Add download_captions function to download YouTube captions with English preference

youtube_transcript_downloader.py

@@ -122,15 +122,10 @@ def main():
     print(f"Video ID: {video_id}")
     print("Fetching transcript...")
     
-    # Try to get English transcript first
-    transcript = get_transcript(video_id, ['en', 'en-US', 'en-GB'])
+    # Download captions with preference for English
+    result = download_captions(video_id)
     
-    if transcript is None:
-        # Try to get transcript in any language
-        print("\nNo English transcript found. Trying to get transcript in original language...")
-        transcript = get_transcript(video_id, [])
-    
-    if transcript is None:
+    if not result['success']:
         print("\nCould not retrieve transcript.")
         print("Possible reasons:")
         print("- Video does not have captions/subtitles")
@@ -139,14 +134,15 @@ def main():
         sys.exit(1)
     
     # Format original transcript
-    original_formatted = format_transcript(transcript, include_timestamps=True)
+    original_formatted = format_transcript(result['transcript'], include_timestamps=True)
     
     # Check if translation is needed (try to detect if it's already English)
-    original_text = get_transcript_text(transcript)
+    original_text = get_transcript_text(result['transcript'])
     
     # Ask user if they want to translate
     print("\n" + "=" * 45)
     print("Transcript successfully retrieved!")
+    print(f"Language: {result['language']}")
     
     # Determine if translation is needed
     # If transcript is in English, no translation needed
@@ -236,5 +232,54 @@ def is_english(text):
     return matches > len(words) * 0.05 if words else False
 
 
+def download_captions(video_id):
+    """
+    Download closed captions from a YouTube video, preferring English.
+    
+    Args:
+        video_id: The YouTube video ID
+        
+    Returns:
+        Dictionary with 'transcript', 'language', and 'success' keys,
+        or None if no transcript is available
+    """
+    try:
+        # First, try to get English transcript (preferably en, en-US, en-GB)
+        english_languages = ['en', 'en-US', 'en-GB']
+        transcript = YouTubeTranscriptApi().fetch(video_id, languages=english_languages)
+        
+        if transcript:
+            return {
+                'transcript': transcript,
+                'language': transcript.language,
+                'success': True
+            }
+        
+        # If no English transcript is available, try any language
+        print("No English transcript found. Trying to get transcript in original language...")
+        transcript = YouTubeTranscriptApi().fetch(video_id, languages=[])
+        
+        if transcript:
+            return {
+                'transcript': transcript,
+                'language': transcript.language,
+                'success': True
+            }
+            
+        return {
+            'transcript': None,
+            'language': None,
+            'success': False
+        }
+        
+    except Exception as e:
+        print(f"Error downloading captions: {e}")
+        return {
+            'transcript': None,
+            'language': None,
+            'success': False
+        }
+
+
 if __name__ == "__main__":
     main()