Benjamin_Loison/BNP_PDF_statement_parser
1
0
Fork 0
Code Issues 2 Pull Requests Packages Projects Releases Wiki Activity

Parse all relevés

Browse Source
This commit is contained in:
Benjamin Loison 2024-10-04 01:15:38 +02:00
parent 2d4a10da05
commit e0c99bc068
Signed by: Benjamin_Loison
SSH Key Fingerprint: SHA256:BtnEgYTlHdOg1u+RmYcDE0mnfz1rhv5dSbQ2gyxW8B8
2 changed files with 13 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
bnp_pdf_statement_parser.py
View File

@ -12,7 +12,7 @@ PATH = f'/home/benjamin/Desktop/bens_folder/bazaar/documents/bnp/bank_statements
os.chdir(PATH) os.chdir(PATH)
PRINT_TRANSACTIONS = False PRINT_TRANSACTIONS = True
MAIN_BANK_ACCOUNT = 'compte_de_cheques' MAIN_BANK_ACCOUNT = 'compte_de_cheques'
@ -34,13 +34,13 @@ for folder in os.listdir():
for file in os.listdir(folder): for file in os.listdir(folder):
filePath = f'{folder}/{file}' filePath = f'{folder}/{file}'
print(filePath) print(filePath)
print(readPdfBankStatement(filePath)) #print(readPdfBankStatement(filePath))
#exit(0) #exit(0)
transactions = readPdfBankStatement(filePath)[3] transactions = readPdfBankStatement(filePath)[3]
pprint(transactions) pprint(transactions)
appendTransactions(transactions, folder) appendTransactions(transactions, folder)
exit(1) #exit(1)
os.chdir(f'{MAIN_BANK_ACCOUNT}/') os.chdir(f'{MAIN_BANK_ACCOUNT}/')
for folder in sorted(os.listdir()): for folder in sorted(os.listdir()):
@ -50,11 +50,11 @@ for folder in sorted(os.listdir()):
transactions = readPdfBankStatement(filePath)[3] transactions = readPdfBankStatement(filePath)[3]
appendTransactions(transactions, MAIN_BANK_ACCOUNT) appendTransactions(transactions, MAIN_BANK_ACCOUNT)
if PRINT_TRANSACTIONS: if PRINT_TRANSACTIONS:
for transaction in transactions: pprint(transactions)
printTransaction(transaction)
#break #break
#break #break
exit(1)
allTransactions.sort(key = operator.itemgetter('date')) allTransactions.sort(key = operator.itemgetter('date'))
print(len(allTransactions)) print(len(allTransactions))
pprint(allTransactions) pprint(allTransactions)

20
utils.py
View File

@ -4,10 +4,11 @@ import re
# Source: [the Stack Overflow answer 766377](https://stackoverflow.com/a/766377) # 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})') 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('P\\. \\d+/\\d+') 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})') 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_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})') 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): def execute(command):
return subprocess.check_output(command).decode('utf-8') return subprocess.check_output(command).decode('utf-8')
@ -53,7 +54,7 @@ def readPdfBankStatement(filePath):
if not started: if not started:
# We are interested in the content after this line:) # We are interested in the content after this line:)
soldeCrediteurAuRegexMatch = SOLDE_CREDITEUR_AU_REGEX.match(line) soldeCrediteurAuRegexMatch = SOLDE_CREDITEUR_AU_REGEX.match(line)
if soldeCrediteurAuRegexMatch is not None or (line.startswith('Date Nature des opérations Valeur Débit Crédit') and not firstPage): if soldeCrediteurAuRegexMatch is not None or (COLUMNS_HEADER.match(line) and not firstPage):
if soldeCrediteurAuRegexMatch is not None: if soldeCrediteurAuRegexMatch is not None:
initialDate = datetime.strptime(soldeCrediteurAuRegexMatch.group(1), '%d.%m.%Y') initialDate = datetime.strptime(soldeCrediteurAuRegexMatch.group(1), '%d.%m.%Y')
initialAmount = toFloat(soldeCrediteurAuRegexMatch.group(2)) initialAmount = toFloat(soldeCrediteurAuRegexMatch.group(2))
@ -69,6 +70,7 @@ def readPdfBankStatement(filePath):
continue continue
# We aren't interested in the content after this line # We aren't interested in the content after this line
else: else:
#print('hey', line)
totalDesOperationsRegexMatch = TOTAL_DES_OPERATIONS_REGEX.match(line) totalDesOperationsRegexMatch = TOTAL_DES_OPERATIONS_REGEX.match(line)
totalDesOperationsCreditOnlyRegexMatch = TOTAL_DES_OPERATIONS_CREDIT_ONLY_REGEX.match(line) totalDesOperationsCreditOnlyRegexMatch = TOTAL_DES_OPERATIONS_CREDIT_ONLY_REGEX.match(line)
if totalDesOperationsRegexMatch is not None or totalDesOperationsCreditOnlyRegexMatch is not None: if totalDesOperationsRegexMatch is not None or totalDesOperationsCreditOnlyRegexMatch is not None:
@ -84,7 +86,7 @@ def readPdfBankStatement(filePath):
break break
firstLineOfPaymentRegexMatch = FIRST_LINE_OF_PAYMENT_REGEX.match(line) firstLineOfPaymentRegexMatch = FIRST_LINE_OF_PAYMENT_REGEX.match(line)
if firstLineOfPaymentRegexMatch is not None: if firstLineOfPaymentRegexMatch is not None:
print(line) #print(line)
if date is not None: if date is not None:
transactions += [{ transactions += [{
'date': getDateFollowing(date, initialDate), 'date': getDateFollowing(date, initialDate),
@ -99,8 +101,8 @@ def readPdfBankStatement(filePath):
#currentAmount -= amount #currentAmount -= amount
comment = [firstCommentLine] comment = [firstCommentLine]
elif line != '': elif line != '':
print(f'comment: {line}') #print(f'comment: {line}')
comment += [line] comment += [line.strip()]
if date is not None: if date is not None:
transactions += [{ transactions += [{
'date': getDateFollowing(date, initialDate), 'date': getDateFollowing(date, initialDate),
@ -109,10 +111,4 @@ def readPdfBankStatement(filePath):
#'currentAmount': currentAmount, #'currentAmount': currentAmount,
'comment': '\n'.join(comment) 'comment': '\n'.join(comment)
}] }]
return initialAmount, totalMonthlyDebit, totalMonthlyCredit, transactions, fileDatetime return initialAmount, totalMonthlyDebit, totalMonthlyCredit, transactions, fileDatetime
def printTransaction(transaction):
# , transaction['currentAmount']
print(transaction['date'], transaction['valeur'], transaction['amount'])
print(transaction['comment'])
print()