app.py

import os
import io
import logging
import json
import re
import traceback
from typing import Optional, List

import gradio as gr
from langchain_core.prompts import PromptTemplate
from langchain_groq import ChatGroq
from langchain_community.vectorstores import Chroma
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langdetect import detect, LangDetectException
from deep_translator import GoogleTranslator
import torch
from PIL import Image
import fitz
from transformers import BlipProcessor, BlipForConditionalGeneration

# ===== LOGGING =====
logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
)
logger = logging.getLogger(__name__)

# ===== CONFIG =====
GROQ_API_KEY = os.getenv("GROQ_API_KEY")
if not GROQ_API_KEY:
    raise ValueError("❌ Missing GROQ_API_KEY - set in Settings → Secrets")

DB_PATH = "./chroma_db"
DEVICE = "cpu"
os.makedirs(DB_PATH, exist_ok=True)

# ===== INITIALIZE COMPONENTS =====
logger.info("Initializing components...")

# Initialize BLIP for image captioning
try:
    logger.info("Loading BLIP image captioning model...")
    blip_processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
    blip_model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base").to(DEVICE)
    logger.info("✅ BLIP model loaded.")
except Exception as e:
    logger.warning(f"⚠️ BLIP model loading failed: {e}")
    blip_model = None
    blip_processor = None

try:
    embedding_model = HuggingFaceEmbeddings(
        model_name="sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2"
    )
    logger.info("✅ Embeddings model loaded.")
except Exception as e:
    logger.error(f"❌ Error loading embeddings: {e}")
    embedding_model = None

try:
    vectorstore = Chroma(
        persist_directory=DB_PATH,
        embedding_function=embedding_model
    )
    logger.info("✅ Vectorstore initialized.")
except Exception as e:
    logger.error(f"❌ Error initializing vectorstore: {e}")
    vectorstore = None

try:
    llm = ChatGroq(
        groq_api_key=GROQ_API_KEY,
        model="openai/gpt-oss-120b",
        temperature=0.2
    )
    logger.info("✅ LLM initialized.")
except Exception as e:
    logger.error(f"❌ Error initializing LLM: {e}")
    llm = None

# ===== FILE EXTRACTION =====
def extract_text_from_pdf(file_path: str) -> str:
    """Extract text from PDF files"""
    try:
        with open(file_path, "rb") as f:
            pdf = fitz.open(stream=f.read(), filetype="pdf")
        text = ""
        for page_num, page in enumerate(pdf):
            text += f"\n--- Page {page_num + 1} ---\n"
            text += page.get_text()
        logger.info(f"✅ PDF extracted: {len(text)} chars")
        return text
    except Exception as e:
        logger.error(f"❌ PDF extraction error: {e}")
        return ""

def extract_text_from_txt(file_path: str) -> str:
    """Extract text from TXT files"""
    try:
        with open(file_path, "r", encoding="utf-8", errors="ignore") as f:
            text = f.read()
        logger.info(f"✅ TXT extracted: {len(text)} chars")
        return text
    except Exception as e:
        logger.error(f"❌ TXT extraction error: {e}")
        return ""

def extract_text_from_image(file_path: str) -> str:
    """Extract description from image using BLIP"""
    try:
        if blip_model is None or blip_processor is None:
            logger.warning("BLIP model not available, using fallback")
            img = Image.open(file_path).convert("RGB")
            width, height = img.size
            return f"Image dimensions: {width}x{height} pixels"
        
        img = Image.open(file_path).convert("RGB")
        
        # Generate caption using BLIP
        inputs = blip_processor(images=img, return_tensors="pt").to(DEVICE)
        output_ids = blip_model.generate(**inputs, max_length=100)
        caption = blip_processor.decode(output_ids[0], skip_special_tokens=True)
        
        logger.info(f"✅ Image caption extracted: {caption}")
        return f"Image Description: {caption}"
    except Exception as e:
        logger.error(f"❌ Image extraction error: {e}")
        return ""

# ===== AGENTS =====

# 1. TRANSLATION AGENT
def translation_agent(text: str, target_lang: str = "en") -> str:
    """Translate text to target language"""
    try:
        if not text or len(text.strip()) < 3:
            return text
        
        if target_lang == "Auto Detect":
            return text
        
        detected_lang = detect(text)
        logger.info(f"Detected language: {detected_lang}, target: {target_lang}")
        
        if detected_lang == target_lang:
            return text
        
        translated = GoogleTranslator(
            source="auto",
            target=target_lang
        ).translate(text)
        logger.info(f"✅ Translation completed")
        return translated
    except Exception as e:
        logger.warning(f"⚠️ Translation failed: {e}")
        return text

# 2. RETRIEVAL AGENT
def retrieval_agent(query: str, k: int = 5) -> List[str]:
    """Retrieve relevant documents from vectorstore"""
    if vectorstore is None:
        logger.warning("Vectorstore is None")
        return []
    
    try:
        logger.info(f"Retrieving top {k} documents for: {query[:50]}")
        docs = vectorstore.similarity_search(query, k=k)
        
        if not docs:
            logger.warning("No documents found")
            return []
        
        results = [d.page_content for d in docs]
        logger.info(f"✅ Retrieved {len(results)} documents")
        return results
    except Exception as e:
        logger.error(f"❌ Retrieval error: {e}")
        return []

# 3. SUMMARIZATION AGENT
def summarization_agent(text: str) -> str:
    """Summarize given text using LLM"""
    if llm is None:
        return "❌ LLM not available"
    
    try:
        if not text or len(text) < 50:
            return text
        
        prompt = f"""Summarize the following text concisely in 2-3 sentences:

{text[:2000]}

Summary:"""
        
        response = llm.invoke(prompt)
        summary = response.content if hasattr(response, "content") else str(response)
        logger.info(f"✅ Summarization completed")
        return summary
    except Exception as e:
        logger.error(f"❌ Summarization error: {e}")
        return f"[Summarization failed: {str(e)[:100]}]"

# 4. FACT-CHECK AGENT
def fact_check_agent(claim: str, context: str) -> str:
    """Fact-check a claim against retrieved context"""
    if llm is None:
        return "❌ LLM not available"
    
    try:
        prompt = f"""You are a fact-checker. Evaluate the following claim based on the context provided.

Context:
{context[:1500]}

Claim:
{claim}

Response format: [TRUE/FALSE/UNVERIFIED] - Brief explanation"""
        
        response = llm.invoke(prompt)
        result = response.content if hasattr(response, "content") else str(response)
        logger.info(f"✅ Fact-check completed")
        return result
    except Exception as e:
        logger.error(f"❌ Fact-check error: {e}")
        return f"[Fact-check failed: {str(e)[:100]}]"

# 5. QUESTION-ANSWERING AGENT
def qa_agent(question: str, context: str) -> str:
    """Answer question based on provided context"""
    if llm is None:
        return "❌ LLM not available"
    
    try:
        prompt = f"""You are a helpful assistant. Answer the following question based ONLY on the context provided.

Context:
{context[:3000]}

Question:
{question}

If the context doesn't contain the answer, say "I cannot find this information in the provided context."

Answer:"""
        
        response = llm.invoke(prompt)
        answer = response.content if hasattr(response, "content") else str(response)
        logger.info(f"✅ QA completed")
        return answer
    except Exception as e:
        logger.error(f"❌ QA error: {e}")
        return f"[Answer generation failed: {str(e)[:100]}]"

# ===== ORCHESTRATOR =====
def multi_agent_orchestrator(question: str, target_language: str) -> str:
    """Main orchestrator - coordinates all agents"""
    
    if not question or not question.strip():
        return "❌ Please enter a question."
    
    logger.info(f"Orchestrator started - Question: {question[:50]}")
    
    try:
        # Step 1: Translate question to English if needed
        logger.info("STEP 1: Translation Agent")
        english_question = translation_agent(question, "en")
        
        # Step 2: Retrieve relevant documents
        logger.info("STEP 2: Retrieval Agent")
        retrieved_docs = retrieval_agent(english_question, k=5)
        
        if not retrieved_docs:
            return "❌ No documents found in vectorstore. Please upload and process files first."
        
        context = "\n\n".join(retrieved_docs)
        logger.info(f"Context length: {len(context)} chars")
        
        # Step 3: Answer question using QA agent
        logger.info("STEP 3: QA Agent")
        answer = qa_agent(english_question, context)
        
        # Step 4: Translate answer to target language if needed
        if target_language and target_language != "Auto Detect" and target_language != "en":
            logger.info("STEP 4: Translation Agent (answer)")
            answer = translation_agent(answer, target_language)
        
        logger.info("✅ Orchestration completed successfully")
        return answer
        
    except Exception as e:
        logger.error(f"❌ Orchestration error: {e}\n{traceback.format_exc()}")
        return f"❌ Error: {str(e)[:300]}"

# ===== FILE INGESTION =====
def ingest_files(files) -> str:
    """Ingest uploaded files into vectorstore"""
    
    if not files:
        return "⚠️ No files uploaded"
    
    if vectorstore is None or embedding_model is None:
        return "❌ Vectorstore not initialized"
    
    logger.info(f"Starting file ingestion for {len(files)} file(s)")
    
    total_chunks = 0
    error_messages = []
    
    try:
        text_splitter = RecursiveCharacterTextSplitter(
            chunk_size=1000,
            chunk_overlap=200
        )
        
        for file_obj in files:
            try:
                # Handle both string paths and file objects
                if isinstance(file_obj, str):
                    file_path = file_obj
                    filename = os.path.basename(file_path).lower()
                else:
                    file_path = file_obj.name if hasattr(file_obj, 'name') else str(file_obj)
                    filename = os.path.basename(file_path).lower()
                
                logger.info(f"Processing: {filename} from {file_path}")
                
                text = ""
                
                # Extract based on file type
                if filename.endswith(".txt"):
                    text = extract_text_from_txt(file_path)
                elif filename.endswith(".pdf"):
                    text = extract_text_from_pdf(file_path)
                elif filename.endswith((".jpg", ".jpeg", ".png")):
                    text = extract_text_from_image(file_path)
                else:
                    error_messages.append(f"⚠️ Unsupported format: {filename}")
                    continue
                
                # Check if text was extracted
                if not text or len(text.strip()) < 20:
                    error_messages.append(f"⚠️ No content extracted from: {filename}")
                    logger.warning(f"No content from {filename}")
                    continue
                
                # Split and add to vectorstore
                chunks = text_splitter.split_text(text)
                logger.info(f"Created {len(chunks)} chunks from {filename}")
                
                vectorstore.add_texts(chunks)
                total_chunks += len(chunks)
                logger.info(f"✅ Added {len(chunks)} chunks from {filename}")
                
            except Exception as e:
                error_msg = f"Error processing {filename}: {str(e)[:100]}"
                error_messages.append(error_msg)
                logger.error(error_msg)
                continue
        
        # Persist vectorstore
        if total_chunks > 0:
            vectorstore.persist()
            logger.info(f"Vectorstore persisted with {total_chunks} chunks")
        
        # Build response message
        message = f"✅ Successfully ingested {total_chunks} chunks from {len([f for f in files if f])} file(s)"
        
        if error_messages:
            message += "\n\n" + "\n".join(error_messages)
        
        logger.info(message)
        return message
        
    except Exception as e:
        error_msg = f"❌ Ingestion error: {str(e)[:200]}"
        logger.error(f"{error_msg}\n{traceback.format_exc()}")
        return error_msg

# ===== GRADIO INTERFACE =====
suggested_questions = [
    "What is the main idea of the text?",
    "Summarize the content.",
    "What are the key points?",
    "¿Cuál es la idea principal?",
    "Quel est le résumé?",
    "Was sind die wichtigsten Punkte?",
]

supported_languages = [
    "Auto Detect", "en", "es", "fr", "de", "it", "pt", "ru", "zh", "ja", "hi", "bn", "ar"
]

with gr.Blocks(title="Nexus RAG- MultiAgent MultiLingual RAG System") as interface:
    gr.Markdown("# 🧠 Nexus RAG- MultiAgent MultiLingual RAG System")
    gr.Markdown("**Upload documents** → **Auto-ingest** → **Ask questions** in any language")
    gr.Markdown("---")
    
    with gr.Row():
        with gr.Column(scale=2):
            gr.Markdown("### 📁 Document Upload")
            file_input = gr.File(
                label="Upload Files (TXT/PDF/Images)",
                file_count="multiple",
                file_types=[".txt", ".pdf", ".jpg", ".jpeg", ".png"]
            )
            ingest_output = gr.Textbox(
                label="Ingestion Status",
                interactive=False,
                lines=4
            )
        
        with gr.Column(scale=3):
            gr.Markdown("### ❓ Ask Your Question")
            question = gr.Textbox(
                label="Your Question",
                placeholder="Ask anything about your documents...",
                lines=3
            )
            language_dropdown = gr.Dropdown(
                choices=supported_languages,
                value="Auto Detect",
                label="Output Language"
            )
            submit_button = gr.Button("💬 Ask Agent", variant="primary", size="lg")
            output_answer = gr.Markdown(label="📝 Answer")
            
            gr.Markdown("### 💡 Example Questions:")
            gr.Examples(suggested_questions, inputs=question)
    
    # Auto-ingest files when uploaded
    file_input.change(
        fn=ingest_files,
        inputs=file_input,
        outputs=ingest_output
    )
    
    # Submit question
    submit_button.click(
        fn=multi_agent_orchestrator,
        inputs=[question, language_dropdown],
        outputs=output_answer
    )

if __name__ == "__main__":
    logger.info("🚀 Launching Gradio interface...")
    interface.launch()