feat: support environment variables in parsers

[wkpark]

support environment variables in parsers

Checklist

• Changes tested locally
• Code reviewed
• Documentation updated (if necessary)
• Unit tests added (if applicable)

Additional Notes

you can set environment variables for Mineru and Docling.

for example, you can set CUDA_VISIBLE_DEVICES=1 to make the parser run on a specific GPU. This prevents the parser from slowing down your main AI model on the first GPU.

# Use the second GPU (ID: 1) for this parsing task
await rag.process_document_complete(
    file_path="manual.pdf",
    env={"CUDA_VISIBLE_DEVICES": "1"}
)

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[LarFii]

Thanks for the contribution.

One potential regression to consider:

• In MineruParser._run_mineru_command, the new **kwargs makes unknown arguments silently ignored (except env) instead of failing fast. Previously, unexpected arguments would raise a TypeError, which was easier to catch during integration.
• This could lead to “runs successfully but config not applied” scenarios.

Suggestion:

1. Validate and explicitly reject unsupported kwargs (or at least log a warning).
2. Add light validation for env (e.g., ensure mapping of string keys/values).
3. Add tests for:
   • env propagation to subprocess
   • behavior when unknown kwargs are passed

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.