Create app.py

app.py

@@ -0,0 +1,187 @@
+import streamlit as st
+import base64
+import requests
+from PIL import Image, ImageDraw
+from io import BytesIO
+import fitz  # PyMuPDF
+import time
+import os
+
+# Configuration
+GEMINI_API_KEY = st.secrets["GEMINI_API_KEY"]
+GEMINI_MODEL = "gemini-1.5-flash"  # Verify with Gemini API documentation
+DOCUMENT_TYPES = ["Land Records", "Caste Certificates", "Property Registrations"]
+
+# Define a class to mimic Streamlit's UploadedFile object for the example image
+class ExampleFile:
+    def __init__(self, file_path):
+        self.name = os.path.basename(file_path)
+        self.type = "image/jpeg"
+        with open(file_path, "rb") as f:
+            self._buffer = BytesIO(f.read())
+    
+    def getvalue(self):
+        return self._buffer.getvalue()
+
+# Initialize session state
+def initialize_session_state():
+    if "processed_doc" not in st.session_state:
+        st.session_state.processed_doc = None
+    if "current_file" not in st.session_state:
+        st.session_state.current_file = None
+
+# Reset session state
+def reset_session_state():
+    st.session_state.processed_doc = None
+    st.session_state.current_file = None
+
+# Encode file to base64 for Gemini API
+def encode_file(file):
+    try:
+        file_content = file.getvalue()
+        return base64.b64encode(file_content).decode('utf-8')
+    except Exception as e:
+        st.error(f"Error encoding file: {str(e)}")
+        return None
+
+# Query Gemini API
+def query_gemini(prompt, image_b64):
+    try:
+        headers = {
+            "Authorization": f"Bearer {GEMINI_API_KEY}",
+            "Content-Type": "application/json"
+        }
+        payload = {
+            "model": GEMINI_MODEL,
+            "prompt": prompt,
+            "image": image_b64
+        }
+        response = requests.post(
+            "https://api.gemini.com/v1/analyze",  # Replace with actual Gemini API endpoint
+            headers=headers,
+            json=payload
+        )
+        response.raise_for_status()
+        return response.json().get("result", "")
+    except Exception as e:
+        st.error(f"Gemini API error: {str(e)}")
+        return None
+
+# Process the document
+def process_document(file):
+    try:
+        with st.spinner("Analyzing document..."):
+            # Encode file to base64
+            image_b64 = encode_file(file)
+            if not image_b64:
+                return
+
+            # Store preview image
+            if file.type == "application/pdf":
+                pdf = fitz.open(stream=BytesIO(file.getvalue()))
+                page = pdf[0]
+                pix = page.get_pixmap()
+                st.session_state.doc_preview = Image.frombytes("RGB", [pix.width, pix.height], pix.samples)
+            elif file.type.startswith('image/'):
+                st.session_state.doc_preview = Image.open(BytesIO(file.getvalue()))
+            elif file.type == "text/plain":
+                text = file.getvalue().decode('utf-8')
+                img = Image.new('RGB', (800, 600), color=(73, 109, 137))
+                d = ImageDraw.Draw(img)
+                d.text((10, 10), text, fill=(255, 255, 0))
+                st.session_state.doc_preview = img
+            else:
+                st.error("Unsupported file format")
+                return
+
+            # Classify document
+            classify_prompt = f"Classify this document into one of these categories: {', '.join(DOCUMENT_TYPES)}. Respond only with the category name."
+            doc_type = query_gemini(classify_prompt, image_b64)
+            
+            # Extract details
+            extract_prompt = """Extract and organize all important details from this document including:
+            - Names
+            - Dates
+            - Identification numbers
+            - Locations
+            - Key terms
+            Format as a bullet-point list with clear headings."""
+            details = query_gemini(extract_prompt, image_b64)
+
+            # Verify authenticity
+            verify_prompt = "Analyze this document for signs of tampering or forgery. Check for: inconsistent fonts, alignment issues, suspicious modifications. Provide verification conclusion."
+            verification = query_gemini(verify_prompt, image_b64)
+
+            st.session_state.processed_doc = {
+                "type": doc_type or "Unclassified",
+                "details": details or "No details extracted",
+                "verification": verification or "Verification failed",
+                "preview": st.session_state.doc_preview
+            }
+
+            st.success("Document processing complete!")
+            time.sleep(1)
+
+    except Exception as e:
+        st.error(f"Document processing failed: {str(e)}")
+        st.session_state.processed_doc = None
+
+# Main application
+def main():
+    st.set_page_config(page_title="DocVerify AI", layout="wide")
+    initialize_session_state()
+    
+    # Sidebar Controls
+    with st.sidebar:
+        st.header("Document Controls")
+        
+        uploaded_file = st.file_uploader(
+            "Upload Document",
+            type=["pdf", "jpg", "jpeg", "png", "txt"],
+            key="uploaded_file"
+        )
+        
+        if st.button("Use Example Image"):
+            example_file_path = "Caste-Certificate.jpg"
+            st.session_state.current_file = ExampleFile(example_file_path)
+        elif uploaded_file:
+            st.session_state.current_file = uploaded_file
+        else:
+            st.session_state.current_file = None
+        
+        if st.button("Process Document"):
+            if st.session_state.current_file:
+                process_document(st.session_state.current_file)
+            else:
+                st.error("Please select a document to process.")
+        
+        if st.button("New Document"):
+            reset_session_state()
+            st.rerun()
+        
+        if st.session_state.processed_doc:
+            st.divider()
+            st.subheader("Document Summary")
+            st.markdown(f"**Type:** {st.session_state.processed_doc['type']}")
+            st.markdown(f"**Verification Status:**\n{st.session_state.processed_doc['verification']}")
+
+    # Main Interface
+    st.title("📄 Automated Document Verifier")
+    
+    if st.session_state.processed_doc and 'preview' in st.session_state.processed_doc:
+        col1, col2 = st.columns([1, 2])
+        with col1:
+            st.subheader("Document Preview")
+            st.image(st.session_state.processed_doc['preview'], use_column_width=True)
+        
+        with col2:
+            st.subheader("Extracted Details")
+            st.markdown(st.session_state.processed_doc['details'])
+            
+            st.subheader("Verification Analysis")
+            st.markdown(st.session_state.processed_doc['verification'])
+    else:
+        st.info("Please select a document and click 'Process Document' to begin verification analysis.")
+
+if __name__ == "__main__":
+    main()