feat: 提取QQ卡片中的Bilibili链接

src/modules/qqbot/plugins/bilibili-card/src/domain/bilibili-card.types.ts

@@ -11,3 +11,9 @@ export type BilibiliVideoReference =
       sourceUrl: string;
       value: string;
     };
+
+export type BilibiliUrlExtractionInput = {
+  messageText?: string;
+  rawEvent?: Record<string, unknown>;
+  rawMessage?: string;
+};

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

@@ -0,0 +1,112 @@
+import type { BilibiliUrlExtractionInput } from './bilibili-card.types';
+import {
+  cleanBilibiliUrlCandidate,
+  isAllowedBilibiliUrl,
+} from './bilibili-url-parser';
+
+const URL_PATTERN = /https?:\/\/[^\s<>"'，。！？；、]+/giu;
+const MAX_DEPTH = 7;
+
+/**
+ * Extracts Bilibili URL candidates from normalized message text and raw QQ card payloads.
+ * @param input - Normalized QQBot message fields and raw OneBot event payload from NapCat.
+ * @returns Unique allowed Bilibili URL strings in discovery order.
+ */
+export function extractBilibiliUrls(input: BilibiliUrlExtractionInput) {
+  const candidates = collectStringCandidates(input);
+  const seen = new Set<string>();
+  const output: string[] = [];
+
+  for (const text of candidates) {
+    for (const rawUrl of text.match(URL_PATTERN) || []) {
+      const cleaned = cleanBilibiliUrlCandidate(rawUrl);
+      if (!isAllowedBilibiliUrl(cleaned) || seen.has(cleaned)) continue;
+      seen.add(cleaned);
+      output.push(cleaned);
+    }
+  }
+
+  return output;
+}
+
+/**
+ * Collects string values that may contain links from text fields and nested QQ card objects.
+ * @param input - Extraction input built from normalized message state.
+ * @returns Candidate strings that may contain URLs.
+ */
+function collectStringCandidates(input: BilibiliUrlExtractionInput) {
+  const output: string[] = [];
+  const seen = new WeakSet<object>();
+  pushText(output, input.messageText);
+  pushText(output, input.rawMessage);
+  collectUnknown(input.rawEvent, output, seen, 0);
+  return output;
+}
+
+/**
+ * Walks unknown card payload data while bounding recursion and parsing JSON-looking strings.
+ * @param value - Unknown raw value from OneBot message segments or nested card fields.
+ * @param output - Mutable candidate string list.
+ * @param seen - Object identity set used to avoid cyclic payloads.
+ * @param depth - Current recursion depth used to bound malformed payloads.
+ */
+function collectUnknown(
+  value: unknown,
+  output: string[],
+  seen: WeakSet<object>,
+  depth: number,
+) {
+  if (depth > MAX_DEPTH || value == null) return;
+  if (typeof value === 'string') {
+    pushText(output, value);
+    collectJsonString(value, output, seen, depth + 1);
+    return;
+  }
+  if (typeof value !== 'object') return;
+  if (seen.has(value)) return;
+  seen.add(value);
+
+  if (Array.isArray(value)) {
+    for (const item of value) {
+      collectUnknown(item, output, seen, depth + 1);
+    }
+    return;
+  }
+
+  for (const nestedValue of Object.values(value)) {
+    collectUnknown(nestedValue, output, seen, depth + 1);
+  }
+}
+
+/**
+ * Parses a JSON card string when possible and ignores invalid JSON without aborting extraction.
+ * @param value - Raw string that may be JSON.
+ * @param output - Mutable candidate string list.
+ * @param seen - Object identity set shared with the recursive walk.
+ * @param depth - Recursion depth for the parsed value.
+ */
+function collectJsonString(
+  value: string,
+  output: string[],
+  seen: WeakSet<object>,
+  depth: number,
+) {
+  const trimmed = value.trim();
+  if (!trimmed.startsWith('{') && !trimmed.startsWith('[')) return;
+  try {
+    collectUnknown(JSON.parse(trimmed), output, seen, depth);
+  } catch {
+    return;
+  }
+}
+
+/**
+ * Adds a non-empty string candidate to the output list.
+ * @param output - Mutable candidate string list.
+ * @param value - Candidate value from normalized text or raw card payload.
+ */
+function pushText(output: string[], value: unknown) {
+  if (typeof value === 'string' && value.trim()) {
+    output.push(value);
+  }
+}

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

@@ -0,0 +1,81 @@
+import { extractBilibiliUrls } from '../../../../../src/modules/qqbot/plugins/bilibili-card/src/domain/bilibili-url-extractor';
+
+describe('Bilibili URL extractor', () => {
+  it('extracts links from messageText and rawMessage while deduplicating them', () => {
+    expect(
+      extractBilibiliUrls({
+        messageText: '看看 https://www.bilibili.com/video/BV1xx411c7mD',
+        rawMessage:
+          '重复 https://www.bilibili.com/video/BV1xx411c7mD?share=qq',
+        rawEvent: {},
+      }),
+    ).toEqual([
+      'https://www.bilibili.com/video/BV1xx411c7mD',
+      'https://www.bilibili.com/video/BV1xx411c7mD?share=qq',
+    ]);
+  });
+
+  it('extracts a QQ share card URL', () => {
+    const urls = extractBilibiliUrls({
+      messageText: '',
+      rawMessage: '',
+      rawEvent: {
+        message: [
+          {
+            data: {
+              content: '夏祭 视频',
+              title: 'Bilibili',
+              url: 'https://www.bilibili.com/video/BV1xx411c7mD',
+            },
+            type: 'share',
+          },
+        ],
+      },
+    });
+
+    expect(urls).toEqual(['https://www.bilibili.com/video/BV1xx411c7mD']);
+  });
+
+  it('extracts nested URLs from json and lightapp cards', () => {
+    const payload = JSON.stringify({
+      app: 'com.tencent.structmsg',
+      meta: {
+        detail: {
+          jumpUrl: 'https://b23.tv/abc123',
+        },
+      },
+    });
+
+    expect(
+      extractBilibiliUrls({
+        messageText: '',
+        rawMessage: '',
+        rawEvent: {
+          message: [
+            { data: { data: payload }, type: 'json' },
+            { data: { data: payload }, type: 'lightapp' },
+          ],
+        },
+      }),
+    ).toEqual(['https://b23.tv/abc123']);
+  });
+
+  it('extracts URLs from xml card text and ignores non-Bilibili URLs', () => {
+    expect(
+      extractBilibiliUrls({
+        messageText: 'https://example.com/video/BV1xx411c7mD',
+        rawMessage: '',
+        rawEvent: {
+          message: [
+            {
+              data: {
+                data: '<msg url="https://m.bilibili.com/video/av170001" />',
+              },
+              type: 'xml',
+            },
+          ],
+        },
+      }),
+    ).toEqual(['https://m.bilibili.com/video/av170001']);
+  });
+});