app.py

import gradio as gr
import numpy as np
import google.generativeai as genai
import faiss
from sentence_transformers import SentenceTransformer
from datasets import load_dataset
import warnings

# Suppress warnings
warnings.filterwarnings("ignore")

# Configuration - PUT YOUR API KEY HERE
GEMINI_API_KEY = "AIzaSyASrFvE3gFPigihza0JTuALzZmBx0Kc3d0"  # ⚠️ REPLACE WITH YOUR KEY
MODEL_NAME = "all-MiniLM-L6-v2"
GENAI_MODEL = "gemini-pro"
DATASET_NAME = "midrees2806/7K_Dataset"
CHUNK_SIZE = 500
TOP_K = 3

# Initialize Gemini with enhanced configuration
genai.configure(
    api_key=GEMINI_API_KEY,
    transport='rest',  # Force REST API
    client_options={
        'api_endpoint': "https://generativelanguage.googleapis.com/v1beta/models/gemini-pro:generateContent"
    }
)

class GeminiRAGSystem:
    def __init__(self):
        self.index = None
        self.chunks = []
        self.dataset_loaded = False
        self.loading_error = None
        
        print("Initializing embedding model...")
        try:
            self.embedding_model = SentenceTransformer(MODEL_NAME)
            print("Embedding model initialized successfully")
        except Exception as e:
            error_msg = f"Failed to initialize embedding model: {str(e)}"
            print(error_msg)
            raise RuntimeError(error_msg)
        
        print("Loading dataset...")
        self.load_dataset()
    
    def load_dataset(self):
        """Load dataset with detailed error handling"""
        try:
            print(f"Downloading dataset: {DATASET_NAME}")
            dataset = load_dataset(
                DATASET_NAME,
                split='train',
                download_mode="force_redownload"
            )
            print("Dataset downloaded successfully")
            
            if 'text' in dataset.features:
                self.chunks = dataset['text'][:1000]
                print(f"Loaded {len(self.chunks)} text chunks")
            elif 'context' in dataset.features:
                self.chunks = dataset['context'][:1000]
                print(f"Loaded {len(self.chunks)} context chunks")
            else:
                raise ValueError("Dataset must have 'text' or 'context' field")
            
            print("Creating embeddings...")
            embeddings = self.embedding_model.encode(
                self.chunks,
                show_progress_bar=False,
                convert_to_numpy=True
            )
            print(f"Created embeddings with shape {embeddings.shape}")
            
            self.index = faiss.IndexFlatL2(embeddings.shape[1])
            self.index.add(embeddings.astype('float32'))
            print("FAISS index created successfully")
            
            self.dataset_loaded = True
            print("Dataset loading complete")
        except Exception as e:
            error_msg = f"Dataset loading failed: {str(e)}"
            print(error_msg)
            self.loading_error = error_msg
    
    def get_relevant_context(self, query: str) -> str:
        """Retrieve context with debugging"""
        if not self.index:
            print("No index available for search")
            return ""
            
        try:
            print(f"Processing query: {query}")
            query_embed = self.embedding_model.encode(
                [query],
                convert_to_numpy=True
            ).astype('float32')
            print("Query embedded successfully")
            
            distances, indices = self.index.search(query_embed, k=TOP_K)
            print(f"Search results - distances: {distances}, indices: {indices}")
            
            context = "\n\n".join([self.chunks[i] for i in indices[0] if i < len(self.chunks)])
            print(f"Context length: {len(context)} characters")
            return context
        except Exception as e:
            print(f"Search error: {str(e)}")
            return ""

    def generate_response(self, query: str) -> str:
        """Generate response with detailed error handling"""
        if not self.dataset_loaded:
            msg = f"⚠️ Dataset loading failed: {self.loading_error}" if self.loading_error else "⚠️ System initializing..."
            print(msg)
            return msg
        
        print(f"\n{'='*40}\nNew Query: {query}\n{'='*40}")
        
        context = self.get_relevant_context(query)
        if not context:
            print("No relevant context found")
            return "No relevant context found"
        
        prompt = f"""Answer based on this context:
        {context}
        
        Question: {query}
        Answer concisely:"""
        
        print(f"\nPrompt sent to Gemini:\n{prompt}\n")
        
        try:
            model = genai.GenerativeModel(GENAI_MODEL)
            response = model.generate_content(
                prompt,
                generation_config=genai.types.GenerationConfig(
                    temperature=0.3,
                    max_output_tokens=1000
                )
            )
            
            print(f"Raw API response: {response}")
            
            if response.candidates and response.candidates[0].content.parts:
                answer = response.candidates[0].content.parts[0].text
                print(f"Answer: {answer}")
                return answer
            print("⚠️ Empty response from API")
            return "⚠️ No response from API"
        except Exception as e:
            error_msg = f"⚠️ API Error: {str(e)}"
            print(error_msg)
            return error_msg

# Initialize system with verbose logging
print("Initializing RAG system...")
try:
    rag_system = GeminiRAGSystem()
    init_status = "✅ System ready" if rag_system.dataset_loaded else f"⚠️ Initializing... {rag_system.loading_error or ''}"
    print(init_status)
except Exception as e:
    init_status = f"❌ Initialization failed: {str(e)}"
    print(init_status)
    rag_system = None

# Create interface with enhanced debugging
with gr.Blocks(title="Document Chatbot") as app:
    gr.Markdown("# Document Chatbot with Gemini")
    
    with gr.Row():
        chatbot = gr.Chatbot(height=500, label="Chat History")
    
    with gr.Row():
        query = gr.Textbox(label="Your question", placeholder="Ask about the documents...")
    
    with gr.Row():
        submit_btn = gr.Button("Submit", variant="primary")
        clear_btn = gr.Button("Clear", variant="secondary")
    
    status = gr.Textbox(label="System Status", value=init_status, interactive=False)

    def respond(message, chat_history):
        print(f"\n{'='*40}\nUser Query: {message}\n{'='*40}")
        if not rag_system:
            error_msg = "System initialization failed"
            print(error_msg)
            return chat_history + [(message, error_msg)]
        
        response = rag_system.generate_response(message)
        return chat_history + [(message, response)]
    
    def clear_chat():
        print("Chat cleared")
        return []
    
    submit_btn.click(respond, [query, chatbot], [chatbot])
    query.submit(respond, [query, chatbot], [chatbot])
    clear_btn.click(clear_chat, outputs=chatbot)

if __name__ == "__main__":
    print("Launching Gradio interface...")
    app.launch(debug=True)