Correctly detect transactions with thousands
This commit is contained in:
parent
9eed49c344
commit
c425ebcb00
@ -10,22 +10,23 @@ PATH = f'/home/benjamin/Desktop/bens_folder/bazaar/documents/bnp/bank_statements
|
||||
|
||||
os.chdir(PATH)
|
||||
|
||||
PRINT_TRANSACTIONS = True
|
||||
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():
|
||||
for folder in os.listdir():#[-2:]:
|
||||
if folder != MAIN_BANK_ACCOUNT:
|
||||
print(folder)
|
||||
for file in os.listdir(folder):
|
||||
for file in os.listdir(folder):#[-1:]:
|
||||
print(file)
|
||||
filePath = f'{folder}/{file}'
|
||||
initialAmount, totalMonthlyDebit, totalMonthlyCredit, transactions, fileDatetime = readPdfBankStatement(filePath)
|
||||
print(totalMonthlyCredit)
|
||||
print(transactions)
|
||||
#print(totalMonthlyCredit)
|
||||
#import json
|
||||
#print(json.dumps(transactions, indent = 4))
|
||||
otherBankAccountsCredits[fileDatetime] = otherBankAccountsCredits.get(fileDatetime, 0) + totalMonthlyCredit
|
||||
|
||||
#exit(1)
|
||||
@ -48,13 +49,16 @@ for folder in sorted(os.listdir()):
|
||||
print(f'Total monthly debit: {totalMonthlyDebit}')
|
||||
print(f'Total monthly credit: {totalMonthlyCredit}')
|
||||
totalMonthlyDebits += [totalMonthlyDebit]
|
||||
totalMonthlyCredit += otherBankAccountsCredits.get(fileDatetime, 0)
|
||||
otherBankAccountsCredit = otherBankAccountsCredits.get(fileDatetime, 0)
|
||||
print(f'{otherBankAccountsCredit=}')
|
||||
totalMonthlyCredit += otherBankAccountsCredit
|
||||
totalMonthlyCredits += [totalMonthlyCredit]
|
||||
totalMonthlyDifference = totalMonthlyCredit - totalMonthlyDebit
|
||||
totalMonthlyDifferences += [totalMonthlyDifference]
|
||||
if PRINT_TRANSACTIONS:
|
||||
for transaction in transactions:
|
||||
print(transaction['date'], transaction['valeur'], transaction['amount'], transaction['currentAmount'])
|
||||
# , transaction['currentAmount']
|
||||
print(transaction['date'], transaction['valeur'], transaction['amount'])
|
||||
print(transaction['comment'])
|
||||
print()
|
||||
if firstDatetime is None:
|
||||
|
29
utils.py
29
utils.py
@ -2,11 +2,11 @@ import subprocess
|
||||
from datetime import datetime
|
||||
import re
|
||||
|
||||
FIRST_LINE_OF_PAYMENT_REGEX = re.compile('\\d{2}\\.\\d{2} \\d{2}\\.\\d{2} \\d+,\\d{2}')
|
||||
FIRST_LINE_OF_PAYMENT_REGEX = re.compile('(\\d{2}\\.\\d{2}) (\\d{2}\\.\\d{2}) ([\\d ]+,\\d{2})')
|
||||
END_PAGE_AFTER_THE_FIRST_ONE_REGEX = re.compile('P\\. \\d+/\\d+')
|
||||
SOLDE_CREDITEUR_AU_REGEX = re.compile('SOLDE CREDITEUR AU \\d{2}\\.\\d{2}\\.\\d{4}')
|
||||
TOTAL_DES_OPERATIONS_REGEX = re.compile('TOTAL\\ DES\\ OPERATIONS\\ ([0-9 ]+,\\d{2})\\ ([0-9 ]+,\\d{2})')
|
||||
TOTAL_DES_OPERATIONS_CREDIT_ONLY_REGEX = re.compile('TOTAL\\ DES\\ OPERATIONS\\ ([0-9 ]+,\\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})')
|
||||
|
||||
def execute(command):
|
||||
return subprocess.check_output(command).decode('utf-8')
|
||||
@ -25,15 +25,19 @@ def getMonthIndexSinceEpoch(aDatetime):
|
||||
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 readPdfBankStatement(filePath):
|
||||
file = filePath.split('/')[-1]
|
||||
fileDatetime = getDatetimeFromFileName(file)
|
||||
content = getTextFromPdf(filePath)
|
||||
lines = content.splitlines()
|
||||
#print('\n'.join(lines))
|
||||
started = False
|
||||
firstPage = True
|
||||
initialAmount = None
|
||||
currentAmount = None
|
||||
#currentAmount = None
|
||||
date = None
|
||||
comment = []
|
||||
transactions = []
|
||||
@ -43,7 +47,7 @@ def readPdfBankStatement(filePath):
|
||||
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
|
||||
#currentAmount = initialAmount
|
||||
started = True
|
||||
continue
|
||||
else:
|
||||
@ -57,27 +61,26 @@ def readPdfBankStatement(filePath):
|
||||
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
|
||||
break
|
||||
if FIRST_LINE_OF_PAYMENT_REGEX.match(line) is not None:
|
||||
#print(line)
|
||||
firstLineOfPaymentRegexMatch = FIRST_LINE_OF_PAYMENT_REGEX.match(line)
|
||||
if firstLineOfPaymentRegexMatch is not None:
|
||||
if date is not None:
|
||||
transactions += [{
|
||||
'date': date,
|
||||
'valeur': valeur,
|
||||
'amount': amount,
|
||||
'currentAmount': currentAmount,
|
||||
#'currentAmount': currentAmount,
|
||||
'comment': '\n'.join(comment)
|
||||
}]
|
||||
date = None
|
||||
date, valeur, amount = line.split()
|
||||
amount = float(amount.replace(',', '.'))
|
||||
currentAmount -= amount
|
||||
date, valeur, amount = firstLineOfPaymentRegexMatch.groups()
|
||||
amount = toFloat(amount)
|
||||
#currentAmount -= amount
|
||||
comment = []
|
||||
else:
|
||||
comment += [line]
|
||||
@ -86,7 +89,7 @@ def readPdfBankStatement(filePath):
|
||||
'date': date,
|
||||
'valeur': valeur,
|
||||
'amount': amount,
|
||||
'currentAmount': currentAmount,
|
||||
#'currentAmount': currentAmount,
|
||||
'comment': '\n'.join(comment)
|
||||
}]
|
||||
return initialAmount, totalMonthlyDebit, totalMonthlyCredit, transactions, fileDatetime
|
Loading…
Reference in New Issue
Block a user