trailwatch 0.3.3

Python SDK for TrailWatch by Kicksaw

• Installation
• Using TrailWatch
  • Decorator
  • Context Manager
• Connectors
  • AWS Connector
  • Salesforce Connector
• Control Execution Status
  • Partial Success
  • Timeout
• Send a File
• Using With Other Decorators

Installation

Install the SDK (supports AWS server):

pip install trailwatch

Install with Salesforce connector support:

pip install trailwatch[salesforce]

Using TrailWatch

Decorator

This is the recommended way to use TrailWatch. Decorator will automatically assign job name and description based on the function name and docstring.

from trailwatch import configure, watch
from trailwatch.connectors.aws import AwsConnectorFactory

configure(
    project="My project name",
    project_description="My project description",
    environment="production",
    connectors=[
        AWSConnectorFactory(
            url="https://<random>.execute-api.us-west-2.amazonaws.com",
            api_key="my_key",
        )
    ],
    loggers=["__main__", "integration"],
)


@watch()
def handler(event, context):
  # Do your thing
  return

Context Manager

Decorator uses this context manager internally. You can use the context manager directly if you need more control over the execution or if you want to report a portion of the execution (code block) as a separate job.

from trailwatch import configure, TrailwatchContext
from trailwatch.connectors.aws import AwsConnectorFactory

configure(
    project="My project name",
    project_description="My project description",
    environment="production",
    connectors=[
        AWSConnectorFactory(
            url="https://<random>.execute-api.us-west-2.amazonaws.com",
            api_key="my_key",
        )
    ],
    loggers=["__main__", "integration"],
)


def handler(event, context):
    # Other code
    with TrailwatchContext(
        job="My Job",
        job_description="My job description",
    ) as execution:
        # Do your thing
        return
    # Other code

Connectors

TwailWatch SDK works by attaching connectors to the execution context. Connectors are responsible for tracking execution flow and for creating a record in their respective systems. Connectors should be configured using the configure function by provividing a list of connector factories in the connectors argument.

AWS Connector

AWS connector is used to send execution information to AWS TrailWatch service deployed in client's AWS account. To use AWS connector, you will need to deploy the TrailWatch service first and then obtain URL and API key from the service.

from trailwatch.connectors.aws import AwsConnectorFactory

configure(
    # Other configuration parameters
    connectors=[
        AWSConnectorFactory(
            url="url",
            api_key="key",
        )
    ],
)

Salesforce Connector

Salesforce connector is used to send execution information to Kicksaw Integration App deployed to client's Salesforce org. To use Salesforce connector, you will need to deploy the Kicksaw Integration App first and then obtain credentials required to sign in to the Salesforce org.

from trailwatch.connectors.salesforce import SalesforceConnectorFactory

configure(
    # Other configuration parameters
    connectors=[
        SalesforceConnectorFactory(
            username="username",
            password="password",
            security_token="token",
            domain="domain",
        )
    ],
)

Control Execution Status

Partial Success

Raise a PartialSuccess exception to indicate that the execution was partially successful. This exception is handled by TrailWatch to set execution status to partial and will not be propagated to the caller.

from trailwatch.exceptions import PartialSuccessError


@watch()
def handler(event, context):
    # Do your thing
    # You find out that only a subset of the work was successful
    # Log information about the failure normally using the logger
    raise PartialSuccessError

Timeout

You can set timeout on a function to force it to stop after a certain amount of time. This will raise TimeoutError and set the execution status to timeout.

:warning: When using timeout inside AWS Lambda or Azure Function you will need to set the TrailWatch timeout to a value lower than the timeout you set on the cloud function. Otherwise, function can timeout before TrailWatch has a chance to set the status.

@watch(timeout=10)
def handler(event, context):
    # Do something that takes more than 10 seconds
    ...

Or using the context manager:

def handler(event, context):
    with TrailwatchContext(
        job="My Job",
        job_description="My job description",
        timeout=10,
    ) as execution:
        # Do something that takes more than 10 seconds
        ...

Send a File

Some connectors support attaching (sending) files to be associated with the execution. To send a file, use the send_file method on the TrailwatchContext object. Three methods are available to send a file:

• send_file - send a file from the local filesystem
• send_fileobj - send a file-like object (stream or open file object)
• send_file_content - send a file content as a string or bytes

def handler(event, context):
    with TrailwatchContext(
        job="My Job",
        job_description="My job description",
    ) as execution:
        execution.send_file("my_file.txt", "/path/to/my_file.txt")
        execution.send_fileobj("my_file.txt", open("my_file.txt", "rb"))
        execution.send_file_content("my_file.txt", "Hello from file!")

When using the decorator, you will need to add an extra argument trailwatch_execution_context to the function you are decorating to receive the TrailwatchContext object.

from trailwatch import TrailwatchContext

@watch()
def handler(event, context, trailwatch_execution_context: TrailwatchContext):
    trailwatch_execution_context.send_file("my_file.txt", "/path/to/my_file.txt")

Connectors supporting sending files:

• AWS Connector

Using With Other Decorators

When using TrailWatch with other decorators, make sure that TrailWatch decorator is the innermost decorator. This is because TrailWatch decorator may inject the context argument into the function and other decorators (like FastAPI) can get confused by this argument.

from fastapi import FastAPI, Path
from pydantic import BaseModel

app = FastAPI()


class Model(BaseModel):
    name: str


@app.post("/my_endpoint/{resource_id}")
@watch()
def my_endpoint(
    trailwatch_execution_context: TrailwatchContext
    model: Model,
    resource_id: str = Path(..., title="The ID of the resource to get"),
):
    # Do your thing
    return