Add WIP monthly debit and credit plot

bnp_pdf_statement_parser.py

@@ -1,10 +1,12 @@
-#!/usr/bin/python3
+#!/usr/bin/env python
 
 # Depends on `pdftotext`.
 
 import os
 import subprocess
 import re
+import matplotlib.pyplot as plt
+import matplotlib.ticker as ticker
 
 path = f'/home/benjamin/Desktop/bens_folder/bazaar/documents/bnp/bank_statements/'
 
@@ -27,12 +29,17 @@ def execute(command):
 def getTextFromPdf(pdfPath):
     return execute(['pdftotext', '-raw', pdfPath, '-'])
 
-firstLineOfPaymentRegex = re.compile('\d{2}\.\d{2} \d{2}\.\d{2} \d+,\d{2}')
+firstLineOfPaymentRegex = re.compile('\\d{2}\\.\\d{2} \\d{2}\\.\\d{2} \\d+,\\d{2}')
-endPageAfterTheFirstOneRegex = re.compile('P\. \d+/\d+')
+endPageAfterTheFirstOneRegex = re.compile('P\\. \\d+/\\d+')
-soldeCrediteurAuRegex = re.compile('SOLDE CREDITEUR AU \d{2}\.\d{2}\.\d{4}')
+soldeCrediteurAuRegex = re.compile('SOLDE CREDITEUR AU \\d{2}\\.\\d{2}\\.\\d{4}')
+totalDesOperationsRegex = re.compile('TOTAL\\ DES\\ OPERATIONS\\ ([0-9 ]+,\\d{2})\\ ([0-9 ]+,\\d{2})')
 
-for folder in os.listdir():
+PRINT_TRANSACTIONS = False
-    for file in os.listdir(folder):
+
+totalMonthlyDebits = []
+totalMonthlyCredits = []
+for folder in sorted(os.listdir()):
+    for file in sorted(os.listdir(folder)):
         #folder = '2022'
         #file = '20220321.pdf'
         filePath = f'{folder}/{file}'
@@ -64,9 +71,15 @@ for folder in os.listdir():
                     continue
                 # We aren't interested in the content after this line
                 elif line.startswith('TOTAL DES OPERATIONS'):
+                    totalDesOperationsRegexMatch = totalDesOperationsRegex.match(line)
+                    totalMonthlyDebit, totalMonthlyCredit = [float(group.replace(',', '.').replace(' ', '')) for group in totalDesOperationsRegexMatch.groups()]
+                    print(f'Total monthly debit: {totalMonthlyDebit}')
+                    print(f'Total monthly credit: {totalMonthlyCredit}')
+                    totalMonthlyDebits += [totalMonthlyDebit]
+                    totalMonthlyCredits += [totalMonthlyCredit]
                     break
                 if firstLineOfPaymentRegex.match(line) is not None:
-                    if date is not None:
+                    if date is not None and PRINT_TRANSACTIONS:
                         print(date, valeur, amount, currentAmount)
                         print('\n'.join(comment))
                         print()
@@ -76,6 +89,19 @@ for folder in os.listdir():
                     comment = []
                 else:
                     comment += [line]
-        break
+        #break
-    break
+    #break
 
+fig, ax = plt.subplots()
+plt.title('Monthly debits and credits')
+plt.xlabel('Date')
+plt.ylabel('€')
+ALPHA = 0.5
+xTicks = range(len(totalMonthlyDebits))
+plt.bar(xTicks, totalMonthlyDebits, alpha = ALPHA, label = 'Debit')
+plt.bar(xTicks, totalMonthlyCredits, alpha = ALPHA, label = 'Credit')
+plt.legend()
+
+ax.yaxis.set_major_formatter(ticker.StrMethodFormatter('{x:,}'))
+
+plt.show()