BNP_PDF_statement_parser/utils.py

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}')
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})')

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

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

def getDatetimeFromFileName(aDatetimeStr):
    #aDatetime = datetime.strptime(aDatetimeStr, '%Y%m%d.pdf')
    #aDatetime = datetime.strptime(aDatetimeStr[:6], '%Y%m')
    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')