first commit

2026-08-13 18:36:31 +09:00
commit cee36bf12d
48 changed files with 4212 additions and 0 deletions
@@ -0,0 +1,88 @@
+import { Innertube } from 'youtubei.js';
+import { AppError } from '../../utils/errors.js';
+import { canonicalSourceUrl, extractYouTubeId } from '../../utils/url.js';
+
+function textValue(value) {
+  if (value == null) return '';
+  return typeof value === 'string' ? value : value.toString();
+}
+
+export function readTranscriptSegments(transcriptInfo) {
+  const segments = transcriptInfo?.transcript?.content?.body?.initial_segments ?? [];
+  return segments
+    .filter((segment) => segment?.snippet && segment?.start_ms != null)
+    .map((segment) => ({
+      startSec: Number(segment.start_ms) / 1000,
+      endSec: Number(segment.end_ms) / 1000,
+      text: textValue(segment.snippet).trim(),
+    }))
+    .filter((segment) => Number.isFinite(segment.startSec) && segment.text);
+}
+
+export class YouTubeExtractor {
+  constructor({ clientFactory } = {}) {
+    this.clientFactory = clientFactory ?? (() => Innertube.create({
+      lang: 'ko',
+      location: 'KR',
+      retrieve_player: false,
+    }));
+  }
+
+  async extract(sourceUrl) {
+    const sourceId = extractYouTubeId(sourceUrl);
+    let info;
+
+    try {
+      const client = await this.clientFactory();
+      info = await client.getInfo(sourceId);
+    } catch (error) {
+      throw new AppError('YouTube 정보를 가져오지 못했습니다.', {
+        statusCode: 502,
+        code: 'YOUTUBE_EXTRACTION_FAILED',
+        cause: error,
+      });
+    }
+
+    const basic = info.basic_info ?? {};
+    const description = basic.short_description?.trim() ?? '';
+    let transcriptSegments = [];
+
+    try {
+      transcriptSegments = readTranscriptSegments(await info.getTranscript());
+    } catch {
+      // 자막이 비활성화된 영상은 설명만으로 계속 분석한다.
+    }
+
+    if (!description && transcriptSegments.length === 0) {
+      throw new AppError('YouTube 자막과 설명을 찾을 수 없습니다.', {
+        statusCode: 422,
+        code: 'YOUTUBE_TEXT_NOT_FOUND',
+      });
+    }
+
+    const transcriptText = transcriptSegments
+      .map((segment) => `[${segment.startSec}] ${segment.text}`)
+      .join('\n');
+    const rawText = [
+      description && `[DESCRIPTION]\n${description}`,
+      transcriptText && `[TRANSCRIPT]\n${transcriptText}`,
+    ].filter(Boolean).join('\n\n');
+    const thumbnails = basic.thumbnail ?? [];
+    const thumbnail = thumbnails.length > 0 ? thumbnails[thumbnails.length - 1] : null;
+
+    return {
+      platform: 'youtube',
+      sourceUrl: canonicalSourceUrl(sourceUrl),
+      sourceId,
+      title: basic.title?.trim() || null,
+      author: basic.author?.trim() || basic.channel?.name?.trim() || null,
+      thumbnailUrl: thumbnail?.url ?? null,
+      rawText,
+      metadata: {
+        durationSec: basic.duration ?? null,
+        transcript: transcriptSegments,
+      },
+    };
+  }
+}
+