Logging#

Structured logging for SpectoPrep, built on structlog.

Call configure_logging() once at application/CLI entry to install the processor chain, then obtain loggers anywhere with get_logger(). Library modules only call get_logger(); they never configure logging at import time, so importing SpectoPrep never mutates a host application’s logging setup.

Examples

>>> from spectoprep.logging import configure_logging, get_logger
>>> configure_logging(level="INFO")
>>> log = get_logger("example")
>>> log.info("cv_evaluated", rmse=0.12, r2=0.98)
spectoprep.logging.configure_logging(level='INFO', json_logs=False, force=False)[source]#

Configure structlog and the standard-library logging bridge.

The level is always validated. If logging has already been configured (for example by an application entry point) this call becomes a no-op unless force=True, so a library component instantiated later never overrides an explicit application configuration.

Parameters:

  • level (str, default="INFO") – Minimum level name (“DEBUG”, “INFO”, “WARNING”, “ERROR”, “CRITICAL”).

  • json_logs (bool, default=False) – Emit newline-delimited JSON (machine-readable) when True, otherwise a colourised, human-readable console renderer.

  • force (bool, default=False) – Reconfigure even if logging was already configured.

Raises:

ValueError – If level is not a recognised logging level name.

Return type:

None

spectoprep.logging.get_logger(name=None)[source]#

Return a structlog logger.

This never configures logging as a side effect, so importing SpectoPrep stays inert. structlog returns a lazy proxy that binds to whatever configuration is active at first log call; if configure_logging() was never called, structlog’s built-in defaults apply.

Parameters:

name (str, optional) – Logger name, bound as the logger key on every event.

Return type:

Any