Update app.py

app.py

@@ -6,15 +6,15 @@ import os
 # Direct API key placement
 API_KEY = 'AIzaSyBMZrhMXRpQKp7M-JcN2Qk73afeta5Mv5Y'
 
-def extract_exact_text(image):
+def extract_malayalam_text(image):
     """
-    Extract text exactly as it appears in the uploaded image
+    Extract text specifically in Malayalam from the image
     
     Args:
         image (PIL.Image): Uploaded image
     
     Returns:
-        str: Extracted text preserving original format
+        str: Extracted Malayalam text
     """
     # Validate input
     if image is None:
@@ -27,22 +27,22 @@ def extract_exact_text(image):
         # Use Gemini 2.0 Flash model
         model = genai.GenerativeModel('gemini-2.0-flash')
         
-        # Prompt to extract text exactly as in the image
+        # Highly specific prompt for Malayalam text extraction
         response = model.generate_content(
             [
-                "CRITICAL INSTRUCTIONS: "
-                "Extract the text EXACTLY as it appears in the image. "
-                "Preserve: "
-                "1. Exact text content "
-                "2. Original formatting "
-                "3. Line breaks "
-                "4. Spacing "
-                "5. Case sensitivity "
-                "Do not modify or interpret the text in any way.", 
+                "CRITICAL INSTRUCTIONS FOR MALAYALAM TEXT EXTRACTION: "
+                "1. This image contains text in Malayalam script ONLY. "
+                "2. Extract ONLY Malayalam characters and script. "
+                "3. Use pure Malayalam Unicode characters. "
+                "4. If the text is in another language, state 'No Malayalam text found'. "
+                "5. Preserve exact Malayalam script, including any special characters. "
+                "6. DO NOT translate or convert to any other script. "
+                "7. If there are multiple lines, preserve line breaks. "
+                "8. Focus EXCLUSIVELY on Malayalam script.", 
                 image
             ],
             generation_config=genai.types.GenerationConfig(
-                temperature=0.1,  # Lowest temperature for precise extraction
+                temperature=0.1,  # Ultra-low temperature for precise extraction
                 max_output_tokens=500  # Increased to capture more text
             )
         )
@@ -50,9 +50,12 @@ def extract_exact_text(image):
         # Extract text from response
         extracted_text = response.text
         
-        # Validate text extraction
-        if not extracted_text or extracted_text.strip() == "":
-            return "No text detected. Please check the image quality."
+        # Validate Malayalam characters
+        malayalam_chars = [char for char in extracted_text if '\u0D00' <= char <= '\u0D7F']
+        
+        # Check if extracted text contains Malayalam characters
+        if not malayalam_chars:
+            return "No Malayalam text detected. Please verify the image contains Malayalam script."
         
         return extracted_text
     
@@ -61,11 +64,11 @@ def extract_exact_text(image):
 
 # Create Gradio Interface
 demo = gr.Interface(
-    fn=extract_exact_text,
-    inputs=gr.Image(type="pil", label="Upload Image with Text"),
-    outputs=gr.Textbox(label="Extracted Text", lines=10),
-    title="Exact Text Extractor",
-    description="Upload an image to extract text exactly as it appears."
+    fn=extract_malayalam_text,
+    inputs=gr.Image(type="pil", label="Upload Malayalam Text Image"),
+    outputs=gr.Textbox(label="Extracted Malayalam Text", lines=10),
+    title="Malayalam Text Extractor",
+    description="Upload an image containing Malayalam text for precise extraction."
 )
 
 # Launch the app