#!/usr/bin/env python3
"""Scan ~/.hermes/skills/ for corrupted .py scripts.

Usage: python corruption-scanner.py [--fix-dir /path/to/output]
Output: JSON summary of clean vs corrupted files.
"""
import json, os, sys
from pathlib import Path

def scan_skills(base_dir):
    results = {"clean": [], "corrupted": [], "skipped": []}
    
    for root, dirs, files in os.walk(base_dir):
        if '__pycache__' in root or 'venv' in root:
            continue
        
        for fname in files:
            if not fname.endswith('.py'):
                continue
            
            fpath = os.path.join(root, fname)
            size = os.path.getsize(fpath)
            
            # Skip empty __init__.py (normal)
            if fname == '__init__.py' and size <= 50:
                results["skipped"].append({"path": fpath, "reason": "empty __init__.py"})
                continue
            
            try:
                with open(fpath, 'r') as f:
                    content = f.read(1000)
                
                first_line = content.strip().split('\n')[0].lower()
                
                if '% partial rewrite' in first_line[:200] or 'latex' in first_line[:200]:
                    results["corrupted"].append({
                        "path": fpath,
                        "size": size,
                        "type": "latex_contamination"
                    })
                elif first_line.startswith('%pdf') or '%PDF' in first_line[:20]:
                    results["corrupted"].append({
                        "path": fpath,
                        "size": size,
                        "type": "pdf_binary"
                    })
                elif len(content) > 200000:
                    results["corrupted"].append({
                        "path": fpath,
                        "size": size,
                        "type": "oversized"
                    })
                else:
                    results["clean"].append({"path": fpath, "size": size})
                    
            except UnicodeDecodeError:
                results["skipped"].append({"path": fpath, "reason": "binary file"})
            except Exception as e:
                results["skipped"].append({"path": fpath, "reason": str(e)})
    
    return results

if __name__ == '__main__':
    base = os.environ.get('HERMES_HOME', str(Path.home() / '.hermes'))
    skills_dir = os.path.join(base, 'skills')
    
    fix_dir = None
    args = sys.argv[1:]
    if '--fix-dir' in args:
        idx = args.index('--fix-dir')
        if idx + 1 < len(args):
            fix_dir = args[idx + 1]
    
    results = scan_skills(skills_dir)
    
    print(json.dumps(results, indent=2, ensure_ascii=False))
    
    # Summary
    print(f"\n{'='*60}")
    print(f"Scan complete:")
    print(f"  Clean: {len(results['clean'])}")
    print(f"  Corrupted: {len(results['corrupted'])}")
    print(f"  Skipped: {len(results['skipped'])}")
    
    if results['corrupted']:
        print(f"\nCorrupted files:")
        for f in results['corrupted']:
            print(f"  ❌ {f['path']} ({f['size']:,} bytes) - {f['type']}")