The GitHub Actions job "CI" on rocketmq-dashboard.git/fix/ai-markdown-thematic-break has succeeded. Run started by GitHub user lizhimins (triggered by lizhimins).
Head commit for run: c5d6740f82563e1836acd6fb704d5293492fe431 / lizhimins <[email protected]> fix(ai): guard the heading and bullet repairs against well-formed markers The upstream PR guarded the two repair regexes in the old whole-content, /gm implementation. Trunk has since rewritten normalizeAiMarkdown into a fence-aware line-by-line pass, so those two literals no longer exist there and the cherry-pick could not carry the change over. Port the same two guards onto the line-by-line version instead of rebasing the contributor branch: - `^(#{1,6})(?=\S)` -> `^(#{1,6})(?=[^\s#])`: a plain (?=\S) lookahead lets the engine backtrack `#{1,6}`, so a WELL-FORMED `## heading` matched as one `#` followed by `#` and was rewritten to `# # heading`, corrupting every level-2+ heading in an assistant answer. - `^([-+*])(?=\S)` -> `^([-+*])(?!\1\1)(?=\S)`: `---`/`+++`/`***` are thematic breaks (and `---` under a title line is a setext heading); they must not be rewritten into a bullet with stray markers. The `g`/`m` flags stay off - the line-by-line pass already anchors per line. The fence-info-string repair is untouched. Verified with the PR's new markdown.test.ts (4 cases) and trunk's markdownFences.test.ts (10 cases). Report URL: https://github.com/apache/rocketmq-dashboard/actions/runs/39247990993 With regards, GitHub Actions via GitBox
