summd 0.1.10 → 0.1.11

package/dist/url-to-md.js

@@ -80,16 +80,29 @@ async function fetchYouTubeTitle(url, videoId) {
 }
 // Browser cookie sources — tried in order, first working one is reused for all runs.
 const BROWSERS = ['chrome', 'chromium', 'firefox', 'safari', 'edge'];
-// Subtitle strategies — tried in order, stop at first that produces output.
-// Each downloads only the requested languages, not everything.
-const SUB_STRATEGIES = [
-    // 1. Original-language ASR: yt-dlp 'orig' matches the video's primary language
-    ['--write-auto-subs', '--sub-langs', 'orig'],
-    // 2. English manual + ASR
+// Subtitle fallback strategies used when native language detection fails.
+// Tried in order after the native-language attempt.
+const SUB_FALLBACKS = [
+    // 1. English manual + ASR
     ['--write-subs', '--write-auto-subs', '--sub-langs', 'en'],
-    // 3. Most-spoken languages by global user count (covers remaining cases)
+    // 2. Most-spoken languages by global user count (covers remaining cases)
     ['--write-subs', '--write-auto-subs', '--sub-langs', 'zh-Hans,zh,ja,ko,fr,de,es,pt,hi,ar,ru'],
 ];
+// Detect the video's primary language via yt-dlp --dump-json.
+// Returns a language code like 'zh-Hans', 'en', 'ja', etc., or null on failure.
+async function getVideoLanguage(url, cookieArgs) {
+    try {
+        const { stdout } = await execFileAsync('yt-dlp', [
+            '--dump-json', '--no-playlist', '--quiet',
+            ...cookieArgs, url,
+        ], { timeout: 15_000 });
+        const meta = JSON.parse(stdout);
+        return meta.language ?? null;
+    }
+    catch {
+        return null;
+    }
+}
 async function ytDlpTranscript(url, videoId) {
     const dir = tmpdir();
     const outTemplate = join(dir, videoId);
@@ -139,25 +152,43 @@ async function ytDlpTranscript(url, videoId) {
             return null;
         }
     };
-    // Phase 1: find a working browser cookie source using the first subtitle strategy.
-    // The result tells us both which browser works AND whether strategy 1 produced output.
+    // Phase 1: find a working browser cookie source.
+    // Try each browser with a no-op probe (--dump-json is light and confirms auth works).
     let cookieArgs = [];
     for (const browser of BROWSERS) {
-        const outcome = await run(SUB_STRATEGIES[0], ['--cookies-from-browser', browser]);
+        try {
+            await execFileAsync('yt-dlp', [
+                '--dump-json', '--no-playlist', '--quiet',
+                '--cookies-from-browser', browser, url,
+            ], { timeout: 15_000 });
+            cookieArgs = ['--cookies-from-browser', browser];
+            break;
+        }
+        catch (e) {
+            const err = e;
+            if (err.code === 'ENOENT')
+                return { transcript: null, reason: 'not-installed' };
+            const stderr = err.stderr ?? '';
+            if (stderr.includes('Could not find') || stderr.includes('cookies from browser'))
+                continue;
+            // Other errors (bot-detection without cookies, etc.) — stop trying browsers
+            break;
+        }
+    }
+    // cookieArgs is empty if no browser found — fall back to cookie-less requests
+    // Phase 2: detect the video's primary language from metadata.
+    const nativeLang = await getVideoLanguage(url, cookieArgs);
+    // Phase 3: try native language subtitles first (most accurate).
+    if (nativeLang) {
+        const outcome = await run(['--write-subs', '--write-auto-subs', '--sub-langs', nativeLang], cookieArgs);
         if (outcome === 'not-installed')
             return { transcript: null, reason: 'not-installed' };
-        if (outcome === 'cookie-error')
-            continue; // browser not available, try next
-        cookieArgs = ['--cookies-from-browser', browser];
-        break;
+        const t = readAndClean();
+        if (t)
+            return { transcript: t };
     }
-    // cookieArgs is empty if no browser found — fall back to cookie-less requests
-    // Check if strategy 1 produced output
-    const t1 = readAndClean();
-    if (t1)
-        return { transcript: t1 };
-    // Phase 2: try remaining subtitle strategies with the confirmed cookie source
-    for (const subArgs of SUB_STRATEGIES.slice(1)) {
+    // Phase 4: fallback strategies with the confirmed cookie source.
+    for (const subArgs of SUB_FALLBACKS) {
         const outcome = await run(subArgs, cookieArgs);
         if (outcome === 'not-installed')
             return { transcript: null, reason: 'not-installed' };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "summd",
-  "version": "0.1.10",
+  "version": "0.1.11",
   "description": "CLI for sum.md — Sum to anything.",
   "license": "MIT",
   "bin": {