ARY / implementation_plan.md
harshitmahour360's picture
Initial commit for stage 3
e331bc2
|
Raw History Blame Contribute Delete
2.14 kB

Implementation Plan: Production-Grade Logging

Replace all print() statements with structured logging using the Python logging module to meet production readiness standards.

User Review Required

The format of the output in inference.py will change from plain text to structured log lines (e.g., 2026-04-03 14:45:00 [INFO] [inference.py] ...). If you use automated tools to parse the "START/STEP/END" markers, they should be updated to handle the log prefix.

Proposed Changes

SupportDesk Environment Core

[NEW] logging_config.py

  • Create a shared get_logger(name) utility.
  • Set a default level of INFO, configurable via the LOG_LEVEL environment variable.
  • Format: %(asctime)s [%(levelname)s] [%(name)s] - %(message)s.

Agent & Utilities

[MODIFY] inference.py

  • Import and initialize the logger.
  • Update _print_start, _print_step, and _print_end to use logger.info().
  • Replace all other print() calls with appropriate levels (info, error, warning).

[MODIFY] smoke_test.py

  • Replace progress prints with logger.info().
  • Replace error prints with logger.error().

[MODIFY] validate.py

  • Replace informative prints with logging.
  • Note: Keep the final JSON result as a print() if it's used for shell piping/automation, or transition to a structured log.

Open Questions

  1. Output Level: Should we default the log level to DEBUG for more verbosity during development, or keep it at INFO?
  2. Clean Logs: Would you like the "OpenEnv" markers ([START], [STEP]) to remain "clean" without the timestamp prefix for easier parsing?

Verification Plan

Automated Tests

  • Run pytest to ensure logging integration doesn't break logic.
  • Run python inference.py --offline to verify log appearance.

Manual Verification

  • Run python smoke_test.py and verify terminal output.