the database working. flaskr/__init__. Continue with Step 4: Database Connections.flaskr. As usual. The flaskr app is now installed in your virtualenv (see output of pip freeze). With that out of the way. the Flask development server needs the location of the app instance. 34 . Do this with the following commands: export FLASK_APP=flaskr export FLASK_DEBUG=true flask run (In case you are on Windows you need to use set instead of export).flaskr import app This import statement brings the application instance into the top-level of the appli- cation package. flaskr/.graft flaskr/templates graft flaskr/static include flaskr/schema. The editable flag allows editing source code without having to reinstall the Flask app each time you make changes. At this point you should be able to install the application. add the following import statement into this file. Never leave debug mode activated in a production system. because it will allow users to execute code on the server! You will see a message telling you that server has started along with the address at which you can access it. but first. When it is time to run the application. This import statement simplifies the location process. Without it the export statement a few steps below would need to be export FLASK_APP=flaskr. That will be addressed a little later. The above installation command assumes that it is run within the projects root direc- tory.sql To simplify locating the application. go ahead and install the application with: pip install --editable . When you head over to the server in your browser. The FLASK_DEBUG flag enables or disables the interactive debugger. With that said. you will get a 404 error because we don’t have any views yet.

For in- stance. you will need to make sure that only one request at a time uses the connection. 35 . Flask provides us with the teardown_appcontext() decorator. it will create a database connection for the current context. An elegant way to do this is by utilizing the application context.sqlite_db = connect_db() return g. the request variable is the request object associated with the current request. the app context is created before the request comes in and is destroyed (torn down) whenever the request finishes. but by itself. and successive calls will return the already established connection: def get_db(): """Opens a new database connection if there is none yet for the current application context. all you have to know is that there are special variables that use these. whereas g is a general purpose variable associated with the current application con- text. so you will need to keep it around for longer. Continue to Step 5: Creating The Database. 'sqlite_db'): g.close() Functions marked with teardown_appcontext() are called every time the app context tears down. What does this mean? Essentially. but how can you properly disconnect? For that. So when do you put it on there? To do that you can make a helper function. Because database connections encapsulate a transaction. """ if not hasattr(g.sqlite_db.5. in which case the error is passed to the teardown function. Curious about what these contexts mean? Have a look at the The Application Context documentation to learn more. It’s executed every time the application context tears down: @app. The first time the function is called.6 Step 4: Database Connections You currently have a function for establishing a database connection with connect_db.teardown_appcontext def close_db(error): """Closes the database again at the end of the request. Creating and closing database connections all the time is very inefficient. all you have to know is that you can store information safely on the g object. 'sqlite_db'): g.""" if hasattr(g. For the time being. Flask provides two contexts: the application context and the request context.sqlite_db Now you know how to connect. For the time being. The tutorial will cover some more details of this later on. it is not particularly useful. A tear- down can happen because of two reasons: either everything went well (the error pa- rameter will be None) or an exception happened.

A logical place is to group these module-level functions together. and if your app grows larger. In Flask. is just below the connect_db function in flaskr. When the command executes. and put your new get_db and close_db functions below your ex- isting connect_db function (following the tutorial line-by-line). Such a schema can be created by piping the schema. you can create a function and hook it into a flask command that initializes the database.cli.""" init_db() print('Initialized the database. and more precisely.') The app. you can then access 36 .py: def init_db(): db = get_db() with app.cli. This also requires that you provide the path to the database.commit() @app.command('initdb') def initdb_command(): """Initializes the database. Flaskr is a database powered application. 5. You don’t have to. A good place to add this function.7 Step 5: Creating The Database As outlined earlier. It’s a good idea to add a function that initializes the database for you. you might be wondering where to put the code from this step and the next.db < schema.executescript(f.open_resource('schema. which can introduce errors. Before starting the server for the first time.cursor(). which is not necessarily the case on every system. If you need a moment to find your bearings. For now just take a look at the code segment below.sql The downside of this is that it requires the sqlite3 command to be installed.read()) db. to the application. Such systems need a schema that tells them how to store that information. you can put all of your application code into a single Python module. it’s a good idea not to. it’s important to create that schema. Within the function.Hint: Where do I put this code? If you’ve been following along in this tutorial. Flask will automatically create an application context which is bound to the right application.sql file into the sqlite3 command as follows: sqlite3 /tmp/flaskr. it is an application powered by a relational database system. take a look at how the example source is organized. To do this.command() decorator registers a new command with the flask script. and command.sql'. mode='r') as f: db.

for example). Finally.8. Continue with Step 6: The View Functions 5. you only have to commit the changes. it is possible to create a database with the flask script: flask initdb Initialized the database. Troubleshooting If you get an exception later on stating that a table cannot be found. there is a method to execute a complete script. so that we can easily create databases in unit tests later on.g and other things as you might expect. though. you can start writing the view func- tions.flask. (For more information see Testing Flask Applications. The one with the highest id (the newest entry) will be on top. plural. The view function will pass the entries to the show_entries.) The open_resource() method of the application object is a convenient helper function that will open a resource that the application provides. You will need four of them: 5. It listens on the root of the application and will select title and text from the database.html template and return the rendered one: @app. the application context tears down and the database connection is released. SQLite3 and other transactional databases will not commit unless you explicitly tell it to.8 Step 6: The View Functions Now that the database connections are working. It is used in this example to execute a script on the database connection. This function opens a file from the resource location (the flaskr/flaskr folder) and allows you to read from it. The connection object provided by SQLite can give you a cursor object.1 Show Entries This view shows all the entries stored in the database.route('/') def show_entries(): 37 . You will want to keep an actual function around that initializes the database. On that cur- sor.Row row factory. When the script ends. The rows returned from the cursor look a bit like dictionaries because we are using the sqlite3. Now. check that you did execute the initdb command and that your table names are correct (singular vs.

it will flash() an information message to the next request and redirect back to the show_entries page: @app. If everything worked out well. text) values (?. 5. methods=['GET'. the actual form is shown on the show_entries page. If an error occurred.get('logged_in'): abort(401) db = get_db() db. if the logged_in key is present in the session and True). db = get_db() cur = db. [request. See Using SQLite 3 with Flask for more.8. a message is flashed that informs the user that he or she was logged in successfully.route('/login'. methods=['POST']) def add_entry(): if not session.route('/add'.execute('insert into entries (title.2 Add New Entry This view lets the user add new entries if they are logged in. Login checks the username and password against the ones from the configuration and sets the logged_in key for the session.form['title']. text from entries order by id desc') entries = cur. that key is set to True.html'.config['USERNAME']: 38 .method == 'POST': if request.form['username'] != app. your app will be vulnerable to SQL injection when you use string formatting to build SQL statements.execute('select title. This only responds to POST requests.commit() flash('New entry was successfully posted') return redirect(url_for('show_entries')) Note that this view checks that the user is logged in (that is. Security Note Be sure to use question marks when building SQL statements.form['text']]) db. and the user is asked again: @app.fetchall() return render_template('show_entries. and the user is redi- rected back to the show_entries page. ?)'.3 Login and Logout These functions are used to sign the user in and out. the template is notified about that. as done in the example above.8. entries=entries) 5. In addition. Otherwise. If the user logged in successfully. request. 'POST']) def login(): error = None if request.

removes that key from the session again. This is helpful because now it is not necessary to check if the user was logged in.9 Step 7: The Templates Now it is time to start working on the templates. The templates are using Jinja2 syntax and have autoescaping enabled by default. error=error) The logout function. This means that unless you mark a value in the code with Markup or with the |safe filter in the template.pop('logged_in'. This tutorial uses plain text passwords for simplicity. you will get an exception that Flask cannot find the templates. If you plan to release a project based off this tutorial out into the world.form['password'] != app. error = 'Invalid username' elif request. None) flash('You were logged out') return redirect(url_for('show_entries')) Security Note Passwords should never be stored in plain text in a production system. 39 .html'. You can find a list of recommended Flask extensions here Continue with Step 7: The Templates. There are also many general python libraries that can be used for hashing. so adding this functionality is fairly straight forward. We are also using template inheritance which makes it possible to reuse the layout of the website in all pages. Fortunately.route('/logout') def logout(): session. @app. if you make requests with the app running. there are Flask extensions for the purpose of hashing passwords and veri- fying passwords against hashes. Jinja2 will ensure that special characters such as < or > are escaped with their XML equivalents. There is a neat trick here: if you use the pop() method of the dict and pass a second parameter to it (the default).config['PASSWORD']: error = 'Invalid password' else: session['logged_in'] = True flash('You were logged in') return redirect(url_for('show_entries')) return render_template('login. As you may have noticed. passwords should be both hashed and salted before being stored in a database or file. 5. on the other hand. the method will delete the key from the dictionary if present or do nothing when that key is not in there.

logged_in %} <a href="{{ url_for('login') }}">log in</a> {% else %} <a href="{{ url_for('logout') }}">log out</a> {% endif %} </div> {% for message in get_flashed_messages() %} <div class=flash>{{ message }}</div> {% endfor %} {% block body %}{% endblock %} </div> 5.Put the following templates into the templates folder: 5.logged_in %} <form action="{{ url_for('add_entry') }}" method=post class=add-entry> <dl> <dt>Title: <dd><input type=text size=30 name=title> <dt>Text: 40 .9. even if there is no 'logged_in' key in the session: <!doctype html> <title>Flaskr</title> <link rel=stylesheet type=text/css href="{{ url_for('static'.1 layout.html This template extends the layout. Note that in Jinja you can access missing attributes and items of objects / dicts which makes the following code work. filename='style.html template from above to display the mes- sages. Note that the for loop iterates over the messages we passed in with the render_template() function.html This template contains the HTML skeleton. the header and a link to log in (or log out if the user was already logged in). The {% block body %} block can be replaced by a block of the same name (body) in a child template.→') }}"> <div class=page> <h1>Flaskr</h1> <div class=metanav> {% if not session.9.html" %} {% block body %} {% if session. Notice that the form is configured to to submit to the add_entry view function and use POST as HTTP method: {% extends "layout. It also displays the flashed messages if there are any.2 show_entries. The session dict is available in the template as well and you can use that to check if the user is logged in or not.css .

10 Step 8: Adding Style Now that everything else works.9. 5. } h1 { border-bottom: 2px solid #eee. } a.title }}</h2>{{ entry. h1. background: #eee. which basically just displays a form to allow the user to login: {% extends "layout. h2 { font-family: 'Georgia'. } 41 .css in the static folder: body { font-family: sans-serif. it’s time to add some style to the application.html" %} {% block body %} <h2>Login</h2> {% if error %}<p class=error><strong>Error:</strong> {{ error }}{% endif %} <form action="{{ url_for('login') }}" method=post> <dl> <dt>Username: <dd><input type=text name=username> <dt>Password: <dd><input type=password name=password> <dd><input type=submit value=Login> </dl> </form> {% endblock %} Continue with Step 8: Adding Style. margin: 0.html This is the login template. } h1.2em. <dd><textarea name=text rows=5 cols=40></textarea> <dd><input type=submit value=Share> </dl> </form> {% endif %} <ul class=entries> {% for entry in entries %} <li><h2>{{ entry.text|safe }} {% else %} <li><em>Unbelievable. No entries here so far</em> {% endfor %} </ul> {% endblock %} 5.3 login. h2 { color: #377ba8. } h2 { font-size: 1. Just create a stylesheet called style. serif.

py static/ templates/ tests/ test_flaskr.in For now go ahead a create the tests/ directory as well as the test_flaskr. The application above is used as a basic example of how to perform unit testing in the Testing Flask Applications section of the documentation.11.8em 1.error { background: #f0d6d6.py MANIFEST. } . padding: 0.2 Running the tests At this point you can run the tests. border: 1px solid #aacbe2. } . 5. width: 35em. } .2em.py setup. border-bottom: 1px solid #ccc. margin-bottom: 1em.5em. 42 .. background: #fafafa. you might be wondering about ways to organize the project. border: 5px solid #ccc. padding: 0.11.1 Adding tests to flaskr Assuming you have seen the Testing Flask Applications section and have either written your own tests for flaskr or have followed along with the examples provided. background: white. margin: 0. padding: 0.flash { background: #cee5F5.11 Bonus: Testing the Application Now that you have finished the application and everything works as expected.add-entry { font-size: 0.entries li { margin: 0. } Continue with Bonus: Testing the Application. font-size: 0.entries { list-style: none.metanav { text-align: right.py file.3em.8em. } . Go there to see how easy it is to test Flask applications. One possible and recommended project structure is: flaskr/ flaskr/ __init__. padding: 0. } .8em. } .page { margin: 2em auto.5em.entries li h2 { margin-left: -1em. } . padding: 0. Here pytest will be used. 5. 5. it’s probably not a bad idea to add automated tests to simplify modifications in the future. } .add-entry dl { font-weight: bold.9em.

within the top-level flaskr/ directory as: py. install_requires=[ 'flask'. tests_require=[ 'pytest'.cfg in the project root (alongside setup.11. packages=['flaskr'].py test This calls on the alias created in setup. as the setup. include_package_data=True.py) Following the standard rules of test-discovery your tests will be found.cfg.py file to contain: from setuptools import setup setup( name='flaskr'. Otherwise pytest test will not be able to import the required components to test the application: pip install -e .py script has been called. ]. pip install pytest Run and watch the tests pass. setup_requires=[ 'pytest-runner'. Here that requires adding a couple of lines to the setup.cfg which in turn runs pytest via pytest-runner. Go ahead and update the setup. ) Now create setup. and hopefully pass. run. ].Note: Make sure that pytest is installed in the same virtualenv as flaskr.py file and creating a new file setup.py): [aliases] test=pytest Now you can run: python setup. (Recall the setup_requires ar- gument in setup.test 5. 43 . ].3 Testing + setuptools One way to handle testing is to integrate it with setuptools. One benefit of running the tests this way is that you do not have to install pytest.

Integrating testing with setuptools is convenient because it is not necessary to actually download pytest or any other testing framework one might use.This is one possible way to run and manage testing. 44 . but there are other options such as nose. Here pytest is used.

If you want information on the template engine’s syntax itself. • autoescaping is enabled for all strings when using render_template_string(). . but you still have to install Jinja2 to run Flask itself. 6.1 Jinja Setup Unless customized. Jinja2 is configured by Flask as follows: • autoescaping is enabled for all templates ending in .xhtml when using render_template().html.xml as well as .config) New in version 0. An extension can depend on Jinja2 being present. • Flask inserts a couple of global functions and helpers into the Jinja2 context. ad- ditionally to the values that are present by default.6. You are obviously free to use a different tem- plate engine. 45 .htm.2 Standard Context The following global variables are available within Jinja2 templates by default: config The current configuration object (flask. . head over to the official Jinja2 Template Documentation for more information.10: This is now always available. even in imported tem- plates. Changed in version 0. • a template has the ability to opt in/out autoescaping with the {% autoescape %} tag. This section only gives a very quick introduction into how Jinja2 is integrated into Flask. This requirement is necessary to enable rich extensions. CHAPTER SIX TEMPLATES Flask leverages Jinja2 as template engine. 6.

This is for example very helpful if you try to generate JavaScript on the fly. get_flashed_messages() The flask. 2. g The request-bound object for global variables (flask. Importing with context looks like this: {% from '_helpers. This is partially caused by performance considerations.html' import my_macro with context %} 6.url_for() function. they are not global variables.3 Standard Filters These filters are available in Jinja2 additionally to the filters provided by Jinja2 itself: tojson() This function converts the given object into JSON representation. partially to keep things explicit. url_for() The flask.10 if you intend to use it inside script tags: 46 .session). or the attribute of the request object you are interested in. This variable is unavailable if the template was rendered without an active request context.g). you import the macro “with context”. you explicitly pass the request to the macro as parameter. Note that inside script tags no escaping must take place.request). session The current session object (flask. The Jinja Context Behavior These variables are added to the context of variables. What does this mean for you? If you have a macro you want to import. so make sure to disable escaping with |safe before Flask 0.request The current request object (flask.get_flashed_messages() function. The difference is that by default these will not show up in the context of imported templates. that needs to access the request object you have two possibilities: 1. This variable is unavail- able if the template was rendered without an active request context. This variable is unavailable if the template was rendered without an active request context.

• Inside the template. Because these characters carry specific meanings in documents on their own you have to replace them by so called “entities” if you want to use them for text. This is in general the recommended way. please be very cautious about the variables you are using in this block. This can be the case if you want to explicitly inject HTML into pages. " as well as '. You can either put them by hand into the jinja_env of the application or use the template_filter() decorator. <script type=text/javascript> doSomethingWith({{ user. but can also lead to security problems. for example if they come from a system that generates secure HTML like a markdown to HTML converter.username|tojson|safe }}). Not doing so would not only cause user frustration by the inability to use these characters in text. (see Cross-Site Scripting (XSS)) Sometimes however you will need to disable autoescaping in templates. The two following examples work the same and both reverse an object: @app. and thus XHTML) are &. </script> 6. wrap the HTML string in a Markup object before passing it to the template. To disable the autoescape system in templates. use the |safe filter to explicitly mark a string as safe HTML ({{ myvariable|safe }}) • Temporarily disable the autoescape system altogether. Spe- cial characters in the sense of HTML (or XML. There are three ways to accomplish that: • In the Python code. you can use the {% autoescape %} block: {% autoescape false %} <p>autoescaping is disabled here <p>{{ will_not_be_escaped }} {% endautoescape %} Whenever you do this. 6. <.template_filter('reverse') def reverse_filter(s): 47 .4 Controlling Autoescaping Autoescaping is the concept of automatically escaping special characters for you.5 Registering Filters If you want to register your own filters in Jinja2 you have two ways to do that. >.

33) }} You could also build format_price as a template filter (see Registering Filters). context processors exist in Flask. 48 . currency) return dict(format_price=format_price) The context processor above makes the format_price function available to all templates: {{ format_price(0.2f}{1}'. Variables are not limited to values.6 Context Processors To inject new variables automatically into the context of a template. A context processor is a function that returns a dictionary.filters['reverse'] = reverse_filter In case of the decorator the argument is optional if you want to use the function name as name of the filter.context_processor def inject_user(): return dict(user=g. you can use the filter in your templates in the same way as Jinja2’s builtin filters.user) The context processor above makes a variable called user available in the template with the value of g.format(amount. Context processors run before the template is rendered and have the ability to inject new values into the template context.jinja_env. but this demonstrates how to pass functions in a context processor. The keys and values of this dictionary are then merged with the template context. for all templates in the app: @app. a context processor can also make functions avail- able to templates (since Python allows passing around functions): @app. This example is not very interesting because g is available in templates anyways. for example if you have a Python list in context called mylist: {% for x in mylist | reverse %} {% endfor %} 6.context_processor def utility_processor(): def format_price(amount. currency=u''): return u'{0:. Once registered.user. return s[::-1] def reverse_filter(s): return s[::-1] app. but it gives an idea how this works.

get the sources from the examples. it is also not far from the truth.py) and create a unittest skeleton there: import os import flaskr import unittest import tempfile class FlaskrTestCase(unittest.app = flaskr.mkstemp() flaskr. In this documentation we will use the unittest package that comes pre-installed with Python. CHAPTER SEVEN TESTING FLASK APPLICATIONS Something that is untested is broken.config['TESTING'] = True self. If you don’t have that application yet.test_client() 49 . If an application has automated tests.TestCase): def setUp(self): self. we add a second module (flaskr_tests. Untested applications make it hard to improve existing code and developers of untested applications tend to become pretty paranoid. 7. we need an application to test.1 The Application First.config['DATABASE'] = tempfile. 7. The origin of this quote is unknown and while it is not entirely correct.db_fd. You can then use that with your favourite testing solution. you can safely make changes and instantly know if anything breaks. we will use the application from the Tutorial.app. flaskr.app.2 The Testing Skeleton In order to test the application.app. Flask provides a way to test your application by exposing the Werkzeug test Client and handling the context locals for you.

close(self. We just have to keep the db_fd around so that we can use the os.db_fd. To delete the database after the test. 7. Let’s check that the application shows “No entries here so far” if we access the root of the application (/).config['DATABASE'] = tempfile. we add a new test method to our class. close() function to close the file. This test client will give us a simple interface to the application. we close the file and remove it from the filesystem in the tearDown() method. we already know that our flaskr applica- tion is syntactically valid. To do this. This function is called before each individual test function is run. otherwise the import would have died with an exception. What it does is disable the error catching during request handling so that you get bet- ter error reports when performing test requests against the application.app.000s OK Even though it did not run any actual tests. If we now run the test suite. flaskr. Additionally during setup the TESTING config flag is activated.app.config['DATABASE']) if __name__ == '__main__': unittest.app.mkstemp() self. We can trigger test requests to the application.main() The code in the setUp() method creates a new test client and initializes a new database.db_fd) os.unlink(flaskr.test_client() 50 . like this: class FlaskrTestCase(unittest.app_context(): flaskr.app. Because SQLite3 is filesystem-based we can easily use the tempfile module to create a temporary database and initialize it.py ---------------------------------------------------------------------- Ran 0 tests in 0. with flaskr. and the client will also keep track of cookies for us. the latter we use as database name. we should see the following output: $ python flaskr_tests. The mkstemp() function does two things for us: it returns a low-level file handle and a random file name.init_db() def tearDown(self): os.app = flaskr.TestCase): def setUp(self): self.3 The First Test Now it’s time to start testing the functionality of the application.


def tearDown(self):

def test_empty_db(self):
rv = self.app.get('/')
assert b'No entries here so far' in rv.data

Notice that our test functions begin with the word test; this allows unittest to auto-
matically identify the method as a test to run.
By using self.app.get we can send an HTTP GET request to the application with the
given path. The return value will be a response_class object. We can now use the
data attribute to inspect the return value (as string) from the application. In this case,
we ensure that 'No entries here so far' is part of the output.
Run it again and you should see one passing test:

$ python flaskr_tests.py
Ran 1 test in 0.034s


7.4 Logging In and Out

The majority of the functionality of our application is only available for the adminis-
trative user, so we need a way to log our test client in and out of the application. To do
this, we fire some requests to the login and logout pages with the required form data
(username and password). And because the login and logout pages redirect, we tell
the client to follow_redirects.
Add the following two methods to your FlaskrTestCase class:

def login(self, username, password):
return self.app.post('/login', data=dict(
), follow_redirects=True)

def logout(self):
return self.app.get('/logout', follow_redirects=True)

Now we can easily test that logging in and out works and that it fails with invalid
credentials. Add this new test to the class:


def test_login_logout(self):
rv = self.login('admin', 'default')
assert b'You were logged in' in rv.data
rv = self.logout()
assert b'You were logged out' in rv.data
rv = self.login('adminx', 'default')
assert b'Invalid username' in rv.data
rv = self.login('admin', 'defaultx')
assert b'Invalid password' in rv.data

7.5 Test Adding Messages

We should also test that adding messages works. Add a new test method like this:

def test_messages(self):
self.login('admin', 'default')
rv = self.app.post('/add', data=dict(
text='<strong>HTML</strong> allowed here'
), follow_redirects=True)
assert b'No entries here so far' not in rv.data
assert b'&lt;Hello&gt;' in rv.data
assert b'<strong>HTML</strong> allowed here' in rv.data

Here we check that HTML is allowed in the text but not in the title, which is the in-
tended behavior.
Running that should now give us three passing tests:

$ python flaskr_tests.py
Ran 3 tests in 0.332s


For more complex tests with headers and status codes, check out the MiniTwit Exam-
ple from the sources which contains a larger test suite.

7.6 Other Testing Tricks

Besides using the test client as shown above, there is also the test_request_context()
method that can be used in combination with the with statement to activate a request
context temporarily. With this you can access the request, g and session objects like
in view functions. Here is a full example that demonstrates this approach:


import flask

app = flask.Flask(__name__)

with app.test_request_context('/?name=Peter'):
assert flask.request.path == '/'
assert flask.request.args['name'] == 'Peter'

All the other objects that are context bound can be used in the same way.
If you want to test your application with different configurations and there does not
seem to be a good way to do that, consider switching to application factories (see
Application Factories).
Note however that if you are using a test request context, the before_request()
and after_request() functions are not called automatically. However
teardown_request() functions are indeed executed when the test request con-
text leaves the with block. If you do want the before_request() functions to be called
as well, you need to call preprocess_request() yourself:
app = flask.Flask(__name__)

with app.test_request_context('/?name=Peter'):

This can be necessary to open database connections or something similar depending
on how your application was designed.
If you want to call the after_request() functions you need to call into
process_response() which however requires that you pass it a response object:
app = flask.Flask(__name__)

with app.test_request_context('/?name=Peter'):
resp = Response('...')
resp = app.process_response(resp)

This in general is less useful because at that point you can directly start using the test

7.7 Faking Resources and Context

New in version 0.10.
A very common pattern is to store user authorization information and database con-
nections on the application context or the flask.g object. The general pattern for this is
to put the object on there on first usage and then to remove it on a teardown. Imagine
for instance this code to get the current user:


def get_user():
user = getattr(g, 'user', None)
if user is None:
user = fetch_current_user_from_database()
g.user = user
return user

For a test it would be nice to override this user from the outside without hav-
ing to change some code. This can be accomplished with hooking the flask.
appcontext_pushed signal:

from contextlib import contextmanager
from flask import appcontext_pushed, g

def user_set(app, user):
def handler(sender, **kwargs):
g.user = user
with appcontext_pushed.connected_to(handler, app):

And then to use it:

from flask import json, jsonify

def users_me():
return jsonify(username=g.user.username)

with user_set(app, my_user):
with app.test_client() as c:
resp = c.get('/users/me')
data = json.loads(resp.data)
self.assert_equal(data['username'], my_user.username)

7.8 Keeping the Context Around

New in version 0.4.
Sometimes it is helpful to trigger a regular request but still keep the context around
for a little longer so that additional introspection can happen. With Flask 0.4 this is
possible by using the test_client() with a with block:

app = flask.Flask(__name__)

with app.test_client() as c:
rv = c.get('/?tequila=42')
assert request.args['tequila'] == '42'


If you were to use just the test_client() without the with block, the assert would
fail with an error because request is no longer available (because you are trying to use
it outside of the actual request).

7.9 Accessing and Modifying Sessions

New in version 0.8.
Sometimes it can be very helpful to access or modify the sessions from the test client.
Generally there are two ways for this. If you just want to ensure that a session has
certain keys set to certain values you can just keep the context around and access

with app.test_client() as c:
rv = c.get('/')
assert flask.session['foo'] == 42

This however does not make it possible to also modify the session or to access the ses-
sion before a request was fired. Starting with Flask 0.8 we provide a so called “session
transaction” which simulates the appropriate calls to open a session in the context of
the test client and to modify it. At the end of the transaction the session is stored. This
works independently of the session backend used:

with app.test_client() as c:
with c.session_transaction() as sess:
sess['a_key'] = 'a value'

# once this is reached the session was stored

Note that in this case you have to use the sess object instead of the flask.session
proxy. The object however itself will provide the same interface.



Applications fail. Here are some situations where perfectly fine code can lead to server errors: • the client terminated the request early and the application was still reading from the incoming data • the database server was overloaded and could not handle the query • a filesystem is full • a harddrive crashed • a backend server overloaded • a programming error in a library you are using • network connection of the server to another system failed And that’s just a small sample of issues you could be facing. captures the full stack trace and local variables for debugging. even if just for critical ones. Flask will display a very simple page for you and log the exception to the logger. Sentry aggregates duplicate errors. can become overwhelming if enough users are hitting the error and log files are typically never looked at. 8. This is why we recommend using Sentry for dealing with application errors. and we will cover some better setups to deal with errors. But there is more you can do. you will still see exceptions from time to time. servers fail. Even if your code is 100% correct.1 Error Logging Tools Sending error mails. So how do we deal with that sort of problem? By default if your application runs in production mode.3. It’s available as an Open Source project on GitHub and is also available as a hosted version which you can try for free. Why? Because everything else involved will fail. and sends you mails based on new errors or frequency thresholds. CHAPTER EIGHT APPLICATION ERRORS New in version 0. To use Sentry you need to install the raven client: 57 . Sooner or later you will see an exception in production.

exceptions. Afterwards failures are automatically reported to Sentry and from there you can re- ceive error notifications. dsn='YOUR_DSN_HERE') Or if you are using factories you can also init it later: from raven. This can be done by registering error handlers. code == 400). 8.2 Error handlers You might want to show custom error pages to the user when an error occurs.2. HTTPException subclasses like BadRequest from the example and their HTTP codes are interchangeable when handed to the registration methods or decorator (BadRequest. but the first one is more clear and leaves you with a function to call on your whim (and in tests).flask import Sentry sentry = Sentry(dsn='YOUR_DSN_HERE') def create_app(): app = Flask(__name__) sentry.flask import Sentry sentry = Sentry(app. return app The YOUR_DSN_HERE value needs to be replaced with the DSN value you get from your Sentry installation..BadRequest) def handle_bad_request(e): return 'bad request!' app. they are registered for exceptions that are raised while trying to do something else..contrib.1 Registering Register error handlers using errorhandler() or register_error_handler(): @app. 58 .$ pip install raven And then add this to your Flask app: from raven. Error handlers are normal Pluggable Views but instead of being registered for routes.errorhandler(werkzeug. Note that werkzeug. lambda e: 'bad request!') Those two ways are equivalent.exceptions.register_error_handler(400.init_app(app) . 8.contrib.

its class hierarchy is traversed.2 Handling Once an exception instance is raised.ERROR) app. Then you get an alert and you can do something about it.com'. and searched for in the exception classes for which handlers are registered. What you probably want instead is a mail the second the exception happened. If your mail server re- quires credentials.setLevel(logging. The most specific handler is selected.0. Here is how you can configure the Flask logger to send you mails for exceptions: ADMINS = ['yourname@example. 59 .0. if an instance of ConnectionRefusedError is raised. ADMINS. and it can actually send you mails for errors which is probably what you want.0.handlers import SMTPHandler mail_handler = SMTPHandler('127. these can also be provided. and its response is shown to the user. Often it’s in your webserver’s log files.You are however not limited to HTTPException or HTTP status codes but can register a handler for every exception class you like. The reason for that is that Flask by default will just report to the WSGI error stream or stderr (depending on what’s available). For that check out the documentation for the SMTPHandler.3 Error Mails If the application runs in production mode (which it will do on your server) you might not see any log messages. and a handler is registered for ConnectionError and ConnectionRefusedError. E.g. 'server-error@example. Flask uses the Python builtin logging system.com'] if not app.1 to all the ADMINS from the address server- error@example. 'YourApplication Failed') mail_handler. Where this ends up is sometimes hard to find. 8. 8. the more specific ConnectionRefusedError handler is called on the exception instance. Changed in version 0.logger.com with the subject “YourApplication Failed”.1'.addHandler(mail_handler) So what just happened? We created a new SMTPHandler that will send mails with the mail server listening on 127.debug: import logging from logging. I can pretty much promise you however that if you only use a logfile for the application errors you will never look at it except for debugging an issue when a user reported it for you.11: Errorhandlers are now prioritized by specificity of the excep- tion classes they are registered for instead of the order they are registered in.2.0.

If you are deploying on a Windows box. Once you picked your log handler. and more importantly. Before you run that in production.5 Controlling the Log Format By default a handler will only write the message string into a file or send you that message as mail. You don’t have to do that in the log formatter format string.logs messages to a file on the filesystem and will rotate after a certain number of messages. Warnings however are not. Because we certainly don’t want to get a mail for warnings or other useless logs that might happen during request handling. It’s a good idea to keep as much information around that might be required to debug a problem.4 Logging to a File Even if you get mails.will log to the system event log of a Windows system. where it did. Note that tracebacks are ap- pended to the log entry automatically. just make sure to use a lower setting (I would recommend WARNING): if not app.. A formatter can be instantiated with a format string.WARNING) app.11. errors are logged to your webserver’s log automatically. this is what you want to use. • RotatingFileHandler . so it’s your responsibility to warn in the code if something seems odd.setLevel(logging.addHandler(file_handler) 8. do like you did with the SMTP handler above.We also tell the handler to only send errors and more critical messages.. That will save you from a lot of frustration. There are a couple of handlers provided by the logging system out of the box but not all of them are useful for basic error logging.sends logs to a UNIX syslog. please also look at Controlling the Log Format to put more information into that error mail. and it makes a lot of sense to configure your logger to also contain that information so that you have a better idea of why that error happened. Please note that Flask itself will not issue any warnings in the core system. • SysLogHandler . The most interesting are probably the following: • FileHandler .debug: import logging from themodule import TheHandlerYouWant file_handler = TheHandlerYouWant(. By default as of Flask 0.logs messages to a file on the filesystem. you probably also want to log warnings. 60 . 8.logger. A log record stores more information. • NTEventLogHandler .) file_handler.

5.2 File logging from logging import Formatter file_handler. 61 . Note that this list is not complete.5.setFormatter(Formatter( '%(asctime)s %(levelname)s: %(message)s ' '[in %(pathname)s:%(lineno)d]' )) 8.1 Email from logging import Formatter mail_handler.5.setFormatter(Formatter(''' Message type: %(levelname)s Location: %(pathname)s:%(lineno)d Module: %(module)s Function: %(funcName)s Time: %(asctime)s Message: %(message)s ''')) 8.Here are some example setups: 8.3 Complex Log Formatting Here is a list of useful formatting variables for the format string. consult the official documentation of the logging package for a full list.

'CRITICAL'). getting the loggers with the getLogger() function and iterating over them to attach handlers: from logging import getLogger loggers = [app. This can be changed by subclassing the formatter and overriding the formatTime() method. %(message)s The logged message. you don’t have to override it.logger. By de- fault this is of the form "2003-07-08 16:49:45. formatTime(): called for asctime formatting. If you want a different time format you can override this method. getLogger('sqlalchemy'). I would recommend figuring out which loggers you are interested in. formatException() called for exception formatting. For more information. 'ERROR'. It is passed a LogRecord object and has to return the formatted string. There might be a situation in which you want to have multiple separate applications running side by side in the same Python interpreter and then it becomes impossible to have different logging setups for those. While there is a method to configure all loggers at once in the logging package. 'WARNING'. For example. getLogger('otherlibrary')] for logger in loggers: logger. SQLAlchemy uses logging heavily in its core. The formatter has three interesting methods: format(): handles the actual formatting.896" (the num- bers after the comma are millisecond portion of the time). %(lineno)d Source line number where the logging call was issued (if avail- able). Other libraries might log themselves as well. I would not recommend using it. %(funcName)s Name of function containing the logging call.6 Other Libraries So far we only configured the logger your application created itself. It is passed an exc_info tuple and has to return a string. head over to the official documentation. Format Description %(levelname)s Text logging level for the message ('DEBUG'. %(pathname)s Full pathname of the source file where the logging call was issued (if available). %(module)s Module (name portion of filename). %(filename)s Filename portion of pathname. Instead. computed as msg % args If you want to further customize the formatting. 'INFO'. The default is usually fine.addHandler(mail_handler) logger.addHandler(file_handler) 62 . 8. %(asctime)s Human-readable time when the LogRecord was created. you can subclass the formatter.

9. of course): 63 . note that debuggers interfere with each other. but be sure to do this temporarily in a controlled environment. This section provides pointers when debugging deployment configuration and digging deeper with a full-featured Python debugger.e. 9. Run Manually Having problems getting your application configured for production? If you have shell access to your host. which is helpful in catching configuration issues.1 When in Doubt. CHAPTER NINE DEBUGGING APPLICATION ERRORS For production applications. If you’re using Aptana/Eclipse for debugging you’ll need to set both use_debugger and use_reloader to False.yaml (change the block as appropriate for your application. You have to set some options in order to use your favorite debugger: • debug . Flask provides a debugger out of the box (see Debug Mode). Do not run in production with debug=True. configure your application with logging and notifications as described in Application Errors. verify that you can run your application manually from the shell in the deployment environment.2 Working with Debuggers To dig deeper. Be sure to run under the same user account as the configured deployment to troubleshoot permission issues.whether to reload and fork the process on exception debug must be True (i.whether to enable debug mode and catch exceptions • use_debugger . If you would like to use another Python debugger.whether to use the internal Flask debugger • use_reloader . A possible useful pattern for configuration is to set the following in your config. possibly to trace code execution.. You can use Flask’s builtin development server with debug=True on your production host. exceptions must be caught) in order for the other two options to have any value.

FLASK: DEBUG: True DEBUG_WITH_APTANA: True Then in your application’s entry-point (main.0') 64 .config.0.yaml") if app. host='0.run(use_debugger=use_debugger. debug=app. you could have something like: if __name__ == "__main__": # To allow aptana to receive errors.0.debug.get('DEBUG_WITH_APTANA')) except: pass app.py). set use_debugger=False app = create_app(config="config. use_reloader=use_debugger.debug: use_debugger = True try: # Disable Flask's debugger if external debugger is requested use_debugger = not(app.

config['DEBUG'] = True Certain configuration values are also forwarded to the Flask object so you can read and write them from there: app. There are different settings you might want to change depending on the application environment like toggling the debug mode.update( DEBUG=True. You can hardcode the configuration in the code.config. which for many small applications is not actually that bad. setting the secret key. but there are better ways. Independent of how you load your config.3.. CHAPTER TEN CONFIGURATION HANDLING New in version 0. Applications need some kind of configuration. there is a config object available which holds the loaded configuration values: The config attribute of the Flask object.. The way Flask is designed usually requires the configuration to be available when the application starts up.debug = True To update multiple keys at once you can use the dict. 10. SECRET_KEY='. and other such environment-specific things.' ) 65 .1 Configuration Basics The config is actually a subclass of a dictionary and can be modified just like any dictionary: app = Flask(__name__) app.update() method: app. But this is also where you can have your own configuration. This is the place where Flask itself puts certain configuration values and also where exten- sions can put their configuration values.

10.2 Builtin Configuration Values The following configuration values are used internally by Flask: 66 .

This can be disabled by this key. If this is not set the cookie will be valid for all of APPLICATION_ROOT or if that is not set for '/'. PERMANENT_SESSION_LIFETIME the lifetime of a permanent session as datetime. If this is not set. dev:5000') Note that localhost does not sup- port subdomains so setting this to “localhost” does not help. SERVER_NAME the name and port number of the server. 'debug' will only activate logging in debug mode.8 this can also be an integer represent- ing seconds. SESSION_COOKIE_PATH the path for the session cookie. 'production' will only log in produc- tion and 'never' disables it entirely. If set to True (which is the de- fault) then the cookie is refreshed each request which automatically bumps the lifetime. APPLICATION_ROOT If the application does not occupy a whole do- . Starting with Flask 0.g. SECRET_KEY the secret key SESSION_COOKIE_NAME the name of the session cookie SESSION_COOKIE_DOMAIN the domain for the session cookie. You can also use this setting to force-enable it for non debug execution which might be useful to debug production applications (but also very risky). Re- quired for subdomain support (e. SESSION_REFRESH_EACH_REQUEST this flag controls how permanent sessions are refreshed. the cookie will be valid for all subdomains of SERVER_NAME. SESSION_COOKIE_SECURE controls if the cookie should be set with the secure flag.: 'myapp. Non permanent sessions are not affected by this. USE_X_SENDFILE enable/disable x-sendfile LOGGER_NAME the name of the logger LOGGER_HANDLER_POLICY the policy of the default logging handler. If not set or explicitly set to None this is implicitly true if either TESTING or DEBUG is true.DEBUG enable/disable debug mode TESTING enable/disable testing mode PROPAGATE_EXCEPTIONS explicitly enable or disable the propagation of exceptions. Setting a SERVER_NAME also by67 default enables URL generation without a re- quest context but with an application context. If set to False a set-cookie header is only sent if the session is modified. Defaults to True. SESSION_COOKIE_HTTPONLY controls if the cookie should be set with the httponly flag.timedelta object. PRESERVE_CONTEXT_ON_EXCEPTION By default if the application is in debug mode the request context is not popped on excep- tions to enable debuggers to introspect the data. Defaults to False. The default is 'always' which means that the default logging handler is always active.

TRAP_HTTP_EXCEPTIONS. New in version 0. your web browser does as well.local' and add this name + the subdomains you want to use into your host config or setup a local bind. This environment variable can be set on Linux or OS X with the export command in the shell before starting the server: 68 .5: SERVER_NAME New in version 0. Please keep in mind that not only Flask has the problem of not knowing what sub- domains are. Most modern web browsers will not allow cross-subdomain cookies to be set on a server name without dots in it.8: TRAP_BAD_REQUEST_ERRORS. SESSION_COOKIE_DOMAIN. This makes packaging and distributing your application possible via various package handling tools (Deploying with Setuptools) and finally modifying the configuration file afterwards. SESSION_COOKIE_SECURE New in version 0. this is required if you want to work with subdomains. So a common pattern is this: app = Flask(__name__) app.7: PROPAGATE_EXCEPTIONS.from_envvar('YOURAPPLICATION_SETTINGS') This first loads the configuration from the yourapplication. JSONIFY_PRETTYPRINT_REGULAR New in version 0.config. ideally located outside the actual application package. LOGGER_HANDLER_POLICY. SESSION_COOKIE_HTTPONLY. Please choose a different server name in that case.default_settings module and then overrides the values with the contents of the file the YOURAPPLICATION_SETTINGS environment variable points to. JSON_SORT_KEYS. PRESERVE_CONTEXT_ON_EXCEPTION New in version 0.More on SERVER_NAME The SERVER_NAME key is used for the subdomain support.3 Configuring from Files Configuration becomes more useful if you can store it in a separate file. APPLICATION_ROOT. SESSION_COOKIE_PATH. This is also used for the session cookie.from_object('yourapplication. TEMPLATES_AUTO_RELOAD.4: LOGGER_NAME New in version 0.11: SESSION_REFRESH_EACH_REQUEST. So if your server name is 'localhost' you will not be able to set a cookie for 'localhost' and every subdomain of it. like 'myapplication. Because Flask cannot guess the subdomain part without the knowledge of the actual server name.default_settings') app. EXPLAIN_TEMPLATE_LOADING 10.config.10: JSON_AS_ASCII.6: MAX_CONTENT_LENGTH New in version 0.9: PREFERRED_URL_SCHEME New in version 0.

$ export YOURAPPLICATION_SETTINGS=/path/to/settings.. If you limit your- self to request-only accesses to the configuration you can reconfigure the object later on as needed.4 Configuration Best Practices The downside with the approach mentioned earlier is that it makes testing a little harder. 10. There are other methods on the config object as well to load from individual files. That way you can create multiple instances of your application with different configura- tions attached which makes unittesting a lot easier. Do not write code that needs the configuration at import time. There is no single 100% solution for this problem in general. On Windows systems use the set builtin instead: >set YOURAPPLICATION_SETTINGS=\path\to\settings.py * Running on http://127. For a complete reference. The easiest way to handle this is to use a default configuration that is always loaded and part of the version control.0. 2. So make sure to use uppercase letters for your config keys.1:5000/ * Restarting with reloader. Here is an example of a configuration file: # Example configuration DEBUG = False SECRET_KEY = '?\xbf. Create your application in a function and register blueprints on it. but there are a couple of things you can keep in mind to improve that experience: 1.. and a separate configuration that overrides the values as necessary as mentioned in the example above: 69 . Only values in uppercase are actually stored in the config object later on.w\xee\x8d$0\x13\x8b83' Make sure to load the configuration very early on.cfg $ python run-app.0. You can use this to pass in configuration as needed. read the Config object’s documentation. 10.\xb4\x8d\xa3"<\x9c\xb0@\x0f5\xab.cfg The configuration files themselves are actual Python files. There should be at least separate configurations for the production server and the one used during development.5 Development / Production Most applications need more than one configuration. so that extensions have the ability to access the configuration when starting up.

However there are alternative ways as well. • Use an environment variable to switch between the configurations. For example you could use imports or subclassing. development etc and import different hardcoded files based on that. An interesting pattern is also to use classes and inheritance for configuration: class Config(object): DEBUG = False TESTING = False DATABASE_URI = 'sqlite://:memory:' class ProductionConfig(Config): DATABASE_URI = 'mysql://user@localhost/foo' class DevelopmentConfig(Config): DEBUG = True class TestingConfig(Config): TESTING = True To enable such a config you just have to call into from_object(): app.py and you are done. head over to the Deploying with Fabric pattern.config. You could also inspect an environment vari- able like YOURAPPLICATION_MODE and set that to production. This can be done from outside the Python interpreter and makes development and deploy- ment much easier because you can quickly and easily switch between different configs without having to touch the code at all.config. For some details about how to do that.ProductionConfig') There are many different ways and it’s up to you how you want to manage your con- figuration files. 70 .config. Either populate the config with this default configuration or import it in your own configuration files before overriding values.default_settings') app.from_object('yourapplication.from_envvar('YOURAPPLICATION_SETTINGS') Then you just have to add a separate config. • Use a tool like fabric in production to push code and configurations separately to the production server(s). What is very popular in the Django world is to make the import explicit in the config file by adding from yourapplication.default_settings import * to the top of the file and then overriding the changes by hand.from_object('configmodule.app = Flask(__name__) app. If you are working often on dif- ferent projects you can even create your own script for sourcing that activates a virtualenv and exports the development configuration for you. However here a list of good recommendations: • Keep a default configuration in version control.py file and export YOURAPPLICATION_SETTINGS=/path/to/config.

If the instance_path parameter is not provided the following default locations are used: • Uninstalled module: /myapp.py /instance • Installed module or package: $PREFIX/lib/python2. For explicit con- figuration use the instance_path parameter: app = Flask(__name__. You can either explicitly provide the path of the instance folder when creating the Flask application or you can let Flask autodetect the instance folder. This was also how many developers loaded configurations stored next to the application.instance_path.X/site-packages/myapp $PREFIX/var/myapp-instance $PREFIX is the prefix of your Python installation. With Flask 0.8.6 Instance Folders New in version 0. This can be /usr or the path to your virtualenv. The instance folder is designed to not be under version control and be deployment specific. Flask for a long time made it possible to refer to paths relative to the application’s folder directly (via Flask.root_path). instance_path='/path/to/instance/folder') Please keep in mind that this path must be absolute when provided. Unfortu- nately however this only works well if applications are not packages in which case the root path refers to the contents of the package.8 introduces instance folders. It refers to a new concept called the “instance folder”.prefix to see what the prefix is set to. It’s the perfect place to drop things that either change at runtime or configuration files.py /instance • Uninstalled package: /myapp /__init__. You can print the value of sys.8 a new attribute was introduced: Flask. Flask 0.10. Since the config object provided loading of configuration files from relative filenames we made it possible to change the loading via filenames to be relative to the instance path if wanted. The behavior of relative paths in config files can be flipped between “relative to the application root” (the default) to “relative to instance folder” via the instance_relative_config switch to the application constructor: 71 .

read() # or via open_instance_resource: with app. instance_relative_config=True) Here is a full example of how to configure Flask to preload the config from a module and then override the config from a file in the config folder if it exists: app = Flask(__name__. open_instance_resource().instance_path.instance_path.join(app.path.default_settings') app. Flask also provides a shortcut to open a file from the instance folder with Flask. silent=True) The path to the instance folder can be found via the Flask.read() 72 . 'application.cfg'.open_instance_resource('application.config.cfg') with open(filename) as f: config = f.app = Flask(__name__.cfg') as f: config = f.config. Example usage for both: filename = os. instance_relative_config=True) app.from_pyfile('application.from_object('yourapplication.

the sender is the application that issued the signal. you can use the disconnect() method. However. This is especially true if you are developing an extension. you can use the connect() method of a signal. signals allow certain senders to notify subscribers that something happened. is executed in a specific order and is able to abort the request early by returning a response.1 Subscribing to Signals To subscribe to a signal. there is integrated support for signalling in Flask. The first ar- gument is the function that should be called when the signal is emitted. 73 . In short. there are differences in how they work. CHAPTER ELEVEN SIGNALS New in version 0. for example. be sure to also provide a sender unless you really want to listen for signals from all applications. When you subscribe to a signal. The big advantage of signals over handlers is that you can safely subscribe to them for just a split second. This support is provided by the excellent blinker library and will gracefully fall back if it is not available. the optional second argument specifies a sender. 11.6. To unsubscribe from a signal. In contrast all signal handlers are executed in undefined order and do not modify any data.6. Also keep in mind that signals are intended to notify subscribers and should not encourage subscribers to modify data. What are signals? Signals help you decouple applications by sending notifications when actions occur elsewhere in the core framework or another Flask extensions. You will notice that there are signals that appear to do the same thing like some of the builtin decorators do (eg: request_started is very similar to before_request()). Starting with Flask 0. Flask comes with a couple of signals and other extensions might provide more. For all core Flask signals. The core before_request() handler. Say you want to know what templates were rendered as part of a request: signals allow you to do exactly that. These temporary subscriptions are helpful for unittesting for example.

Additionally there is a convenient helper method (connected_to()) that allows you to temporarily subscribe a function to a signal with a context manager on its own. context)) return template_rendered.. app) try: yield recorded finally: template_rendered. the template object as well as context are appended to it. recorded. **extra): .disconnect(record. **extra): def record(sender.name == 'index.test_client(). you have to pass the list in as an argument: from flask import template_rendered def captured_templates(app. **extra): recorded. template. context = templates[0] assert template. context): recorded.append((template. here is a helper context manager that can be used in a unittest to deter- mine which templates were rendered and what variables were passed to the template: from flask import template_rendered from contextlib import contextmanager @contextmanager def captured_templates(app): recorded = [] def record(sender. app) This can now easily be paired with a test client: with captured_templates(app) as templates: rv = app.For example. template. template..status_code == 200 assert len(templates) == 1 template. app) The example above would then look like this: templates = [] with captured_templates(app. All the template rendering in the code issued by the application app in the body of the with block will now be recorded in the templates variable.get('/') assert rv.html' assert len(context['items']) == 10 Make sure to subscribe with an extra **extra argument so that your calls don’t fail if Flask introduces new arguments to the signals.connected_to(record. Because the return value of the context manager cannot be specified that way. Whenever a template is rendered.append((template. context.connect(record. context)) template_rendered. context = templates[0] 74 . templates.

Blinker API Changes The connected_to() method arrived in Blinker with version 1.. def save(self): model_saved.. 11.Namespace class.send(self) Try to always pick a good sender. you can use the blinker library directly. you can do so by calling the send() method. pass self as sender. If you are emitting a signal from a random function. This is what is recommended most of the time: from blinker import Namespace my_signals = Namespace() Now you can create new signals like this: model_saved = my_signals.3 Sending Signals If you want to emit a signal. For Extension Developers If you are writing a Flask extension and you want to gracefully degrade for missing blinker installations..2 Creating Signals If you want to use signals in your own application. Passing Proxies as Senders 75 .signals._get_current_object() as sender. 11. you can pass current_app. You can access the name of the signal with the name attribute.signal('model-saved') The name for the signal here makes it unique and also simplifies debugging. If you have a class that is emitting a signal. It accepts a sender as first argument and optionally some keyword arguments that are forwarded to the signal subscribers: class Model(object): . you can do so by using the flask.1. The most common use case are named signals in a custom Namespace.

context) 11. Note the limitations described in Send- ing Signals and the request_tearing_down signal. The reason for this is that current_app is a proxy and not the real application object. so you can rely on flask. _get_current_object() instead.1 you can also easily subscribe to signals by using the new connect_via() decorator: from flask import template_rendered @template_rendered.6 Core Signals Take a look at Signals for a list of all builtin signals. Context-local vari- ables are consistently available between request_started and request_finished.g and others as needed. 11.4 Signals and Flask’s Request Context Signals fully support The Request Context when receiving signals. 11.Never pass current_app as sender to a signal. 76 .5 Decorator Based Signal Subscriptions With Blinker 1. **extra): print 'Template %s is rendered with %s' % (template.connect_via(app) def when_template_rendered(sender.name. context. template. Use current_app.

views import View class ShowUsers(View): def dispatch_request(self): users = User. but if you want to provide this view in a generic fashion that can be adapted to other models and templates as well you might want more flex- ibility. CHAPTER TWELVE PLUGGABLE VIEWS New in version 0. Then we have to convert that class into an actual view function by using the as_view() class method. so let’s refactor the code a bit: 77 .query.html'. objects=users) app. But this by itself is not helpful.1 Basic Principle Consider you have a function that loads a list of objects from the database and renders into a template: @app.views.query.all() return render_template('users. As the first step to convert this into a class based view you would do this: from flask.add_url_rule('/users/'. Flask 0. 12. This is where pluggable class-based views come into place.all() return render_template('users. view_func=ShowUsers. The main intention is that you can replace parts of the implementations and this way have customizable pluggable views.as_view('show_users')) As you can see what you have to do is to create a subclass of flask. The string you pass to that function is the name of the endpoint that view will then have. users=users) This is simple and flexible.route('/users/') def show_users(page): users = User.7.7 introduces pluggable views inspired by the generic views from Django which are based on classes instead of functions.html'.View and im- plement dispatch_request().

add_url_rule('/about'. The class itself is instantiated with the parameters passed to the as_view() function.from flask. view_func=RenderTemplateView.template_name) And then you can register it like this: app. **context) def dispatch_request(self): context = {'objects': self. The way this works is that whenever the request is dispatched a new instance of the class is created and the dispatch_request() method is called with the parameters from the URL rule. In order to move that information to the class you can provide a methods attribute that has this information: 78 .query.as_view( 'about_page'.render_template(context) class UserView(ListView): def get_template_name(self): return 'users. context): return render_template(self. For instance you can write a class like this: class RenderTemplateView(View): def __init__(self.get_template_name().html')) 12.get_objects()} return self. but it’s good enough to explain the basic principle.template_name = template_name def dispatch_request(self): return render_template(self.2 Method Hints Pluggable views are attached to the application like a regular function by either using route() or better add_url_rule().views import View class ListView(View): def get_template_name(self): raise NotImplementedError() def render_template(self.html' def get_objects(self): return User. template_name='about. template_name): self. When you have a class-based view the question comes up what self points to.all() This of course is not that helpful for such a small example. That however also means that you would have to provide the names of the HTTP methods the view supports when you attach this.

form) .query.... 'POST'] def dispatch_request(self): if request.4 Decorating Views Since the view class itself is not the view function that is added to the routing system it does not make much sense to decorate the class itself. Each HTTP method maps to a function with the same name (just in lowercase): from flask. It’s automatically set based on the methods defined in the class. view_func=MyView. app..user: abort(401) return f(*args.views import MethodView class UserAPI(MethodView): def get(self): users = User.3 Method Based Dispatching For RESTful APIs it’s especially helpful to execute a different function for each HTTP method.""" def decorator(*args. Instead you either have to decorate the return value of as_view() by hand: def user_required(f): """Checks whether user is logged in or raises error 401... view_func=UserAPI.add_url_rule('/users/'.method == 'POST': . .from_form_data(request.all() .as_view('users')) That way you also don’t have to provide the methods attribute.as_view('myview')) 12. **kwargs): if not g. def post(self): user = User.class MyView(View): methods = ['GET'..add_url_rule('/myview'.MethodView you can easily do that.. **kwargs) return decorator 79 . With the flask. app.views. 12.

user_id): # delete a single user pass 80 . For instance consider that you are exposing a user object on the web: URL Method Description /users/ GET Gives a list of all users /users/ POST Creates a new user /users/<id> GET Shows a single user /users/<id> PUT Updates a single user /users/<id> DELETE Deletes a single user So how would you go about doing that with the MethodView? The trick is to take advantage of the fact that you can provide multiple rules to the same view.8 there is also an alternative way where you can specify a list of decorators to apply in the class declaration: class UserAPI(MethodView): decorators = [user_required] Due to the implicit self from the caller’s perspective you cannot use regular view dec- orators on the individual methods of the view however. That said. view_func=view) Starting with Flask 0.5 Method Views for APIs Web APIs are often working very closely with HTTP verbs so it makes a lot of sense to implement such an API based on the MethodView. you will notice that the API will require different URL rules that go to the same method view most of the time.view = user_required(UserAPI.as_view('users')) app. user_id): if user_id is None: # return a list of users pass else: # expose a single user pass def post(self): # create a new user pass def delete(self.add_url_rule('/users/'. Let’s assume for the moment the view would look like this: class UserAPI(MethodView): def get(self. 12. keep this in mind.

'PUT'. 'DELETE']) If you have a lot of APIs that look similar you can refactor that registration code: def register_api(view. view_func=user_view.add_url_rule(url.]) app.add_url_rule('/users/'. endpoint. view_func=user_view. 'DELETE']) register_api(UserAPI. view_func=view_func. 'user_api'.add_url_rule('/users/<int:user_id>'. methods=['GET'. view_func=view_func.add_url_rule('%s<%s:%s>' % (url.as_view(endpoint) app. user_id): # update a single user pass So how do we hook this up with the routing system? By adding two rules and explic- itly mentioning the methods for each: user_view = UserAPI. methods=['GET'. methods=['GET'.as_view('user_api') app. def put(self. 'PUT'.]) app.add_url_rule('/users/'. pk='id'.]) app. methods=['POST'.]) app. methods=['GET'. defaults={'user_id': None}. pk='user_id') 81 . pk_type. pk). defaults={pk: None}. view_func=user_view.add_url_rule(url. '/users/'. url. pk_type='int'): view_func = view. view_func=view_func. methods=['POST'.

82 .

CHAPTER THIRTEEN THE APPLICATION CONTEXT New in version 0.1 Purpose of the Application Context The main reason for the application’s context existence is that in the past a bunch of functionality was attached to the request context for lack of a better solution. The application setup state in which the application implicitly is on the module level. • no request handling happened so far • you have to have a reference to the application object in order to modify it. or a command line application. There is a third state which is sitting in between a little bit. for instance. One of the design ideas behind Flask is that there are two different “states” in which code is executed. during request handling. Sometimes you are dealing with an application in a way that is similar to how you interact with applications during request handling. and it implicitly ends when the first request comes in. the context local objects (flask. • any code can get hold of these objects at any time.request and others) point to the current request.9. Since one of the pillars of Flask’s design is that you can have more than one application in the same Python process. Consider. 13. a couple of other rules exist: • while a request is active. It starts when the Flask object is instantiated. The application context is what powers the current_app context local. While the application is in this state a few assumptions are true: • the programmer can modify the application object safely. 83 . In contrast. just that there is no request active. that you’re sitting in an interactive Python shell and interacting with the application. there is no magic proxy that can give you a reference to the application object you’re currently creating or modifying.

assuming they pick a sufficiently unique name and should put their information there. A common workaround for that problem was to use the current_app proxy later on. For more information about that. a RuntimeError will be raised. the application context was introduced. see Flask Extension Development.2 Creating an Application Context There are two ways to make an application context.app_context(): # within this block. Extensions are free to store additional information on the topmost level. 13. The first one is implicit: whenever a request context is pushed. It never moves between threads and it will not be shared between requests.name The application context is also used by the url_for() function in case a SERVER_NAME was configured. an application context will be created alongside if this is necessary. which was bound to the current request’s application reference. 13. print current_app. current_app points to app. The second way is the explicit way using the app_context() method: from flask import Flask.g object which is reserved for user code. RuntimeError: Working outside of application context._app_ctx_stack. you can ignore the existence of the application context unless you need it. This allows you to generate URLs even in the absence of a request. As such it is the perfect place to store database connection information and other things. instead of on the flask. Since creating such a request context is an unnecessarily expensive operation in case there is no request around. The internal stack object is called flask. 84 . current_app app = Flask(__name__) with app.3 Locality of the Context The application context is created and destroyed as necessary. As a result.So how does the code find the “right” application? In the past we recommended pass- ing applications around explicitly. If no request context has been pushed and an application context has not been explic- itly set. but that caused issues with libraries that were not designed with that in mind.

None) if db is not None: db.local import LocalProxy db = LocalProxy(get_db) That way a user can directly access db which internally calls get_db(). To make this im- plicit a LocalProxy can be used: from werkzeug.teardown_appcontext def teardown_db(exception): db = getattr(g.4 Context Usage The context is typically used to cache resources that need to be created on a per-request or usage case._database = connect_to_database() return db @app.close() The first time get_db() is called the connection will be established. 85 . The most common usage is to split resource management into two parts: 1. This is an example that connects to a database: import sqlite3 from flask import g def get_db(): db = getattr(g. an implicit resource caching on the context. None) if db is None: db = g. '_database'. a context teardown based resource deallocation. '_database'. Generally there would be a get_X() function that creates resource X if it does not ex- ist yet and otherwise returns the same resource. database connections are destined to go there. and a teardown_X() function that is registered as teardown handler. 2. For instance. When storing things on the application context unique names should be chosen as this is a place that is shared between Flask applications and extensions.13.

86 .

referrer or \ url_for('index') As you can see. Either with the with statement or by calling the push() and pop() methods: 87 . It is recommended that you read the The Application Context chapter first. this is the exception you will see: >>> redirect_url() Traceback (most recent call last): File "<stdin>".test_request_context('/?next=http://example. 14.args.get('next') or \ request. If you try to run this from a plain Python shell. So we have to make a request and bind it to the current context. url_for def redirect_url(): return request. in <module> AttributeError: 'NoneType' object has no attribute 'request' That makes a lot of sense because we currently do not have a request we could access.7 which is mostly in line with the old behavior but has some small.com/') This context can be used in two ways. it accesses the request object. CHAPTER FOURTEEN THE REQUEST CONTEXT This document describes the behavior in Flask 0. The test_request_context method can create us a RequestContext: >>> ctx = app. Imagine it would always redirect to the URL’s next parameter or the HTTP referrer or the index page: from flask import request. line 1.1 Diving into Context Locals Say you have a utility function that returns the URL the user should be redirected to. subtle differences.

start_response) The method request_context() returns a new RequestContext object and uses it in combination with the with statement to bind the context. On popping the application’s teardown_request() functions are also executed.request and others). head over to the Working with the Shell chapter.handle_exception(e)) return response(environ. 88 . you will find a piece of code that looks very much like this: def wsgi_app(self.>>> ctx.pop() Because the request context is internally maintained as a stack you can push and pop multiple times. The request context internally works like a stack: The topmost level on the stack is the current active request.com/' Until you call pop: >>> ctx. Another thing of note is that the request context will automatically also create an appli- cation context when it’s pushed and there is no application context for that application so far. push() adds the context to the stack on the very top. Everything that is called from the same thread from this point onwards until the end of the with statement will have access to the request globals (flask.push() From that point onwards you can work with the request object: >>> redirect_url() u'http://example. pop() removes it from the stack again.make_response(self. For more information of how to utilize the request context from the interactive Python shell.2 How the Context Works If you look into how the Flask WSGI application internally works.full_dispatch_request() except Exception as e: response = self. environ): with self.request_context(environ): try: response = self. 14. This is very handy to implement things like internal redirects.

In any case however the return value is treated as a replacement for the view’s return value. Now what happens on errors? In production mode if an exception is not caught. The new teardown functions are supposed to be used as a replacement for things that absolutely need to happen at the end of request. the other functions are no longer called. This is important to know if the life of the request context is prolonged by using the test client in a with statement or when using the request context from the command line: with app. When the request context is popped.test_client() as client: resp = client. Before each request. An important change in 0.get('/foo') # the teardown functions are still not called at that point 89 . This way the internal dispatching code looks cleaner and is easier to customize and understand. 14. Strictly speaking they are independent of the actual request handling as they are bound to the lifecycle of the RequestContext object. If one of these functions return a response.4 Teardown Callbacks The teardown callbacks are special callbacks in that they are executed at a different point. 2.7 because we wanted to make it easier to understand what is actually happening. At the end of the request the teardown_request() functions are executed. This always happens.3 Callbacks and Errors What happens if an error occurs in Flask during request processing? This particular behavior changed in 0. In development mode however the exception is not further processed and bubbles up to the WSGI server. the teardown_request() functions are called. If the before_request() functions did not return a response. 4. even in case of an unhandled exception down the road or if a before-request handler was not executed yet or at all (for example in test envi- ronments sometimes you might want to not execute before-request callbacks). the 500 internal server handler is called.14. before_request() functions are executed. the regular request handling kicks in and the view function that was matched has the chance to return a response. That way things like the interactive debugger can provide helpful debug information. The return value of the view is then converted into an actual response object and handed over to the after_request() functions which have the chance to replace it or modify it in place.7 is that the internal server error is now no longer post processed by the after request callbacks and after request callbacks are no longer guar- anteed to be executed. The new behavior is quite simple: 1. 3.

... # even though the response ended and you have the response # object in your hand # only when the code reaches this point the teardown functions # are called. you can use the _get_current_object() method: app = current_app. >>> ctx = app.pop() this runs after request >>> Keep in mind that teardown callbacks are always executed. Alternatively the same thing happens if another # request was triggered from the test client It’s easy to see the behavior from the command line: >>> app = Flask(__name__) >>> @app. so if you want to perform actual instance checks. 14. you have to do that on the instance that is being proxied (see _get_current_object below).teardown_request .._get_current_object() my_signal. but there are some exceptions where it is good to know that this object is an actual proxy: • The proxy objects do not fake their inherited types.. • if the object reference is important (so for example for sending Signals) If you need to get access to the underlying object that is proxied. Most of the time you don’t have to care about that. print 'this runs after request' . The reason behind this is that these proxies are shared between threads and they have to dispatch to the actual object bound to a thread behind the scenes as necessary. Certain parts of the test sys- tem might also temporarily create a request context without calling the before-request handlers.5 Notes On Proxies Some of the objects provided by Flask are proxies to other objects..push() >>> ctx. Make sure to write your teardown-request handlers in a way that they will never fail.test_request_context() >>> ctx.send(app) 90 . even if before-request call- backs were not executed yet but an exception happened. def teardown_request(exception=None): .

However it can be useful during development to get the same error preserving behavior as in development mode when attempting to debug an error that only occurs under production settings. Starting with Flask 0. in pro- duction mode it’s not. By default it’s linked to the setting of DEBUG. Do not force activate PRESERVE_CONTEXT_ON_EXCEPTION in production mode as it will cause your application to leak memory on exceptions.6 Context Preservation on Error If an error occurs or not. In Flask 0. at the end of the request the request context is popped and all data associated with it is destroyed. During development however that can be problematic as you might want to have the information around for a longer time in case an exception occurred. if an exception occurred.7 you have finer control over that behavior by setting the PRESERVE_CONTEXT_ON_EXCEPTION configuration variable.6 and earlier in debug mode.14. 91 . the request context was not popped so that the interactive debugger can still provide you with important information. If the application is in debug mode the context is preserved.

92 .

A blueprint in Flask is not a pluggable app because it is not actually an application – it’s a set of operations which can be registered on an application. even multiple times. • Register a blueprint on an application for any of these cases when initializing a Flask extension. Blueprints can greatly simplify how large applications work and provide a central means for Flask extensions to register operations on applications. A blueprint does not have to implement applications or view func- tions. initialize several extensions. 15. templates.7. Why not have multiple application objects? You can do that (see Application Dispatching). A Blueprint object works similarly to a Flask application object. and other utilities through blueprints. Flask uses a concept of blueprints for making application components and supporting common patterns within an application or across applications. 93 . Rather it is a blueprint of how to construct or extend an application. • Provide template filters. • Register a blueprint on an application at a URL prefix and/or subdomain. static files. • Register a blueprint multiple times on an application with different URL rules. and register a collection of blueprints. CHAPTER FIFTEEN MODULAR APPLICATIONS WITH BLUEPRINTS New in version 0. Pa- rameters in the URL prefix/subdomain become common view arguments (with defaults) across all view functions in the blueprint. but it is not actually an application.1 Why Blueprints? Blueprints in Flask are intended for these cases: • Factor an application into a set of blueprints. but your applications will have separate configs and will be managed at the WSGI layer. This is ideal for larger applications. a project could instantiate an application object.

simple_page import simple_page app = Flask(__name__) app. 15.3 My First Blueprint This is what a very basic blueprint looks like.2 The Concept of Blueprints The basic concept of blueprints is that they record operations to execute when reg- istered on an application. Additionally it will prefix the endpoint of the function with the name of the blueprint which was given to the Blueprint constructor (in this case also simple_page). __name__. template_folder='templates') @simple_page. Flask associates view functions with blueprints when dis- patching requests and generating URLs from one endpoint to another. The downside is that you cannot unregister a blueprint once an application was created without having to destroy the whole application object.4 Registering Blueprints So how do you register that blueprint? Like this: from flask import Flask from yourapplication.register_blueprint(simple_page) 94 .html' % page) except TemplateNotFound: abort(404) When you bind a function with the help of the @simple_page. render_template.route('/<page>') def show(page): try: return render_template('pages/%s.route decorator the blueprint will record the intention of registering the function show on the application when it’s later registered. defaults={'page': 'index'}) @simple_page. and can change an application object as necessary with being registered. abort from jinja2 import TemplateNotFound simple_page = Blueprint('simple_page'.Blueprints instead provide separation at the Flask level. 15. In this case we want to implement a blueprint that does simple rendering of static templates: from flask import Blueprint. 15. share application config.route('/'.

GET) -> static>.1 Blueprint Resource Folder Like for regular applications. they are also prefixed with the name of the blueprint and separated by a dot (. <Rule '/<page>' (HEAD. 15.show>. OPTIONS.If you check the rules registered on the application. blueprints are considered to be contained in a folder. you will find these: [<Rule '/static/<filename>' (HEAD. Sometimes you might want to introduce a blueprint only for the resources it provides. Blueprints however can also be mounted at different locations: app. If it’s a module. GET) -> simple_page. <Rule '/' (HEAD.root_path property to see what the resource folder is: >>> simple_page. OPTIONS. url_prefix='/pages') And sure enough.show>.show>] On top of that you can register blueprints multiple times though not every blueprint might respond properly to that.5 Blueprint Resources Blueprints can provide resources as well. <Rule '/pages/' (HEAD. This argument specifies what logical Python module or package corre- sponds to the blueprint.register_blueprint(simple_page. OPTIONS. The other two are for the show function of the simple_page blueprint. OPTIONS. 15. While multiple blueprints can originate from the same folder.show>] The first one is obviously from the application itself for the static files. The folder is inferred from the second argument to Blueprint which is usually __name__. OPTIONS.).5. these are the generated rules: [<Rule '/static/<filename>' (HEAD. As you can see. GET) -> simple_page. You can access the Blueprint. In fact it depends on how the blueprint is imple- mented if it can be mounted more than once. GET) -> simple_page. GET) -> simple_page.root_path '/Users/username/TestProject/yourapplication' To quickly open sources from this folder you can use the open_resource() function: 95 . If it points to an actual Python package that package (which is a folder on the filesystem) is the resource folder. GET) -> static>. OPTIONS. it does not have to be the case and it’s usually not recommended. the package the mod- ule is contained in will be the resource folder. <Rule '/pages/<page>' (HEAD.

This also means that if you don’t want a blueprint template to be accidentally overridden.5. Because the folder is called static here it will be available at the location of the blueprint + /static. That way you can easily override templates that a blueprint provides in the actual application.static'.css') 15. So if you have a blueprint in the folder yourapplication/admin and you want to ren- der the template 'admin/index.static so you can generate URLs to it like you would do to the static folder of the application: url_for('admin. The endpoint is named blueprint_name. Say the blueprint is registered for /admin the static folder will be at /admin/ static. make sure that no other blueprint or actual application template has the same relative path. It can either be an absolute path or one relative to the folder of the blueprint: admin = Blueprint('admin'.html which is specific to this blueprint.html' and you have provided templates as a tem- plate_folder you will have to create a file like this: yourapplication/admin/templates/ admin/index. __name__.with simple_page. template_folder='templates') For static files. When mul- tiple blueprints provide the same relative template path the first blueprint registered takes precedence over the others.css') as f: code = f.3 Templates If you want the blueprint to expose templates you can do that by providing the tem- plate_folder parameter to the Blueprint constructor: admin = Blueprint('admin'.open_resource('static/style. To further reiterate this: if you have a blueprint named admin and you want to render a template called index. __name__. filename='style. The reason for the extra admin folder is to avoid getting our tem- plate overridden by a template named index. the best idea is to lay out your templates like this: 96 .read() 15. the path can be absolute or relative to the blueprint resource folder.html in the actual application template folder.2 Static Files A blueprint can expose a folder with static files by providing a path to a folder on the filesystem via the static_folder keyword argument.5. The template folder is added to the search path of templates but with a lower prior- ity than the actual application’s template folder.html. static_folder='static') By default the rightmost part of the path is where it is exposed on the web.

15.errorhandler(404) def page_not_found(e): return render_template('pages/404.html as the name to look up the template by. so it is easy to make Blueprint-specific custom error pages. use admin/index. 97 . you can use relative redi- rects by prefixing the endpoint with a dot only: url_for('.index') This will link to admin. Here is an example for a “404 Page Not Found” exception: @simple_page.yourpackage/ blueprints/ admin/ templates/ admin/ index.index for instance in case the current request was dispatched to any other admin blueprint endpoint.7 Error Handlers Blueprints support the errorhandler decorator just like the Flask application object.html') More information on error handling see Custom Error Pages.): url_for('admin. If you encounter problems loading the correct templates enable the EXPLAIN_TEMPLATE_LOADING config variable which will instruct Flask to print out the steps it goes through to locate templates on every render_template call. 15.6 Building URLs If you want to link from one page to another you can use the url_for() function just like you normally would do just that you prefix the URL endpoint with the name of the blueprint and a dot (.html __init__.py And then when you want to render the template.index') Additionally if you are in a view function of a blueprint or a rendered template and you want to link to another endpoint of the same blueprint.

98 .

There are no general rules in how extensions are supposed to behave but they are imported from common locations. If you add a Flask extension as dependency to your requirements. If this is the case.foo or flask_foo depending on how the extension is 99 . 16.txt or setup.4 Flask Before 0. instead you have to import from flaskext.7 or earlier the flask.ext package will not exist. you can always create your own.1 Finding Extensions Flask extensions are listed on the Flask Extension Registry and can be downloaded with easy_install or pip. you may not find an extension that fits your need. Consider reading Flask Extension Development to develop your own Flask extension. 16. For in- stance they add support for databases and other common tasks.2 Using Extensions Extensions typically have documentation that goes along that shows how to use it. If you have an extension called Flask-Foo or Foo-Flask it should be always importable from flask_foo: import flask_foo 16.8 If you are using Flask 0.py file they are usually installed with a simple command or when your application installs.3 Building Extensions While Flask Extension Registry contains many Flask extensions. CHAPTER SIXTEEN FLASK EXTENSIONS Flask extensions extend the functionality of Flask in various different ways. 16.

If you want to develop an application that supports Flask 0.py And here is how you can use it: import flaskext_compat flaskext_compat.ext import foo Once the flaskext_compat module is activated the flask. We provide you with a compatibility module that provides this package for older versions of Flask. 100 .ext package.distributed.ext will exist and you can start importing from there.7 or earlier you should still import from the flask.activate() from flask. You can download it from GitHub: flaskext_compat.

This is achieved by exporting the FLASK_APP environment variable.py flask run 101 .py file with the application in it named app this is how it can be run. The way this script works is by providing access to all the commands on your Flask application’s Flask. For instance mymodule:application would tell it to use the application object in the mymodule. One of the nice new features in Flask 0. It can be either set to an import path or to a filename of a Python module that contains a Flask application. Flask extensions can also register more commands there if they desire so. In that imported file the name of the app needs to be called app or optionally be speci- fied after a colon. This enables a wide range of new features for the Flask ecosystem and your own applications.cli instance as well as some built-in commands that are always there.11 is the built-in integration of the click com- mand line interface. Environment variables (On Windows use set instead of export): export FLASK_APP=hello flask run Or with a filename: export FLASK_APP=/path/to/hello.11. 17. For the flask script to work. If you don’t want to install Flask or you have a special use-case you can also use python -m flask to accomplish exactly the same. Given a hello.1 Basic Usage After installation of Flask you will now find a flask script installed into your vir- tualenv. CHAPTER SEVENTEEN COMMAND LINE INTERFACE New in version 0.py file. an application needs to be discovered.

For instance if you want a shell command to initialize the database you can do this: import click from flask import Flask app = Flask(__name__) @app. Flask uses click for the command interface which makes creating custom commands very easy.5 Custom Commands If you want to add more commands to the shell script you can do this easily. 17.""" click.2 Virtualenv Integration If you are constantly working with a virtualenv you can also put the export FLASK_APP into your activate script by adding it to the bottom of the file.echo('Init the db') The command will then show up on the command line: 102 .3 Debug Flag The flask script can also be instructed to enable the debug mode of the application automatically by exporting FLASK_DEBUG. This is done by invoking the Flask.cli. make_shell_context() method of the application.17. If set to 1 debug is enabled or 0 disables it: export FLASK_DEBUG=1 17. That way every time you activate your virtualenv you automatically also activate the correct application name.4 Running a Shell To run an interactive Python shell you can use the shell command: flask shell This will start up an interactive Python shell.command() def initdb(): """Initialize the database. By default you have access to your app and g. setup the correct application context and setup the local variables in the shell. 17.

py with these contents: import os from yourapplication import create_app app = create_app(os.environ['YOURAPPLICATION_CONFIG']) Once this has happened you can make the flask command automatically pick it up: export YOURAPPLICATION_CONFIG=/path/to/config. Because of this.cfg export FLASK_APP=/path/to/autoapp. Be- cause of this reason the recommendation is to create a separate file that instantiates applications. This is not the only way to make this work. This behavior is not available if a command is added later with add_command() or through other means.command(with_appcontext=False) def example(): pass 17. Flask won’t be able to figure out how to instantiate your application properly by itself.cli.6 Application Context Most commands operate on the application so it makes a lot of sense if they have the application context setup.$ flask initdb Init the db 17.py From this point onwards flask will find your application. with_appcontext() which informs the cli system to ensure that an application con- text is set up. It can also be disabled by passing with_appcontext=False to the decorator: @app. This could be a file named autoapp.7 Factory Functions In case you are using factory functions to create your application (see Application Facto- ries) you will discover that the flask command cannot work with them directly. 103 . cli with the command() the callback will automatically be wrapped through cli. For instance if you have a factory function that creates an application from a filename you could make a separate file that creates such an application from an environment variable. if you register a callback on app. Another is the Custom Scripts support.

cli package. Next step is to create a FlaskGroup. create_app=create_wiki_app) def cli(): """This is a management script for the wiki application. This is very useful if you write reusable applications that you want to ship to users and they should be presented with a custom management script.py script that manages a hypothetical wiki application. This can either directly import an application object or create it (see Application Factories). 3. This is simple but it has some limitations. so let’s go through all parts step by step.FlaskGroup click group.environ. you can also make your own “driver scripts”. This is necessary so that the commands know which Flask application they have to work with. With a custom script you don’t have this problem as you can fully customize how the application will be created.cli will expect to be (indirectly at least) launched from a flask. 'wikiconfig.cli.py')) @click. Since Flask uses click for the scripts there is no reason you cannot hook these scripts into any click application. Primarily we are here interested in the FlaskGroup click group. In this case we just make an empty function with a help doc string that just does nothing and then pass the create_wiki_app 104 . Primarily it does not work with application factory functions (see Application Factories). here is an example manage. To explain all of this. If you use the flask script you specify the application to work with on the command line or environment variable as an import name. 1. In this case we load the config from an environment vari- able. 2. There is one big caveat and that is.8 Custom Scripts While the most common way is to use the flask command.get('WIKI_CONFIG'. First we import the click library as well as the click extensions from the flask. To understand why you might want custom scripts you need to understand how click finds and executes the Flask application. The next thing we do is defining a function that is invoked with the script info object (ScriptInfo) from Flask and its purpose is to fully import and create the application.17.cli import FlaskGroup def create_wiki_app(info): from yourwiki import create_app return create_app( config=os. that commands registered to Flask.group(cls=FlaskGroup.""" if __name__ == '__main__': cli() That’s a lot of code for not much. We will go through the details afterwards: import os import click from flask.

All is rounded up by invoking the script. function as a factory function.cli instance with more commands if they want.9 CLI Plugins Flask extensions can always patch the Flask. 105 . However there is a second way to add CLI plugins to Flask which is through setuptools. entry_points=''' [flask. This is useful to provide extra functionality that Flask itself cannot ship. 4. 17. If you make a Python package that should export a Flask command line plugin you can ship a setup.commands:cli '''.py file that declares an entrypoint that points to a click command: Example setup... ) Inside mypackage/commands.py you can then export a Click object: import click @click.""" Once that package is installed in the same virtualenv as Flask itself you can run flask my-command to invoke your command.commands] my-command=mypackage. .command() def cli(): """This is an example command.py: from setuptools import setup setup( name='flask-my-extension'. Whenever click now needs to operate on a Flask application it will call that func- tion with the script info and ask for it to be created.

106 .

run() method. For instance the reloader can be disabled: $ flask run --no-reload 18.run() method. This will immediately launch a local server exactly the same way the flask script does. The individual features of the server can be controlled by passing more arguments to the run option. The best one is the flask command line utility but you can also continue using the Flask.1 Command Line The flask command line script (Command Line Interface) is strongly recommended for development because it provides a superior reload experience due to how it loads the application. The reason for this 107 .11 onwards the flask method is recommended.11 there are multiple built-in ways to run a development server. 18.2 In Code The alternative way to start the application is through the Flask. The basic usage is like this: $ export FLASK_APP=my_application $ export FLASK_DEBUG=1 $ flask run This will enable the debugger. the reloader and then start the server on http://localhost:5000/. CHAPTER EIGHTEEN DEVELOPMENT SERVER Starting with Flask 0.run() This works well for the common case but it does not work well for development which is why from Flask 0. Example: if __name__ == '__main__': app.

sometimes crashing without message or dying when a syntax or import error happens). It is however still a perfectly valid method for invoking a non automatic reloading application.is that due to how the reload mechanism works there are some bizarre side-effects (like executing certain code twice. 108 .

but also for unittesting and other situations that require a faked request context. Flask itself does not come with an interactive shell. It basically allows you to execute Python commands in real time and immediately get results back.2 Creating a Request Context The easiest way to create a proper request context from the shell is by using the test_request_context method which creates us a RequestContext: >>> ctx = app. so what can you do? This is where some helper functions come in handy. 19. For more information see Command Line Interface.1 Command Line Interface Starting with Flask 0. One of the reasons everybody loves Python is the interactive shell.3. For instance the shell is automatically initialized with a loaded application context. Generally it’s recommended that you read the The Request Context chapter of the doc- umentation first. There are however some handy helpers to make playing around in the shell a more pleasant experience.test_request_context() 109 . 19. But the code you want to test might depend on them.11 the recommended way to work with the shell is the flask shell command which does a lot of this automatically for you. just import your application and start playing around. request and others are not available. The main issue with interactive console sessions is that you’re not triggering a request like a browser does which means that g. Keep in mind however that these functions are not only there for interactive shell usage. because it does not require any specific setup upfront. CHAPTER NINETEEN WORKING WITH THE SHELL New in version 0.

3 Firing Before/After Request By just creating a request context.test_request_context() >>> ctx. in that case just ignore it. create yourself a module with stuff you want to star import into your interactive session. dropping tables etc.pop() The functions registered as teardown_request() are automatically called when the context is popped. This however can easily be done yourself. There you could also define some more helper methods for common things such as initializing the database. This might result in your database being unavailable if you are connecting to the database in a before-request callback or the current user not being stored on the g object etc. you need to trick a bit before the after request functions (trig- gered by process_response()) operate on a response object: >>> app. but in the shell it’s easier to use the push() and pop() methods by hand: >>> ctx.push() >>> app.preprocess_request() Keep in mind that the preprocess_request() function might return a response object.response_class()) <Response 0 bytes [200 OK]> >>> ctx. Just put them into a module (like shelltools) and import from there: >>> from shelltools import * 110 . you still don’t have run the code that is normally run before a request.process_response(app. 19.push() From that point onwards you can work with the request object until you call pop: >>> ctx. So this is the perfect place to automatically tear down resources that were needed by the request context (such as database connections).Normally you would use the with statement to make this request object active.4 Further Improving the Shell Experience If you like the idea of experimenting in a shell.pop() 19. Just call preprocess_request(): >>> ctx = app. To shutdown a request.

py. CHAPTER TWENTY PATTERNS FOR FLASK Certain things are common enough that the chances are high you will find them in most web applications. That is quite simple. the database connection is closed again. (Make sure to delete all . For example quite a lot of applications are using relational databases and user authentication..1. 20. 20. just create a new folder yourapplication inside the existing one and move everything below it. chances are they will open a database connection at the beginning of the request and get the information of the currently logged in user.pyc files first.css /templates layout.html .py to __init__. otherwise things would most likely break) You should then end up with something like that: /yourapplication /yourapplication 111 . There are more user contributed snippets and patterns in the Flask Snippet Archives..1 Simple Packages To convert that into a larger one.html index. At the end of the request.1 Larger Applications For larger applications it’s a good idea to use a package instead of a module.html login. In that case. Then rename yourapplication.py /static style. Imagine a small application looks like this: /yourapplication yourapplication.

html index. ) In order to run the application you need to export an environment variable that tells Flask where to find the application instance: export FLASK_APP=yourapplication If you are outside of the project directory make sure to provide the exact path to your application directory..html login. packages=['yourapplication'].css /templates layout. That way each module can import it safely and the __name__ variable will resolve to the correct package.py file. __init__.. Similiarly you can turn on “debug mode” with this environment variable: export FLASK_DEBUG=true In order to install and run the application you need to issue the following commands: pip install -e .py next to the inner yourapplication folder with the following contents: from setuptools import setup setup( name='yourapplication'. ]. But that is not a big problem.py file. But how do you run your application now? The naive python yourapplication/ __init__.py will not work. The only thing you have to remember is the following quick checklist: 1. include_package_data=True.html . 112 . all the view functions (the ones with a route() decorator on top) have to be imported in the __init__.py /static style. but the module it is in. Not the object itself. the Flask application object creation has to be in the __init__. install_requires=[ 'flask'. Let’s just say that Python does not want modules in pack- ages to be the startup file. Import the view module after the application object is created. flask run What did we gain from this? Now we can restructure the application a bit into multiple modules. just add a new file called setup. 2.

Circular Imports Every Python programmer hates them..py views. In this case views.py).css /templates layout.html index. 113 .Here’s an example __init__.views And this is what views. For a gentle intro- duction into this topic refer to the Modular Applications with Blueprints chapter of the documentation. Check out the Becoming Big section for some inspiration how to deal with that.1. 20..2 Working with Blueprints If you have larger applications it’s recommended to divide them into smaller groups where each group is implemented with the help of a blueprint.py: from flask import Flask app = Flask(__name__) import yourapplication. and yet we just added some: circular imports (That’s when two modules depend on each other.html login. There are still some problems with that approach but if you want to use decorators there is no way around that. The reason for this is that we are not actually using the views in __init__.py /static style.py /yourapplication __init__.route('/') def index(): return 'Hello World!' You should then end up with something like that: /yourapplication setup.html .py and just ensuring the module is imported and we are doing that at the bottom of the file.py would look like: from yourapplication import app @app. Be advised that this is a bad idea in general but here it is actually fine.py depends on __init__.

you can have multiple instances of the same application running in the same application process which can be handy.views.20.route('/') def index(): return render_template(current_app. Imagine you want to run different versions of the same ap- plication. you can then create multiple instances of this app later.register_blueprint(admin) app.register_blueprint(frontend) return app The downside is that you cannot use the application object in the blueprints at import time. Testing.config.2. So why would you want to do this? 1. __name__.init_app(app) from yourapplication.admin import admin from yourapplication. but if you use factories. Of course you could have multiple instances with different configs set up in your webserver. url_prefix='/admin') @admin. You can however use it from within a request.config['INDEX_TEMPLATE']) 114 . Blueprint. A common pattern is creating the application object when the blueprint is imported.2 Application Factories If you are already using packages and blueprints for your application (Modular Appli- cations with Blueprints) there are a couple of really nice ways to further improve the experience.views.from_pyfile(config_filename) from yourapplication. 2.frontend import frontend app. Like this: def create_app(config_filename): app = Flask(__name__) app.1 Basic Factories The idea is to set up the application in a function. But if you move the creation of this object into a function. You can have instances of the application with different settings to test every case. How do you get access to the application with the config? Use current_app: from flask import current_app. So how would you then actually implement that? 20. Multiple instances.model import db db. render_template admin = Blueprint('admin'.

as an example.py (or equivalent): db = SQLAlchemy() and in your application. 20. in model.Here we look up the name of a template in the config.py (or equivalent): def create_app(config_filename): app = Flask(__name__) app. so one extension object can be used for multiple apps.2.from_pyfile(config_filename) db = SQLAlchemy(app) But.init_app(app) Using this design pattern.config.model import db db. For more information about the design of extensions refer to Flask Extension Development.3 Using Applications So to use such an application you then have to create the application first in a sep- arate file otherwise the flask command won’t be able to find it. no application-specific state is stored on the extension ob- ject.from_pyfile(config_filename) from yourapplication.2 Factories & Extensions It’s preferable to create your extensions and app factories so that the extension object does not initially get bound to the application. rather.cfg') It can then be used with the flask command: export FLASK_APP=exampleapp flask run 115 . 20.py file that creates such an application: from yourapplication import create_app app = create_app('/path/to/config. Using Flask-SQLAlchemy.2. Here an example exampleapp.config. you should not do something along those lines: def create_app(config_filename): app = Flask(__name__) app.

call a function from a blueprint when the application is setting up so that you have a place to modify attributes of the application (like hooking in before / after request handlers etc.20. You can combine not only Flask applications but any WSGI application. The fundamental difference from the module approach is that in this case you are run- ning the same or different Flask applications that are entirely isolated from each other. For production.1 Working with this Document Each of the techniques and examples below results in an application object that can be run with any WSGI server.4 Factory Improvements The factory function from above is not very clever so far. 20.) 3. you can improve it. 20.2. debugging must be enabled both on the ap- plication and the simple server. They run different configurations and are dispatched on the WSGI level.serving. make it possible to pass in configuration values for unittests so that you don’t have to create config files on the filesystem 2.3. This would allow you to run a Django and a Flask application in the same interpreter side by side if you want. application. The usefulness of this depends on how the applications work internally.debug = True 116 . The following changes are straightforward and possible: 1. Werkzeug provides a builtin server for development available at werkzeug. see Deployment Options. run_simple(): from werkzeug. Here is the “hello world” example with debugging and run_simple: from flask import Flask from werkzeug.serving import run_simple run_simple('localhost'. For development. Use a full-blown WSGI server. use_reloader=True) Note that run_simple is not intended for use in production. 5000. In order to use the interactive debugger.3 Application Dispatching Application dispatching is the process of combining multiple Flask applications on the WSGI level. Add in WSGI middlewares when the application is creating if necessary.serving import run_simple app = Flask(__name__) app.

You write your own WSGI application that looks at the request that comes and delegates it to your Flask application. The idea here is that each Flask application is a valid WSGI application and they are combined by the dispatcher middleware into a larger one that is dispatched based on prefix. use_evalex=True) 20. it is dynamically created and remem- bered: 117 .3 Dispatch by Subdomain Sometimes you might want to use multiple instances of the same application with different configurations.2 Combining Applications If you have entirely separated applications and you want them to work next to each other in the same Python interpreter process you can take advantage of the werkzeug.@app. use_debugger=True. Once you have your server set up to listen on all subdomains you can use a very simple WSGI application to do the dynamic application creation.wsgi import DispatcherMiddleware from frontend_app import application as frontend from backend_app import application as backend application = DispatcherMiddleware(frontend. 5000. that is really easy to implement. For example you could have your main application run on / and your backend inter- face on /backend: from werkzeug.DispatcherMiddleware. If that application does not exist yet. For instance you configure your webserver to dispatch all requests for all subdomains to your application and you then use the subdomain information to create user-specific in- stances. A very common example would be creating applications per subdomain.3.route('/') def hello_world(): return 'Hello World!' if __name__ == '__main__': run_simple('localhost'. Assuming the application is created inside a function and you can call that function to instantiate it. The perfect level for abstraction in that regard is the WSGI layer.3. app. In order to develop your application to support creating new instances in functions have a look at the Application Factories pattern. { '/backend': backend }) 20. wsgi. use_reloader=True.

') with self. # We can then just return the NotFound() exception as # application which will render a default 404 page. create_app): self.rstrip('. start_response) This dispatcher can then be used like this: from myapplication import create_app.instances[subdomain] = app return app def __call__(self. Instead of looking at the Host header to figure out the subdomain one simply looks at the request path up to the first slash: 118 .split(':')[0] assert host.create_app = create_app self. 'Configuration error' subdomain = host[:-len(self. start_response): app = self. environ.lock: app = self.exceptions import NotFound def make_app(subdomain): user = get_user_for_subdomain(subdomain) if user is None: # if there is no user for that subdomain we still have # to return a WSGI application that handles that request.endswith(self. domain.get_application(environ['HTTP_HOST']) return app(environ.domain)].get(subdomain) if app is None: app = self.4 Dispatch by Path Dispatching by a path on the URL is very similar. make_app) 20.instances = {} def get_application(self.3.create_app(subdomain) self.instances. get_user_for_subdomain from werkzeug.domain).lock = Lock() self.from threading import Lock class SubdomainDispatcher(object): def __init__(self.com'.domain = domain self. host): host = host. # You might also redirect the user to the main page then return NotFound() # otherwise create the application for the specific user return create_app(user) application = SubdomainDispatcher('example.

make_app) 20. One of the first things that developers run into is the realization that the builtin exceptions are not expressive enough for APIs and that the content type of text/html they are emitting is not very useful for API consumers.4 Implementing API Exceptions It’s very common to implement RESTful APIs on top of Flask.instances[prefix] = app return app def __call__(self.create_app(prefix) if app is not None: self.instances = {} def get_application(self.lock: app = self. The better solution than using abort to signal errors for invalid API usage is to im- 119 . get_user_for_prefix def make_app(prefix): user = get_user_for_prefix(prefix) if user is not None: return create_app(user) application = PathDispatcher(default_app. start_response): app = self. default_app.instances.get_application(peek_path_info(environ)) if app is not None: pop_path_info(environ) else: app = self. peek_path_info class PathDispatcher(object): def __init__(self.default_app return app(environ. environ.lock = Lock() self.default_app = default_app self.wsgi import pop_path_info.from threading import Lock from werkzeug. prefix): with self. create_app): self.create_app = create_app self.get(prefix) if app is None: app = self. start_response) The big difference between this and the subdomain one is that this one falls back to another application if the creator function returns None: from myapplication import create_app. default_app.

plement your own exception type and install an error handler for it that produces the errors in the format the user is expecting. The reason for this is that there is no handler registered for this error class.4.message = message if status_code is not None: self. status_code=None. This is a simple example: from flask import jsonify class InvalidUsage(Exception): status_code = 400 def __init__(self. a status code for the error and some optional payload to give more context for the error. payload=None): Exception. Additionally some extra payload can be provided as a dictionary through the payload parameter. 20. but it would immediately result in an internal server error.4.status_code = status_code self.to_dict()) response.status_code = error.1 Simple Exception Class The basic idea is to introduce a new exception that can take a proper human readable message. message.2 Registering an Error Handler At that point views can raise that error.4.payload = payload def to_dict(self): rv = dict(self.__init__(self) self.errorhandler(InvalidUsage) def handle_invalid_usage(error): response = jsonify(error.message return rv A view can now raise that exception with an error message.3 Usage in Views Here is how a view can use that functionality: 120 . 20.status_code return response 20. That however is easy to add: @app.payload or ()) rv['message'] = self.

The code below checks if the language code is not yet in the dictionary of URL values and if the endpoint wants a value named 'lang_code': @app. URL processors are especially helpful when combined with blueprints.. The idea is that you might have a bunch of resources with common parts in the URL that you don’t always explicitly want to provide.5. 20. @app.route('/<lang_code>/') def index(lang_code): g. For the latter..7 introduces the concept of URL processors. status_code=410) 20.. They can automatically inject values into a call for url_for() automatically. We will handle both application specific URL processors here as well as blueprint specifics.route('/<lang_code>/about') def about(lang_code): g. For instance you might have a bunch of URLs that have the language code in it but you don’t want to have to handle it in every single function yourself.lang_code = lang_code .5 Using URL Processors New in version 0.lang_code: 121 . Sure.7. values): if 'lang_code' in values or not g.lang_code = lang_code . This is an awful lot of repetition as you have to handle the language code setting on the g object yourself in every single function.1 Internationalized Application URLs Consider an application like this: from flask import Flask. g app = Flask(__name__) @app. this is where url_defaults() functions come in..route('/foo') def get_foo(): raise InvalidUsage('This view is gone'. a decorator could be used to simplify this.@app. Flask 0.url_defaults def add_language_code(endpoint. but if you want to generate URLs from one function to another you would have to still provide the language code explicitly which can be annoying.

values): if 'lang_code' in values or not g. Furthermore blueprints can have per- 122 . return if app. Once the 'lang_code' is popped from the values dictionary and it will no longer be forwarded to the view function reducing the code to this: from flask import Flask. values): g.2 Internationalized Blueprint URLs Because blueprints can automatically prefix all URLs with a common string it’s easy to automatically do that for every function..url_value_preprocessor def pull_lang_code(endpoint.url_map.lang_code = values.is_endpoint_expecting(endpoint.5.route('/<lang_code>/') def index(): .lang_code The method is_endpoint_expecting() of the URL map can be used to figure out if it would make sense to provide a language code for the given endpoint. They are executed right after the request was matched and can execute code based on the URL values.. values): g. The reverse of that function are url_value_preprocessor()s. 'lang_code'): values['lang_code'] = g. The idea is that they pull information out of the values dictionary and put it somewhere else: @app.. g app = Flask(__name__) @app.pop('lang_code'.pop('lang_code'.route('/<lang_code>/about') def about(): .url_map. @app.. 20. 'lang_code'): values['lang_code'] = g. but the more beautiful solution is using a blueprint.lang_code: return if app.url_value_preprocessor def pull_lang_code(endpoint.lang_code = values.lang_code @app. You can further improve that by writing your own decorator that prefixes URLs with the language code.is_endpoint_expecting(endpoint. None) @app. None) That way you no longer have to do the lang_code assignment to g in every function.url_defaults def add_language_code(endpoint.

. is an extension library that is commonly used to distribute Python libraries and extensions.4) installed from python.route('/about') def about(): . This makes it possible to query information provided by one package from another package.lang_code = values. you will need to install them yourself. It extends distutils. you will already have pip and setuptools on your system. The best known feature of this system is the entry point sup- port which allows one package to declare an “entry point” that another package can hook into to extend the other package.pop('lang_code') @bp.9) or Python 3 (>=3. __name__. Flask itself.. 20. a basic module installation system shipped with Python to also support various more complex constructs that make larger applications easier to distribute: • support for dependencies: a library or application can declare a list of other libraries it depends on which will be installed automatically for you.blueprint URL processors which removes a whole lot of logic from the url_defaults() function because it no longer has to check if the URL is really interested in a 'lang_code' parameter: from flask import Blueprint. g bp = Blueprint('frontend'.lang_code) @bp. @bp. g.. • package registry: setuptools registers your package with your Python installa- tion. Otherwise. values): values. and all the libraries you can find on PyPI are distributed with either setup- tools or distutils.url_value_preprocessor def pull_lang_code(endpoint.route('/') def index(): .setdefault('lang_code'.6 Deploying with Setuptools Setuptools. If you have Python 2 (>=2. 123 . values): g. • installation manager: pip can install other libraries for you.org.7.url_defaults def add_language_code(endpoint. url_prefix='/<lang_code>') @bp..

If you have not yet converted your application into a package. find_packages setup( .. include_package_data=True. version='1.py next to your application. packages=find_packages() ) Most parameters to the setup function should be self explanatory. head over to the Larger Applications pattern to see how this can be done.6. also read the Deploying with Fabric chapter. A basic setup. The zip_safe flag can be used to force or prevent zip Archive creation.py and you are not using a module. 20. If you want to fully automate the process. If you want setuptools to lookup the packages for you automatically.In this case we assume your application is called yourapplication. long_description=__doc__. We will use this to distribute the static files and templates along with the Python module (see Distributing Resources).1 Basic Setup Script Because you have Flask installed. 124 . but because everybody will look for a file with that name. packages=['yourapplication'].. Your setup code always goes into a file named setup. install_requires=['Flask'] ) Please keep in mind that you have to list subpackages explicitly. include_package_data and zip_safe might not be.in file and install all the entries that match as package data. Flask already depends upon setuptools. The name of the file is only convention. In general you probably don’t want your packages to be installed as zip files because some tools do not support them and they make debugging a lot harder. but a package. you better not change it. zip_safe=False. Standard disclaimer applies: you better use a virtualenv. include_package_data tells setuptools to look for a MANIFEST. you have setuptools available on your system. A working deployment with setuptools is the first step into more complex and more automated deployment scenarios. you can use the find_packages function: from setuptools import setup.0'.py file for a Flask application looks like this: from setuptools import setup setup( name='Your Application'.

py sdist will create a development package with ”.in file. What if you want to depend on a package that cannot be found on PyPI and won’t be because it is an internal package you don’t want to share with anyone? Just do it as if there was a PyPI entry and provide a list of alternative locations where setuptools should look for tarballs: dependency_links=['http://example. Each item in that list is the name of a package that should be pulled from PyPI on installation.0.6'.in file next to your setup.6.tar. gz.dev20160314.0.2'.<=1. 'BrokenPackage>=0.2 Tagging Builds It is useful to distinguish between release and development builds. dependencies are pulled from PyPI.cfg file to configure these options.6.7. Here some examples: install_requires=[ 'Flask>=0.com/yourfiles'] 125 .py file. is to create a MANIFEST. but you can also provide minimum and maximum version requirements. Running python setup. 'SQLAlchemy>=0. Add a setup.gz. you will notice that folders like static or templates are not installed for you. By default it will always use the most recent version. This file lists all the files that should be added to your tarball: recursive-include yourapplication/templates * recursive-include yourapplication/static * Don’t forget that even if you enlist them in your MANIFEST. 20. [egg_info] tag_build = .dev” and the current date appended: flaskr-1. The reason for this is that setuptools does not know which files to add for you. py release sdist will create a release package with only the version: flaskr-1. they won’t be in- stalled for you unless you set the include_package_data parameter of the setup function to True! 20.6.3 Distributing Resources If you try to install the package you just created.dev tag_date = 1 [aliases] release = egg_info -RDb ‘’ Running python setup. What you should do.0' ] As mentioned earlier.tar.20.4 Declaring Dependencies Dependencies are declared in the install_requires parameter as a list.

5 Installing / Developing To install your application (ideally into a virtualenv) just run the setup. All the functions defined in that file will show up as fab subcom- 126 .1 Creating the first Fabfile A fabfile is what controls what Fabric executes. 20.py script with the install parameter.7. This tutorial assumes the latest version of Fabric. It will install your application into the virtualenv’s site- packages folder and also download and install all dependencies: $ python setup. 20.py install If you are developing on the package and also want the requirements to be installed.7 Deploying with Fabric Fabric is a tool for Python similar to Makefiles but with the ability to execute com- mands on a remote server.py file (Deploying with Setuptools). You can of course use your own favourite server there. Before we get started. 20. but for this example we chose Apache + mod_wsgi because it’s very easy to setup and has a simple way to reload applications without root access.py and executed by the fab command. If you have an internal company server that contains the packages. provide the URL to that server. In combination with a properly set up Python package (Larger Applications) and a good concept for configurations (Configuration Handling) it is very easy to deploy Flask applications to external servers. here a quick checklist of things we have to ensure upfront: • Fabric 1.Make sure that page has a directory listing and the links on the page are pointing to the actual tarballs with their correct filenames as this is how setuptools will find the files. It is named fabfile.6. • In the following example we are using mod_wsgi for the remote servers.0 has to be installed locally.py develop This has the advantage of just installing a link to the site-packages folder instead of copying the data over. you can use the develop command instead: $ python setup. • The application already has to be a package and requires a working setup. You can then continue to work on the code without having to run install again after each change.

These hosts can be defined either in the fabfile or on the command line.com'] def pack(): # build the package local('python setup.py --fullname'.api import * # the user to use for the remote commands env. '/tmp/%s' % filename) # install the package in the application's virtualenv with pip run('/var/www/yourapplication/env/bin/pip install /tmp/%s' % filename) # remove the uploaded package run('rm -r /tmp/%s' % filename) # touch the .hosts = ['server1. capture=False) def deploy(): # figure out the package name and version dist = local('python setup.tar. But obviously you can do that.gz' % dist # upload the package to the temporary folder on the server put('dist/%s' % filename.py sdist --formats=gztar'.wsgi file on the server. If we just have one application server (which the majority of applications will have).wsgi file to trigger a reload in mod_wsgi run('touch /var/www/yourapplication.strip() filename = '%s. To deploy the current version of the code on the remote server you would use this command: $ fab pack deploy However this requires that our server already has the /var/www/yourapplication folder created and /var/www/yourapplication/env to be a virtual environment.2 Running Fabfiles Now how do you execute that fabfile? You use the fab command. This is a basic first example that has the ability to upload the current source code to the server and install it into a pre-existing virtual environment: from fabric.wsgi') 20. creating a command in the fabfile for this is overkill.com'. capture=True). 'server2.example. In that case you 127 . In this case we will add them to the fabfile. They are executed on one or more hosts.example. So how do we bootstrap a new server into our infrastructure? This now depends on the number of servers we want to set up.7.mands.user = 'appuser' # the servers where the commands are executed env. Fur- thermore are we not creating the configuration or .

(See mod_wsgi (Apache) for more information) So now the question is.would probably call it setup or bootstrap and then pass the servername explicitly on the command line: $ fab -H newserver.from_object('yourapplication. So we have to put the configuration in a place where the application will able to find it.cfg' from yourapplication import app The application itself then has to initialize itself like this to look for the config at that environment variable: app = Flask(__name__) app. Create a new Apache config for yourapplication and activate it.environ['YOURAPPLICATION_CONFIG'] = '/var/www/yourapplication/application.7.3 The WSGI File The WSGI file has to import the application and also to set an environment variable so that the application knows where to look for the config.com bootstrap To setup a new server you would roughly do these steps: 1. Create the directory structure in /var/www: $ mkdir /var/www/yourapplication $ cd /var/www/yourapplication $ virtualenv --distribute env 2.7. the application will find the correct configuration file by looking up the YOURAPPLICATION_CONFIG environment variable.cfg) 3.from_envvar('YOURAPPLICATION_CONFIG') This approach is explained in detail in the Configuration Handling section of the docu- mentation.default_config') app.config.cfg files come from? 20. Make sure to ac- tivate watching for changes of the . where do the application.wsgi and application.4 The Configuration File Now as mentioned above.example.wsgi file to the server and the configuration file for the application (eg: application.config.wsgi file so that we can automatically reload the application by touching it. This is a short example that does exactly that: import os os. Configuration files 128 . 20. Upload a new application.

Thanks to the setup. We have set up the servers so that they have their virtual environments and activated apache configs. so you do not version them usually. • Put configuration files into a separate version control repository and symlink the active configs into place. in our case here we only expect one or two servers and we can upload them ahead of time by hand. setup apache appropriately etc. 129 .5 First Deployment Now we can do our first deployment. A popular approach is to store configuration files for different servers in a sepa- rate version control repository and check them out on all servers.7. 20. That way you can also easily go back to older versions.have the unfriendly quality of being different on all computers.7. It could initialize a new virtual environment. 20. First it will execute pack so that we have our tarball ready and then it will execute deploy and upload the source code to all servers and install it there. • hook in testing functionality so that you can deploy to an external server and run the test suite. • You could also put your application code into a repository and check out the latest version on the server and then install. Now we can pack up the application and deploy it: $ fab pack deploy Fabric will now connect to all servers and run the commands as written down in the fabfile. Either way.py file we will automatically pull in the required libraries into our virtual environment.6 Next Steps From that point onwards there is so much that can be done to make deployment actu- ally fun: • Create a bootstrap command that initializes new servers. Working with Fabric is fun and you will notice that it’s quite magical to type fab deploy and see your application being deployed automatically to one or more remote servers. Then symlink the file that is active for the server into the location where it’s expected (eg: /var/www/ yourapplication).

If you want to use this code outside a request context you can use it in a Python shell by opening the application context by hand: 130 ._app_ctx_stack.9 or older you need to use flask.._database = sqlite3.connect(DATABASE) return db @app. Because of this we have to make sure here that the database is there before we close it.top instead of g as the flask. '_database'. Note: Please keep in mind that the teardown request and appcontext functions are al- ways executed. to use the database.close() Now..20.8.db' def get_db(): db = getattr(g. even if a before-request handler failed or was never executed. None) if db is not None: db. '_database'.route('/') def index(): cur = get_db().cursor() . 20.8 Using SQLite 3 with Flask In Flask you can easily implement the opening of database connections on demand and closing them when the context dies (usually at the end of the request). None) if db is None: db = g.teardown_appcontext def close_connection(exception): db = getattr(g.1 Connect on Demand The upside of this approach (connecting on first use) is that this will only open the connection if truly necessary. At that point the get_db function can be used to get the current database connec- tion. Note: if you use Flask 0. the application must either have an active application con- text (which is always true if there is a request in flight) or create an application context itself. Example: @app. Here is a simple example of how you can use SQLite 3 with Flask: import sqlite3 from flask import g DATABASE = '/path/to/database. Whenever the context is destroyed the database connection will be terminated.g object was bound to the request and not application context.

execute(query. value in enumerate(row)) db.. value) for idx. it is a good idea to provide a query function that combines getting the cursor.Row called r for the rows id. we could place this in get_db instead: db. LastName.description[idx][0].. It is executed for every result returned from the database to convert the result. FirstName. this could be inserted into the get_db function we created above: def make_dicts(cursor.row_factory = sqlite3. so we can access them either by index or by key. Even more simply. as- suming we have a sqlite3.db to get the current open database connection. a row factory function is use- ful.8. print(value) 1 John Doe M Additionally. For example. one=False): cur = get_db(). row): return dict((cursor.row_factory = make_dicts This will make the sqlite3 module return dicts for this database connection. args) rv = cur.fetchall() cur. executing and fetching the results: def query_db(query. For instance. args=().2 Easy Querying Now in each request handling function you can access g.close() return (rv[0] if rv else None) if one else rv 131 .with app.app_context(): # now you can use get_db() 20. you can get them based on index >>> r[1] John # Row objects are also iterable: >>> for value in r: . To simplify working with SQLite. in order to get dictionaries instead of tuples.Row This would use Row objects rather than dicts to return the results of queries. These are namedtuple s. and MiddleInitial: >>> # You can get values based on the row's name >>> r['FirstName'] John >>> # Or. which are much nicer to deal with.

in combination with a row factory.app_context(): db = get_db() with app. It’s a good idea to provide a function that creates the database based on that schema.commit() You can then create such a database from the Python shell: >>> from yourapplication import init_db >>> init_db() 20. 20. 'has the id'. Here is how you can use it: for user in query_db('select * from users'): print user['username']. one=True) if user is None: print 'No such user' else: print the_username.sql'. user['user_id'] To pass variable parts to the SQL statement. so applications often ship a schema. Never directly add them to the SQL statement with string formatting because this makes it possible to attack the application using SQL Injections.3 Initial Schemas Relational databases need schemas. 'has the id'. [the_username]. user['user_id'] Or if you just want a single result: user = query_db('select * from users where username = ?'. This function can do that for you: def init_db(): with app.sql file that creates the database.open_resource('schema.9 SQLAlchemy in Flask Many people prefer SQLAlchemy for database access.This handy little function.executescript(f. mode='r') as f: db. use a question mark in the statement and pass in the arguments as a list. makes working with the database much more pleasant than it is by just using the raw cursor and connection objects.cursor().8.read()) db. In this case it’s encouraged to use a package instead of a module for your flask application and drop the models into 132 .

sessionmaker from sqlalchemy. You can download Flask-SQLAlchemy from PyPI.models Base.9. If you are wondering why we don’t have to care about threads here (like we did in the SQLite3 example above with the g object): that’s because SQLAlchemy does that for us already with the scoped_session. Here’s the example database. autoflush=False. While that is not necessary.2 Declarative The declarative extension in SQLAlchemy is the most recent method of using SQLAlchemy. you just have to put the following code into your application module. It allows you to define tables and models in one go.query = db_session.a separate module (Larger Applications).query_property() def init_db(): # import all modules here that might define models so that # they will be registered properly on the metadata. 20. just subclass the Base class that was created by the code above.orm import scoped_session. To use SQLAlchemy in a declarative way with your application. there is a Flask extension that handles that for you. Flask will automatically remove database sessions at the end of the request or when the application shuts down: 133 . similar to how Django works. This is recommended if you want to get started quickly.9.1 Flask-SQLAlchemy Extension Because SQLAlchemy is a common database abstraction layer and object relational mapper that requires a little bit of configuration effort. Otherwise # you will have to import them first before calling init_db() import yourapplication.declarative import declarative_base engine = create_engine('sqlite:////tmp/test. it makes a lot of sense.py module for your application: from sqlalchemy import create_engine from sqlalchemy.create_all(bind=engine) To define your models. bind=engine)) Base = declarative_base() Base.ext.db'. convert_unicode=True) db_session = scoped_session(sessionmaker(autocommit=False. There are four very common ways to use SQLAlchemy. I will outline each of them here: 20.metadata. In addition to the following text I recommend the official documenta- tion on the declarative extension.

134 .9.name == 'admin').email = email def __repr__(self): return '<User %r>' % (self.all() [<User u'admin'>] >>> User. It’s more flexible but a little more to type. primary_key=True) name = Column(String(50). String from yourapplication. e.database import db_session @app.commit() Querying is simple as well: >>> User.add(u) >>> db_session.py. Integer.): from sqlalchemy import Column.filter(User.query.database import Base class User(Base): __tablename__ = 'users' id = Column(Integer.query.models import User >>> u = User('admin'.first() <User u'admin'> 20. unique=True) def __init__(self.database import db_session >>> from yourapplication. The main difference is that you define tables and classes separately and map them together. email=None): self.name) To create the database you can use the init_db function: >>> from yourapplication.remove() Here is an example model (put this into models. unique=True) email = Column(String(120). 'admin@localhost') >>> db_session.g.3 Manual Object Relational Mapping Manual object relational mapping has a few upsides and a few downsides versus the declarative approach from above. name=None.from yourapplication.database import init_db >>> init_db() You can insert entries into the database like this: >>> from yourapplication.teardown_appcontext def shutdown_session(exception=None): db_session.name = name self.

name=None. db_session class User(object): query = db_session. autoflush=False. Column('email'. email=None): self. convert_unicode=True) metadata = MetaData() db_session = scoped_session(sessionmaker(autocommit=False. bind=engine)) def init_db(): metadata. unique=True) ) mapper(User. 135 . String from sqlalchemy.database import metadata. metadata. users) Querying and inserting works exactly the same as in the example above.remove() Here is an example table and model (put this into models. Put this into your application module: from yourapplication.teardown_appcontext def shutdown_session(exception=None): db_session. String(50).name) users = Table('users'. Integer.database import db_session @app. sessionmaker engine = create_engine('sqlite:////tmp/test.query_property() def __init__(self.email = email def __repr__(self): return '<User %r>' % (self. String(120). Here is an example database. MetaData from sqlalchemy.In general it works like the declarative approach. Column('id'. Integer.orm import mapper from yourapplication.orm import scoped_session.py module for your application: from sqlalchemy import create_engine. primary_key=True).py): from sqlalchemy import Table. Column.db'. you need to close the session after each request or application context shutdown. so make sure to also split up your application into multiple modules in a package. Column('name'.name = name self. unique=True).create_all(bind=engine) As in the declarative approach.

[1]).db'. convert_unicode=True) metadata = MetaData(bind=engine) Then you can either declare the tables in your code like in the examples above. 136 . Table engine = create_engine('sqlite:////tmp/test.id == 1).execute(). We have to get a connection first so that we can use a transaction: >>> con = engine.id == 1).execute(users.first() (1. u'admin'.10 Uploading Files Ah yes. metadata. The basic idea of file uploads is actually quite simple. It basically works like this: 1. u'admin'.first() (1. the good old problem of file uploads. autoload=True) To insert data you can use the insert method.connect() >>> con. MetaData.9.20. or automatically load them: from sqlalchemy import Table users = Table('users'. To query your database.c. you use the engine directly or use a connection: >>> users.insert(). email='admin@localhost') SQLAlchemy will automatically commit for us.4 SQL Abstraction Layer If you just want to use the database system (and SQL) abstraction layer you basically only need the engine: from sqlalchemy import create_engine. head over to the website.first() >>> r['name'] u'admin' You can also pass strings of SQL statements to the execute() method: >>> engine.c. u'admin@localhost') For more information about SQLAlchemy.execute(). name='admin'. u'admin@localhost') These results are also dict-like tuples: >>> r = users.select(users. A <form> tag is marked with enctype=multipart/form-data and an <input type=file> is placed in that form. 20.select(users.execute('select * from users where id = :1'.

Why do we limit the extensions that are allowed? You probably don’t want your users to be able to upload everything there if the server is directly sending out the data to the client. 1)[1]. the werkzeug. 3. use the save() method of the file to save the file permanently somewhere on the filesystem.'. right? :) Next the functions that check if an extension is valid and that uploads the file and redirects the user to the URL for the uploaded file: def allowed_file(filename): return '. 'png'. methods=['GET'.files: flash('No file part') return redirect(request. Also make sure to disallow .filename == '': 137 . Most should be straightforward. 'jpg'.config['UPLOAD_FOLDER'] = UPLOAD_FOLDER So first we need a couple of imports. 2. That way you can make sure that users are not able to upload HTML files that would cause XSS problems (see Cross-Site Scripting (XSS)). 20. Let’s look at the bootstrapping code for our application: import os from flask import Flask. 'POST']) def upload_file(): if request.utils import secure_filename UPLOAD_FOLDER = '/path/to/the/uploads' ALLOWED_EXTENSIONS = set(['txt'. but who has PHP installed on their server.files['file'] # if user does not select file. request.url) file = request.10.rsplit('. url_for from werkzeug. redirect. browser also # submit a empty part without filename if file.route('/'. secure_filename() is explained a little bit later. The application accesses the file from the files dictionary on the request object.php files if the server executes them. 'gif']) app = Flask(__name__) app. 'pdf'. The UPLOAD_FOLDER is where we will store the uploaded files and the ALLOWED_EXTENSIONS is the set of allowed file exten- sions.' in filename and \ filename.method == 'POST': # check if the post request has the file part if 'file' not in request. 'jpeg'.lower() in ALLOWED_EXTENSIONS @app.1 A Gentle Introduction Let’s start with a very basic application that uploads a file to a specific upload folder and displays a file to the user.

/ uploads/filename.filename) file.. Information for the Pros So you’re interested in what that secure_filename() function does and what the prob- lem is if you’re not using it? So just imagine someone would send the following infor- mation as filename to your application: filename = ".route('/uploads/<filename>') 138 ./. As of Flask 0./. All submitted form data can be forged. flash('No selected file') return redirect(request.bashrc" Assuming the number of .config['UPLOAD_FOLDER'].bashrc') 'home_username_.path.. filename=filename)../home/username/. So we write the uploaded_file() function to return the file of that name./home/username/... For the moment just remember: always use that function to secure a filename before storing it directly on the filesystem. and filenames can be dan- gerous..save(os. filename)) return redirect(url_for('uploaded_file'. that is../. filename=filename)) return ''' <!doctype html> <title>Upload new File</title> <h1>Upload new File</h1> <form method=post enctype=multipart/form-data> <p><input type=file name=file> <input type=submit value=Upload> </form> ''' So what does that secure_filename() function actually do? Now the problem is that there is that principle called “never trust user input”..filename): filename = secure_filename(file./..5 we can use a function that does that for us: from flask import send_from_directory @app. This is also true for the filename of an uploaded file. This does require some knowledge about how the application looks like.url) if file and allowed_file(file.join(app.bashrc' Now one last thing is missing: the serving of the uploaded files. but trust me./ is correct and you would join this with the UPLOAD_FOLDER the user might have the ability to modify a file on the server’s filesystem he or she should not modify. In the upload_file() we redirect the user to url_for('uploaded_file'. hackers are patient :) Now let’s look how that function works: >>> secure_filename('././.

139 . Request app = Flask(__name__) app.wsgi_app. 20. But how do you specify the maximum file size after which an upload is aborted? By default Flask will happily accept file uploads to an un- limited amount of memory.def uploaded_file(filename): return send_from_directory(app. For more information on that consult the Werkzeug documentation on file handling. 'uploaded_file'.3 Upload Progress Bars A while ago many developers had the idea to read the incoming file in small chunks and store the upload progress in the database to be able to poll the progress with JavaScript from the client.6 but can be achieved in older versions as well by subclassing the request object.config['UPLOAD_FOLDER'] }) If you now run the application everything should work as expected. Long story short: the client asks the server every 5 seconds how much it has transmitted already. If a larger file is transmitted. 20.6.config['MAX_CONTENT_LENGTH'] = 16 * 1024 * 1024 The code above will limited the maximum allowed payload to 16 megabytes.add_url_rule('/uploads/<filename>'. build_only=True) app.10. This also works with older versions of Flask: from werkzeug import SharedDataMiddleware app. Do you realize the irony? The client is asking for something it should already know.config['UPLOAD_FOLDER']. but you can limit that by setting the MAX_CONTENT_LENGTH config key: from flask import Flask. So how exactly does Flask handle uploads? Well it will store them in the webserver’s memory if the files are reasonable small otherwise in a temporary location (as returned by tempfile.2 Improving Uploads New in version 0. Flask will raise an RequestEntityTooLarge exception.wsgi_app = SharedDataMiddleware(app.gettempdir()). This feature was added in Flask 0. filename) Alternatively you can register uploaded_file as build_only rule and use the SharedDataMiddleware.10. { '/uploads': app.

It supports multiple cache backends.11 Caching When your application runs slow. Flask itself does not provide caching for you.contrib.0. has some very basic cache support. at least it’s the easiest way to speed up things.11. Because the common pattern for file uploads exists almost unchanged in all applica- tions dealing with uploads. What does a cache do? Say you have a function that takes some time to complete but the results would still be good enough if they were 5 min- utes old. similar to how Flask objects are created.cache import MemcachedCache cache = MemcachedCache(['127.1 Setting up a Cache You create a cache object once and keep it around.cache import GAEMemcachedCache cache = GAEMemcachedCache() 140 . normally you want to use a memcached server.contrib. Well. There are JavaScript libraries like jQuery that have form plugins to ease the construction of progress bar. but Werkzeug. that one is a simple cache that keeps the item stored in the memory of the Python interpreter: from werkzeug. make sure to have one of the memcache modules supported (you get them from PyPI) and a memcached server running somewhere. throw some caches in. 20.cache import SimpleCache cache = SimpleCache() If you want to use memcached.0.1:11211']) If you are using App Engine. This is how you connect to such an memcached server then: from werkzeug. So then the idea is that you actually put the result of that calculation into a cache for some time. one of the libraries it is based on.4 An Easier Solution Now there are better solutions that work faster and are more reliable. there is also a Flask extension called Flask-Uploads that implements a full fledged upload mechanism with white and blacklisting of exten- sions and more. you can connect to the App Engine memcache server easily: from werkzeug. If you are using the development server you can create a SimpleCache object.10.20.contrib. 20.

Use functools.set('my-item'.get('my-item') To add items to the cache. This example assumes that the login page is called 'login' and that the current user is stored in g. But there are use cases for implementing your own decorator. Since the original function is replaced.12 View Decorators Python has a really interesting feature called function decorators. dec- orators can be used to inject additional functionality to one or more functions.wraps() to handle this for you. A decorator is a function that wraps and replaces another function. from functools import wraps from flask import g. Otherwise that function will return None: rv = cache. timeout=5 * 60) return rv 20. url_for def login_required(f): 141 . For instance. The first argument is the key and the second the value that should be set. This allows some really neat things for web applications. The route() decorator is the one you probably used already. If a user goes to the site and is not logged in.1 Login Required Decorator So let’s implement such a decorator. If something is in the cache.12.get('my-item') if rv is None: rv = calculate_value() cache. 20. request. you need to remember to copy the original function’s information to the new function. Also a timeout can be provided after which the cache will automatically remove item.2 Using a Cache Now how can one use such a cache? There are two very important operations: get() and set().11. Because each view in Flask is a function. This is a good example of a use case where a decorator is an excellent solution. This is how to use them: To get an item from the cache call get() with a string as key name. they should be redirected to the login page. imagine you have a view that should only be used by people that are logged in. rv. Here a full example how this looks like normally: def get_my_item(): rv = cache. it is returned.user and is None if there is no-one logged in. redirect.20. use the set() method instead.

You can do this with a hidden input tag. but the code should still be straight- forward to read. You’ll have to pass it along when sending the POST request from the login form. Here the code: 142 . get the unique cache key for the current request base on the current path. apply it as innermost decorator to a view function.route('/secret_page') @login_required def secret_page(): pass Note: The next value will exist in request. Sounds aw- ful? Unfortunately it is a little bit more complex. always remember that the route() decorator is the outermost. @wraps(f) def decorated_function(*args. Here is an example cache function. When apply- ing further decorators. The decorated function will then work as follows 1. otherwise the original function is called and the return value is stored in the cache for the timeout provided (by default 5 minutes). If the cache returned something we will return that value. get the value for that key from the cache. **kwargs): if g. 2.args.form when logging the user in.get('next'. A decorator would be nice for that. '') }}"/> 20.12. It generates the cache key from a specific prefix (actually a format string) and the current path of the request.2 Caching Decorator Imagine you have a view function that does an expensive calculation and because of that you would like to cache the generated results for a certain amount of time. then retrieve it from request. We’re assuming you have set up a cache like mentioned in Caching. <input type="hidden" value="{{ request.args after a GET request for the login page. next=request.user is None: return redirect(url_for('login'. @app. **kwargs) return decorated_function To use the decorator. 3.url)) return f(*args. Notice that we are using a function that first creates the decorator that then decorates the function.

That way you can still use the redirect function or return simple strings.route('/') @templated('index. With that.route('/') @templated() def index(): return dict(value=42) As you can see. if something else than a dictionary is returned we return it from the function unchanged. Otherwise the provided template name is used.set(cache_key. Here is the code for that decorator: 143 .html') def index(): return dict(value=42) @app.3 Templating Decorator A common pattern invented by the TurboGears guys a while back is a templating decorator.get(cache_key) if rv is not None: return rv rv = f(*args.path rv = cache. the dictionary returned is passed to the template rendering function. if no template name is provided it will use the endpoint of the URL map with dots converted to slashes + '. the following three examples do exactly the same: @app. rv. 20. **kwargs) cache.12. When the decorated function returns. key='view/%s'): def decorator(f): @wraps(f) def decorated_function(*args.html'.from functools import wraps from flask import request def cached(timeout=5 * 60.route('/') def index(): return render_template('index. see Caching for more information. If None is returned. timeout=timeout) return rv return decorated_function return decorator Notice that this assumes an instantiated cache object is available. The idea of that decorator is that you return a dictionary with the values passed to the template from the view function and the template is automatically ren- dered. an empty dictionary is assumed. **kwargs): cache_key = key % request.html'. value=42) @app.

When you are working with WTForms you have to define your forms as classes first.add(Rule('/'.'. endpoint='index')) @app. dict): return ctx return render_template(template_name. code quickly becomes very hard to read.13 Form Validation with WTForms When you have to work with form data submitted by a browser view.replace('.endpoint('index') def my_index(): return "Hello world" 20. **kwargs): template_name = template if template_name is None: template_name = request.from functools import wraps from flask import request. There are libraries out there designed to make this process easier to manage. you might want to give it a try. If you find yourself in the situation of having many forms.12. '/') + '. One of them is WTForms which we will handle here. **ctx) return decorated_function return decorator 20. 144 .url_map.4 Endpoint Decorator When you want to use the werkzeug routing system for more flexibility you need to map the endpoint as defined in the Rule to a view function. I recommend breaking up the application into multiple modules (Larger Applications) for that and adding a separate module for the forms. **kwargs) if ctx is None: ctx = {} elif not isinstance(ctx.routing import Rule app = Flask(__name__) app.html' ctx = f(*args.endpoint \ . render_template def templated(template=None): def decorator(f): @wraps(f) def decorated_function(*args. For example: from flask import Flask from werkzeug. This is possible with this decorator.

'POST']) def register(): form = RegistrationForm(request. methods=['GET'.add(user) flash('Thanks for registering') return redirect(url_for('login')) return render_template('register. False otherwise. You can get it from PyPI.validate(): user = User(form. 145 . message='Passwords must match') ]) confirm = PasswordField('Repeat Password') accept_tos = BooleanField('I accept the TOS'. [ validators. call the validate() method.13.DataRequired(). the usage of this form looks like this: @app.13. [validators.Getting the most out of WTForms with an Extension The Flask-WTF extension expands on this pattern and adds a few little helpers that make working with forms and Flask more fun. which will return True if the data validates. of course.data.data. validators class RegistrationForm(Form): username = StringField('Username'. BooleanField. validators.DataRequired()]) 20. access form.route('/register'.Length(min=4. [validators. StringField.form) if request. but that’s not a requirement. PasswordField.html'. form=form) Notice we’re implying that the view is using SQLAlchemy here (SQLAlchemy in Flask).password.<NAME>.EqualTo('confirm'. max=25)]) email = StringField('Email Address'. 2.email. to access individual values from the form.username. 20.data) db_session. 3.data. form. to validate the data.2 In the View In the view function. Things to remember: 1.1 The Forms This is an example form for a typical registration page: from wtforms import Form. max=35)]) password = PasswordField('New Password'.method == 'POST' and form.Length(min=6. form. create the form from the request form value if the data is submitted via the HTTP POST method and args if the data is submitted as GET. [validators. Adapt the code as necessary.

errors %} <li>{{ error }}</li> {% endfor %} </ul> {% endif %} </dd> {% endmacro %} This macro accepts a couple of keyword arguments that are forwarded to WTForm’s field function. so we have to tell Jinja2 that this data is already HTML-escaped with the |safe filter.html" import render_field %} <form method=post> <dl> {{ render_field(form. which takes ad- vantage of the _formhelpers.html template with such a macro: {% macro render_field(field) %} <dt>{{ field. head over to the WTForms website. The keyword arguments will be inserted as HTML attributes.13. When you pass the form to the templates. for example. which renders the field for us. Look at the following example template to see how easy this is. 146 . So. we can write a macro that renders a field with label and a list of errors if there are any. To make it even nicer. you can easily render them there.3 Forms in Templates Now to the template side.password) }} {{ render_field(form.html template for the function we used above. Note that WTForms returns standard Python unicode strings. class='username') to add a class to the input element.confirm) }} {{ render_field(form.html template: {% from "_formhelpers. Here is the register.username.label }} <dd>{{ field(**kwargs)|safe }} {% if field.email) }} {{ render_field(form. WTForms does half the form generation for us already.20.username) }} {{ render_field(form. Here’s an example _formhelpers. you can call render_field(form.errors %} <ul class=errors> {% for error in field.accept_tos) }} </dl> <p><input type=submit value=Register> </form> For more information about WTForms.

All the block tag does is tell the template engine that a child template may override those portions of the template.html.My Webpage</title> {% endblock %} </head> <body> <div id="content">{% block content %}{% endblock %}</div> <div id="footer"> {% block footer %} &copy. 20. {% endblock %} </div> </body> </html> In this example. which we’ll call layout. Sounds complicated but is very basic.html" %} {% block title %}Index{% endblock %} {% block head %} {{ super() }} <style type="text/css"> . It’s the job of “child” tem- plates to fill the empty blocks with content: <!doctype html> <html> <head> {% block head %} <link rel="stylesheet" href="{{ url_for('static'.14 Template Inheritance The most powerful part of Jinja is template inheritance. Copyright 2010 by <a href="http://domain. It’s easiest to understand it by starting with an example.14.20. defines a simple HTML skeleton docu- ment that you might use for a simple two-column page. filename='style. Template inheritance allows you to build a base “skeleton” template that contains all the common elements of your site and defines blocks that child templates can override. } </style> 147 . the {% block %} tags define four blocks that child templates can fill in. 20.1 Base Template This template.14.css') }}"> <title>{% block title %}{% endblock %} .important { color: #336699.2 Child Template A child template might look like this: {% extends "layout.invalid/">you</a>.

form['username'] != 'admin' or \ request. Note that browsers and sometimes web servers enforce a limit on cookie sizes. Flask provides a really simple way to give feedback to a user with the flashing system.form['password'] != 'secret': error = 'Invalid credentials' else: flash('You were successfully logged in') 148 . It tells the template engine that this template “extends” another template.15 Message Flashing Good applications and user interfaces are all about feedback.route('/login'. use {{ super() }}. render_template. flash. \ request. url_for app = Flask(__name__) app.route('/') def index(): return render_template('index. first it locates the parent. To render the contents of a block defined in the parent template.{% endblock %} {% block content %} <h1>Index</h1> <p class="important"> Welcome on my awesome homepage.html') @app. redirect. This is usually combined with a layout template that does this.15. When the template system evaluates this template. This means that flashing messages that are too large for session cookies causes message flashing to fail silently.1 Simple Flashing So here is a full example: from flask import Flask. 20. methods=['GET'.method == 'POST': if request. {% endblock %} The {% extends %} tag is the key here. The extends tag must be the first tag in the template. The flashing system basically makes it possible to record a message at the end of a request and access it next request and only next request. If the user does not get enough feedback they will probably end up hating the application. 20. 'POST']) def login(): error = None if request.secret_key = 'some_secret' @app.

html: {% extends "layout.html" %} {% block body %} <h1>Overview</h1> <p>Do you want to <a href="{{ url_for('login') }}">log in?</a> {% endblock %} And here is the login.html template which inherits from layout. error=error) And here is the layout.html template which does the magic: <!doctype html> <title>My Application</title> {% with messages = get_flashed_messages() %} {% if messages %} <ul class=flashes> {% for message in messages %} <li>{{ message }}</li> {% endfor %} </ul> {% endif %} {% endwith %} {% block body %}{% endblock %} Here is the index.15.html: {% extends "layout. 149 .form.3. return redirect(url_for('index')) return render_template('login.username }}"> <dt>Password: <dd><input type=password name=password> </dl> <p><input type=submit value=Login> </form> {% endblock %} 20.2 Flashing With Categories New in version 0.html" %} {% block body %} <h1>Login</h1> {% if error %} <p class=error><strong>Error:</strong> {{ error }} {% endif %} <form method=post> <dl> <dt>Username: <dd><input type=text name=username value="{{ request.html template which also inherits from layout.html'.

It is also possible to provide categories when flashing a message. 20. message in messages %} <li class="{{ category }}">{{ message }}</li> {% endfor %} </ul> {% endif %} {% endwith %} This is just one example of how to render these flashed messages. The loop looks slightly different in that situation then: {% with messages = get_flashed_messages(with_categories=true) %} {% if messages %} <ul class=flashes> {% for category.3 Filtering Flash Messages New in version 0. The default cate- gory if nothing is provided is 'message'. Alternative categories can be used to give the user better feedback. One might also use the category to add a prefix such as <strong>Error:</strong> to the message.9. To flash a message with a different category. 'error') Inside the template you then have to tell the get_flashed_messages() function to also return the categories.15.for msg in errors %} <li>{{ msg }}</li> {% endfor -%} </ul> </div> {% endif %} {% endwith %} 150 . This is useful if you wish to render each category in a sepa- rate block. {% with errors = get_flashed_messages(category_filter=["error"]) %} {% if errors %} <div class="alert-message block-message error"> <a class="close" href="#">×</a> <ul> {%. Optionally you can pass a list of categories which filters the results of get_flashed_messages(). For example error messages could be displayed with a red background. just use the second argument to the flash() function: flash(u'Invalid password provided'.

But what if you later decide to move your application to a different location? For example to http://example.1 Loading jQuery In order to use jQuery.googleapis.com/ajax/libs/jquery/1.20.js') }}">\x3C/script>')</script> In this case you have to put jQuery into your static folder as a fallback. It became popular a few years ago and quickly replaced XML as trans- port format in web applications.write('<script src="{{ url_for('static'. filename='jquery.9. It is the perfect tool to make web applications more dynamic by exchanging JSON between server and client.com/myapp? On the server side this never was a problem because we were using the handy url_for() function that could answer that question for us.jQuery || document. but it will first try to load it directly from Google. dicts and lists) look like which is widely supported and very easy to parse. JSON itself is a very lightweight transport format. This has the advantage that your website will probably load faster for users if they went to at least one other website before using the same jQuery version from Google because it will already be in the browser cache.16 AJAX with jQuery jQuery is a small JavaScript library commonly used to simplify working with the DOM and JavaScript in general. you have to download it first and place it in the static folder of your application and then ensure it’s loaded.16. strings. so how can we do that? A simple method would be to add a script tag to our page that sets a global variable to the prefix to the root of the application.script_root|tojson|safe }}. filename='jquery.16.1/jquery. but if we are using jQuery we should not hardcode the path to the application but make that dynamic.js') }}"></script> Another method is using Google’s AJAX Libraries API to load jQuery: <script src="//ajax. very similar to how Python primi- tives (numbers. 20. Ideally you have a layout template that is used for all pages where you just have to add a script statement to the bottom of your <body> to load jQuery: <script type=text/javascript src="{{ url_for('static'.2 Where is My Site? Do you know where your application is? If you are developing the answer is quite simple: it’s on localhost port something and directly on the root of that server. 20. Something like this: <script type=text/javascript> $SCRIPT_ROOT = {{ request.js"></script> <script>window. </script> 151 .min.

jsonify. type=int) return jsonify(result=a + b) @app. This is especially handy for code that is triggered by a script (APIs.10 it goes a step further and escapes all HTML tags with unicode escapes.args. This tem- plate will load jQuery as above and have a little form we can add two numbers and a link to trigger the function on the server side.10 so that Jinja does not escape the JSON encoded string with HTML rules.The |safe is necessary in Flask before 0. Everything until </script> is handled as script.) because you don’t need special error reporting in that case. but a simple example that shows how you would use jQuery and Flask nonetheless: from flask import Flask. 0.route('/') def index(): return render_template('index. 152 . Note that we are using the get() method here which will never fail. If the key is missing a default value (here 0) is returned. Information for Pros In HTML the script tag is declared CDATA which means that entities will not be parsed. request app = Flask(__name__) @app. type=int) b = request. |tojson is kind enough to do the right thing here and escape slashes for you ({{ "</script>"|tojson|safe }} is rendered as "<\/ script>").get('a'. In Flask 0. This makes it possible for Flask to automatically mark the result as HTML safe.get('b'.3 JSON View Functions Now let’s create a server side function that accepts two URL arguments of numbers which should be added together and then sent back to the application in a JSON object.16. This is a really ridiculous example and is something you usually would do on the client side alone.route('/_add_numbers') def add_numbers(): a = request. 0. 20. but we are inside a script block here where different rules apply. This also means that there must never be any </ between the script tags.args. Usually this would be necessary. Furthermore it can convert values to a specific type (like in our case int).html') As you can see I also added an index method here that renders a template. render_template. JavaScript etc.

val() }. navigate to the # URL). $(function() { . { a: $('input[name="a"]').. return false.bind('click'. }).val(). or do that on the top. Here’s the HTML code needed for our little application (index. }) specifies code that should run once the browser is done loading the basic parts of the page. }).text(data. $. 2.. just a very quick explanation of the little bit of code above: 1.html template with jQuery loaded and the $SCRIPT_ROOT variable set. function(data) { $("#result"). 3.result). data. It is usually a better idea to have that in a separate script file: <script type=text/javascript> $(function() { $('a#calculate'). 4. Notice that we also drop the script directly into the HTML here. If you don’t get the whole picture. }).html). download the sourcecode for this example from GitHub.16.getJSON($SCRIPT_ROOT + '/_add_numbers'. function() { $. 153 . b: $('input[name="b"]').20. $('selector') selects an element and lets you operate on it. func) specifies a function that should run when the user clicked on the element. it will call the given function with the return value as argument.html template either has to extend a layout.bind('event'. If that function returns false.4 The HTML Your index. Note that we can use the $SCRIPT_ROOT variable here that we set earlier. func) sends a GET request to url and will send the con- tents of the data object as query parameters. the default behavior will not kick in (in this case. element. </script> <h1>jQuery Example</h1> <p><input type=text size=5 name=a> + <input type=text size=5 name=b> = <span id=result>?</span> <p><a href=# id=calculate>calculate server side</a> I won’t go into detail here about how jQuery works. Once the data arrived.getJSON(url.

If you are not deleting documents permanently from the database but just mark them as deleted. you will have to send a 403 code for disallowed resources. at least a link back to the index. even if the application behaves correctly: 404 Not Found The good old “chap.17. so make sure to also provide the HTTP status code when returning a response. Depending on the error code it is less or more likely for the user to actually see such an error. Keep in mind that Flask will not set the error code for you. It will also provide a plain black and white error page for you with a basic description. Flask will not trigger it if it’s running in Debug mode. 500 Internal Server Error Usually happens on programming errors or if the server is overloaded. So common that even novices to the internet know that 404 means: damn. just like a view function. Here an example implementation for a “404 Page Not Found” exception: 154 . A terribly good idea is to have a nice page there. but it is called when an error happens and is passed that error. Please note that if you add an error handler for “500 Internal Server Error”. the thing I was looking for is not there. 403 Forbidden If you have some kind of access control on your website. It’s a very good idea to make sure there is actually something useful on a 404 page.17. So make sure the user is not lost when they try to access a forbidden resource.17 Custom Error Pages Flask comes with a handy abort() function that aborts a request with an HTTP error code early. but in one case it can be a different error: a handler for internal server errors will be passed other exception instances as well if they are uncaught.2 Error Handlers An error handler is a function. 20. do the user a favour and use the 410 code instead and display a message that what they were looking for was deleted for all eternity. because your application will fail sooner or later (see also: Application Errors). An error handler is registered with the errorhandler() decorator and the error code of the exception. you made a mistake typing that URL” message. but nothing fancy. 20. The error is most likely a HTTPException. 410 Gone Did you know that there the “404 Not Found” has a brother named “410 Gone”? Few people actually implement that.20. but the idea is that resources that previously existed and got deleted answer with 410 instead of 404.1 Common Error Codes The following error codes are some that are often displayed to the user.

html" %} {% block title %}Page Not Found{% endblock %} {% block body %} <h1>Page Not Found</h1> <p>What you were looking for is just not there.from flask import render_template @app. The system that enables having a central URL map is the add_url_rule() function. Instead of using decorators.route('/') def index(): pass @app.1 Converting to Centralized URL Map Imagine the current application looks somewhat like this: from flask import Flask app = Flask(__name__) @app. 20.18.18 Lazily Loading Views Flask is usually used with the decorators. 404 An example template might be this: {% extends "layout.errorhandler(404) def page_not_found(e): return render_template('404. Decorators are simple and you have the URL right next to the function that is called for that specific URL.html'). So if you suddenly notice that your application outgrows this approach you can fall back to a centralized URL mapping. you have a file that sets up the application with all URLs. It might have to do that on systems like Google’s App Engine or other systems. <p><a href="{{ url_for('index') }}">go somewhere nice</a> {% endblock %} 20. However there is a downside to this approach: it means all your code that uses decorators has to be imported upfront or Flask will never actually find your function. This can be a problem if your application has to import quick.route('/user/<username>') def user(username): pass 155 .

index')) 156 . self. *args. view_func=LazyView('yourapplication. This is used by Flask internally to figure out how to name the URL rules in case you don’t provide a name for the rule yourself.import_name = import_name @cached_property def view(self): return import_string(self.Then. py) but without any decorator: def index(): pass def user(username): pass And then a file that sets up an application which maps the functions to URLs: from flask import Flask from yourapplication import views app = Flask(__name__) app.__module__.add_url_rule('/'.__name__ = import_name.add_url_rule('/user/<username>'.index) app. cached_property class LazyView(object): def __init__(self.helpers import LazyView app = Flask(__name__) app. Then you can define your central place to combine the views like this: from flask import Flask from yourapplication. view_func=views.views. The trick is to actually load the view function as needed.rsplit('. view_func=views. This can be accom- plished with a helper class that behaves just like a function but internally imports the real function on first use: from werkzeug import import_string.18. with the centralized approach you would have one file with the views (views. import_name): self.user) 20. but the module is still loaded up- front.view(*args.import_name) def __call__(self. **kwargs): return self.add_url_rule('/'. 1) self.'.2 Loading Late So far we only split up the views and the routing. **kwargs) What’s important here is is that __module__ and __name__ are properly set.

url_rules) One thing to keep in mind is that before and after request handlers have to be in a file that is imported upfront to work properly on the first request. url_rules=[]. and by wrapping view_func in a LazyView as needed.py module for your application: from flask import Flask from mongokit import Connection.19.19 MongoKit in Flask Using a document database rather than a full DBMS gets more common these days.views. **options): view = LazyView('yourapplication. Document # configuration MONGODB_HOST = 'localhost' MONGODB_PORT = 27017 # create the little application object 157 .user')) You can further optimize this in terms of amount of keystrokes needed to write this by having a function that calls into add_url_rule() by prefixing a string with the project name and a dot. view_func=LazyView('yourapplication.user'. 20. to integrate with MongoDB.add_url_rule('/user/<username>'.'/user/<username>'] url('views.app. I will outline each of them here: 20.' + import_name) for url_rule in url_rules: app.add_url_rule(url_rule.1 Declarative The default behavior of MongoKit is the declarative one that is based on common ideas from Django or the SQLAlchemy declarative extension. The same goes for any kind of remaining decorator. a document mapper library. This pattern requires a running MongoDB server and the MongoKit library installed.index'. view_func=view. **options) # add a single route to the index view url('views. There are two very common ways to use MongoKit. def url(import_name. Here an example app. This pattern shows how to use MongoKit. ['/']) # add two routes to a single function endpoint url_rules = ['/user/'.

config['MONGODB_PORT']) To define your models.config['MONGODB_HOST']. 158 . just subclass the Document class that is imported from Mon- goKit. app.config. . This sometimes makes it more to type but also makes it blazingly fast. e. MongoKit does not have something like a session.register([User]) This example shows you how to define your schema (named structure). 'email': max_length(120) } use_dot_notation = True def __repr__(self): return '<User %r>' % (self. MongoKit is just schemaless too. If you’ve seen the SQLAlchemy pattern you may wonder why we do not have a session and even do not define a init_db function here.py. Per default MongoKit behaves like a python dictionary but with use_dot_notation set to True you can use your documents like you use models in nearly any other ORM by using dots to separate between attributes.→format(length)) return validate class User(Document): structure = { 'name': unicode.): from mongokit import ValidationError def max_length(length): def validate(value): if len(value) <= length: return True # must have %s in error format string to have mongokit place key in there raise ValidationError('%s must be at most {} characters long'. On the other hand. 'email': unicode. Here is an example document (put this also into app. a valida- tor for the maximum character length and uses a special MongoKit feature called use_dot_notation. This means you can modify the data structure from one insert query to the next without any prob- lem.app = Flask(__name__) app.g. On the one hand.from_object(__name__) # connect to the database connection = Connection(app.name) # register the User document with our current connection connection. MongoDB is schemaless. } validators = { 'name': max_length(50). but implements some validation to ensure data integrity.

u'email': u . you can do that with MongoKit as well.database import connection >>> from yourapplication.models import User >>> collection = connection['test'].find_one({'name': u'admin'}) <User u'admin'> 20. 'email': u'admin@localhost'} >>> collection.19.→'admin@localhost'}] >>> collection. see the above MongoKit code for examples: from MongoKit import Connection connection = Connection() To insert data you can use the insert method. To query your database.insert(user) MongoKit will automatically commit for us. We have to get a collection first.save() Note that MongoKit is kinda strict with used column types. you use the collection directly: >>> list(collection.2 PyMongo Compatibility Layer If you just want to use PyMongo.User() >>> user['name'] = u'admin' >>> user['email'] = u'admin@localhost' >>> user. Querying is simple as well: >>> list(collection. You may use this process if you need the best performance to get. u'email': u . you must not use a com- mon str type for either name or email but unicode.find()) [{u'_id': ObjectId('4c271729e13823182f000000'). u'name': u'admin'. Note that this example does not show how to couple it with Flask.User. u'name': u'admin'.users >>> user = collection.find_one({'name': u'admin'}) {u'_id': ObjectId('4c271729e13823182f000000').users >>> user = {'name': u'admin'. >>> collection = connection['test'].You can insert entries into the database like this: >>> from yourapplication.User.→'admin@localhost'} These results are also dict-like objects: 159 . this is somewhat the same as a table in the SQL world.find()) [<User u'admin'>] >>> collection.

160 .path.root_path. It should be 16 × 16 pixels and in the ICO file format.20 Adding a favicon A “favicon” is an icon used by browsers for tabs and bookmarks. This is not a requirement but a de-facto standard supported by all relevant browsers. refer to the web server’s documentation.ico. 20. at the website root. This helps to distin- guish your website and to give it a unique brand.ico')) If you want to save the extra redirect request you can also write a view using send_from_directory(): import os from flask import send_from_directory @app. to get browsers to find your icon. 'static').ico'. A common question is how to add a favicon to a Flask application. If your application is not mounted at the root path of the domain you either need to configure the web server to serve the icon at the root or if you can’t do that you’re out of luck.microsoft.ico') }}"> That’s all you need for most browsers. you need an icon.find_one({'name': u'admin'}) >>> r['email'] u'admin@localhost' For more information about MongoKit.join(app. but we may as well specify it to avoid the extra guessing. the correct way is to add a link tag in your HTML.ico') def favicon(): return send_from_directory(os. First. The old de-facto standard is to serve this file. redirect_to=url_for('static'. So. as it will always be the same. If however your application is the root you can simply route a redirect: app. for example: <link rel="shortcut icon" href="{{ url_for('static'.ico'.route('/favicon. Now. 'favicon. with this name. The above will serve the icon via your application and if possible it’s better to config- ure your dedicated web server to serve it. however some really old ones do not support this standard.add_url_rule('/favicon. Put the icon in your static directory as favicon.icon') We can leave out the explicit mimetype and it will be guessed. of course. filename='favicon. mimetype='image/vnd. filename='favicon. head over to the website.>>> r = collection.

Note though that some WSGI middlewares might break streaming. 20.20.'. mimetype='text/csv') Each yield expression is directly sent to the browser. so be careful there in debug environments with profilers and other things you might have enabled.join(row) + '\n' return Response(generate(). **context): app.21. much more than you want to keep in memory.enable_buffering(5) return rv 161 .20.1 Basic Usage This is a basic view function that generates a lot of CSV data on the fly.csv') def generate_large_csv(): def generate(): for row in iter_all_rows(): yield '.update_template_context(context) t = app. This functionality is not directly exposed by Flask because it is quite uncommon.get_template(template_name) rv = t.21. The trick is to have an inner function that uses a generator to generate data and to then invoke that function and pass it to a response object: from flask import Response @app.2 Streaming from Templates The Jinja2 template engine also supports rendering templates piece by piece.route('/large.21 Streaming Contents Sometimes you want to send an enormous amount of data to the client. 20.jinja_env. but you can easily do it yourself: from flask import Response def stream_template(template_name. When you are generating the data on the fly though.stream(context) rv.1 See also • The Favicon article on Wikipedia 20. how do you send that back to the client without the roundtrip to the filesystem? The answer is by using generators and direct responses.

html') def render_large_template(): rows = iter_all_rows() return Response(stream_template('the_template. Flask 0.9. For instance you can try to move that logic into an after-request callback instead. there is no response object yet.route('/stream') def streamed_response(): def generate(): yield 'Hello ' yield request.3 Streaming with Context New in version 0. 20.args['name'] yield '!' return Response(stream_with_context(generate())) Without the stream_with_context() function you would get a RuntimeError at that point.22 Deferred Request Callbacks One of the design principles of Flask is that response objects are created and passed down a chain of potential callbacks that can modify them or replace them. Very often that is possible. Response @app.enable_buffering(size). Since we’re bypassing the Flask template render functions and using the template object itself we have to make sure to update the render context ourselves by calling update_template_context().9 provides you with a helper that can keep the request context around during the execution of the generator: from flask import stream_with_context.html'.route('/my-large-page.@app. It is created as necessary either by a view function or by some other component in the system.21. request. rows=rows)) The trick here is to get the template object from the Jinja2 environment on the appli- cation and to call stream() instead of render() which returns a stream object instead of a string. When the request handling starts. Since each time you do a yield the server will flush the content to the client you might want to buffer up a few items in the template which you can do with rv. Note that when you stream data. 20. The template is then evaluated as the stream is iterated over. the request context is already gone the moment the function executes. One way is to avoid the situation. Sometimes however moving 162 . 5 is a sane default. But what happens if you want to modify the response at a point where the response does not exist yet? A common example for that would be a before-request function that wants to set a cookie on the response object.

For example you can remember the current language of the user in a cookie in the before-request function: from flask import request @app.get('user_lang') if language is None: language = guess_language_from_request() @after_this_request def remember_language(response): 163 .22.2 Calling the Deferred Now you can use the after_this_request decorator to mark a function to be called at the end of the request. we can register a function to be called at the end of the request.that code there is just not a very pleasant experience or makes code look very awk- ward. This way you can defer code execution from anywhere in the application. It registers a function on a list on the g object: from flask import g def after_this_request(f): if not hasattr(g. As an alternative possibility you can attach a bunch of callback functions to the g object and call them at the end of the request. ()): callback(response) return response 20.after_request_callbacks = [] g. For this the following function needs to be registered as after_request() callback: @app. But we still need to call them.22.after_request def call_after_request_callbacks(response): for callback in getattr(g.cookies. 'after_request_callbacks'.3 A Practical Example At any time during a request. 20.after_request_callbacks.22.before_request def detect_user_language(): language = request.1 The Decorator The following decorator is the key. 'after_request_callbacks'): g.append(f) return f 20.

'HEAD'. 'PATCH'.set_cookie('user_lang'. 'replace') environ['REQUEST_METHOD'] = method if method in self. 'POST'. 'DELETE']) def __init__(self.23 Adding HTTP Method Overrides Some HTTP proxies do not support arbitrary HTTP methods or newer HTTP methods (such as PATCH). '').upper() if method in self. The way this works is by letting the client do an HTTP POST request and set the X-HTTP-Method-Override header and set the value to the intended HTTP method (such as PATCH).encode('ascii'. 'OPTIONS' ]) bodyless_methods = frozenset(['GET'.wsgi_app = HTTPMethodOverrideMiddleware(app. This can easily be accomplished with an HTTP middleware: class HTTPMethodOverrideMiddleware(object): allowed_methods = frozenset([ 'GET'. 'HEAD'.allowed_methods: method = method.bodyless_methods: environ['CONTENT_LENGTH'] = '0' return self. start_response): method = environ. environ. language) g. In that case it’s possible to “proxy” HTTP methods through another HTTP method in total violation of the protocol.get('HTTP_X_HTTP_METHOD_OVERRIDE'.app(environ.app = app def __call__(self. 'OPTIONS'. 'PUT'. app): self. start_response) To use this with Flask this is all that is necessary: from flask import Flask app = Flask(__name__) app.wsgi_app) 164 . response. 'DELETE'.language = language 20.

route('/special-api'. For instance JSON data ends up on the request object already read and processed._hash. This seems inconvenient when you want to calculate the checksum of the incoming request data.hexdigest() return 'Hash was: %s' % checksum 165 . before_request_handlers for instance should be careful not to access it)._hash = hashlib.readline(size_hint) self.input']) env['wsgi.environ stream = ChecksumCalcStream(env['wsgi.20. (Eg: be careful accessing request._hash To use this.read(bytes) self.24 Request Content Checksums Various pieces of code can consume the request data and preprocess it. size_hint): rv = self._stream._stream = stream self. Fortunately this is however very simple to change by wrapping the input stream.update(rv) return rv def readline(self. The following example calculates the SHA1 checksum of the incoming data as it gets read and stores it in the WSGI environment: import hashlib class ChecksumCalcStream(object): def __init__(self. This is necessary sometimes for some APIs. methods=['POST']) def special_api(): hash = generate_checksum(request) # Accessing this parses the input stream files = request.form or anything of that nature. bytes): rv = self._hash.files # At this point the hash is fully constructed. Example usage: @app.update(rv) return rv def generate_checksum(request): env = request. stream): self._stream. form data ends up there as well but goes through a different code path.input'] = stream return stream.sha1() def read(self. checksum = hash. all you need to do is to hook the calculating stream in before the request starts consuming data.

20.config['CELERY_RESULT_BACKEND'].25. For instance you can place this in a tasks module. It used to have a Flask inte- gration but it became unnecessary after some restructuring of the internals of Celery with Version 3.Task = ContextTask return celery The function creates a new Celery object. configures it with the broker from the ap- plication config. like creating tasks and managing workers. **kwargs): with app.1 Installing Celery Celery is on the Python Package Index (PyPI). backend=app. *args.config['CELERY_BROKER_URL']) celery. While you can use Celery without any reconfiguration with Flask. it must be possible for other modules to import it. 166 . this is called the celery application.config) TaskBase = celery. just for Celery. It serves the same purpose as the Flask object in Flask. it becomes a bit nicer by subclassing tasks and adding support for Flask’s application contexts and hooking it up with the Flask configura- tion. updates the rest of the Celery config from the Flask config and then creates a subclass of the task that wraps the task execution in an application context. 20. This guide fills in the blanks in how to properly use Celery with Flask but assumes that you generally already read the First Steps with Celery guide in the official Celery documentation.conf. Since this instance is used as the entry-point for everything you want to do in Celery.25 Celery Based Background Tasks Celery is a task queue for Python with batteries included.Task class ContextTask(TaskBase): abstract = True def __call__(self. This is all that is necessary to properly integrate Celery with Flask: from celery import Celery def make_celery(app): celery = Celery(app.update(app. **kwargs) celery. so it can be installed with standard Python tools like pip or easy_install: $ pip install celery 20. broker=app.app_context(): return TaskBase.import_name. *args.__call__(self.25.2 Configuring Celery The first thing you need is a Celery instance.

Request from werkzeug.20.update( CELERY_BROKER_URL='redis://localhost:6379'.wait() will never actually return. For example.config. That’s because you also need to run celery.26 Subclassing Flask The Flask class is designed for subclassing.25.celery worker The your_application string has to point to your application’s package or module that creates the celery object.wait() 65 20.25. b): return a + b This task can now be called in the background: >>> result = add_together.4 Running the Celery Worker Now if you jumped in and already executed the above code you will be disappointed to learn that your .datastructures import ImmutableOrderedMultiDict class MyRequest(Request): """Request subclass to override request parameter storage""" parameter_storage_class = ImmutableOrderedMultiDict class MyFlask(Flask): 167 . CELERY_RESULT_BACKEND='redis://localhost:6379' ) celery = make_celery(flask_app) @celery. 42) >>> result. you may want to override how request parameters are handled to pre- serve their order: from flask import Flask.3 Minimal Example With what we have above this is the minimal example of using Celery with Flask: from flask import Flask flask_app = Flask(__name__) flask_app. You can do that by running celery as a worker: $ celery -A your_application.task() def add_together(a.delay(23. 20.

"""Flask subclass using the custom request class""" request_class = MyRequest This is the recommended approach for overriding or augmenting Flask’s internal func- tionality. 168 .

consider using mod_wsgi.run() calls you might have in your applica- tion file are inside an if __name__ == '__main__': block or moved to a separate file.1 Hosted options • Deploying Flask on Heroku • Deploying Flask on OpenShift • Deploying Flask on Webfaction • Deploying Flask on Google App Engine • Deploying Flask on AWS Elastic Beanstalk • Sharing your Localhost Server with Localtunnel • Deploying on Azure (IIS) • Deploying on PythonAnywhere 21. Watch Out Please make sure in advance that any app.2.2 Self-hosted options 21. Some of the options available for properly running Flask in production are documented here. 21. Just remember that your Flask application object is the actual WSGI application. 169 .1 mod_wsgi (Apache) If you are using the Apache webserver. If you want to deploy your Flask application to a WSGI server not listed here. Flask’s built-in server is not suitable for produc- tion as it doesn’t scale well and by default serves only one request at a time. look up the server documentation about how to use a WSGI app with it. CHAPTER TWENTYONE DEPLOYMENT OPTIONS While lightweight and easy to use.

etc.wsgi file. If you encounter segfaulting child processes after the first apache reload you can safely ignore them. Just restart the server. This file contains the code mod_wsgi is executing on startup to get the application object. OpenSUSE..wsgi file before the import: 170 .: /var/www/yourapplication) and make sure that yourapplication and all the libraries that are in use are on the python load path. For most applications the following file should be sufficient: from yourapplication import app as application If you don’t have a factory function for application creation but a singleton instance you can directly import that one as application. Keep in mind that you will have to actually install your application into the virtualenv as well.) you can install it as follows: # yum install mod_wsgi On FreeBSD install mod_wsgi by compiling the www/mod_wsgi port or by using pkg_add: # pkg install ap22-mod_wsgi2 If you are using pkgsrc you can install mod_wsgi by compiling the www/ap2-wsgi pack- age. If you don’t want to install it system wide consider using a virtual python instance.Just make sure it’s not called because this will always start a local WSGI server which we do not want if we deploy that application to mod_wsgi. Store that file somewhere that you will find it again (e. The object called application in that file is then used as application. The mod_wsgi installation instructions cover source installations on UNIX systems. If you are using Ubuntu/Debian you can apt-get it and activate it as follows: # apt-get install libapache2-mod-wsgi If you are using a yum based distribution (Fedora.g. Installing mod_wsgi If you don’t have mod_wsgi installed yet you have to either install it using a package manager or compile it yourself. Alternatively there is the option to just patch the path in the . Creating a .wsgi file To run your application you need a yourapplication.

wsgi <Directory C:\yourdir> Order deny.wsgi <Directory /var/www/yourapplication> WSGIProcessGroup yourapplication WSGIApplicationGroup %{GLOBAL} Order deny.2 Order allow. the syntax for directory permissions has changed from httpd 2.deny Allow from all to httpd 2.4 syntax Require all granted For more information consult the mod_wsgi documentation.path.insert(0.allow Allow from all </Directory> </VirtualHost> Note: WSGIDaemonProcess isn’t implemented in Windows and Apache will refuse to run with the above configuration.allow Allow from all </Directory> </VirtualHost> Note: There have been some changes in access control configuration for Apache 2.com WSGIDaemonProcess yourapplication user=user1 group=group1 threads=5 WSGIScriptAlias / /var/www/yourapplication/yourapplication.import sys sys. '/path/to/the/application') Configuring Apache The last thing you have to do is to create an Apache configuration file for your ap- plication. In this example we are telling mod_wsgi to execute the application under a different user for security reasons: <VirtualHost *> ServerName example. 171 . On a Windows system.com WSGIScriptAlias / C:\yourdir\yourapp.4. eliminate those lines: <VirtualHost *> ServerName example. Most notably.

just add the following directive to your Directory section: WSGIScriptReloading On Working with Virtual Environments Virtual environments have the advantage that they never install the required depen- dencies system wide so you have a better control over what is used where.stdout = sys.run() call in your application file that is not guarded by an if __name__ == '__main__': condition. errorlog shows SystemExit ignored You have an app. 172 . Problem: application gives permission errors Probably caused by your application running as the wrong user. Either remove that run() call from the file and move it into a separate run.wsgi file slightly. Make sure the folders the application needs access to have the proper privileges set and the application runs as the correct user (user and group parameter to the WSGIDaemonProcess directive) Problem: application dies with an error on print Keep in mind that mod_wsgi dis- allows doing anything with sys. If you want to use a virtual environment with mod_wsgi you have to modify your . The reason for this is that for non-installed packages. For that.stdout and sys. Please be aware that this does not work. or convert your application into a package.py file you symlinked into the site-packages folder.stderr.stderr Problem: accessing resources gives IO errors Your application probably is a single . mod_wsgi will reload all the daemon processes for us. follow this guide to troubleshoot: Problem: application does not run. You can disable this protection from the config by setting the WSGIRestrictStdout to off: WSGIRestrictStdout Off Alternatively you can also replace the standard out in the .py file or put it into such an if block. Support for Automatic Reloading To help deployment tools you can activate support for automatic reloading. instead you either have to put the folder into the pythonpath the file is stored in.Troubleshooting If your application does not run. the module filename is used to locate the resources and for symlinks the wrong filename is picked up.wsgi file with a differ- ent stream: import sys sys. Whenever something changes the .wsgi file.

read().0.0. 21. dict(__file__=activate_this)) This sets up the load paths according to the settings of the virtual environment.0. Keep in mind that the path has to be absolute.2.wsgi import WSGIServer from yourapplication import app http_server = WSGIServer((''. It’s a pre-fork worker model ported from Ruby’s Unicorn project.wsgi file: activate_this = '/path/to/env/bin/activate_this.wsgi file: activate_this = '/path/to/env/bin/activate_this.Add the following lines to the top of your . you can proxy to them from your web server.0.serve_forever() 173 . These servers stand alone when they run. dict(__file__=activate_this)) For Python 3 add the following lines to the top of your . 5000). Note the section on Proxy Setups if you run into issues. It supports both eventlet and greenlet.1:4000 myproject:app Gevent Gevent is a coroutine-based Python networking library that uses greenlet to provide a high-level synchronous API on top of libev event loop: from gevent.2 Standalone WSGI Containers There are popular servers written in Python that contain WSGI applications and serve HTTP. app) http_server. Running a Flask application on this server is quite simple: gunicorn myproject:app Gunicorn provides many command-line options – see gunicorn -h.1:4000): gunicorn -w 4 -b 127.py' execfile(activate_this. For example.py' with open(activate_this) as file_: exec(file_. Gunicorn Gunicorn ‘Green Unicorn’ is a WSGI HTTP Server for UNIX. to run a Flask application with 4 worker processes (-w 4) binding to localhost port 4000 (-b 127.

non-blocking event- driven networking library. proxy_redirect off. the most common setup invokes the host being set from X-Forwarded-Host and the remote address from X-Forwarded-For: 174 . see twistd -h and twistd web -h for more information. proxy_set_header Host $host.log. location / { proxy_pass http://127. and the twistd utility does as well. on port 8080. Werkzeug ships a fixer that will solve some common setups. } } If your httpd is not providing these headers.0.log. Here’s a simple nginx configuration which proxies to an application served on local- host at port 8000. with an application from myproject: twistd -n web --port 8080 --wsgi myproject. server_name _. but you might want to write your own WSGI middleware for specific setups.app Proxy Setups If you deploy your application using one of these servers behind an HTTP proxy you will need to rewrite a few headers in order for the application to work. proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for. setting appropriate headers: server { listen 80. Twisted Web comes with a standard WSGI container which can be controlled from the command line using the twistd utility: twistd web --wsgi myproject. proxy_set_header X-Real-IP $remote_addr. You can configure your httpd to pass these headers. proxy_set_header X-Forwarded-Proto $scheme. to run a Twisted Web server in the foreground. or you can fix them in middleware.Twisted Web Twisted Web is the web server shipped with Twisted. Twisted Web supports many flags and options.0. a mature. access_log /var/log/nginx/access. The two prob- lematic values in the WSGI environment usually are REMOTE_ADDR and HTTP_HOST. For example.app This example will run a Flask application called app from a module named myproject.1:8000/. error_log /var/log/nginx/error.

fixers import ProxyFix app. start_response): host = environ.2.run() calls you might have in your applica- tion file are inside an if __name__ == '__main__': block or moved to a separate file.3 uWSGI uWSGI is a deployment option on servers like nginx. If you want to rewrite the headers from another header.contrib. 175 . lighttpd. app): self. To use your WSGI appli- cation with uWSGI protocol you will need a uWSGI server first. Make sure to have it installed to follow along. FastCGI.wsgi_app = ProxyFix(app. The most popular uWSGI server is uwsgi.app = app def __call__(self.from werkzeug. you might want to use a fixer like this: class CustomProxyFix(object): def __init__(self. environ. see FastCGI and Standalone WSGI Containers for other options. Just make sure it’s not called because this will always start a local WSGI server which we do not want if we deploy that application to uWSGI.wsgi_app) 21. start_response) app.app(environ.wsgi_app) Trusting Headers Please keep in mind that it is a security issue to use such a middleware in a non-proxy setup because it will blindly trust the incoming headers which might be forged by malicious clients. the application server can serve uWSGI. and HTTP protocols.wsgi_app = CustomProxyFix(app. which we will use for this guide. and cherokee. uWSGI is both a protocol and an application server. Watch Out Please make sure in advance that any app.get('HTTP_X_FHOST'. '') if host: environ['HTTP_HOST'] = host return self.

Starting your app with uwsgi uwsgi is designed to operate on WSGI callables found in python modules.4 FastCGI FastCGI is a deployment option on servers like nginx. If you want to deploy your flask application inside of a virtual environment. use the following command: $ uwsgi -s /tmp/yourapplication. } location @yourapplication { include uwsgi_params. If you want to have it in the URL root its a bit simpler: location / { try_files $uri @yourapplication. Configuring nginx A basic flask nginx configuration looks like this: location = /yourapplication { rewrite ^ /yourapplication/. you can use a single / instead of /yourapplication. see uWSGI and Standalone WSGI Containers for other options. If your application is accessi- ble at root level.sock --manage-script-name --mount / . uwsgi_pass unix:/tmp/yourapplication. To use your WSGI applica- tion with any of them you will need a FastCGI server first. myapp refers to the name of the file of your flask application (without extension) or the module which provides app. app is the callable inside of your application (usually the line reads app = Flask(__name__). and cherokee. you need to also add --virtualenv /path/to/virtual/environment. } location @yourapplication { include uwsgi_params. lighttpd. } location /yourapplication { try_files $uri @yourapplication. since its smarter about that. Given a flask application in myapp.sock. Make sure to have it installed to follow along. } This configuration binds the application to /yourapplication. It is used together with the --mount directive which will make requests to /yourapplication be directed to myapp:app.→yourapplication=myapp:app The --manage-script-name will move the handling of SCRIPT_NAME to uwsgi.2. } 21. You might also need to add --plugin python or --plugin python3 depending on which python version you use for your project. uwsgi_pass unix:/tmp/yourapplication.sock. The most popular one is flup which we will use for this guide.py. 176 .

It makes sense to have that in /var/www/yourapplication or something similar. however nginx and older versions of lighttpd need a socket to be explicitly passed to communicate with the FastCGI server.fcgi -idle-timeout 300 -processes 5 177 .so FastCgiServer /var/www/html/yourapplication/app. For that to work you need to pass the path to the socket to the WSGIServer: WSGIServer(application.fcgi Configuring Apache The example above is good enough for a basic Apache deployment but your . A preferable way is to use the ScriptAlias and SetHandler config- uration directives to route requests to the FastCGI server.sock').run() calls you might have in your applica- tion file are inside an if __name__ == '__main__': block or moved to a separate file.fcgi import WSGIServer from yourapplication import app if __name__ == '__main__': WSGIServer(app). bindAddress='/path/to/fcgi.run() This is enough for Apache to work.fcgi file First you need to create the FastCGI server file. Save the yourapplication.Watch Out Please make sure in advance that any app.g.fcgi/news/. Creating a .fcgi does not appear in the URL. There are few ways to configure your application so that yourapplication. example.run() The path has to be the exact same path you define in the server config.server. Make sure to set the executable bit on that file so that the servers can execute it: # chmod +x /var/www/yourapplication/yourapplication.com/yourapplication. The following example uses FastCgiServer to start 5 instances of the application which will handle all incoming requests: LoadModule fastcgi_module /usr/lib64/httpd/modules/mod_fastcgi.fcgi file somewhere you will find it again.fcgi file will appear in your application URL e. Just make sure it’s not called because this will always start a local WSGI server which we do not want if we deploy that application to FastCGI. Let’s call it yourapplication.fcgi: #!/usr/bin/python from flup.

com DocumentRoot /var/www/html/yourapplication AddHandler fastcgi-script fcgi ScriptAlias / /var/www/html/yourapplication/app.1:3000 If you cannot set ScriptAlias.fcgi from the URLs.L] </IfModule> Set yourapplication. you can use WSGI middleware to remove yourapplication. Set .mydomain. app): 178 .fcgi import WSGIServer from yourapplication import app class ScriptNameStripper(object): def __init__(self.path.c> Options +FollowSymlinks RewriteEngine On RewriteBase / RewriteCond %{REQUEST_FILENAME} !-f RewriteRule ^(.server. it’s simply used as an identifier to other directives such as AliasMatch: FastCgiServer /var/www/html/yourapplication -host 127. you can use the FastCgiExternalServer directive instead.fcgi/$1 [QSA.0. Note that in the fol- lowing the path is not real.<VirtualHost *> ServerName webapp1.*)$ yourapplication.insert(0.c> AddHandler fcgid-script . If you’re using a standalone FastCGI server.fcgi <Files ~ (\.htaccess: <IfModule mod_fcgid. '<your_local_path>/lib/python2.6/site-packages') from flup.fcgi)> SetHandler fcgid-script Options +FollowSymLinks +ExecCGI </Files> </IfModule> <IfModule mod_rewrite.0.fcgi/ <Location /> SetHandler fastcgi-script </Location> </VirtualHost> These processes will be managed by Apache. for example on a shared web host.fcgi: #!/usr/bin/python #: optional path to your local python site-packages folder import sys sys.

Also.fcgi" => (( "socket" => "/tmp/yourapplication-fcgi. A basic Flask FastCGI configuration for nginx looks like this: 179 . start_response): environ['SCRIPT_NAME'] = '' return self. environ. Make sure to apply it only if you are mounting the application the URL root.app(environ.app = app def __call__(self.*))$" => "$1".sock".*)$" => "/yourapplication. This configuration binds the application to /yourapplication. see the Lighty docs for more information on FastCGI and Python (note that explicitly passing a socket to run() is no longer necessary). Configuring nginx Installing FastCGI applications on nginx is a bit different because by default no FastCGI parameters are forwarded. self. start_response) app = ScriptNameStripper(app) if __name__ == '__main__': WSGIServer(app). "max-procs" => 1 )) ) alias.fcgi". "bin-path" => "/var/www/yourapplication/yourapplication.url = ( "/static/" => "/path/to/your/static" ) url. If you want the application to work in the URL root you have to work around a lighttpd bug with the LighttpdCGIRootFix middle- ware.server = ("/yourapplication.run() Configuring lighttpd A basic FastCGI configuration for lighttpd looks like that: fastcgi. "^(/. alias and rewrite modules.fcgi$1" ) Remember to enable the FastCGI.rewrite-once = ( "^(/static($|/. "check-local" => "disable".

fastcgi_param SCRIPT_NAME "". and note that this is a manual solution which does not persist across system restart: $ screen $ /var/www/yourapplication/yourapplication.g. fastcgi_pass unix:/tmp/yourapplication-fcgi. You can look around for other FastCGI process managers or write a script to run your . fastcgi_pass unix:/tmp/yourapplication-fcgi. e. Super- visor can manage FastCGI processes. Very often the only thing the server log tells you is something along the lines of “premature end of headers”. using a SysV init. fastcgi_param SCRIPT_NAME $fastcgi_script_name. } location @yourapplication { include fastcgi_params. you can always run the . fastcgi_split_path_info ^(/yourapplication)(. fastcgi_param PATH_INFO $fastcgi_script_name.*)$. fastcgi_param PATH_INFO $fastcgi_path_info. This example assumes your application is called application.fcgi script inside GNU screen.location = /yourapplication { rewrite ^ /yourapplication/ last. } Running FastCGI Processes Since nginx and others do not load FastCGI apps.fcgi Debugging FastCGI deployments tend to be hard to debug on most web servers. If you want to have it in the URL root it’s a bit simpler because you don’t have to figure out how to calculate PATH_INFO and SCRIPT_NAME: location / { try_files $uri @yourapplication. See man screen for details.fcgi file at boot. } This configuration binds the application to /yourapplication.d script. } location @yourapplication { include fastcgi_params.sock. In order to debug the application the only thing that can really give you ideas why it breaks is switching to the correct user and executing the application by hand. For a temporary solution.sock.fcgi and that your web server user is www-data: $ su www-data $ cd /var/www/yourapplication 180 . } location /yourapplication { try_files $uri @yourapplication. you have to do it by yourself.

CGI is sup- ported by all major servers but usually has a sub-optimal performance.run() calls you might have in your applica- tion file are inside an if __name__ == '__main__': block or moved to a separate file. With CGI. Watch Out Please make sure in advance that any app. 21. CGI will work for sure.cgi into a cgi-bin (and use mod_rewrite or something similar to rewrite the URL) or let the 181 .$ python application. Creating a . Let’s call it yourapplication.stdout is overridden by something that doesn’t write into the HTTP response.fcgi". Just make sure it’s not called because this will always start a local WSGI server which we do not want if we deploy that application to CGI / app engine. line 4.cgi file First you need to create the CGI application file. Either just copy the . where execution happens in a CGI-like environment. • Different python interpreters being used. or that sys. Common problems are: • Relative paths being used. Don’t rely on the current working directory.run(app) Server Setup Usually there are two ways to configure the server.handlers import CGIHandler from yourapplication import app CGIHandler().cgi: #!/usr/bin/python from wsgiref. • The code depending on environment variables that are not set by the web server.fcgi Traceback (most recent call last): File "yourapplication. you will also have to make sure that your code does not contain any print statements. in <module> ImportError: No module named yourapplication In this case the error seems to be “yourapplication” not being on the python path. This is also the way you can use a Flask application on Google’s App Engine.2.5 CGI If all other deployment methods do not work.

server point to the file directly.*)$ /path/to/the/application.htaccess. works too but the ScriptAlias directive won’t work in that case: RewriteEngine On RewriteCond %{REQUEST_FILENAME} !-f # Don't interfere with static files RewriteRule ^(. though. sitting in the public directory you want your app to be available.cgi/$1 [L] For more information consult the documentation of your webserver. 182 . In this case. In Apache for example you can put something like this into the config: ScriptAlias /app /path/to/the/application. a file called . you might not have access to your Apache config.cgi On shared webhosting.

it became useful to a wide audience. Flask started in part to demonstrate how to build your own framework on top of existing well-used tools Werkzeug (WSGI) and Jinja (templating). Extend. Dig deeper on the APIs you use. and look for the customizations which are available out of the box in a Flask release. Flask’s code is written to be read. and documents its internal utilities so that you can find the hook points needed for your project. and look for patterns to build your own extensions if you do not find the tools you need. Look for ways in which your project can be refactored into a collection of utilities and Flask extensions. and as it developed. hook points. 22. and Signals. You can quickly add or customize behavior by subclassing Flask (see the linked method docs) and using that subclass wherever you instantiate an application class. Read the source. its documentation is published so you can use its internal APIs. You can provide custom classes for things like the request and response objects. As you grow your codebase. CHAPTER TWENTYTWO BECOMING BIG Here are your options when growing your codebase or scaling your application.2 Hook. See Subclassing Flask for an example. Explore the many extensions in the community. 22. Flask sticks to documented APIs in upstream libraries. 183 . 22.3 Subclass. don’t just use Flask – understand it.1 Read the Source. This works well with Applica- tion Factories. The Flask class has many methods designed for subclassing. The API docs are full of available overrides.

Werkzeug includes several middlewares. However the majority of servers are using either threads. For many web applications the complexity of the code is less an issue than the scaling for the number of users or data entries expected. Many people use patched and modified versions of their framework to counter shortcomings. the framework would be a very complex system to begin with which causes a steep learning curve and a lot of user frustration. This is not unique to Flask. forking should be the very last resort. This is natural because if this would not be the case. fork Flask. You don’t have to contribute any changes back if you decide to modify the framework. Be- cause of that. Scaling well means for example that if you double the amount of servers you get about twice the performance. the data store you want to use and the Python implementation and webserver you are running on.6 Scale like a pro.22.5 Fork. Flask by itself is only limited in terms of scaling by your application code. These libraries do the majority of the work. There is only one limiting factor regarding scaling in Flask which are the context local proxies. 184 .4 Wrap with middleware. 22. The Application Dispatching chapter shows in detail how to apply middleware. The majority of code of Flask is within Werkzeug and Jinja2. For every project there is the point where the underlying framework gets in the way (due to assumptions the original developers had). The downside of forking is of course that Flask extensions will most likely break be- cause the new framework has a different import name. If your server uses some kind of concurrency that is not based on threads or greenlets. 22. If none of the above options work. Flask will no longer be able to support these global proxies. They depend on context which in Flask is defined as being either a thread. Scaling bad means that if you add a new server the application won’t perform any better or would not even support a second server. Furthermore integrating up- stream changes can be a complex process. Flask is just the paste that glues those together. greenlets or separate pro- cesses to achieve concurrency which are all methods well supported by the underlying Werkzeug library. process or greenlet. You can introduce WSGI middleware to wrap your Flask instances and introduce fixes and changes at the layer between your Flask application and your HTTP server. This idea is also reflected in the license of Flask. depending on the number of changes.

don’t hesitate to contact the developers on the mailinglist or IRC channel. The Flask developers keep the framework accessible to users with codebases big and small.22. caused by Flask. If you find an obstacle in your way.7 Discuss with the community. The best way for the Flask and Flask extension developers to improve the tools for larger applications is getting feedback from users. 185 .

186 .

Part II

If you are looking for information on a specific function, class or method, this part of
the documentation is for you.






This part of the documentation covers all the interfaces of Flask. For parts where Flask
depends on external libraries, we document the most important right here and provide
links to the canonical documentation.

23.1 Application Object

class flask.Flask(import_name, static_path=None, static_url_path=None,
static_folder=’static’, template_folder=’templates’, in-
stance_path=None, instance_relative_config=False,
The flask object implements a WSGI application and acts as the central object.
It is passed the name of the module or package of the application. Once it is
created it will act as a central registry for the view functions, the URL rules,
template configuration and much more.
The name of the package is used to resolve resources from inside the package
or the folder the module is contained in depending on if the package parameter
resolves to an actual python package (a folder with an __init__.py file inside)
or a standard module (just a .py file).
For more information about resource loading, see open_resource().
Usually you create a Flask instance in your main module or in the __init__.py
file of your package like this:

from flask import Flask
app = Flask(__name__)

About the First Parameter
The idea of the first parameter is to give Flask an idea of what belongs to your
application. This name is used to find resources on the filesystem, can be used
by extensions to improve debugging information and a lot more.
So it’s important what you provide there. If you are using a single module,
__name__ is always the correct value. If you however are using a package, it’s


usually recommended to hardcode the name of your package there.
For example if your application is defined in yourapplication/app.py you
should create it with one of the two versions below:

app = Flask('yourapplication')
app = Flask(__name__.split('.')[0])

Why is that? The application will work even with __name__, thanks to how re-
sources are looked up. However it will make debugging more painful. Certain
extensions can make assumptions based on the import name of your application.
For example the Flask-SQLAlchemy extension will look for the code in your ap-
plication that triggered an SQL query in debug mode. If the import name is not
properly set up, that debugging information is lost. (For example it would only
pick up SQL queries in yourapplication.app and not yourapplication.views.frontend)

New in version 0.7: The static_url_path, static_folder, and template_folder parame-
ters were added.
New in version 0.8: The instance_path and instance_relative_config parameters
were added.
New in version 0.11: The root_path parameter was added.
• import_name – the name of the application package
• static_url_path – can be used to specify a different path
for the static files on the web. Defaults to the name of the
static_folder folder.
• static_folder – the folder with static files that should be
served at static_url_path. Defaults to the 'static' folder in the
root path of the application.
• template_folder – the folder that contains the templates that
should be used by the application. Defaults to 'templates'
folder in the root path of the application.
• instance_path – An alternative instance path for the applica-
tion. By default the folder 'instance' next to the package or
module is assumed to be the instance path.
• instance_relative_config – if set to True relative filenames
for loading the config are assumed to be relative to the instance
path instead of the application root.
• root_path – Flask by default will automatically calculate the
path to the root of the application. In certain situations this
cannot be achieved (for instance if the package is a Python 3
namespace package) and needs to be manually defined.


add_template_filter(f, name=None)
Register a custom template filter. Works exactly like the template_filter()
Parameters name – the optional name of the filter, otherwise the
function name will be used.
add_template_global(f, name=None)
Register a custom template global function. Works exactly like the
template_global() decorator.
New in version 0.10.
Parameters name – the optional name of the global function, other-
wise the function name will be used.
add_template_test(f, name=None)
Register a custom template test. Works exactly like the template_test()
New in version 0.10.
Parameters name – the optional name of the test, otherwise the
function name will be used.
add_url_rule(rule, endpoint=None, view_func=None, **options)
Connects a URL rule. Works exactly like the route() decorator. If a
view_func is provided it will be registered with the endpoint.
Basically this example:

def index():

Is equivalent to the following:

def index():
app.add_url_rule('/', 'index', index)

If the view_func is not provided you will need to connect the endpoint to a
view function like so:

app.view_functions['index'] = index

Internally route() invokes add_url_rule() so if you want to customize the
behavior via subclassing you only need to change this method.
For more information refer to URL Route Registrations.
Changed in version 0.2: view_func parameter added.
Changed in version 0.6: OPTIONS is added automatically as method.


• rule – the URL rule as string
• endpoint – the endpoint for the registered URL rule. Flask
itself assumes the name of the view function as endpoint
• view_func – the function to call when serving a request to the
provided endpoint
• options – the options to be forwarded to the underlying Rule
object. A change to Werkzeug is handling of method options.
methods is a list of methods this rule should be limited to
(GET, POST etc.). By default a rule just listens for GET (and
implicitly HEAD). Starting with Flask 0.6, OPTIONS is implicitly
added and handled by the standard request handling.
after_request(f )
Register a function to be run after each request.
Your function must take one parameter, an instance of response_class and
return a new response object or the same (see process_response()).
As of Flask 0.7 this function might not be executed at the end of the request
in case an unhandled exception occurred.
after_request_funcs = None
A dictionary with lists of functions that should be called after each request.
The key of the dictionary is the name of the blueprint this function is active
for, None for all requests. This can for example be used to close database
connections. To register a function here, use the after_request() decorator.
Binds the application only. For as long as the application is bound to the
current context the flask.current_app points to that application. An appli-
cation context is automatically created when a request context is pushed if
Example usage:

with app.app_context():

New in version 0.9.
The class that is used for the g instance.
Example use cases for a custom class:
1.Store arbitrary attributes on flask.g.
2.Add a property for lazy per-request database connectors.
3.Return None instead of AttributeError on unexpected attributes.
4.Raise exception if an unexpected attr is set, a “controlled” flask.g.


The function will be called without any arguments. To register a function here. Blueprints can be at- tached multiple times so this dictionary does not tell you how often they got attached.g object is now applica- tion context scoped. In Flask 0. blueprints = None all the attached blueprints in a dictionary by name. The function will be called without any arguments and its return value is ignored. None for all requests. New in version 0. before_first_request_funcs = None A lists of functions that should be called at the beginning of the first request to this instance. This can for example be used to open database connections or getting hold of the currently logged in user. New in version 0.10 to app_ctx_globals_class because the flask.8. Commands registered 193 . use the before_request() decorator. it’s handled as if it was the return value from the view and further request handling is stopped. If the function returns a non-None value. New in version 0. It will basically calculate the path to a folder named instance next to your main file or the package. cli = None The click command line context for this application.9 this property was called request_globals_class but it was changed in 0. To register a function here.8. alias of _AppCtxGlobals auto_find_instance_path() Tries to locate the instance path if it was not provided to the constructor of the application class.7.8. use the before_first_request() decorator. before_request_funcs = None A dictionary with lists of functions that should be called at the beginning of the request. The key of the dictionary is the name of the blueprint this function is active for. New in version 0. New in version 0.10. before_first_request(f ) Registers a function to be run before the first request to this instance of the application. before_request(f ) Registers a function to run before each request.

New in version 0. New in version 0. create_url_adapter(request) Creates a URL adapter for the given request. The URL adapter is created at a point where the request context is not yet set up so the request is passed explicitly.5. alias of Config context_processor(f ) Registers a template context processor function. Changed in version 0. Example use cases for a custom class: 1. Since 0.11: Environment. This behaves exactly like a regular dictionary but supports additional methods to load a config from files.9: This can now also be called without a request object when the URL adapter is created for the application context.6. config_class The class that is used for the config attribute of this app. This is an instance of a click. Instead one should override the jinja_loader() function instead. 194 . The global loader dispatches between the loaders of the application and the individual blueprints. New in version 0. Can be used to override just the loader and keeping the rest unchanged. create_global_jinja_loader() Creates the loader for the Jinja2 environment. Override this function to customize the behavior.Default values for certain config options. Defaults to Config.7 this also adds the Jinja2 globals and filters after initialization.Access to config values through attributes in addition to keys. create_jinja_environment() Creates the Jinja2 environment based on jinja_options and select_jinja_autoescape().7. The default commands are provided by Flask itself and can be overridden.11. It’s discouraged to override this function. here show up in the flask command once the application has been dis- covered. 2.Group object. Changed in version 0.auto_reload set in accordance with TEMPLATES_AUTO_RELOAD configuration option. New in version 0. config = None The configuration dictionary as Config.

endpoint('example. default_config = ImmutableDict({‘JSON_AS_ASCII’: True.7: This no longer does the exception handling. Changed in version 0.9. The key is None for error handlers active on the application. This does not have to be a response object. This is not actually called by the Flask object itself but is always triggered when the request context is popped. The special key None points to a list of tuples 195 . Defaults to False. This attribute can also be configured from the config with the DEBUG config- uration key. This works pretty much the same as do_teardown_request() but for the application context. ‘USE_X_SENDFILE’: False. Matches the URL and returns the return value of the view or error handler.9: Added the exc argument. New in version 0. do_teardown_request(exc=<object object>) Called after the actual request dispatching and will call every as teardown_request() decorated function. endpoint(endpoint) A decorator to register a function as an endpoint. Example: @app. this code was moved to the new full_dispatch_request(). otherwise the key is the name of the blueprint. In order to convert the return value to a proper response object. call make_response(). ‘SE Default configuration parameters. Set this to True to enable debugging of the application.endpoint') def example(): return "example" Parameters endpoint – the name of the endpoint error_handler_spec = None A dictionary of all registered error handlers. In debug mode the debugger will kick in when an unhandled exception occurs and the integrated server will automatically reload the application if changes in the code are detected. That way we have a tighter control over certain resources under testing environments. dispatch_request() Does the request dispatching. Each key points to another dictionary where the key is the status code of the http exception.debug The debug flag. Changed in version 0. Previously this was al- ways using the current exception information. do_teardown_appcontext(exc=<object object>) Called when an application context is popped.

or an arbitrary exception extensions = None a place where extensions can store application specific state.7: Use register_error_handler() instead of modifying error_handler_spec directly.7: One can now additionally also register custom excep- tion types that do not necessarily have to be a subclass of the HTTPException class.errorhandler(404) def page_not_found(error): return 'This page does not exist'. errorhandler(code_or_exception) A decorator that is used to register a function given an error code. If the error handler should be application wide None shall be used. 500 You can also register a function as error handler without using the errorhandler() decorator. The following example is equivalent to the one above: def page_not_found(error): return 'This page does not exist'. use the errorhandler() decorator.error_handler_spec[None][404] = page_not_found Setting error handlers via assignments to error_handler_spec however is discouraged as it requires fiddling with nested dictionaries and the special case for arbitrary exception types. For exam- ple this is where an extension could store database engines and similar things.extensions['extensionname'] = SomeObject() 196 . New in version 0. To register a error handler. The first None refers to the active blueprint. 404 You can also register handlers for arbitrary exceptions: @app. 'extensions'): app. For backwards compatibility extensions should register themselves like this: if not hasattr(app. Example: @app. for application wide error handlers. Parameters code_or_exception – the code as integer for the han- dler.extensions = {} app. New in version 0. 404 app. where the first item is the class for the instance check and the second the error handler function.errorhandler(DatabaseError) def special_exception_handler(error): return 'Database connection failed'.

Flask. name) New in version 0. this function returns SEND_FILE_MAX_AGE_DEFAULT from the con- figuration of current_app. endpoint. to set the cache timeout for .7.9. name): if name. For example. got_first_request This attribute is set to True if the application started handling the first re- quest. In debug mode the exception will be re-raised immediately. full_dispatch_request() Dispatches the request and on top of that performs request pre and post- processing as well as HTTP exception catching and error handling. this method is called. the key would be 'foo'. By default. For example in case of a “Flask-Foo” extension in flask_foo. values) Handle BuildError on url_for(). If no such handler exists. The key must match the name of the extension module.lower(). New in version 0. 197 .Flask): def get_send_file_max_age(self.js files to 60 seconds: class MyFlask(flask. If a cache_timeout is given in send_file().3.get_send_file_max_age(self. This allows subclasses to change the behavior when sending files based on the filename. and send_file() calls this function on current_app when the given cache_timeout is None. otherwise. New in version 0. oth- erwise it is logged and the handler for a 500 internal server error is used.js'): return 60 return flask. handle_http_exception(e) Handles an HTTP exception. that time- out is used.3. get_send_file_max_age(filename) Provides default cache_timeout for the send_file() functions.7.8. handle_url_build_error(error. New in version 0. Static file functions such as send_from_directory() use this function. handle_exception(e) Default exception handling that kicks in when an exception occurs that is not caught.endswith('. New in version 0. a default 500 internal server error message is dis- played. New in version 0. By default this will invoke the registered error handlers and fall back to returning the exception as response.

11.7. A special case are HTTPExceptions which are forwarded by this function to the handle_http_exception() method. jinja_options = ImmutableDict({‘extensions’: [’jinja2.7. Changed in version 0.with_’]}) Options that are passed directly to the Jinja2 environment. handle_user_exception(e) This method is called whenever an exception occurs that should be handled. New in version 0.8. Used to initialize the Jinja2 globals.7.11. 198 . New in version 0.5. New in version 0. This is used internally and automatically called on URL building. init_jinja_globals() Deprecated.ext. New in version 0.autoescape’.7: This method is deprecated with 0. Override create_jinja_environment() instead. alias of Environment jinja_loader The Jinja loader for this package bound object. jinja_environment The class that is used for the Jinja environment. New in version 0.5. jinja_env The Jinja2 environment used to load templates.5. New in version 0. inject_url_defaults(endpoint. ‘jinja2. values) Injects the URL defaults for the given endpoint directly into the values dic- tionary passed.ext. instance_path = None Holds the path to the instance folder. This function will either return a response value or reraise the exception with the same traceback. New in version 0. iter_blueprints() Iterates over all blueprints by the order they were registered. New in version 0. has_static_folder This is True if the package bound object’s container has a folder for static files.

New in version 0. New in version 0. logger_name The name of the logger to use.logger. New in version 0.debug('A value for debugging') app. This is called by handle_exception() if debugging is disabled and right before the handler is called.10. By default the logger name is the package name passed to the constructor. 42) app. Here some examples: app. The default implementation logs the exception as error on the logger. make_config(instance_relative=False) Used to create the config attribute by the Flask constructor. make_default_options_response() This method is called to create the default OPTIONS response.7.3.Logger object for this application.4.10. make_null_session() Creates a new instance of a missing session. 199 .logger. alias of JSONEncoder log_exception(exc_info) Logs an exception. This logger can be used to (surprise) log messages.8. Instead of overriding this method we recommend replacing the session_interface. New in version 0.error('An error occurred') New in version 0. The default configuration is to log to stderr if the application is in debug mode. alias of JSONDecoder json_encoder The JSON encoder class to use. New in version 0.json_decoder The JSON decoder class to use.logger. This can be changed through subclassing to change the default behavior of OPTIONS re- sponses. New in version 0. Defaults to JSONEncoder.warning('A warning occurred (%d apples)'.8. logger A logging. The in- stance_relative parameter is passed in from the constructor of Flask (there named instance_relative_config) and indicates if the config should be relative to the instance path or the root path of the application. Defaults to JSONDecoder.

It can be set and overridden to change the value. To access resources within subfolders use forward slashes as separator. • mode – resource file opening mode. This name is used as a display name when Flask needs the name of the applica- tion. status.7. headers) or (response. consider the following folder structure: 200 . name The name of the application. This runs all the registered shell context processors. mode=’rb’) Opens a resource from the application’s resource folder. Otherwise works like open_resource(). The following types are allowed for rv: response_class the object is returned unchanged str a response object is created with the string as body unicode a response object is created with the string encoded to utf-8 as body a WSGI function the function is called as WSGI application and buffered as response object tuple A tuple in the form (response. Parameters • resource – the name of the resource. make_response(rv) Converts the return value from a view function to a real response object that is an instance of response_class.11. To see how this works. open_resource(resource. This is usually the import name with the dif- ference that it’s guessed from the run file if the import name is main. mode=’rb’) Opens a resource from the application’s instance folder (instance_path). Instance resources can also be opened for writing.9: Previously a tuple was interpreted as the arguments for the response object.8. default is ‘rb’. New in version 0. Parameters rv – the return value from the view function Changed in version 0. New in version 0. status is a string or an inte- ger and headers is a list or a dictionary with header values. make_shell_context() Returns the shell context for an interactive shell for this application. open_instance_resource(resource. headers) where response is any of the types defined here. New in version 0.

html /index. /myapplication.open_resource('schema.html If you want to open the schema.sql file you would do the following: with app. New in version 0. 201 . This also triggers the url_value_preprocessor() functions before the actual before_request() functions are called. passing no arguments. Instead of overriding this method we recommend replacing the session_interface. default is ‘rb’. This attribute can also be configured from the config with the PERMANENT_SESSION_LIFETIME configuration key. permanent_session_lifetime A timedelta which is used to set the expiration date of a permanent session. open_session(request) Creates or opens a new session. • mode – resource file opening mode. If any of these functions returns a value. otherwise a sensible default is returned.sql') as f: contents = f.read() do_something_with(contents) Parameters • resource – the name of the resource.py /schema. Default implementation stores all session data in a signed cookie.7.sql /static /style. preserve_context_on_exception Returns the value of the PRESERVE_CONTEXT_ON_EXCEPTION configuration value in case it’s set.css /templates /layout. This requires that the secret_key is set. it’s handled as if it was the return value from the view and further request handling is stopped. The default is 31 days which makes a permanent session survive for roughly one month. Parameters request – an instance of request_class. Defaults to timedelta(days=31) preprocess_request() Called before the actual request dispatching and will call each before_request() decorated function. To access resources within subfolders use forward slashes as separator.

This must be used in combination with the with statement because the request is only bound to the current context for the duration of the with block. New in version 0.7. Parameters response – a response_class object.7. request_class The class that is used for request objects. Returns a new response object or the same. register_blueprint(blueprint. process_response(response) Can be overridden in order to modify the response object before it’s sent to the WSGI server. register_error_handler(code_or_exception. propagate_exceptions Returns the value of the PROPAGATE_EXCEPTIONS configuration value in case it’s set.request_context(environ): do_something_with(request) The object returned can also be used without the with statement which is useful for working in the shell. The example above is doing exactly the same as this code: ctx = app. Changed in version 0. See Request for more information.push() try: do_something_with(request) 202 . Example usage: with app.5: As of Flask 0. **options) Registers a blueprint on the application. By default this will call all the after_request() decorated functions. has to be an instance of response_class.7.5 the functions registered for after request execution are called in reverse order of registration. alias of Request request_context(environ) Creates a RequestContext from the given environment and binds it to the current context. f ) Alternative error attach function to the errorhandler() decorator that is more straightforward to use for non decorator usage. New in version 0. New in version 0.request_context(environ) ctx. otherwise a sensible default is returned.

Do not use run() in a production setting. Parameters • rule – the URL rule as string • endpoint – the endpoint for the registered URL rule. Instead. 203 . If the debug flag is set the server will automatically reload for code changes and show a debugger in case an exception happened. POST etc. but disable the code ex- ecution on the interactive debugger. See Response for more informa- tion.pop() Changed in version 0.). Flask itself assumes the name of the view function as endpoint • options – the options to be forwarded to the underlying Rule object.6. methods is a list of methods this rule should be limited to (GET. By default a rule just listens for GET (and implicitly HEAD).3: Added support for non-with statement usage and with statement is now passed the ctx object. alias of Response route(rule. **options) A decorator that is used to register a view function for a given URL rule. OPTIONS is implicitly added and handled by the standard request handling. **options) Runs the application on a local development server. Parameters environ – a WSGI environment response_class The class that is used for response objects. see Deploy- ment Options for WSGI server recommendations. If you want to run the application in debug mode. This does the same thing as add_url_rule() but is intended for decorator usage: @app. finally: ctx. This will keep the debugger’s traceback screen active. run(host=None. you can pass use_evalex=False as pa- rameter. port=None. but disable code execution. It is not intended to meet security and performance requirements for a production server. debug=None. Starting with Flask 0. A change to Werkzeug is handling of method options.route('/') def index(): return 'Hello World' For more information refer to URL Route Registrations.

select_jinja_autoescape(filename) Returns True if autoescaping should be active for the given template name.0' to have the server available externally as well. Defaults to '127. Set this to '0. As such to enable just the interactive debugger with- out the code reloading. Instead you should be using the flask command line script’s run support.5. Instead of overriding this method we recommend replac- ing the session_interface. • debug – if given.0.0. enable or disable debug mode. save_session(session. This attribute can also be configured from the config with the SECRET_KEY configuration key. If no template name is given. It is not recommended to use this function for development with automatic reloading as this is badly supported. you have to invoke run() with debug=True and use_reloader=False. Setting use_debugger to True without being in debug mode won’t catch any exceptions because there won’t be any to catch. Defaults to 5000 or the port defined in the SERVER_NAME config variable if present. returns True. • options – the options to be forwarded to the underlying Werkzeug server.serving.10: The default port is now picked from the SERVER_NAME variable. For the default implementation. • port – the port of the webserver. cryptographic components can use this to sign cookies and other things. 204 . Keep in Mind Flask will suppress any server error with a generic error page unless it is in debug mode.run_simple() for more information. See debug. Changed in version 0.0. Defaults to None. Parameters • session – the session to be saved (a SecureCookie object) • response – an instance of response_class secret_key If a secret key is set. New in version 0. check open_session(). response) Saves the session if it needs updates. 1'. Set this to a complex random value when you want to use the secure cookie for instance. See werkzeug. Parameters • host – the hostname to listen on.0.

If this function returns True then the teardown handlers will not be passed the error. New in version 0.5.11. By default an instance of SecureCookieSessionInterface is used here. New in version 0. shell_context_processor(f ) Registers a shell context processor function. These functions are typically also called when the request context is popped. should_ignore_error(error) This is called to figure out if an error should be ignored or not as far as the teardown system is concerned.app_context() ctx. teardown_appcontext(f ) Registers a function to be called when the application context ends. static_folder The absolute path to the configured static folder.11. This attribute can also be configured from the config with the SESSION_COOKIE_NAME configuration key.send_file_max_age_default A timedelta which is used as default cache_timeout for the send_file() functions. session_cookie_name The secure cookie uses this for the name of the session cookie. New in version 0. New in version 0. Defaults to timedelta(hours=12) send_static_file(filename) Function used internally to send static files from the static folder to the browser. shell_context_processors = None A list of shell context processor functions that should be run when a shell context is created. Example: ctx = app. Defaults to 'session' session_interface = <flask.sessions. This configuration vari- able can also be set with an integer value used as seconds.8. The default is 12 hours. New in version 0.SecureCookieSessionInterface object> the session interface to use.push() 205 .10. This attribute can also be configured from the config with the SEND_FILE_MAX_AGE_DEFAULT configuration key.

Instead it will keep it alive so that the interactive de- 206 . regardless of whether there was an exception or not.test_request_context() ctx. ctx. New in version 0.pop() is executed in the above example. This becomes relevant if you are using such constructs in tests..9. Generally teardown functions must take every necessary step to avoid that they will fail. the teardown functions are called just before the app context moves from the stack of active con- texts. When a teardown function was called because of an exception it will be passed an error object.pop() When ctx. These functions are executed when the request context is popped. If they do execute code that might fail they will have to sur- round the execution of these code by try/except statements and log occur- ring errors. ctx..pop() When ctx. .9. teardown_request(f ) Register a function to be run at the end of each request. Debug Note In debug mode Flask will not tear down a request on an exception immediately. The return values of teardown functions are ignored.. teardown_appcontext_funcs = None A list of functions that are called when the application context is destroyed.pop() is executed in the above example. even if not an actual request was performed. This becomes relevant if you are using such constructs in tests. the teardown functions are called just before the request context moves from the stack of active con- texts. New in version 0.. The return values of teardown functions are ignored. Since a request context typically also manages an application context it would also be called when you pop a request context. When a teardown function was called because of a exception it will be passed an error object. Since the application context is also torn down if the request ends this is the place to store code that disconnects from databases.push() . Example: ctx = app.

Example: @app. To register a function here. You can specify a name for the filter. template_context_processors = None A dictionary with list of functions that are called without argument to pop- ulate the template context. teardown_request_funcs = None A dictionary with lists of functions that are called after each request. Example: 207 . otherwise the function name will be used. Example: @app. You can specify a name for the test.template_filter() def reverse(s): return s[::-1] Parameters name – the optional name of the filter. None for all requests. If an exception occurred while processing the request. Parameters name – the optional name of the global function. even if an exception has occurred. The key of the dictionary is the name of the blueprint this function is active for. and their return values are ig- nored. Each returns a dictionary that the template context is updated with.7. otherwise the function name will be used. To register a function here. it gets passed to each teardown_request function. The key of the dictionary is the name of the blueprint this function is active for. otherwise the function name will be used. other- wise the function name will be used. None for all requests. New in version 0. bugger can still access it. template_test(name=None) A decorator that is used to register custom template test. template_global(name=None) A decorator that is used to register a custom template global function.10. This behavior can be controlled by the PRESERVE_CONTEXT_ON_EXCEPTION configuration variable. use the teardown_request() decorator. use the context_processor() decorator. otherwise the function name will be used. template_filter(name=None) A decorator that is used to register custom template filter. You can specify a name for the global function. These func- tions are not allowed to modify the request.template_global() def double(n): return 2 * n New in version 0.

**kwargs): self. Changed in version 0.test_client_class = CustomClient client = app.testing import FlaskClient class CustomClient(FlaskClient): def __init__(self. For example: from flask. Note that if you are testing for assertions or exceptions in your application code. **kwargs) app..') See FlaskClient for more information.10. This is useful if you want to access the context locals for testing: with app.test_client() The test client can be used in a with block to defer the closing down of the context until the end of the with block. *args.get('/?vodka=42') assert request. int(math.ceil(math..test_client(authentication='Basic . For example: app._authentication = kwargs. you must set app. 208 .testing = True in order for the exceptions to prop- agate to the test client. you may pass optional keyword arguments that will then be passed to the application’s test_client_class constructor.self). Otherwise.template_test() def is_prime(n): if n == 2: return True for i in range(2. Parameters name – the optional name of the test.test_client() as c: rv = c.sqrt(n))) + 1): if n % i == 0: return False return True New in version 0.pop("authentication") super(CustomClient. @app. See the testing attribute.4: added support for with block usage for the client. test_client(use_cookies=True. otherwise the function name will be used. **kwargs) Creates a test client for this application.__init__( *args. For information about unit testing head over to Testing Flask Applications.args['vodka'] == '42' Additionally.. the exception will be handled by the application (not visible to the test client) and the only indication of an As- sertionError or other exception will be a 500 status code response to the test client.testing = True client = app.

New in version 0. **kwargs) Creates a WSGI environment from the given values (see werkzeug. It also returns True if TRAP_HTTP_EXCEPTIONS is set to True. testing The testing flag. For example this might activate unittest helpers that have an additional runtime cost which should not be enabled by default. EnvironBuilder for more information. config and g into the template context as well as every- thing template context processors want to inject. If this is enabled and PROPAGATE_EXCEPTIONS is not changed from the default it’s implicitly enabled. test_client_class = None the test client that is used with when test_client is used. This in- jects request. Set this to True to enable the test mode of Flask extensions (and in the future probably also Flask itself). New in version 0. test_request_context(*args.6.7. the original values in the context will not be overridden if a context processor decides to return a value with the same key.7: The use_cookies parameter was added as well as the ability to override the client to be used by setting the test_client_class attribute. url_build_error_handlers = None A list of functions that are called when url_for() raises a BuildError. Defaults to False. session. This is called for all HTTP exceptions raised by a view function. this function accepts the same argu- ments). New in version 0. Note that the as of Flask 0. This is helpful for debug- ging implicitly raised HTTP exceptions.11: Added **kwargs to support passing additional key- word arguments to the constructor of test_client_class. This attribute can also be configured from the config with the TESTING con- figuration key. update_template_context(context) Update the template context with some commonly used variables. Changed in version 0.8. If it returns True for any exception the error handler for this exception is not called and it shows up as regular exception in the traceback. trap_http_exception(e) Checks if an HTTP exception should be trapped or not. By default this will return False for all exceptions except for a bad request key er- ror if TRAP_BAD_REQUEST_ERRORS is set to True. Each 209 .test. Parameters context – the context as a dictionary that is updated in place to add extra variables.

') def to_url(self. You can use this to change the routing converters after the class was created but before any routes are connected. otherwise the key is the name of the blueprint.join(super(ListConverter.9. url_map = None The Map for this instance.'. It’s called with the endpoint and values and should update the values passed in place.7. endpoint and values. If a function returns None or raises a BuildError the next function is tried. New in version 0.split('. Whenever a URL is built these functions are called to modify 210 . self). Defaults to werkzeug. The key None here is used for application wide callbacks. This is used by add_url_rule().routing import BaseConverter class ListConverter(BaseConverter): def to_python(self. url_defaults(f ) Callback function for URL defaults for all view functions of the applica- tion.converters['list'] = ListConverter url_rule_class The rule object to use for URL rules created. function registered here is called with error. url_value_preprocessors = None A dictionary with lists of functions that can be used as URL value processor functions. New in version 0.url_map. For each function registered this one should also provide a url_defaults() function that adds the parameters automatically again that were removed that way.7. Each of these functions has the chance to modify the dictionary of URL values before they are used as the key- word arguments of the view function.to_url(value) for value in values) app = Flask(__name__) app. value): return value. values): return '. It’s called before the view functions are called and can modify the url values provided.routing.Rule. url_default_functions = None A dictionary with lists of functions that can be used as URL value prepro- cessors. New in version 0. alias of Rule url_value_preprocessor(f ) Registers a function as URL value preprocessor for all view functions of the application. Example: from werkzeug.

view_functions = None A dictionary of all view functions registered. The key None here is used for application wide callbacks. The keys will be function names which are also used to generate URLs and the values are the function objects themselves. otherwise the key is the name of the blueprint. This attribute can also be configured from the config with the USE_X_SENDFILE configuration key. This is not implemented in __call__ so that middlewares can be applied without losing a reference to the class.wsgi_app = MyMiddleware(app. Defaults to False.wsgi_app) Then you still have the original application object around and can continue to call methods on it. use the route() decorator. New in version 0. the dictionary of values in place. Changed in version 0. Keep in mind that the server has to support this. So in- stead of doing this: app = MyMiddleware(app) It’s a better idea to do this instead: app. Parameters • environ – a WSGI environment • start_response – a callable accepting a status code.2. use_x_sendfile Enable this if you want to use the X-Sendfile feature. To register a view function. a list of headers and an optional exception context to start the re- sponse 211 .7. See Callbacks and Errors. wsgi_app(environ. Each of these functions has the chance to modify the dictionary New in version 0. independent on if an error occurred or not.7: The behavior of the before and after request call- backs was changed under error conditions and a new callback was added that will always execute at the end of the request. This only affects files sent with the send_file() method. start_response) The actual WSGI application.

name=None) Register a custom template test. available application wide.23. endpoint=None. even if outside of the blueprint. See Modular Applications with Blueprints for more infor- mation. Like Flask. available application wide.10.2 Blueprint Objects class flask. New in version 0. otherwise the function name will be used. url_prefix=None. otherwise the function name will be used. import_name. Like Flask. add_app_template_test(f. Works exactly like the app_template_filter() decorator. add_app_template_filter(f. name=None) Register a custom template filter. add_template_filter() but for a blueprint.add_template_global() but for a blueprint. New in version 0. after_app_request(f ) Like Flask. Works exactly like the app_template_global() decorator. New in version 0. url_defaults=None. This function is only exe- cuted after each request that is handled by a function of that blueprint. Like Flask.add_url_rule() but for a blueprint. static_url_path=None. A blueprint is an object that records functions that will be called with the BlueprintSetupState later to register functions or other things on the main application. Parameters name – the optional name of the filter. subdomain=None. available application wide. name=None) Register a custom template global. add_app_template_global(f. static_folder=None. after_request(f ) Like Flask. The endpoint for the url_for() function is prefixed with the name of the blueprint. **options) Like Flask. Parameters name – the optional name of the test. Parameters name – the optional name of the global. add_url_rule(rule. Works exactly like the app_template_test() decorator.add_template_test() but for a blueprint. otherwise the function name will be used.after_request() but for a blueprint. Such a function is executed after each request. view_func=None. root_path=None) Represents a blueprint. 212 . template_folder=None.Blueprint(name.after_request() but for a blueprint.7.10.

available application wide. Parameters name – the optional name of the global. Parameters name – the optional name of the filter. 213 . app_url_value_preprocessor(f ) Same as url_value_preprocessor() but application wide. even if outside of the blueprint. Such a function is executed before the first request to the application.app_context_processor(f ) Like Flask. This function is only exe- cuted before each request that is handled by a function of that blueprint.context_processor() but for a blueprint. even if outside of the blueprint.context_processor() but for a blueprint. Parameters name – the optional name of the test.before_request() but for a blueprint. New in version 0. New in version 0. Like Flask. available application wide. template_test() but for a blueprint. template_global() but for a blueprint.errorhandler() but for a blueprint. app_errorhandler(code) Like Flask. otherwise the function name will be used. This handler is used for all requests. before_request(f ) Like Flask.before_first_request(). otherwise the function name will be used. even if outside of a blueprint. otherwise the function name will be used. Like Flask. Like Flask. before_app_request(f ) Like Flask. Such a function is ex- ecuted each request. app_template_test(name=None) Register a custom template test.10. app_template_filter(name=None) Register a custom template filter. This function is only executed for requests handled by a blueprint.before_request().10. Such a function is executed before each re- quest. before_app_first_request(f ) Like Flask. template_filter() but for a blueprint. app_template_global(name=None) Register a custom template global. context_processor(f ) Like Flask. app_url_defaults(f ) Same as url_defaults() but application wide. available application wide.

it will be registered to the current blueprint.Flask): def get_send_file_max_age(self.9. options. jinja_loader The Jinja loader for this package bound object. Static file functions such as send_from_directory() use this function. open_resource(resource. Otherwise works as the errorhandler() decorator of the Flask object. mode=’rb’) Opens a resource from the application’s resource folder. New in version 0. that time- out is used.endswith('. name): if name.Flask. endpoint(endpoint) Like Flask. to set the cache timeout for . otherwise it’s an application independent endpoint.js files to 60 seconds: class MyFlask(flask. For example. Please be aware that routing does not happen local to a blueprint so an error han- dler for 404 usually is not handled by a blueprint unless it is caused inside a view function. has_static_folder This is True if the package bound object’s container has a folder for static files. By default.lower(). first_registration=False) Creates an instance of BlueprintSetupState() object that is later passed to the register callback functions. and send_file() calls this function on current_app when the given cache_timeout is None. If the endpoint is prefixed with a .endpoint() but for a blueprint. this function returns SEND_FILE_MAX_AGE_DEFAULT from the con- figuration of current_app. otherwise. this method is called. errorhandler(code_or_exception) Registers an error handler that becomes active for this blueprint only. Another special case is the 500 internal server error which is always looked up from the application. If a cache_timeout is given in send_file(). get_send_file_max_age(filename) Provides default cache_timeout for the send_file() functions.get_send_file_max_age(self.5. this has to be done explicitly by the user of this method.js'): return 60 return flask. This does not prefix the end- point with the blueprint name. make_setup_state(app. name) New in version 0.5. This allows subclasses to change the behavior when sending files based on the filename. Subclasses can override this to return a sub- class of the setup state. New in version 0. To see how this 214 .

first_registration=False) Called by Flask. This can be overridden to customize the register behavior.css /templates /layout. 215 .html If you want to open the schema. If the blueprint is registered a second time on the application. the function passed is not called. The endpoint for the url_for() function is prefixed with the name of the blueprint. akin to the register_error_handler() application-wide function of the Flask ob- ject but for error handlers limited to this blueprint. • mode – resource file opening mode. register(app. consider the following folder structure: /myapplication. record_once(func) Works like record() but wraps the function in another function that will ensure the function is only called once.sql /static /style.sql file you would do the following: with app. To access resources within subfolders use forward slashes as separator.sql') as f: contents = f. record(func) Registers a function that is called when the blueprint is registered on the application. **options) Like Flask. options. send_static_file(filename) Function used internally to send static files from the static folder to the browser. f ) Non-decorator version of the errorhandler() error attach function. works.py /schema.11. default is ‘rb’. Key- word arguments from register_blueprint() are directly forwarded to this method in the options dictionary.read() do_something_with(contents) Parameters • resource – the name of the resource. register_error_handler(code_or_exception.open_resource('schema.route() but for a blueprint. New in version 0. route(rule.register_blueprint() to register a blueprint on the ap- plication. This function is called with the state as argument as returned by the make_setup_state() method.html /index.

The request object is a Request subclass and provides all of the attributes Werkzeug defines plus a few Flask specific ones. This function is only executed when tearing down requests handled by a function of that blueprint. args A MultiDict with the parsed contents of the query string.3 Incoming Request Data class flask. Such a function is exe- cuted when tearing down each request. teardown_app_request(f ) Like Flask. 23.teardown_request() but for a blueprint. It’s called with the endpoint and values and should update the values passed in place. cookies A dict with the contents of all cookies transmitted with the request.Request(environ. form A MultiDict with the parsed form data from POST or PUT requests. Teardown request functions are executed when the request con- text is popped. (The part in the URL after the question mark).5. static_folder The absolute path to the configured static folder. teardown_request(f ) Like Flask. url_value_preprocessor(f ) Registers a function as URL value preprocessor for this blueprint. If you want to replace the request object used you can subclass this and set request_class to your subclass. Please keep in mind that file uploads will not end up here. but instead in the files attribute. Remembers the matched endpoint and view arguments. even if outside of the blueprint. shallow=False) The request object used by default in Flask. url_defaults(f ) Callback function for URL defaults for this blueprint. populate_request=True. New in version 0. even when no actual request was performed. values A CombinedMultiDict with the contents of both form and args.teardown_request() but for a blueprint. It’s called before the view functions are called and can modify the url values provided. It is what ends up as request. stream If the incoming form data was not encoded with a known mimetype the 216 .

html' url u'http://www. Imagine your application is listening on the following application root: http://www.html?x=y' script_root u'/myapplication' base_url u'http://www.com/myapplication/π/page.com/myapplication/π/page. Each file is stored as FileStorage object.example.com/myapplication/%CF%80/page.example. method The current request method (POST. The stream only returns the data once. headers The incoming request headers as a dictionary like object.com/myapplication/' 217 . environ The underlying WSGI environment.example. with the difference that it also has a save() function that can store the file on the filesystem.) path full_path script_root url base_url url_root Provides different ways to look at the current IRI. data Contains the incoming request data as string in case it came with a mime- type Flask does not handle.example. files A MultiDict with files uploaded as part of a POST or PUT request. data is stored unmodified in this stream for consumption.example. It basically behaves like a standard file object you know from Python.html? x=y' url_root u'http://www.html?x=y In this case the values of the above mentioned attributes would be the fol- lowing: path u'/π/page.com/myapplication And a user requests the following URI: http://www. GET etc.html' full_path u'/π/page. Most of the time it is a better idea to use data which will give you that data as a string.

cache=True) Parses the incoming JSON request data and returns it. on_json_loading_failed(e) Called if decoding of the JSON data failed. If an exception happened when matching. blueprint The name of the current blueprint endpoint The endpoint that matched the request. module The name of the current module if the request was dispatched to an actual module. Parameters • force – if set to True the mimetype is ignored. This only works with libraries that support the X-Requested-With header and set it to XMLHttpRequest. json If the mimetype is applicationhttps://www.scribd.com/json this will contain the parsed JSON data. • cache – if set to True the parsed JSON data is remembered on the request.11. get_json(force=False. If parsing fails the on_json_loading_failed() method on the request object will be invoked. By default a request is considered to include JSON data if the mimetype is applicationhttps://www.scribd.com/json or application/ *+json. New in version 0. use blueprints instead. This is deprecated functionality. Otherwise this will be None. The return value of this method is used by get_json() when an error occurred. This in combination with view_args can be used to reconstruct the same or a modified URL. The get_json() method should be used instead. max_content_length Read-only view of the MAX_CONTENT_LENGTH config key. jQuery and Mochikit and probably some more. By default this function will return None if the mimetype is not applicationhttps://www.scribd.com/json but this can be overridden by the force parameter. 218 . The default implementation just raises a BadRequest exception. this will be None. • silent – if set to True this method will fail silently and return None. is_xhr True if the request was triggered via a JavaScript XMLHttpRequest. silent=False. Libraries that do that are prototype. is_json Indicates if this request is JSON or not.

methods) etc. This is usually a NotFound exception or something similar.8.request To access incoming request data. If an exception hap- pened when matching. headers A Headers object representing the response headers.6. This can be useful to in- spect which methods are allowed for the URL from a before/after handler (request. this will be None. status A string with a response status. this is the exception that will be raised / was raised as part of the request handling. See Notes On Proxies for more information.4 Response Objects class flask. mime- type=None. Quite often you don’t have to create this object yourself because make_response() will take care of that for you. This just shows a quick overview of the most im- portant ones. Works like the response object from Werkzeug but is set to have an HTML mimetype by default. New in version 0. Internally Flask makes sure that you always get the correct data for the active thread if you are in a multithreaded environment. 23. headers=None. url_rule = None The internal URL rule that matched the request. routing_exception = None If matching the URL failed.url_rule. status=None. class flask. If you want that behavior back you can trivially add it by subclassing. you can use the global request object. Flask parses incoming request data for you and gives you access to it through that global object. The request object is an instance of a Request subclass and provides all of the attributes Werkzeug defines. 219 . view_args = None A dict of view arguments that matched the request.10: Removed buggy previous behavior of generating a random JSON response. If you want to replace the response object used you can subclass this and set response_class to your subclass. direct_passthrough=False) The response object that is used by default in Flask. This is a proxy. Changed in version 0. content_type=None.Response(response=None. New in version 0.

This is a proxy.secret_key set you can use sessions in Flask applications. do- main=None. The parameters are the same as in the cookie Morsel object in the Python standard library but it accepts unicode data. So the user can look at the session contents.session The session object works pretty much like an ordinary dict. This should not be used and will eventually get deprecated. too. To access the current session you can use the session object: class flask.com" will set a cookie that is readable by the domain www. expires=None. a cookie will only be readable by the domain that set it. domain=".example. mimetype The mimetype (content type without charset etc. secure=None. See Notes On Proxies for more information. A session basically makes it possible to remember information from one request to an- other. • max_age – should be a number of seconds. 23.) set_cookie(key. • value – the value of the cookie. but not modify it unless they know the secret key. status_code The response status as integer. The way Flask does this is by using a signed cookie. value=’‘. path=’/’.5 Sessions If you have the Flask. • expires – should be a datetime object or UNIX timestamp. httponly=False) Sets a cookie.example. • domain – if you want to set a cross-domain cookie. per default it will span the whole domain. For exam- ple. so make sure to set that to something complex and unguessable. The following attributes are interesting: 220 . foo. Oth- erwise. with the difference that it keeps track on modifications. Parameters • key – the key (name) of the cookie to be set. • path – limits the cookie to a given path.com etc.example. or None (default) if the cookie should last only as long as the client’s browser session.com. max_age=None. data A descriptor that calls get_data() and set_data().

session_interface: app = Flask(__name__) app. in that situation you have to explicitly set the attribute to True yourself.6 Session Interface New in version 0. The session interface provides a simple way to replace the session implementation that Flask is using. The default NullSession class that is created will complain that the secret key was not set. If set to False (which is the default) the session will be deleted when the user closes the browser. modified True if the session object detected a modification.sessions. The session object returned by the open_session() method has to provide a dic- tionary like interface plus the properties and methods from the SessionMixin.session_interface = MySessionInterface() 221 . SessionMixin): pass If open_session() returns None Flask will call into make_null_session() to create a session that acts as replacement if the session support cannot work because some requirement is not fulfilled. class flask.append(42) # so mark it as modified yourself session.modified = True permanent If set to True the session lives for permanent_session_lifetime seconds.8. new True if the session is new. the others have useful defaults which you don’t need to change.SessionInterface The basic interface you have to implement in order to replace the default session interface which uses werkzeug’s securecookie implementation. session['objects']. To replace the session interface on an application all you have to do is to assign flask. 23. Be advised that modifica- tions on mutable structures are not picked up automatically. We recommend just subclassing a dict and adding that mixin: class Session(dict.Flask. False otherwise. The default is 31 days. The only meth- ods you have to implement are open_session() and save_session(). Here an example: # this change is not picked up because a mutable object (here # a list) is changed.

8. Likewise the is_null_session() method will perform a typecheck against this type. The default implementation returns now + the permanent session lifetime configured on the application. The default imple- mentation uses the value from the SESSION_COOKIE_PATH config var if it’s set. is_null_session(obj) Checks if a given object is a null session. get_cookie_secure(app) Returns True if the cookie should be secure. Null sessions are not asked to be saved. get_cookie_domain(app) Helpful helper method that returns the cookie domain that should be used for the session cookie if session cookies are used. get_cookie_path(app) Returns the path for which the cookie should be valid. null_session_class make_null_session() will look here for the class that should be created when a null session is requested. session) A helper method that returns an expiration date for the session or None if the session is linked to the browser session. and falls back to APPLICATION_ROOT or uses / if it’s None. get_expiration_time(app. make_null_session(app) Creates a null session which acts as a replacement object if the real session support could not be loaded due to a configuration error. pickle_based = False A flag that indicates if the session interface is pickle based. This currently just returns the value of the SESSION_COOKIE_HTTPONLY config var. request) This method has to be implemented and must either return None in case the loading failed because of a configuration error or an instance of a ses- sion object which implements a dictionary like interface + the methods and attributes on SessionMixin. This creates an instance of null_session_class by default. This currently just returns the value of the SESSION_COOKIE_SECURE setting. This can be used 222 . This mainly aids the user experience because the job of the null session is to still support lookup without complaining but modifications are answered with a helpful error message of what failed. This checks if the object is an instance of null_session_class by default. alias of NullSession open_session(app. get_cookie_httponly(app) Returns True if the session cookie should be httponly. New in version 0.

New in version 0. This is used by session backends to figure out if they should emit a set-cookie header or not.sessions.sessions. by Flask extensions to make a decision in regards to how to deal with the session object.SecureCookieSession(initial=None) Base class for sessions based on signed cookies. Will still allow read-only access to the empty session but fail on setting. The default is sha1 key_derivation = ‘hmac’ the name of the itsdangerous supported key derivation.sessions. The default is hmac. if set to True it’s always set if the session is permanent.11.NullSession(initial=None) Class used to generate nicer error messages if sessions are not available. session) Indicates whether a cookie should be set now or not. This check is usually skipped if sessions get deleted.SessionMixin Expands a basic dictionary with an accessors that are expected by Flask exten- sions and users for the session. static digest_method() the hash function to use for the signature. class flask. response) This is called for actual sessions returned by open_session() at the end of the request. This is still called during a request context so if you absolutely need access to the request you can do that. The default is a compact JSON derived serializer with support for some extra Python types such as datetime objects or tuples. salt = ‘cookie-session’ the salt that should be applied on top of the secret key for the signing of cookie based sessions. serializer = <flask. session. save_session(app. class flask. 223 . The default behavior is controlled by the SESSION_REFRESH_EACH_REQUEST config variable.TaggedJSONSerializer object> A python serializer for the payload.SecureCookieSessionInterface The default session interface that stores sessions in signed cookies through the itsdangerous module.sessions. New in version 0.sessions. If it’s set to False then a cookie is only set if the session is modified.10. session_class alias of SecureCookieSession class flask. should_set_cookie(app. class flask.

markup objects. which can be set after instantiation of the app. new = False some session backends can tell you if a session is new.session_json_serializer = <flask. 224 .test_client() object in client.sessions. Use with caution. Either catch this down yourself or use the permanent_session_lifetime attribute on the app which converts the result to an integer automatically.test.12: app. datetime). Changed in version 0. For general information about how to use this class refer to werkzeug. Cur- rently the following extended values are supported in the JSON it dumps: •Markup objects •UUID objects •datetime objects •tuples Notice The PERMANENT_SESSION_LIFETIME config key can also be an integer starting with Flask 0.environ_base. permanent this reflects the '_permanent' key in the dict.test_client() includes preset default environ- ment. The default mixin implementation just hardcodes False in. flask. 23.FlaskClient(*args.sessions.8.7 Test Client class flask.Client. **kwargs) Works like a regular Werkzeug test client but has some knowledge about how Flask works to defer the cleanup of the request context stack to the end of a with body when used in a with statement. but that is not nec- essarily guaranteed. The default mixin imple- mentation just hardcodes True in. This object provides dumping and loading methods similar to simplejson but it also tags certain builtin Python objects that commonly appear in sessions. modified = True for some backends this will always be True. Basic usage is outlined in the Testing Flask Applications chapter. but some backends will default this to false and detect changes in the dictionary for as long as changes do not happen on mutable structures in the session.TaggedJSONSerializer object> A customized JSON serializer that supports a few extra types that we take for granted when serializing (tuples.testing.

This is useful for extensions that 225 . Once the with block is left the session is stored back. See Notes On Proxies for more information. Starting with Flask 0. flask.session_transaction() as session: session['value'] = 42 Internally this is implemented by going through a temporary test request context and since session handling could depend on request variables this function accepts the same arguments as test_request_context() which are directly passed through. For example a database connection or the user that is currently logged in. None) It’s now also possible to use the in operator on it to see if an attribute is defined and it yields all keys on iteration. This is especially useful when combined with the Faking Resources and Context pattern for testing. As of 0. This can be used to modify the session that the test client uses.10 you can use the get() method to get an attribute or None (or the second argument) if it’s not set. with client.g. These two usages are now equivalent: user = getattr(flask. like it does for request and session.10 this is stored on the application context and no longer on the request context which means it becomes available if only the application context is bound and not yet a request. a global variable is not good enough because it would break in threaded environments.get('user'.g. 'user'. 23.11 you can use pop() and setdefault() in the same way you would use them on a dictionary. 23. None) user = flask.g Just store on this whatever you want. **kwargs) When used in combination with a with statement this opens a session trans- action. This is a proxy.9 Useful Functions and Classes flask. Flask provides you with a special object that ensures it is only valid for the active request and that will return different values for each request.8 Application Globals To share data that is valid for one request only from one function to another.current_app Points to the application handling the request. Additionally as of 0. session_transaction(*args. In a nutshell: it does the right thing.

class User(db.. but fail silently if it is unavailable. .has_request_context() If you have code that wants to test if a request context is there or not this function can be used. flask. See Notes On Proxies for more information.username = username if remote_addr is None and request: remote_addr = request. it can access flask.remote_addr self. flask. For instance. username.spawn(do_some_work) return 'Regular response' 226 . gevent.remote_addr self.Model): def __init__(self.request like you # would otherwise in the view function.remote_addr = remote_addr Alternatively you can also just test any of the context bound objects (such as request or g for truthness): class User(db. This is a proxy.username = username if remote_addr is None and has_request_context(): remote_addr = request.remote_addr = remote_addr New in version 0.7. so you can change the value of this proxy by using the app_context() method. you may want to take advantage of request informa- tion if the request object is available. username. This is useful when working with greenlets..route('/') def index(): @copy_current_request_context def do_some_work(): # do some work here. This is powered by the application context and not by the request context. Example: import gevent from flask import copy_current_request_context @app. want to support multiple applications running side by side. The moment the function is deco- rated a copy of the request context is created and then pushed when the function is called.Model): def __init__(self. remote_addr=None): self. remote_addr=None): self.copy_current_request_context(f ) A helper function that decorates a function to retain the current request context.

url = lookup_url(endpoint. exc_value. head over to the Quickstart. flask. An example: def external_url_handler(error. 227 . When it does. To integrate applications.). the whole pair is skipped. Note that this is for building URLs outside the current application. The url_for function results in a BuildError when the current app does not have a URL for the given endpoint and values. exc_value. **values) if url is None: # External lookup did not have a URL. If the value of a query argument is None. tb = sys. lookup_url is some utility function you've built # which looks up the endpoint in some external URL registry.has_app_context() Works like has_request_context() but for the application context.index') For more information. flask.9.append(external_url_handler) Here. values): "Looks up an external URL when `url_for` cannot build a URL.10. New in version 0. and endpoint and values are the argu- ments passed into url_for. Flask has a hook to intercept URL build errors through Flask. **values) Generates a URL to the given endpoint with the method provided.exc_info() if exc_value is error: raise exc_type. In case blueprints are active you can shortcut references to the same blueprint by prefixing the local endpoint with a dot (. # Re-raise the BuildError. tb else: raise error # url_for will use this result. and not for handling 404 NotFound errors.url_build_error_handlers. This will reference the index function local to the current blueprint: url_for('. error is the instance of BuildError. Variable arguments that are unknown to the target endpoint are appended to the generated URL as query arguments. the current_app calls its url_build_error_handlers if it is not None. which can return a string to use as the result of url_for (instead of url_for‘s default to raise the BuildError exception) or re-raise the exception. exc_type. # Here. instead of raising BuildError. New in version 0.url_for(endpoint. in context of original traceback.url_build_error_handlers. New in version 0. You can also just do a boolean check on the current_app object instead." # This is an example of hooking the build_error_handler.10: The _scheme parameter was added. return url app. endpoint.

wrappers. If the first argument to the callable is an integer it will be looked up in the mapping. The rest of the arguments are forwarded to the exception constructor. As of Werkzeug 0. redirects the client to the target location. 300 is not supported because it’s not a real redirect and 304 because it’s the answer for a request with a request with defined If-Modified-Since headers. flask. The default behavior uses the same scheme as the current re- quest. this also can be set to an empty string to build protocol-relative URLs. New in version 0.Response if un- specified. Server address can be changed via SERVER_NAME configuration variable which defaults to localhost. Because views do not have to return response objects but can return a value that is converted 228 . New in version 0. • _scheme – a string specifying the desired URL scheme. an absolute URL is generated. if it’s a WSGI application it will be raised in a proxy exception. Parameters • endpoint – the endpoint of the URL (name of the function) • values – the variable arguments of the URL rule • _external – if set to True. code=302. New in version 0. and 307. Supported codes are 301. Parameters • location – the location the response should redirect to. New in version 0. flask. • _method – if provided this explicitly specifies an HTTP method.10. 303. • _anchor – if provided this is added as anchor to the URL. Response=None) Returns a response object (a WSGI application) that.10: The class used for the Response object can now be passed in. The default is werkzeug.6: The location can now be a unicode string that is encoded using the iri_to_uri() function.9: The _anchor and _method parameters were added. if called.abort() When passed a dict of code -> exception items it can be used as callable that raises exceptions.handle_build_error() on BuildError. 305. defaults to 302.make_response(*args) Sometimes it is necessary to set additional headers in a view.9: Calls Flask. or PREFERRED_URL_SCHEME from the app configuration if no request context is available. • code – the redirect status code.redirect(location. flask. The _ex- ternal parameter must be set to True or a ValueError is raised. • Response (class) – a Response class to use when instantiating a response. 302.

Flask. This for example creates a response with a 404 error code: response = make_response(render_template('not_found. 404) The other use case of this function is to force the return value of a view function into a response which is helpful with view decorators: response = make_response(view_function()) response. foo=42)) response.headers['X-Parachutes'] = 'parachutes are cool' return response This function accepts the very same arguments you can return from a view func- tion.route('/') def index(): @after_this_request def add_header(response): response. flask. into a response object by Flask itself.headers['X-Parachutes'] = 'parachutes are cool' Internally this function does the following things: •if no arguments are passed.html'.html'). foo=42) You can now do something like this: def index(): response = make_response(render_template('index. New in version 0.make_response() is invoked with it. This function can be called instead of using a return and you will get a response object which you can use to attach headers.make_response() function as tuple. it creates a new response argument •if one argument is passed. If view looked like this and you want to add a new header: def index(): return render_template('index.html'. the arguments are passed to the flask.6. This is useful to modify response objects. •if more than one argument is passed. flask. it becomes tricky to add headers to it. Example: @app.headers['X-Foo'] = 'Parachute' return response return 'Hello World!' 229 .Flask. The function is passed the response object and has to return the same or a new one.after_this_request(f ) Executes a function after this request.

you should use send_from_directory() instead. Parameters • filename_or_fp – the filename of the file to send in latin-1. when None.12: The filename is no longer automatically inferred from file objects. This is more useful if a function other than the view function wants to modify a response. Changed in version 0.9: cache_timeout pulls its default from application config. this method will try to upgrade the response stream to support range requests. Pass a filename if you are able to. This functionality will be removed in Flask 1. The default behavior is now to attach etags. ETags will also be attached automatically if a filename is provided. pass a filepath via filename_or_fp or attachment_filename. By default it will try to use the WSGI server’s file_wrapper support. add_etags=True.9. This will use the most efficient method available and configured. attach- ment_filename=None.2. For instance think of a decorator that wants to add some headers without converting the return value into a response object. con- ditional=False. New in version 0. but you can also explicitly provide one. By default it will try to guess the mimetype for you. Alternatively you can set the application’s use_x_sendfile attribute to True to directly emit an X-Sendfile header. The mimetype guessing requires a filename or an attachment_filename to be provided.7: mimetype guessing and etag support for file objects was deprecated because it was unreliable. otherwise attach an etag yourself. Changed in version 0.0 Changed in version 0. cache_timeout and conditional parameters were added. Alternatively a file object might be provided in which case X-Sendfile might not work and fall back to the traditional 230 . This however requires support of the underlying webserver for X-Sendfile.send_file(filename_or_fp.5: The add_etags. You can turn this off by setting add_etags=False. New in version 0. Please never pass filenames to this function from user sources. If you want to use automatic mimetype and etag support. cache_timeout=None. New in version 0. Changed in version 0. flask. This will allow the request to be answered with partial content response. mimetype=None. last_modified=None) Sends the contents of a file to the client. If conditional=True and filename is provided. as_attachment=False. This is relative to the root_path if a relative path is speci- fied.12: The attachment_filename is preferred over filename for MIME-type detection. For extra security you probably want to send certain files as at- tachment (HTML for instance).

• add_etags – set to False to disable attaching of etags. this overrides its mtime. filename. Make sure that the file pointer is positioned at the start of data to send before calling send_file(). Parameters • directory – the directory where all the files are stored.config['UPLOAD_FOLDER']. Example usage: @app. • cache_timeout – the timeout in seconds for the headers. 231 .5. • filename – the filename relative to that directory to download.send_from_directory(directory. • last_modified – set the Last-Modified header to this value. a datetime or timestamp. • conditional – set to True to enable conditional responses. This is a secure way to quickly expose static files from an upload folder or something similar. • options – optional keyword arguments that are directly for- warded to send_file(). as_attachment=True) Sending files and Performance It is strongly recommended to activate either X-Sendfile support in your web- server or (if no authentication happens) to tell the webserver to serve files for the given path on its own without calling into the web application for improved performance. • attachment_filename – the filename for the attachment if it dif- fers from the file’s filename.route('/uploads/<path:filename>') def download_file(filename): return send_from_directory(app. If a file path is given. method. flask. If a file was passed. filename. New in version 0. **options) Send a file from a given directory with send_file(). auto detection happens as fallback. • as_attachment – set to True if you want to send this file with a Content-Disposition: attachment header. otherwise an error will be raised. this value is set by get_send_file_max_age() of current_app. • mimetype – the mimetype of the file if provided. When None (default).

.config['WIKI_FOLDER'].. The constructor of the Markup class can be used for three different things: When passed an unicode object it’s assumed to be safe. 'rb') as fd: content = fd. def __html__(self): . This implements the __html__ interface a couple of frame- works and web applications use. Markup is a direct subclass of unicode and pro- vides all the methods of unicode just that it escapes arguments passed and always returns Markup. return '<a href="#">foo</a>' .Markup Marks a string as being safe for inclusion in HTML/XML output without need- ing to be escaped. <.read() # Read and process the file content.flask. The escape function returns markup objects so that double escaping can’t happen.route('/wiki/<path:filename>') def wiki_page(filename): filename = safe_join(app. Marks return value as markup string. >. flask.. and ” in string s to HTML-safe sequences. Use this if you need to display text that might contain such characters in HTML. *pathnames) Safely join directory and zero or more untrusted pathnames components.escape(s) → markup Convert the characters &. • pathnames – the untrusted pathnames relative to that directory.. class flask. when passed an object with an HTML representation (has an __html__ method) that representation is used. Parameters • directory – the trusted base directory. Raises NotFound if one or more passed paths fall out of its boundaries. >>> Markup(Foo()) Markup(u'<a href="#">foo</a>') If you want object passed being always treated as unsafe you can use the escape() classmethod to create a Markup object: 232 . ‘... otherwise the object passed is converted into a unicode string and then assumed to be safe: >>> Markup("Hello <em>World</em>!") Markup(u'Hello <em>World</em>!') >>> class Foo(object): .. filename) with open(filename..safe_join(directory. Example usage: @app.

>>> Markup.escape("Hello <em>World</em>!")
Markup(u'Hello &lt;em&gt;World&lt;/em&gt;!')

Operations on a markup string are markup aware which means that all argu-
ments are passed through the escape() function:

>>> em = Markup("<em>%s</em>")
>>> em % "foo & bar"
Markup(u'<em>foo &amp; bar</em>')
>>> strong = Markup("<strong>%(text)s</strong>")
>>> strong % {'text': '<blink>hacker here</blink>'}
Markup(u'<strong>&lt;blink&gt;hacker here&lt;/blink&gt;</strong>')
>>> Markup("<em>Hello</em> ") + "<foo>"
Markup(u'<em>Hello</em> &lt;foo&gt;')

classmethod escape(s)
Escape the string. Works like escape() with the difference that for sub-
classes of Markup this function would return the correct subclass.
Unescape markup into an text_type string and strip all tags. This also re-
solves known HTML4 and XHTML entities. Whitespace is normalized to

>>> Markup("Main &raquo; <em>About</em>").striptags()
u'Main \xbb About'

Unescape markup again into an text_type string. This also resolves known
HTML4 and XHTML entities:

>>> Markup("Main &raquo; <em>About</em>").unescape()
u'Main \xbb <em>About</em>'

23.10 Message Flashing

flask.flash(message, category=’message’)
Flashes a message to the next request. In order to remove the flashed mes-
sage from the session and to display it to the user, the template has to call
Changed in version 0.3: category parameter added.
• message – the message to be flashed.
• category – the category for the message. The following values
are recommended: 'message' for any kind of message, 'error'


for errors, 'info' for information messages and 'warning' for
warnings. However any kind of string can be used as category.
flask.get_flashed_messages(with_categories=False, category_filter=[])
Pulls all flashed messages from the session and returns them. Further calls in the
same request to the function will return the same messages. By default just the
messages are returned, but when with_categories is set to True, the return value
will be a list of tuples in the form (category, message) instead.
Filter the flashed messages to one or more categories by providing those cate-
gories in category_filter. This allows rendering categories in separate html blocks.
The with_categories and category_filter arguments are distinct:
•with_categories controls whether categories are returned with message text
(True gives a tuple, where False gives just the message text).
•category_filter filters the messages down to only those matching the pro-
vided categories.
See Message Flashing for examples.
Changed in version 0.3: with_categories parameter added.
Changed in version 0.9: category_filter parameter added.
• with_categories – set to True to also receive categories.
• category_filter – whitelist of categories to limit return values

23.11 JSON Support

Flask uses simplejson for the JSON implementation. Since simplejson is provided by
both the standard library as well as extension, Flask will try simplejson first and then
fall back to the stdlib json module. On top of that it will delegate access to the current
application’s JSON encoders and decoders for easier customization.
So for starters instead of doing:

import simplejson as json
except ImportError:
import json

You can instead just do this:

from flask import json

For usage examples, read the json documentation in the standard library. The follow-
ing extensions are by default applied to the stdlib’s JSON module:
1. datetime objects are serialized as RFC 822 strings.


2. Any object with an __html__ method (like Markup) will have that method called
and then the return value is serialized as string.
The htmlsafe_dumps() function of this json module is also available as filter called
|tojson in Jinja2. Note that inside script tags no escaping must take place, so make
sure to disable escaping with |safe if you intend to use it inside script tags unless
you are using Flask 0.10 which implies that:

<script type=text/javascript>
doSomethingWith({{ user.username|tojson|safe }});

Auto-Sort JSON Keys
The configuration variable JSON_SORT_KEYS (Configuration Handling) can be set to false
to stop Flask from auto-sorting keys. By default sorting is enabled and outside of the
app context sorting is turned on.
Notice that disabling key sorting can cause issues when using content based HTTP
caches and Python’s hash randomization feature.

flask.json.jsonify(*args, **kwargs)
This function wraps dumps() to add a few enhancements that make life easier. It
turns the JSON output into a Response object with the applicationhttps://www.scribd.com/json mime-
type. For convenience, it also converts multiple arguments into an array or mul-
tiple keyword arguments into a dict. This means that both jsonify(1,2,3) and
jsonify([1,2,3]) serialize to [1,2,3].
For clarity, the JSON serialization behavior has the following differences from
1.Single argument: Passed straight through to dumps().
2.Multiple arguments: Converted to an array before being passed to dumps().
3.Multiple keyword arguments: Converted to a dict before being passed to
4.Both args and kwargs: Behavior undefined and will throw an exception.
Example usage:

from flask import jsonify

def get_current_user():
return jsonify(username=g.user.username,

This will send a JSON response like this to the browser:


"username": "admin",
"email": "admin@localhost",
"id": 42

Changed in version 0.11: Added support for serializing top-level arrays. This
introduces a security risk in ancient browsers. See JSON Security for details.
This function’s response will be pretty printed if it was not requested
with X-Requested-With: XMLHttpRequest to simplify debugging unless the
JSONIFY_PRETTYPRINT_REGULAR config parameter is set to false. Compressed (not
pretty) formatting currently means no indents and no spaces after separators.
New in version 0.2.
flask.json.dumps(obj, **kwargs)
Serialize obj to a JSON formatted str by using the application’s configured en-
coder (json_encoder) if there is an application on the stack.
This function can return unicode strings or ascii-only bytestrings by default
which coerce into unicode strings automatically. That behavior by default is
controlled by the JSON_AS_ASCII configuration variable and can be overridden
by the simplejson ensure_ascii parameter.
flask.json.dump(obj, fp, **kwargs)
Like dumps() but writes into a file object.
flask.json.loads(s, **kwargs)
Unserialize a JSON object from a string s by using the application’s configured
decoder (json_decoder) if there is an application on the stack.
flask.json.load(fp, **kwargs)
Like loads() but reads from a file object.
class flask.json.JSONEncoder(skipkeys=False, ensure_ascii=True,
check_circular=True, allow_nan=True,
sort_keys=False, indent=None, separators=None,
encoding=’utf-8’, default=None)
The default Flask JSON encoder. This one extends the default simplejson encoder
by also supporting datetime objects, UUID as well as Markup objects which are
serialized as RFC 822 datetime strings (same as the HTTP date format). In order
to support more data types override the default() method.
Implement this method in a subclass such that it returns a serializable object
for o, or calls the base implementation (to raise a TypeError).
For example, to support arbitrary iterators, you could implement default
like this:

def default(self, o):


iterable = iter(o)
except TypeError:
return list(iterable)
return JSONEncoder.default(self, o)

class flask.json.JSONDecoder(encoding=None, object_hook=None,
parse_float=None, parse_int=None,
parse_constant=None, strict=True, ob-
The default JSON decoder. This one does not change the behavior from the de-
fault simplejson decoder. Consult the json documentation for more informa-
tion. This decoder is not only used for the load functions of this module but also

23.12 Template Rendering

flask.render_template(template_name_or_list, **context)
Renders a template from the template folder with the given context.
• template_name_or_list – the name of the template to be ren-
dered, or an iterable with template names the first one existing
will be rendered
• context – the variables that should be available in the context
of the template.
flask.render_template_string(source, **context)
Renders a template from the given template source string with the given context.
Template variables will be autoescaped.
• source – the source code of the template to be rendered
• context – the variables that should be available in the context
of the template.
flask.get_template_attribute(template_name, attribute)
Loads a macro (or variable) a template exports. This can be used to invoke a
macro from within Python code. If you for example have a template named
_cider.html with the following contents:

{% macro hello(name) %}Hello {{ name }}!{% endmacro %}

You can access this from Python code like this:


hello = get_template_attribute('_cider.html', 'hello')
return hello('World')

New in version 0.2.
• template_name – the name of the template
• attribute – the name of the variable of macro to access

23.13 Configuration

class flask.Config(root_path, defaults=None)
Works exactly like a dict but provides ways to fill it from files or special dictio-
naries. There are two common patterns to populate the config.
Either you can fill the config from a config file:


Or alternatively you can define the configuration options in the module that calls
from_object() or provide an import path to a module that should be loaded.
It is also possible to tell it to use the same module and with that provide the
configuration values just before the call:

DEBUG = True
SECRET_KEY = 'development key'

In both cases (loading from any Python file or loading from modules), only up-
percase keys are added to the config. This makes it possible to use lowercase
values in the config file for temporary values that are not added to the config or
to define the config keys in the same file that implements the application.
Probably the most interesting way to load configurations is from an environment
variable pointing to a file:


In this case before launching the application you have to set this environment
variable to the file you want to use. On Linux and OS X use the export statement:

export YOURAPPLICATION_SETTINGS='/path/to/config/file'

On windows use set instead.


11. This can either be an absolute filename or a filename relative to the root path. This is basically just a shortcut with nicer error messages for this line of code: app.from_pyfile(os. **kwargs) Updates the config like update() ignoring items with non-upper keys. from_json(filename. When the config object is created by the application. True if able to load config.config.environ['YOURAPPLICATION_SETTINGS']) Parameters • variable_name – name of the environment variable • silent – set to True if you want silent failure for missing files. Parameters • filename – the filename of the JSON file. A dict object will not work with from_object() because the keys of a dict are not attributes of the dict class. from_object() loads only the uppercase attributes of the module/class. • silent – set to True if you want silent failure for missing files. New in version 0. from_object(obj) Updates the values from the given object. Example of module-based configuration: 239 . • root_path – path to which files are read relative from. silent=False) Updates the values in the config from a JSON file. This function behaves as if the JSON object was a dictionary and passed to the from_mapping() function. New in version 0.11. An object can be of one of the following two types: •a string: in this case the object with that name will be imported •an actual object reference: that object is used directly Objects are usually either modules or classes. False otherwise. Returns bool. silent=False) Loads a configuration from an environment variable pointing to a configu- ration file. • defaults – an optional dictionary of default values from_envvar(variable_name. from_mapping(*mapping. this is the appli- cation’s root_path.

website.config['IMAGE_STORE_PATH'] = '/var/apphttps://www.scribd.com/images' app. 'base_url': 'http://img. 'path': '/var/apphttps://www.scribd.com/images'. Example usage: app.config.config.get_namespace('IMAGE_STORE_') The resulting dictionary image_store_config would look like: { 'type': 'fs'. The actual config should be loaded with from_pyfile() and ideally from a location not within the package because the package might be installed system wide.com' } This is often useful when configuration options map directly to keyword arguments in functions or class constructors. get_namespace(namespace. Parameters • filename – the filename of the config. This can either be an absolute filename or a filename relative to the root path.config['IMAGE_STORE_BASE_URL'] = 'http://img. This function behaves as if the file was imported as module with the from_object() function.from_object(default_config) You should not use this function to load the actual configuration but rather configuration defaults.config.com' image_store_config = app. Parameters obj – an import name or object from_pyfile(filename.from_object('yourapplication. New in version 0. app.7: silent parameter. lowercase=True. See Development / Production for an example of class-based configuration using from_object(). trim_namespace=True) Returns a dictionary containing a subset of configuration options that match the specified namespace/prefix.config['IMAGE_STORE_TYPE'] = 'fs' app. Parameters • namespace – a configuration namespace • lowercase – a flag indicating if the keys of the resulting dic- tionary should be lowercase 240 .website.default_config') from yourapplication import default_config app. silent=False) Updates the values in the config from a Python file. • silent – set to True if you want silent failure for missing files.

23.ext This module acts as redirect import module to Flask extensions. This function however can help you keep the context around for longer: from flask import stream_with_context.args['name'] yield '!' return Response(generate()) Alternatively it can also be used around a specific generator: from flask import stream_with_context. • trim_namespace – a flag indicating if the keys of the resulting dictionary should not include the namespace New in version 0. The downside is that if you are using streamed responses.route('/stream') def streamed_response(): def generate(): 241 . Response @app. request.11. request. This is done for efficiency reasons and to make it less likely to encounter memory leaks with badly written WSGI middlewares.15 Stream Helpers flask. Response @app. If you want to use an extension named “Flask-Foo” you would import it from ext as follows: from flask.ext import foo New in version 0. 23. It was added in 0.8.stream_with_context(generator_or_function) Request contexts disappear when the response is started on the server.8 as the canonical way to import Flask extensions and makes it possible for us to have more flexibility in how we distribute extensions.14 Extensions flask.route('/stream') def streamed_response(): @stream_with_context def generate(): yield 'Hello ' yield request. the generator cannot access request bound information any more.

environ. This will also trigger the execution of functions registered by the teardown_request() decorator.16 Useful Internals class flask.RequestContext(app.10. Changed in version 0. copy() Creates a copy of this request context with the same request object.ctx. instead use test_request_context() and request_context() to create this object._preserve_context' to True on the WSGI environment the context will not pop itself at the end of the request. pop(exc=<object object>) Pops the request context and unbinds it by doing that. request=None) The request context contains all request relevant information. With 0. Make sure to properly pop() the stack yourself in that situation. yield 'Hello ' yield request. match_request() Can be overridden by a subclass to hook into the matching of the request. You might find this helpful for unittests where you need the information from the context local around for a little longer. New in version 0. By setting 'flask. This can be used to move a request context to a different greenlet. 242 .9. When the request context is popped. 23. The request context is automatically popped at the end of the request for you. It will create the URL adapter and request object for the WSGI environment provided.9: Added the exc argument. it will evaluate all the functions registered on the application for teardown execution (teardown_request()). Because the actual request object is the same this cannot be used to move a request context to a different thread unless access to the request object is locked. In debug mode the request context is kept around if exceptions happen so that interactive debuggers have a chance to introspect the data. It is created at the beginning of the request and pushed to the _request_ctx_stack and removed at the end of it.args['name'] yield '!' return Response(stream_with_context(generate())) New in version 0. This is used by the test_client() for example to implement the deferred cleanup functionality. Do not attempt to use this class directly. otherwise your unittests will leak memory.4 this can also be forced for requests that did not fail and outside of DEBUG mode.

similar to how the RequestContext binds request informa- tion. This is a documented instance and can be used by extensions and application code but the use is discouraged in general. session the active session object. This is mainly there for extensions to store data.session class flask._app_ctx_stack Works similar to the request context but only binds the application.blueprints.AppContext(app) The application context binds an application object implicitly to the current thread or greenlet. pop(exc=<object object>) Pops the app context. push() Binds the app context to the current context. options. app. New in version 0. flask.ctx.g object. push() Binds the request context to the current context. Example usage: from flask import _request_ctx_stack def get_session(): ctx = _request_ctx_stack. An in- stance of this class is created by the make_setup_state() method and later passed to all register callback functions.9. 243 . The following attributes are always present on each layer of the stack: app the active Flask application. first_registration) Temporary holder object for registering a blueprint with the application. request the current request object.top if ctx is not None: return ctx._request_ctx_stack The internal LocalStack that is used to implement all the context local objects used in Flask. class flask. The application context is also implicitly created if a request context is created but the application is not on top of the individual application context.BlueprintSetupState(blueprint. flashes an internal cache for the flashed messages. flask. url_adapter the URL adapter that was used to match the request. g an object with all the attributes of the flask.

url_prefix = None The prefix that should be used for all URLs defined on the blueprint.signals_available True if the signaling system is available. The endpoint is automatically prefixed with the blueprint’s name. add_url_rule(rule. options = None a dictionary with all options that were passed to the register_blueprint() method. context. Example subscriber: def log_template_renders(sender.debug('Rendering template "%s" with context %s'. None otherwise.logger. The signal is in- voked with the instance of the template as template and the context as dictionary (named context).17 Signals New in version 0. url_defaults = None A dictionary with URL defaults that is added to each and every URL that was defined with the blueprint. This is the case when blinker is installed. first_registration = None as blueprints can be registered multiple times with the application and not everything wants to be registered multiple times on it. app = None a reference to the current application blueprint = None a reference to the blueprint that created this setup state. view_func=None. subdomain = None The subdomain that the blueprint should be active for. 23.name or 'string template'.6. signals. endpoint=None. template. this attribute can be used to figure out if the blueprint was registered in the past already. **options) A helper method to register a rule (and optionally a view function) to the application. **extra): sender. context) 244 . The following signals exist in Flask: flask. template.template_rendered This signal is sent when a template was successfully rendered.

got_request_exception This signal is sent when an exception happens during request processing. Example subscriber: 245 .debug('Rendering template "%s" with context %s'.name or 'string template'. before any request process- ing happens. Example subscriber: def log_response(sender. the subscriber can access the request with the standard global proxies such as request. app) flask. response) from flask import request_finished request_finished.logger. template.connect(log_request. **extra): sender. It is sent before the standard exception handling kicks in and even in debug mode.connect(log_template_renders.request_finished This signal is sent right before the response is sent to the client.request_started This signal is sent when the request context is set up. ' 'Response: %s'. context. where no exception handling happens. **extra): sender.connect(log_template_renders. The exception itself is passed to the sub- scriber as exception.debug('Request context is set up') from flask import request_started request_started. Example subscriber: def log_template_renders(sender. response. app) flask. from flask import template_rendered template_rendered. app) flask. Because the request context is already bound.debug('Request context is about to close down. template. It is passed the response to be sent named response. Example subscriber: def log_request(sender.logger. context) from flask import before_render_template before_render_template.logger. The signal is invoked with the instance of the template as template and the context as dictionary (named context). **extra): sender.connect(log_response.before_render_template This signal is sent before template rendering process. app) flask.

Currently functions listening to this signal are called after the regular teardown handlers.appcontext_pushed This signal is sent when an application context is pushed. app) flask. This is usually useful for unittests in order to temporarily hook in information. even if an exception is caused. Example subscriber: def close_db_connection(sender.appcontext_tearing_down This signal is sent when the app context is tearing down. Example subscriber: def close_db_connection(sender. exception. app) This will also be passed an exc keyword argument that has a reference to the exception that caused the teardown if there was one. Example usage: from contextlib import contextmanager from flask import appcontext_pushed @contextmanager def user_set(app. **extra): sender. but this is not something you can rely on. This is always called.connect(close_db_connection. **kwargs): 246 . exception) from flask import got_request_exception got_request_exception. **extra): session. **extra): session.debug('Got exception during processing: %s'.connect(log_exception. flask.9. even if an exception is caused.close() from flask import appcontext_tearing_down appcontext_tearing_down.connect(close_db_connection. Currently functions listening to this signal are called after the regular teardown handlers. app) As of Flask 0. This is always called. this will also be passed an exc keyword argument that has a reference to the exception that caused the teardown if there was one.logger. user): def handler(sender.close() from flask import request_tearing_down request_tearing_down. flask. but this is not something you can rely on.request_tearing_down This signal is sent when the request is tearing down. For instance it can be used to set a resource early onto the g object. The sender is the application. def log_exception(sender.

10. New in version 0. This usually falls in line with the appcontext_tearing_down signal. category.connected_to(handler. The sender is the application. otherwise a dummy class that creates fake signals. **extra): recorded.connect(record.data == 'username=john' New in version 0.7. class flask.message_flashed This signal is sent when the application is flashing a message.Namespace if blinker is available. signal(name. message. app) New in version 0. 23. 'john'): c = app. category)) from flask import message_flashed message_flashed. flask. This class is available for Flask extensions that want to provide the same fallback system as Flask itself.base.10. doc=None) Creates a new signal for this namespace if blinker is available.test_client() resp = c.appcontext_popped This signal is sent when an application context is popped. class signals. g.18 Class-Based Views New in version 0. flask.View Alternative way to use view functions.10.Namespace An alias for blinker.append((message. Example subscriber: recorded = [] def record(sender.views. otherwise returns a fake signal that has a send method that will do nothing but will fail with a RuntimeError for all other operations. including connecting.get('/users/me') assert resp. app): yield And in the testcode: def test_user_me(self): with user_set(app. A subclass has to implement 247 .user = user with appcontext_pushed. The messages is sent as message keyword argument and the category as category.

*class_args. New in version 0. This method is called with all the arguments from the URL rule. view_func=MyView. **class_kwargs) Converts the class into an actual view function that can be used with the routing system. name): return 'Hello %s!' % name app. dispatch_request() Subclasses have to override this method to implement the actual view func- tion code. You can place one or more decorators in this list and whenever the view function is created the result is automatically decorated. The decorators stored in the decorators list are applied one after another when the view function is created.. methods = None A list of methods this view can handle. If methods is provided the methods do not have to be passed to the add_url_rule() method explicitly: class MyView(View): methods = ['GET'] def dispatch_request(self. Internally this generates a function on the fly which will in- stantiate the View on each request and call the dispatch_request() method on it. decorators = () The canonical way to decorate class-based views is to decorate the return value of as_view(). 248 . However since this moves parts of the logic from the class declaration to the place where it’s hooked into the routing system..add_url_rule('/hello/<name>'. Note that you can not use the class based decorators since those would decorate the view class and not the generated view function! classmethod as_view(name. The arguments passed to as_view() are forwarded to the constructor of the class.8. dispatch_request() which is called with the view arguments from the URL rout- ing system.as_view('myview')) When you want to decorate a pluggable view you will have to either do that when the view function is created (by wrapping the return value of as_view()) or you can use the decorators attribute: class SecretView(View): methods = ['GET'] decorators = [superuser_required] def dispatch_request(self): .

For instance if you implement a method called get() it means it will re- spond to 'GET' requests and the dispatch_request() implementation will auto- matically forward your request to that.Flask. You can use the flask. Variable parts in the route can be specified with angular brackets (/user/<username>).get('counter'. 0) def post(self): session['counter'] = session.add_url_rule() function. By default a variable part in the URL accepts any string without a slash however a different converter can be specified as well by using <converter:name>.url_map.as_view('counter')) 23.add_url_rule('/counter'. view_func=CounterAPI. Variable parts are passed to the view function as keyword arguments.Flask. You can directly access the underlying Werkzeug routing system which is ex- posed as flask.MethodView Like a regular class-based view but that dispatches requests to particular meth- ods. 0) + 1 return 'OK' app. Here are some examples: @app. You can use the flask.get('counter'.class flask.19 URL Route Registrations Generally there are three ways to define rules for the routing system: 1.route() decorator. The following converters are available: string accepts any text without a slash (the default) int accepts integers float like int but for floating point values path like the default but also accepts slashes any matches one of the items provided uuid accepts UUID strings Custom converters can be defined using flask. 3.route('/') def index(): pass 249 . 2.Flask. Also options is set for you automatically: class CounterAPI(MethodView): def get(self): return session.views.Flask.url_map.

If a rule ends with a slash and is requested without a slash by the user.route('/post/<int:post_id>') def show_post(post_id): pass An important detail to keep in mind is how Flask deals with trailing slashes. 2.@app. the user is automatically redirected to the same page with a trailing slash attached. This is consistent with how web servers deal with static files. 250 .route('/<username>') def show_user(username): pass @app.route('/users/page/<int:page>') def show_users(page): pass This specifies that /users/ will be the URL for page one and /users/page/N will be the URL for page N.route('/users/'. defaults={'page': 1}) @app. They have to be unique however. Here for example is a definition for a URL that accepts an optional page: @app. You can also define multiple rules for the same function. Here are the parameters that route() and add_url_rule() accept. Defaults can also be specified. a 404 not found is raised. This also makes it possi- ble to use relative link targets safely. If a rule does not end with a trailing slash and the user requests the page with a trailing slash. The only difference is that with the route parameter the view function is defined with the decorator instead of the view_func parameter. The idea is to keep each URL unique so the following rules apply: 1.

Full example: def index(): if request. • provide_automatic_options: if this attribute is set Flask will either force enable or disable the automatic implementation of the HTTP OPTIONS response. Flask itself assumes that the name point of the view function is the name of the endpoint if not explicitly stated. This can be useful when working with decorators that want to customize the OPTIONS response on a per-view basis.. methods is a list of methods this rule should be limited to (GET.specifies the rule for the subdomain in case subdomain matching is in use. 23. • methods: If methods are not provided when the URL rule is added. POST etc. rule the URL rule as string end. view_func the function to call when serving a request to the provided endpoint. it will pull the information for the methods from there. Flask will always add these methods when registering a URL rule even if the methods were explicitly overridden in the route() call. They have to be specified as keyword arguments. sub.20 View Function Options For internal usage the view functions can have some attributes attached to customize behavior the view function would normally not have control over. A change to tions Werkzeug is handling of method options. If this is not provided one can specify the function later by storing it in the view_functions dictionary with the endpoint as key. de. OPTIONS is implicitly added and handled by the standard request handling. main **op.method == 'OPTIONS': # custom options handling here .the options to be forwarded to the underlying Rule object.the endpoint for the registered URL rule. If do. See the example above for how faults defaults work.6. Starting with Flask 0. A dictionary with defaults for this rule. 251 . If it does. Flask will look on the view function object itself if a methods attribute exists. Additionally this will be prefixed with the name of the blueprint by default which cannot be customized from the function itself. • required_methods: if this attribute is set. not specified the default subdomain is assumed.. The following at- tributes can be provided optionally to either override some defaults to add_url_rule() or general behavior: • __name__: The name of a function is by default used as endpoint.). If endpoint is provided explicitly this value is used. By default a rule just listens for GET (and implicitly HEAD).

class flask. return 'Hello World!' index. Parameters • add_default_commands – if this is True then the default run and shell commands wil be added. • add_version_option – adds the --version option. 'OPTIONS'] app. group(*args.21 Command Line Interface class flask. • create_app – an optional callback that is passed the script info and returns the loaded app.8: The provide_automatic_options functionality was added. commands=None. Typically it’s created au- tomatically by the FlaskGroup but you can also manually create it and pass it onwards as click object. **extra) Special subclass of the AppGroup group that supports loading more commands from the configured Flask app.add_url_rule('/'. **kwargs) This works exactly like the method of the same name on a regular click. Normally a developer does not have to interface with this class but there are some very advanced use cases for which it makes sense to create an instance of this. Group but it wraps callbacks in with_appcontext() unless it’s disabled by passing with_appcontext=False. Not to be confused with FlaskGroup. add_version_option=True. **kwargs) This works exactly like the method of the same name on a regular click. For information as of why this is useful see Custom Scripts. create_app=None) Help object to deal with Flask applications. This is usually not necessary to in- terface with as it’s used internally in the dispatching to click.AppGroup(name=None. 23. index) New in version 0.provide_automatic_options = False index.methods = ['GET'. 252 .cli.FlaskGroup(add_default_commands=True. create_app=None.cli. **attrs) This works similar to a regular click Group but it changes the behavior of the command() decorator so that it automatically wraps the functions in with_appcontext(). In future versions of Flask this object will most likely play a bigger role.cli. command(*args. Group but it defaults the group class to AppGroup.ScriptInfo(app_import_path=None. class flask.

cli. app_import_path = None Optionally the import path for the Flask application.cli. This can be changed with the –with- threads option which will enable basic multithreading.Command object> Runs an interactive Python shell in the context of a given Flask application.cli object then they are wrapped with this function by default unless it’s disabled. load_app() Loads the Flask app (if not yet loaded) and returns it. flask. flask. flask. 253 . create_app = None Optionally a function that is passed the script info to create the instance of the application.cli. flask. The reloader and debugger are by default enabled if the debug flag of Flask is enabled and disabled otherwise. By default it will not support any sort of concurrency at all to simplify debugging.Command object> Runs a local development server for the Flask application. Calling this multiple times will just result in the already loaded app to be returned. If callbacks are registered directly to the app.run_command = <click.with_appcontext(f ) Wraps a callback so that it’s guaranteed to be executed with the script’s applica- tion context. data = None A dictionary with arbitrary data that can be associated with this script info. This local server is recommended for development purposes only but it can also be used for simple intranet deployments. This is useful for executing small snippets of management code without having to manually configuring the application.core.shell_command = <click. The application will populate the default namespace of this shell according to it’s configuration.core.pass_script_info(f ) Marks a function so that an instance of ScriptInfo is passed as first argument to the click callback.cli.

254 .

legal information and changelog are here for the interested.Part III ADDITIONAL NOTES Design notes. 255 .

256 .

When you want to test something 257 . especially in direct comparison with other frameworks. but why can’t Flask do that itself? Without such an explicit application object the following code: from flask import Flask app = Flask(__name__) @app. Now the question is: when does a microframework need more than one application at the same time? A good example for this is unittesting. but this causes some problems I won’t outline here in detail. Each Flask application has to create an instance of this class itself and pass it the name of the module. this section is for you. The most important one is that implicit appli- cation objects require that there may only be one instance at the time. like maintaining a stack of applications.route('/') def index(): return 'Hello World!' Would look like this instead: from hypothetical_flask import route @route('/') def index(): return 'Hello World!' There are three major reasons for this. CHAPTER TWENTYFOUR DESIGN DECISIONS IN FLASK If you are curious why Flask does certain things the way it does and not differently. In Flask this is an instance of the Flask class.1 The Explicit Application Object A Python web application based on WSGI has to have one central callable object that implements the actual application. 24. This should give you an idea about some of the design decisions that may appear arbitrary and surprising at first. There are ways to fake multiple applications with a single application object.

When the application object is deleted everything it allocated will be freed again. Werkzeug will go quite far with that in that it will automatically redirect to a canonical URL if a route is ambiguous. Whenever you create a Flask instance you usually pass it __name__ as package name. just wrap it and you’re done (though there are better ways to do that so that you do not lose the reference to the application object wsgi_app()). which is a very unreliable way to determine where the application is. With Python’s outstanding support for reflection it can then access the package to figure out where the templates and static files are stored (see open_resource()). The third reason is “explicit is better than implicit”. 24.it can be very helpful to create a minimal application to test specific behavior. That object is your WSGI ap- plication. Another design decision with the Werkzeug routing system is that routes in Werkzeug try to ensure that URLs are unique. If you want to apply a WSGI middleware. Furthermore this design makes it possible to use a factory function to create the appli- cation which is very helpful for unittesting and similar things (Application Factories). 24. This would not be possible without hacks if the object were created ahead of time for you based on a class that is not exposed to you. Now obviously there are frameworks around that do not need any configuration and will still be able to load templates relative to your application module. but Flask will still configure Jinja2 for you. Flask depends on that information to properly load resources relative to your module. This is a requirement if you want to properly imple- ment decorator based routing since decorators could be fired in undefined order when the application is split into multiple modules. While that limitation that Jinja2 is always configured 258 .3 One Template Engine Flask decides on one template engine: Jinja2. Why doesn’t Flask have a pluggable tem- plate engine interface? You can obviously use a different template engine. This means that you can declare routes in arbitrary order and they will still work as expected. But there is another very important reason why Flask depends on an explicit instantia- tion of that class: the package name. But they have to use the current working directory for that. you don’t have to remember anything else. Another thing that becomes possible when you have an explicit object lying around in your code is that you can subclass the base class (Flask) to alter specific behavior.2 The Routing System Flask uses the Werkzeug routing system which was designed to automatically order routes by complexity. Worse: many web- servers do not set the working directory to the directory of your application but to the document root which does not have to be the same folder. The current working directory is process-wide and if you are running multiple applications in one process (which could happen in a webserver without you knowing) the paths will be off.

On the surface they all work the same: you tell the engine to evaluate a template with a set of variables and take the return value as string. packages with dependencies are no longer an issue and there are very few reasons against having libraries that depend on others. template inheritance by taking the availability of XPath into account and more. Flask is a framework that takes advantage of the work already done by Werkzeug to properly interface WSGI (which can be a complex task at times). A template abstraction layer that would not take the unique features of the template engines away is a science on its own and a too large undertaking for a microframework like Flask. but an extension could still depend on Jinja itself. On the other hand an engine like Genshi is based on XML stream evaluation.will probably go away. For instance. Also it provides ways to access macros from Jinja2 templates. support for reusable blocks (macros) that can be used from inside templates and also from Python code. the decision to bundle one template engine and use that will not. Flask uses Jinja2’s extensive autoescaping support. configurable syntax and more. Why shouldn’t it? If we look over to the Ruby side of web development there we have a protocol very similar to WSGI. Just that it’s called Rack there. 24.4 Micro with Dependencies Why does Flask call itself a microframework and yet it depends on two libraries (namely Werkzeug and Jinja2). a certain way to do template inheritance. but besides that it looks very much like a WSGI rendition for Ruby. Furthermore extensions can then easily depend on one template language being present. But that’s about where similarities end. This Rack library has two equivalents in Python: WebOb (formerly Paste) and Werkzeug. uses Unicode for all operations. supports iterative template rendering. You can easily use your own templating language. When it comes to connecting a template engine with an application or framework there is more than just rendering templates. but on top of a library with the same name. Paste is still around but from my understanding it’s sort of deprecated in favour of WebOb. The development of WebOb and Werkzeug started side by side with similar ideas in mind: be a good implementation of WSGI for other applications to take advantage. 259 . Template engines are like programming languages and each of those engines has a certain understanding about how things work. Mako on the other hand treats templates similar to Python modules. Jinja2 for example has an extensive filter sys- tem. But nearly all applications in Ruby land do not work with Rack directly. Thanks to recent developments in the Python package infrastructure.

Why is this the case? Because people have different preferences and requirements and Flask could not meet those if it would force any of this into the core. Everything else is up to you or extensions.5 Thread Locals Flask uses thread local objects (context local objects in fact. Why is that and isn’t that a bad idea? Yes it is usually not such a bright idea to use thread locals. What Flask is Not Flask will never have a database layer. Flask wants to make it quick and easy to write a traditional web application. The majority of web applications will need a template engine in some sort. Also see the Becoming Big section of the documentation for some inspiration for larger applications based on Flask. they support greenlet con- texts as well) for request. 260 . They cause troubles for servers that are not based on the concept of threads and make large applications harder to maintain. session and an extra object you can put your own things on (g). Everything else is up for extensions. Flask itself just bridges to Werkzeug to implement a proper WSGI application and to Jinja2 to handle templating. However Flask is just not designed for large applications or asynchronous servers. The idea of Flask is to build a good foundation for all applications. It will not have a form library or anything else in that direction.24.6 What Flask is. 24. It also binds to a few common standard library packages such as logging. However not every application needs a SQL database.

However. XHTML also changed the way JavaScript is used. The majority of users. While it is relatively easy to configure Web servers to serve XHTML properly. One of the most important causes of pain is XML’s draconian (strict and ruthless) er- ror handling. so that the HTML is cleaner and faster to load. There are a couple of major reasons why this is the case. but Internet Ex- plorer refuses to read files with that MIME type. There are XML based template en- gines. the browser is supposed to show the user an ugly error message. This is likely because properly using XHTML can be quite painful. To properly work with XHTML. such as Kid and the popular Genshi. One of them is Internet Explorer’s lack of proper XHTML support. but they often come with a larger runtime overhead and are not as straightforward to use because they have to obey XML rules. They wrote an XHTML doctype at the top of the document and self-closed all the necessary tags (<br> becomes <br/> or <br></br> in XHTML). when end tags are optional they are not used. Most of the (X)HTML generation on the web is based on non-XML template engines (such as Jinja. this document tries to answer some of the major ques- tions. The XHTML spec states that XHTML must be served with the MIME type application/xhtml+xml. 25. few people do. instead of attempting to recover from the error and display what it can. it appeared that HTML was about to be replaced by XHTML. However. even if the document properly validates as XHTML. So the valid XHTML was being treated as invalid HTML. however. barely any websites on the Internet are actual XHTML (which is HTML processed us- ing XML rules). When an XML parsing error is encountered. assumed they were properly using XHTML. which as said before is often not set properly. Because there is much confusion about HTML and XHTML among developers. CHAPTER TWENTYFIVE HTML/XHTML FAQ The Flask documentation and example applications are using HTML5.1 History of XHTML For a while. You may no- tice that in many situations. what really determines XHTML/HTML processing in browsers is the MIME type. pro- grammers have to use the namespaced DOM interface with the XHTML namespace 261 . the one used in Flask) which do not protect you from accidentally creating invalid XHTML.

01. Currently.0” by the Web Hypertext Application Technology Working Group.1 and HTML5. it appears that XHTML is losing traction.0 is not included. known as HTML5. since people were using XHTML-like tags along the lines of <link />. or WHATWG (which was formed by the major browser vendors Apple.) 262 . 25. (XHTML 1. the specification was adopted as the basis of a new HTML specification under the umbrella of the W3C.1 and the barely-used XHTML5. For example. browser vendors implemented the XHTML syntax over the syntax defined by the specification. as it was superseded by XHTML 1.3 HTML versus XHTML The following table gives you a quick overview of features available in HTML 4. Mozilla. and Opera) with the goal of writing a new and improved HTML specification. However.to query for HTML elements. In 2007. based on existing browser behavior instead of unrealistic and backwards-incompatible specifi- cations. as the XHTML 2 working group has been disbanded and HTML5 is being implemented by all major browser vendors. XHTML 1. in HTML4 <title/Hello/ theoretically parses exactly the same as <title>Hello</title>. 25.2 History of HTML5 Development of the HTML5 specification was started in 2004 under the name “Web Applications 1.

but it also specifies exactly how a browser should react to parsing errors .1 HTML5 <tag/value/ == <tag>value</tag> 1 <br/> supported 2 <script/> supported should be served as text/html 3 should be served as application/xhtml+xml strict error handling inline SVG inline MathML <video> tag <audio> tag New semantic tags like <article> 25. others are actually specified. • Requiring boolean attributes to have a value. >.01 XHTML1.0 is the last XHTML standard that allows to be served as text/html for backwards com- patibility reasons. <th>. which simply states parsing should abort. <tbody>.4 What does “strict” mean? HTML5 has strictly defined parsing rules.unlike XHTML. HTML4. so long as they contain no whitespace or special characters (like <. '. • Quoting attributes. missing end tags or unquoted attribute values). or "). Some people are confused by apparently invalid syntax that still generates the expected results (for example. It should not be used in new code. Some of these work because of the lenient error handling most browsers use when they encounter a markup error. for reasons detailed above. It is usually not supported by browsers. <td>. <dt>. 3 XHTML 1. 263 . <thead>. <dd>. This means the following page in HTML5 is perfectly valid: <!doctype html> <title>Hello HTML5</title> 1 This is an obscure feature inherited from SGML. 2 This is for compatibility with server code that generates XHTML for tags such as <br>. or <tfoot> tags. but have to be supported by browsers: • Wrapping the document in an <html> tag • Wrapping header elements in <head> or the body elements in <body> • Closing the <p>. The following constructs are optional in HTML5 by standard. <li>. <tr>.

Web Workers.<div class=header> <h1>Hello HTML5</h1> <p class=tagline>HTML5 is awesome </div> <ul class=nav> <li><a href=/index>Index</a> <li><a href=/downloads>Downloads</a> <li><a href=/about>About</a> </ul> <div class=body> <h2>HTML5 is probably the future</h2> <p> There might be some other things around but in terms of browser vendor support. Dive Into HTML5. Web Sockets. 264 .5 New technologies in HTML5 HTML5 adds many new features that make Web applications easier to write and to use. however. A good guide to new features in HTML5 is Mark Pilgrim’s soon-to-be-published book. • New form control types like <input type="date"> that allow user agents to make entering and validating values easier. Many other features have been added. and <time> that make con- tent easier to understand. geolo- cation. as well. reducing the need for server-generated images to present data graphically. <dl> <dt>Key 1 <dd>Value 1 <dt>Key 2 <dd>Value 2 </dl> </div> 25. and offline applications. <nav>. • The <canvas> tag. HTML5 is hard to beat. • Semantic elements like <article>. <header>. • The <audio> and <video> tags provide a way to embed audio and video without complicated add-ons like QuickTime or Flash. • Advanced JavaScript APIs like Web Storage. which supports a powerful drawing API. so use caution. Not all of them are supported in browsers yet.

• It is much easier to write. the answer is HTML5.25. • HTML5 adds several new features. • Many JavaScript libraries also do not support XHTML. • It has the support of most browser vendors behind it. currently leads in market share) has poor sup- port for XHTML. 265 . it is undoubtedly better to use HTML5 than XHTML. For most applications. There are very few reasons to use XHTML consider- ing the latest developments in Web browsers. including semantic tags and the long-awaited <audio> and <video> tags. To summarize the reasons given above: • Internet Explorer (which.6 What should be used? Currently. and more compact. sadly. due to the more compli- cated namespacing API it requires.

266 .

never do that. Another thing that is very important are unquoted attributes. To counter this possible attack vector. Flask tries to solve a few of these things for you. Some browsers are using content-type guessing based on the first few bytes so users could trick a browser to execute HTML. To remedy this. For example an attacker could inject this piece of HTML+JavaScript: 267 . there is one thing it cannot protect you from: XSS by attribute injection. • sending out textfiles from uploaded files. Flask configures Jinja2 to automatically escape all values unless explicitly told other- wise. CHAPTER TWENTYSIX SECURITY CONSIDERATIONS Web applications usually face all kinds of security problems and it’s very hard to get everything right. 26.1 Cross-Site Scripting (XSS) Cross site scripting is the concept of injecting arbitrary HTML (and with it JavaScript) into the context of a website. be sure to always quote your attributes with either double or single quotes when using Jinja expressions in them: <a href="{{ href }}">the text</a> Why is this necessary? Because if you would not be doing that. an attacker could easily inject custom JavaScript handlers. While Jinja2 can protect you from XSS issues by escaping HTML. developers have to properly escape text so that it cannot include arbitrary HTML tags. use the Content-Disposition: attachment header to prevent that problem. but there are still other places where you have to be careful: • generating HTML without the help of Jinja2 • calling Markup on data submitted by users • sending out HTML from uploaded files. For more information on that have a look at the Wikipedia article on Cross-Site Scripting. but there are a couple more you have to take care of yourself. This should rule out all XSS problems caused in templates.

268 . How can you prevent that? Basically for each request that modifies content on the server you would have to either use a one-time token and store that in the cookie and also transmit it with the form data. Unfortunately that includes requests triggered by 3rd party sites. After receiving the data on the server again. Why does Flask not do that for you? The ideal place for this to happen is the form validation framework. so only extremely old browsers are still vul- nerable. Say you have a specific URL that. The state of “being logged in” is controlled by a cookie. This was because of a security vulnerability in ECMAScript 4. some people might be able to trick your application’s users with social engineering to do stupid things without them knowing. If your authentication information is stored in cookies. which does not exist in Flask.onmouseover=alert(document. their profiles would get deleted while they are looking at images of fluffy cats. If you don’t keep that in mind. so this behavior was changed and jsonify() now supports serializing arrays.com/user/delete). you have implicit state man- agement.3 JSON Security In Flask 0. 26.2 Cross-Site Request Forgery (CSRF) Another big problem is CSRF. If an attacker now creates a page that sends a post request to that page with some JavaScript they just have to trick some users to load that page and their profiles will end up being deleted.10 and lower. and that cookie is sent with each request to a page. when you sent POST requests to will delete a user’s profile (say http://example. you would then have to compare the two tokens and ensure they are equal. Imagine you were to run Facebook with millions of concurrent users and someone would send out links to images of little kittens. In combination with CSS injections the attacker might even make the element fill out the entire page so that the user would just have to have the mouse anywhere on the page to trigger the attack.cookie) When the user would then move with the mouse over the link. This is a very complex topic and I won’t outline it here in detail just mention what it is and how to theoretically prevent it. the cookie would be presented to the user in an alert window. But instead of showing the cookie to the user. jsonify() did not serialize top-level arrays to JSON. 26. All of these browsers have other more serious vulnerabilities. ECMAScript 5 closed this vulnerability. When users would go to that page. a good attacker might also execute any other JavaScript code.

is totally Unicode based when it comes to text. you should probably read The Absolute Min- imum Every Software Developer Absolutely. However HTML which is usually trans- mitted on top of HTTP supports a large variety of character sets and which ones are used. Some databases have a protocol that already transmits Unicode and if they do not. To not make this too complex Flask just assumes that if you are sending Unicode out you want it to be UTF-8 encoded. are transmitted in an HTTP header. also the system used to address doc- uments on servers (so called URIs or URLs). • encoding and decoding happens whenever you are talking over a protocol that requires bytes to be transmitted. Not only the protocol. Positively Must Know About Unicode and Character Sets. 269 . So what does this mean to you? HTTP is based on bytes. Not only these libraries. This part of the documentation just tries to cover the very basics so that you have a pleasant experience with Unicode related things. If you don’t know Unicode so far. like Jinja2 and Werkzeug. SQLAlchemy or your other ORM should take care of that. Flask will do the encoding and setting of the appropriate headers for you. CHAPTER TWENTYSEVEN UNICODE IN FLASK Flask. The same is true if you are talking to databases with the help of SQLAlchemy or a similar ORM system. 27.1 Automatic Conversion Flask has a few assumptions about your application (which you can change of course) that give you basic and painless Unicode support: • the encoding for text on your website is UTF-8 • internally you will always use Unicode exclusively for text except for literal strings with only ASCII character points. also the majority of web related Python libraries that deal with text.

read(). encode() method: def write_file(filename.write(contents.coding: utf-8 -*. 'r') as f: return f. So make sure to tell your editor to save the file as UTF-8 there as well. charset='utf-8'): with open(filename.decode(charset) To go from Unicode into a specific charset such as UTF-8 you can use the unicode.x mean? • as long as you are using ASCII charpoints only (basically numbers.3 Encoding and Decoding Yourself If you are talking with a filesystem or something that is not really based on Unicode you will have to ensure that you decode properly when working with Unicode inter- face. 'w') as f: f. 27. • Jinja is configured to decode the template files from UTF-8. What does working with Unicode in Python 2.encode(charset)) 27.4 Configuring Editors Most editors save as UTF-8 by default nowadays but in case your editor is not config- ured to do this you have to change it. Here some common ways to set your editor to 270 . To load such a file with Unicode you can use the built-in str. contents. Anyways.27. some special characters of latin letters without umlauts or anything fancy) you can use regular string literals ('Hello World').2 The Golden Rule So the rule of thumb: if you are not dealing with binary data. • if you need anything else than ASCII in a string you have to mark this string as Unicode string by prefixing it with a lowercase u.decode() method: def read_file(filename. I recommend UTF-8 for this pur- pose.into the first or second line of your Python source file. charset='utf-8'): with open(filename. So for example if you want to load a file on the filesystem and embed it into a Jinja2 template you will have to decode it from the encoding of that file. (like u'Hänsel und Gretel') • if you are using non-Unicode characters in your Python files you have to tell Python which encoding your file uses. Again. work with Unicode. To tell the interpreter your encoding you can put the # -*. Here the old problem that text files do not specify their encoding comes into play. So do yourself a favour and limit yourself to UTF-8 for text files as well.

. • Emacs: either use an encoding cookie or put this into your .emacs file: (prefer-coding-system 'utf-8) (setq default-buffer-file-coding-system 'utf-8) • Notepad++: 1.vimrc file. 2.store as UTF-8: • Vim: put set enc=utf-8 to your . Select the “New Document/Default Directory” tab 3. 271 . Select “UTF-8 without BOM” as encoding It is also recommended to use the Unix newline format.. you can select it in the same panel but this is not a requirement. Go to Settings -> Preferences .

272 .

Because very often these steps could be abstracted to support multi- ple projects the Flask Extension Registry was created. So for example if you plan to add support for a library named simplexml to Flask. If you want to create your own Flask extension for something that does not exist yet.1 Anatomy of an Extension Extensions are all located in a package called flask_something where “something” is the name of the library you want to bridge. See Extension Import Transition for more details.something. The name of the actual extension (the human readable name) however would be some- thing like “Flask-SimpleXML”. This is done to transition from the old namespace packages. CHAPTER TWENTYEIGHT FLASK EXTENSION DEVELOPMENT Flask. Flask sets up a redirect package called flask. often requires some repetitive steps to get a third party library working.ext where users should import the ex- tensions from. you would name your extension’s package flask_simplexml. If you for instance have a package called flask_something users would import it as flask. 273 .py file and registered on PyPI. being a microframework. this guide to extension development will help you get your extension running in no time and to feel like users would expect your extension to behave. Because of that it is crucial that your application supports that kind of behavior. Also the development checkout link should work so that people can easily install the development version into their virtualenv without having to download the library by hand. Most importantly the extension must be shipped with a setup. This is a requirement because many people will use patterns like the Application Factories pattern to create their application as needed to aid unittests and to support multiple configurations.ext. 28. But what do extensions look like themselves? An extension has to ensure that it works with multiple Flask application instances at once.py files. This is how users can then register dependencies to your extension in their setup. Make sure to include the name “Flask” somewhere in that name and that you check the capitalization.

author='Your Name'.py file which is used to install your Flask extension. version='1. description='Very short description'. 274 . license='BSD'. Keep in mind that the Flask Extension Registry is a moderated place and libraries will be reviewed upfront if they behave as required.com'. The extension we want to create here will provide very basic support for SQLite3. # if you would be using a package instead use packages instead # of py_modules: # packages=['flask_sqlite3'].Flask extensions must be licensed under a BSD. long_description=__doc__.2. MIT or more liberal license to be able to be enlisted in the Flask Extension Registry. The following contents are something you can work with: """ Flask-SQLite3 ------------- This is the description for that library """ from setuptools import setup setup( name='Flask-SQLite3'. author_email='your-email@example. url='http://example. 28.py The next file that is absolutely required is the setup.py LICENSE README Here’s the contents of the most important files: 28. py_modules=['flask_sqlite3']. First we create the following folder structure: flask-sqlite3/ flask_sqlite3.0'. zip_safe=False.com/flask-sqlite3/'. include_package_data=True.1 setup.2 “Hello Flaskext!” So let’s get started with creating such a Flask extension.

For an example look at how the OAuth exten- sion works: there is an OAuth object that provides some helper functions like OAuth. For example. install_requires=[ 'Flask' ].2. But how exactly should such an exten- sion look like? What are the best practices? Continue reading for some insight. 'License :: OSI Approved :: BSD License'. 'Operating System :: OS Independent'. 28.py Now this is where your extension code goes. 275 . 'Programming Language :: Python'. classes: Classes work mostly like initialization functions but can later be used to further change the behavior.3 Initializing Extensions Many extensions will need some kind of initialization step. 'Topic :: Internet :: WWW/HTTP :: Dynamic Content'. classifiers=[ 'Environment :: Web Environment'.remote_app to create a reference to a remote application that uses OAuth. 28. It could attach before / after handlers etc. consider an application that’s currently connecting to SQLite like the documentation suggests (Us- ing SQLite 3 with Flask). There are two recommended ways for an extension to initialize: initialization functions: If your extension is called helloworld you might have a function called init_helloworld(app[. 'Intended Audience :: Developers'. 'Topic :: Software Development :: Libraries :: Python Modules' ] ) That’s a lot of code but you can really just copy/paste that from existing extensions and adapt.2 flask_sqlite3. So how does the extension know the name of the application object? Quite simple: you pass it to it. platforms='any'. extra_args]) that initializes the extension for that application.

config['SQLITE3_DATABASE']) def teardown(self. app): app.teardown) def connect(self): return sqlite3.setdefault('SQLITE3_DATABASE'. # Starting with Flask 0.top if hasattr(ctx. # otherwise fall back to the request context if hasattr(app.9. For the SQLite 3 extension we will use the class-based approach because it will provide users with an object that handles opening and closing database connections.teardown) else: app.py for copy/paste: import sqlite3 from flask import current_app # Find the stack on which we want to store the database connection.config. app=None): self. # before that we need to use the _request_ctx_stack.What to use depends on what you have in mind.teardown_request(self.sqlite3_db.app = app if app is not None: self.teardown_appcontext(self.connect(current_app. 'teardown_appcontext'): app.init_app(app) def init_app(self. 28. ':memory:') # Use the newstyle teardown_appcontext if it's available. What’s important about classes is that they encourage to be shared around on mod- ule level. exception): ctx = stack. try: from flask import _app_ctx_stack as stack except ImportError: from flask import _request_ctx_stack as stack class SQLite3(object): def __init__(self. the _app_ctx_stack is the correct one.close() @property 276 .4 The Extension Code Here’s the contents of the flask_sqlite3. In that case. the object itself must not under any circumstances store any application specific state and must be shareable between different application. 'sqlite3_db'): ctx.

sqlite3_db = self. Next. we add a connection property that on first access opens the database connection and stores it on the context. falls back to the request context one.top if the application is using an older version of Flask that does not support it.sqlite3_db So here’s what these lines of code do: 1.top. de- faulting to an in memory database if no configuration is supplied. if supplied. 'sqlite3_db'): ctx. 3. This method supports the factory pattern for cre- ating applications.connect() return ctx. Note that we’re falling back to the _request_ctx_stack. we define a connect method that opens a database connection. The init_app method exists so that the SQLite3 object can be instantiated with- out requiring an app object. you can use the database in the same way: 277 . the init_app method attaches the teardown handler. This is also the recommended way to handling resources: fetch resources lazily the first time they are used. So why did we decide on a class-based approach here? Because using our extension looks something like this: from flask import Flask from flask_sqlite3 import SQLite3 app = Flask(__name__) app. 2. 4. will call init_app.) Likewise if you are outside of a request but you are using Flask 0. The __init__ method takes an optional app object and. def connection(self): ctx = stack.cursor() cur.config.. The init_app will set the configuration for the database.cfg') db = SQLite3(app) You can then use the database from views like this: @app. Finally. Note here that we’re attaching our database connection to the top application context via _app_ctx_stack.execute(. In addition.connection. It will try to use the new- style app context handler and if it does not exist. Extensions should use the top context for stor- ing their own information with a sufficiently complex name.route('/') def show_all(): cur = db.from_pyfile('the-config.top if ctx is not None: if not hasattr(ctx.9 or later with the app context support..

app_context(): cur = db.cursor() cur.init_app(app) Keep in mind that supporting this factory pattern for creating apps is required for approved flask extensions (described below).connection. it is recommended to fall back to _request_ctx_stack which is bound to a request. 28. init_app does not assign app to self. During the teardown of a request. 28. When the extension needs to find the current application and it does not have a refer- ence to it. app = create_app('the-config. The following pattern is a good way to support both: 278 .top. This is intentional! Class based Flask extensions must only store the application on the object when the application was passed to the constructor. this variable is accessible using the connection property of SQLite3.) At the end of the with block the teardown handles will be executed automatically.cfg') db.execute(.. In a view function. Additionally. the same connection to the sqlite3 database is accessible to anything that needs it for the duration of the request. By using this pattern.6 and older Due to the change in Flask 0. This tells the extension: I am not interested in using multiple applications. the sqlite3_db connection is closed. before every request.6 Teardown Behavior This is only relevant if you want to support Flask 0.with app.5 Using _app_ctx_stack In the example above. the init_app method is used to support the factory pattern for creating apps: db = Sqlite3() # Then later on. Note on init_app As you noticed. a sqlite3_db variable is assigned to _app_ctx_stack. it must either use the current_app context local or change the API in a way that you can pass the application explicitly. If the _app_ctx_stack does not exist because the user uses an old version of Flask..7 regarding functions that are run at the end of the request your extension will have to be extra careful there if it wants to continue to support older versions of Flask.

teardown_request(close_connection) else: app. Especially if you do something nobody before you did. The best Flask extensions are extensions that share common idioms for the API. 1. Remember: good API design is hard. because teardown functions are passed the exception and typically don’t return anything. In the event an extension author would like to move beyond the project. the project should find a new maintainer including full source hosting transition and PyPI access. give access to the Flask core team.def close_connection(response): ctx = _request_ctx_stack.close() return response if hasattr(app. However because the return value is discarded this will just work assuming that the code in between does not touch the passed parameter. 'teardown_request'): app.7 Learn from Others This documentation only touches the bare minimum for extension development. 28. These approved extensions are listed on the Flask Extension Registry and marked appropriately. Approved extensions are tested as part of Flask itself to ensure extensions do not break on new releases. If no main- tainer is available.top ctx. 279 . If you want to learn more. but also to avoid having multiple developers working on pretty much the same side by side.after_request(close_connection) Strictly speaking the above code is wrong. And this can only work if collaboration happens early. This not only to get an idea about what people might want to have from an extension. An approved Flask extension requires a maintainer. If you want your own extension to be approved you have to follow these guidelines: 0. and let other developers give you a helping hand with designing the API.8 Approved Extensions Flask also has the concept of approved extensions. it’s a very good idea to check out existing extensions on the Flask Extension Registry. An approved Flask extension must provide exactly one package or module named flask_extensionname. If you feel lost there is still the mailinglist and the IRC chan- nel to get some ideas for nice looking APIs.sqlite3_db. 28. so introduce your project on the mailinglist. it might be a very good idea to get some more input.

py test. The test suite also has to be part of the distribution. Approved extensions must define all their dependencies in the setup.11. 5. An extension currently has to support Python 2. 9. This turned out to be problematic in practice be- cause it meant that multiple flaskext packages coexist. For test suites invoked with make test the extension has to ensure that all dependencies for the test are installed automatically.foo.ext.foo compatibility alias is still in Flask 0.foo would try flask_foo and flaskext. 6.ext.11 but is now deprecated – you should use flask_foo. even if the extension would be safe for zipping. As of Flask 0.7 28.6 as well as Python 2. The flask. The naming scheme for official extensions is Flask-ExtensionName or ExtensionName-Flask. The extension must have documentation that uses one of the two Flask themes for Sphinx documentation.9 Extension Import Transition In early versions of Flask we recommended using namespace packages for Flask ex- tensions. The setup.py file un- less a dependency cannot be met because it is not available on PyPI. of the form flaskext. 10. The zip_safe flag in the setup script must be set to False.8 introduced a redirect import system as a compatibility aid for app develop- ers: Importing flask. Consequently we have rec- ommended to name extensions flask_foo over flaskext. Flask 0. If tests are invoked with python setup. 3. • it must be possible to use the factory pattern for creating applications. website (if there is one) and there must be a link to automatically install the development version (PackageName==dev). 7. 280 .py test.foo for a long time. The license must be BSD/MIT/WTFPL licensed. 8.foo in that order. It must ship a testing suite that can either be invoked with make test or python setup. 4. most Flask extensions have transitioned to the new naming schema. APIs of approved extensions will be checked for the following characteristics: • an approved extension has to support multiple applications running in the same Python process.py file.py description (and thus the PyPI description) has to link to the doc- umentation. 2. test dependencies can be specified in the setup.

42.scalar > 120) \ . align to the braces: this_is_a_very_long(function_call.filter(MyModel. Try to avoid too nested code by cleverly placing break. Maximum line length: 79 characters with a soft limit for 84 if absolutely necessary. continue and return state- ments. In general the Pocoo Styleguide closely follows PEP 8 with some small differences and extensions. no exceptions.query.order_by(MyModel. 'set of items'. 'like this' ] 281 .that_returns_an_object_with_an_attribute MyModel.desc()) \ . or indent four spaces: this_is_a_very_long(function_call.name. 'with many parameters'. including Flask. This styleguide is a requirement for Patches to Flask and a recommendation for Flask ex- tensions.1 General Layout Indentation: 4 real spaces. No tabs. break immediately after the opening brace: items = [ 'this is the first'.limit(10) If you break in a statement with parentheses or braces. CHAPTER TWENTYNINE POCOO STYLEGUIDE The Pocoo styleguide is the styleguide for all Pocoo Projects. 'with more items'. 'with many parameters') \ . Continuing long statements: To continue a statement you can use backslashes in which case you should align the next line with the last dot or equal sign. 'to come in this line'. 'and even more') For lists or tuples with many items. 23. 29.

: -.05 value = (item_value / item_count) * offset / exp value = my_list[index] value = my_dict['key'] Bad: exp = . 1. Example: def hello(name): print 'Hello %s!' % name def goodbye(name): print 'See you %s. • Whitespace is placed between binary operators.2 Expressions and Statements General whitespace rules: • No whitespace for unary operators that are not words (e. name): self. everything else by one.g.name.05 value = ( item_value / item_count ) * offset / exp value = (item_value/item_count)*offset/exp value=( item_value/item_count ) * offset/exp value = my_list[ index ] value = my_dict ['key'] Yoda statements are a no-go: Never compare constant with variable.) as well on the inner side of parentheses. Good: exp = -1. Do not use too many blank lines to separate logical segments in code. always variable with constant: 282 .name = name def get_annoying_name(self): return self.upper() + '!!!!111' 29.' % name class MyClass(object): """This is a simple docstring""" def __init__(self.Blank lines: Top level functions and classes are separated by two lines. ~ etc.

Double underscores are reserved for mixin classes. Function and method arguments: • class methods: cls as first parameter • instance methods: self as first parameter • lambdas for properties might have the first parameter replaced with x like in display_name = property(lambda x: x. do not foo instead) Negated containment checks: use foo not in bar instead of not foo in bar Instance checks: isinstance(a. with acronyms kept uppercase (HTTPWriter and not HttpWriter) • Variable names: lowercase_with_underscores • Method and function names: lowercase_with_underscores • Constants: UPPERCASE_WITH_UNDERSCORES • precompiled regular expressions: name_re Protected members are prefixed with a single underscore.real_name or x.username) 283 . If the function needs to access a shadowed builtin. but try to avoid instance checks in general. Clashes with builtins are allowed and must not be resolved by appending an underline to the variable name. rebind the builtin to a different name instead. Good: if method == 'md5': pass Bad: if 'md5' == method: pass Comparisons: • against arbitrary types: == and != • against singletons with is and is not (eg: foo is not None) • never compare something with True or False (for example never do foo == False. trailing underscores are appended. C) instead of type(A) is C. 29. On classes with keywords. Check for features.3 Naming Conventions • Class names: CamelCase.

:copyright: (c) YEAR by AUTHOR.module ~~~~~~~~~~~~~~ A brief description goes here.coding: utf-8 -*- """ package. """ Module header: The module header consists of an utf-8 encoding declaration (if non ASCII letters are used. Both are formatted with reStructured- Text. otherwise the text is on the same line as the opening quote and the triple quote that closes the string on its own line: def foo(): """This is a simple docstring""" def bar(): """This is a longer docstring with so much information in there that it spans three lines. Depending on the number of lines in the docstring. If a comment is used to document an attribute. put a colon after the opening pound sign (#): class User(object): #: the name of the user as unicode string name = Column(String) #: the sha1 hash of the password + inline salt pw_hash = Column(String) 284 .29. In this case the closing triple quote is on its own line. If it’s just one line. but it is recommended all the time) and a standard doc- string: # -*. the closing triple quote is on the same line as the opening. 29.5 Comments Rules for comments are similar to docstrings. """ Please keep in mind that proper copyrights and license files are a requirement for approved Flask extensions. see LICENSE_FILE for more details.4 Docstrings Docstring conventions: All docstrings are formatted with reStructuredText as under- stood by Sphinx. :license: LICENSE_NAME. they are laid out differently.

2 and older are not supported. The majority of the upgrade work is in the lower-level libraries like Flask and Werkzeug. You should start using Python 3 for your next project. Werkzeug wraps that information in high-level helpers. 285 . This mainly affects WSGI middleware interacting with the WSGI environ data. all of the examples in the Flask repository work on both Python 2 and 3 and did not require a single line of code changed. 3. which complicates how low level code handles HTTP data. its dependencies. You need to use Python 3. and most Flask extensions support Python 3.9 were the first versions to introduce Python 3 support. CHAPTER THIRTY PYTHON 3 SUPPORT Flask. You should use the latest versions of all Flask-related packages.3 or higher.10 and Werkzeug 0. but there are a few things to be aware of. Flask 0. Python 3 changed how unicode and bytes are handled. not the high-level application code. so encoding issues should not affect you. For example.

286 .

1 Changes to send_file The filename is no longer automatically inferred from file-like objects.1.12 31. However every once in a while there are changes that do require some changes in your code or there are changes that make it possible for you to improve your own code quality by taking advantage of new features in Flask.txt')) Any of the following is functionally equivalent: fname = '/path/to/file. This means that the following code will no longer automatically have X-Sendfile support.) 287 . This section of the documentation enumerates all the changes in Flask from release to release and how you can change your code to have a painless updating experience. etag generation or MIME-type guessing: response = send_file(open('/path/to/file.txt' # Just pass the filepath directly response = send_file(fname) # Set the MIME-type and ETag explicitly response = send_file(open(fname).1 Version 0.. the kind where you don’t have to change anything in your code to profit from a new release.set_etag(. Use the pip command to upgrade your existing Flask installation by providing the --upgrade parameter: $ pip install --upgrade Flask 31.. Most of the changes are the nice kind. CHAPTER THIRTYONE UPGRADING TO NEWER RELEASES Flask itself is changing like any software is changing over time. mimetype='text/plain') response.

In case you did track the master branch you will notice that flask --app is removed now. Instead the new LOGGER_HANDLER_POLICY configuration can be used to disable the default log handlers and custom log handlers can be set up. and therefore was replaced with the intended behavior of registering handlers only using exception classes and HTTP error codes. Now the inheritance hierarchy takes precedence and handlers for more specific exception classes are executed instead of more general ones.1 Debugging Flask 0.11 release first with some changes removed to make the transition easier. Silently swallowing errors in such cases was not a satisfying solution.) The reason for this is that some file-like objects have a invalid or even misleading name attribute.. See Error handlers for specifics. Note: There used to be a logic error allowing you to register handlers only for excep- tion instances.0 release. 31. You need to use the environment variable to specify an application. Additionally the default of falling back to application/octet-stream has been re- stricted. respectively.set_etag(.2. 288 .2. The precedence of handlers used to be based on the decoration/call order of errorhandler() and register_error_handler(). Trying to register a handler on an instance now raises ValueError. If Flask can’t guess one or the user didn’t provide one.# Set `attachment_filename` for MIME-type guessing # ETag still needs to be manually set response = send_file(open(fname).11 0.0 you might see some unexpected changes.2 Version 0.2 Error handling The behavior of error handlers was changed. 31.11 is an odd release in the Flask release cycle because it was supposed to be the 1. attachment_filename=fname) response. However because there was such a long lead time up to the release we decided to push out a 0. the function fails if no filename information was provided.. If you have been tracking the master branch which was 1. This was unintended and plain wrong. 31.11 removed the debug_log_format attribute from Flask applications.

3 Version 0.9 to 0. This change has been done in order to avoid the damage an attacker can do if the secret key is leaked.app_ctx_globals_class. In order to not break people’s sessions it is possible to continue using the old session system by using the Flask-OldSessions extension. rv) 289 . headers) where at least one item has to be provided. This change should be transparent for you but it means that you now can store things on the g object when there is no request context yet but an applica- tion context.foo are deprecated.4 Version 0. Flask also started storing the flask.2.10 is that the cookie serialization format changed from pickle to a specialized JSON format.10 The biggest change going from 0. We also provide a migration utility called flask-ext-migrate that is supposed to automatically rewrite your imports for this.ext.response_class(*rv) return Flask. This better matches the behavior of render_template(). but Flask will issue a flask.Flask.exthook. status.3 Templating The render_template_string() function has changed to autoescape template vari- ables by default.4 Extension imports Extension imports of the form flask. you can add it easily by subclassing Flask: class TraditionalFlask(Flask): def make_response(self. If you return a tuple it no longer defines the arguments for the response object you’re creating.ExtDeprecationWarning for each extension you import the old way.2. 31. it’s now always a tuple in the form (response. rv): if isinstance(rv.31.make_response(self. you should use flask_foo.Flask.g object on the application context instead of the request context. The new sessions are by design much more restricted to only allow JSON with a few small extensions for tuples and strings with HTML markup. The old form still works. 31. When you upgrade you will notice two major changes: all sessions that were issued before the upgrade are invalidated and you can only store a limited amount of types in the session.request_globals_class attribute was renamed to flask. 31. If you depend on the old behavior. tuple): return self.9 The behavior of returning tuples from a function was simplified. The old flask.

31. Due to a bug in the test client Flask 0.8 Flask introduced a new session interface system. Using the app context stack instead of the request context stack will make extensions more readily handle use cases outside of requests. please con- sider changing to _app_ctx_stack if it makes sense for your extension. To apply the upgrade script do the following: 290 . If you were catching this down explicitly in the past as ValueError you will need to change this. With that introduction we moved the implementation details for the session system into a new module called flask. the app context stack makes sense for extensions which connect to databases. If invalid JSON data was submitted Flask will now raise a BadRequest exception in- stead of letting the default ValueError bubble up.diff and then using the interactive commit feature to only apply the chunks that look good. If you used the previously undocumented session support we urge you to upgrade. For instance. Because we want to make upgrading as easy as possible we tried to counter the prob- lems arising from these changes by providing a script that can ease the transition.7 did not trigger teardown handlers when the test client was used in a with statement.session the module that implements sessions and flask. We strongly recommend that you hand review the generated patchfile and only apply the chunks that look good. This has the advantage that you no longer have to handle that error to avoid an internal server error showing up for the user. We also noticed that there was a naming collision between flask. The script scans your whole application and generates an unified diff with changes it assumes are safe to apply.6 Version 0.7 we cleaned up the code base internally a lot and did some backwards in- compatible changes that make it easier to implement larger applications with Flask. However as this is an automated tool it won’t be able to find all use cases and it might miss some.session which is the global session object. 31. sessions. This was since fixed but might require some changes in your test suites if you relied on this behavior. We internally spread a lot of deprecation warnings all over the place to make it easy to find pieces of code that it was unable to upgrade. If you are using git as version control system for your project we recommend applying the patch with path -p1 < patchfile.5 Version 0.7 In Flask 0.If you maintain an extension that was using _request_ctx_stack before.

html. Also source control systems like svn. 1. make sure to update your application to work with either filenames there or disable etag attaching and attach them yourself.diff 3.7. Previously if you had a folder named templates next to a blueprint named admin the implicit template path automatically was admin/index.diff 5. msysgit or ming32. In order to see the deprecation warnings that might be emitted you have to enabled them with the warnings module.11. If you get a deprecation warning. If you were using per-module template folders you need to move some templates around. Review the generated patchfile.html. Download the script: flask-07-upgrade. The tool will not detect this so you will have to do that on your own. Now you need to name the template templates/admin/index. If you are working with windows and you lack the patch command line utility you can get it as part of various Unix runtime environments for windows including cygwin.py 2. Check the manual of your version control system for more information. Run it in the directory of your application: python flask-07-upgrade. Previously it was possible to use etags and mimetypes when file objects were passed.html for a template file called templates/index. This no longer is the case. hg or git have builtin support for applying unified diffs as generated by the tool.py > patchfile. Please note that deprecation warnings are disabled by default starting with Python 2. 4. This was unreliable and caused issues for a few setups.1 Bug in Request Locals Due to a bug in earlier implementations the request local proxies now raise a RuntimeError instead of an AttributeError when they are unbound. If you caught these exceptions with AttributeError before. Old code: return send_file(my_file_object) return send_file(my_file_object) New code: 291 .6. Apply the patch: patch -p1 < patchfile. Additionally the send_file() function is now issuing deprecation warnings if you depend on functionality that will be removed in Flask 0. 31. you should catch them with RuntimeError now.

You can take advantage of the fact that the teardown callbacks are called when the re- sponse context is removed from the stack so a test can query the database after request handling: with app.close() return response You are now encouraged to use this instead: @app.3 Manual Error Handler Attaching While it is still possible to attach error handlers to Flask.db is still bound if there is such a thing # and here it's gone 31.errorhandler() for more information.get('/') # g. In general we no longer recommend custom error handler attaching via assignments to the underlying dictionary due to the more complex internal handling to support arbitrary exception classes and blueprints. but for things that absolutely must happen at the end of request we introduced the new teardown_request() decorator.error_handlers it’s discour- aged to do so and in fact deprecated. For things that modify the response the after_request() decorators continue to work as expected. See Flask.teardown_request def after_request(exception): if hasattr(g.2 Upgrading to new Teardown Handling We streamlined the behavior of the callbacks for request handling.6. 'db'): g.db. add_etags=False) 31.return send_file(my_file_object. Unfortunately that change also made after-request work differently under error conditions. It’s not consistently skipped if exceptions happen whereas previously it might have been called twice to ensure it is executed at the end of the request. This makes it now a lot easier to write unit tests as you can prevent closing down of database connections for a while.after_request def after_request(response): g.6. 292 .test_client() as client: resp = client. If you have database connection code that looks like this: @app.db.close() On the upside this change greatly improves the internal code flow and makes it easier to customize the dispatching and error handling.

Flask will restore the previous behavior as good as possible.register_error_handler(403. Now the blueprints can provide template folders which are added to a general template searchpath.7) 31. 293 . The update script provided should be able to upgrade your applications automatically. Now it means “relative to current module”..html as the template name. The upgrade script tries to guess that name but it might fail as this information could change at runtime. If you continue to use the Module object which is deprecated. They will also no longer automatically export templates from a folder called templates next to their loca- tion however but it can be enabled from the constructor. They provide better semantics for various features and work better with large applications. blueprint specific error handlers and a lot more. It will do this in a very eager way so you might end up with some unnecessary leading dots in your code if you’re not using modules. • Blueprints have an inverse behavior for url_for().4 Blueprint Support Blueprints replace the previous concept of “Modules” in Flask.error_handlers[403] = handle_error Into this: app. handle_error) Alternatively you should just attach the function with a decorator: @app.. • Blueprints do not automatically provide static folders. Modules had an automatic name guessing scheme where the shortname for the module was taken from the last part of the import module. What changed? • Blueprints need explicit names.errorhandler(403) def handle_error(e): . Previously .6. but there might be some cases where it fails to upgrade.foo told url_for() that it should look for the endpoint foo on the application. This means that you need to add another subfolder with the blueprint’s name into that folder if you want blueprintname/template. The script will inverse all calls to url_for() automatically for you. (Note that register_error_handler() is new in Flask 0. However we strongly recommend upgrading to the new blueprints as they provide a lot of useful improvement such as the ability to attach a blueprint multiple times. • Rendering templates was simplified. Same with static files: if you want to continue serving static files you need to tell the constructor explic- itly the path to the static folder (which can be relative to the blueprint’s module path).The proper upgrade is to change this: app.

Instead it is configured to only happen on files ending with . If you want to customize the Jinja loader now. There were a couple of internal refactoring so if you depend on undocumented internal details you probably have to adapt the imports.html.9 Version 0. the default context processor will not override it with the current request object.6 Flask 0.4 For application developers there are no changes that require changes in your code.htm. use the create_jinja_environment() method instead.5 is the first release that comes as a Python package instead of a single mod- ule.6 comes with a backwards incompatible change which affects the order of after- request handlers. 31. . Removing support for this makes the Flask internal code easier to understand and fixes a couple of small issues that make debugging harder than necessary. If you have templates with different extensions you should override the select_jinja_autoescape() method. The following changes may be relevant to your application: • autoescaping no longer happens for all templates.5 Flask 0.7 Version 0.xml and . • Flask no longer supports zipped applications in this release. This change was made so that Flask behaves more like people expected it to work and how other systems handle request pre.8 Version 0. be sure to change the order. • The create_jinja_loader function is gone. now they are called in reverse order. 294 . If you depend on the order of execution of post-request functions. If for example request is as variable passed directly to the template. Another change that breaks backwards compatibility is that context processors will no longer override values passed directly to the template rendering function.and post- processing.xhtml. and that extension has a unittest-mode you might want to link the activation of that mode to the new TESTING flag. In case you are developing on a Flask extension however. This makes it easier to extend context processors later to inject additional variables without breaking existing template not expecting them. 31. This functionality might come back in future releases if there is demand for this feature. Previously they were called in the order of the registration.31. .

3 Categories for Flash Messages Flash messages can now have categories attached. For more information on that.10. 31.secret_key = SECRET_KEY You no longer have to do that. If you previously had code like this: app. All these are features that are 100% backwards compatible but you might want to take advantage of them. read Application Errors.10.debug = DEBUG app. Also you can start logging warnings and errors into the logger when appropriately.3 introduces configuration support and logging as well as categories for flash- ing messages. Read all about that in the Message Flashing pattern.10. This makes it possible to render errors. instead you can just load a configuration into the config object. How this works is outlined in Configuration Handling. (Which most likely is the case for any application out there). 295 .1 Configuration Support The configuration support makes it easier to write any kind of application that requires some sort of configuration. If you run your application in production and want to profit from automatic error logging.31. This is an opt-in feature because it requires some rethinking in the code.3 Flask 0. you might be interested in attaching a proper log handler. warnings or regular messages differently for example. 31.2 Logging Integration Flask now configures a logger for you with some basic and useful defaults.10 Version 0. 31.

296 .

12 Released on December 21st 2016.safe_join able to join multiple paths like os. unreleased • Fix crash when running under PyPy3. • Revert a behavior change that made the dev server crash instead of returning a Internal Server Error (pull request #2006). 32. • Disable logger propagation by default for the app logger. instead of per client.path. see pull request #1814. which can now be di- rectly set. See pull request #1988. See pull request #1849.2 Version 0.2 Bugfix release.1 Version 0.get. 297 . codename Punsch. • Mimetype guessing in send_file now fails loudly and doesn’t fall back to application/octet-stream. • app.join (pull re- quest #1730). • the cli command now responds to –version.test_client includes preset default environment. • Make flask. • Add support for range requests in send_file. • Correctly invoke response handlers for both regular request dispatching as well as error handlers. • Mimetype guessing and ETag generation for file-like objects in send_file has been removed. as per issue #104.11. CHAPTER THIRTYTWO FLASK CHANGELOG Here you can see the full list of changes between each Flask release. 32.

thereby allow- ing catching exceptions that are subclasses of HTTP exceptions (in werkzeug.test_client_class.3’s namespace loader.Config.config_class. • Added a workaround for a limitation in Python 3.32. See JSON Security for details. headers) from a view function. • Templates are no longer automatically reloaded outside of debug mode. If set to True a permanent session will be refreshed each request and get their lifetime extended. • Added SESSION_REFRESH_EACH_REQUEST config key that controls the set-cookie behavior. • Added support to serializing top-level arrays to flask. This intro- duces a security risk in ancient browsers.Flask. • Added flask. • Added flask and the flask.jsonify().cli module to start the local debug server through the click CLI system. codename Absinthe. • Added support for returning tuples in the form (response.run() method as it works faster and more reliable due to a different design and also replaces Flask-Script.Config.11.4 Version 0.3 Version 0. if set to False it will only be modified if the session actually modifies. • Error handlers that match specific classes are now checked first.test_client() to support passing additional keyword arguments to the constructor of flask.get_namespace(). • Fixed a bug that prevented FLASK_APP=foobar/__init__.Flask. • Added flask. This makes it possible for an extension author to create exceptions 298 . • Added flask. exceptions). • Added support for explicit root paths when using Python 3. • Added before_render_template signal. See pull request #1872. • Made Flask support custom JSON mimetypes for incoming data. 32.py from working. Non permanent sessions are not affected by this and will al- ways expire if the browser window closes. This can be configured with the new TEMPLATES_AUTO_RELOAD config key.Test. • Added **kwargs to flask.1 Bugfix release. released on June 7th 2016. This is recommended over the old flask.3’s namespace pack- ages.11 Released on May 29th 2016.from_json().

from_mapping(). • Allow custom Jinja environment subclasses (pull request #1422). • Added flask.templating.ext is now deprecated (pull request #1484). • Flask will now log by default even if debug is disabled.json. • flask. The log format is now hardcoded but the default log handling can be disabled through the LOGGER_HANDLER_POLICY configuration key.g now has pop() and setdefault methods.jsonify now supports the datetime. • send_from_directory now raises BadRequest if the filename is invalid on the server OS (pull request #1763).render_template_string by default (pull request #1515).get_json(). • Add “pretty” and “compressed” separators definitions in jsonify() method. that will by default result in the HTTP error of their choosing. • flask. • JSON responses are now terminated with a newline character.test.Config. See https://github. • Added the JSONIFY_MIMETYPE configuration variable (pull request #1728).date type (pull request #1326). • Don’t leak exception info of already catched exceptions to context teardown han- dlers (pull request #1393). This should help users debug when the wrong templates are loaded. • Deprecated request. Re- duces JSON response size when JSONIFY_PRETTYPRINT_REGULAR=False by removing unnecessary white space included by default after separators. • Exceptions during teardown handling will no longer leave bad application con- texts lingering around. • flask.com/pallets/flask/pull/ 1262 – this came up originally as a part of https://github. • Ported test suite to py. • Turn on autoescape for flask. but may be caught with a custom error handler if desired. because it is a convention that UNIX text files end with a newline and some clients don’t deal well when this newline is missing. • Enforce blueprint handling in the order they were registered for template load- ing.com/kennethreitz/ httpbin/issues/168 • The automatically provided OPTIONS method is now correctly disabled if the user registered an overriding rule with the lowercase-version options (issue #1288). 299 .json in favour of request. • Removed deprecated module functionality. • Added the EXPLAIN_TEMPLATE_LOADING config flag which when enabled will in- struct Flask to explain how it locates templates.

• Added support for byte strings back to the session system. Now it’s possible to use that filter in single quoted attributes. released on June 14th 2013) • Fixed an issue where |tojson was not quoting single quotes which made the filter not work properly in HTML attributes. 32.5 Version 0.find_package() with a useful mes- sage explaining why it is raised when a PEP 302 import hook is used without an is_package() method. • Fixed Python 3 bug when a handler from app.7 Version 0. • Changed logic of before first request handlers to flip the flag after invoking. • Made @before_first_request into a decorator as intended. • Raise an AttributeError in flask. • Fixed send_from_directory not expanding to the application root path correctly. • Fixed an issue where registering the same method twice for the same endpoint would trigger an exception incorrectly. 300 .10.2 (bugfix release. • Fixed an issue causing exceptions raised before entering a request or app context to be passed to teardown handlers. codename Limoncello.url_build_error_handlers reraises the BuildError. This should make using that filter with angular.32.10.6 Version 0. This will allow some uses that are potentially dangerous but should probably be per- mitted. 32.helpers.1 (bugfix release. • Fixed an issue with query parameters getting removed from requests in the test client when absolute URLs were requested. This broke compati- bility with the common case of people putting binary data for token verification into the session.js easier. • Fixed an etags bug when sending a file streams with a name. release date to be announced) • Fixed broken test_appcontext_signals() test case.10 Released on June 13th 2013.

See Version 0. This is not to be confused with flask. 301 .Flask.10.10 for more infor- mation.dumps to return bytestrings by default. • flask.sessions instead to get the session module.json. • The config object is now available to the template as a real global and not through a context processor which makes it available even in imported templates by de- fault. session and g are now also added as proxies to the template context which makes them available in imported templates.• Changed default cookie serialization format from pickle to JSON to limit the impact an attacker can do if the secret key leaks.g is now stored on the app context instead of the request context.g now gained a get() method for not erroring out on non existing items. • Removed deprecated internal flask. It’s disabled by default to not cause confusion with existing libraries that might expect flask. • Set the content-length header for x-sendfile. The behavior is slightly different as the request.Flask. • flask. One has to be very careful with those though because usage outside of macros might cause caching.g now can be used with the in operator to see what’s defined and it now is iterable and will yield all attributes stored. This allows much easier customization of how JSON is handled by Flask or any Flask extension. • Templates can now be rendered without request context. • flask.session module alias. session and g objects will not be available and blueprint’s context processors are not called. • Flask will now raise an error if you attempt to register a new function on an already used endpoint. app_ctx_globals_class which is a better name to what it does since 0. • request. Use flask. • Added template_test methods in addition to the already existing template_filter method family. This was allowed due to the different escaping behavior. • tojson used in templates is now safe by default due.session the session proxy. • Added wrapper module around simplejson and added default serialization of datetime objects. • flask.request_globals_class got renamed to flask. • tojson filter now does not escape script blocks in HTML5 parsers. • Added template_global methods in addition to the already existing template_filter method family. • Added an option to generate non-ascii encoded JSON which should result in less bytes being transmitted over the network.

• Added the JSONIFY_PRETTYPRINT_REGULAR configuration variable. • Removed custom JSON HTTP exception subclasses.json prop- erty. Using them however is strongly discouraged as the interface was flawed.3 port.Request.9 Released on July 1st 2012. • The builtin run method now takes the SERVER_NAME into account when picking the default port to run on.url_for() function now can also explicitly generate URL rules spe- cific to a given HTTP method. • Added a workaround for chrome’s cookies in localhost not working as intended with domain names. • Flask will no longer invoke the wrong error handlers if a proxy exception is passed through.8 Version 0.url_for() function now can generate anchors to the generated links. codename Campari. • Added flask. • The flask. • Added support for copying of request contexts for better working with greenlets. • Python requirements changed: requiring Python 2.get_json() as a replacement for the old flask. • Added message_flashed signal that simplifies flashing testing. • Changed logic for picking defaults for cookie values from sessions to work better with Google Chrome. This is now more reliable in case something handles an exception halfway through the error handling process.on_json_loading_failed() now returns a JSON formatted response by default.request. • The flask. If you were relying on them you can reintroduce them again yourself trivially. • Added appcontext_pushed and appcontext_popped signals. • The flask. • Logger now only returns the debug log setting if it was not set explicitly. • Changed how the teardown system is informed about exceptions. • Request context preservation in debug mode now keeps the exception informa- tion around which means that teardown handlers are able to distinguish error from success cases.7 now to prepare for Python 3. 32. 302 . • Flask now orders JSON keys by default to not trash HTTP caches due to different hash seeds between different workers.request.6 or 2.

• Unregister a circular dependency between the WSGI environment and the re-
quest object when shutting down the request. This means that environ werkzeug.
request will be None after the response was returned to the WSGI server but has
the advantage that the garbage collector is not needed on CPython to tear down
the request unless the user created circular dependencies themselves.
• Session is now stored after callbacks so that if the session payload is stored in the
session you can still modify it in an after request callback.
• The flask.Flask class will avoid importing the provided import name if it can
(the required first parameter), to benefit tools which build Flask instances pro-
grammatically. The Flask class will fall back to using import on systems with
custom module hooks, e.g. Google App Engine, or when the import name is
inside a zip archive (usually a .egg) prior to Python 2.7.
• Blueprints now have a decorator to add custom template filters application wide,
• The Flask and Blueprint classes now have a non-decorator method for adding
custom template filters application wide, flask.Flask.add_template_filter()
and flask.Blueprint.add_app_template_filter().
• The flask.get_flashed_messages() function now allows rendering flashed mes-
sage categories in separate blocks, through a category_filter argument.
• The flask.Flask.run() method now accepts None for host and port arguments,
using default values when None. This allows for calling run using configuration
values, e.g. app.run(app.config.get('MYHOST'), app.config.get('MYPORT')),
with proper behavior whether or not a config file is provided.
• The flask.render_template() method now accepts a either an iterable of tem-
plate names or a single template name. Previously, it only accepted a single
template name. On an iterable, the first template found is rendered.
• Added flask.Flask.app_context() which works very similar to the request con-
text but only provides access to the current application. This also adds support
for URL generation without an active request context.
• View functions can now return a tuple with the first instance being an instance
of flask.Response. This allows for returning jsonify(error="error msg"), 400
from a view function.
• Flask and Blueprint now provide a get_send_file_max_age() hook for sub-
classes to override behavior of serving static files from Flask when using
flask.Flask.send_static_file() (used for the default static file handler) and
send_file(). This hook is provided a filename, which for example allows chang-
ing cache controls by file extension. The default max-age for send_file and static
files can be configured through a new SEND_FILE_MAX_AGE_DEFAULT configuration
variable, which is used in the default get_send_file_max_age implementation.
• Fixed an assumption in sessions implementation which could break message
flashing on sessions implementations which use external storage.


• Changed the behavior of tuple return values from functions. They are no longer
arguments to the response object, they now have a defined meaning.
• Added flask.Flask.request_globals_class to allow a specific class to be used
on creation of the g instance of each request.
• Added required_methods attribute to view functions to force-add methods on reg-
• Added flask.after_this_request().
• Added flask.stream_with_context() and the ability to push contexts multiple
times without producing unexpected behavior.

32.9 Version 0.8.1

Bugfix release, released on July 1st 2012
• Fixed an issue with the undocumented flask.session module to not work properly
on Python 2.5. It should not be used but did cause some problems for package

32.10 Version 0.8

Released on September 29th 2011, codename Rakija
• Refactored session support into a session interface so that the implementation of
the sessions can be changed without having to override the Flask class.
• Empty session cookies are now deleted properly automatically.
• View functions can now opt out of getting the automatic OPTIONS implementa-
• HTTP exceptions and Bad Request errors can now be trapped so that they show
up normally in the traceback.
• Flask in debug mode is now detecting some common problems and tries to warn
you about them.
• Flask in debug mode will now complain with an assertion error if a view was
attached after the first request was handled. This gives earlier feedback when
users forget to import view code ahead of time.
• Added the ability to register callbacks that are only triggered once at the begin-
ning of the first request. (Flask.before_first_request())
• Malformed JSON data will now trigger a bad request HTTP exception instead
of a value error which usually would result in a 500 internal server error if not
handled. This is a backwards incompatible change.


• Applications now not only have a root path where the resources and modules are
located but also an instance path which is the designated place to drop files that
are modified at runtime (uploads etc.). Also this is conceptually only instance de-
pending and outside version control so it’s the perfect place to put configuration
files etc. For more information see Instance Folders.
• Added the APPLICATION_ROOT configuration variable.
• Implemented session_transaction() to easily modify sessions from the test en-
• Refactored test client internally. The APPLICATION_ROOT configuration variable as
well as SERVER_NAME are now properly used by the test client as defaults.
• Added flask.views.View.decorators to support simpler decorating of plug-
gable (class-based) views.
• Fixed an issue where the test client if used with the “with” statement did not
trigger the execution of the teardown handlers.
• Added finer control over the session cookie parameters.
• HEAD requests to a method view now automatically dispatch to the get method
if no handler was implemented.
• Implemented the virtual flask.ext package to import extensions from.
• The context preservation on exceptions is now an integral component of Flask
itself and no longer of the test client. This cleaned up some internal logic and
lowers the odds of runaway request contexts in unittests.

32.11 Version 0.7.3

Bugfix release, release date to be decided
• Fixed the Jinja2 environment’s list_templates method not returning the correct
names when blueprints or modules were involved.

32.12 Version 0.7.2

Bugfix release, released on July 6th 2011
• Fixed an issue with URL processors not properly working on blueprints.

32.13 Version 0.7.1

Bugfix release, released on June 29th 2011
• Added missing future import that broke 2.5 compatibility.


• Fixed an infinite redirect issue with blueprints.

32.14 Version 0.7

Released on June 28th 2011, codename Grappa
• Added make_default_options_response() which can be used by subclasses to
alter the default behavior for OPTIONS responses.
• Unbound locals now raise a proper RuntimeError instead of an AttributeError.
• Mimetype guessing and etag support based on file objects is now deprecated for
flask.send_file() because it was unreliable. Pass filenames instead or attach
your own etags and provide a proper mimetype by hand.
• Static file handling for modules now requires the name of the static folder to
be supplied explicitly. The previous autodetection was not reliable and caused
issues on Google’s App Engine. Until 1.0 the old behavior will continue to work
but issue dependency warnings.
• fixed a problem for Flask to run on jython.
• added a PROPAGATE_EXCEPTIONS configuration variable that can be used to flip
the setting of exception propagation which previously was linked to DEBUG alone
and is now linked to either DEBUG or TESTING.
• Flask no longer internally depends on rules being added through the add_url_rule
function and can now also accept regular werkzeug rules added to the url map.
• Added an endpoint method to the flask application object which allows one to
register a callback to an arbitrary endpoint with a decorator.
• Use Last-Modified for static file sending instead of Date which was incorrectly
introduced in 0.6.
• Added create_jinja_loader to override the loader creation process.
• Implemented a silent flag for config.from_pyfile.
• Added teardown_request decorator, for functions that should run at the end of
a request regardless of whether an exception occurred. Also the behavior for
after_request was changed. It’s now no longer executed when an exception is
raised. See Upgrading to new Teardown Handling
• Implemented flask.has_request_context()
• Deprecated init_jinja_globals. Override the create_jinja_environment() method
instead to achieve the same functionality.
• Added flask.safe_join()
• The automatic JSON request data unpacking now looks at the charset mimetype


• Don’t modify the session on flask.get_flashed_messages() if there are no mes-
sages in the session.
• before_request handlers are now able to abort requests with errors.
• it is not possible to define user exception handlers. That way you can provide
custom error messages from a central hub for certain errors that might occur dur-
ing request processing (for instance database connection errors, timeouts from
remote resources etc.).
• Blueprints can provide blueprint specific error handlers.
• Implemented generic Pluggable Views (class-based views).

32.15 Version 0.6.1

Bugfix release, released on December 31st 2010
• Fixed an issue where the default OPTIONS response was not exposing all valid
methods in the Allow header.
• Jinja2 template loading syntax now allows ”./” in front of a template load path.
Previously this caused issues with module setups.
• Fixed an issue where the subdomain setting for modules was ignored for the
static folder.
• Fixed a security problem that allowed clients to download arbitrary files if the
host server was a windows based operating system and the client uses back-
slashes to escape the directory the files where exposed from.

32.16 Version 0.6

Released on July 27th 2010, codename Whisky
• after request functions are now called in reverse order of registration.
• OPTIONS is now automatically implemented by Flask unless the application
explicitly adds ‘OPTIONS’ as method to the URL rule. In this case no automatic
OPTIONS handling kicks in.
• static rules are now even in place if there is no static folder for the module. This
was implemented to aid GAE which will remove the static folder if it’s part of a
mapping in the .yml file.
• the config is now available in the templates as config.
• context processors will no longer override values passed directly to the render
• added the ability to limit the incoming request data with the new
MAX_CONTENT_LENGTH configuration value.


• the endpoint for the flask.Module.add_url_rule() method is now optional to
be consistent with the function of the same name on the application object.
• added a flask.make_response() function that simplifies creating response object
instances in views.
• added signalling support based on blinker. This feature is currently optional and
supposed to be used by extensions and applications. If you want to use it, make
sure to have blinker installed.
• refactored the way URL adapters are created. This process is now fully customiz-
able with the create_url_adapter() method.
• modules can now register for a subdomain instead of just an URL prefix. This
makes it possible to bind a whole module to a configurable subdomain.

32.17 Version 0.5.2

Bugfix Release, released on July 15th 2010
• fixed another issue with loading templates from directories when modules were

32.18 Version 0.5.1

Bugfix Release, released on July 6th 2010
• fixes an issue with template loading from directories when modules where used.

32.19 Version 0.5

Released on July 6th 2010, codename Calvados
• fixed a bug with subdomains that was caused by the inability to specify the
server name. The server name can now be set with the SERVER_NAME config key.
This key is now also used to set the session cookie cross-subdomain wide.
• autoescaping is no longer active for all templates. Instead it is only active for .
html, .htm, .xml and .xhtml. Inside templates this behavior can be changed with
the autoescape tag.
• refactored Flask internally. It now consists of more than a single file.
• flask.send_file() now emits etags and has the ability to do conditional re-
sponses builtin.
• (temporarily) dropped support for zipped applications. This was a rarely used
feature and led to some confusing behavior.


This can also be used to trigger custom requests that do not pop the request stack for testing.git folder for themes.3. 309 . • the logger switches to DEBUG mode now if debug is enabled. • removed support for create_jinja_loader which is no longer used in 0. 32. built documentation in zip and pdf file and some .Config. • added TESTING switch that can activate unittesting helpers. • test client has not the ability to preserve the request context for a little longer. • because the Python standard library caches loggers. codename Schnaps • added support for categories for flashed messages. • added support for context binding that does not require the use of the with state- ment for playing in the console.1 Bugfix release.pyc files) 32.3 Released on May 28th 2010.5 due to the improved module support. • the application now configures a logging. • added support for per-package template and static-file directories.21 Version 0. the name of the logger is configurable now to better support unittests. codename Rakia • added the ability to register application wide error handlers from modules. • after_request() handlers are now also invoked if the request dies with an ex- ception and an error handling page kicks in. released on May 28th 2010 • fixed a error reporting bug with flask.from_envvar() • removed some unused code from flask • release does no longer include development leftover files (.20 Version 0. 32. • added a helper function to expose files from any directory.4 Released on June 18th 2010.22 Version 0.Handler and will log request handling exceptions to that logger when not in debug mode. This makes it possible to receive mails on server errors for example.

• added support for Google Appengine.1 First public preview release.24 Version 0.1 by default now to fix issues with chrome. • server listens on 127.2 Released on May 12th 2010. • sessions can be set to be permanent now on a per-session basis. 32. • add_url_rule() can now also register a view function. • added support for send_file() • module support and internal request handling refactoring to better support pluggable applications. 32. • better error reporting on missing secret keys. 310 . • refactored internal request dispatching. • added external URL support.0. • the request context is now available within the with statement making it possible to further push the request context or pop it. codename Jägermeister • various bugfixes • integrated JSON support • added get_template_attribute() helper function.23 Version 0. • added support for configurations.0.

1 Development Lead • Armin Ronacher <armin.1.1 Authors Flask is written and maintained by Armin Ronacher and various contributors: 33. It basically means: do whatever you want with it as long as the copyright in Flask sticks around. CHAPTER THIRTYTHREE LICENSE Flask is licensed under a three clause BSD License. The full license text can be found below (Flask License).com> 33. the conditions are not modified and the disclaimer is present. Furthermore you must not use the names of the authors to promote derivatives of the software without written consent. For the documentation and artwork different licenses apply. 33.ronacher@active-4.1.2 Patches and Suggestions • Adam Zapletal • Ali Afshar • Chris Edgemon • Chris Grindstaff • Christopher Grebs • Daniel Neuhäuser • Dan Sully • David Lord @davidism • Edmond Burnett • Florent Xicluna 311 .

• Georg Brandl • Jeff Widman @jeffwidman • Justin Quick • Kenneth Reitz • Keyan Pishdadian • Marian Sigler • Martijn Pieters • Matt Campell • Matthew Frazier • Michael van Tellingen • Ron DuPlain • Sebastien Estienne • Simon Sapin • Stephane Wirtel • Thomas Schranz • Zhao Xiaohong 33. • “AUTHORS” hereby refers to all the authors listed in the Authors section. Redistribution and use in source and binary forms of the software as well as docu- mentation. See AUTHORS for more details.2 General License Definitions The following section contains the full license texts for Flask and the documentation. • The “Flask Artwork License” applies to the project’s Horn-Logo. are permitted provided that the following conditions are met: 312 . 33. Some rights reserved. with or without modification.3 Flask License Copyright (c) 2015 by Armin Ronacher and contributors. • The “Flask License” applies to all the source code shipped as part of Flask (Flask itself as well as the examples and the unittests) as well as documentation.

SPECIAL. but does not indicate endorsement by the project. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT. this list of conditions and the following disclaimer. OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE AND DOCUMENTATION. • The names of the contributors may not be used to endorse or promote products derived from this software without specific prior written permission.4 Flask Artwork License Copyright (c) 2010 by Armin Ronacher. DATA. INDIRECT. THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DIS- CLAIMED. Note: we would appreciate that you make the image a link to http://flask. OR PROFITS. • The names of the contributors to the Flask software (see AUTHORS) may not be used to endorse or promote products derived from this software without specific prior written permission. 313 . LOSS OF USE. with or without modification. STRICT LIABILITY. 33.) of the image. BUT NOT LIMITED TO.org/ if you use it on a web page. INCIDENTAL. Some rights reserved. OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY. This logo or a modified version may be used by anyone to refer to the Flask project. EXEMPLARY. • Redistributions of source code must retain the above copyright notice. Redistribution and use in source (the SVG file) and binary forms (rendered PNG files etc. OR CONSEQUENTIAL DAMAGES (INCLUDING. are permitted provided that the fol- lowing conditions are met: • Redistributions of source code must retain the above copyright notice and this list of conditions. INCLUDING. PRO- CUREMENT OF SUBSTITUTE GOODS OR SERVICES. this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. WHETHER IN CONTRACT. BUT NOT LIMITED TO. EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.pocoo. • Redistributions in binary form must reproduce the above copyright notice. THIS SOFTWARE AND DOCUMENTATION IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS IS” AND ANY EXPRESS OR IMPLIED WAR- RANTIES.

Blueprint method). 228 tribute). 247 add_url_rule() (flask.Request attribute). 213 A app_import_path (flask.cli).Blueprint method). 243 args (flask.Flask method).Blueprint method).Flask method). 213 add_template_filter() (flask. 248 method).Flask attribute). 246 212 appcontext_tearing_down (in module add_url_rule() flask).Blueprint method).BlueprintSetupState AppGroup (class in flask. 243 tribute). AppContext (class in flask.Blueprint (flask. 212 method). 243 191 appcontext_popped (in module flask). 212 auto_find_instance_path() (flask.Blueprint app_template_test() (flask. 244 method). 229 (flask. 213 add_template_global() (flask. 243 app_errorhandler() (flask. base_url (flask. 213 add_template_test() (flask. 216 add_url_rule() (flask. 212 method).Blueprint method).ScriptInfo at- abort() (in module flask). 213 add_app_template_test() (flask.ctx). 192 before_first_request() (flask.Blueprint method).Blueprint method). 246 (flask.Flask method). 190 method). method).Blueprint (flask.Request attribute).View class after_app_request() (flask. 192 B after_request_funcs (flask. 217 192 before_app_first_request() after_this_request() (in module flask).Flask app_url_defaults() (flask.Blueprint method). appcontext_pushed (in module flask).Blueprint method).Blueprint method). 193 314 . 213 add_app_template_global() app_template_global() (flask. 193 method).blueprints. 191 as_view() (flask.blueprints.Flask at- _app_ctx_stack (in module flask). 212 before_first_request_funcs (flask. INDEX Symbols app_ctx_globals_class (flask.views. 193 212 after_request() (flask.Flask app_context_processor() (flask.Flask method).cli.Flask at- tribute). 252 add_app_template_filter() app_template_filter() (flask.Flask app_url_value_preprocessor() method).Blueprint attribute). 192 _request_ctx_stack (in module flask). 191 (flask. 252 method). 213 app (flask. 213 app_context() (flask.Blueprint method).Flask after_request() (flask.BlueprintSetupState before_app_request() (flask. 212 method).

193 method).View attribute). 253 flask (module).Flask method).RequestContext method).Flask method). 236 C dumps() (in module flask. 225 attribute).Flask attribute).Flask attribute).Request attribute). 252 default() (flask. 32 194 YOURAPPLICATION_SETTINGS. 34 context_processor() (flask.Flask method).Flask method).Flask flask. 223 tribute). 32 method).Flask attribute).ext (in module flask). 217 context_processor() (flask. 194 files (flask.JSONEncoder FLASKR_SETTINGS.json. 193 command() (flask. 216 error_handler_spec (flask.before_request() (flask. 195 213 digest_method() before_request() (flask. first_registration 194 (flask. 234 debug (flask. 241 data (flask. FlaskClient (class in flask.cli). 217 create_url_adapter() (flask.json (module).Markup class method).Request attribute). 232 create_global_jinja_loader() (flask.Flask attribute).Request attribute).View attribute). 189 data (flask. 195 BlueprintSetupState (class in do_teardown_request() (flask.cli.Request attribute).ScriptInfo attribute).ScriptInfo attribute).Request attribute).Flask blueprints (flask. 218 Config (class in flask). 195 242 errorhandler() (flask.Flask F method). 194 create_jinja_environment() (flask.blueprints.blueprints). 238 endpoint() (flask.json). 213 config (flask. 220 flask.Flask attribute).SecureCookieSessionInterface before_request_funcs (flask.Blueprint method). 224 248 FlaskGroup (class in flask.Flask attribute). 233 253 escape() (in module flask). static method). 236 form (flask. 216 315 . 195 config_class (flask.Flask attribute).json). copy() (flask. 214 ule flask). 34 decorators (flask. 243 method). FLASKR_SETTINGS.Flask at.Response attribute). 233 D Flask (class in flask). 236 cli (flask. 195 dump() (in module flask. default_config (flask. 244 method).Flask extensions (flask. 196 method). 226 errorhandler() (flask. 194 FLASK_DEBUG. 193 dispatch_request() (flask. 244 flash() (in module flask).AppGroup method).blueprints.sessions.testing). E 252 endpoint (flask.Blueprint method). 217 flask. 212 195 blueprint (flask. copy_current_request_context() (in mod. 196 create_app (flask. 213 FLASK_DEBUG. 194 endpoint() (flask.Request attribute). 194 environ (flask.views. 248 blueprint (flask. escape() (flask. 218 do_teardown_appcontext() (flask.cli.BlueprintSetupState current_app (in module flask).Blueprint method).Flask method).Flask method). 68 cookies (flask.Flask attribute).ctx. 189 data (flask.Blueprint environment variable method). 193 (flask.Request attribute).cli.views. Blueprint (class in flask).BlueprintSetupState dispatch_request() (flask.

Flask method). 197 from_mapping() (flask. jinja_env (flask. get_send_file_max_age() (flask. 227 method). 217 method). 252 jsonify() (in module flask.Request method).Flask method).Flask attribute).Flask attribute). 225 has_static_folder (flask. full_path (flask. 198 197 json_encoder (flask.SessionInterface headers (flask.SecureCookieSessionInterface attribute). json_decoder (flask. 197 from_object() (flask.Blueprint at- G tribute).Flask attribute). 214 jinja_loader (flask. 198 flask).json). 222 headers (flask. 237 245 JSONEncoder (class in flask.Flask attribute).Request attribute).SessionInterface (flask.Config method). 222 198 get_cookie_secure() instance_path (flask. get_cookie_path() 198 (flask. 236 group() (flask. 222 method).SessionInterface I method). 197 has_request_context() (in module flask).Flask attribute). 214 g (in module flask).json). method). 235 H K handle_exception() (flask.Blueprint 198 method).sessions. 197 jinja_loader (flask.Flask 239 method). 218 method).SessionInterface method). get_send_file_max_age() (flask. 240 method). 198 240 jinja_environment (flask. 239 handle_http_exception() (flask. 198 full_dispatch_request() (flask. handle_url_build_error() (flask.Flask method). get_cookie_domain() 198 (flask. 218 J get_namespace() (flask. JSONDecoder (class in flask.Config method).Request attribute).sessions.SessionInterface inject_url_defaults() (flask.Flask attribute). 198 (flask. key_derivation 197 (flask.Flask from_pyfile() (flask.Flask attribute).sessions. 217 get_flashed_messages() (in module flask). 218 got_first_request (flask.Flask from_json() (flask.Request attribute).sessions.sessions.Flask method).Flask attribute). 237 json (flask. 199 got_request_exception (in module flask). 217 226 has_static_folder (flask.Request attribute). 239 method). 239 handle_user_exception() (flask. 222 init_jinja_globals() (flask.json). 223 316 . 198 234 get_json() (flask.Response attribute).Config method).sessions.Blueprint attribute). 222 is_xhr (flask.SessionInterface is_json (flask.Config method).cli.Request attribute).Config method).from_envvar() (flask.sessions. 198 get_template_attribute() (in module jinja_options (flask. 219 get_cookie_httponly() (flask.Flask attribute).Flask 214 method).AppGroup method). 222 is_null_session() get_expiration_time() (flask.Config method).Flask has_app_context() (in module flask). iter_blueprints() (flask.

Flask method).sessions.Request log_exception() (flask. 224 method).sessions.Flask method). 221 process_response() (flask. 199 open_resource() (flask.SessionMixin make_shell_context() (flask. 222 make_null_session() options (flask.Flask methods (flask. 201 match_request() pickle_based (flask.Flask attribute). 242 attribute).Request at- pop() (flask.Blueprint permanent (flask.Flask method).cli.SessionMixin at- 201 tribute). message_flashed (in module flask).Request attribute). 232 attribute).Flask method).Flask attribute).SessionInterface attribute). 200 253 make_response() (in module flask).RequestContext (flask.L attribute).Flask method).session attribute). 243 tribute). 253 O loads() (in module flask.Flask Markup (class in flask).Request attribute). 220 method).Flask attribute). 202 N push() (flask. 221 method).session attribute).SessionMixin at- Python Enhancement Proposals tribute). 224 PEP 8. 200 make_default_options_response() open_session() (flask. modified (flask.session attribute).sessions.AppContext method). 218 pop() (flask. pass_script_info() (in module flask.ScriptInfo method). 217 make_setup_state() (flask.Flask attribute).ctx. 200 permanent_session_lifetime (flask. M 214 make_config() (flask.View attribute). 248 method).Flask method).sessions. (flask. 218 logger (flask.Flask method).json). 200 push() (flask. 199 open_session() make_null_session() (flask. 222 max_content_length (flask.Flask logger_name (flask. 214 permanent (flask.json). 223 propagate_exceptions (flask. 243 name (flask.Flask at- module (flask.ctx. 218 tribute). 199 method).Blueprint method).RequestContext method).Response attribute). 201 (flask.cli).BlueprintSetupState (flask. 248 preserve_context_on_exception mimetype (flask. 200 open_resource() (flask.Flask attribute).SessionInterface 199 method). 247 242 method (flask. 222 P make_response() (flask.ctx. 244 method).Flask method). 242 new (flask. 217 preprocess_request() (flask.views).Request attribute). 236 on_json_loading_failed() (flask.views.blueprints.SessionInterface 317 . 228 path (flask. 222 load() (in module flask. 223 load_app() (flask.sessions. 199 open_instance_resource() (flask. 201 MethodView (class in flask.sessions.RequestContext new (flask.sessions).ctx. 220 (flask.ctx. 201 modified (flask. 199 method).sessions.AppContext method).SessionInterface method). 236 NullSession (class in flask. 281 null_session_class (flask.

223 run() (flask. 204 signals.sessions. 234 (flask.R SecureCookieSession (class in record() (flask.Flask attribute). 223 request_class (flask.FlaskClient method).Response method).Flask S method). 215 send_from_directory() (in module flask). 223 request_started (in module flask).Blueprint attribute). 245 session_cookie_name (flask. 204 method). 215 render_template_string() (in module send_static_file() (flask. 247 signals.Flask method).sessions). 253 shell_command (in module flask.SecureCookieSessionInterface tribute).Flask save_session() (flask. SecureCookieSessionInterface (class in 215 flask.sessions).SecureCookieSessionInterface request_finished (in module flask). 215 224 route() (flask.Flask method). 237 205 Request (class in flask).SessionInterface (flask. flask).Flask method). routing_exception (flask.cli). 220 request_context() (flask. 253 shell_context_processor() (flask.SecureCookieSessionInterface request (class in flask).Blueprint render_template() (in module flask).Flask method).testing.Blueprint method). 204 register_blueprint() (flask. 205 save_session() should_set_cookie() (flask.cli). 223 record_once() (flask. 237 method).Flask attribute). 230 202 send_file_max_age_default (flask. 223 script_root (flask.Flask attribute). send_file() (in module flask). 219 attribute).sessions.Flask method).Flask method). 228 select_jinja_autoescape() (flask. 203 flask.Namespace ScriptInfo (class in flask. RequestContext (class in flask.Flask at- salt (flask.Flask register_error_handler() (flask.Request at. 231 shell_context_processors (flask. 205 safe_join() (in module flask).signals. session_class 202 (flask. 224 RFC session_transaction() RFC 822.Flask 231 method). 205 attribute). route() (flask. 202 send_static_file() (flask. 217 signal() (flask. tribute). 216 serializer (flask. 221 tribute).sessions. 223 method). 204 method). 247 secret_key (flask.sessions.cli). 203 set_cookie() (flask.Flask at- request_tearing_down (in module flask).Flask attribute). 242 205 Response (class in flask). 223 should_ignore_error() (flask. 252 method).Request attribute).sessions). 215 flask.Flask register() (flask.signals_available (in module 318 . 219 SessionMixin (class in flask. 223 redirect() (in module flask).Namespace (class in flask). register_error_handler() (flask.sessions).Blueprint method). 245 attribute). 205 246 session_interface (flask.sessions). 220 run_command (in module flask. 202 session (class in flask). 219 session_json_serializer (in module response_class (flask.SessionInterface method).Blueprint method).ctx).Blueprint method). 215 method). 203 SessionInterface (class in flask.sessions.

Flask method).Flask method). 210 method). 209 U unescape() (flask.Flask method).Flask attribute).Markup method). 208 test_client_class (flask. tribute).Request attribute).Flask teardown_appcontext_funcs (flask.Request attribute). Y 209 YOURAPPLICATION_SETTINGS. 216 url_value_preprocessor() (flask. 209 trap_http_exception() (flask.Flask at. attribute). 211 test_client() (flask. 210 status (flask.blueprints. 216 use_x_sendfile (flask.Flask attribute). 210 241 url_for() (in module flask).Flask method). W 207 with_appcontext() (in module flask.BlueprintSetupState attribute).Flask method).Flask attribute). 227 striptags() (flask.values (flask. 244 attribute). 207 template_global() (flask.views). 211 teardown_request() (flask. 207 View (class in flask. 233 update_template_context() (flask. url_defaults() (flask.Flask view_args (flask.Response attribute). 210 method).Flask attribute).cli). 216 tribute). 207 wsgi_app() (flask. 210 subdomain (flask.Flask method). 206 V teardown_request_funcs (flask.Flask method). 219 teardown_app_request() (flask. 244 253 template_test() (flask.Request attribute).Request attribute). 247 template_context_processors (flask.blueprints. 216 216 stream_with_context() (in module flask).Flask at- static_folder (flask. 207 view_functions (flask. 209 216 url_default_functions (flask.Blueprint url_rule_class (flask. 210 attribute).Blueprint attribute).blueprints.Response attribute).Flask method).Flask attribute).Flask method).Flask attribute). 205 url_value_preprocessor() (flask.Blueprint tribute). 217 T url_rule (flask.Blueprint method). 209 url (flask.BlueprintSetupState status_code (flask.Request attribute). 217 319 .Flask method). 206 url_value_preprocessors (flask. 244 219 url_defaults() (flask.Flask attribute). 211 template_filter() (flask.Blueprint teardown_appcontext() (flask. 216 method).Markup method). 219 attribute).Flask method). 244 url_build_error_handlers (flask. 233 url_map (flask. 68 test_request_context() (flask. 244 url_root (flask.Request attribute). stream (flask. 209 testing (flask.Flask at- static_folder (flask. 219 url_defaults (flask.Flask at- teardown_request() (flask. flask).BlueprintSetupStateurl_prefix (flask. template_rendered (in module flask).Flask method). 205 tribute).