Update app.py

app.py

@@ -1,6 +1,99 @@
 # Install dependencies
 
+import gradio as gr
+import fitz
+import numpy as np
+import requests
+import faiss
+import re
+import json
+import pandas as pd
+from docx import Document
+from pptx import Presentation
+from sentence_transformers import SentenceTransformer
+from concurrent.futures import ThreadPoolExecutor
+
+# Configuration
+GEMINI_API_KEY = "AIzaSyAPF8eVHU2jRWrQfwD8J9HPz4DrfIWK4GQ"  # 🔑 REPLACE WITH YOUR GEMINI KEY
+MODEL_NAME = "all-MiniLM-L6-v2"
+CHUNK_SIZE = 1024
+MAX_TOKENS = 4096
+MODEL = SentenceTransformer(MODEL_NAME)
+WORKERS = 8
 
+class DocumentProcessor:
+    def __init__(self):
+        self.index = faiss.IndexFlatIP(MODEL.get_sentence_embedding_dimension())
+        self.chunks = []
+        self.processor_pool = ThreadPoolExecutor(max_workers=WORKERS)
+
+    # ... (keep all existing document processing methods unchanged) ...
+
+    def query(self, question):
+        if not self.chunks:
+            return "Please process documents first", False
+
+        try:
+            print("\n" + "="*40 + " QUERY PROCESSING " + "="*40)
+            print(f"Question: {question}")
+
+            # Generate embedding for the question
+            question_embedding = MODEL.encode([question], convert_to_tensor=True).cpu().numpy().astype('float32')
+
+            # Search FAISS index
+            _, indices = self.index.search(question_embedding, 3)
+            print(f"Top indices: {indices}")
+
+            # Get context from top chunks
+            context = "\n".join([self.chunks[i] for i in indices[0] if i < len(self.chunks)])
+            print(f"Context length: {len(context)} characters")
+
+            # Gemini API Call
+            url = f"https://generativelanguage.googleapis.com/v1beta/models/gemini-pro:generateContent?key={GEMINI_API_KEY}"
+            headers = {"Content-Type": "application/json"}
+
+            payload = {
+                "contents": [{
+                    "parts": [{
+                        "text": f"Answer concisely based on this context: {context}\n\nQuestion: {question}"
+                    }]
+                }],
+                "generationConfig": {
+                    "temperature": 0.3,
+                    "maxOutputTokens": MAX_TOKENS
+                }
+            }
+
+            response = requests.post(
+                url,
+                headers=headers,
+                json=payload,
+                timeout=20
+            )
+
+            print(f"API Status Code: {response.status_code}")
+
+            if response.status_code != 200:
+                return f"API Error: {response.text}", False
+
+            # Parse Gemini response
+            response_json = response.json()
+            try:
+                final_answer = response_json['candidates'][0]['content']['parts'][0]['text']
+            except (KeyError, IndexError) as e:
+                print(f"Response parsing error: {str(e)}")
+                return "Error: Could not parse API response", False
+
+            print(f"Final Answer: {final_answer}")
+            return final_answer, True
+
+        except Exception as e:
+            print(f"Query Error: {str(e)}")
+            return f"Error: {str(e)}", False
+
+# ... (keep the rest of the Gradio interface code unchanged) ...
+
+''' 
 import gradio as gr
 import fitz
 import numpy as np
@@ -234,4 +327,4 @@ with gr.Blocks(title="RAG System") as app:
     )
 
 app.launch(share=True, debug=True)
-
+'''