Analyze transaction messages from bank Mellat, exported by Telexporter to HTML

mellat_sms.py

import matplotlib.pyplot as plt
import os.path
import re
import sys
from bs4 import BeautifulSoup
from datetime import datetime
from persiantools.jdatetime import JalaliDateTime
from pytz import timezone
from typing import List

tz = timezone('Asia/Tehran')


class Message:
    def __init__(self, text: str, date: str, time: str):
        self.text = text
        self.date = date
        self.time = time
        self.jalali = self.gregory = datetime.fromtimestamp(0, tz=tz)
        if date != "" and time != "":
            century = "13" if int(date[:2]) > 50 else "14"
            splTime = time.split(":")
            self.jalali = JalaliDateTime(int(century + date[:2]), int(date[3:5]), int(date[6:8]),
                                         int(splTime[0]), int(splTime[1]),
                                         tzinfo=tz)
            self.gregory = self.jalali.to_gregorian()


class Transaction(Message):
    def __init__(self, text: str, amount: str, remain: str, date: str, time: str):
        Message.__init__(self, text, date, time)
        self.amount = int(amount.replace(",", "")) if isinstance(amount, str) else amount
        self.remain = int(remain.replace(",", "")) if isinstance(remain, str) else remain
        # self.amount /= 220000
        # self.remain /= 220000


class Income(Transaction):
    def __init__(self, text: str, amount: str, remain: str, date: str, time: str):
        Transaction.__init__(self, text, amount, remain, date, time)


class Expense(Transaction):
    def __init__(self, text: str, amount: str, remain: str, date: str, time: str):
        Transaction.__init__(self, text, amount, remain, date, time)


root = os.path.dirname(os.path.realpath(__file__)) + "/"
incomes: List[Transaction] = list()
expenses: List[Transaction] = list()
twoFactor, bullshit = list(), list()

for html in sys.argv[1:]:
    with open(root + html, "r", encoding="utf-8") as f:
        soup = BeautifulSoup(f.read(), 'html.parser')
    every = soup.find_all('div', attrs={'class': 'message'})
    for e in every:
        e = str(e)[21:-6].strip()
        bTime, eTime = re.search(r'<p class="time">', e), re.search(r'</p>', e)
        sTime = e[bTime.end():eTime.start()]
        sDate = ""
        if "واريز به حساب" in e:
            bAmo, eAmo = re.search(r"مبلغ : ", e), re.search(r" ريال", e)
            bRem = re.search(r"موجودي : ", e)
            bRem = bRem if bRem is not None else re.search(r"موجودي: ", e)
            bDat = re.search(r"99/", e)
            bDat = bDat if bDat is not None else re.search(r"00/", e)
            if bAmo is not None and eAmo is not None and bRem is not None and bDat is not None:
                incomes.append(Income(e, e[bAmo.end():eAmo.start()],
                                      e[bRem.end():bRem.end() + re.search(r" ريال", e[bRem.end():]).start()],
                                      e[bDat.start():bDat.start() + 8], sTime))
            else:
                bullshit.append(Message(e, sDate, sTime))
        elif "برداشت از حساب" in e:
            bAmo, eAmo = re.search(r"مبلغ : ", e), re.search(r" ريال", e)
            bRem = re.search(r"موجودي : ", e)
            bRem = bRem if bRem is not None else re.search(r"موجودي: ", e)
            bDat = re.search(r"99/", e)
            bDat = bDat if bDat is not None else re.search(r"00/", e)
            if bAmo is not None and eAmo is not None and bRem is not None and bDat is not None:
                expenses.append(Expense(e, e[bAmo.end():eAmo.start()],
                                        e[bRem.end():bRem.end() + re.search(r" ريال", e[bRem.end():]).start()],
                                        e[bDat.start():bDat.start() + 8], sTime))
            else:
                bullshit.append(Message(e, sDate, sTime))
        elif "رمز برداشت وجه از کارت شما" in e or "رمز پويا" in e:
            twoFactor.append(Message(e, sDate, sTime))
        elif " سود " in e:
            interest = Income(e, e[re.search(r"به مبلغ ", e).end():re.search(r" واريز", e).start()],
                              incomes[-1].remain if len(incomes) > 1 else "-1", sDate, sTime)
            interest.remain += interest.amount
            incomes.append(interest)
        else:
            bullshit.append(Message(e, sDate, sTime))

print("INCOME MESSAGES:", len(incomes))
print("EXPENSE MESSAGES:", len(expenses))
print("TWO-FACTOR MESSAGES:", len(twoFactor))
print("BULLSHIT MESSAGES:", len(bullshit))
print()

all_transactions = incomes + expenses
valid_tas = list()
for t in all_transactions:
    if t.gregory.year != 1970:
        valid_tas.append(t)
valid_tas.sort(key=lambda k: k.gregory)
initial = None
if isinstance(valid_tas[0], Income):
    initial = valid_tas[0].remain - valid_tas[0].amount
else:
    initial = valid_tas[0].remain + valid_tas[0].amount
print("INITIAL:", '{:,}'.format(initial))
print("INCOME:", '{:,}'.format(sum([x.amount for x in incomes])))
print("EXPENSE:", '{:,}'.format(sum([x.amount for x in expenses])))
print("IN", valid_tas[-1].gregory - valid_tas[0].gregory)
print()

while True:
    to_show = input("SHOW WHAT? (b => Balance, e => Expenses, i => Incomes, q => Quit):\n")
    transactions = list()
    if to_show == "b":
        transactions = all_transactions
    elif to_show == "e":
        transactions = expenses
    elif to_show == "i":
        transactions = incomes
    else:
        quit()

    transactions.sort(key=lambda k: k.gregory)
    rems, tims = list(), list()
    for t in transactions:
        if t.gregory.year == 1970:
            continue
        if to_show == "e" or to_show == "i":
            rems.append(t.amount)
        else:
            rems.append(t.remain)
        tims.append(t.gregory)
    plt.plot(tims, rems)
    plt.show()