Python SDK Reference

ConfigCat Python SDK on GitHub

info

This documentation applies to the v8.x version of the ConfigCat Python SDK. For the documentation of the latest release, please refer to this page.

Getting started

1. Install ConfigCat SDK

pip install configcat-client

2. Import the package

import configcatclient

3. Create the ConfigCat client with your SDK Key

client = configcatclient.get('#YOUR-SDK-KEY#')

4. Get your setting value

is_my_awesome_feature_enabled = client.get_value('isMyAwesomeFeatureEnabled', False)
if is_my_awesome_feature_enabled:
    do_the_new_thing()
else:
    do_the_old_thing()

5. Stop ConfigCat client

You can safely shut down all clients at once or individually and release all associated resources on application exit.

configcatclient.close_all() # closes all clients

client.close() # closes a specific client

Creating the ConfigCat Client

ConfigCat Client is responsible for:

• managing the communication between your application and ConfigCat servers.
• caching your setting values and feature flags.
• serving values quickly in a failsafe way.

configcatclient.get('#YOUR-SDK-KEY#') returns a client with default options.

Client options	Description	Default
base_url	Obsolete, sets the CDN base url (forward proxy, dedicated subscription) from where the sdk will download the config JSON.	None
polling_mode	Sets the polling mode for the client. More about polling modes.	PollingMode.auto_poll()
config_cache	Sets a custom config cache implementation for the client. More about cache.	None
proxies	Sets custom proxies for the client. More about proxy.	None
proxy_auth	Sets proxy authentication for the custom proxies. More about proxy.	None
connect_timeout_seconds	The number of seconds to wait for the server to make the initial connection (i.e. completing the TCP connection handshake).	10
read_timeout_seconds	The number of seconds to wait for the server to respond before giving up.	30
flag_overrides	Local feature flag & setting overrides. More about feature flag overrides	None
data_governance	Describes the location of your feature flag and setting data within the ConfigCat CDN. This parameter needs to be in sync with your Data Governance preferences. More about Data Governance. Available options: Global, EuOnly.	Global
default_user	Sets the default user. More about default user.	None
hooks	Used to subscribe events that the SDK sends in specific scenarios. More about hooks.	None
offline	Indicates whether the SDK should be initialized in offline mode. More about offline mode.	False

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        polling_mode=PollingMode.auto_poll()
    )
)

caution

We strongly recommend you to use the ConfigCat Client as a Singleton object in your application. The configcatclient.get() static factory method constructs singleton client instances for your SDK keys. These clients can be closed all at once with the configcatclient.close_all() method or individually with client.close().

Anatomy of get_value()

Parameters	Description
key	REQUIRED. Setting-specific key. Set on ConfigCat Dashboard for each setting.
default_value	REQUIRED. This value will be returned in case of an error.
user	Optional, User Object. Essential when using Targeting. Read more about Targeting.

value = client.get_value(
    'keyOfMySetting', # Setting Key
    False, # Default value
    User('#UNIQUE-USER-IDENTIFIER#') # Optional User Object
)

Anatomy of get_value_details()

get_value_details() is similar to get_value() but instead of returning the evaluated value only, it gives an EvaluationDetails object with more detailed information about the evaluation result.

Parameters	Description
key	REQUIRED. Setting-specific key. Set on ConfigCat Dashboard for each setting.
defaultValue	REQUIRED. This value will be returned in case of an error.
user	Optional, User Object. Essential when using Targeting. Read more about Targeting.

details = client.get_value_details(
    'keyOfMySetting', # Setting Key
    False, # Default value
    User('#UNIQUE-USER-IDENTIFIER#') # Optional User Object
)

The details result contains the following information:

Property	Description
value	The evaluated value of the feature flag or setting.
key	The key of the evaluated feature flag or setting.
is_default_value	True when the default value passed to get_value_details() is returned due to an error.
error	In case of an error, this field contains the error message.
user	The User Object that was used for evaluation.
matched_evaluation_rule	If the evaluation was based on a Targeting Rule, this field contains that specific rule.
matched_evaluation_percentage_rule	If the evaluation was based on a Percentage Rule, this field contains that specific rule.
fetch_time	The last download time (UTC datetime) of the current config.

User Object

The User Object is essential if you'd like to use ConfigCat's Targeting feature.

user_object = User('#UNIQUE-USER-IDENTIFIER#')

user_object = User('[email protected]')

Customized User Object creation

Parameters	Description
identifier	REQUIRED. Unique identifier of a user in your application. Can be any string value, even an email address.
email	Optional parameter for easier Targeting Rule definitions.
country	Optional parameter for easier Targeting Rule definitions.
custom	Optional dictionary for custom attributes of a user for advanced Targeting Rule definitions. E.g. User role, Subscription type.

user_object = User(
  '#UNIQUE-USER-IDENTIFIER#',
  'john@example',
  'United Kingdom',
  { SubscriptionType: 'Pro', UserRole: 'Admin' }
)

Default user

There's an option to set a default User Object that will be used at feature flag and setting evaluation. It can be useful when your application has a single user only, or rarely switches users.

You can set the default User Object either on SDK initialization:

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        default_user=User('[email protected]')
    )
)

or with the set_default_user() method of the ConfigCat client.

client.set_default_user(User('[email protected]'))

Whenever the get_value(), get_value_details(), get_variation_id(), get_all_variation_ids(), or get_all_values() methods are called without an explicit user parameter, the SDK will automatically use the default user as a User Object.

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        default_user=User('[email protected]')
    )
)
# The default user will be used at the evaluation process.
value = client.get_value('keyOfMySetting', False)

When the user parameter is specified on the requesting method, it takes precedence over the default user.

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        default_user=User('[email protected]')
    )
)
other_user = User('[email protected]')
# otherUser will be used at the evaluation process.
value = client.get_value('keyOfMySetting', False, other_user)

For deleting the default user, you can do the following:

client.clear_default_user()

Polling Modes

The ConfigCat SDK supports 3 different polling mechanisms to acquire the setting values from ConfigCat. After latest setting values are downloaded, they are stored in the internal cache, then all get_value() calls are served from there. With the following polling modes, you can customize the SDK to best fit to your application's lifecycle.
More about polling modes.

Auto polling (default)

The ConfigCat SDK downloads the latest values and stores them automatically every 60 seconds.

Use the poll_interval_seconds option parameter of the PollingMode.auto_poll() to change the polling interval.

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        polling_mode=PollingMode.auto_poll(poll_interval_seconds=95)
    )
)

Available options:

Option Parameter	Description	Default
poll_interval_seconds	Polling interval.	60
max_init_wait_time_seconds	Maximum waiting time between the client initialization and the first config acquisition in seconds.	5

caution

Auto polling mode utilizes its polling job in a threading.Thread object. If you are running your application behind an uWSGI web server, the auto polling mode may not work as expected because the uWSGI web server disables Python's threading by default. Please enable threading or switch to another polling mode in this case.

Lazy loading

When calling get_value(), the ConfigCat SDK downloads the latest setting values if they are not present or expired in the cache. In this case get_value() will return the setting value after the cache is updated.

Use cache_refresh_interval_seconds option parameter to set cache lifetime.

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        polling_mode=PollingMode.lazy_load(cache_refresh_interval_seconds=600)
    )
)

Available options:

Option Parameter	Description	Default
cache_refresh_interval_seconds	Cache TTL.	60

Manual polling

Manual polling gives you full control over when the config JSON (with the setting values) is downloaded. ConfigCat SDK will not update them automatically. Calling force_refresh() is your application's responsibility.

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        polling_mode=PollingMode.manual_poll()
    )
)
client.force_refresh()

get_value() returns default_value if the cache is empty. Call force_refresh() to update the cache.

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        polling_mode=PollingMode.manual_poll()
    )
)
value = client.get_value('key', 'my default value') # Returns "my default value"
client.force_refresh()
value = client.get_value('key', 'my default value') # Returns "value from server"

Hooks

With the following hooks you can subscribe to particular events fired by the SDK:

• on_client_ready(): This event is sent when the SDK reaches the ready state. If the SDK is set up with lazy load or manual polling it's considered ready right after instantiation. If it's using auto polling, the ready state is reached when the SDK has a valid config JSON loaded into memory either from cache or from HTTP. If the config couldn't be loaded neither from cache nor from HTTP the on_client_ready event fires when the auto polling's max_init_wait_time_seconds is reached.

• on_config_changed(config: dict): This event is sent when the SDK loads a valid config JSON into memory from cache, and each subsequent time when the loaded config JSON changes via HTTP.

• on_flag_evaluated(evaluation_details: EvaluationDetails): This event is sent each time when the SDK evaluates a feature flag or setting. The event sends the same evaluation details that you would get from get_value_details().

• on_error(error: str): This event is sent when an error occurs within the ConfigCat SDK.

You can subscribe to these events either on SDK initialization:

def on_flag_evaluated(evaluation_details):
    # handle the event
    pass

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        hooks=Hooks(on_flag_evaluated=on_flag_evaluated)
    )
)

or with the get_hooks() method of the ConfigCat client:

client.get_hooks().add_on_flag_evaluated(on_flag_evaluated)

Online / Offline mode

In cases when you'd want to prevent the SDK from making HTTP calls, you can put it in offline mode:

client.set_offline()

In offline mode, the SDK won't initiate HTTP requests and will work only from its cache.

To put the SDK back in online mode, you can do the following:

client.set_online()

With client.is_offline() you can check whether the SDK is in offline mode.

Flag Overrides

With flag overrides you can overwrite the feature flags & settings downloaded from the ConfigCat CDN with local values. Moreover, you can specify how the overrides should apply over the downloaded values. The following 3 behaviours are supported:

• Local only (OverrideBehaviour.LocalOnly): When evaluating values, the SDK will not use feature flags & settings from the ConfigCat CDN, but it will use all feature flags & settings that are loaded from local-override sources.

• Local over remote (OverrideBehaviour.LocalOverRemote): When evaluating values, the SDK will use all feature flags & settings that are downloaded from the ConfigCat CDN, plus all feature flags & settings that are loaded from local-override sources. If a feature flag or a setting is defined both in the downloaded and the local-override source then the local-override version will take precedence.

• Remote over local (OverrideBehaviour.RemoteOverLocal): When evaluating values, the SDK will use all feature flags & settings that are downloaded from the ConfigCat CDN, plus all feature flags & settings that are loaded from local-override sources. If a feature flag or a setting is defined both in the downloaded and the local-override source then the downloaded version will take precedence.

You can set up the SDK to load your feature flag & setting overrides from a file or a dictionary.

JSON File

The SDK can load your feature flag & setting overrides from a file.

File

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        flag_overrides=LocalFileFlagOverrides(
            file_path='path/to/the/local_flags.json',  # path to the file
            override_behaviour=OverrideBehaviour.LocalOnly
        )
    )
)

JSON File Structure

The SDK supports 2 types of JSON structures to describe feature flags & settings.

1. Simple (key-value) structure

{
  "flags": {
    "enabledFeature": true,
    "disabledFeature": false,
    "intSetting": 5,
    "doubleSetting": 3.14,
    "stringSetting": "test"
  }
}

2. Complex (full-featured) structure

This is the same format that the SDK downloads from the ConfigCat CDN. It allows the usage of all features that are available on the ConfigCat Dashboard.

You can download your current config JSON from ConfigCat's CDN and use it as a baseline.

The URL to your current config JSON is based on your Data Governance settings:

• GLOBAL: https://cdn-global.configcat.com/configuration-files/{YOUR-SDK-KEY}/config_v5.json
• EU: https://cdn-eu.configcat.com/configuration-files/{YOUR-SDK-KEY}/config_v5.json

{
  "f": {
    // list of feature flags & settings
    "isFeatureEnabled": {
      // key of a particular flag
      "v": false, // default value, served when no rules are defined
      "i": "430bded3", // variation id (for analytical purposes)
      "t": 0, // feature flag's type, possible values:
      // 0 -> BOOLEAN
      // 1 -> STRING
      // 2 -> INT
      // 3 -> DOUBLE
      "p": [
        // list of Percentage Rules
        {
          "o": 0, // rule's order
          "v": true, // value served when the rule is selected during evaluation
          "p": 10, // % value
          "i": "bcfb84a7" // variation id (for analytical purposes)
        },
        {
          "o": 1, // rule's order
          "v": false, // value served when the rule is selected during evaluation
          "p": 90, // % value
          "i": "bddac6ae" // variation id (for analytical purposes)
        }
      ],
      "r": [
        // list of Targeting Rules
        {
          "o": 0, // rule's order
          "a": "Identifier", // comparison attribute
          "t": 2, // comparator, possible values:
          // 0  -> 'IS ONE OF',
          // 1  -> 'IS NOT ONE OF',
          // 2  -> 'CONTAINS',
          // 3  -> 'DOES NOT CONTAIN',
          // 4  -> 'IS ONE OF (SemVer)',
          // 5  -> 'IS NOT ONE OF (SemVer)',
          // 6  -> '< (SemVer)',
          // 7  -> '<= (SemVer)',
          // 8  -> '> (SemVer)',
          // 9  -> '>= (SemVer)',
          // 10 -> '= (Number)',
          // 11 -> '<> (Number)',
          // 12 -> '< (Number)',
          // 13 -> '<= (Number)',
          // 14 -> '> (Number)',
          // 15 -> '>= (Number)',
          // 16 -> 'IS ONE OF (Hashed)',
          // 17 -> 'IS NOT ONE OF (Hashed)'
          "c": "@example.com", // comparison value
          "v": true, // value served when the rule is selected during evaluation
          "i": "bcfb84a7" // variation id (for analytical purposes)
        }
      ]
    }
  }
}

Dictionary

You can set up the SDK to load your feature flag & setting overrides from a dictionary.

dictionary = {
    'enabledFeature': True,
    'disabledFeature': False,
    'intSetting': 5,
    'doubleSetting': 3.14,
    'stringSetting': 'test'
}

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        flag_overrides=LocalDictionaryFlagOverrides(source=dictionary, override_behaviour=OverrideBehaviour.LocalOnly)
    )
)

Logging

The ConfigCat SDK uses Python's built-in logging library.

With the logging.basicConfig() method, you can set up the logging system. The following example shows how to set the default root logger to write logs to the standard output with INFO log level.

import logging
logging.basicConfig(level=logging.INFO)

The ConfigCat SDK specifies an internal logger called 'configcat'. The following example shows how to set the Log Level on the internal ConfigCat logger.

logger = logging.getLogger('configcat')
logger.setLevel(logging.INFO)

Available log levels:

Level	Description
ERROR	Only error level events are logged.
WARN	Default. Errors and Warnings are logged.
INFO	Errors, Warnings and feature flag evaluation is logged.
DEBUG	All of the above plus debug info is logged. Debug logs can be different for other SDKs.

Info level logging helps to inspect the feature flag evaluation process:

INFO -- : Evaluating get_value('isPOCFeatureEnabled').
User object:
{
    "Identifier" : "#UNIQUE-USER-IDENTIFIER#",
    "Email" : "[email protected]"
}
Evaluating rule: [Email] [CONTAINS] [@something.com] => no match
Evaluating rule: [Email] [CONTAINS] [@example.com] => match, returning: True

get_all_keys()

You can query the keys from your config JSON in the SDK with the get_all_keys() method.

client = configcatclient.get('#YOUR-SDK-KEY#')
keys = client.get_all_keys()

get_all_values()

Evaluates and returns the values of all feature flags and settings. Passing a User Object is optional.

Parameters	Description
user	Optional, User Object. Essential when using Targeting. Read more about Targeting.

client = configcatclient.get('#YOUR-SDK-KEY#')
all_values = client.get_all_values(User('#UNIQUE-USER-IDENTIFIER#'))  # Optional User Object

get_all_value_details()

Evaluates and returns the detailed values of all feature flags and settings. Passing a User Object is optional.

client = configcatclient.get('#YOUR-SDK-KEY#')
all_value_details = client.get_all_value_details(User('#UNIQUE-USER-IDENTIFIER#'))  # Optional User Object

Custom cache

The ConfigCat SDK stores the downloaded config data in a local cache to minimize network traffic and enhance client performance. If you prefer to use your own cache solution, such as an external or distributed cache in your system, you can implement the ConfigCache interface and set the config_cache parameter in the options passed to configcatclient.get. This allows you to seamlessly integrate ConfigCat with your existing caching infrastructure.

from configcatclient.interfaces import ConfigCache

class InMemoryConfigCache(ConfigCache):
    def __init__(self):
        self._value = {}

    def get(self, key):
        # you should return the cached value
        return self._value.get(key)

    def set(self, key, value):
        # you should cache the new value
        self._value[key] = value

Then use your custom cache implementation:

client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        config_cache=InMemoryConfigCache()
    )
)

info

The Python SDK supports shared caching. You can read more about this feature and the required minimum SDK versions here.

Using ConfigCat behind a proxy

Provide your own network credentials (username/password), and proxy server settings (proxy server/port) by passing the proxy details to the creator method.

proxies = {'https': '127.0.0.1:8080'}
proxy_auth = HTTPProxyAuth('user', 'password')
client = configcatclient.get('#YOUR-SDK-KEY#',
    ConfigCatOptions(
        proxies=proxies,
        proxy_auth=proxy_auth
    )
)

Sample Applications

• Sample Console App
• Django Web App

Guides

See this guide on how to use ConfigCat's Python SDK in a Flask application.

Look under the hood

• ConfigCat's Python SDK on GitHub
• ConfigCat's Python SDK in PyPI