fix: 保留Bilibili链接HTML实体边界

src/modules/qqbot/plugins/bilibili-card/src/domain/bilibili-url-extractor.ts

@@ -4,7 +4,8 @@ import {
   isAllowedBilibiliUrl,
 } from './bilibili-url-parser';
 
-const URL_PATTERN = /https?:\/\/[^\s<>"',;:!，。！？；、\]\)}]+/giu;
+const URL_PATTERN = /https?:\/\/[^\s<>"',:!，。！？；、\]\)}]+/giu;
+const HTML_ENTITY_BEFORE_SEMICOLON = /&(?:amp|quot|lt|gt|#34|#60|#62)$/iu;
 const MAX_DEPTH = 7;
 const MAX_NODES = 200;
 const MAX_STRINGS = 200;
@@ -31,7 +32,9 @@ export function extractBilibiliUrls(input: BilibiliUrlExtractionInput) {
 
   for (const text of candidates) {
     for (const rawUrl of text.match(URL_PATTERN) || []) {
-      const cleaned = cleanBilibiliUrlCandidate(rawUrl);
+      const cleaned = cleanBilibiliUrlCandidate(
+        trimNonEntitySemicolonTail(rawUrl),
+      );
       if (!isAllowedBilibiliUrl(cleaned) || seen.has(cleaned)) continue;
       seen.add(cleaned);
       output.push(cleaned);
@@ -42,6 +45,20 @@ export function extractBilibiliUrls(input: BilibiliUrlExtractionInput) {
   return output;
 }
 
+/**
+ * Removes semicolon-delimited CQ tail text while keeping common HTML entities intact.
+ * @param rawUrl - Raw URL token collected by the extractor regex.
+ * @returns URL token that can still be decoded by `cleanBilibiliUrlCandidate`.
+ */
+function trimNonEntitySemicolonTail(rawUrl: string) {
+  for (let index = 0; index < rawUrl.length; index += 1) {
+    if (rawUrl[index] !== ';') continue;
+    if (HTML_ENTITY_BEFORE_SEMICOLON.test(rawUrl.slice(0, index))) continue;
+    return rawUrl.slice(0, index);
+  }
+  return rawUrl;
+}
+
 /**
  * Collects string values that may contain links from text fields and nested QQ card objects.
  * @param input - Extraction input built from normalized message state.

test/modules/qqbot/plugins/bilibili-card/bilibili-url-extractor.spec.ts

@@ -135,4 +135,36 @@ describe('Bilibili URL extractor', () => {
       }),
     ).toEqual(['https://b23.tv/abc123']);
   });
+
+  it('keeps html entity wrappers intact until URL cleanup runs', () => {
+    expect(
+      extractBilibiliUrls({
+        messageText: '',
+        rawEvent: {
+          message: [
+            {
+              data: {
+                data: '<【https://www.bilibili.com/video/BV1xx411c7mD】>',
+              },
+              type: 'xml',
+            },
+          ],
+        },
+        rawMessage: '',
+      }),
+    ).toEqual(['https://www.bilibili.com/video/BV1xx411c7mD']);
+  });
+
+  it('preserves ampersand query entities while removing CQ separators', () => {
+    expect(
+      extractBilibiliUrls({
+        messageText:
+          '[CQ:share,url=https://www.bilibili.com/video/BV1xx411c7mD?foo=1&bar=2,title=视频]',
+        rawEvent: {},
+        rawMessage: '',
+      }),
+    ).toEqual([
+      'https://www.bilibili.com/video/BV1xx411c7mD?foo=1&bar=2',
+    ]);
+  });
 });