document_status_logger.py

"""
Enhanced startup logging to show actual document usage
"""

import logging
from lancedb_service import lancedb_service

logger = logging.getLogger("app")

async def log_document_status():
    """Log comprehensive document status during startup"""
    try:
        logger.info("📊 Document Database Status Check...")
        
        total_documents = 0
        
        # Check rajasthan_documents table (main voice bot documents)
        if "rajasthan_documents" in lancedb_service.db.table_names():
            raj_table = lancedb_service.db.open_table("rajasthan_documents")
            raj_count = raj_table.count_rows()
            total_documents += raj_count
            logger.info(f"🏛️ Rajasthan Documents: {raj_count:,} (Voice Bot Primary Source)")
        
        # Check general documents table
        if "documents" in lancedb_service.db.table_names():
            doc_table = lancedb_service.db.open_table("documents")
            doc_count = doc_table.count_rows()
            total_documents += doc_count
            logger.info(f"📚 General Documents: {doc_count:,}")
        
        # Summary
        logger.info(f"🎯 TOTAL AVAILABLE DOCUMENTS: {total_documents:,}")
        
        if total_documents >= 1000:
            logger.info("✅ Voice Bot has EXCELLENT document coverage (1000+ docs)")
        elif total_documents >= 100:
            logger.info("⚠️ Voice Bot has GOOD document coverage (100+ docs)")
        elif total_documents >= 10:
            logger.info("⚠️ Voice Bot has LIMITED document coverage (<100 docs)")
        else:
            logger.info("❌ Voice Bot has MINIMAL document coverage")
            
        # Show sample document topics if available
        if "rajasthan_documents" in lancedb_service.db.table_names():
            raj_table = lancedb_service.db.open_table("rajasthan_documents")
            if raj_table.count_rows() > 0:
                sample = raj_table.head(3).to_pylist()
                logger.info("📄 Sample document topics available:")
                for i, doc in enumerate(sample, 1):
                    filename = doc.get('filename', 'Unknown')
                    content_preview = doc.get('content', '')[:100] + "..."
                    logger.info(f"   {i}. {filename}: {content_preview}")
        
        return total_documents
        
    except Exception as e:
        logger.error(f"❌ Error checking document status: {e}")
        return 0