Fix YouTube API compatibility and update Gemini model to the new one

[wiecon-dev]

Summary

Fixed multiple compatibility issues with updated API dependencies and improved project configuration.

Changes Made

Bug Fixes

1. Fixed YouTube Transcript API compatibility (v1.2.3)

   • Updated extract_languages() function to use instance method ytt_api.list() instead of deprecated static
     method YouTubeTranscriptApi.list_transcripts()
   • Updated extract_transcript() function to use new API workflow: list() → find_transcript() → fetch()
   • Changed data access from dictionary notation i['text'] to object attribute i.text

2. Fixed Gemini AI model compatibility

   • Updated model name from deprecated gemini-pro to current gemini-2.5-flash
   • Resolved 404 error: "models/gemini-pro is not found for API version v1beta"

3. Fixed Streamlit deprecation warning

   • Replaced deprecated parameter use_column_width=True with use_container_width=True in st.image() calls
     (2 occurrences)

Configuration Improvements

4. Added .gitignore

   • Environment variables (.env)
   • Python cache files
   • Virtual environments
   • IDE files
   • OS-specific files
   • Streamlit secrets

5. Added .env.example

   • Provides template for required environment variables
   • Helps new users set up the application correctly

6. Removed sensitive files from repository

   • Removed .env file containing actual API keys from version control

Dependencies

7. Updated requirements.txt
   • Specified minimum version: youtube-transcript-api>=0.6.0

Testing

• ✅ Successfully tested with YouTube video URL
• ✅ Transcript extraction working correctly
• ✅ AI summary generation functional
• ✅ No deprecation warnings

Breaking Changes

None - all changes are backward compatible improvements.