summd 0.1.9 → 0.1.11

package/dist/url-to-md.js

@@ -78,90 +78,124 @@ async function fetchYouTubeTitle(url, videoId) {
     catch { }
     return `YouTube: ${videoId}`;
 }
-// Browser cookie sources — tried in order, first one that succeeds is used.
-// Cookies bypass bot-detection and geo-blocks; required on most IPs.
+// Browser cookie sources — tried in order, first working one is reused for all runs.
 const BROWSERS = ['chrome', 'chromium', 'firefox', 'safari', 'edge'];
-// yt-dlp — uses browser cookies by default; falls back to no-cookies for public videos.
-// --sub-langs all: accept whatever subtitle language the video provides, no preference.
+// 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'],
+    // 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);
-    const run = (cookieArgs) => execFileAsync('yt-dlp', [
-        '--write-subs',
-        '--write-auto-subs',
-        '--sub-langs', 'all', // accept any language — no hardcoded preference
-        '--sub-format', 'json3',
-        '--skip-download',
-        '--quiet',
-        '--no-progress',
-        ...cookieArgs,
-        '-o', outTemplate,
-        url,
-    ], { timeout: 30_000 });
-    // Try with browser cookies first (handles bot-detection, auth, geo-block).
-    // Fall back to no-cookies only if no browser is available.
-    let ran = false;
-    for (const browser of BROWSERS) {
+    const run = async (subArgs, cookieArgs) => {
         try {
-            await run(['--cookies-from-browser', browser]);
-            ran = true;
-            break;
+            await execFileAsync('yt-dlp', [
+                '--sub-format', 'json3',
+                '--skip-download', '--quiet', '--no-progress',
+                ...subArgs, ...cookieArgs,
+                '-o', outTemplate, url,
+            ], { timeout: 30_000 });
+            return 'ok';
         }
         catch (e) {
             const err = e;
             if (err.code === 'ENOENT')
-                return { transcript: null, reason: 'not-installed' };
+                return 'not-installed';
             const stderr = err.stderr ?? '';
-            // Cookie access failed (browser not installed / locked DB) — try next browser
-            if (stderr.includes('browser') || stderr.includes('cookie') || stderr.includes('Could not find'))
-                continue;
-            // yt-dlp ran but failed for another reason (video unavailable, no captions…)
-            ran = true;
-            break;
+            if (stderr.includes('Could not find') || stderr.includes('cookies from browser'))
+                return 'cookie-error';
+            return 'other-error';
         }
-    }
-    // No browser found with usable cookies — try without (works for fully public videos)
-    if (!ran) {
+    };
+    const readAndClean = () => {
+        try {
+            const files = readdirSync(dir).filter(f => f.startsWith(videoId) && f.endsWith('.json3'));
+            let transcript = null;
+            for (const file of files) {
+                const filePath = join(dir, file);
+                try {
+                    if (!transcript) {
+                        const text = parseTimedText(JSON.parse(readFileSync(filePath, 'utf8')));
+                        if (text)
+                            transcript = text;
+                    }
+                }
+                finally {
+                    try {
+                        unlinkSync(filePath);
+                    }
+                    catch { }
+                }
+            }
+            return transcript;
+        }
+        catch {
+            return null;
+        }
+    };
+    // 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) {
         try {
-            await run([]);
+            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' };
-            return { transcript: null, reason: 'no-transcript' };
+            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;
         }
     }
-    // Pick the best subtitle file and clean up all temp files.
-    // Priority: original-language ASR (-orig suffix) > English > first available.
-    // yt-dlp names the original-language auto-generated track with an "-orig" infix,
-    // e.g. zh-Hans-orig.json3 — this is the most reliable source for non-English videos.
-    try {
-        const files = readdirSync(dir).filter(f => f.startsWith(videoId) && f.endsWith('.json3'));
-        const pick = files.find(f => f.includes('-orig.')) ?? // original language ASR
-            files.find(f => /\.en(-\w+)?\.json3$/.test(f)) ?? // English (manual or ASR)
-            files[0]; // first available
-        let transcript = null;
-        for (const file of files) {
-            const filePath = join(dir, file);
-            try {
-                if (file === pick) {
-                    const text = parseTimedText(JSON.parse(readFileSync(filePath, 'utf8')));
-                    if (text)
-                        transcript = text;
-                }
-            }
-            finally {
-                try {
-                    unlinkSync(filePath);
-                }
-                catch { } // clean up every file, not just the picked one
-            }
-        }
-        if (transcript)
-            return { transcript };
+    // 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' };
+        const t = readAndClean();
+        if (t)
+            return { transcript: t };
+    }
+    // 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' };
+        const t = readAndClean();
+        if (t)
+            return { transcript: t };
     }
-    catch { }
     return { transcript: null, reason: 'no-transcript' };
 }
 // json3 parser — shared algorithm with browser extension

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "summd",
-  "version": "0.1.9",
+  "version": "0.1.11",
   "description": "CLI for sum.md — Sum to anything.",
   "license": "MIT",
   "bin": {