Features:
- Delete entire channels with all videos and downloaded files
- Delete individual video files while keeping database entries
- Scheduled automatic cleanup of videos older than 7 days
- Proper cascading deletes with file cleanup
Channel Deletion:
- New DELETE endpoint at /api/channels/<id>
- Removes channel, all video entries, and downloaded files
- User ownership verification
- Returns count of deleted files
- UI button on channels page with detailed confirmation dialog
Video File Deletion:
- New DELETE endpoint at /api/videos/<id>/file
- Celery async task to remove file from disk
- Resets download status to pending (allows re-download)
- UI button on watch page for completed videos
- Confirmation dialog with clear warnings
Scheduled Cleanup:
- Celery beat configuration for periodic tasks
- cleanup_old_videos task runs daily at midnight
- Automatically deletes videos completed more than 7 days ago
- Removes files and resets database status
- scheduled_tasks.py for beat schedule configuration
- verify_schedule.py helper to check task scheduling
UI Improvements:
- Added .btn-danger CSS class (black/white theme)
- Delete buttons with loading states
- Detailed confirmation dialogs warning about permanent deletion
- Dashboard now filters to show only completed videos
Bug Fixes:
- Fixed navbar alignment issues
- Added proper error handling for file deletion
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Switch to light mode with black and white color scheme
- Simplify channel subscription to use channel ID only instead of RSS URL
- Add Downloads page to track all video download jobs
- Fix Flask-Login session management bug in user loader
- Always filter YouTube Shorts from feeds (case-insensitive)
- Fix download service video URL attribute error
- Fix watch page enum comparison for download status display
UI Changes:
- Update CSS to pure black/white/grayscale theme
- Remove colored text and buttons
- Use underlines for hover states instead of color changes
- Improve visual hierarchy with grayscale shades
Channel Subscription:
- Accept channel ID directly instead of full RSS URL
- Add validation for channel ID format (UC/UU prefix)
- Update help text and examples for easier onboarding
Downloads Page:
- New route at /downloads showing all video download jobs
- Display status, progress, and metadata for each download
- Sortable by status (downloading, pending, failed, completed)
- Actions to download, retry, or watch videos
- Responsive grid layout with thumbnails
Bug Fixes:
- Fix user loader to properly use database session context manager
- Fix download service accessing wrong attribute (link → video_url)
- Fix watch page template enum value comparisons
- Fix session detachment issues when accessing channel data
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
- Added yt-dlp, celery, and redis dependencies to pyproject.toml
- Extended VideoEntry model with download tracking fields:
- download_status (enum: pending, downloading, completed, failed)
- download_path, download_started_at, download_completed_at
- download_error, file_size
- Created celery_app.py with Redis broker configuration
- Created download_service.py with async download tasks:
- download_video() task downloads as MP4 format
- Configured yt-dlp for best MP4 quality with fallback
- Automatic retries on failure (max 3 attempts)
- Progress tracking and database updates
- Added Flask API endpoints in main.py:
- POST /api/download/<video_id> to trigger download
- GET /api/download/status/<video_id> to check status
- POST /api/download/batch for bulk downloads
- Generated and applied Alembic migration for new fields
- Created downloads/ directory for video storage
- Updated .gitignore to exclude downloads/ directory
- Updated CLAUDE.md with comprehensive documentation:
- Redis and Celery setup instructions
- Download workflow and architecture
- yt-dlp configuration details
- New API endpoint examples
Videos are downloaded as MP4 files using Celery workers.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
