#!/usr/bin/env python3
"""
PST 邮件清理规则执行脚本 v1.2
用法: python3 apply_rules.py <csv_file>
"""
import csv
import json
import sys
from collections import defaultdict

RULES_FILE = '/Users/weishen/pst-processing/rules/delete_rules.json'

def load_rules():
    with open(RULES_FILE) as f:
        return json.load(f)['rules']

def apply_rules(csv_path):
    rules = load_rules()
    rows = list(csv.DictReader(open(csv_path, encoding='utf-8')))
    
    for r in rows:
        r['delete_flag'] = 'N'
    
    total_deleted = 0
    
    for rule in rules:
        folder_pattern = rule['folder_contains']
        action = rule['action']
        
        folder_rows = [r for r in rows if folder_pattern in r['folder']]
        if not folder_rows:
            continue
        
        deleted_count = 0
        kept_count = 0
        
        if action == 'keep':
            kept_count = len(folder_rows)
        
        elif action == 'delete_all':
            for r in folder_rows:
                r['delete_flag'] = 'Y'
                deleted_count += 1
        
        elif action == 'keep_sample':
            keep_limit = rule['keep_count']
            subject_map = defaultdict(list)
            for r in folder_rows:
                subj = r['subject'].strip()[:80]
                subject_map[subj].append(r)
            sorted_subjects = sorted(subject_map.items(), key=lambda x: -len(x[1]))
            
            for subj, emails in sorted_subjects:
                if kept_count >= keep_limit:
                    for r in emails:
                        r['delete_flag'] = 'Y'
                        deleted_count += 1
                else:
                    for i, r in enumerate(emails):
                        if i == 0:
                            kept_count += 1
                        else:
                            r['delete_flag'] = 'Y'
                            deleted_count += 1
        
        elif action == 'keep_if_attachment':
            for r in folder_rows:
                if r['has_attachment'] == 'Y':
                    kept_count += 1
                else:
                    r['delete_flag'] = 'Y'
                    deleted_count += 1
        
        print(f"  {folder_pattern}: {len(folder_rows)} | 保留{kept_count} | 删除{deleted_count}")
        total_deleted += deleted_count
    
    total_kept = len(rows) - total_deleted
    print(f"📊 总计: 保留{total_kept} | 删除{total_deleted}")
    
    fieldnames = list(rows[0].keys())
    out_marked = csv_path.replace('.csv', '_marked.csv')
    out_delete = csv_path.replace('.csv', '_delete_list.csv')
    
    with open(out_marked, 'w', newline='', encoding='utf-8') as f:
        writer = csv.DictWriter(f, fieldnames=fieldnames)
        writer.writeheader()
        writer.writerows(rows)
    
    with open(out_delete, 'w', newline='', encoding='utf-8') as f:
        writer = csv.DictWriter(f, fieldnames=fieldnames)
        writer.writeheader()
        writer.writerows([r for r in rows if r['delete_flag'] == 'Y'])
    
    print(f"✅ 输出: {out_marked}")
    print(f"✅ 删除清单: {out_delete} ({total_deleted} 封)")

if __name__ == '__main__':
    if len(sys.argv) < 2:
        print("用法: python3 apply_rules.py <csv_file>")
        sys.exit(1)
    apply_rules(sys.argv[1])