app.py

import gradio as gr
import logging
import os
import numpy as np
from sentence_transformers import SentenceTransformer
from langchain.text_splitter import RecursiveCharacterTextSplitter
import faiss
from simple_salesforce import Salesforce
from dotenv import load_dotenv
import zipfile
from pathlib import Path

# Setup logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

# Load environment variables from .env file
load_dotenv()  # Load the .env file

# Get the Salesforce credentials from environment variables
sf_username = os.getenv("SF_USERNAME")
sf_password = os.getenv("SF_PASSWORD")
sf_security_token = os.getenv("SF_SECURITY_TOKEN")
sf_instance_url = os.getenv("SF_INSTANCE_URL")

# Check if the environment variables are correctly set
if not sf_username or not sf_password or not sf_security_token or not sf_instance_url:
    logger.error("❌ Salesforce credentials are missing from environment variables!")
    raise ValueError("Salesforce credentials are not properly set.")

# Salesforce connection
try:
    sf = Salesforce(
        username=sf_username,
        password=sf_password,
        security_token=sf_security_token,
        instance_url=sf_instance_url
    )
    logger.info("✅ Connected to Salesforce")
except Exception as e:
    logger.error(f"❌ Salesforce connection failed: {str(e)}")
    raise

# --- Extract zip files and read documents ---
def extract_zip(zip_path, extract_to):
    try:
        with zipfile.ZipFile(zip_path, 'r') as zip_ref:
            zip_ref.extractall(extract_to)
        logger.info(f"Extracted {zip_path} to {extract_to}")
    except Exception as e:
        logger.error(f"Failed to extract {zip_path}: {str(e)}")
        raise

def load_documents(folder_path):
    documents = []
    sources = []
    for file in Path(folder_path).rglob("*.txt"):
        text = file.read_text(encoding="utf-8", errors="ignore")
        documents.append(text)
        sources.append(file.name)
    return documents, sources

# --- Chunking ---
text_splitter = RecursiveCharacterTextSplitter(chunk_size=300, chunk_overlap=50)

# --- Load model ---
model = SentenceTransformer("all-MiniLM-L6-v2")

# --- Preprocessing ---
data_dir = Path("./data")
data_dir.mkdir(exist_ok=True)

doc_folders = [
    ("Company_Policies.zip", "Company_Policies"),
    ("HR_Policies.zip", "Hr_Policies"),
    ("Contract_Clauses.zip", "Contract_Clauses")
]

all_chunks = []
metadata = []

for zip_name, folder in doc_folders:
    zip_path = Path(zip_name)
    if not zip_path.exists():
        logger.error(f"Zip file {zip_name} not found")
        raise FileNotFoundError(f"Zip file {zip_name} not found")
    extract_path = data_dir / folder
    extract_path.mkdir(exist_ok=True)
    extract_zip(zip_path, extract_path)
    docs, sources = load_documents(extract_path)
    if not docs:
        logger.error(f"No documents found in {extract_path}")
        raise ValueError(f"No documents found in {extract_path}")
    for doc, src in zip(docs, sources):
        chunks = text_splitter.split_text(doc)
        all_chunks.extend(chunks)
        src_url = f"https://company.com/{folder}/{src}"
        metadata.extend([src_url] * len(chunks))

# --- Embeddings + FAISS index ---
embeddings = model.encode(all_chunks)
index = faiss.IndexFlatL2(embeddings.shape[1])
index.add(np.array(embeddings))
logger.info("FAISS index built successfully")

# --- Create Record in Salesforce ---
def create_salesforce_record(query, answer, confidence_percentage, source_link):
    try:
        # Convert the confidence_percentage to Python float (to avoid numpy float32)
        confidence_percentage = float(confidence_percentage)

        # Data with correctly mapped field names
        data = {
            "Query__c": query,  # Field for User Query
            "Answer__c": answer,  # Field for Answer
            "Confidence_Percentage__c": confidence_percentage,  # Field for Confidence Score
            "Document_link__c": source_link,  # Field for Document Link
        }

        # Creating the record in Salesforce
        response = sf.chat_query_log__c.create(data)
        
        # Check if record was created successfully
        if 'id' in response:  # If the response contains an 'id', the record is created successfully
            record_id = response['id']
            logger.info(f"✅ Record created successfully in Salesforce with ID: {record_id}")
            return record_id  # Return the Salesforce record ID
        else:
            # Log the failure response
            logger.error(f"❌ Failed to create Salesforce record. Response: {response}")
            return None
    except Exception as e:
        # Log any error during record creation
        logger.error(f"Error creating Salesforce record: {str(e)}")
        return None

# --- Search & Answer ---
def answer_query(query):
    try:
        logger.info(f"Processing query: {query}")
        query_embedding = model.encode([query])
        D, I = index.search(np.array(query_embedding), k=3)
        top_chunks = [all_chunks[i] for i in I[0]]
        top_sources = [metadata[i] for i in I[0]]
        distances = D[0]

        relevant_chunks = [
            chunk for chunk, dist in zip(top_chunks, distances) if dist < 0.8
        ]
        relevant_sources = [
            src for src, dist in zip(top_sources, distances) if dist < 0.8
        ]

        if not relevant_chunks:
            return "No relevant information found.", "Confidence: 0%", "Source Link: None"

        answer = relevant_chunks[0].strip()
        min_distance = min(distances)
        confidence_percentage = max(0, 100 - (min_distance * 100))
        source_link = relevant_sources[0] if relevant_sources else "None"

        # Create Salesforce record for the query response
        record_id = create_salesforce_record(query, answer, confidence_percentage, source_link)

        if record_id:
            return (
                answer,
                f"Confidence: {confidence_percentage:.2f}%",
                f"Source Link: {source_link}",
                f"Salesforce Record ID: {record_id}"  # Display the Salesforce record ID
            )
        else:
            return (
                answer,
                f"Confidence: {confidence_percentage:.2f}%",
                f"Source Link: {source_link}",
                "Failed to create record in Salesforce"
            )
    except Exception as e:
        logger.error(f"Error in answer_query: {str(e)}")
        return f"Error: {str(e)}", "", "", ""

# --- Gradio Chatbot UI Design ---
def process_question(q, chat_history):
    if not q.strip():
        return chat_history + [("User", "Please enter a question.")], "", ""
    
    answer, confidence, source, record_id = answer_query(q)
    chat_history.append(("User", q))
    chat_history.append(("Bot", answer))
    
    return chat_history, confidence, source, record_id

# --- Chatbot UI with dynamic styling using elem_id ---
with gr.Blocks(title="Company Documents Q&A Chatbot", theme=gr.themes.Soft()) as demo:
    gr.Markdown("## 📚 **Company Policies Q&A Chatbot**")

    with gr.Row():
        with gr.Column(scale=3):
            question = gr.Textbox(
                label="Ask a Question",
                placeholder="What are the conditions for permanent employment status?",
                lines=1,
                interactive=True,
                elem_id="user-question",
                visible=True
            )
        with gr.Column(scale=1):
            submit_btn = gr.Button("Submit", variant="primary", elem_id="submit-btn")

    with gr.Row():
        with gr.Column():
            chat_history = gr.Chatbot(
                label="Chat History", 
                elem_id="chatbox",
                height=400,  # Set a fixed height
                show_label=False  # Hide the label to make the chat more clean
            )
            conf_out = gr.Markdown(label="Confidence", elem_id="confidence")
            source_out = gr.Markdown(label="Source Link", elem_id="source-link")
            record_out = gr.Markdown(label="Salesforce Record ID", elem_id="salesforce-id")

    submit_btn.click(fn=process_question, inputs=[question, chat_history], outputs=[chat_history, conf_out, source_out, record_out])

demo.launch(server_name="0.0.0.0", server_port=7860, share=True)