glam/scripts/migrate_web_dirs_to_ghcid.py

#!/usr/bin/env python3
"""
Migrate web directories from index-based naming to GHCID-based naming.

This script reads the _entry_to_ghcid.txt mapping file and renames directories
from numeric indices (e.g., 0002, 0681) to their corresponding GHCID names
(e.g., NL-DR-ASS-A-DA, NL-NH-HEE-S-HKH).

Usage:
    python scripts/migrate_web_dirs_to_ghcid.py [--dry-run] [--verbose]

Options:
    --dry-run   Show what would be done without making changes
    --verbose   Show detailed progress information
"""

import os
import sys
import argparse
from pathlib import Path
from collections import defaultdict


def parse_mapping_file(mapping_path: Path) -> dict[str, str]:
    """Parse the entry to GHCID mapping file.
    
    Returns dict mapping index (as string with leading zeros) to GHCID.
    For duplicate indices, uses the first mapping encountered.
    """
    mapping = {}
    seen_indices = set()
    
    with open(mapping_path, 'r') as f:
        for line in f:
            line = line.strip()
            if not line:
                continue
            
            parts = line.split(maxsplit=1)
            if len(parts) != 2:
                print(f"Warning: Skipping malformed line: {line}")
                continue
            
            index_str, ghcid = parts
            
            # Handle both "0" and "0002" style indices
            # Normalize to 4-digit zero-padded format
            try:
                index_int = int(index_str)
                index_normalized = str(index_int)  # Without padding for directory matching
                index_padded = f"{index_int:04d}"  # With padding for some directories
            except ValueError:
                print(f"Warning: Invalid index '{index_str}' in line: {line}")
                continue
            
            # Skip duplicates - use first mapping
            if index_normalized in seen_indices:
                continue
            
            seen_indices.add(index_normalized)
            mapping[index_normalized] = ghcid
            mapping[index_padded] = ghcid  # Also store padded version
    
    return mapping


def find_index_directories(web_dir: Path) -> list[Path]:
    """Find all directories with numeric names."""
    index_dirs = []
    
    for item in web_dir.iterdir():
        if item.is_dir() and item.name.isdigit():
            index_dirs.append(item)
        elif item.is_dir() and item.name.lstrip('0').isdigit():
            # Handle zero-padded names like "0002"
            index_dirs.append(item)
    
    return sorted(index_dirs, key=lambda p: int(p.name.lstrip('0') or '0'))


def migrate_directories(
    web_dir: Path,
    mapping: dict[str, str],
    dry_run: bool = False,
    verbose: bool = False
) -> tuple[int, int, int]:
    """Migrate directories from index to GHCID naming.
    
    Returns (success_count, skip_count, error_count).
    """
    index_dirs = find_index_directories(web_dir)
    
    success_count = 0
    skip_count = 0
    error_count = 0
    
    for dir_path in index_dirs:
        dir_name = dir_path.name
        # Try both padded and unpadded versions
        index_key = dir_name.lstrip('0') or '0'
        
        if index_key not in mapping and dir_name not in mapping:
            if verbose:
                print(f"Skip: No mapping for index {dir_name}")
            skip_count += 1
            continue
        
        ghcid = mapping.get(index_key) or mapping.get(dir_name)
        new_path = web_dir / ghcid
        
        # Check if target already exists
        if new_path.exists():
            if verbose:
                print(f"Skip: Target already exists: {ghcid}")
            skip_count += 1
            continue
        
        # Perform the rename
        if dry_run:
            print(f"Would rename: {dir_name} -> {ghcid}")
            success_count += 1
        else:
            try:
                dir_path.rename(new_path)
                if verbose:
                    print(f"Renamed: {dir_name} -> {ghcid}")
                success_count += 1
            except OSError as e:
                print(f"Error renaming {dir_name}: {e}")
                error_count += 1
    
    return success_count, skip_count, error_count


def main():
    parser = argparse.ArgumentParser(
        description="Migrate web directories from index to GHCID naming"
    )
    parser.add_argument(
        "--dry-run",
        action="store_true",
        help="Show what would be done without making changes"
    )
    parser.add_argument(
        "--verbose",
        action="store_true",
        help="Show detailed progress information"
    )
    args = parser.parse_args()
    
    # Paths
    project_root = Path(__file__).parent.parent
    web_dir = project_root / "data" / "custodian" / "web"
    mapping_file = web_dir / "_entry_to_ghcid.txt"
    
    # Validate paths
    if not web_dir.exists():
        print(f"Error: Web directory not found: {web_dir}")
        sys.exit(1)
    
    if not mapping_file.exists():
        print(f"Error: Mapping file not found: {mapping_file}")
        sys.exit(1)
    
    # Parse mapping
    print(f"Reading mapping from: {mapping_file}")
    mapping = parse_mapping_file(mapping_file)
    print(f"Found {len(mapping) // 2} unique index-to-GHCID mappings")
    
    # Find directories to migrate
    index_dirs = find_index_directories(web_dir)
    print(f"Found {len(index_dirs)} index-based directories")
    
    if args.dry_run:
        print("\n=== DRY RUN MODE ===\n")
    
    # Perform migration
    success, skip, error = migrate_directories(
        web_dir, mapping, dry_run=args.dry_run, verbose=args.verbose
    )
    
    # Summary
    print(f"\n=== Summary ===")
    print(f"Successfully {'would rename' if args.dry_run else 'renamed'}: {success}")
    print(f"Skipped (no mapping or target exists): {skip}")
    print(f"Errors: {error}")
    
    if args.dry_run and success > 0:
        print(f"\nRun without --dry-run to perform actual migration")
    
    return 0 if error == 0 else 1


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