Improve YouTube download with CLI subprocess and multi-browser cookie support

app.py

@@ -266,43 +266,92 @@ footer { display: none !important; }
 # ============================================================================
 
 def download_youtube(url: str) -> tuple:
-    """Download audio from YouTube"""
-    if not HAS_YTDLP:
+    """Download audio from YouTube using CLI for better cookie support"""
+    import subprocess
+    import shutil
+
+    yt_dlp_path = shutil.which('yt-dlp')
+    if not yt_dlp_path:
         return None, "yt-dlp not installed"
 
     try:
-        output_path = OUTPUT_DIR / f"yt_{datetime.now().strftime('%H%M%S')}.wav"
-        ydl_opts = {
-            'format': 'bestaudio/best',
-            'outtmpl': str(output_path).replace('.wav', ''),
-            'postprocessors': [{
-                'key': 'FFmpegExtractAudio',
-                'preferredcodec': 'wav',
-            }],
-            'quiet': True,
-            'no_warnings': True,
-            'socket_timeout': 60,
-            'retries': 5,
-            'extractor_args': {
-                'youtube': {
-                    'player_client': ['web', 'android', 'ios'],
-                }
-            },
-        }
-
-        # Try with browser cookies on local, skip on HF Space
+        output_dir = OUTPUT_DIR / f"yt_{datetime.now().strftime('%H%M%S')}"
+        output_dir.mkdir(parents=True, exist_ok=True)
+        audio_path = output_dir / "audio.wav"
+
+        # Check for cookies.txt file
+        cookies_file = Path(__file__).parent / "cookies.txt"
+
+        def build_cmd(with_cookies=None):
+            cmd = [yt_dlp_path, '--socket-timeout', '60', '--retries', '5', '--no-warnings']
+            if with_cookies == 'file' and cookies_file.exists():
+                cmd.extend(['--cookies', str(cookies_file)])
+            elif with_cookies and not IS_HF_SPACE:
+                cmd.extend(['--cookies-from-browser', with_cookies])
+            return cmd
+
+        # Try different cookie sources (skip browser cookies on HF Space)
+        cookie_sources = []
+        if cookies_file.exists():
+            cookie_sources.append('file')
         if not IS_HF_SPACE:
-            ydl_opts['cookiesfrombrowser'] = ('chrome',)
+            cookie_sources.extend(['chrome', 'edge', 'firefox', 'brave'])
+        cookie_sources.append(None)  # Try without cookies last
+
+        title = 'Unknown'
+        download_success = False
+        last_error = ""
+
+        for cookie_source in cookie_sources:
+            source_name = "cookies.txt" if cookie_source == 'file' else (cookie_source or 'no cookies')
+            print(f"Trying YouTube download with {source_name}...")
+
+            base_cmd = build_cmd(cookie_source)
+
+            # Get title
+            title_cmd = base_cmd + ['--print', 'title', '--no-download', url]
+            result = subprocess.run(title_cmd, capture_output=True, text=True, timeout=30)
+
+            if result.returncode == 0 and result.stdout.strip():
+                title = result.stdout.strip()
+
+                # Download audio
+                audio_cmd = base_cmd + [
+                    '-f', 'bestaudio/best',
+                    '-x', '--audio-format', 'wav',
+                    '-o', str(audio_path).replace('.wav', '.%(ext)s'),
+                    url
+                ]
+
+                print(f"Downloading: {title}")
+                result = subprocess.run(audio_cmd, capture_output=True, text=True, timeout=300)
+
+                if result.returncode == 0:
+                    download_success = True
+                    break
+
+            last_error = result.stderr if result.stderr else "Unknown error"
+            if 'Sign in' not in last_error and 'bot' not in last_error.lower():
+                break
+
+        if not download_success:
+            if IS_HF_SPACE:
+                return None, "YouTube requires authentication on HuggingFace. Please upload the audio file directly."
+            return None, f"{last_error}\n\nTry uploading the audio file directly instead."
+
+        # Find the audio file
+        if not audio_path.exists():
+            for f in output_dir.glob('audio.*'):
+                if f.suffix == '.wav':
+                    audio_path = f
+                    break
+
+        return str(audio_path), title
 
-        with yt_dlp.YoutubeDL(ydl_opts) as ydl:
-            info = ydl.extract_info(url, download=True)
-            title = info.get('title', 'Unknown')
-        return str(output_path), title
+    except subprocess.TimeoutExpired:
+        return None, "Download timed out. Try a shorter video."
     except Exception as e:
-        error_msg = str(e)
-        if 'Sign in to confirm' in error_msg or 'bot' in error_msg.lower():
-            return None, "YouTube requires authentication. Please upload the audio file directly instead, or try a different video."
-        return None, error_msg
+        return None, str(e)
 
 @spaces.GPU(duration=120)
 def separate_stems(audio_path: str, progress=None) -> dict: