BNP_PDF_statement_parser/utils.py

import subprocess
from datetime import datetime
import re

# For not-greedy `?`, see Source: [the Stack Overflow answer 766377](https://stackoverflow.com/a/766377)
FIRST_LINE_OF_PAYMENT_REGEX = re.compile('\\ +(\\d{2}\\.\\d{2})\\ +([A-Z\\d /.()*\\-,]+?)\\ +(\\d{2}\\.\\d{2})\\ +([\\d ]+,\\d{2})')
END_PAGE_AFTER_THE_FIRST_ONE_REGEX = re.compile(' +RELEVE ((DE (COMPTE (CHEQUES|D\'EPARGNE LOGEMENT|LEP))|LIVRET (A|JEUNE))|LIVRET DEV. DURABLE ET SOLIDAIRE) +P\\. \\d+/\\d+')
SOLDE_CREDITEUR_AU_REGEX = re.compile('\\ +SOLDE CREDITEUR AU (\\d{2}\\.\\d{2}\\.\\d{4})\\ +([\\d ]+,\\d{2})')
TOTAL_DES_OPERATIONS_REGEX = re.compile('\\ +TOTAL\\ DES\\ OPERATIONS\\ +([\\d ]+,\\d{2})\\ +([\\d ]+,\\d{2})')
TOTAL_DES_OPERATIONS_CREDIT_ONLY_REGEX = re.compile('\\ +TOTAL\\ DES\\ OPERATIONS\\ +([\\d ]+,\\d{2})')
COLUMNS_HEADER = re.compile(' +Date +Nature des opérations +Valeur +Débit +Crédit')

def execute(command):
    return subprocess.check_output(command).decode('utf-8')

def getTextFromPdf(pdfPath):
    return execute(['pdftotext', '-layout', pdfPath, '-'])

def getDatetimeFromFileName(aDatetimeStr):
    aDatetime = datetime(int(aDatetimeStr[:4]), int(aDatetimeStr[4:6]), 1)
    return aDatetime

def getMonthIndexSinceEpoch(aDatetime):
    return aDatetime.year * 12 + aDatetime.month

def getMonthNameFromMonthIndex(monthIndex):
    return datetime((monthIndex - 1) // 12, 1 + (monthIndex - 1) % 12, 1).strftime('%b %Y')

def toFloat(group):
    return float(group.replace(',', '.').replace(' ', ''))

def getDateFollowing(date, initialDate):
    date = datetime.strptime(date, '%d.%m').replace(year = initialDate.year)
    # To support new year.
    if date < initialDate:
        date = date.replace(year = date.year + 1)
    return date

def readPdfBankStatement(filePath):
    file = filePath.split('/')[-1]
    fileDatetime = getDatetimeFromFileName(file)
    content = getTextFromPdf(filePath)
    lines = content.splitlines()
    started = False
    firstPage = True
    initialAmount = None
    initialDate = None
    currentAmount = None
    date = None
    comment = []
    transactions = []
    debitIndex = None
    creditIndex = None
    for line in lines:
        if not started:
            # We are interested in the content after this line:)
            soldeCrediteurAuRegexMatch = SOLDE_CREDITEUR_AU_REGEX.match(line)
            if COLUMNS_HEADER.match(line) is not None:
                getIndex = lambda line, type_: line.index(type_) + len(type_)
                debitIndex = getIndex(line, 'Débit')
                creditIndex = getIndex(line, 'Crédit')
            if soldeCrediteurAuRegexMatch is not None or (COLUMNS_HEADER.match(line) is not None and not firstPage):
                if soldeCrediteurAuRegexMatch is not None:
                    initialDate = datetime.strptime(soldeCrediteurAuRegexMatch.group(1), '%d.%m.%Y')
                    initialAmount = toFloat(soldeCrediteurAuRegexMatch.group(2))
                    print(f'{initialAmount=}')
                    currentAmount = 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
            else:
                totalDesOperationsRegexMatch = TOTAL_DES_OPERATIONS_REGEX.match(line)
                totalDesOperationsCreditOnlyRegexMatch = TOTAL_DES_OPERATIONS_CREDIT_ONLY_REGEX.match(line)
                if totalDesOperationsRegexMatch is not None or totalDesOperationsCreditOnlyRegexMatch is not None:
                    # 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.
                    if totalDesOperationsRegexMatch is not None:
                        totalMonthlyDebit, totalMonthlyCredit = [toFloat(group) for group in totalDesOperationsRegexMatch.groups()]
                    else:
                        totalMonthlyCredit = toFloat(totalDesOperationsCreditOnlyRegexMatch.group(1))
                        totalMonthlyDebit = 0
                    print(f'{totalMonthlyDebit=}')
                    print(f'{totalMonthlyCredit=}')
                    break
            firstLineOfPaymentRegexMatch = FIRST_LINE_OF_PAYMENT_REGEX.match(line)
            if firstLineOfPaymentRegexMatch is not None:
                if date is not None:
                    transactions += [{
                        'date': getDateFollowing(date, initialDate),
                        'valeur': getDateFollowing(valeur, initialDate),
                        'amount': amount,
                        'current amount': currentAmount,
                        'comment': '\n'.join(comment)
                    }]
                    date = None
                date, firstCommentLine, valeur, amount = firstLineOfPaymentRegexMatch.groups()
                lineLen = len(line)
                amount = toFloat(amount)
                if abs(debitIndex - lineLen) < abs(creditIndex - lineLen):
                    amount *= -1
                currentAmount += amount
                comment = [firstCommentLine]
            elif line != '':
                comment += [line.strip()]
    if date is not None:
        transactions += [{
            'date': getDateFollowing(date, initialDate),
            'valeur': getDateFollowing(valeur, initialDate),
            'amount': amount,
            'current amount': currentAmount,
            'comment': '\n'.join(comment)
        }]
    return initialAmount, totalMonthlyDebit, totalMonthlyCredit, transactions, fileDatetime