100 lines
3.3 KiB
Python
100 lines
3.3 KiB
Python
#!/usr/bin/env python3
|
|
"""
|
|
Update NDE YAML with Batch 2 enrichment results (records 11-17).
|
|
|
|
Wikidata Q-numbers found through manual searches and SPARQL queries.
|
|
"""
|
|
|
|
import yaml
|
|
from pathlib import Path
|
|
from datetime import datetime, timezone
|
|
|
|
# Batch 2 results: Records 11-17 (municipalities in Drenthe)
|
|
BATCH_2_MAPPING = {
|
|
"Gemeente Hoogeveen": "Q208012",
|
|
"Gemeente Emmen": "Q14641",
|
|
"Gemeente Meppel": "Q60425",
|
|
"Gemeente Midden-Drenthe": "Q835125",
|
|
"Gemeente Noordenveld": "Q835083",
|
|
"Gemeente Westerveld": "Q747920",
|
|
"Gemeente Tynaarlo": "Q840457",
|
|
}
|
|
|
|
def main():
|
|
"""Update YAML with Batch 2 results."""
|
|
base_dir = Path(__file__).parent.parent
|
|
data_path = base_dir / "data" / "nde" / "voorbeeld_lijst_organisaties_en_diensten-totaallijst_nederland.yaml"
|
|
|
|
print("=" * 80)
|
|
print("NDE WIKIDATA ENRICHMENT - BATCH 2 UPDATE")
|
|
print("=" * 80)
|
|
print()
|
|
|
|
# Load data
|
|
print(f"Loading {data_path.name}...")
|
|
with open(data_path, 'r', encoding='utf-8') as f:
|
|
orgs = yaml.safe_load(f)
|
|
|
|
print(f"Total organizations: {len(orgs)}")
|
|
print()
|
|
|
|
# Create backup
|
|
backup_path = data_path.parent / f"{data_path.stem}.backup.{datetime.now(timezone.utc).strftime('%Y%m%d_%H%M%S')}.yaml"
|
|
print(f"Creating backup: {backup_path.name}")
|
|
with open(backup_path, 'w', encoding='utf-8') as f:
|
|
yaml.dump(orgs, f, allow_unicode=True, default_flow_style=False, sort_keys=False)
|
|
print()
|
|
|
|
# Update records
|
|
updated_count = 0
|
|
already_enriched = 0
|
|
|
|
print("Updating records:")
|
|
print("-" * 80)
|
|
|
|
for i, org in enumerate(orgs):
|
|
org_name = org.get('organisatie', '')
|
|
|
|
if org_name in BATCH_2_MAPPING:
|
|
q_number = BATCH_2_MAPPING[org_name]
|
|
|
|
if org.get('wikidata_id'):
|
|
already_enriched += 1
|
|
print(f"{i+1:3d}. ○ {org_name[:50]:50s} (already enriched: {org['wikidata_id']})")
|
|
else:
|
|
orgs[i]['wikidata_id'] = q_number
|
|
updated_count += 1
|
|
print(f"{i+1:3d}. ✓ {org_name[:50]:50s} → {q_number}")
|
|
|
|
print()
|
|
print("=" * 80)
|
|
|
|
# Save if updates made
|
|
if updated_count > 0:
|
|
print(f"Saving {updated_count} updates to {data_path.name}...")
|
|
with open(data_path, 'w', encoding='utf-8') as f:
|
|
yaml.dump(orgs, f, allow_unicode=True, default_flow_style=False, sort_keys=False)
|
|
print("✅ Saved successfully!")
|
|
else:
|
|
print("⚠️ No new updates to save")
|
|
|
|
print()
|
|
print("=" * 80)
|
|
print("BATCH 2 SUMMARY")
|
|
print("=" * 80)
|
|
print(f"Records in mapping: {len(BATCH_2_MAPPING)}")
|
|
print(f"Newly enriched: {updated_count}")
|
|
print(f"Already enriched: {already_enriched}")
|
|
print(f"Total enriched so far: {sum(1 for o in orgs if o.get('wikidata_id'))}")
|
|
print(f"Remaining records: {len(orgs) - sum(1 for o in orgs if o.get('wikidata_id'))}")
|
|
print()
|
|
|
|
# Calculate overall progress
|
|
total_enriched = sum(1 for o in orgs if o.get('wikidata_id'))
|
|
progress_pct = (total_enriched / len(orgs)) * 100
|
|
|
|
print(f"OVERALL PROGRESS: {total_enriched}/{len(orgs)} ({progress_pct:.1f}%)")
|
|
print()
|
|
|
|
if __name__ == "__main__":
|
|
main()
|