fix(web): recognize markdown file links with angle brackets and parentheses - #5222
fix(web): recognize markdown file links with angle brackets and parentheses#5222yashranaway wants to merge 3 commits into
Conversation
…theses The file-link classifier scanned message text with a single regex that captured fewer CommonMark link destination forms than react-markdown renders. Angle-bracket destinations ([x](</a b/c.tsv>)) captured nothing and bare destinations with balanced parentheses ([x](/tmp/a(1).txt)) were truncated at the first ')', so those links rendered as plain anchors that did nothing when clicked. Replace the regex with extractMarkdownLinkHrefs, a small CommonMark-aware scanner that understands both destination forms, and add normalizeMarkdownLinkHrefKey which decodes percent-encoding so a rendered href (spaces become %20) matches the unencoded destination scanned from the source text. File URIs keep their single-decode semantics. Both helpers move to markdown-links.ts with focused unit tests. Fixes pingdotgg#5158
|
Important Review skippedAuto reviews are disabled on this repository. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Repository UI Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
github-actions
Bot
added
vouch:unvouched
![macroscopeapp[bot]](https://avatars.githubusercontent.com/in/900172?s=60&v=4){kind=small}
Resolve conflicts in apps/web with upstream inline-code file-link work: keep the shared extractMarkdownLinkHrefs/normalizeMarkdownLinkHrefKey helpers (now imported from markdown-links.ts) alongside upstream's extractInlineCodeSpans and resolveInlineCodeFileLinkMeta.
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
ApprovabilityVerdict: Needs human review This PR introduces substantial new markdown link parsing logic. There is an unresolved medium-severity comment identifying that unclosed brackets stop the entire link scan, potentially missing valid links that appear later in text. This correctness concern warrants human review. You can customize Macroscope's approvability policy. Learn more. |
Address review findings on the file-link scanner: - Resolve the link label end across escaped (\]) and balanced nested ([foo [bar]]) brackets instead of stopping at the first ], so those CommonMark links are classified correctly. - Skip the optional link title and closing paren after the destination so a [label](url) sequence embedded inside a title is no longer extracted as a real link. - Decode file:// keys consistently in normalizeMarkdownLinkHrefKey so the pre-scan key (from the raw file:// href) matches the render-time key (from the already-rewritten path). The file path is now resolved from the original href to keep single-decode semantics and avoid double decoding percent-encoded octets. Extend markdown-links tests to cover escaped/nested labels, titles that contain link-looking text, and file:// key matching.
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes using high effort and found 2 potential issues.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit c9550c9. Configure here.
| const labelStart = text.indexOf("[", index); | ||
| if (labelStart === -1) break; | ||
|
|
||
| const labelEnd = findMarkdownLinkLabelEnd(text, labelStart); |
There was a problem hiding this comment.
Unclosed label stops link scan
Medium Severity
When findMarkdownLinkLabelEnd can't find a balanced closing bracket for an opening [, extractMarkdownLinkHrefs stops scanning the entire text. This means any valid inline links appearing later in the message are missed, preventing them from being recognized as file links and causing them to render incorrectly, even if react-markdown parses them fine.
Reviewed by Cursor Bugbot for commit c9550c9. Configure here.
| } | ||
|
|
||
| if (text[pos] === ")") pos += 1; | ||
| return pos; |
There was a problem hiding this comment.
Parenthesized title breaks on first paren
Low Severity
skipMarkdownLinkTitleAndClose treats a parenthesized link title by scanning until the first ), without balancing nested parentheses. Destinations or nested [text](url) inside a (title) stop the scan early, so the scanner can treat title content as later links or advance the index incorrectly compared to quoted-title handling.
Reviewed by Cursor Bugbot for commit c9550c9. Configure here.
1 participant
What Changed
File-link classification in
ChatMarkdownscanned message text with a singleregex that recognized fewer CommonMark link destinations than react-markdown
renders. Angle-bracket destinations (
[x](</a b/c.tsv>)) matched nothing andbare destinations containing balanced parentheses (
[x](/tmp/a(1).txt)) weretruncated at the first
), so both rendered as dead plain anchors.extractMarkdownLinkHrefs, a small CommonMark-awarescanner handling both the angle-bracket and bare (balanced-paren) forms.
normalizeMarkdownLinkHrefKey, decoding percent-encoding so a renderedhref (a space becomes
%20) matches the unencoded destination scanned fromsource text; file URIs keep single-decode semantics.
markdown-links.tswith focused unit tests.Why
Paths agents emit routinely contain spaces, and duplicate downloads produce
names like
report (1).pdf. Those links previously did nothing when clicked.Fixes #5158.
UI Changes
Sending the exact repro from the issue
markdown_file_link_chips_fix_demo.mp4
Checklist
Note
Fix markdown file link detection to support angle-bracket and parenthesized destinations
extractMarkdownLinkHrefsandnormalizeMarkdownLinkHrefKeyfrom markdown-links.tsextractMarkdownLinkHrefsnow parses CommonMark inline link destinations, including angle-bracket form (with spaces) and bare destinations with balanced parentheses, and skips destinations embedded in link titlesnormalizeMarkdownLinkHrefKeyproduces stable lookup keys that align raw markdown destinations with react-markdown-rendered hrefs, includingfile://URIsMacroscope summarized c9550c9.
Note
Low Risk
Localized markdown link parsing and lookup-key normalization in the web chat UI, with broad unit tests and no auth or data-layer changes.
Overview
Chat markdown file links now work for paths with spaces and balanced parentheses in link destinations—cases that previously rendered as plain anchors because pre-scan used a regex that didn’t match CommonMark-style destinations.
extractMarkdownLinkHrefsinmarkdown-links.tsreplaces that regex with a small CommonMark-aware scanner (angle-bracket destinations, bare destinations with balanced(), escaped/nested label brackets, optional titles).normalizeMarkdownLinkHrefKeyaligns pre-scan keys with react-markdown’s renderedhref(e.g.%20vs literal space) and with rewrittenfile://paths, whileChatMarkdownstill resolves file paths from the original href to avoid double-decoding.File link chips use the render-time
hrefin copy markdown when available. Unit tests cover extraction and key normalization.Reviewed by Cursor Bugbot for commit c9550c9. Bugbot is set up for automated code reviews on this repo. Configure here.