BNP_PDF_statement_parser/bnp_pdf_statement_parser.py

#!/usr/bin/env python

import os
import matplotlib.pyplot as plt
import matplotlib.ticker as ticker
from datetime import datetime
from utils import getTextFromPdf, getDatetimeFromFileName, getMonthIndexSinceEpoch, getMonthNameFromMonthIndex, FIRST_LINE_OF_PAYMENT_REGEX, END_PAGE_AFTER_THE_FIRST_ONE_REGEX, SOLDE_CREDITEUR_AU_REGEX, TOTAL_DES_OPERATIONS_REGEX, TOTAL_DES_OPERATIONS_CREDIT_ONLY_REGEX

PATH = f'/home/benjamin/Desktop/bens_folder/bazaar/documents/bnp/bank_statements/'

os.chdir(PATH)

PRINT_TRANSACTIONS = False

MAIN_BANK_ACCOUNT = 'compte_de_cheques'

# As far as I know there was no debit yet.
otherBankAccountsCredits = {}

for folder in os.listdir():
    if folder != MAIN_BANK_ACCOUNT:
        print(folder)
        for file in os.listdir(folder):
            print(file)
            fileDatetime = getDatetimeFromFileName(file)
            print(fileDatetime)

            #otherBankAccountsCredits[fileDatetime] = otherBankAccountsCredits.get(fileDatetime, 0) +

os.chdir(f'{MAIN_BANK_ACCOUNT}/')

totalMonthlyDebits = []
totalMonthlyCredits = []
totalMonthlyDifferences = []
totals = []
firstDatetime = None
lastDatetime = None
for folder in sorted(os.listdir()):
    for file in sorted(os.listdir(folder)):
        filePath = f'{folder}/{file}'
        print(filePath)
        fileDatetime = getDatetimeFromFileName(file)
        if firstDatetime is None:
            firstDatetime = fileDatetime
        content = getTextFromPdf(filePath)
        lines = content.splitlines()
        started = False
        firstPage = True
        initialAmount = None
        currentAmount = None
        date = None
        comment = []
        for line in lines:
            if not started:
                # We are interested in the content after this line:
                if SOLDE_CREDITEUR_AU_REGEX.match(line) is not None or (line.startswith('Date Nature des opérations Valeur Débit Crédit') and not firstPage):
                    if SOLDE_CREDITEUR_AU_REGEX.match(line):
                        initialAmount = float(SOLDE_CREDITEUR_AU_REGEX.sub('', line).replace(',', '.').replace(' ', ''))
                        currentAmount = initialAmount
                        print('Initial amount', initialAmount)
                        print()
                        totals += [initialAmount]
                    started = True
                    continue
            else:
                # We aren't interested in the content after this line:
                if line.startswith('BNP PARIBAS SA au capital de') or END_PAGE_AFTER_THE_FIRST_ONE_REGEX.match(line) is not None:
                    firstPage = False
                    started = False
                    continue
                # We aren't interested in the content after this line
                elif line.startswith('TOTAL DES OPERATIONS'):
                    totalDesOperationsRegexMatch = TOTAL_DES_OPERATIONS_REGEX.match(line)
                    # Note that transfer between accounts will be noted in both debits and credits, as trying to cancel would make benefits show as negative debit which does not make sense.
                    # Cannot just consider January as benefits only as `20240122.pdf` also contains an additional transfer between my accounts.
                    toFloat = lambda group: float(group.replace(',', '.').replace(' ', ''))
                    if totalDesOperationsRegexMatch is not None:
                        totalMonthlyDebit, totalMonthlyCredit = [toFloat(group) for group in totalDesOperationsRegexMatch.groups()]
                    else:
                        totalMonthlyCredit = toFloat(TOTAL_DES_OPERATIONS_CREDIT_ONLY_REGEX.match(line).group(1))
                        totalMonthlyDebit = 0
                    if os.getcwd().endswith(f'/{MAIN_BANK_ACCOUNT}'):
                        totalMonthlyCredit += otherBankAccountsCredits[fileDatetime]
                    print(f'Total monthly debit: {totalMonthlyDebit}')
                    print(f'Total monthly credit: {totalMonthlyCredit}')
                    totalMonthlyDebits += [totalMonthlyDebit]
                    totalMonthlyCredits += [totalMonthlyCredit]
                    totalMonthlyDifference = totalMonthlyCredit - totalMonthlyDebit
                    totalMonthlyDifferences += [totalMonthlyDifference]
                    break
                if FIRST_LINE_OF_PAYMENT_REGEX.match(line) is not None:
                    if date is not None and PRINT_TRANSACTIONS:
                        print(date, valeur, amount, currentAmount)
                        print('\n'.join(comment))
                        print()
                    date, valeur, amount = line.split()
                    amount = float(amount.replace(',', '.'))
                    currentAmount -= amount
                    comment = []
                else:
                    comment += [line]
        #break
    #break
lastDatetime = getDatetimeFromFileName(file)

fig, ax = plt.subplots()
plt.title('Monthly debits and credits')
plt.xlabel('Date')
plt.ylabel('€')
ALPHA = 0.5

xTicks = range(getMonthIndexSinceEpoch(firstDatetime), getMonthIndexSinceEpoch(lastDatetime) + 1)
totalMonthlyAmountAndLabel = (
    #(totalMonthlyDebits, 'Debit'),
    #(totalMonthlyCredits, 'Credit'),
    (totalMonthlyDifferences, 'Difference'),
    (totals, 'Total'),
)
for totalMonthlyAmount, totalMonthlyLabel in totalMonthlyAmountAndLabel:
    plt.bar(xTicks, totalMonthlyAmount, alpha = ALPHA, label = totalMonthlyLabel)
plt.legend()

#plt.yscale('symlog')
ax.yaxis.set_major_formatter(ticker.StrMethodFormatter('{x:,}'))

ticksLabels = [getMonthNameFromMonthIndex(monthIndex) for monthIndex in xTicks]
plt.xticks(xTicks, ticksLabels, rotation = 90)
#plt.tight_layout()
# How to show the horizontal lines for subticks?
plt.grid(axis = 'y')

plt.show()