ci: add manual OCR integration workflow - #90
Conversation
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 3be3527b33
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| run: python -m pip install --upgrade pip && python -m pip install -r requirements.txt | ||
|
|
||
| - name: Run real OCR integration tests | ||
| run: python -m pytest -m integration --runintegration --runslow -v |
There was a problem hiding this comment.
Fail the workflow when real OCR returns an error
When model download, initialization, or the actual OCR call fails, this workflow still reports success: the selected integration test calls process_image, which converts every exception into an OCRResult(status="error"), and then explicitly accepts "error" as a valid status in tests/test_engine.py:614-618. Consequently, the new manual workflow cannot provide the advertised real-engine integration proof; the test should require successful OCR output or otherwise propagate the failure.
Useful? React with 👍 / 👎.
What changed
workflow_dispatchjob for the real PaddleOCR integration test;Why
The required OCR CI suite must remain deterministic and fast, while the real
PaddleOCR engine requires model downloads and is slower. This workflow makes
that integration proof reproducible on demand without making pull requests
depend on external model downloads.
Validation
git diff --cached --checkpytestandpydantic-settingsinstalled; the workflow installs the declared dependenciesThis partially addresses #72. The issue remains open for the complete OCR test
matrix, coverage evidence, and remaining contract/integration work.