glam/scripts/fix_argentina_region_codes.py

#!/usr/bin/env python3
"""Fix Argentina region codes from numeric/invalid codes to ISO 3166-2:AR codes."""

import os
import re
import yaml
from datetime import datetime, timezone
from pathlib import Path

CUSTODIAN_DIR = Path("/Users/kempersc/apps/glam/data/custodian")

# Argentina GeoNames admin1 -> ISO 3166-2:AR mapping
# Reference: https://en.wikipedia.org/wiki/ISO_3166-2:AR
# GeoNames uses 2-digit numeric codes
AR_CODE_MAPPING = {
    # Numeric GeoNames codes to ISO
    "01": "B",   # Buenos Aires Province
    "02": "K",   # Catamarca
    "03": "H",   # Chaco
    "04": "U",   # Chubut
    "05": "X",   # Córdoba
    "06": "W",   # Corrientes
    "07": "E",   # Entre Ríos
    "08": "P",   # Formosa
    "09": "Y",   # Jujuy
    "10": "L",   # La Pampa
    "11": "F",   # La Rioja
    "12": "M",   # Mendoza
    "13": "N",   # Misiones
    "14": "Q",   # Neuquén
    "15": "R",   # Río Negro
    "16": "A",   # Salta
    "17": "J",   # San Juan
    "18": "D",   # San Luis
    "19": "Z",   # Santa Cruz
    "20": "S",   # Santa Fe
    "21": "G",   # Santiago del Estero
    "22": "V",   # Tierra del Fuego
    "23": "T",   # Tucumán
    "24": "C",   # Ciudad Autónoma de Buenos Aires
    # Invalid 2-letter codes that need fixing
    "BU": "B",   # Buenos Aires (should be single letter)
    "SA": "A",   # Salta (should be single letter)
    "EN": "E",   # Entre Ríos (should be single letter)
    "CI": "C",   # CABA (should be single letter)
    "LA": "L",   # La Pampa (should be single letter)
    "CO": "X",   # Córdoba (should be single letter X, not CO)
    "ME": "M",   # Mendoza (should be single letter)
    "CF": "C",   # Ciudad Federal - alias for CABA
    "TU": "T",   # Tucumán (should be single letter)
    "SF": "S",   # Santa Fe (should be single letter)
    "MI": "N",   # Misiones (should be N, not MI)
    "RI": "R",   # Río Negro (should be single letter)
    "NE": "Q",   # Neuquén (should be Q, not NE)
    "JU": "Y",   # Jujuy (should be Y, not JU)
    "CH": "U",   # Chubut (should be U, not CH)
    "CA": "K",   # Catamarca (should be K, not CA)
    # Already correct single-letter codes
    "A": "A", "B": "B", "C": "C", "D": "D", "E": "E", "F": "F", "G": "G",
    "H": "H", "J": "J", "K": "K", "L": "L", "M": "M", "N": "N", "P": "P",
    "Q": "Q", "R": "R", "S": "S", "T": "T", "U": "U", "V": "V", "W": "W",
    "X": "X", "Y": "Y", "Z": "Z",
}

def fix_file(filepath: Path) -> tuple[bool, str]:
    """Fix Argentina region code in file."""
    with open(filepath, 'r', encoding='utf-8') as f:
        content = f.read()
        data = yaml.safe_load(content)
    
    if not data or 'ghcid' not in data:
        return False, "No GHCID"
    
    current = data['ghcid'].get('ghcid_current', '')
    match = re.match(r'^AR-([A-Z0-9]{1,2})-(.+)$', current)
    if not match:
        return False, "Invalid format"
    
    old_region = match.group(1)
    rest = match.group(2)
    
    if old_region not in AR_CODE_MAPPING:
        return False, f"Unknown region code {old_region}"
    
    new_region = AR_CODE_MAPPING[old_region]
    
    if old_region == new_region:
        return False, "Already correct"
    
    new_ghcid = f"AR-{new_region}-{rest}"
    new_filename = f"{new_ghcid}.yaml"
    new_filepath = CUSTODIAN_DIR / new_filename
    
    # Check for collision
    if new_filepath.exists() and new_filepath != filepath:
        return False, f"COLLISION: {new_ghcid}"
    
    # Update GHCID
    timestamp = datetime.now(timezone.utc).isoformat()
    data['ghcid']['ghcid_current'] = new_ghcid
    
    # Update location_resolution
    if 'location_resolution' in data['ghcid']:
        data['ghcid']['location_resolution']['region_code'] = new_region
    
    # Add history entry
    if 'ghcid_history' not in data['ghcid']:
        data['ghcid']['ghcid_history'] = []
    data['ghcid']['ghcid_history'].insert(0, {
        'ghcid': new_ghcid,
        'valid_from': timestamp,
        'reason': f"Fixed region code: {old_region} -> {new_region} (ISO 3166-2:AR)"
    })
    
    # Update location.region_code if present
    if 'location' in data and isinstance(data['location'], dict):
        if data['location'].get('region_code') == old_region:
            data['location']['region_code'] = new_region
    
    # Update identifiers
    if 'identifiers' in data:
        for ident in data['identifiers']:
            if ident.get('identifier_scheme') == 'GHCID':
                ident['identifier_value'] = new_ghcid
    
    # Write updated content
    with open(filepath, 'w', encoding='utf-8') as f:
        yaml.dump(data, f, allow_unicode=True, default_flow_style=False, sort_keys=False)
    
    # Rename file
    if new_filepath != filepath:
        os.rename(filepath, new_filepath)
    
    return True, f"{current} -> {new_ghcid}"

def main():
    files = list(CUSTODIAN_DIR.glob("AR-*.yaml"))
    print(f"Found {len(files)} Argentina files")
    
    fixed = 0
    errors = 0
    collisions = []
    
    for f in sorted(files):
        # Check if region needs correction
        match = re.match(r'^AR-([A-Z0-9]{1,2})-', f.name)
        if match:
            old_region = match.group(1)
            # Check if it needs fixing (numeric or 2-letter invalid)
            if old_region in AR_CODE_MAPPING and old_region != AR_CODE_MAPPING[old_region]:
                success, msg = fix_file(f)
                if success:
                    print(f"  Fixed: {msg}")
                    fixed += 1
                elif "COLLISION" in msg:
                    print(f"  {msg}")
                    collisions.append((f.name, msg))
                else:
                    print(f"  Error: {f.name}: {msg}")
                    errors += 1
    
    print(f"\nSummary: Fixed {fixed}, Collisions {len(collisions)}, Errors {errors}")
    
    if collisions:
        print("\nCollisions to resolve:")
        for name, msg in collisions:
            print(f"  {name}: {msg}")

if __name__ == "__main__":
    main()