fix(ci): normalize Windows CRLF line endings in PR governance script (headroomlabs-ai#1012)

Parideboy · JerrettDavis · web-flow · commit 5194388b6652 · 2026-06-22T18:45:12.000-05:00
## Description

The `CODE_BLOCK_RE` regex in `scripts/pr-governance.py` expects LF after
the opening fenced code block. PR bodies authored on Windows can arrive
with CRLF line endings, which leaves a `\r` before the `\n` and prevents
`has_test_output()` from detecting a valid Test Output block.

This normalizes CRLF to LF once when loading the pull request body,
before section extraction and code-block matching. A regression test now
verifies that a valid PR body with CRLF line endings still passes
governance.

## Type of Change

- [x] Bug fix (non-breaking change that fixes an issue)
- [x] Tests only

## Changes Made

- Normalize Windows CRLF line endings in `scripts/pr-governance.py`
before regex-based validation runs.
- Added `test_validate_pull_request_accepts_crlf_test_output_code_block`
to prevent regressions.

## Testing

- [x] Unit tests pass (`pytest`)
- [x] Linting passes (`ruff check`)
- [x] Formatting passes (`ruff format --check`)
- [x] New tests added for new functionality

### Test Output

```text
python -m pytest scripts/tests/test_pr_governance.py scripts/tests/test_pr_health_labels.py scripts/tests/test_pr_health_workflow.py -q
8 passed in 0.06s

ruff check scripts/pr-governance.py scripts/tests/test_pr_governance.py scripts/tests/test_pr_health_labels.py scripts/tests/test_pr_health_workflow.py
All checks passed!

ruff format --check scripts/pr-governance.py scripts/tests/test_pr_governance.py scripts/tests/test_pr_health_labels.py scripts/tests/test_pr_health_workflow.py
4 files already formatted
```

## Real Behavior Proof

- Environment: local Windows 11 checkout, Python 3.13.13.
- Exact command / steps: Converted the known-valid governance test body
to CRLF line endings and passed it through `validate_pull_request` in
the new regression test.
- Observed result: The report is valid with no problems, proving the
fenced Test Output block is recognized after normalization.
- Not tested: GitHub-hosted Windows PR authoring path end to end; the
unit test covers the exact CRLF body shape consumed by the validator.

## Review Readiness

- [x] I have performed a self-review
- [x] This PR is ready for human review

---------

Co-authored-by: JerrettDavis &lt;mxjerrett@gmail.com&gt;
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -155,7 +155,6 @@ jobs:
       matrix:
         shard: [1, 2, 3, 4]
     env:
-      HF_HUB_OFFLINE: "1"
       TRANSFORMERS_OFFLINE: "1"
     steps:
       - uses: actions/checkout@v6
diff --git a/scripts/pr-governance.py b/scripts/pr-governance.py
@@ -142,6 +142,9 @@ def validate_pull_request(event: dict[str, Any]) -> GovernanceReport:
     is_draft = bool(pull_request.get("draft", False))
     is_bot_pr = author.endswith("[bot]")
     body = pull_request.get("body") or ""
+    # Normalize Windows line endings so regex patterns expecting \n
+    # (particularly the code-block fence regex) match correctly.
+    body = body.replace("\r\n", "\n")
 
     if is_bot_pr:
         summary = "### PR governance\n\nBot-authored PR detected; template enforcement is skipped."
diff --git a/scripts/tests/test_pr_governance.py b/scripts/tests/test_pr_governance.py
@@ -87,6 +87,16 @@ def test_validate_pull_request_marks_ready_pr_valid() -> None:
     assert module.AUTHOR_ACTION_LABEL in report.labels_to_remove
 
 
+def test_validate_pull_request_accepts_crlf_test_output_code_block() -> None:
+    module = _load_module()
+    body = VALID_BODY.replace("\n", "\r\n")
+
+    report = module.validate_pull_request(_event(body))
+
+    assert report.valid is True
+    assert report.problems == []
+
+
 def test_validate_pull_request_allows_draft_without_ready_checkboxes() -> None:
     module = _load_module()
     body = VALID_BODY.replace(
