Update app.py

app.py

@@ -13,7 +13,7 @@ from huggingface_hub import HfApi, hf_hub_download
 from typing import Optional, List, Dict, Any
 
 print("=" * 60)
-print("🚀 TEXT STYLING API - WITHOUT GRADIO")
+print("🚀 TEXT STYLING API - ASS SUBTITLE METHOD")
 print("=" * 60)
 
 # =============================================
@@ -151,8 +151,11 @@ def get_font_path(font_family):
             return font_info["path"]
     return None
 
+# =============================================
+# ASS SUBTITLE METHOD - WORKS FOR CHINESE
+# =============================================
 def create_text_overlay(input_video, output_video, text_style):
-    """Add text overlay to video - FIXED for Chinese fonts"""
+    """Add text overlay using ASS subtitles - WORKS FOR CHINESE"""
     font_path = get_font_path(text_style.font_family)
     if not font_path:
         print(f"⚠️ Font not found: {text_style.font_family}")
@@ -160,75 +163,88 @@ def create_text_overlay(input_video, output_video, text_style):
     
     print(f"✅ Using font: {font_path}")
     
-    # Register font with fontconfig (so FFmpeg can find it properly)
-    font_dir = os.path.dirname(font_path)
-    font_name = os.path.basename(font_path).split('.')[0]
+    # Create working directory for ASS file
+    work_dir = os.path.dirname(output_video)
+    ass_file = os.path.join(work_dir, "subtitle.ass")
     
-    # Create fontconfig configuration
-    fc_config = f"""<?xml version="1.0"?>
-<!DOCTYPE fontconfig SYSTEM "fonts.dtd">
-<fontconfig>
-    <dir>{font_dir}</dir>
-    <alias>
-        <family>myfont</family>
-        <prefer>
-            <family>{font_name}</family>
-        </prefer>
-    </alias>
-</fontconfig>"""
+    # Color mapping
+    color_map = {
+        "white": "FFFFFF", "black": "000000", "red": "FF0000",
+        "green": "00FF00", "blue": "0000FF", "yellow": "FFFF00",
+        "gold": "FFD700", "purple": "800080", "cyan": "00FFFF"
+    }
     
-    fc_path = "/tmp/fonts.conf"
-    with open(fc_path, 'w') as f:
-        f.write(fc_config)
+    # Get font color
+    font_color_hex = color_map.get(text_style.color.lower(), "FFFFFF")
     
-    os.environ['FONTCONFIG_FILE'] = fc_path
+    # Parse background color
+    bg_parts = text_style.bg_color.split('@')
+    bg_color_name = bg_parts[0]
+    bg_opacity = float(bg_parts[1]) if len(bg_parts) > 1 else 0.5
+    bg_color_hex = color_map.get(bg_color_name.lower(), "000000")
+    bg_alpha = int((1 - bg_opacity) * 255)  # ASS opacity: 0=transparent, 255=opaque
     
-    # Position mapping
+    # Map position to ASS alignment (1-9)
     pos_map = {
-        "top-left": f"x={text_style.margin}:y={text_style.margin}",
-        "top-right": f"x=w-tw-{text_style.margin}:y={text_style.margin}",
-        "center": f"x=(w-tw)/2:y=(h-th)/2",
-        "bottom-left": f"x={text_style.margin}:y=h-th-{text_style.margin}",
-        "bottom-right": f"x=w-tw-{text_style.margin}:y=h-th-{text_style.margin}",
-        "top-center": f"x=(w-tw)/2:y={text_style.margin}",
-        "bottom-center": f"x=(w-tw)/2:y=h-th-{text_style.margin}"
+        "bottom-left": 1,
+        "bottom-center": 2,
+        "bottom-right": 3,
+        "left": 4,
+        "center": 5,
+        "right": 6,
+        "top-left": 7,
+        "top-center": 8,
+        "top-right": 9
     }
-    position = pos_map.get(text_style.position, pos_map["center"])
+    alignment = pos_map.get(text_style.position, 5)
     
-    # Parse background color
-    bg_parts = text_style.bg_color.split('@')
-    bg_color = bg_parts[0]
-    bg_opacity = float(bg_parts[1]) if len(bg_parts) > 1 else 0.5
+    # Calculate margins
+    margin_l = text_style.margin if alignment in [1,4,7] else 0
+    margin_r = text_style.margin if alignment in [3,6,9] else 0
+    margin_v = text_style.margin
+    
+    # Get font name from path (remove extension)
+    font_name = os.path.basename(font_path).split('.')[0]
+    
+    # Create ASS file content with proper Chinese support
+    ass_content = f"""[Script Info]
+; Script generated by Video Styling Space
+ScriptType: v4.00+
+PlayResX: 1920
+PlayResY: 1080
+ScaledBorderAndShadow: yes
+
+[V4+ Styles]
+Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
+Style: Default,{font_name},{text_style.font_size},&H00{font_color_hex},&H000000FF,&H00000000,&H{bg_alpha:02X}{bg_color_hex},0,0,0,0,100,100,0,0,1,1,0,{alignment},{margin_l},{margin_r},{margin_v},1
+
+[Events]
+Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
+Dialogue: 0,0:00:00.00,0:00:10.00,Default,,0,0,0,,{text_style.text}"""
     
-    # Use font family name instead of direct path
-    drawtext_filter = (
-        f"drawtext=text='{text_style.text}':"
-        f"fontfamily={font_name}:"
-        f"fontsize={text_style.font_size}:"
-        f"fontcolor={text_style.color}:"
-        f"{position}:"
-        f"box=1:"
-        f"boxcolor={bg_color}@{bg_opacity}:"
-        f"boxborderw={text_style.padding}"
-    )
+    # Write ASS file with UTF-8 encoding (essential for Chinese)
+    with open(ass_file, 'w', encoding='utf-8') as f:
+        f.write(ass_content)
     
-    print(f"🎬 FFmpeg filter: {drawtext_filter}")
+    print(f"📝 Created ASS subtitle file")
     
+    # Use FFmpeg to burn subtitles
     cmd = [
         'ffmpeg', '-y',
         '-i', input_video,
-        '-vf', drawtext_filter,
+        '-vf', f"ass={ass_file}",
         '-c:a', 'copy',
         output_video
     ]
     
+    print(f"🎬 Running FFmpeg with ASS filter...")
     result = subprocess.run(cmd, capture_output=True, text=True)
     
     if result.returncode != 0:
         print(f"❌ FFmpeg error: {result.stderr}")
         return False
     
-    print(f"✅ Video styled successfully")
+    print(f"✅ Video styled successfully with ASS subtitles")
     return True
 
 # =============================================
@@ -251,29 +267,46 @@ async def style_video(request: StylingRequest):
         work_dir = f"/tmp/styling/{request.project_id}_{uuid.uuid4().hex[:8]}"
         os.makedirs(work_dir, exist_ok=True)
         
+        # Download video
         video_path = os.path.join(work_dir, "input.mp4")
         download_file(request.video_url, video_path)
         
         current_video = video_path
         
+        # Add title overlay if provided
         if request.title_overlay:
             titled_path = os.path.join(work_dir, "titled.mp4")
             if create_text_overlay(current_video, titled_path, request.title_overlay):
                 current_video = titled_path
         
+        # Upload styled video
         timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
         styled_filename = f"styled_{timestamp}.mp4"
         styled_url = upload_to_dataset(current_video, request.project_id, styled_filename, "videos")
         
+        # Cleanup
         shutil.rmtree(work_dir, ignore_errors=True)
         
         if styled_url:
-            return StylingResponse(status="success", project_id=request.project_id, styled_video_url=styled_url)
+            return StylingResponse(
+                status="success", 
+                project_id=request.project_id, 
+                styled_video_url=styled_url
+            )
         else:
-            return StylingResponse(status="error", project_id=request.project_id, error="Upload failed")
+            return StylingResponse(
+                status="error", 
+                project_id=request.project_id, 
+                error="Failed to upload styled video"
+            )
         
     except Exception as e:
-        return StylingResponse(status="error", project_id=request.project_id, error=str(e))
+        print(f"❌ Error: {e}")
+        return StylingResponse(
+            status="error", 
+            project_id=request.project_id, 
+            error=str(e)
+        )
 
 @app.get("/")
 async def root():
@@ -288,64 +321,6 @@ async def root():
         "fonts_loaded": len(FONTS)
     }
 
-@app.get("/test-font/{font_name}")
-async def test_font(font_name: str):
-    """Generate a test image with the specified font"""
-    try:
-        # Create a simple test image
-        from PIL import Image, ImageDraw, ImageFont
-        import io
-        import base64
-        
-        # Get font path
-        font_path = None
-        for key, font_info in FONTS.items():
-            if font_name in key or font_name in font_info["name"]:
-                font_path = font_info["path"]
-                break
-        
-        if not font_path:
-            return {"error": f"Font {font_name} not found"}
-        
-        # Create image
-        img = Image.new('RGB', (800, 400), color='white')
-        d = ImageDraw.Draw(img)
-        
-        # Test texts
-        test_texts = [
-            "English Text Test",
-            "中文测试文本",
-            "荆南麦圆体测试",
-            f"Font: {font_name}"
-        ]
-        
-        y_position = 50
-        for text in test_texts:
-            try:
-                font = ImageFont.truetype(font_path, 36)
-                d.text((50, y_position), text, fill='black', font=font)
-                y_position += 60
-            except:
-                d.text((50, y_position), f"Failed to render: {text}", fill='red', font=ImageFont.load_default())
-                y_position += 60
-        
-        # Save to bytes
-        img_bytes = io.BytesIO()
-        img.save(img_bytes, format='PNG')
-        img_bytes = img_bytes.getvalue()
-        
-        # Return as base64
-        return {
-            "font_name": font_name,
-            "font_path": font_path,
-            "test_image": base64.b64encode(img_bytes).decode('utf-8'),
-            "message": "Image contains: English, Chinese, and font name"
-        }
-        
-    except Exception as e:
-        return {"error": str(e)}
-
-
 # =============================================
 # RUN
 # =============================================