Update app.py

app.py

@@ -12,21 +12,25 @@ from PIL import Image
 import io
 import json
 
-# Hugging Face imports
-from huggingface_hub import InferenceClient
-from transformers import pipeline
-
 # Configure logging
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
-# Initialize Hugging Face
+# Try to import Hugging Face (will work in HF Spaces)
+try:
+    from huggingface_hub import InferenceClient
+    HF_AVAILABLE = True
+except ImportError:
+    HF_AVAILABLE = False
+    logger.warning("huggingface_hub not available - using mock mode")
+
+# Initialize Hugging Face (token will be available in HF Spaces secrets)
 HF_TOKEN = os.getenv("HF_TOKEN") or os.getenv("HUGGINGFACE_TOKEN")
-if HF_TOKEN:
+if HF_TOKEN and HF_AVAILABLE:
     client = InferenceClient(token=HF_TOKEN)
     logger.info("✓ Hugging Face client initialized")
 else:
-    logger.warning("⚠ HF_TOKEN not found - using mock mode")
+    logger.warning("⚠ HF_TOKEN not found or huggingface_hub not available - using mock mode")
     client = None
 
 # Initialize FastAPI
@@ -45,7 +49,7 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
-# Mock data (keep your existing COSMETICS and PRODUCTS)
+# Mock data
 COSMETICS = [
     {
         "id": "cos001",
@@ -133,7 +137,10 @@ class ConfidenceMeter:
         avg_confidence = sum(model_scores) / len(model_scores)
         
         # Calculate variance (how much models agree)
-        variance = np.var(model_scores) if len(model_scores) > 1 else 0
+        if len(model_scores) > 1:
+            variance = sum((x - avg_confidence) ** 2 for x in model_scores) / len(model_scores)
+        else:
+            variance = 0
         
         # Adjust confidence based on agreement
         agreement_factor = 1 - (variance / 100)  # Higher agreement = higher confidence
@@ -199,7 +206,7 @@ class ConfidenceMeter:
         total_weight = 0
         
         for c in condition_confidences:
-            weight = weights.get(c["condition"].lower(), weights["default"])
+            weight = weights.get(c.get("condition", "").lower(), weights["default"])
             weighted_sum += c["confidence"] * weight
             total_weight += weight
         
@@ -222,8 +229,7 @@ class ConfidenceMeter:
             "level": level,
             "average_per_condition": round(avg_confidence, 2),
             "weighted_score": round(weighted_avg, 2),
-            "conditions_analyzed": len(condition_confidences),
-            "confidence_breakdown": condition_confidences
+            "conditions_analyzed": len(condition_confidences)
         }
     
     @staticmethod
@@ -253,92 +259,9 @@ class ConfidenceMeter:
 confidence_meter = ConfidenceMeter()
 
 # ============================================
-# REAL HUGGING FACE INTEGRATION
+# MOCK ANALYSIS FUNCTION (for HF Spaces)
 # ============================================
 
-async def analyze_with_huggingface(image_bytes: bytes) -> Dict[str, Any]:
-    """
-    Use Hugging Face models for skin analysis
-    """
-    if not client:
-        logger.warning("Using mock analysis (HF token not configured)")
-        return mock_skin_analysis_with_confidence()
-    
-    try:
-        # 1. Skin type classification
-        # Using a vision transformer model
-        skin_type_result = client.image_classification(
-            image=image_bytes,
-            model="microsoft/resnet-50"  # Replace with actual skin analysis model
-        )
-        
-        # 2. Skin condition detection
-        # Using a specialized model
-        condition_result = client.image_classification(
-            image=image_bytes,
-            model="google/vit-base-patch16-224"  # Replace with actual condition model
-        )
-        
-        # 3. Process results with confidence
-        conditions_with_confidence = []
-        model_scores_map = {}
-        
-        # Process skin type
-        skin_type = skin_type_result[0]["label"] if skin_type_result else "combination"
-        skin_type_confidence = skin_type_result[0]["score"] * 100 if skin_type_result else 75
-        
-        # Process conditions (simplified for demo)
-        conditions = [
-            {"name": "hyperpigmentation", "score": random.uniform(0.6, 0.95)},
-            {"name": "oiliness", "score": random.uniform(0.5, 0.9)},
-            {"name": "texture", "score": random.uniform(0.55, 0.85)},
-        ]
-        
-        # Calculate confidence for each condition
-        for condition in conditions:
-            # Simulate multiple model scores (in production, you'd use actual ensemble results)
-            model_scores = [
-                condition["score"] * 100,
-                condition["score"] * random.uniform(0.9, 1.1) * 100,
-                condition["score"] * random.uniform(0.85, 1.05) * 100
-            ]
-            
-            condition_confidence = confidence_meter.calculate_condition_confidence(
-                condition["name"],
-                model_scores
-            )
-            conditions_with_confidence.append(condition_confidence)
-        
-        # Calculate overall confidence
-        overall_confidence = confidence_meter.calculate_overall_confidence(conditions_with_confidence)
-        
-        # Detect skin tone (you'd use a color analysis model)
-        skin_tone = {
-            "primary_tone": "medium",
-            "undertone": "warm",
-            "confidence": random.uniform(70, 95),
-            "hex": "#B47850"
-        }
-        
-        return {
-            "conditions": conditions_with_confidence,
-            "skin_tone": skin_tone,
-            "skin_type": {
-                "type": skin_type,
-                "confidence": skin_type_confidence
-            },
-            "confidence": overall_confidence,
-            "dermatologist_warning": overall_confidence["overall"] < 70,
-            "model_info": {
-                "models_used": ["resnet-50", "vit-base", "custom-ensemble"],
-                "ensemble_size": 3
-            }
-        }
-        
-    except Exception as e:
-        logger.error(f"Hugging Face analysis failed: {e}")
-        return mock_skin_analysis_with_confidence()
-
 def mock_skin_analysis_with_confidence() -> Dict[str, Any]:
     """
     Generate mock skin analysis results with realistic confidence scores
@@ -421,6 +344,7 @@ async def root():
         "version": "2.0.0",
         "status": "operational",
         "hf_token_configured": bool(HF_TOKEN),
+        "hf_available": HF_AVAILABLE,
         "confidence_meter": "enabled",
         "endpoints": [
             "/health",
@@ -449,7 +373,7 @@ async def analyze_skin(request: Request):
         conditions = body.get("conditions", [])
         
         # Get analysis with confidence
-        analysis = mock_skin_analysis_with_confidence() if not client else await analyze_with_huggingface(b"")
+        analysis = mock_skin_analysis_with_confidence()
         
         # Override with manual conditions if provided
         if conditions:
@@ -502,8 +426,8 @@ async def analyze_skin_with_image(file: UploadFile = File(...)):
         # Read image
         contents = await file.read()
         
-        # Get analysis with confidence using Hugging Face
-        analysis = await analyze_with_huggingface(contents) if client else mock_skin_analysis_with_confidence()
+        # Get analysis with confidence (mock for HF Spaces)
+        analysis = mock_skin_analysis_with_confidence()
         
         # Generate product recommendations
         recommendations = {
@@ -542,12 +466,15 @@ async def analyze_skin_with_image(file: UploadFile = File(...)):
 @app.get("/api/confidence/test")
 async def test_confidence():
     """Test endpoint to see confidence meter in action"""
+    # Create sample condition confidences
+    condition1 = confidence_meter.calculate_condition_confidence("acne", [85, 82, 88])
+    condition2 = confidence_meter.calculate_condition_confidence("hyperpigmentation", [75, 72, 78])
+    condition3 = confidence_meter.calculate_condition_confidence("wrinkles", [65, 62, 58])
+    
     return {
         "confidence_meter": {
             "current": confidence_meter.calculate_overall_confidence([
-                confidence_meter.calculate_condition_confidence("acne", [85, 82, 88]),
-                confidence_meter.calculate_condition_confidence("hyperpigmentation", [75, 72, 78]),
-                confidence_meter.calculate_condition_confidence("wrinkles", [65, 62, 58])
+                condition1, condition2, condition3
             ]),
             "color_map": {
                 "high": confidence_meter.get_confidence_color(90),
@@ -560,7 +487,8 @@ async def test_confidence():
                 "moderate": confidence_meter.get_confidence_emoji(75),
                 "fair": confidence_meter.get_confidence_emoji(60),
                 "low": confidence_meter.get_confidence_emoji(40)
-            }
+            },
+            "sample_conditions": [condition1, condition2, condition3]
         }
     }
 
@@ -630,17 +558,8 @@ async def get_products(category: str = None, limit: int = 20):
 
 def skin_analysis_ui(image, manual_conditions):
     """Gradio interface for skin analysis with confidence meter"""
-    if image:
-        # Convert PIL image to bytes
-        img_byte_arr = io.BytesIO()
-        image.save(img_byte_arr, format='PNG')
-        img_byte_arr = img_byte_arr.getvalue()
-        
-        # Analyze with image
-        result = mock_skin_analysis_with_confidence() if not client else analyze_with_huggingface(img_byte_arr)
-    else:
-        # Analyze with manual conditions
-        result = mock_skin_analysis_with_confidence()
+    # Analyze with image or manual conditions
+    result = mock_skin_analysis_with_confidence()
     
     # Build HTML with confidence meter
     confidence = result["confidence"]["overall"]
@@ -670,7 +589,7 @@ def skin_analysis_ui(image, manual_conditions):
         <div style="background: white; border-radius: 15px; padding: 20px; box-shadow: 0 4px 6px rgba(0,0,0,0.1);">
             <h2 style="color: #333; margin-bottom: 20px;">📊 Analysis Results</h2>
             
-            {/* Overall Confidence Meter */}
+            <!-- Overall Confidence Meter -->
             <div style="margin-bottom: 30px;">
                 <div style="display: flex; justify-content: space-between; align-items: center; margin-bottom: 10px;">
                     <span style="font-size: 1.2em; font-weight: bold;">Overall Confidence {confidence_emoji}</span>
@@ -685,7 +604,7 @@ def skin_analysis_ui(image, manual_conditions):
                 </div>
             </div>
             
-            {/* Skin Type */}
+            <!-- Skin Type -->
             <div style="margin-bottom: 20px; padding: 15px; background: #f8f9fa; border-radius: 10px;">
                 <div style="display: flex; justify-content: space-between; margin-bottom: 10px;">
                     <strong>🧴 Skin Type</strong>
@@ -699,7 +618,7 @@ def skin_analysis_ui(image, manual_conditions):
                 </div>
             </div>
             
-            {/* Skin Tone */}
+            <!-- Skin Tone -->
             <div style="margin-bottom: 20px; padding: 15px; background: #f8f9fa; border-radius: 10px;">
                 <div style="display: flex; justify-content: space-between; margin-bottom: 10px;">
                     <strong>🎨 Skin Tone</strong>
@@ -718,21 +637,19 @@ def skin_analysis_ui(image, manual_conditions):
                 </div>
             </div>
             
-            {/* Detected Conditions */}
+            <!-- Detected Conditions -->
             <h3 style="margin: 20px 0 10px;">🔍 Detected Conditions</h3>
             {conditions_html}
             
-            {/* Models Used */}
+            <!-- Models Used -->
             <div style="margin-top: 20px; font-size: 0.9em; color: #999;">
                 Analysis using {result['model_info']['models_used'][0] if 'model_info' in result else 'ensemble model'}
-                {', ' + result['model_info']['models_used'][1] if 'model_info' in result and len(result['model_info']['models_used']) > 1 else ''}
-                {', ' + result['model_info']['models_used'][2] if 'model_info' in result and len(result['model_info']['models_used']) > 2 else ''}
             </div>
             
-            {/* Disclaimer */}
+            <!-- Disclaimer -->
             <div style="margin-top: 30px; padding: 15px; background: #fff3cd; border: 1px solid #ffeeba; border-radius: 10px; color: #856404;">
                 <strong>⚠️ Important:</strong> {result['disclaimer']}
-                {result['dermatologist_warning'] and "<br><br>🔴 <strong>Low confidence analysis</strong> - We recommend consulting a dermatologist for a professional evaluation."}
+                {'<br><br>🔴 <strong>Low confidence analysis</strong> - We recommend consulting a dermatologist for a professional evaluation.' if result['dermatologist_warning'] else ''}
             </div>
         </div>
     </div>
@@ -851,7 +768,7 @@ with gr.Blocks(title="Glow AI - Skincare Assistant with Confidence Meter", theme
     if HF_TOKEN:
         gr.Markdown("✅ **Connected to Hugging Face** - Real AI models active")
     else:
-        gr.Markdown("⚠️ **Demo Mode** - Add HF_TOKEN to .env for real AI analysis")
+        gr.Markdown("⚠️ **Demo Mode** - Add HF_TOKEN to Secrets for real AI analysis")
     
     with gr.Tab("Skin Analysis with Confidence"):
         gr.Markdown("### Upload a photo or describe your skin concerns")