Skip to content

absmartly/python3-sdk

Repository files navigation

A/B Smartly SDK

A/B Smartly - Python SDK

Compatibility

The A/B Smartly Python SDK is compatible with Python 3. It provides both a blocking and an asynchronous interfaces.

Getting Started

Install the SDK

pip install absmartly==0.2.3

Dependencies

setuptools~=60.2.0  
requests~=2.28.1  
urllib3~=1.26.12  
jsons~=1.6.3

Import and Initialize the SDK

Once the SDK is installed, it can be initialized in your project.

    def main():
        client_config = ClientConfig()
        client_config.endpoint = "https://sandbox.test.io/v1"
        client_config.api_key = "test"
        client_config.application = "www"
        client_config.environment = "prod"
    
        default_client_config = DefaultHTTPClientConfig()
        default_client = DefaultHTTPClient(default_client_config)
        sdk_config = ABSmartlyConfig()
        sdk_config.client = Client(client_config, default_client)
        sdk = ABSmartly(sdk_config)
    
        context_config = ContextConfig()
        ctx = sdk.create_context(context_config)

SDK Options

Config Type Required? Default Description
endpoint string undefined The URL to your API endpoint. Most commonly "your-company.absmartly.io"
apiKey string undefined Your API key which can be found on the Web Console.
environment "production" or "development" undefined The environment of the platform where the SDK is installed. Environments are created on the Web Console and should match the available environments in your infrastructure.
application string undefined The name of the application where the SDK is installed. Applications are created on the Web Console and should match the applications where your experiments will be running.
max_retries number 5 The number of retries before the SDK stops trying to connect.
connection_timeout number 3 An amount of time, in seconds, before the SDK will stop trying to connect.
context_event_logger (self, event_type: EventType, data: object) See "Using a Custom Event Logger" below A callback function which runs after SDK events.

Using a custom Event Logger

The A/B Smartly SDK can be instantiated with an event logger used for all contexts. In addition, an event logger can be specified when creating a particular context, in the ContextConfig.

    class EventType(Enum):
        ERROR = "error"
        READY = "ready"
        REFRESH = "refresh"
        PUBLISH = "publish"
        EXPOSURE = "exposure"
        GOAL = "goal"
        CLOSE = "close"
    
    
    class ContextEventLogger:
    
        @abstractmethod
        def handle_event(self, event_type: EventType, data: object):
            raise NotImplementedError

The data parameter depends on the type of event. Currently, the SDK logs the following events:

event when data
Error Context receives an error Throwable object
Ready Context turns ready ContextData used to initialize the context
Refresh Context.refresh() method succeeds ContextData used to refresh the context
Publish Context.publish() method succeeds PublishEvent sent to the A/B Smartly event collector
Exposure Context.getTreatment() method succeeds on first exposure Exposure enqueued for publishing
Goal Context.track() method succeeds GoalAchievement enqueued for publishing
Close Context.close() method succeeds the first time null

Create a New Context Request

Synchronously

# define a new context request
    context_config = ContextConfig()
    context_config.publish_delay = 10
    context_config.refresh_interval = 5

    context_config = ContextConfig()
    ctx = sdk.create_context(context_config)
    ctx.wait_until_ready()

Asynchronously

# define a new context request
    context_config = ContextConfig()
    context_config.publish_delay = 10
    context_config.refresh_interval = 5

    context_config = ContextConfig()
    ctx = sdk.create_context(context_config)
    ctx.wait_until_ready_async()

With Prefetched Data

# define a new context request
    context_config = ContextConfig()
    context_config.publish_delay = 10
    context_config.refresh_interval = 5
    context_config.units = {"session_id": "bf06d8cb5d8137290c4abb64155584fbdb64d8",
                            "user_id": "12345"}
    
    context_config = ContextConfig()
    ctx = sdk.create_context(context_config)
    ctx.wait_until_ready_async()

Refreshing the Context with Fresh Experiment Data For long-running contexts, the context is usually created once when the application is first started. However, any experiments being tracked in your production code, but started after the context was created, will not be triggered. To mitigate this, we can use the set_refresh_interval() method on the context config.

    default_client_config = DefaultHTTPClientConfig()
    default_client_config.refresh_interval = 5

Alternatively, the refresh() method can be called manually. The refresh() method pulls updated experiment data from the A/B Smartly collector and will trigger recently started experiments when get_treatment() is called again.

    context.refresh()

Setting Extra Units You can add additional units to a context by calling the set_unit() or the set_units() method. This method may be used for example, when a user logs in to your application, and you want to use the new unit type to the context. Please note that you cannot override an already set unit type as that would be a change of identity, and will throw an exception. In this case, you must create a new context instead. The SetUnit() and SetUnits() methods can be called before the context is ready.

    context.set_unit("db_user_id", "1000013")
    
    context.set_units({
            "db_user_id": "1000013"
    })

Basic Usage

Selecting a treatment

    res, _ = context.get_treatment("exp_test_experiment")
    if res == 0:
            # user is in control group (variant 0)
    else:
           # user is in treatment group

Treatment Variables

     res = context.get_variable_value(key, 17)

Peek at treatment variants

Although generally not recommended, it is sometimes necessary to peek at a treatment or variable without triggering an exposure. The A/B Smartly SDK provides a peek_treament() method for that.

    res = context.peek_treament("exp_test_experiment")
    if res == 0:
        # user is in control group (variant 0)
    else:
        # user is in treatment group
Peeking at variables
    variable = context.peek_variable("my_variable")

Overriding treatment variants

During development, for example, it is useful to force a treatment for an experiment. This can be achieved with the override() and/or overrides() methods. The set_override() and set_overrides() methods can be called before the context is ready.

    context.set_override("exp_test_experiment", 1) # force variant 1 of treatment
    context.set_overrides({
        "exp_test_experiment": 1,
        "exp_another_experiment": 0
    })

Advanced

Context Attributes

Attributes are used to pass meta-data about the user and/or the request. They can be used later in the Web Console to create segments or audiences. The set_attributes() and set_attributes() methods can be called before the context is ready.

    context.set_attributes("user_agent", req.get_header("User-Agent"))
    
    context.set_attributes({
            "customer_age": "new_customer"
    })

Custom Assignments

Sometimes it may be necessary to override the automatic selection of a variant. For example, if you wish to have your variant chosen based on data from an API call. This can be accomplished using the set_custom_assignment() method.

    context.set_custom_assignment("exp_test_not_eligible", 3)

If you are running multiple experiments and need to choose different custom assignments for each one, you can do so using the set_custom_assignments() method.

    context.set_custom_assignments({"db_user_id2": 1})

Publish

Sometimes it is necessary to ensure all events have been published to the A/B Smartly collector, before proceeding. You can explicitly call the publish() or publish_async() methods.

    context.publish()

Finalize

The close() and close_async() methods will ensure all events have been published to the A/B Smartly collector, like publish(), and will also "seal" the context, throwing an error if any method that could generate an event is called.

    context.close()

Tracking Goals

Goals are created in the A/B Smartly web console.

    context.track("payment", {
            "item_count": 1,
            "total_amount": 1999.99
    })

About A/B Smartly

A/B Smartly is the leading provider of state-of-the-art, on-premises, full-stack experimentation platforms for engineering and product teams that want to confidently deploy features as fast as they can develop them. A/B Smartly's real-time analytics helps engineering and product teams ensure that new features will improve the customer experience without breaking or degrading performance and/or business metrics.

Have a look at our growing list of clients and SDKs: