fix(tui): keep sentence punctuation out of terminal hyperlink targets (#128727)

src/tui/osc8-hyperlinks.test.ts

@@ -8,6 +8,15 @@ describe("extractUrls", () => {
     expect(urls).toEqual(["https://example.com"]);
   });
 
+  it.each([".", ",", ";", "!", "?", ":", "*", "_", "~", ".?!"])(
+    "excludes trailing GFM punctuation %s from bare URLs",
+    (punctuation) => {
+      expect(extractUrls(`Visit https://example.com/path${punctuation}`)).toEqual([
+        "https://example.com/path",
+      ]);
+    },
+  );
+
   it("stops bare URLs before bidi formatting controls", () => {
     const url = "https://example.com/path";
     expect(extractUrls(`مرحبا ${url}\u2069`)).toEqual([url]);
@@ -21,9 +30,9 @@ describe("extractUrls", () => {
     expect(urls).toHaveLength(2);
   });
 
-  it("extracts markdown link hrefs", () => {
-    const urls = extractUrls("[Click here](https://example.com/path)");
-    expect(urls).toEqual(["https://example.com/path"]);
+  it.each(["", ".", "?", ";"])("preserves authored markdown href suffix %s", (suffix) => {
+    const url = `https://example.com/path${suffix}`;
+    expect(extractUrls(`[Click here](${url})`)).toEqual([url]);
   });
 
   it("extracts markdown links with angle brackets and title text", () => {
@@ -64,6 +73,11 @@ describe("extractUrls", () => {
     expect(extractUrls(`[Wikipedia](${url})`)).toEqual([url]);
   });
 
+  it("keeps balanced URL parentheses and internal query punctuation", () => {
+    const url = "https://example.com/v1.0/URL_(disambiguation)?a=1&b=2#section";
+    expect(extractUrls(`Visit ${url}.`)).toEqual([url]);
+  });
+
   it("does not extract an incomplete markdown link destination", () => {
     expect(extractUrls("[broken](https://)")).toEqual([]);
   });
@@ -108,6 +122,41 @@ describe("addOsc8Hyperlinks", () => {
     expect(result[0]).toBe(`Visit \x1b]8;;${url}\x07${url}\x1b]8;;\x07 for info`);
   });
 
+  it.each([".", ",", ";", "!", "?", ":", "*", "_", "~", ".?!"])(
+    "keeps trailing GFM punctuation %s outside terminal hyperlink targets",
+    (punctuation) => {
+      const url = "https://example.com/path";
+      const line = `Visit ${url}${punctuation}`;
+
+      expect(addOsc8Hyperlinks([line], extractUrls(line))).toEqual([
+        `Visit \x1b]8;;${url}\x07${url}\x1b]8;;\x07${punctuation}`,
+      ]);
+    },
+  );
+
+  it("preserves punctuation explicitly authored in markdown hyperlink targets", () => {
+    const url = "https://example.com/path.";
+    const line = `Docs (${url})`;
+
+    expect(addOsc8Hyperlinks([line], extractUrls(`[Docs](${url})`))).toEqual([
+      `Docs (\x1b]8;;${url}\x07${url}\x1b]8;;\x07)`,
+    ]);
+  });
+
+  it.each([".", ","])(
+    "keeps authored and bare hyperlink occurrences distinct before %s",
+    (punctuation) => {
+      const bareUrl = "https://example.com/path";
+      const authoredUrl = `${bareUrl}${punctuation}`;
+      const markdown = `[Docs](${authoredUrl}) and ${bareUrl}${punctuation}`;
+      const rendered = `Docs (${authoredUrl}) and ${bareUrl}${punctuation}`;
+
+      expect(addOsc8Hyperlinks([rendered], extractUrls(markdown))).toEqual([
+        `Docs (\x1b]8;;${authoredUrl}\x07${authoredUrl}\x1b]8;;\x07) and \x1b]8;;${bareUrl}\x07${bareUrl}\x1b]8;;\x07${punctuation}`,
+      ]);
+    },
+  );
+
   it("keeps bidi isolation outside the exact OSC 8 target", () => {
     const url = "https://example.com/path";
     expect(addOsc8Hyperlinks([`\u2067مرحبا ${url}\u2069`], [url])).toEqual([

src/tui/osc8-hyperlinks.ts

@@ -13,11 +13,11 @@ const OSC8_START_RE = new RegExp(`^${OSC8_PATTERN}`);
 const URL_PATH_WITH_PARENS =
   /https?:\/\/[^()\s<>\u061c\u200e\u200f\u202a-\u202e\u2066-\u2069]+(?:\([^()\s<>\u061c\u200e\u200f\u202a-\u202e\u2066-\u2069]*\)[^()\s<>\u061c\u200e\u200f\u202a-\u202e\u2066-\u2069]*)*/g;
 
-/** Strip the suffix starting at a `)` without a matching `(` in the URL.
- *  Bare URLs in prose can pick up a trailing `)` that belongs to surrounding
- *  punctuation, e.g. `(see https://example.com/path)` — the `)` after `path`
- *  and anything after it are sentence punctuation, not part of the URL. */
-function trimUnbalancedTrailingParens(url: string): string {
+/** Strip GFM sentence punctuation and unmatched closing parentheses from bare
+ *  URLs, while preserving balanced parentheses and exact authored Markdown
+ *  destinations. `(see https://example.com/path).` must link only the URL,
+ *  but `[label](https://example.com/path.)` must retain its authored dot. */
+function trimUrlTrailingPunctuation(url: string, knownUrls?: string[]): string {
   let open = 0;
   for (let index = 0; index < url.length; index++) {
     const ch = url[index];
@@ -25,12 +25,16 @@ function trimUnbalancedTrailingParens(url: string): string {
       open++;
     } else if (ch === ")") {
       if (open === 0) {
-        return url.slice(0, index);
+        const authoredUrl = url.slice(0, index);
+        return knownUrls?.includes(authoredUrl)
+          ? authoredUrl
+          : trimUrlTrailingPunctuation(authoredUrl, knownUrls);
       }
       open--;
     }
   }
-  return url;
+  const trimmed = url.replace(/[?!.,:;*_~]+$/u, "");
+  return knownUrls?.includes(url) && !knownUrls.includes(trimmed) ? url : trimmed;
 }
 
 function hasUrlContent(url: string): boolean {
@@ -65,7 +69,7 @@ export function extractUrls(markdown: string): string[] {
   const bareRe =
     /https?:\/\/(?:\[[0-9a-f:.]+\](?::\d+)?[^\s\]>\u061c\u200e\u200f\u202a-\u202e\u2066-\u2069]*|[^\s[\]>\u061c\u200e\u200f\u202a-\u202e\u2066-\u2069]+)/gi;
   while ((m = bareRe.exec(stripped)) !== null) {
-    const url = trimUnbalancedTrailingParens(m[0]);
+    const url = trimUrlTrailingPunctuation(m[0]);
     if (hasUrlContent(url)) {
       urls.add(url);
     }
@@ -134,7 +138,7 @@ function findUrlRanges(
   let match: RegExpExecArray | null;
 
   while ((match = urlRe.exec(visibleText)) !== null) {
-    const fragment = trimUnbalancedTrailingParens(match[0]);
+    const fragment = trimUrlTrailingPunctuation(match[0], knownUrls);
     const start = match.index;
 
     // Resolve fragment to a known URL (exact > prefix > superstring)
@@ -154,7 +158,7 @@ function findUrlRanges(
         nextVisibleText
           ?.trimStart()
           .match(/^[^\s\]>\u061c\u200e\u200f\u202a-\u202e\u2066-\u2069]+/)?.[0] ?? "";
-      const nextFragment = trimUnbalancedTrailingParens(nextToken);
+      const nextFragment = trimUrlTrailingPunctuation(nextToken);
       for (const known of knownUrls) {
         if (!known.startsWith(fragment)) {
           continue;
@@ -171,14 +175,8 @@ function findUrlRanges(
       }
     }
 
-    if (!found) {
-      for (const known of knownUrls) {
-        if (known === fragment) {
-          resolvedUrl = known;
-          found = true;
-          break;
-        }
-      }
+    if (!found && knownUrls.includes(fragment)) {
+      found = true;
     }
     if (!found) {
       let bestLen = 0;