shuke987 opened a new pull request, #68657: URL: https://github.com/apache/doris/pull/68657
### What problem does this PR solve? Issue Number: N/A Related PR: #68599 Problem Summary: In [the review run for #67395](https://github.com/apache/doris/actions/runs/36549706235), a temporary greedy extraction captured subsequent findings and internal ledger sections in each inline body. Those bodies still passed the submission schema. Editing the posted comments afterward made them differ from the frozen submission, so final verification correctly failed. Add a local `submit_review.py --check-only` mode that reuses the existing schema validation and reports each comment's path, diff position, character/line counts, bounded start/end previews, and advisory warnings for long bodies or known ledger markers. It only reads the input file: it does not access GitHub, read the run context, or freeze a submission. Warnings do not reject or truncate legitimate long SQL/code examples or quoted ledger text. Update the runner prompt to assemble explicitly bounded finding text, run the preflight, inspect every complete body in the actual JSON, and rerun after edits before submitting. It also explains why directly editing posted bodies cannot repair a frozen submission. This is a preventive diagnostic, not a guarantee of semantic correctness or an amendment protocol. Review notes: - Scope and compatibility: three files; the JSON schema and existing submission command remain compatible. The new flag takes effect per invocation. - Lifecycle, concurrency and persistence: the preflight has no write or network side effects. Exclusive declaration, duplicate prevention, exact readback, and resume verification remain on the existing submission path. - Conditions, observability and performance: warnings have concrete extraction-failure signals and bounded diagnostic previews; processing is linear in the input size. The actual payload is unchanged. - Coverage: regression cases reproduce the greedy extraction and exercise malformed inputs, absent/existing context, correction and recheck, the existing CLI submission path, and legitimate long multiline bodies. ### Release note None ### Check List (For Author) - Test - [ ] Regression test - [x] Unit Test - [x] Manual test (details below) - [ ] No need to test or manual test. Validation: - `python3 -B -m unittest discover -s .github/scripts -p 'test_submit_review.py'`: 27 passed. - `python3 -B -m unittest discover -s .github/scripts -p 'test_*review*.py'`: 163 tests, 144 passed and 19 skipped on macOS for Linux process supervision. - Parsed the workflow with Ruby YAML, checked Python syntax, and ran `git diff --check`. - No live review was posted during validation; GitHub submission/readback tests use the existing mocked API. - Behavior changed: - [ ] No. - [x] Yes. CI review instructions now require local preflight and full-body inspection before submission. - Does this need documentation? - [x] No. The runner prompt and CLI help describe the added mode. - [ ] Yes. ### Check List (For Reviewer who merge this PR) - [ ] Confirm the release note - [ ] Confirm test cases - [ ] Confirm document - [ ] Add branch pick label -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
