fix(weixin): strip typographic quotes from pasted URLs

Reviewed-by: codex-mini0
Reviewed-by: First-principles-0

clis/weixin/download.js

@@ -14,17 +14,65 @@ import { downloadArticle } from '@jackwener/opencli/download/article-download';
 /**
  * Normalize a pasted WeChat article URL.
  */
+// Wrapping quote characters to strip from a pasted URL. Covers ASCII plus
+// CJK typographic / smart quotes, which are common when users copy URLs from
+// Chinese-language environments (WeChat itself, macOS smart-quote
+// substitution, Word / Pages, …).
+//
+// Pairs:
+//   "  "         ASCII straight double
+//   '  '         ASCII straight single
+//   “  ”         curly double (U+201C / U+201D)
+//   ‘  ’         curly single (U+2018 / U+2019)
+//   「  」         CJK corner brackets (U+300C / U+300D)
+//   『  』         CJK white corner brackets (U+300E / U+300F)
+//   „  ‟         German-style double quotes (U+201E / U+201F)
+//   ‹  ›         single guillemets (U+2039 / U+203A)
+//   «  »         double guillemets (U+00AB / U+00BB)
+const WRAPPING_QUOTE_PAIRS = [
+    ['"', '"'],
+    ["'", "'"],
+    ['“', '”'],
+    ['‘', '’'],
+    ['「', '」'],
+    ['『', '』'],
+    ['„', '‟'],
+    ['‹', '›'],
+    ['«', '»'],
+];
+const LEADING_WRAP_CHARS = new Set(WRAPPING_QUOTE_PAIRS.map(([open]) => open).concat('<'));
+const TRAILING_WRAP_CHARS = new Set(WRAPPING_QUOTE_PAIRS.map(([, close]) => close).concat('>'));
+
+function stripBoundaryWrapChars(value) {
+    let s = value;
+    for (let i = 0; i < 4; i += 1) {
+        const before = s;
+        for (const [open, close] of WRAPPING_QUOTE_PAIRS) {
+            if (s.length >= 2 && s.startsWith(open) && s.endsWith(close)) {
+                s = s.slice(open.length, s.length - close.length).trim();
+                break;
+            }
+        }
+        while (s && LEADING_WRAP_CHARS.has(s[0])) {
+            s = s.slice(1).trimStart();
+        }
+        while (s && TRAILING_WRAP_CHARS.has(s[s.length - 1])) {
+            s = s.slice(0, -1).trimEnd();
+        }
+        if (s === before)
+            break;
+    }
+    return s;
+}
+
 export function normalizeWechatUrl(raw) {
     let s = (raw || '').trim();
     if (!s)
         return s;
-    // Strip wrapping quotes / angle brackets
-    if ((s.startsWith('"') && s.endsWith('"')) || (s.startsWith("'") && s.endsWith("'"))) {
-        s = s.slice(1, -1).trim();
-    }
-    if (s.startsWith('<') && s.endsWith('>')) {
-        s = s.slice(1, -1).trim();
-    }
+    // Strip quote / angle-bracket characters only at the pasted boundary. This
+    // handles both paired wrappers ("<url>", "“url”") and common one-sided
+    // trailing punctuation ("url”") without touching encoded URL content.
+    s = stripBoundaryWrapChars(s);
     // Remove backslash escapes before URL-significant characters
     s = s.replace(/\\+([:/&?=#%])/g, '$1');
     // Decode HTML entities

clis/weixin/download.test.js

@@ -0,0 +1,106 @@
+import { describe, expect, it } from 'vitest';
+import { normalizeWechatUrl } from './download.js';
+
+describe('normalizeWechatUrl', () => {
+    const canonical = 'https://mp.weixin.qq.com/s/oBz-oik0i9YM2Uia_aadjw';
+
+    it('returns the input unchanged when already canonical', () => {
+        expect(normalizeWechatUrl(canonical)).toBe(canonical);
+    });
+
+    it('returns empty string for empty / nullish input', () => {
+        expect(normalizeWechatUrl('')).toBe('');
+        expect(normalizeWechatUrl(null)).toBe('');
+        expect(normalizeWechatUrl(undefined)).toBe('');
+    });
+
+    it('strips ASCII straight double quotes', () => {
+        expect(normalizeWechatUrl(`"${canonical}"`)).toBe(canonical);
+    });
+
+    it('strips ASCII straight single quotes', () => {
+        expect(normalizeWechatUrl(`'${canonical}'`)).toBe(canonical);
+    });
+
+    it('strips CJK curly double quotes (U+201C / U+201D)', () => {
+        // Wrap the URL in left/right curly double quotes — common when
+        // copy-pasting from WeChat / macOS smart-quote substitution.
+        expect(normalizeWechatUrl(`“${canonical}”`)).toBe(canonical);
+    });
+
+    it('strips CJK curly single quotes (U+2018 / U+2019)', () => {
+        expect(normalizeWechatUrl(`‘${canonical}’`)).toBe(canonical);
+    });
+
+    it('strips CJK corner brackets 「 」 (U+300C / U+300D)', () => {
+        expect(normalizeWechatUrl(`「${canonical}」`)).toBe(canonical);
+    });
+
+    it('strips CJK white corner brackets 『 』 (U+300E / U+300F)', () => {
+        expect(normalizeWechatUrl(`『${canonical}』`)).toBe(canonical);
+    });
+
+    it('strips German-style double quotes „ ‟ (U+201E / U+201F)', () => {
+        expect(normalizeWechatUrl(`„${canonical}‟`)).toBe(canonical);
+    });
+
+    it('strips single guillemets ‹ › (U+2039 / U+203A)', () => {
+        expect(normalizeWechatUrl(`‹${canonical}›`)).toBe(canonical);
+    });
+
+    it('strips double guillemets « » (U+00AB / U+00BB)', () => {
+        expect(normalizeWechatUrl(`«${canonical}»`)).toBe(canonical);
+    });
+
+    it('strips wrapping angle brackets < >', () => {
+        expect(normalizeWechatUrl(`<${canonical}>`)).toBe(canonical);
+    });
+
+    it('handles whitespace around the wrapping quotes', () => {
+        expect(normalizeWechatUrl(`   “${canonical}”   `)).toBe(canonical);
+    });
+
+    it('strips one-sided trailing smart quotes from pasted URL text', () => {
+        expect(normalizeWechatUrl(`${canonical}”`)).toBe(canonical);
+        expect(normalizeWechatUrl(`${canonical}」`)).toBe(canonical);
+        expect(normalizeWechatUrl(`${canonical}>`)).toBe(canonical);
+    });
+
+    it('strips one-sided leading smart quotes from pasted URL text', () => {
+        expect(normalizeWechatUrl(`“${canonical}`)).toBe(canonical);
+        expect(normalizeWechatUrl(`「${canonical}`)).toBe(canonical);
+        expect(normalizeWechatUrl(`<${canonical}`)).toBe(canonical);
+    });
+
+    it('strips asymmetric boundary quote punctuation without touching the URL body', () => {
+        expect(normalizeWechatUrl(`“${canonical}"`)).toBe(canonical);
+    });
+
+    it('does not strip encoded quote-like characters inside the URL', () => {
+        const withEncodedQuote = 'https://mp.weixin.qq.com/s/oBz-oik0i9YM2Uia_aadjw?note=%E2%80%9D#frag';
+        expect(normalizeWechatUrl(withEncodedQuote)).toBe(withEncodedQuote);
+    });
+
+    it('removes backslash escapes inserted by some shells', () => {
+        const escaped = 'https\\://mp.weixin.qq.com/s/oBz-oik0i9YM2Uia_aadjw';
+        expect(normalizeWechatUrl(escaped)).toBe(canonical);
+    });
+
+    it('decodes &amp; HTML entity', () => {
+        const html = 'https://mp.weixin.qq.com/s?foo=1&amp;bar=2';
+        expect(normalizeWechatUrl(html)).toBe('https://mp.weixin.qq.com/s?foo=1&bar=2');
+    });
+
+    it('handles bare mp.weixin.qq.com hostname (no protocol)', () => {
+        expect(normalizeWechatUrl('mp.weixin.qq.com/s/oBz-oik0i9YM2Uia_aadjw')).toBe(canonical);
+    });
+
+    it('handles //mp.weixin.qq.com/... protocol-relative URL', () => {
+        expect(normalizeWechatUrl('//mp.weixin.qq.com/s/oBz-oik0i9YM2Uia_aadjw')).toBe(canonical);
+    });
+
+    it('forces https:// for mp.weixin.qq.com http:// links', () => {
+        const http = 'http://mp.weixin.qq.com/s/oBz-oik0i9YM2Uia_aadjw';
+        expect(normalizeWechatUrl(http)).toBe(canonical);
+    });
+});