glam/scripts/test_alternative_names.py

#!/usr/bin/env python3
"""
Test alternative name matching for Tunisia enrichment.
Tests Diocesan Library and Kerkouane Museum before full run.
"""

import yaml
from pathlib import Path
import sys

# Import the search function
sys.path.insert(0, str(Path(__file__).parent))
from enrich_tunisia_wikidata_validated import search_wikidata_with_validation

def test_diocesan_library():
    """Test Diocesan Library of Tunis with alternative names."""
    print("\n" + "="*60)
    print("TEST 1: Diocesan Library of Tunis")
    print("="*60)
    
    primary_name = "Diocesan Library of Tunis"
    alternative_names = [
        "Bibliothèque Diocésaine de Tunis",
        "Tunis Diocesan Library"
    ]
    
    print(f"Primary name: {primary_name}")
    print(f"Alternative names: {alternative_names}")
    print()
    
    # Test WITHOUT alternative names (baseline)
    print("🔍 Test 1a: Searching with PRIMARY name only...")
    result_primary = search_wikidata_with_validation(
        name=primary_name,
        inst_type='LIBRARY',
        city='Tunis',
        alternative_names=None
    )
    
    if result_primary:
        print(f"  ✅ Found: {result_primary['qid']} - {result_primary['name']}")
        print(f"  Match score: {result_primary['match_score']:.1f}%")
        print(f"  Matched name: {result_primary.get('matched_name', 'N/A')}")
    else:
        print(f"  ❌ NOT FOUND with primary name")
    
    # Test WITH alternative names
    print("\n🔍 Test 1b: Searching with ALTERNATIVE names...")
    result_alt = search_wikidata_with_validation(
        name=primary_name,
        inst_type='LIBRARY',
        city='Tunis',
        alternative_names=alternative_names
    )
    
    if result_alt:
        print(f"  ✅ Found: {result_alt['qid']} - {result_alt['name']}")
        print(f"  Match score: {result_alt['match_score']:.1f}%")
        print(f"  Matched name: {result_alt.get('matched_name', 'N/A')}")
        
        if result_alt['qid'] == 'Q28149782':
            print(f"  ✅✅ CORRECT Q-NUMBER FOUND!")
        else:
            print(f"  ⚠️  Wrong Q-number (expected Q28149782)")
    else:
        print(f"  ❌ NOT FOUND even with alternative names")
    
    return result_alt is not None and result_alt.get('qid') == 'Q28149782'

def test_kerkouane_museum():
    """Test Kerkouane Archaeological Site and Museum."""
    print("\n" + "="*60)
    print("TEST 2: Kerkouane Archaeological Site and Museum")
    print("="*60)
    
    primary_name = "Kerkouane Archaeological Site and Museum"
    alternative_names = [
        "Musée de Kerkouane",
        "Kerkouane Museum"
    ]
    
    print(f"Primary name: {primary_name}")
    print(f"Alternative names: {alternative_names}")
    print()
    
    # Test WITHOUT alternative names (baseline)
    print("🔍 Test 2a: Searching with PRIMARY name only...")
    result_primary = search_wikidata_with_validation(
        name=primary_name,
        inst_type='MUSEUM',
        city='El Haouaria',
        alternative_names=None
    )
    
    if result_primary:
        print(f"  ✅ Found: {result_primary['qid']} - {result_primary['name']}")
        print(f"  Match score: {result_primary['match_score']:.1f}%")
        print(f"  Entity type: {result_primary.get('entity_type', 'N/A')}")
    else:
        print(f"  ❌ NOT FOUND with primary name")
    
    # Test WITH alternative names
    print("\n🔍 Test 2b: Searching with ALTERNATIVE names...")
    result_alt = search_wikidata_with_validation(
        name=primary_name,
        inst_type='MUSEUM',
        city='El Haouaria',
        alternative_names=alternative_names
    )
    
    if result_alt:
        print(f"  ✅ Found: {result_alt['qid']} - {result_alt['name']}")
        print(f"  Match score: {result_alt['match_score']:.1f}%")
        print(f"  Matched name: {result_alt.get('matched_name', 'N/A')}")
        print(f"  Entity type: {result_alt.get('entity_type', 'N/A')}")
        
        if result_alt['qid'] == 'Q3329437':
            print(f"  ✅✅ CORRECT Q-NUMBER FOUND!")
        else:
            print(f"  ⚠️  Different Q-number (might be correct)")
    else:
        print(f"  ❌ NOT FOUND even with alternative names")
    
    return result_alt is not None

def main():
    print("\n🧪 TESTING ALTERNATIVE NAME MATCHING")
    print("Testing whether French alternative names improve match scores")
    print()
    
    test1_passed = test_diocesan_library()
    test2_passed = test_kerkouane_museum()
    
    print("\n" + "="*60)
    print("TEST SUMMARY")
    print("="*60)
    print(f"Diocesan Library: {'✅ PASS' if test1_passed else '❌ FAIL'}")
    print(f"Kerkouane Museum: {'✅ PASS' if test2_passed else '❌ FAIL'}")
    
    if test1_passed and test2_passed:
        print("\n✅✅ ALL TESTS PASSED! Ready for full enrichment run.")
        return 0
    elif test1_passed or test2_passed:
        print("\n⚠️  PARTIAL SUCCESS - review results before full run")
        return 0
    else:
        print("\n❌ TESTS FAILED - alternative name approach needs adjustment")
        return 1

if __name__ == '__main__':
    sys.exit(main())