@cablate/banini-tracker 2.0.11 → 2.0.12

package/dist/cli.js

@@ -5,6 +5,8 @@ import { fetchFacebookPosts } from './facebook.js';
 import { sendTelegramMessage } from './telegram.js';
 import { filterNewPosts, markPostsSeen, listSeenIds, clearSeen } from './seen.js';
 import { readFileSync } from 'fs';
+import { createTranscriber, transcribeVideoPosts, isVideoPost } from './transcribe.js';
+import { getDb } from './db.js';
 const program = new Command();
 program
     .name('banini-tracker')
@@ -18,10 +20,13 @@ program
     .option('--tg-bot-token <token>', 'Telegram Bot token')
     .option('--tg-channel-id <id>', 'Telegram Channel ID')
     .option('--fb-page-url <url>', 'Facebook 粉專網址', 'https://www.facebook.com/DieWithoutBang/')
+    .option('--groq-api-key <key>', 'Groq API key（影片轉錄用）')
     .action((opts) => {
     const config = defaultConfig();
     if (opts.apifyToken)
         config.apifyToken = opts.apifyToken;
+    if (opts.groqApiKey)
+        config.groqApiKey = opts.groqApiKey;
     if (opts.tgBotToken || opts.tgChannelId) {
         config.telegram = {
             botToken: opts.tgBotToken ?? '',
@@ -66,6 +71,8 @@ program
     .option('--until <date>', '只抓此時間之前的貼文')
     .option('--no-dedup', '不做去重，抓到什麼就輸出什麼')
     .option('--mark-seen', '輸出後自動標記為已讀')
+    .option('--transcribe', '自動轉錄影片（captionText 為空時走 Groq Whisper）')
+    .option('--save-db', '抓取後直接存入 SQLite')
     .action(async (opts) => {
     try {
         const config = loadConfig();
@@ -80,6 +87,53 @@ program
         if (opts.dedup !== false) {
             posts = filterNewPosts(posts);
         }
+        // 影片轉錄：captionText 為空的影片走 Groq
+        if (opts.transcribe) {
+            const groqKey = config.groqApiKey || process.env.GROQ_API_KEY;
+            if (!groqKey) {
+                console.error('⚠ --transcribe 需要 Groq API key，請用 init --groq-api-key 設定或設定環境變數 GROQ_API_KEY');
+            }
+            else {
+                const needsTranscribe = posts.filter((p) => isVideoPost(p.mediaType) && !p.captionText);
+                if (needsTranscribe.length > 0) {
+                    console.error(`[轉錄] ${needsTranscribe.length} 篇影片需要轉錄...`);
+                    if (!process.env.GROQ_API_KEY)
+                        process.env.GROQ_API_KEY = groqKey;
+                    const transcriber = createTranscriber('groq');
+                    const transcripts = await transcribeVideoPosts(needsTranscribe, transcriber);
+                    for (const p of needsTranscribe) {
+                        const result = transcripts.get(p.id);
+                        if (result)
+                            p.captionText = result.text;
+                    }
+                }
+            }
+        }
+        // 存入 DB
+        if (opts.saveDb && posts.length > 0) {
+            const db = getDb();
+            const upsert = db.prepare(`
+          INSERT INTO posts (id, source, text, ocr_text, transcript_text, media_type, media_url, url, like_count, comment_count, post_timestamp, fetched_at)
+          VALUES (@id, @source, @text, @ocr_text, @transcript_text, @media_type, @media_url, @url, @like_count, @comment_count, @post_timestamp, @fetched_at)
+          ON CONFLICT(id) DO UPDATE SET
+            transcript_text = CASE WHEN excluded.transcript_text != '' THEN excluded.transcript_text ELSE posts.transcript_text END,
+            like_count = excluded.like_count,
+            comment_count = excluded.comment_count
+        `);
+            const now = new Date().toISOString();
+            db.transaction(() => {
+                for (const p of posts) {
+                    upsert.run({
+                        id: p.id, source: p.source, text: p.text,
+                        ocr_text: p.ocrText || '', transcript_text: p.captionText || '',
+                        media_type: p.mediaType, media_url: p.mediaUrl, url: p.url,
+                        like_count: p.likeCount, comment_count: p.commentCount || 0,
+                        post_timestamp: p.timestamp, fetched_at: now,
+                    });
+                }
+            })();
+            console.error(`[DB] ${posts.length} 篇已存入`);
+        }
         // 標記已讀
         if (opts.markSeen && posts.length > 0) {
             markPostsSeen(posts.map((p) => p.id));

package/dist/config.d.ts

@@ -1,5 +1,6 @@
 export interface Config {
     apifyToken: string;
+    groqApiKey?: string;
     telegram?: {
         botToken: string;
         channelId: string;

package/dist/transcribe.js

@@ -66,7 +66,7 @@ export class GroqTranscriber {
         };
     }
     async transcribeViaDownload(videoUrl) {
-        console.log(`[轉錄] 下載音訊: ${videoUrl.slice(0, 60)}...`);
+        console.error(`[轉錄] 下載音訊: ${videoUrl.slice(0, 60)}...`);
         const audioFile = await downloadAudio(videoUrl);
         try {
             const result = await this.client.audio.transcriptions.create({
@@ -115,14 +115,14 @@ export async function transcribeVideoPosts(posts, transcriber) {
         if (!isVideoPost(post.mediaType) || !post.mediaUrl)
             continue;
         try {
-            console.log(`[轉錄][${transcriber.name}] 處理影片: ${post.id}`);
+            console.error(`[轉錄][${transcriber.name}] 處理影片: ${post.id}`);
             const result = await transcriber.transcribe(post.mediaUrl);
             if (result.text.trim().length > 0) {
                 results.set(post.id, result);
-                console.log(`[轉錄] ${post.id}: ${result.text.slice(0, 50)}...（${result.durationSec ?? '?'}s）`);
+                console.error(`[轉錄] ${post.id}: ${result.text.slice(0, 50)}...（${result.durationSec ?? '?'}s）`);
             }
             else {
-                console.log(`[轉錄] ${post.id}: 無可辨識內容`);
+                console.error(`[轉錄] ${post.id}: 無可辨識內容`);
             }
         }
         catch (err) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@cablate/banini-tracker",
-  "version": "2.0.11",
+  "version": "2.0.12",
   "description": "巴逆逆反指標追蹤器 — 常駐排程 + CLI 雙模式",
   "type": "module",
   "bin": {