Cancellation and logging

Cancelling long jobs

Extracting imagery for thousands of assets or running a local model over them takes minutes to hours. Every long-running component accepts a cancel_event, a standard threading.Event, and checks it between assets, tiles or batches. Setting the event from another thread, a GUI button or a signal handler stops the job at the next boundary and raises OperationCancelled, while keeping everything produced so far.

import threading
from rapidtools import AerialImageryExtractor, AssetAnalyzer, OperationCancelled, Pipeline

stop = threading.Event()
pipeline = Pipeline(
    [
        AerialImageryExtractor('ortho.tif', save_directory='crops', cancel_event=stop),
        AssetAnalyzer(model, prompt='...', cancel_event=stop),
    ],
    cancel_event=stop,
)

try:
    buildings = pipeline.run(buildings)
except OperationCancelled:
    print('Stopped by the user; partial results are on the assets.')

Sharing one event between the pipeline and its steps means the pipeline stops between steps and each step stops within itself. The exception derives from RuntimeError, so generic handlers still catch it while specific ones can tell a user-requested stop from a failure. Your own components can join the scheme with raise_if_cancelled().

Logging

Importing rAPIdtools never touches logging configuration. To see progress messages in a script or notebook, call configure_logging() once:

import rapidtools as rt
rt.configure_logging()          # INFO to stdout
rt.configure_logging('DEBUG')

The rapidtools logger then stops propagating to the root logger, so messages print once even when your application configured root logging. Skip the call if you would rather route the library’s messages through your own handlers.

HTTP sessions

Network components share get_configured_session(), a requests.Session with retries, exponential back-off and a default timeout. Use it for your own calls to the same providers so they behave consistently:

from rapidtools.config import get_configured_session
session = get_configured_session(retries=3, backoff_factor=0.5)