Update app.py

app.py

@@ -106,100 +106,155 @@ def verify_password(password, hashed):
 class ImageProcessor:
     """Handles image preprocessing for better OCR results"""
     
+    @staticmethod
+    def preprocess_receipt_image(image_path):
+        """
+        Preprocess receipt image for optimal OCR
+        Returns: processed image path and preprocessing info
+        """
+        try:
+            if not PIL_AVAILABLE:
+                return image_path, "No preprocessing - PIL not available"
+            
+            # Load image
+            image = Image.open(image_path)
+            
+            # Convert to RGB if needed
+            if image.mode != 'RGB':
+                image = image.convert('RGB')
+            
+            # Enhance contrast
+            enhancer = ImageEnhance.Contrast(image)
+            image = enhancer.enhance(1.5)
+            
+            # Enhance sharpness
+            enhancer = ImageEnhance.Sharpness(image)
+            image = enhancer.enhance(2.0)
+            
+            # Convert to grayscale
+            image = image.convert('L')
+            
+            # Apply Gaussian blur to reduce noise
+            image = image.filter(ImageFilter.GaussianBlur(radius=0.5))
+            
+            # Convert to numpy array for OpenCV processing
+            if 'cv2' in globals() and cv2 is not None:
+                img_array = np.array(image)
+                
+                # Apply threshold to get binary image
+                _, binary = cv2.threshold(img_array, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
+                
+                # Morphological operations to clean up the image
+                kernel = np.ones((1,1), np.uint8)
+                binary = cv2.morphologyEx(binary, cv2.MORPH_CLOSE, kernel)
+                
+                # Convert back to PIL Image
+                image = Image.fromarray(binary)
+            
+            # Save processed image
+            processed_path = image_path.replace('.', '_processed.')
+            image.save(processed_path)
+            
+            return processed_path, "Enhanced contrast, sharpness, applied thresholding"
+            
+        except Exception as e:
+            print(f"Image preprocessing error: {e}")
+            return image_path, f"Preprocessing failed: {str(e)}"
+    
     @staticmethod
     def process_receipt_image(image_file, phone):
-    """
-    Complete receipt processing pipeline that handles Gradio file objects
-    Returns: (success, status_message, extracted_data, image_preview)
-    """
-    try:
-        if not phone:
-            return False, "❌ Please sign in first", {}, None
-        
-        if not image_file:
-            return False, "❌ No image uploaded", {}, None
+        """
+        Complete receipt processing pipeline that handles Gradio file objects
+        Returns: (success, status_message, extracted_data, image_preview)
+        """
+        try:
+            if not phone:
+                return False, "❌ Please sign in first", {}, None
+            
+            if not image_file:
+                return False, "❌ No image uploaded", {}, None
 
-        # Debug input type
-        print(f"\n📁 Input type: {type(image_file)}")
-        
-        # Handle different input types
-        if isinstance(image_file, str):
-            # Case 1: Direct file path (local testing)
-            image_path = image_file
-        elif hasattr(image_file, 'name'):
-            # Case 2: Gradio NamedString object (Hugging Face Spaces)
-            image_path = image_file.name
-        else:
-            return False, "❌ Unsupported file input type", {}, None
+            # Debug input type
+            print(f"\n📁 Input type: {type(image_file)}")
+            
+            # Handle different input types
+            if isinstance(image_file, str):
+                # Case 1: Direct file path (local testing)
+                image_path = image_file
+            elif hasattr(image_file, 'name'):
+                # Case 2: Gradio NamedString object (Hugging Face Spaces)
+                image_path = image_file.name
+            else:
+                return False, "❌ Unsupported file input type", {}, None
 
-        # Create receipts directory if needed
-        os.makedirs(RECEIPTS_DIR, exist_ok=True)
-        
-        # Generate unique filename
-        timestamp = int(time.time())
-        filename = f"receipt_{phone}_{timestamp}{os.path.splitext(image_path)[1]}"
-        save_path = os.path.join(RECEIPTS_DIR, filename)
-        
-        # Copy the uploaded file (works for both Gradio and direct paths)
-        with open(image_path, 'rb') as src, open(save_path, 'wb') as dst:
-            dst.write(src.read())
-        
-        print(f"📄 Saved receipt to: {save_path}")
-        
-        # Preprocess image
-        processed_path, preprocessing_info = ImageProcessor.preprocess_receipt_image(save_path)
-        print(f"🖼️ Preprocessing: {preprocessing_info}")
-        
-        # Extract text using OCR
-        raw_text, confidence, extracted_data = ocr_service.extract_text_from_receipt(processed_path)
-        print(f"🔍 OCR Confidence: {confidence:.1%}")
-        
-        # Auto-categorize
-        if extracted_data.get('merchant'):
-            suggested_category = db.auto_categorize_receipt(
-                phone, 
-                extracted_data['merchant'], 
-                extracted_data.get('total_amount', 0)
-            )
-            extracted_data['suggested_category'] = suggested_category
-            print(f"🏷️ Suggested category: {suggested_category}")
-        
-        # Prepare receipt data for database
-        receipt_data = {
-            'image_path': save_path,
-            'processed_image_path': processed_path,
-            'merchant': extracted_data.get('merchant', ''),
-            'amount': extracted_data.get('total_amount', 0.0),
-            'date': extracted_data.get('date', ''),
-            'category': extracted_data.get('suggested_category', 'Miscellaneous'),
-            'confidence': confidence,
-            'raw_text': raw_text,
-            'extracted_data': extracted_data,
-            'is_validated': False
-        }
-        
-        # Save to database
-        receipt_id = db.save_receipt(phone, receipt_data)
-        extracted_data['receipt_id'] = receipt_id
-        print(f"💾 Saved to DB with ID: {receipt_id}")
-        
-        # Create image preview
-        try:
-            image_preview = Image.open(save_path)
-            image_preview.thumbnail((400, 600))  # Resize for display
+            # Create receipts directory if needed
+            os.makedirs(RECEIPTS_DIR, exist_ok=True)
+            
+            # Generate unique filename
+            timestamp = int(time.time())
+            filename = f"receipt_{phone}_{timestamp}{os.path.splitext(image_path)[1]}"
+            save_path = os.path.join(RECEIPTS_DIR, filename)
+            
+            # Copy the uploaded file (works for both Gradio and direct paths)
+            with open(image_path, 'rb') as src, open(save_path, 'wb') as dst:
+                dst.write(src.read())
+            
+            print(f"📄 Saved receipt to: {save_path}")
+            
+            # Preprocess image
+            processed_path, preprocessing_info = ImageProcessor.preprocess_receipt_image(save_path)
+            print(f"🖼️ Preprocessing: {preprocessing_info}")
+            
+            # Extract text using OCR
+            raw_text, confidence, extracted_data = ocr_service.extract_text_from_receipt(processed_path)
+            print(f"🔍 OCR Confidence: {confidence:.1%}")
+            
+            # Auto-categorize
+            if extracted_data.get('merchant'):
+                suggested_category = db.auto_categorize_receipt(
+                    phone, 
+                    extracted_data['merchant'], 
+                    extracted_data.get('total_amount', 0)
+                )
+                extracted_data['suggested_category'] = suggested_category
+                print(f"🏷️ Suggested category: {suggested_category}")
+            
+            # Prepare receipt data for database
+            receipt_data = {
+                'image_path': save_path,
+                'processed_image_path': processed_path,
+                'merchant': extracted_data.get('merchant', ''),
+                'amount': extracted_data.get('total_amount', 0.0),
+                'date': extracted_data.get('date', ''),
+                'category': extracted_data.get('suggested_category', 'Miscellaneous'),
+                'confidence': confidence,
+                'raw_text': raw_text,
+                'extracted_data': extracted_data,
+                'is_validated': False
+            }
+            
+            # Save to database
+            receipt_id = db.save_receipt(phone, receipt_data)
+            extracted_data['receipt_id'] = receipt_id
+            print(f"💾 Saved to DB with ID: {receipt_id}")
+            
+            # Create image preview
+            try:
+                image_preview = Image.open(save_path)
+                image_preview.thumbnail((400, 600))  # Resize for display
+            except Exception as e:
+                print(f"⚠️ Preview generation failed: {e}")
+                image_preview = None
+            
+            status_msg = f"✅ Receipt processed! Confidence: {confidence:.1%}"
+            if confidence < 0.7:
+                status_msg += " ⚠️ Low confidence - please verify"
+            
+            return True, status_msg, extracted_data, image_preview
+            
         except Exception as e:
-            print(f"⚠️ Preview generation failed: {e}")
-            image_preview = None
-        
-        status_msg = f"✅ Receipt processed! Confidence: {confidence:.1%}"
-        if confidence < 0.7:
-            status_msg += " ⚠️ Low confidence - please verify"
-        
-        return True, status_msg, extracted_data, image_preview
-        
-    except Exception as e:
-        print(f"❌ Processing error: {traceback.format_exc()}")
-        return False, f"❌ Processing failed: {str(e)}", {}, None
+            print(f"❌ Processing error: {traceback.format_exc()}")
+            return False, f"❌ Processing failed: {str(e)}", {}, None
     
     @staticmethod
     def extract_text_regions(image_path):