check_agentic_docs.py

#!/usr/bin/env python3
"""
Check documents in the agentic_chatbot LanceDB database
"""
import os
import sys

# Add the path to access lancedb_service
sys.path.append('/Users/abhishekchoudhary/Abhi Project/Voice_Bot/PensionBot')

# Change to agentic_chatbot directory to use its database
os.chdir('/Users/abhishekchoudhary/Abhi Project/Voice_Bot/agentic_chatbot')

import asyncio
import logging
import lancedb

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

async def check_agentic_documents():
    """Check how many documents are in the agentic_chatbot database"""
    try:
        logger.info("🔍 Checking agentic_chatbot LanceDB document count...")
        
        # Connect to the local LanceDB
        db = lancedb.connect("lancedb_data")
        table_names = db.table_names()
        
        logger.info(f"📊 Found {len(table_names)} tables in agentic_chatbot database:")
        
        total_documents = 0
        for table_name in table_names:
            try:
                table = db.open_table(table_name)
                count = table.count_rows()
                logger.info(f"  📄 {table_name}: {count} documents")
                total_documents += count
            except Exception as e:
                logger.error(f"  ❌ Error checking table {table_name}: {e}")
        
        logger.info(f"📈 Total documents in agentic_chatbot: {total_documents}")
            
    except Exception as e:
        logger.error(f"❌ Error checking documents: {e}")

if __name__ == "__main__":
    asyncio.run(check_agentic_documents())