feat(cli): support MARKITDOWN_CU_ENDPOINT and MARKITDOWN_DOCINTEL_ENDPOINT env vars

[Rafli Surya Wijaya (loulanyue)]

Summary

Adds environment-variable fallbacks for Azure Content Understanding and Document Intelligence endpoints in the CLI:

• --cu-endpoint falls back to MARKITDOWN_CU_ENDPOINT
• -e / --endpoint falls back to MARKITDOWN_DOCINTEL_ENDPOINT

Fixes #2326

Rationale

Credentials (AZURE_API_KEY or DefaultAzureCredential) are already ambiently resolved from the environment. Allowing endpoints to be configured via environment variables enables operator-configured deployments and clean tool invocations for automated agents and scripts without needing resource URLs hardcoded in command-line arguments.

Explicit CLI flags (--cu-endpoint, -e / --endpoint) take precedence over environment variables.

Changes

• packages/markitdown/src/markitdown/__main__.py:
  • Resolved docintel_endpoint = args.endpoint or os.environ.get("MARKITDOWN_DOCINTEL_ENDPOINT").
  • Resolved cu_endpoint = args.cu_endpoint or os.environ.get("MARKITDOWN_CU_ENDPOINT").
  • Updated error messages and argument help text to mention the environment variables.
• packages/markitdown/tests/test_cli_misc.py:
  • Added unit tests for missing endpoint error messages and environment variable fallbacks.

Verification

• Ran pytest packages/markitdown/tests/test_cli_misc.py -v:

  test_version PASSED
  test_invalid_flag PASSED
  test_docintel_missing_endpoint_error PASSED
  test_docintel_env_endpoint_fallback PASSED
  test_cu_missing_endpoint_error PASSED
  test_cu_env_endpoint_fallback PASSED
  6 passed in 3.56s
  

[Rafli Surya Wijaya (loulanyue)]

@microsoft-github-policy-service agree