run_chatbot_chat.py

import os
import json
from chatbot_model import ChatbotConfig, RetrievalChatbot
from response_quality_checker import ResponseQualityChecker
from environment_setup import EnvironmentSetup
from logger_config import config_logger

logger = config_logger(__name__)
            
def run_chatbot_chat():
    env = EnvironmentSetup()
    env.initialize()
    
    MODEL_DIR = "models"
    FAISS_INDICES_DIR = os.path.join(MODEL_DIR, "faiss_indices")
    FAISS_INDEX_PRODUCTION_PATH = os.path.join(FAISS_INDICES_DIR, "faiss_index_production.index")
    FAISS_INDEX_TEST_PATH = os.path.join(FAISS_INDICES_DIR, "faiss_index_test.index")
    
    # Toggle 'production' or 'test' env
    ENVIRONMENT = "production"
    if ENVIRONMENT == "test":
        FAISS_INDEX_PATH = FAISS_INDEX_TEST_PATH
        RESPONSE_POOL_PATH = FAISS_INDEX_TEST_PATH.replace(".index", "_responses.json")
    else:
        FAISS_INDEX_PATH = FAISS_INDEX_PRODUCTION_PATH
        RESPONSE_POOL_PATH = FAISS_INDEX_PRODUCTION_PATH.replace(".index", "_responses.json")
    
    # Load the config
    config_path = os.path.join(MODEL_DIR, "config.json")
    if os.path.exists(config_path):
        with open(config_path, "r", encoding="utf-8") as f:
            config_dict = json.load(f)
        config = ChatbotConfig.from_dict(config_dict)
        logger.info(f"Loaded ChatbotConfig from {config_path}")
    else:
        config = ChatbotConfig()
        logger.warning("No config.json found. Using default ChatbotConfig.")
        
    # Load RetrievalChatbot in 'inference' mode
    try:
        chatbot = RetrievalChatbot.load_model(load_dir=MODEL_DIR, mode="inference")
    except Exception as e:
        logger.error(f"Failed to load RetrievalChatbot: {e}")
        return
    
    # Confirm FAISS index & response pool exist
    if not os.path.exists(FAISS_INDEX_PATH) or not os.path.exists(RESPONSE_POOL_PATH):
        logger.error("FAISS index or response pool file is missing.")
        return
    
    # Load FAISS index and response pool
    try:
        chatbot.data_pipeline.load_faiss_index(FAISS_INDEX_PATH)
        
        with open(RESPONSE_POOL_PATH, "r", encoding="utf-8") as f:
            chatbot.data_pipeline.response_pool = json.load(f)
        logger.info(f"FAISS index loaded from {FAISS_INDEX_PATH}.")    
        # Validate dimension consistency
        chatbot.data_pipeline.validate_faiss_index()
        
    except Exception as e:
        logger.error(f"Failed to load or validate FAISS index: {e}")
        return
    
    # Init QualityChecker and Validator
    quality_checker = ResponseQualityChecker(data_pipeline=chatbot.data_pipeline)
    
    # Run interactive chat loop
    logger.info("\nStarting interactive chat session...")
    chatbot.run_interactive_chat(quality_checker)

if __name__ == "__main__":
    run_chatbot_chat()