BNP_PDF_statement_parser/bnp_pdf_statement_parser.py

79 lines
2.4 KiB
Python
Raw Normal View History

2024-10-01 20:35:43 +02:00
#!/usr/bin/env python
2023-06-21 00:36:27 +02:00
2023-11-05 21:10:45 +01:00
import os
2024-10-01 20:35:43 +02:00
import matplotlib.pyplot as plt
import matplotlib.ticker as ticker
2024-10-01 20:57:36 +02:00
from datetime import datetime
2024-10-03 19:51:54 +02:00
from utils import getDatetimeFromFileName, getMonthIndexSinceEpoch, getMonthNameFromMonthIndex, readPdfBankStatement
2023-06-21 00:36:27 +02:00
2024-10-03 19:33:54 +02:00
PATH = f'/home/benjamin/Desktop/bens_folder/bazaar/documents/bnp/bank_statements/'
2023-06-21 00:36:27 +02:00
2024-10-03 17:43:44 +02:00
os.chdir(PATH)
2023-06-21 00:36:27 +02:00
PRINT_TRANSACTIONS = False
2024-10-01 20:35:43 +02:00
2024-10-03 19:33:54 +02:00
MAIN_BANK_ACCOUNT = 'compte_de_cheques'
2024-10-03 21:22:30 +02:00
debits = []
credits_ = []
# Precise bank account
allTransactions = []
2024-10-03 19:33:54 +02:00
2024-10-03 21:22:30 +02:00
def appendTransactions()
for folder in os.listdir():
2024-10-03 19:33:54 +02:00
if folder != MAIN_BANK_ACCOUNT:
print(folder)
2024-10-03 21:22:30 +02:00
for file in os.listdir(folder):
2024-10-03 19:33:54 +02:00
print(file)
2024-10-03 20:10:06 +02:00
filePath = f'{folder}/{file}'
2024-10-03 21:22:30 +02:00
_initialAmount, _totalMonthlyDebit, _totalMonthlyCredit, transactions, fileDatetime = readPdfBankStatement(filePath)
for transaction in transactions:
transaction['bank account'] = folder
allTransactions += [transaction]
2024-10-03 19:33:54 +02:00
2024-10-03 20:10:06 +02:00
#exit(1)
2024-10-03 19:33:54 +02:00
os.chdir(f'{MAIN_BANK_ACCOUNT}/')
2024-10-01 20:35:43 +02:00
for folder in sorted(os.listdir()):
for file in sorted(os.listdir(folder)):
2023-06-21 00:36:27 +02:00
filePath = f'{folder}/{file}'
print(filePath)
2024-10-03 19:51:54 +02:00
initialAmount, totalMonthlyDebit, totalMonthlyCredit, transactions, fileDatetime = readPdfBankStatement(filePath)
2024-10-03 21:22:30 +02:00
for transaction in transactions:
transaction['bank account'] = MAIN_BANK_ACCOUNT
allTransactions += [transaction]
2024-10-03 19:51:54 +02:00
if PRINT_TRANSACTIONS:
for transaction in transactions:
2024-10-03 21:22:30 +02:00
printTransaction(transaction)
2024-10-01 20:35:43 +02:00
#break
#break
lastDatetime = getDatetimeFromFileName(file)
2024-10-01 20:35:43 +02:00
fig, ax = plt.subplots()
plt.title('Monthly debits and credits')
plt.xlabel('Date')
plt.ylabel('')
ALPHA = 0.5
2024-10-01 20:57:36 +02:00
xTicks = range(getMonthIndexSinceEpoch(firstDatetime), getMonthIndexSinceEpoch(lastDatetime) + 1)
2024-10-02 00:33:11 +02:00
totalMonthlyAmountAndLabel = (
#(totalMonthlyDebits, 'Debit'),
#(totalMonthlyCredits, 'Credit'),
(totalMonthlyDifferences, 'Difference'),
(totals, 'Total'),
)
for totalMonthlyAmount, totalMonthlyLabel in totalMonthlyAmountAndLabel:
plt.bar(xTicks, totalMonthlyAmount, alpha = ALPHA, label = totalMonthlyLabel)
2024-10-01 20:35:43 +02:00
plt.legend()
2024-10-02 00:33:11 +02:00
#plt.yscale('symlog')
2024-10-01 20:35:43 +02:00
ax.yaxis.set_major_formatter(ticker.StrMethodFormatter('{x:,}'))
2023-06-21 00:36:27 +02:00
ticksLabels = [getMonthNameFromMonthIndex(monthIndex) for monthIndex in xTicks]
2024-10-01 20:57:36 +02:00
plt.xticks(xTicks, ticksLabels, rotation = 90)
#plt.tight_layout()
# How to show the horizontal lines for subticks?
plt.grid(axis = 'y')
2024-10-01 20:35:43 +02:00
plt.show()