#!/usr/bin/env python3
# -*- coding: utf-8 -*-
import pymysql, json
cfg = json.load(open('db_config.json'))
conn = pymysql.connect(host=cfg['host'], port=cfg['port'], user=cfg['user'],
                       password=cfg.get('password'), database=cfg['database'], charset='utf8mb4')
cur = conn.cursor()
cur.execute("""SELECT note FROM variable_expenses
               WHERE member_id=3 AND note LIKE '%[微信]%' AND recorded_date BETWEEN '2026-01-01' AND '2026-06-30'""")
notes = [r[0] for r in cur.fetchall()]
# 分析 note 格式种类
import collections
fmt = collections.Counter()
for n in notes:
    if n.startswith('[微信] 支出'):
        fmt['导入格式-支出'] += 1
    elif '微信' in n and '转账' in n:
        fmt['含微信+转账'] += 1
    else:
        fmt['其他:' + n[:30]] += 1
print("note格式分布:")
for k, v in fmt.most_common(10):
    print(f"  {v}笔  {k}")
# 看几条历史微信的note样例
print("\n样例(非本次导入格式):")
shown = 0
for n in notes:
    if not n.startswith('[微信] 支出'):
        print("  ", n[:60])
        shown += 1
        if shown >= 5: break
conn.close()