glam/scripts/add_alternative_names_category2.py

#!/usr/bin/env python3
"""
Add alternative_names to Category 2 entries (legal name vs trade name variations).
These are VALID entries where custodian_name is correct but we need to capture
the legal name as an alternative.
"""
import yaml
import os
from datetime import datetime

# Define entries to update with their alternative names
CATEGORY2_UPDATES = {
    "0040_museum_contemporary_tibetan_art.yaml": {
        "alternative_names": [
            {"name": "Stichting Tibet House Holland", "source": "original_entry", "note": "Legal foundation name (KvK registration)"},
            {"name": "Tibet House", "source": "inferred", "note": "Short name variant"}
        ],
        "note": "Museum of Contemporary Tibetan Art is the public name; Stichting Tibet House Holland is the legal entity"
    },
    "0138_Q13447121.yaml": {
        "alternative_names": [
            {"name": "Stichting CODA", "source": "original_entry", "note": "Legal foundation name"},
            {"name": "CODA Museum", "source": "google_maps", "note": "Google Maps listing name"}
        ],
        "note": "CODA Apeldoorn is the museum name; Stichting CODA is the legal foundation"
    },
    "0399_heemkundevereniging_roggel.yaml": {
        "alternative_names": [
            {"name": "Museumboerderij De Leimskoel", "source": "google_maps", "note": "Museum farm operated by the society"}
        ],
        "note": "Heemkundevereniging Roggel operates Museumboerderij De Leimskoel"
    },
    "0424_oce_museum.yaml": {
        "alternative_names": [
            {"name": "Océ Museum", "source": "inferred", "note": "Common short name"},
            {"name": "Stichting Behoud Erfgoed Nederlands Lucht- en Ruimtevaartcentrum NLR", "source": "inferred", "note": "Related heritage foundation (if applicable)"}
        ],
        "note": "Heritage organization preserving Océ (now Canon) history, located at Canon Customer Experience Center"
    },
    "0432_Q56460988.yaml": {
        "alternative_names": [
            {"name": "Stichting De Domijnen", "source": "original_entry", "note": "Legal foundation name"},
            {"name": "De Domijnen", "source": "inferred", "note": "Short name variant"},
            {"name": "Filmhuis ZICHT", "source": "google_maps", "note": "Cinema venue in same building"}
        ],
        "note": "Museum De Domijnen is the public name; Stichting De Domijnen is the legal foundation"
    },
    "0697_Q2218933.yaml": {
        "alternative_names": [
            {"name": "Museumhuis Barnaart", "source": "google_maps", "note": "Public museum name"},
            {"name": "Museumhuis Hendrick de Keyser - Huis Barnaart", "source": "inferred", "note": "Full official name"}
        ],
        "note": "Historic house museum, part of Vereniging Hendrick de Keyser museum houses"
    },
    "0752_unknown.yaml": {
        "alternative_names": [
            {"name": "Oudheidkundige Vereniging Medenblick", "source": "original_entry", "note": "Parent historical society"},
            {"name": "OVM", "source": "inferred", "note": "Abbreviation"}
        ],
        "note": "Historisch Museum Medemblik is operated by Oudheidkundige Vereniging Medenblick"
    },
    "0770_unknown.yaml": {
        "alternative_names": [
            {"name": "Stichting Behoud Erfgoed Nederlands Lucht- en Ruimtevaartcentrum NLR", "source": "original_entry", "note": "Full legal name"},
            {"name": "NLR Heritage Museum", "source": "inferred", "note": "English name variant"}
        ],
        "note": "Abbreviated name vs full legal foundation name"
    },
    "0843_Q85311353.yaml": {
        "alternative_names": [
            {"name": "Museum De Waag", "source": "google_maps", "note": "Museum location in De Waag building"},
            {"name": "Speelgoedmuseum Deventer", "source": "original_entry", "note": "Toy museum location"},
            {"name": "Deventer Verhaal (locaties: Speelgoedmuseum & Museum De Waag)", "source": "original_entry", "note": "Full name with locations"}
        ],
        "note": "Deventer Verhaal is the umbrella organization operating multiple museum locations"
    },
    "0893_unknown.yaml": {
        "alternative_names": [
            {"name": "Historische Kring Ommen", "source": "original_entry", "note": "Parent historical society"}
        ],
        "note": "Museum-Ommen is operated by Historische Kring Ommen"
    },
    "1517_rijksmuseum_amsterdam.yaml": {
        "alternative_names": [
            {"name": "Rijksmuseum Amsterdam", "source": "original_entry", "note": "Full name with city"},
            {"name": "Rijks", "source": "inferred", "note": "Colloquial short name"}
        ],
        "note": "Rijksmuseum is the common public name"
    },
    "1610_eye_film_instituut_nederland.yaml": {
        "alternative_names": [
            {"name": "EYE Film Instituut Nederland", "source": "original_entry", "note": "Full legal name"},
            {"name": "EYE", "source": "inferred", "note": "Common abbreviation"},
            {"name": "Eye Film Institute Netherlands", "source": "inferred", "note": "English name"}
        ],
        "note": "Eye Filmmuseum is the public museum name; EYE Film Instituut Nederland is the legal entity"
    }
}

ENTRIES_DIR = "/Users/kempersc/apps/glam/data/nde/enriched/entries"

def update_entry(filename: str, updates: dict):
    """Update an entry with alternative names."""
    filepath = os.path.join(ENTRIES_DIR, filename)
    
    if not os.path.exists(filepath):
        print(f"  ⚠ File not found: {filename}")
        return False
    
    with open(filepath, 'r', encoding='utf-8') as f:
        data = yaml.safe_load(f)
    
    # Add alternative_names if not present
    if 'alternative_names' not in data:
        data['alternative_names'] = []
    
    # Add new alternative names (avoid duplicates)
    existing_names = {alt.get('name', '') for alt in data.get('alternative_names', [])}
    for alt in updates['alternative_names']:
        if alt['name'] not in existing_names:
            data['alternative_names'].append(alt)
            existing_names.add(alt['name'])
    
    # Update custodian_name with note if present
    if 'custodian_name' in data:
        if 'note' in updates:
            data['custodian_name']['name_variation_note'] = updates['note']
        # Boost confidence for validated entries
        if data['custodian_name'].get('confidence', 0) < 0.8:
            data['custodian_name']['confidence'] = 0.85
            data['custodian_name']['manual_review'] = True
            data['custodian_name']['manual_review_timestamp'] = datetime.now().isoformat()
            data['custodian_name']['manual_review_note'] = "Category 2: Legal name vs trade name variation - both names valid"
    
    with open(filepath, 'w', encoding='utf-8') as f:
        yaml.dump(data, f, default_flow_style=False, allow_unicode=True, sort_keys=False)
    
    return True

def main():
    print("Adding alternative_names to Category 2 entries...")
    print("=" * 60)
    
    updated = 0
    for filename, updates in CATEGORY2_UPDATES.items():
        print(f"Processing: {filename}")
        if update_entry(filename, updates):
            print(f"  ✓ Updated with {len(updates['alternative_names'])} alternative names")
            updated += 1
    
    print("=" * 60)
    print(f"Updated {updated}/{len(CATEGORY2_UPDATES)} entries")

if __name__ == "__main__":
    main()