Configuration Handling

Applications need some kind of configuration. There are different settings you might want to change depending on the application environment like toggling the debug mode, setting the secret key, and other such environment-specific things.

The way Flask is designed usually requires the configuration to be available when the application starts up. You can hard code the configuration in the code, which for many small applications is not actually that bad, but there are better ways.

Independent of how you load your config, there is a config object available which holds the loaded configuration values: The config attribute of the Flask object. This is the place where Flask itself puts certain configuration values and also where extensions can put their configuration values. But this is also where you can have your own configuration.

Configuration Basics

The config is actually a subclass of a dictionary and can be modified just like any dictionary:

  1. app = Flask(__name__)
  2. app.config['TESTING'] = True

Certain configuration values are also forwarded to the Flask object so you can read and write them from there:

  1. app.testing = True

To update multiple keys at once you can use the dict.update() method:

  1. app.config.update(
  2. TESTING=True,
  3. SECRET_KEY='192b9bdd22ab9ed4d12e236c78afcb9a393ec15f71bbf5dc987d54727823bcbf'
  4. )

Environment and Debug Features

The ENV and DEBUG config values are special because they may behave inconsistently if changed after the app has begun setting up. In order to set the environment and debug mode reliably, Flask uses environment variables.

The environment is used to indicate to Flask, extensions, and other programs, like Sentry, what context Flask is running in. It is controlled with the FLASK_ENV environment variable and defaults to production.

Setting FLASK_ENV to development will enable debug mode. flask run will use the interactive debugger and reloader by default in debug mode. To control this separately from the environment, use the FLASK_DEBUG flag.

Changelog

Changed in version 1.0: Added FLASK_ENV to control the environment separately from debug mode. The development environment enables debug mode.

To switch Flask to the development environment and enable debug mode, set FLASK_ENV:

BashFishCMDPowershell

  1. $ export FLASK_ENV=development
  2. $ flask run
  1. $ set -x FLASK_ENV development
  2. $ flask run
  1. > set FLASK_ENV=development
  2. > flask run
  1. > $env:FLASK_ENV = "development"
  2. > flask run

Using the environment variables as described above is recommended. While it is possible to set ENV and DEBUG in your config or code, this is strongly discouraged. They can’t be read early by the flask command, and some systems or extensions may have already configured themselves based on a previous value.

Builtin Configuration Values

The following configuration values are used internally by Flask:

ENV

What environment the app is running in. Flask and extensions may enable behaviors based on the environment, such as enabling debug mode. The env attribute maps to this config key. This is set by the FLASK_ENV environment variable and may not behave as expected if set in code.

Do not enable development when deploying in production.

Default: 'production'

Changelog

New in version 1.0.

DEBUG

Whether debug mode is enabled. When using flask run to start the development server, an interactive debugger will be shown for unhandled exceptions, and the server will be reloaded when code changes. The debug attribute maps to this config key. This is enabled when ENV is 'development' and is overridden by the FLASK_DEBUG environment variable. It may not behave as expected if set in code.

Do not enable debug mode when deploying in production.

Default: True if ENV is 'development', or False otherwise.

TESTING

Enable testing mode. Exceptions are propagated rather than handled by the the app’s error handlers. Extensions may also change their behavior to facilitate easier testing. You should enable this in your own tests.

Default: False

PROPAGATE_EXCEPTIONS

Exceptions are re-raised rather than being handled by the app’s error handlers. If not set, this is implicitly true if TESTING or DEBUG is enabled.

Default: None

PRESERVE_CONTEXT_ON_EXCEPTION

Don’t pop the request context when an exception occurs. If not set, this is true if DEBUG is true. This allows debuggers to introspect the request data on errors, and should normally not need to be set directly.

Default: None

TRAP_HTTP_EXCEPTIONS

If there is no handler for an HTTPException-type exception, re-raise it to be handled by the interactive debugger instead of returning it as a simple error response.

Default: False

TRAP_BAD_REQUEST_ERRORS

Trying to access a key that doesn’t exist from request dicts like args and form will return a 400 Bad Request error page. Enable this to treat the error as an unhandled exception instead so that you get the interactive debugger. This is a more specific version of TRAP_HTTP_EXCEPTIONS. If unset, it is enabled in debug mode.

Default: None

SECRET_KEY

A secret key that will be used for securely signing the session cookie and can be used for any other security related needs by extensions or your application. It should be a long random bytes or str. For example, copy the output of this to your config:

  1. $ python -c 'import secrets; print(secrets.token_hex())'
  2. '192b9bdd22ab9ed4d12e236c78afcb9a393ec15f71bbf5dc987d54727823bcbf'

Do not reveal the secret key when posting questions or committing code.

Default: None

SESSION_COOKIE_NAME

The name of the session cookie. Can be changed in case you already have a cookie with the same name.

Default: 'session'

SESSION_COOKIE_DOMAIN

The domain match rule that the session cookie will be valid for. If not set, the cookie will be valid for all subdomains of SERVER_NAME. If False, the cookie’s domain will not be set.

Default: None

SESSION_COOKIE_PATH

The path that the session cookie will be valid for. If not set, the cookie will be valid underneath APPLICATION_ROOT or / if that is not set.

Default: None

SESSION_COOKIE_HTTPONLY

Browsers will not allow JavaScript access to cookies marked as “HTTP only” for security.

Default: True

SESSION_COOKIE_SECURE

Browsers will only send cookies with requests over HTTPS if the cookie is marked “secure”. The application must be served over HTTPS for this to make sense.

Default: False

SESSION_COOKIE_SAMESITE

Restrict how cookies are sent with requests from external sites. Can be set to 'Lax' (recommended) or 'Strict'. See Set-Cookie options.

Default: None

Changelog

New in version 1.0.

PERMANENT_SESSION_LIFETIME

If session.permanent is true, the cookie’s expiration will be set this number of seconds in the future. Can either be a datetime.timedelta or an int.

Flask’s default cookie implementation validates that the cryptographic signature is not older than this value.

Default: timedelta(days=31) (2678400 seconds)

SESSION_REFRESH_EACH_REQUEST

Control whether the cookie is sent with every response when session.permanent is true. Sending the cookie every time (the default) can more reliably keep the session from expiring, but uses more bandwidth. Non-permanent sessions are not affected.

Default: True

USE_X_SENDFILE

When serving files, set the X-Sendfile header instead of serving the data with Flask. Some web servers, such as Apache, recognize this and serve the data more efficiently. This only makes sense when using such a server.

Default: False

SEND_FILE_MAX_AGE_DEFAULT

When serving files, set the cache control max age to this number of seconds. Can be a datetime.timedelta or an int. Override this value on a per-file basis using get_send_file_max_age() on the application or blueprint.

If None, send_file tells the browser to use conditional requests will be used instead of a timed cache, which is usually preferable.

Default: None

SERVER_NAME

Inform the application what host and port it is bound to. Required for subdomain route matching support.

If set, will be used for the session cookie domain if SESSION_COOKIE_DOMAIN is not set. Modern web browsers will not allow setting cookies for domains without a dot. To use a domain locally, add any names that should route to the app to your hosts file.

  1. 127.0.0.1 localhost.dev

If set, url_for can generate external URLs with only an application context instead of a request context.

Default: None

APPLICATION_ROOT

Inform the application what path it is mounted under by the application / web server. This is used for generating URLs outside the context of a request (inside a request, the dispatcher is responsible for setting SCRIPT_NAME instead; see Application Dispatching for examples of dispatch configuration).

Will be used for the session cookie path if SESSION_COOKIE_PATH is not set.

Default: '/'

PREFERRED_URL_SCHEME

Use this scheme for generating external URLs when not in a request context.

Default: 'http'

MAX_CONTENT_LENGTH

Don’t read more than this many bytes from the incoming request data. If not set and the request does not specify a CONTENT_LENGTH, no data will be read for security.

Default: None

JSON_AS_ASCII

Serialize objects to ASCII-encoded JSON. If this is disabled, the JSON returned from jsonify will contain Unicode characters. This has security implications when rendering the JSON into JavaScript in templates, and should typically remain enabled.

Default: True

JSON_SORT_KEYS

Sort the keys of JSON objects alphabetically. This is useful for caching because it ensures the data is serialized the same way no matter what Python’s hash seed is. While not recommended, you can disable this for a possible performance improvement at the cost of caching.

Default: True

JSONIFY_PRETTYPRINT_REGULAR

jsonify responses will be output with newlines, spaces, and indentation for easier reading by humans. Always enabled in debug mode.

Default: False

JSONIFY_MIMETYPE

The mimetype of jsonify responses.

Default: 'application/json'

TEMPLATES_AUTO_RELOAD

Reload templates when they are changed. If not set, it will be enabled in debug mode.

Default: None

EXPLAIN_TEMPLATE_LOADING

Log debugging information tracing how a template file was loaded. This can be useful to figure out why a template was not loaded or the wrong file appears to be loaded.

Default: False

MAX_COOKIE_SIZE

Warn if cookie headers are larger than this many bytes. Defaults to 4093. Larger cookies may be silently ignored by browsers. Set to 0 to disable the warning.

Changelog

Changed in version 1.0: LOGGER_NAME and LOGGER_HANDLER_POLICY were removed. See Logging for information about configuration.

Added ENV to reflect the FLASK_ENV environment variable.

Added SESSION_COOKIE_SAMESITE to control the session cookie’s SameSite option.

Added MAX_COOKIE_SIZE to control a warning from Werkzeug.

New in version 0.11: SESSION_REFRESH_EACH_REQUEST, TEMPLATES_AUTO_RELOAD, LOGGER_HANDLER_POLICY, EXPLAIN_TEMPLATE_LOADING

New in version 0.10: JSON_AS_ASCII, JSON_SORT_KEYS, JSONIFY_PRETTYPRINT_REGULAR

New in version 0.9: PREFERRED_URL_SCHEME

New in version 0.8: TRAP_BAD_REQUEST_ERRORS, TRAP_HTTP_EXCEPTIONS, APPLICATION_ROOT, SESSION_COOKIE_DOMAIN, SESSION_COOKIE_PATH, SESSION_COOKIE_HTTPONLY, SESSION_COOKIE_SECURE

New in version 0.7: PROPAGATE_EXCEPTIONS, PRESERVE_CONTEXT_ON_EXCEPTION

New in version 0.6: MAX_CONTENT_LENGTH

New in version 0.5: SERVER_NAME

New in version 0.4: LOGGER_NAME

Configuring from Python Files

Configuration becomes more useful if you can store it in a separate file, ideally located outside the actual application package. This makes packaging and distributing your application possible via various package handling tools (Deploying with Setuptools) and finally modifying the configuration file afterwards.

So a common pattern is this:

  1. app = Flask(__name__)
  2. app.config.from_object('yourapplication.default_settings')
  3. app.config.from_envvar('YOURAPPLICATION_SETTINGS')

This first loads the configuration from the yourapplication.default_settings module and then overrides the values with the contents of the file the YOURAPPLICATION_SETTINGS environment variable points to. This environment variable can be set in the shell before starting the server:

BashFishCMDPowershell

  1. $ export YOURAPPLICATION_SETTINGS=/path/to/settings.cfg
  2. $ flask run
  3. * Running on http://127.0.0.1:5000/
  1. $ set -x YOURAPPLICATION_SETTINGS /path/to/settings.cfg
  2. $ flask run
  3. * Running on http://127.0.0.1:5000/
  1. > set YOURAPPLICATION_SETTINGS=\path\to\settings.cfg
  2. > flask run
  3. * Running on http://127.0.0.1:5000/
  1. > $env:YOURAPPLICATION_SETTINGS = "\path\to\settings.cfg"
  2. > flask run
  3. * Running on http://127.0.0.1:5000/

The configuration files themselves are actual Python files. Only values in uppercase are actually stored in the config object later on. So make sure to use uppercase letters for your config keys.

Here is an example of a configuration file:

  1. # Example configuration
  2. SECRET_KEY = '192b9bdd22ab9ed4d12e236c78afcb9a393ec15f71bbf5dc987d54727823bcbf'

Make sure to load the configuration very early on, so that extensions have the ability to access the configuration when starting up. There are other methods on the config object as well to load from individual files. For a complete reference, read the Config object’s documentation.

Configuring from Data Files

It is also possible to load configuration from a file in a format of your choice using from_file(). For example to load from a TOML file:

  1. import toml
  2. app.config.from_file("config.toml", load=toml.load)

Or from a JSON file:

  1. import json
  2. app.config.from_file("config.json", load=json.load)

Configuring from Environment Variables

In addition to pointing to configuration files using environment variables, you may find it useful (or necessary) to control your configuration values directly from the environment. Flask can be instructed to load all environment variables starting with a specific prefix into the config using from_prefixed_env().

Environment variables can be set in the shell before starting the server:

BashFishCMDPowershell

  1. $ export FLASK_SECRET_KEY="5f352379324c22463451387a0aec5d2f"
  2. $ export FLASK_MAIL_ENABLED=false
  3. $ flask run
  4. * Running on http://127.0.0.1:5000/
  1. $ set -x FLASK_SECRET_KEY "5f352379324c22463451387a0aec5d2f"
  2. $ set -x FLASK_MAIL_ENABLED false
  3. $ flask run
  4. * Running on http://127.0.0.1:5000/
  1. > set FLASK_SECRET_KEY="5f352379324c22463451387a0aec5d2f"
  2. > set FLASK_MAIL_ENABLED=false
  3. > flask run
  4. * Running on http://127.0.0.1:5000/
  1. > $env:FLASK_SECRET_KEY = "5f352379324c22463451387a0aec5d2f"
  2. > $env:FLASK_MAIL_ENABLED = "false"
  3. > flask run
  4. * Running on http://127.0.0.1:5000/

The variables can then be loaded and accessed via the config with a key equal to the environment variable name without the prefix i.e.

  1. app.config.from_prefixed_env()
  2. app.config["SECRET_KEY"] # Is "5f352379324c22463451387a0aec5d2f"

The prefix is FLASK_ by default. This is configurable via the prefix argument of from_prefixed_env().

Values will be parsed to attempt to convert them to a more specific type than strings. By default json.loads() is used, so any valid JSON value is possible, including lists and dicts. This is configurable via the loads argument of from_prefixed_env().

When adding a boolean value with the default JSON parsing, only “true” and “false”, lowercase, are valid values. Keep in mind that any non-empty string is considered True by Python.

It is possible to set keys in nested dictionaries by separating the keys with double underscore (__). Any intermediate keys that don’t exist on the parent dict will be initialized to an empty dict.

  1. $ export FLASK_MYAPI__credentials__username=user123
  1. app.config["MYAPI"]["credentials"]["username"] # Is "user123"

On Windows, environment variable keys are always uppercase, therefore the above example would end up as MYAPI__CREDENTIALS__USERNAME.

For even more config loading features, including merging and case-insensitive Windows support, try a dedicated library such as Dynaconf, which includes integration with Flask.

Configuration Best Practices

The downside with the approach mentioned earlier is that it makes testing a little harder. There is no single 100% solution for this problem in general, but there are a couple of things you can keep in mind to improve that experience:

  1. Create your application in a function and register blueprints on it. That way you can create multiple instances of your application with different configurations attached which makes unit testing a lot easier. You can use this to pass in configuration as needed.

  2. Do not write code that needs the configuration at import time. If you limit yourself to request-only accesses to the configuration you can reconfigure the object later on as needed.

  3. Make sure to load the configuration very early on, so that extensions can access the configuration when calling init_app.

Development / Production

Most applications need more than one configuration. There should be at least separate configurations for the production server and the one used during development. The easiest way to handle this is to use a default configuration that is always loaded and part of the version control, and a separate configuration that overrides the values as necessary as mentioned in the example above:

  1. app = Flask(__name__)
  2. app.config.from_object('yourapplication.default_settings')
  3. app.config.from_envvar('YOURAPPLICATION_SETTINGS')

Then you just have to add a separate config.py file and export YOURAPPLICATION_SETTINGS=/path/to/config.py and you are done. However there are alternative ways as well. For example you could use imports or subclassing.

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. You could also inspect an environment variable like YOURAPPLICATION_MODE and set that to production, development etc and import different hard-coded files based on that.

An interesting pattern is also to use classes and inheritance for configuration:

  1. class Config(object):
  2. TESTING = False
  3. class ProductionConfig(Config):
  4. DATABASE_URI = 'mysql://user@localhost/foo'
  5. class DevelopmentConfig(Config):
  6. DATABASE_URI = "sqlite:////tmp/foo.db"
  7. class TestingConfig(Config):
  8. DATABASE_URI = 'sqlite:///:memory:'
  9. TESTING = True

To enable such a config you just have to call into from_object():

  1. app.config.from_object('configmodule.ProductionConfig')

Note that from_object() does not instantiate the class object. If you need to instantiate the class, such as to access a property, then you must do so before calling from_object():

  1. from configmodule import ProductionConfig
  2. app.config.from_object(ProductionConfig())
  3. # Alternatively, import via string:
  4. from werkzeug.utils import import_string
  5. cfg = import_string('configmodule.ProductionConfig')()
  6. app.config.from_object(cfg)

Instantiating the configuration object allows you to use @property in your configuration classes:

  1. class Config(object):
  2. """Base config, uses staging database server."""
  3. TESTING = False
  4. DB_SERVER = '192.168.1.56'
  5. @property
  6. def DATABASE_URI(self): # Note: all caps
  7. return f"mysql://user@{self.DB_SERVER}/foo"
  8. class ProductionConfig(Config):
  9. """Uses production database server."""
  10. DB_SERVER = '192.168.19.32'
  11. class DevelopmentConfig(Config):
  12. DB_SERVER = 'localhost'
  13. class TestingConfig(Config):
  14. DB_SERVER = 'localhost'
  15. DATABASE_URI = 'sqlite:///:memory:'

There are many different ways and it’s up to you how you want to manage your configuration files. However here a list of good recommendations:

  • Keep a default configuration in version control. Either populate the config with this default configuration or import it in your own configuration files before overriding values.

  • Use an environment variable to switch between the configurations. This can be done from outside the Python interpreter and makes development and deployment much easier because you can quickly and easily switch between different configs without having to touch the code at all. If you are working often on different projects you can even create your own script for sourcing that activates a virtualenv and exports the development configuration for you.

  • Use a tool like fabric in production to push code and configurations separately to the production server(s). For some details about how to do that, head over to the Deploying with Fabric pattern.

Instance Folders

Changelog

New in version 0.8.

Flask 0.8 introduces instance folders. Flask for a long time made it possible to refer to paths relative to the application’s folder directly (via Flask.root_path). This was also how many developers loaded configurations stored next to the application. Unfortunately however this only works well if applications are not packages in which case the root path refers to the contents of the package.

With Flask 0.8 a new attribute was introduced: Flask.instance_path. It refers to a new concept called the “instance folder”. The instance folder is designed to not be under version control and be deployment specific. It’s the perfect place to drop things that either change at runtime or configuration files.

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. For explicit configuration use the instance_path parameter:

  1. app = Flask(__name__, instance_path='/path/to/instance/folder')

Please keep in mind that this path must be absolute when provided.

If the instance_path parameter is not provided the following default locations are used:

  • Uninstalled module:

    1. /myapp.py
    2. /instance
  • Uninstalled package:

    1. /myapp
    2. /__init__.py
    3. /instance
  • Installed module or package:

    1. $PREFIX/lib/pythonX.Y/site-packages/myapp
    2. $PREFIX/var/myapp-instance

    $PREFIX is the prefix of your Python installation. This can be /usr or the path to your virtualenv. You can print the value of sys.prefix to see what the prefix is set to.

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:

  1. app = Flask(__name__, 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 instance folder if it exists:

  1. app = Flask(__name__, instance_relative_config=True)
  2. app.config.from_object('yourapplication.default_settings')
  3. app.config.from_pyfile('application.cfg', silent=True)

The path to the instance folder can be found via the Flask.instance_path. Flask also provides a shortcut to open a file from the instance folder with Flask.open_instance_resource().

Example usage for both:

  1. filename = os.path.join(app.instance_path, 'application.cfg')
  2. with open(filename) as f:
  3. config = f.read()
  4. # or via open_instance_resource:
  5. with app.open_instance_resource('application.cfg') as f:
  6. config = f.read()