app_gradio.py

#!/usr/bin/env python3
"""
Textilindo AI Assistant - Hugging Face Spaces (Gradio Version)
"""

import gradio as gr
import os
import json
import requests
from difflib import SequenceMatcher
import logging

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

def load_system_prompt(default_text):
    """Load system prompt from configs/system_prompt.md if available"""
    try:
        base_dir = os.path.dirname(__file__)
        prompt_path = os.path.join(base_dir, "configs", "system_prompt.md")
        if os.path.exists(prompt_path):
            with open(prompt_path, 'r', encoding='utf-8') as f:
                return f.read().strip()
    except Exception as e:
        logger.warning(f"Could not load system prompt: {e}")
    return default_text

class TextilindoAI:
    def __init__(self):
        self.dataset = []
        self.system_prompt = load_system_prompt(
            "You are a helpful AI assistant for Textilindo, a textile company. "
            "Provide accurate and helpful information about Textilindo's products, services, and business information."
        )
        self.load_all_datasets()
        logger.info(f"Total examples loaded: {len(self.dataset)}")

    def load_all_datasets(self):
        """Load all JSONL datasets from the data directory"""
        base_dir = os.path.dirname(__file__)
        data_dir = os.path.join(base_dir, "data")
        
        if not os.path.exists(data_dir):
            logger.warning(f"Data directory not found: {data_dir}")
            return
        
        logger.info(f"Found data directory: {data_dir}")
        
        # Load all JSONL files
        for filename in os.listdir(data_dir):
            if filename.endswith('.jsonl'):
                filepath = os.path.join(data_dir, filename)
                file_examples = 0
                try:
                    with open(filepath, 'r', encoding='utf-8') as f:
                        for line in f:
                            line = line.strip()
                            if line:
                                try:
                                    data = json.loads(line)
                                    data['source'] = filename  # Add source tracking
                                    self.dataset.append(data)
                                    file_examples += 1
                                except json.JSONDecodeError as e:
                                    logger.warning(f"Invalid JSON in {filename}: {e}")
                                    continue
                    
                    logger.info(f"Loaded {filename}: {file_examples} examples")
                except Exception as e:
                    logger.error(f"Error loading {filename}: {e}")

    def find_most_similar(self, query, top_k=3):
        """Find most similar examples using sequence matching"""
        similarities = []
        
        for example in self.dataset:
            # Try different field names that might exist in the dataset
            question = example.get('question', example.get('instruction', '')).lower()
            similarity = SequenceMatcher(None, query.lower(), question).ratio()
            similarities.append((similarity, example))
        
        # Sort by similarity and return top_k
        similarities.sort(key=lambda x: x[0], reverse=True)
        return [example for _, example in similarities[:top_k]]

    def chat(self, message, temperature=0.7, max_tokens=300):
        """Generate AI response using RAG"""
        try:
            # Find similar examples
            similar_examples = self.find_most_similar(message, top_k=3)
            
            # Build context from similar examples
            context = ""
            for example in similar_examples:
                # Try different field names that might exist in the dataset
                question = example.get('question', example.get('instruction', ''))
                answer = example.get('answer', example.get('output', ''))
                context += f"Q: {question}\nA: {answer}\n\n"
            
            # Create prompt
            prompt = f"""System: {self.system_prompt}

Context from Textilindo knowledge base:
{context}

User Question: {message}

Please provide a helpful response based on the context above. If the context doesn't contain relevant information, provide a general helpful response about Textilindo."""

            # For now, return a simple response based on context
            if similar_examples:
                # Return the most similar answer
                best_answer = similar_examples[0].get('answer', similar_examples[0].get('output', ''))
                if best_answer:
                    return f"Based on our knowledge base: {best_answer}"
                else:
                    return "I found some relevant information but couldn't extract a proper answer. Please try rephrasing your question."
            else:
                return "I'm sorry, I don't have specific information about that. Please contact Textilindo directly for more details."
                
        except Exception as e:
            logger.error(f"Error in chat: {e}")
            return f"Error: {str(e)}"

# Initialize AI assistant (will be created when needed)
ai = None

def get_ai_assistant():
    """Get or create the AI assistant instance"""
    global ai
    if ai is None:
        try:
            logger.info("Initializing Textilindo AI Assistant...")
            ai = TextilindoAI()
            logger.info("AI Assistant initialized successfully")
        except Exception as e:
            logger.error(f"Failed to initialize AI Assistant: {e}")
            # Create a minimal fallback
            ai = type('FallbackAI', (), {
                'dataset': [],
                'chat': lambda self, message, **kwargs: f"AI Assistant is not available. Error: {str(e)}"
            })()
    return ai

def chat_function(message, temperature=0.7, max_tokens=300):
    """Chat function for Gradio interface"""
    try:
        if not message:
            return "Please enter a message."
        
        # Get AI assistant (initializes if needed)
        ai_assistant = get_ai_assistant()
        
        # Get AI response
        response = ai_assistant.chat(message, temperature=temperature, max_tokens=max_tokens)
        return response
        
    except Exception as e:
        logger.error(f"Error in chat function: {str(e)}")
        return f"Error: {str(e)}"

# Create Gradio interface
def create_interface():
    """Create the Gradio interface"""
    with gr.Blocks(title="Textilindo AI Assistant") as interface:
        gr.Markdown("# 🤖 Textilindo AI Assistant")
        gr.Markdown("AI-powered customer service for Textilindo")
        
        with gr.Row():
            with gr.Column():
                message_input = gr.Textbox(
                    label="Your Message",
                    placeholder="Ask me anything about Textilindo...",
                    lines=3
                )
                
                with gr.Row():
                    temperature = gr.Slider(
                        minimum=0.1,
                        maximum=2.0,
                        value=0.7,
                        step=0.1,
                        label="Temperature"
                    )
                    max_tokens = gr.Slider(
                        minimum=50,
                        maximum=1000,
                        value=300,
                        step=50,
                        label="Max Tokens"
                    )
                
                submit_btn = gr.Button("Send Message", variant="primary")
            
            with gr.Column():
                response_output = gr.Textbox(
                    label="AI Response",
                    lines=10,
                    interactive=False
                )
        
        # Event handlers
        submit_btn.click(
            fn=chat_function,
            inputs=[message_input, temperature, max_tokens],
            outputs=response_output
        )
        
        # Allow Enter key to submit
        message_input.submit(
            fn=chat_function,
            inputs=[message_input, temperature, max_tokens],
            outputs=response_output
        )
        
        # Add examples
        gr.Examples(
            examples=[
                "Dimana lokasi Textilindo?",
                "Apa saja produk yang dijual di Textilindo?",
                "Jam berapa Textilindo buka?",
                "Bagaimana cara menghubungi Textilindo?"
            ],
            inputs=message_input
        )
        
        # Add footer with stats
        try:
            ai_assistant = get_ai_assistant()
            dataset_size = len(ai_assistant.dataset) if hasattr(ai_assistant, 'dataset') else 0
            gr.Markdown(f"**Dataset loaded:** {dataset_size} examples")
        except:
            gr.Markdown("**Dataset:** Loading...")
    
    return interface

if __name__ == '__main__':
    try:
        logger.info("Starting Textilindo AI Assistant...")
        
        # Try to initialize AI assistant early to catch any issues
        try:
            ai_assistant = get_ai_assistant()
            logger.info(f"Dataset loaded: {len(ai_assistant.dataset)} examples")
        except Exception as e:
            logger.warning(f"AI Assistant initialization failed: {e}")
            logger.info("Continuing with fallback mode...")
        
        # Create and launch the interface
        logger.info("Creating Gradio interface...")
        interface = create_interface()
        logger.info("Gradio interface created successfully")
        
        # Get server configuration from environment variables
        server_name = os.environ.get("GRADIO_SERVER_NAME", "0.0.0.0")
        server_port = int(os.environ.get("GRADIO_SERVER_PORT", "7860"))
        
        logger.info(f"Launching Gradio interface on {server_name}:{server_port}")
        
        # Launch with basic configuration for Hugging Face Spaces
        interface.launch(
            server_name=server_name,
            server_port=server_port,
            share=False,  # Hugging Face Spaces handles tunneling automatically
            show_error=True,
            quiet=False,
            inbrowser=False,
            prevent_thread_lock=False  # Essential for keeping the process alive
        )
        
        logger.info("Gradio interface launched successfully")
        
        # Keep the application running
        try:
            import time
            while True:
                time.sleep(1)
        except KeyboardInterrupt:
            logger.info("Application stopped by user")
            
    except Exception as e:
        logger.error(f"Failed to start application: {e}")
        # Create a simple fallback interface
        try:
            with gr.Blocks() as fallback:
                gr.Markdown("# Textilindo AI Assistant")
                gr.Markdown("Application is starting... Please wait.")
                gr.Markdown(f"Error: {str(e)}")
            
            # Get server configuration from environment variables
            server_name = os.environ.get("GRADIO_SERVER_NAME", "0.0.0.0")
            server_port = int(os.environ.get("GRADIO_SERVER_PORT", "7860"))
            
            fallback.launch(
                server_name=server_name,
                server_port=server_port,
                share=False,
                inbrowser=False,
                prevent_thread_lock=False  # Essential for keeping the process alive
            )
            
            # Keep the fallback running
            import time
            while True:
                time.sleep(1)
        except Exception as fallback_error:
            logger.error(f"Fallback interface also failed: {fallback_error}")
            # If everything fails, just keep the process alive
            import time
            while True:
                time.sleep(1)