fix(cli): stop bare-URL hyperlinks at full-width CJK punctuation

packages/cli/src/ui/utils/InlineMarkdownRenderer.test.tsx

@@ -5,6 +5,7 @@
  */
 
 import { afterEach, beforeEach, describe, expect, it } from 'vitest';
+import stripAnsi from 'strip-ansi';
 import { renderWithProviders } from '../../test-utils/render.js';
 import { getPlainTextLength, RenderInline } from './InlineMarkdownRenderer.js';
 import { HYPERLINK_ENV_KEYS } from './osc8.js';
@@ -248,6 +249,45 @@ math then literal: $x^2\$$`;
       expect(out).not.toContain(`\x1b]8;;${url}.\x07`);
     });
 
+    it('stops the bare-URL hyperlink at glued-on full-width CJK punctuation', () => {
+      enableHyperlinks();
+      const url = 'https://github.com/QwenLM/qwen-code/pull/8742';
+      const { lastFrame } = renderWithProviders(
+        <RenderInline text={`PR：${url}（2 commits，等 CI）`} />,
+      );
+
+      const out = lastFrame() ?? '';
+      // OSC 8 target is exactly the URL — the （2 run is not swallowed into
+      // the link (https://github.com/QwenLM/qwen-code/issues/8750).
+      expect(out).toContain(`\x1b]8;;${url}\x07`);
+      expect(out).not.toContain(`\x1b]8;;${url}（`);
+      // The glued-on punctuation renders as plain text after the link.
+      expect(out.replace(/\s+/g, ' ')).toContain('（2 commits，等 CI）');
+    });
+
+    it('does not treat underscores around a later URL as emphasis', () => {
+      enableHyperlinks();
+      const firstUrl = 'https://a.com/x';
+      const secondUrl = 'https://b.com/y';
+      const { lastFrame } = renderWithProviders(
+        <RenderInline text={`见 ${firstUrl}（说明_1）和 ${secondUrl}_。`} />,
+      );
+
+      const out = lastFrame() ?? '';
+      expect(out).toContain(`\x1b]8;;${firstUrl}\x07`);
+      expect(out).toContain(`\x1b]8;;${secondUrl}\x07`);
+      expect(out.replace(/\s+/g, ' ')).toContain('（说明_1）和');
+      expect(stripAnsi(out)).toContain(`${secondUrl}_`);
+    });
+
+    it('preserves dunder identifiers as visible text', () => {
+      const { lastFrame } = renderWithProviders(
+        <RenderInline text="Python 的 __init__ 方法" />,
+      );
+
+      expect(stripAnsi(lastFrame() ?? '')).toContain('__init__');
+    });
+
     it('leaves bare URLs unwrapped when unsupported', () => {
       const url = 'https://example.com/plain';
       const { lastFrame } = renderWithProviders(

packages/cli/src/ui/utils/InlineMarkdownRenderer.tsx

@@ -11,6 +11,7 @@ import stringWidth from 'string-width';
 import { createDebugLogger } from '@qwen-code/qwen-code-core';
 import { renderInlineLatex } from './latexRenderer.js';
 import {
+  BARE_URL_PATTERN,
   MD_LINK_CAPTURE,
   MD_LINK_PATTERN,
   isSafeOscScheme,
@@ -37,8 +38,8 @@ const INLINE_CODE_MARKER_LENGTH = 1; // For "`"
 const UNDERLINE_TAG_START_LENGTH = 3; // For "<u>"
 const UNDERLINE_TAG_END_LENGTH = 4; // For "</u>"
 const INLINE_MARKDOWN_REGEX = new RegExp(
-  String.raw`(\*\*.*?\*\*|\*.*?\*|_.*?_|~~.*?~~|${MD_LINK_PATTERN}|` +
-    String.raw`${INLINE_CODE_SPAN_PATTERN_SOURCE}|<u>.*?<\/u>|https?:\/\/\S+)`,
+  String.raw`(\*\*.*?\*\*|\*.*?\*|(?<![\w\u3400-\u9fff])_(?!_)[^_]*_(?![\w\u3400-\u9fff])|~~.*?~~|${MD_LINK_PATTERN}|` +
+    String.raw`${INLINE_CODE_SPAN_PATTERN_SOURCE}|<u>.*?<\/u>|${BARE_URL_PATTERN})`,
   'g',
 );
 
@@ -249,7 +250,10 @@ const RenderInlineInternal: React.FC<RenderInlineProps> = ({
         // alternative is anchored on `https?://`, so `isSafeOscScheme` is
         // redundant but kept as a cheap defense-in-depth assertion.
         const trimmedUrl = canHyperlink
-          ? trimTrailingUrlPunctuation(fullMatch)
+          ? trimTrailingUrlPunctuation(
+              fullMatch,
+              text[index + fullMatch.length],
+            )
           : fullMatch;
         const wrapOsc8 = canHyperlink && isSafeOscScheme(trimmedUrl);
         renderedNode = (

packages/cli/src/ui/utils/TableRenderer.test.tsx

@@ -649,6 +649,38 @@ describe('<TableRenderer />', () => {
       expectAllLinesToHaveSameVisibleWidth(output);
     });
 
+    it('stops a bare URL before glued-on CJK punctuation', () => {
+      enableHyperlinks();
+      const url = 'https://github.com/QwenLM/qwen-code/pull/8742';
+      const suffix = '（2 commits，等 CI）';
+      const output = renderTable(['PR'], [[`PR：${url}${suffix}`]], 100);
+      expect(output).toContain(`\x1b]8;;${url}\x07`);
+      expect(output).not.toContain(`\x1b]8;;${url}（`);
+      expect(stripAnsi(output)).toContain(suffix);
+    });
+
+    it('does not treat underscores around a later URL as emphasis', () => {
+      enableHyperlinks();
+      const firstUrl = 'https://a.com/x';
+      const secondUrl = 'https://b.com/y';
+      const output = renderTable(
+        ['PR'],
+        [[`见 ${firstUrl}（说明_1）和 ${secondUrl}_。`]],
+        100,
+      );
+
+      expect(output).toContain(`\x1b]8;;${firstUrl}\x07`);
+      expect(output).toContain(`\x1b]8;;${secondUrl}\x07`);
+      expect(stripAnsi(output)).toContain('（说明_1）和');
+      expect(stripAnsi(output)).toContain(`${secondUrl}_。`);
+    });
+
+    it('preserves dunder identifiers as visible text', () => {
+      const output = renderTable(['Value'], [['Python 的 __init__ 方法']], 60);
+
+      expect(stripAnsi(output)).toContain('__init__');
+    });
+
     it('falls back to legacy `label (url)` in cells on unsupported terminals', () => {
       // isTTY=false from the suite-wide beforeEach disables hyperlinks.
       const url = 'https://example.com/page';

packages/cli/src/ui/utils/TableRenderer.tsx

@@ -19,6 +19,7 @@ import {
   unescapeMarkdownDollars,
 } from './inline-math.js';
 import {
+  BARE_URL_PATTERN,
   MD_LINK_CAPTURE,
   MD_LINK_PATTERN,
   isSafeOscScheme,
@@ -49,8 +50,8 @@ const ABSOLUTE_MIN_HORIZONTAL_TABLE_WIDTH = 24;
 const SAFETY_MARGIN = 4;
 
 const INLINE_MARKDOWN_REGEX = new RegExp(
-  String.raw`(\*\*.*?\*\*|\*.*?\*|_.*?_|~~.*?~~|${MD_LINK_PATTERN}|` +
-    String.raw`${INLINE_CODE_SPAN_PATTERN_SOURCE}|<u>.*?<\/u>|https?:\/\/\S+)`,
+  String.raw`(\*\*.*?\*\*|\*.*?\*|(?<![\w\u3400-\u9fff])_(?!_)[^_]*_(?![\w\u3400-\u9fff])|~~.*?~~|${MD_LINK_PATTERN}|` +
+    String.raw`${INLINE_CODE_SPAN_PATTERN_SOURCE}|<u>.*?<\/u>|${BARE_URL_PATTERN})`,
   'g',
 );
 
@@ -371,7 +372,10 @@ function renderMarkdownToAnsi(text: string, enableInlineMath = false): string {
     } else if (/^https?:\/\//.test(fullMatch)) {
       const visible = applyColor(fullMatch, theme.text.link);
       if (canHyperlink) {
-        const trimmedUrl = trimTrailingUrlPunctuation(fullMatch);
+        const trimmedUrl = trimTrailingUrlPunctuation(
+          fullMatch,
+          text[index + fullMatch.length],
+        );
         rendered = isSafeOscScheme(trimmedUrl)
           ? `${osc8Open(trimmedUrl)}${visible}${osc8Close()}`
           : visible;

packages/cli/src/ui/utils/osc8.test.ts

@@ -6,6 +6,7 @@
 
 import { afterEach, beforeEach, describe, expect, it } from 'vitest';
 import {
+  BARE_URL_PATTERN,
   HYPERLINK_ENV_KEYS,
   isSafeOscScheme,
   labelMayDeceive,
@@ -294,6 +295,46 @@ describe('osc8 helpers', () => {
     });
   });
 
+  describe('BARE_URL_PATTERN', () => {
+    it.each([
+      'https://ja.wikipedia.org/wiki/人々',
+      'https://ja.wikipedia.org/wiki/〆切',
+      'https://example.com/二〇二六年報',
+      'https://ja.example.com/〼道と〻次',
+      'https://example.com/〱〢',
+      'https://example.com/a︳b︴c',
+      'https://zh.wikipedia.org/wiki/北京',
+    ])('keeps word-forming CJK characters in %s', (url) => {
+      expect(new RegExp(BARE_URL_PATTERN).exec(url)?.[0]).toBe(url);
+    });
+
+    it.each([
+      'https://en.wikipedia.org/wiki/Mexico–United_States_border',
+      'https://example.com/it’s',
+      'https://example.com/thing…',
+    ])('keeps typographic punctuation inside %s', (url) => {
+      expect(new RegExp(BARE_URL_PATTERN).exec(url)?.[0]).toBe(url);
+    });
+
+    it.each([
+      ['https://example.com/page。下一步', 'https://example.com/page'],
+      ['https://example.com/page、続き', 'https://example.com/page'],
+      ['https://example.com/page：説明', 'https://example.com/page'],
+      ['https://example.com/page？質問', 'https://example.com/page'],
+      ['https://example.com/page！注意', 'https://example.com/page'],
+    ])('stops before CJK punctuation in %s', (input, expected) => {
+      expect(new RegExp(BARE_URL_PATTERN).exec(input)?.[0]).toBe(expected);
+    });
+
+    it.each([
+      ['https://example.com/x﹙備註﹚', 'https://example.com/x'],
+      ['https://example.com/a︒後続', 'https://example.com/a'],
+      ['https://example.com/p﹖説明', 'https://example.com/p'],
+    ])('stops before CJK compat/vertical forms in %s', (input, expected) => {
+      expect(new RegExp(BARE_URL_PATTERN).exec(input)?.[0]).toBe(expected);
+    });
+  });
+
   describe('supportsHyperlinks', () => {
     function setTTY(value: boolean) {
       Object.defineProperty(process.stdout, 'isTTY', {

packages/cli/src/ui/utils/osc8.ts

@@ -88,6 +88,11 @@ export function isSafeOscScheme(url: string): boolean {
   return SAFE_OSC8_SCHEMES.has(match[1]!.toLowerCase());
 }
 
+const BARE_URL_BREAK_CHARACTERS = String.raw`\u3001-\u3004\u3008-\u3020\u302e-\u3030\u3036-\u3037\u303d-\u303f\uff01-\uff0f\uff1a-\uff20\uff3b-\uff40\uff5b-\uff65\ufe10-\ufe1f\ufe30-\ufe32\ufe35-\ufe6f`;
+// The escaped CJK ranges do not contain literal combining characters.
+// eslint-disable-next-line no-misleading-character-class
+const BARE_URL_BREAK_PATTERN = new RegExp(`[${BARE_URL_BREAK_CHARACTERS}]`);
+
 /**
  * Trim trailing sentence punctuation off a bare URL run before it becomes
  * an OSC 8 target. Models routinely produce `see https://example.com.` and
@@ -101,10 +106,13 @@ export function isSafeOscScheme(url: string): boolean {
  * the URL so URLs that legitimately end with `)` (Wikipedia disambiguation,
  * MSDN) aren't truncated.
  */
-export function trimTrailingUrlPunctuation(url: string): string {
-  // Count `( [ {` opens once up-front; we then decrement running `)`/`]`/`}`
-  // close counts as we trim, keeping the whole trim O(n) instead of O(n²)
-  // for adversarial inputs like `https://x.com))))…`.
+export function trimTrailingUrlPunctuation(
+  url: string,
+  nextCharacter = '',
+): string {
+  // Count `( [ {` opens once up-front; we then
+  // decrement running `)`/`]`/`}` close counts as we trim, keeping the whole
+  // trim O(n) instead of O(n²) for adversarial inputs like `https://x.com))))…`.
   let openParen = 0;
   let openBracket = 0;
   let openBrace = 0;
@@ -122,6 +130,12 @@ export function trimTrailingUrlPunctuation(url: string): string {
   }
 
   let end = url.length;
+  if (
+    url.charCodeAt(end - 1) === 0x5f &&
+    BARE_URL_BREAK_PATTERN.test(nextCharacter)
+  ) {
+    end--;
+  }
   while (end > 0) {
     const c = url.charCodeAt(end - 1);
     // .,;:!?'"`> — `>` covers CommonMark autolinks (`<https://x.com>`)
@@ -178,6 +192,21 @@ export const MD_LINK_PATTERN = String.raw`\[.*?\]\((?:[^()]|\([^()]*\))*\)`;
  */
 export const MD_LINK_CAPTURE = /^\[(.*?)\]\(((?:[^()]|\([^()]*\))*)\)$/;
 
+/**
+ * Bare-URL pattern shared between the React and ANSI renderers. Unlike a
+ * plain `\S+` run it stops at CJK / full-width punctuation: Chinese prose
+ * routinely glues `（…）`/`。` onto a URL with no space
+ * (`https://x.com（2 commits）`), and `\S` swallows the punctuation plus
+ * everything up to the next ASCII space, turning the OSC 8 target into a
+ * 404. Raw CJK ideographs, U+3005 々, U+3006 〆, and U+3007 〇 stay in the
+ * match because they are word-forming IRI characters. The exclusion also
+ * covers punctuation in CJK Compatibility Forms and Vertical Forms while
+ * preserving their word-forming repeat marks and vertical low lines. ASCII
+ * and other typographic punctuation stays matched and is left to
+ * `trimTrailingUrlPunctuation`.
+ */
+export const BARE_URL_PATTERN = String.raw`https?:\/\/[^\s${BARE_URL_BREAK_CHARACTERS}]+`;
+
 /**
  * Should the markdown renderers wrap a `[label](url)` token in an OSC 8
  * envelope? Returns true only when (a) the host terminal advertises OSC 8,