#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""分析汪礼平跨月重复的转账/红包对象"""
import json, pymysql, re
from collections import defaultdict
cfg = json.load(open('db_config.json'))
conn = pymysql.connect(host=cfg['host'], port=cfg['port'], user=cfg['user'],
                       password=str(cfg['password']), database=cfg['database'], charset='utf8mb4')
cur = conn.cursor()

cur.execute("""
  SELECT recorded_date, amount, note FROM variable_expenses
  WHERE member_id=1 AND (category='转账/红包' OR category LIKE '%红包%' OR note LIKE '%微信转账%' OR note LIKE '%微信红包%')
""")
rows = cur.fetchall()

def peer_of(note):
    # note 形如  [微信] 转账|对方名|备注  或 转账 | 对方 | 备注
    parts = [p.strip() for p in note.replace('[微信] ', '').split('|')]
    if len(parts) >= 2:
        return parts[1]
    m = re.search(r'转账\s*[:|]\s*([^|\s]+)', note)
    return None

# 按对方+月份统计
by_peer_month = defaultdict(set)   # peer -> set of 'YYYY-MM'
by_peer_total = defaultdict(float)
by_peer_count = defaultdict(int)

for d, amt, note in rows:
    p = peer_of(note)
    if not p:
        p = '(未知)'
    month = d.strftime('%Y-%m')
    by_peer_month[p].add(month)
    by_peer_total[p] += float(amt)
    by_peer_count[p] += 1

# 按月数排序，找出跨月重复的
print('=== 转账对象出现月份数（>=2个月视为规律性） ===')
multi = []
for p, months in by_peer_month.items():
    if len(months) >= 2:
        multi.append((len(months), by_peer_count[p], by_peer_total[p], p, sorted(months)))
multi.sort(key=lambda x: (-x[0], -x[2]))
for nmon, ncnt, tot, p, months in multi:
    print(f'  {p:16s} 出现{nmon}个月共{ncnt}笔 累计¥{tot:.0f}  月:{months}')
conn.close()