@cablate/banini-tracker 2.0.8 → 2.0.10

package/dist/cli.js

@@ -62,6 +62,8 @@ program
     .description('抓取最新貼文（輸出 JSON 到 stdout）')
     .option('-s, --source <source>', '來源：fb', 'fb')
     .option('-n, --limit <n>', '每個來源抓幾篇', '3')
+    .option('--since <date>', '只抓此時間之後的貼文（YYYY-MM-DD 或 ISO 時間戳或相對時間如 "2 months"）')
+    .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')
     .action(async (opts) => {
@@ -69,7 +71,8 @@ program
         const config = loadConfig();
         const limit = parseInt(opts.limit, 10);
         let posts = [];
-        const fp = await fetchFacebookPosts(config.targets.facebookPageUrl, config.apifyToken, limit);
+        const fetchOpts = (opts.since || opts.until) ? { since: opts.since, until: opts.until } : undefined;
+        const fp = await fetchFacebookPosts(config.targets.facebookPageUrl, config.apifyToken, limit, fetchOpts);
         posts.push(...fp);
         // 按時間從新到舊
         posts.sort((a, b) => new Date(b.timestamp).getTime() - new Date(a.timestamp).getTime());

package/dist/db.js

@@ -19,6 +19,21 @@ export function getDb() {
 }
 function migrate(db) {
     db.exec(`
+    CREATE TABLE IF NOT EXISTS posts (
+      id TEXT PRIMARY KEY,
+      source TEXT NOT NULL DEFAULT 'facebook',
+      text TEXT NOT NULL DEFAULT '',
+      ocr_text TEXT NOT NULL DEFAULT '',
+      transcript_text TEXT NOT NULL DEFAULT '',
+      media_type TEXT NOT NULL DEFAULT 'text',
+      media_url TEXT NOT NULL DEFAULT '',
+      url TEXT NOT NULL DEFAULT '',
+      like_count INTEGER NOT NULL DEFAULT 0,
+      comment_count INTEGER NOT NULL DEFAULT 0,
+      post_timestamp TEXT NOT NULL,
+      fetched_at TEXT NOT NULL
+    );
+
     CREATE TABLE IF NOT EXISTS predictions (
       id INTEGER PRIMARY KEY AUTOINCREMENT,
       post_id TEXT NOT NULL,

package/dist/facebook.d.ts

@@ -3,6 +3,7 @@ export interface FacebookPost {
     source: 'facebook';
     text: string;
     ocrText: string;
+    captionText: string;
     timestamp: string;
     likeCount: number;
     commentCount: number;
@@ -11,4 +12,8 @@ export interface FacebookPost {
     mediaType: string;
     mediaUrl: string;
 }
-export declare function fetchFacebookPosts(pageUrl: string, token: string, maxPosts?: number): Promise<FacebookPost[]>;
+export interface FetchOptions {
+    since?: string;
+    until?: string;
+}
+export declare function fetchFacebookPosts(pageUrl: string, token: string, maxPosts?: number, options?: FetchOptions): Promise<FacebookPost[]>;

package/dist/facebook.js

@@ -1,16 +1,22 @@
-export async function fetchFacebookPosts(pageUrl, token, maxPosts = 3) {
+export async function fetchFacebookPosts(pageUrl, token, maxPosts = 3, options) {
     const actorId = 'apify~facebook-posts-scraper';
     const url = `https://api.apify.com/v2/acts/${actorId}/run-sync-get-dataset-items`;
+    const body = {
+        startUrls: [{ url: pageUrl }],
+        resultsLimit: maxPosts,
+        captionText: true,
+    };
+    if (options?.since)
+        body.onlyPostsNewerThan = options.since;
+    if (options?.until)
+        body.onlyPostsOlderThan = options.until;
     const res = await fetch(url, {
         method: 'POST',
         headers: {
             'Content-Type': 'application/json',
             Authorization: `Bearer ${token}`,
         },
-        body: JSON.stringify({
-            startUrls: [{ url: pageUrl }],
-            resultsLimit: maxPosts,
-        }),
+        body: JSON.stringify(body),
         signal: AbortSignal.timeout(180_000),
     });
     if (!res.ok) {
@@ -23,11 +29,17 @@ export async function fetchFacebookPosts(pageUrl, token, maxPosts = 3) {
         const ocrTexts = (item.media ?? [])
             .map((m) => m.ocrText ?? '')
             .filter((t) => t.length > 0);
+        // captionText 可能在 media item 或頂層
+        const captionTexts = (item.media ?? [])
+            .map((m) => m.captionText ?? '')
+            .filter((t) => t.length > 0);
+        const captionText = captionTexts.join('\n') || item.captionText || '';
         return {
             id: `fb_${item.postId ?? item.id ?? ''}`,
             source: 'facebook',
             text: item.text ?? item.message ?? '',
             ocrText: ocrTexts.join('\n'),
+            captionText,
             timestamp: item.time ?? new Date().toISOString(),
             likeCount: item.likes ?? 0,
             commentCount: item.comments ?? 0,

package/dist/index.js

@@ -18,6 +18,7 @@ import { filterNewPosts as filterNew, markPostsSeen } from './seen.js';
 import { withRetry } from './retry.js';
 import { createTranscriber, transcribeVideoPosts } from './transcribe.js';
 import { recordPredictions, updateTracking } from './tracker.js';
+import { getDb } from './db.js';
 // ── Config ──────────────────────────────────────────────────
 const FB_PAGE_URL = 'https://www.facebook.com/DieWithoutBang/';
 const DATA_DIR = process.env.DATA_DIR || join(process.cwd(), 'data');
@@ -34,7 +35,7 @@ function fromFacebook(p) {
         source: 'facebook',
         text: p.text,
         ocrText: p.ocrText,
-        transcriptText: '',
+        transcriptText: p.captionText || '',
         timestamp: p.timestamp,
         likeCount: p.likeCount,
         replyCount: p.commentCount,
@@ -65,7 +66,8 @@ async function runInner(opts) {
     const allPosts = [];
     // 1. 抓取 Facebook（含 retry）
     try {
-        const fbPosts = await withRetry(() => fetchFacebookPosts(FB_PAGE_URL, apifyToken, opts.maxPosts), { label: 'Facebook', maxRetries: 2, baseDelayMs: 5000 });
+        const fetchOpts = (opts.since || opts.until) ? { since: opts.since, until: opts.until } : undefined;
+        const fbPosts = await withRetry(() => fetchFacebookPosts(FB_PAGE_URL, apifyToken, opts.maxPosts, fetchOpts), { label: 'Facebook', maxRetries: 2, baseDelayMs: 5000 });
         allPosts.push(...fbPosts.map(fromFacebook));
     }
     catch (err) {
@@ -81,17 +83,56 @@ async function runInner(opts) {
         console.log('沒有新貼文，結束');
         return;
     }
-    // 2.5. 影片轉錄
+    // 2.5. 影片轉錄（captionText 有值則跳過 Groq）
     const transcriberType = (process.env.TRANSCRIBER ?? 'noop');
     const transcriber = createTranscriber(transcriberType);
     if (transcriber.name !== 'noop') {
-        const transcripts = await transcribeVideoPosts(newPosts, transcriber);
-        for (const p of newPosts) {
-            const result = transcripts.get(p.id);
-            if (result)
-                p.transcriptText = result.text;
+        const needsTranscribe = newPosts.filter((p) => !p.transcriptText);
+        if (needsTranscribe.length > 0) {
+            const transcripts = await transcribeVideoPosts(needsTranscribe, transcriber);
+            for (const p of needsTranscribe) {
+                const result = transcripts.get(p.id);
+                if (result)
+                    p.transcriptText = result.text;
+            }
         }
     }
+    // 2.6. 貼文入庫
+    try {
+        const db = getDb();
+        const upsertPost = 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();
+        const insertMany = db.transaction(() => {
+            for (const p of newPosts) {
+                upsertPost.run({
+                    id: p.id,
+                    source: p.source,
+                    text: p.text,
+                    ocr_text: p.ocrText,
+                    transcript_text: p.transcriptText,
+                    media_type: p.mediaType,
+                    media_url: p.mediaUrl,
+                    url: p.url,
+                    like_count: p.likeCount,
+                    comment_count: p.replyCount,
+                    post_timestamp: p.timestamp,
+                    fetched_at: now,
+                });
+            }
+        });
+        insertMany();
+        console.log(`[DB] 已存入 ${newPosts.length} 篇貼文`);
+    }
+    catch (err) {
+        console.error(`[DB] 貼文入庫失敗: ${err instanceof Error ? err.message : err}`);
+    }
     // 按時間從新到舊排序
     newPosts.sort((a, b) => new Date(b.timestamp).getTime() - new Date(a.timestamp).getTime());
     // 標記當天貼文
@@ -226,16 +267,42 @@ async function runInner(opts) {
     writeFileSync(outFile, JSON.stringify({ timestamp: new Date().toISOString(), posts: newPosts, analysis }, null, 2), 'utf-8');
     console.log(`結果已存檔: ${outFile}`);
 }
+/**
+ * 產生台北時間今天指定時分的 ISO 時間戳
+ * 用於 Apify onlyPostsNewerThan 參數
+ */
+function taipeiToday(hours, minutes = 0) {
+    const now = new Date();
+    const taipeiStr = now.toLocaleString('en-US', { timeZone: 'Asia/Taipei' });
+    const taipeiNow = new Date(taipeiStr);
+    taipeiNow.setHours(hours, minutes, 0, 0);
+    // 轉回 UTC：台北 = UTC+8
+    const utc = new Date(taipeiNow.getTime() - 8 * 60 * 60 * 1000);
+    return utc.toISOString();
+}
+function taipeiYesterday(hours, minutes = 0) {
+    const now = new Date();
+    const taipeiStr = now.toLocaleString('en-US', { timeZone: 'Asia/Taipei' });
+    const taipeiNow = new Date(taipeiStr);
+    taipeiNow.setDate(taipeiNow.getDate() - 1);
+    taipeiNow.setHours(hours, minutes, 0, 0);
+    const utc = new Date(taipeiNow.getTime() - 8 * 60 * 60 * 1000);
+    return utc.toISOString();
+}
 // ── 入口 ────────────────────────────────────────────────────
 if (isCronMode) {
-    // 盤中：週一到五 09:00-13:30，每 30 分鐘，FB only 抓 1 篇
-    // cron 不支援半小時結束，用 9:00-13:00 每 30 分 + 13:30 單獨一個
+    // 早晨補漏：每天 08:00，抓前一晚 22:00 之後的貼文
+    cron.schedule('0 8 * * *', () => {
+        run({ maxPosts: 3, isDryRun: false, label: '早晨', since: taipeiYesterday(22, 0) })
+            .catch((err) => console.error('[早晨] 執行失敗:', err));
+    }, { timezone: 'Asia/Taipei' });
+    // 盤中：週一到五 09:00-13:30，每 30 分鐘，抓 08:30 之後的貼文
     cron.schedule('7,37 9-12 * * 1-5', () => {
-        run({ maxPosts: 1, isDryRun: false, label: '盤中' })
+        run({ maxPosts: 1, isDryRun: false, label: '盤中', since: taipeiToday(8, 30) })
             .catch((err) => console.error('[盤中] 執行失敗:', err));
     }, { timezone: 'Asia/Taipei' });
     cron.schedule('7 13 * * 1-5', () => {
-        run({ maxPosts: 1, isDryRun: false, label: '盤中' })
+        run({ maxPosts: 1, isDryRun: false, label: '盤中', since: taipeiToday(8, 30) })
             .catch((err) => console.error('[盤中] 執行失敗:', err));
     }, { timezone: 'Asia/Taipei' });
     // 追蹤更新：週一到五 15:00（收盤後更新預測追蹤）
@@ -243,15 +310,16 @@ if (isCronMode) {
         updateTracking()
             .catch((err) => console.error('[追蹤更新] 執行失敗:', err));
     }, { timezone: 'Asia/Taipei' });
-    // 盤後：每天晚上 23:00，FB 3 篇
+    // 盤後：每天晚上 23:03，抓 13:30 之後的貼文
     cron.schedule('3 23 * * *', () => {
-        run({ maxPosts: 3, isDryRun: false, label: '盤後' })
+        run({ maxPosts: 3, isDryRun: false, label: '盤後', since: taipeiToday(13, 30) })
             .catch((err) => console.error('[盤後] 執行失敗:', err));
     }, { timezone: 'Asia/Taipei' });
     console.log('=== 巴逆逆排程已啟動 ===');
-    console.log('  盤中：週一~五 09:07/09:37/10:07/.../13:07（FB, 1 篇）');
+    console.log('  早晨：每天 08:00（前晚 22:00 起，3 篇）');
+    console.log('  盤中：週一~五 09:07/09:37/10:07/.../13:07（08:30 起，1 篇）');
     console.log('  追蹤更新：週一~五 15:00（預測追蹤判定）');
-    console.log('  盤後：每天 23:03（FB, 3 篇）');
+    console.log('  盤後：每天 23:03（13:30 起，3 篇）');
     console.log('  按 Ctrl+C 停止\n');
 }
 else {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@cablate/banini-tracker",
-  "version": "2.0.8",
+  "version": "2.0.10",
   "description": "巴逆逆反指標追蹤器 — 常駐排程 + CLI 雙模式",
   "type": "module",
   "bin": {