verify_truth_domains.py

import sys
import os
import asyncio
import numpy as np

# Add server directory to path
sys.path.append('/Users/mac/Projects/customerAgent/server')

from app.services.nlp_processor import NLPService

async def test_truth_domains():
    nlp = NLPService()
    
    test_cases = [
        {
            "text": "What are your opening hours? We are open from 9 AM to 9 PM daily.",
            "url": "https://example.com/faq",
            "title": "Frequently Asked Questions",
            "expected_truth": "high",
            "expected_yes_no": True
        },
        {
            "text": "CareHome provides premium telemedicine services in Lahore and Karachi.",
            "url": "https://example.com/services",
            "title": "Our Services",
            "expected_truth": "medium",
            "expected_yes_no": False
        }
    ]
    
    for case in test_cases:
        chunks = nlp.semantic_chunk_v3(case['text'], url=case['url'], title=case['title'])
        if not chunks:
            print(f"FAILED: No chunks generated for {case['url']}")
            continue
            
        chunk = chunks[0]
        truth = chunk.get('truth_level')
        yes_no = chunk.get('can_answer_yes_no')
        
        print(f"URL: {case['url']}")
        print(f"  Truth: {truth} (Expected: {case['expected_truth']})")
        print(f"  Yes/No: {yes_no} (Expected: {case['expected_yes_no']})")
        
        assert truth == case['expected_truth']
        assert yes_no == case['expected_yes_no']

    print("\n✅ Metadata enrichment verification PASSED!")

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