server/scripts/debug_index.py

import os
import sys
import asyncio

# Add server directory to path
sys.path.append(os.path.join(os.getcwd(), 'server'))

from app.services.medical_retriever import get_medical_retriever

async def check_index():
    retriever = get_medical_retriever()
    
    for i, meta in enumerate(retriever.doc_metadata):
        src = meta.get('source', 'Unknown')
        if src in ['PubMedQA', 'WHO/CDC']:
            print(f"Source: {src}")
            print(f"Index: {i}")
            print(f"Document Text: {retriever.documents[i]}")
            print(f"Metadata: {meta}")
            print("-" * 20)

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