fix(web): recognize markdown file links with angle brackets and parentheses #5222
+273
−22
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -108,6 +108,192 @@ export function rewriteMarkdownFileUriHref(href: string | undefined): string | n | |
| return `${target.path}${target.hash}`; | ||
| } | ||
|
|
||
| /** | ||
| * Find the `]` that closes the link label starting at `labelStart` (`[`). | ||
| * | ||
| * Tracks bracket depth and honors backslash escapes so escaped brackets (`\]`, | ||
| * `\[`) and balanced nested runs (`[foo [bar]]`) are handled the way CommonMark | ||
| * does, instead of stopping at the first `]`. Returns -1 when no matching close | ||
| * bracket exists. | ||
| */ | ||
| function findMarkdownLinkLabelEnd(text: string, labelStart: number): number { | ||
| let depth = 0; | ||
| for (let index = labelStart; index < text.length; index += 1) { | ||
| const char = text[index]; | ||
| if (char === "\\") { | ||
| index += 1; | ||
| continue; | ||
| } | ||
| if (char === "[") { | ||
| depth += 1; | ||
| continue; | ||
| } | ||
| if (char === "]") { | ||
| depth -= 1; | ||
| if (depth === 0) return index; | ||
| } | ||
| } | ||
| return -1; | ||
| } | ||
|
|
||
| /** | ||
| * Advance past an inline link's optional title and its closing `)`, starting | ||
| * from the character after the destination. Returns the index just after the | ||
| * closing parenthesis so a `[label](url)` sequence embedded in a title is not | ||
| * mistaken for a real link on the next scan iteration. | ||
| */ | ||
| function skipMarkdownLinkTitleAndClose(text: string, start: number): number { | ||
| const length = text.length; | ||
| const isSpace = (char: string | undefined): boolean => | ||
| char === " " || char === "\t" || char === "\n" || char === "\r"; | ||
|
|
||
| let pos = start; | ||
| while (pos < length && isSpace(text[pos])) pos += 1; | ||
|
|
||
| const opener = text[pos]; | ||
| if (opener === '"' || opener === "'" || opener === "(") { | ||
| const closer = opener === "(" ? ")" : opener; | ||
| pos += 1; | ||
| while (pos < length) { | ||
| const char = text[pos]; | ||
| if (char === undefined) break; | ||
| if (char === "\\" && pos + 1 < length) { | ||
| pos += 2; | ||
| continue; | ||
| } | ||
| if (char === closer) { | ||
| pos += 1; | ||
| break; | ||
| } | ||
| pos += 1; | ||
| } | ||
| while (pos < length && isSpace(text[pos])) pos += 1; | ||
| } | ||
|
|
||
| if (text[pos] === ")") pos += 1; | ||
| return pos; | ||
| } | ||
|
|
||
| /** | ||
| * Extract the destination of every markdown inline link found in `text`. | ||
| * | ||
| * This mirrors the CommonMark link grammar closely enough for file link | ||
| * classification: it resolves the label end across escaped/nested brackets, and | ||
| * understands both the angle-bracket destination form (`[label](<dest with | ||
| * spaces>)`) and the bare form, where the destination may contain balanced | ||
| * parentheses (`[label](/tmp/a(1).txt)`). It also skips the optional title so a | ||
| * `[..](..)` sequence inside a title is not extracted. The previous single-regex | ||
| * implementation captured none of this, so links whose paths held spaces or | ||
| * parentheses were never recognized as file links. | ||
| */ | ||
| export function extractMarkdownLinkHrefs(text: string): string[] { | ||
|
macroscopeapp[bot] marked this conversation as resolved.
|
||
| const hrefs: string[] = []; | ||
| const length = text.length; | ||
| let index = 0; | ||
|
|
||
| while (index < length) { | ||
| const labelStart = text.indexOf("[", index); | ||
| if (labelStart === -1) break; | ||
|
|
||
| const labelEnd = findMarkdownLinkLabelEnd(text, labelStart); | ||
|
Contributor
There was a problem hiding this comment. Unclosed label stops link scanMedium Severity When   Reviewed by Cursor Bugbot for commit c9550c9. Configure here. |
||
| if (labelEnd === -1) break; | ||
|
|
||
| if (text[labelEnd + 1] !== "(") { | ||
| index = labelEnd + 1; | ||
| continue; | ||
| } | ||
|
|
||
| let pos = labelEnd + 2; | ||
| while (pos < length && (text[pos] === " " || text[pos] === "\t")) { | ||
| pos += 1; | ||
| } | ||
|
|
||
| let destination = ""; | ||
|
|
||
| if (text[pos] === "<") { | ||
| // Angle-bracket destination: may contain spaces, ends at an unescaped ">". | ||
| pos += 1; | ||
| let closed = false; | ||
| while (pos < length) { | ||
| const char = text[pos]; | ||
| if (char === undefined) break; | ||
| if (char === "\\" && pos + 1 < length) { | ||
| destination += text[pos + 1]; | ||
| pos += 2; | ||
| continue; | ||
| } | ||
| if (char === "\n") break; | ||
| if (char === ">") { | ||
| closed = true; | ||
| pos += 1; | ||
| break; | ||
| } | ||
| destination += char; | ||
| pos += 1; | ||
| } | ||
| if (!closed) { | ||
| index = labelEnd + 2; | ||
| continue; | ||
| } | ||
| } else { | ||
| // Bare destination: balanced parentheses are part of the destination; it | ||
| // ends at whitespace, a control character, or an unbalanced ")". | ||
| let depth = 0; | ||
| while (pos < length) { | ||
| const char = text[pos]; | ||
| if (char === undefined) break; | ||
| if (char === "\\" && pos + 1 < length) { | ||
| destination += text[pos + 1]; | ||
| pos += 2; | ||
| continue; | ||
| } | ||
| if (char === " " || char === "\t" || char === "\n" || char.charCodeAt(0) < 0x20) { | ||
| break; | ||
| } | ||
| if (char === "(") { | ||
| depth += 1; | ||
| destination += char; | ||
| pos += 1; | ||
| continue; | ||
| } | ||
| if (char === ")") { | ||
| if (depth === 0) break; | ||
| depth -= 1; | ||
| destination += char; | ||
| pos += 1; | ||
| continue; | ||
| } | ||
| destination += char; | ||
| pos += 1; | ||
| } | ||
| } | ||
|
|
||
| const href = destination.trim(); | ||
| if (href.length > 0) hrefs.push(href); | ||
| index = Math.max(skipMarkdownLinkTitleAndClose(text, pos), labelEnd + 2); | ||
| } | ||
|
|
||
| return hrefs; | ||
| } | ||
|
|
||
| /** | ||
| * Canonical lookup key for a markdown link destination. | ||
| * | ||
| * react-markdown percent-encodes some destination characters when it renders a | ||
| * link (a literal space becomes `%20`), while the raw text the link was authored | ||
| * from may contain the unencoded character. Decoding both the rendered href and | ||
| * the destination scanned out of the source text lets the file-link classifier | ||
| * match the two. `file://` destinations are decoded the same way after rewriting | ||
| * so the pre-scan key (built from the raw `file://` href) agrees with the | ||
| * render-time key (built from the already-rewritten path `markdownUrlTransform` | ||
| * hands to the anchor). This key is only for map lookup; the file path itself is | ||
| * resolved separately from the original href to avoid double-decoding. | ||
| */ | ||
| export function normalizeMarkdownLinkHrefKey(href: string): string { | ||
| const normalizedHref = normalizeMarkdownLinkDestination(href); | ||
| return safeDecode(rewriteMarkdownFileUriHref(normalizedHref) ?? normalizedHref); | ||
| } | ||
|
cursor[bot] marked this conversation as resolved.
|
||
|
|
||
| function looksLikePosixFilesystemPath(path: string): boolean { | ||
| if (!path.startsWith("/")) return false; | ||
| if (POSIX_FILE_ROOT_PREFIXES.some((prefix) => path.startsWith(prefix))) return true; | ||
|
|
||
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Parenthesized title breaks on first paren
Low Severity
skipMarkdownLinkTitleAndClosetreats 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.