summd 0.1.5 → 0.1.6

package/dist/url-to-md.js

@@ -78,29 +78,54 @@ async function fetchYouTubeTitle(url, videoId) {
     catch { }
     return `YouTube: ${videoId}`;
 }
+// Subtitle language priority: English first, then common languages, then anything available.
+// yt-dlp picks the first match; --write-auto-subs covers ASR variants (en-orig etc.)
+const SUB_LANGS = 'en,zh-Hans,zh-Hant,zh,ja,ko,fr,de,es,pt,ru';
 // yt-dlp — works from any IP, supports auth/geo-block via --cookies-from-browser
 async function ytDlpTranscript(url, videoId) {
     const dir = tmpdir();
-    const outTemplate = join(dir, videoId); // output: {videoId}.en.json3
+    const outTemplate = join(dir, videoId);
+    const run = (...extra) => execFileAsync('yt-dlp', [
+        '--write-subs',
+        '--write-auto-subs',
+        '--sub-langs', SUB_LANGS,
+        '--sub-format', 'json3',
+        '--skip-download',
+        '--quiet',
+        '--no-progress',
+        ...extra,
+        '-o', outTemplate,
+        url,
+    ], { timeout: 30_000 });
     try {
-        await execFileAsync('yt-dlp', [
-            '--write-subs',
-            '--write-auto-subs',
-            '--sub-langs', 'en',
-            '--sub-format', 'json3',
-            '--skip-download',
-            '--quiet',
-            '--no-progress',
-            '-o', outTemplate,
-            url,
-        ], { timeout: 30_000 });
+        await run();
     }
     catch (e) {
-        // Distinguish "yt-dlp not in PATH" from "ran but video has no captions"
-        const isNotFound = e.code === 'ENOENT';
-        return { transcript: null, reason: isNotFound ? 'not-installed' : 'no-transcript' };
+        const err = e;
+        if (err.code === 'ENOENT')
+            return { transcript: null, reason: 'not-installed' };
+        // YouTube bot-detection: retry with browser cookies
+        // Triggered when running from cloud/residential IPs without a session
+        const stderr = err.stderr ?? '';
+        if (stderr.includes('Sign in') || stderr.includes('bot')) {
+            const browsers = ['chrome', 'chromium', 'firefox', 'safari', 'edge'];
+            let authed = false;
+            for (const browser of browsers) {
+                try {
+                    await run('--cookies-from-browser', browser);
+                    authed = true;
+                    break;
+                }
+                catch {
+                    continue;
+                }
+            }
+            if (!authed)
+                return { transcript: null, reason: 'no-transcript' };
+        }
+        // Other errors (geo-block, video unavailable): fall through to file check
     }
-    // Find output file: {videoId}.en.json3, {videoId}.en-US.json3, etc.
+    // Find output file: {videoId}.en.json3, {videoId}.zh-Hans.json3, etc.
     try {
         const files = readdirSync(dir).filter(f => f.startsWith(videoId) && f.endsWith('.json3'));
         for (const file of files) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "summd",
-  "version": "0.1.5",
+  "version": "0.1.6",
   "description": "CLI for sum.md — Sum to anything.",
   "license": "MIT",
   "bin": {