招行信用卡邮件抓取

parse_cmb.py

import re
import datetime
from imap_tools import MailBox, AND


def parse_html(html):
    list_day = re.findall(re.compile('120%;">2022/(.*?) 您的消费明细如下：', re.S), html)
    # print(list_day)

    list_time = re.findall(re.compile("""<font face="Awesome Font" style="font-size:12px;line-height:120%;">(.*?)</font>""", re.S), html)
    list_time = [i.strip() for i in list_time if ':' in i]
    # print(list_time)

    list_amt = re.findall(re.compile('CNY&nbsp;(.*?)</font>', re.S), html)
    # print(list_amt)

    list_text = re.findall(re.compile('尾号1167&nbsp;消费&nbsp;(.*?)</font>', re.S), html)
    #print(list_text)

    list_total = re.findall(re.compile('￥(.*?)</font>', re.S), html)
    list_total = [i.strip().replace(',', '') for i in list_total if i]
    #print(list_total)

    x = zip(list_day*len(list_time), list_time, list_amt, list_text)
    return [i for i in x]


def write_bean(x):
    if not x:
        return
    month = x[0][0].split('/')[0]
    with open(f'cmb_2022_{month}.bean', 'a') as f:
        for i in x:
            print(i)
            r = f'''

2022-{i[0].replace('/', '-')} * "{i[3]}"
  Liabilities:CreditCard:AMEX               -{i[2]} CNY
  Expenses:Food:Dinner:Supper
  payTime: "2022-{i[0].replace('/', '-')} {i[1]}"
'''
            print(r)
            f.write(r)


def main():
  mailbox =  MailBox('imap.gmail.com').login('[email protected]', 'password', initial_folder='bill')
  for msg in mailbox.fetch(criteria=AND(date=datetime.datetime.now().date(), from_="cmbchina.com"), charset='utf8'):
      if '每日信用管家' not in msg.subject:
          continue
      print("Message id:",msg.uid)
      print("Message Subject:",msg.subject)
      print("Message Date:", msg.date)
      write_bean(parse_html(msg.html))


if __name__ == '__main__':
    main()