Update .forgejo/workflows/generate-m3u.yml

.forgejo/workflows/generate-m3u.yml

@@ -1,4 +1,4 @@
-name: Generate M3U Playlist
+name: Generate M3U Playlist with Auto-Cleanup
 
 on:
   push:
@@ -7,7 +7,7 @@ on:
   workflow_dispatch:
 
 jobs:
-  build:
+  build-and-cleanup:
     runs-on: ubuntu-22.04
     steps:
     - name: Checkout Repository
@@ -23,104 +23,224 @@ jobs:
         git config --local user.email "actions@forgejo.plainrock127.xyz"
         git config --local user.name "IPTV Playlist Bot"
     
-    - name: Debug File Structure
+    - name: Pre-Cleanup Health Check
       run: |
-        echo "=== Root Directory ==="
+        echo "=== Repository Health Check ==="
-        ls -la
+        python -c "
-        echo ""
+        import sys, os
-        echo "=== Scripts Directory ==="
+        sys.path.insert(0, 'scripts')
-        ls -la scripts/
+        from repo_health import RepoHealthMonitor
-        echo ""
+        monitor = RepoHealthMonitor()
-        echo "=== Check for Import File ==="
+        health = monitor.run_health_check()
-        if [ -f bulk_import.m3u ]; then
+        print(f'Organization Score: {health[\"organization_score\"]}/100')
-          echo "✅ bulk_import.m3u found in root ($(wc -l < bulk_import.m3u) lines)"
+        print(f'Total Size: {health[\"repository_size\"][\"total_mb\"]:.1f} MB')
-        else
+        print(f'Files: {health[\"file_counts\"][\"total_files\"]}')
-          echo "❌ bulk_import.m3u not found in root"
+        if health['cleanup_suggestions']:
-        fi
+            print('Cleanup needed:')
+            for suggestion in health['cleanup_suggestions'][:3]:
+                print(f'  - {suggestion}')
+        " 2>/dev/null || echo "Health check skipped (first run)"
     
+    - name: Auto-Cleanup Repository
+      run: |
+        echo "=== Auto-Cleanup Phase ==="
+        
+        # Remove Python cache thoroughly
+        find . -type d -name "__pycache__" -exec rm -rf {} + 2>/dev/null || true
+        find . -name "*.pyc" -delete 2>/dev/null || true
+        find . -name "*.pyo" -delete 2>/dev/null || true
+        
+        # Remove temporary files
+        find . -name "*_temp*" -type f -delete 2>/dev/null || true
+        find . -name "*.tmp" -delete 2>/dev/null || true
+        find . -name "*~" -delete 2>/dev/null || true
+        
+        # Clean backup files older than 30 days
+        find backups -name "*.txt" -type f -mtime +30 -delete 2>/dev/null || true
+        
+        # Organize log files
+        mkdir -p reports/logs reports/archive
+        find . -maxdepth 1 -name "*.log" -exec mv {} reports/logs/ \; 2>/dev/null || true
+        
+        # Compress old backups (older than 7 days)
+        find backups -name "*.txt" -type f -mtime +7 -exec gzip {} \; 2>/dev/null || true
+        
+        echo "✅ Cleanup completed"
+
     - name: Setup Directories
       run: |
-        mkdir -p config
+        echo "=== Directory Setup ==="
-        mkdir -p backups
+        mkdir -p config backups reports/logs reports/archive templates
-        mkdir -p reports/logs
-        mkdir -p templates
         
         # Create scripts/__init__.py if missing
         if [ ! -f scripts/__init__.py ]; then
           echo '# Scripts package' > scripts/__init__.py
         fi
+        
+        # Ensure proper directory structure
+        ls -la scripts/ || echo "Scripts directory missing"
+
+    - name: Debug File Structure
+      run: |
+        echo "=== File Structure Debug ==="
+        echo "Root files:"
+        ls -la | head -20
+        echo ""
+        echo "Scripts directory:"
+        ls -la scripts/ 2>/dev/null || echo "Scripts directory not found"
+        echo ""
+        echo "Import file status:"
+        if [ -f bulk_import.m3u ]; then
+          echo "✅ bulk_import.m3u found ($(wc -l < bulk_import.m3u) lines, $(du -h bulk_import.m3u | cut -f1))"
+          echo "First few lines:"
+          head -5 bulk_import.m3u
+        else
+          echo "❌ bulk_import.m3u not found"
+        fi
 
     - name: Run Playlist Generation
       run: |
-        echo "=== Running playlist generation ==="
+        echo "=== Playlist Generation ==="
         python scripts/generate_playlist.py
         
-    - name: Check Results
+    - name: Post-Generation Health Check
       run: |
-        echo "=== Final Results ==="
+        echo "=== Post-Generation Analysis ==="
-        ls -la
         
         if [ -f playlist.m3u ]; then
-          CHANNEL_COUNT=$(grep -c "^#EXTINF" playlist.m3u || echo "0")
+          CHANNEL_COUNT=$(grep -c "^#EXTINF" playlist.m3u 2>/dev/null || echo "0")
-          echo "✅ playlist.m3u generated with $CHANNEL_COUNT channels"
+          FILE_SIZE=$(du -h playlist.m3u | cut -f1)
-          echo "File size: $(du -h playlist.m3u | cut -f1)"
+          echo "✅ playlist.m3u generated:"
+          echo "   - Channels: $CHANNEL_COUNT"
+          echo "   - Size: $FILE_SIZE"
+          
+          # Show top countries
+          echo "   - Top countries:"
+          grep 'group-title=' playlist.m3u | sed 's/.*group-title="//; s/".*//' | sort | uniq -c | sort -nr | head -5 || echo "     No country data"
         else
           echo "❌ playlist.m3u not generated"
         fi
         
         if [ -f channels.txt ]; then
-          echo "=== channels.txt status ==="
+          CHANNELS_SIZE=$(du -h channels.txt | cut -f1)
-          echo "Size: $(du -h channels.txt | cut -f1)"
+          CHANNELS_LINES=$(wc -l < channels.txt)
-          echo "Lines: $(wc -l < channels.txt)"
+          echo "📁 channels.txt: $CHANNELS_SIZE ($CHANNELS_LINES lines)"
         fi
         
-        # Check logs
+        # Check for any error logs
-        if [ -d reports/logs ]; then
+        if [ -f reports/logs/playlist_update.log ]; then
-          echo "=== Logs directory ==="
+          echo "📋 Recent log entries:"
-          ls -la reports/logs/
+          tail -10 reports/logs/playlist_update.log || tail -10 playlist_update.log 2>/dev/null || echo "No logs found"
+        fi
+
+    - name: Repository Cleanup & Organization
+      run: |
+        echo "=== Final Repository Organization ==="
+        
+        # Remove any remaining cache files
+        find . -type d -name "__pycache__" -exec rm -rf {} + 2>/dev/null || true
+        
+        # Move any stray log files
+        find . -maxdepth 1 -name "*.log" -exec mv {} reports/logs/ \; 2>/dev/null || true
+        
+        # Clean up any editor backup files
+        find . -name "*~" -delete 2>/dev/null || true
+        find . -name "*.swp" -delete 2>/dev/null || true
+        
+        # Ensure bulk_import.m3u is clean for next use
+        if [ -f bulk_import.m3u ]; then
+          # If it has more than just the header, it means it wasn't cleaned by the script
+          LINE_COUNT=$(wc -l < bulk_import.m3u)
+          if [ "$LINE_COUNT" -gt 2 ]; then
+            echo "⚠️  Manually cleaning bulk_import.m3u (had $LINE_COUNT lines)"
+            echo '#EXTM3U' > bulk_import.m3u
+            echo '' >> bulk_import.m3u
+          fi
         fi
         
+        echo "✅ Repository organized"
+
     - name: Commit Changes
       run: |
-        echo "=== Preparing to commit (excluding docs) ==="
+        echo "=== Committing Changes ==="
         
-        # Remove docs folder from tracking if it exists
+        # Add specific files/directories (exclude unnecessary items)
-        if [ -d docs ]; then
-          echo "Removing docs folder from repository"
-          git rm -rf docs/ || true
-        fi
-        
-        # Remove old playlist_update.log from root if it exists
-        if [ -f playlist_update.log ]; then
-          echo "Removing old playlist_update.log from root"
-          git rm playlist_update.log || true
-        fi
-        
-        # Add specific files/directories only
         git add bulk_import.m3u || true
         git add channels.txt || true
         git add playlist.m3u || true
         git add scripts/ || true
         git add config/ || true
         git add reports/ || true
-        git add backups/ || true
+        git add backups/*.gz || true  # Only compressed backups
         git add templates/ || true
         git add .forgejo/ || true
         git add README.md || true
+        git add .gitignore || true
         
-        # Explicitly do NOT add docs folder or old logs
+        # Remove any files that shouldn't be tracked
-        echo "Skipping docs/ folder and old logs"
+        git rm --cached *.log 2>/dev/null || true
+        git rm --cached **/__pycache__/** 2>/dev/null || true
+        git rm --cached **/*.pyc 2>/dev/null || true
+        
+        # Check what we're about to commit
+        echo "Files to be committed:"
+        git diff --staged --name-only || echo "No staged changes"
         
         if ! git diff --staged --quiet; then
+          # Calculate channel count for commit message
           CHANNEL_COUNT="0"
           if [ -f playlist.m3u ]; then
-            CHANNEL_COUNT=$(grep -c "^#EXTINF" playlist.m3u || echo "0")
+            CHANNEL_COUNT=$(grep -c "^#EXTINF" playlist.m3u 2>/dev/null || echo "0")
           fi
           
-          git commit -m "📺 Updated playlist: $CHANNEL_COUNT channels ($(date '+%Y-%m-%d %H:%M')) - Clean repo"
+          # Calculate repository size
+          REPO_SIZE=$(du -sh . 2>/dev/null | cut -f1 || echo "unknown")
+          
+          # Create informative commit message
+          COMMIT_MSG="📺 Updated playlist: $CHANNEL_COUNT channels ($(date '+%Y-%m-%d %H:%M')) 
+
+Repository Status:
+- Size: $REPO_SIZE
+- Channels: $CHANNEL_COUNT
+- Auto-cleaned and organized
+- $(date '+%Y-%m-%d %H:%M:%S UTC')"
+          
+          git commit -m "$COMMIT_MSG"
           git push
           
-          echo "✅ Repository updated (docs excluded)"
+          echo "✅ Repository updated successfully"
+          echo "   📺 Channels: $CHANNEL_COUNT"
+          echo "   📁 Size: $REPO_SIZE"
         else
-          echo "ℹ️ No changes to commit"
+          echo "ℹ️  No changes to commit"
-        fi
+        fi
+
+    - name: Final Status Report
+      run: |
+        echo "=== Final Status Report ==="
+        echo "🎯 Workflow completed successfully!"
+        echo ""
+        echo "📊 Repository Statistics:"
+        echo "   - Total files: $(find . -type f | wc -l)"
+        echo "   - Directory size: $(du -sh . | cut -f1)"
+        echo "   - Git status: $(git status --porcelain | wc -l) pending changes"
+        echo ""
+        
+        if [ -f playlist.m3u ]; then
+          CHANNEL_COUNT=$(grep -c "^#EXTINF" playlist.m3u 2>/dev/null || echo "0")
+          echo "✅ Playlist Status:"
+          echo "   - Channels: $CHANNEL_COUNT"
+          echo "   - File size: $(du -h playlist.m3u | cut -f1)"
+          echo "   - Last updated: $(date)"
+        else
+          echo "❌ No playlist generated"
+        fi
+        
+        echo ""
+        echo "🧹 Cleanup Summary:"
+        echo "   - Python cache: Removed"
+        echo "   - Log files: Organized in reports/logs/"
+        echo "   - Backup files: Compressed if older than 7 days"
+        echo "   - Temporary files: Cleaned"
+        echo ""
+        echo "🚀 Ready for next import!"