def process_transaction(json_formatted_data, line, transaction_regex, desc_regex, ignorable_regexes): transaction_pattern = re.compile(transaction_regex) desc_pattern = re.compile(desc_regex) ignorable_patterns = [ re.compile(ignorable_regex) for ignorable_regex in ignorable_regexes ] m = transaction_pattern.match(line) if transaction_pattern.match(line): withdraw = float( bsr_utils.pretty_format(m.group( constants.WITHDRAW_AMOUNT_STR)).replace(',', '')) deposit = float( bsr_utils.pretty_format(m.group( constants.DEPOSIT_AMOUNT_STR)).replace(',', '')) amount = withdraw + deposit transaction_type = get_transaction(deposit) json_formatted_data[constants.TRANSACTIONS_STR].append({ constants.DATE_STR: bsr_utils.pretty_format(m.group(constants.DATE_STR)), constants.DESCRIPTION_STR: bsr_utils.pretty_format(m.group(constants.DESCRIPTION_STR)), constants.TYPE_STR: transaction_type, constants.AMOUNT_STR: amount, constants.CLOSING_BALANCE_STR: bsr_utils.pretty_format(m.group(constants.CLOSING_BALANCE_STR)) }) elif bsr_utils.is_ignorable(ignorable_patterns, line): pass elif desc_pattern.match(line): process_desc(json_formatted_data, desc_pattern, line)
def process_transaction(json_formatted_data, line, transaction_regex, desc_regex, ignorable_regexes): transaction_pattern = re.compile(transaction_regex) desc_pattern = re.compile(desc_regex) ignorable_patterns = [ re.compile(ignorable_regex) for ignorable_regex in ignorable_regexes ] m = transaction_pattern.match(line) if transaction_pattern.match(line): opening_balance = bsr_utils.get_opening_balance(json_formatted_data) transaction_type = bsr_utils.get_transaction_type( opening_balance, bsr_utils.pretty_format(m.group(constants.CLOSING_BALANCE_STR))) json_formatted_data[constants.TRANSACTIONS_STR].append({ constants.DATE_STR: bsr_utils.pretty_format(m.group(constants.DATE_STR)), constants.DESCRIPTION_STR: bsr_utils.pretty_format(m.group(constants.DESCRIPTION_STR)), constants.TYPE_STR: transaction_type, constants.AMOUNT_STR: bsr_utils.pretty_format(m.group(constants.AMOUNT_STR)), constants.CLOSING_BALANCE_STR: bsr_utils.pretty_format(m.group(constants.CLOSING_BALANCE_STR)) }) elif bsr_utils.is_ignorable(ignorable_patterns, line): pass elif desc_pattern.match(line): process_desc(json_formatted_data, desc_pattern, line)
def extract(_file, password): header_pattern = re.compile(constants.SCH_BANK_HEADER_REGEX) file_end_patterns = [ re.compile(regex) for regex in constants.SCH_BANK_STATEMENT_END_REGEX ] file_content = bsr_utils.get_file_content(_file, password) date = None json_formatted_data = {constants.TRANSACTIONS_STR: []} is_transaction_started = False acc_details = '' if file_content == 'wrongpassword': return 'wrongpassword' elif file_content == 'pdfnotreadable': return 'pdfnotreadable' for line in file_content: if bsr_utils.is_ignorable(file_end_patterns, line): is_transaction_started = False elif is_transaction_started: date = process_transaction(date, json_formatted_data, line, constants.SCH_BANK_TRANSACTION_REGEX, constants.SCH_BANK_DATE_CHANGE_REGEX, constants.SCH_BANK_DESC_REGEX, constants.SCH_BANK_IGNORABLE_REGEXS) elif header_pattern.match(line): is_transaction_started = True if len(json_formatted_data[constants.TRANSACTIONS_STR]) == 0: bsr_utils.put_acc_details( json_formatted_data, acc_details, constants.SCH_BANK_ACCOUNT_DETAILS_REGEX) else: acc_details += line + '\n' return json_formatted_data
def process_transaction(json_formatted_data, line, transaction_regex, desc_regex, ignorable_regexes): transaction_pattern = re.compile(transaction_regex) desc_pattern = re.compile(desc_regex) ignorable_patterns = [ re.compile(ignorable_regex) for ignorable_regex in ignorable_regexes ] m = transaction_pattern.match(line) if transaction_pattern.match(line): json_formatted_data[business_constants.TRANSACTIONS_STR].append({ business_constants.DATE_STR: bsr_utils.pretty_format(m.group(business_constants.DATE_STR)), business_constants.DESCRIPTION_STR: bsr_utils.pretty_format(m.group( business_constants.DESCRIPTION_STR)), business_constants.AMOUNT_STR: bsr_utils.pretty_format(m.group(business_constants.AMOUNT_STR)), business_constants.CLOSING_BALANCE_STR: bsr_utils.pretty_format( m.group(business_constants.CLOSING_BALANCE_STR)) }) elif bsr_utils.is_ignorable(ignorable_patterns, line): pass elif desc_pattern.match(line): process_desc(json_formatted_data, desc_pattern, line)
def process_transaction(json_formatted_data, i, existing_desc, transaction_regex, file_content, desc_regex, ignorable_regexes): line = file_content[i] return_statement = False Description_value = '' transaction_pattern = re.compile(transaction_regex) desc_pattern = re.compile(desc_regex) ignorable_patterns = [ re.compile(ignorable_regex) for ignorable_regex in ignorable_regexes ] m = transaction_pattern.match(line) if transaction_pattern.match(line): if (len(bsr_utils.pretty_format(m.group(constants.DATE_STR))) > 0) and len( bsr_utils.pretty_format(m.group( constants.DESCRIPTION_STR))) > 0: Description_value = bsr_utils.pretty_format( m.group(constants.DESCRIPTION_STR)) else: Description_value = '' return_statement = True opening_balance = bsr_utils.get_opening_balance(json_formatted_data) transaction_type = bsr_utils.get_transaction_type( opening_balance, bsr_utils.pretty_format(m.group(constants.CLOSING_BALANCE_STR))) json_formatted_data[constants.TRANSACTIONS_STR].append({ constants.DATE_STR: bsr_utils.pretty_format(m.group(constants.DATE_STR)), # constants.DESCRIPTION_STR: bsr_utils.pretty_format(m.group(constants.DESCRIPTION_STR)), constants.DESCRIPTION_STR: Description_value, constants.TYPE_STR: transaction_type, constants.AMOUNT_STR: bsr_utils.pretty_format(m.group(constants.AMOUNT_STR)), constants.CLOSING_BALANCE_STR: bsr_utils.pretty_format(m.group(constants.CLOSING_BALANCE_STR)) }) if return_statement: return existing_desc elif bsr_utils.is_ignorable(ignorable_patterns, line): pass elif desc_pattern.match(line): if existing_desc is None or existing_desc == '': existing_desc = existing_desc + line else: process_desc_custum(json_formatted_data, desc_pattern, line, existing_desc) existing_desc = '' return existing_desc
def extract(_file, password): header_pattern = re.compile(constants.INDUSLAND_BANK_HEADER_REGEX) file_end_pattern = re.compile(constants.INDUSLAND_BANK_STATEMENT_END_REGEX) file_content = bsr_utils.get_file_content(_file, password) json_formatted_data = {constants.TRANSACTIONS_STR: []} is_transaction_started = False transaction_flag = False acc_details = '' transaction_pattern = re.compile( constants.INDUSLAND_BANK_TRANSACTION_REGEX) ignorable_patterns = [ re.compile(ignorable_regex) for ignorable_regex in constants.INDUSLAND_BANK_IGNORABLE_REGEXS ] if file_content == 'wrongpassword': return 'wrongpassword' elif file_content == 'pdfnotreadable': return 'pdfnotreadable' for line in file_content: if file_end_pattern.match(line): break if bsr_utils.is_ignorable(ignorable_patterns, line): transaction_flag = False continue elif is_transaction_started and transaction_flag: process_transaction(json_formatted_data, line, constants.INDUSLAND_BANK_TRANSACTION_REGEX, constants.INDUSLAND_BANK_DESC_REGEX, constants.INDUSLAND_BANK_IGNORABLE_REGEXS) elif is_transaction_started: if transaction_pattern.match(line): transaction_flag = True process_transaction(json_formatted_data, line, constants.INDUSLAND_BANK_TRANSACTION_REGEX, constants.INDUSLAND_BANK_DESC_REGEX, constants.INDUSLAND_BANK_IGNORABLE_REGEXS) elif header_pattern.match(line): is_transaction_started = True bsr_utils.put_acc_details( json_formatted_data, acc_details, constants.INDUSLAND_BANK_ACCOUNT_DETAILS_REGEX) else: acc_details += line + '\n' return json_formatted_data
def process_transaction(json_formatted_data, i, file_content, transaction_regex_1, transaction_regex_2, desc_regex, ignorable_regexes): line = file_content[i] transaction_pattern_1 = re.compile(transaction_regex_1) transaction_pattern_2 = re.compile(transaction_regex_2) desc_pattern = re.compile(desc_regex) ignorable_patterns = [ re.compile(ignorable_regex) for ignorable_regex in ignorable_regexes ] m = transaction_pattern_1.match(line) if transaction_pattern_1.match(line): json_formatted_data[constants.TRANSACTIONS_STR].append({ constants.DATE_STR: bsr_utils.pretty_format(m.group(constants.DATE_STR)), constants.DESCRIPTION_STR: bsr_utils.pretty_format(m.group(constants.DESCRIPTION_STR)), constants.AMOUNT_STR: bsr_utils.pretty_format(m.group(constants.AMOUNT_STR)), constants.CLOSING_BALANCE_STR: bsr_utils.pretty_format(m.group(constants.CLOSING_BALANCE_STR)) }) elif transaction_pattern_2.match(line): m = transaction_pattern_2.match(line) json_formatted_data[constants.TRANSACTIONS_STR].append({ constants.DATE_STR: bsr_utils.pretty_format(m.group(constants.DATE_STR)), constants.DESCRIPTION_STR: bsr_utils.pretty_format(file_content[i - 1]) + ' ' + bsr_utils.pretty_format(m.group(constants.DESCRIPTION_STR)), constants.AMOUNT_STR: bsr_utils.pretty_format(m.group(constants.AMOUNT_STR)), constants.CLOSING_BALANCE_STR: bsr_utils.pretty_format(m.group(constants.CLOSING_BALANCE_STR)) }) elif bsr_utils.is_ignorable(ignorable_patterns, line): pass elif desc_pattern.match(line): process_desc(json_formatted_data, desc_pattern, line)
def process_transaction(json_formatted_data, line, existing_desc, transaction_regex, desc_regex, ignorable_regexes, append_down): transaction_pattern = re.compile(transaction_regex) desc_pattern = re.compile(desc_regex) ignorable_patterns = [ re.compile(ignorable_regex) for ignorable_regex in ignorable_regexes ] m = transaction_pattern.search(line) if transaction_pattern.search(line): transaction_type = get_transaction_type( bsr_utils.pretty_format( m.group(business_constants.TRANSACTION_TYPE_STR))) json_formatted_data[business_constants.TRANSACTIONS_STR].append({ business_constants.DATE_STR: bsr_utils.pretty_format(m.group(business_constants.DATE_STR)), business_constants.DESCRIPTION_STR: existing_desc + bsr_utils.pretty_format( m.group(business_constants.DESCRIPTION_STR)), business_constants.TYPE_STR: transaction_type, business_constants.AMOUNT_STR: bsr_utils.pretty_format(m.group(business_constants.AMOUNT_STR)), business_constants.CLOSING_BALANCE_STR: bsr_utils.pretty_format( m.group(business_constants.CLOSING_BALANCE_STR)) }) existing_desc = '' elif bsr_utils.is_ignorable(ignorable_patterns, line): pass elif desc_pattern.search(line): if append_down: existing_desc = process_desc(desc_pattern, line, existing_desc) else: process_desc_down(json_formatted_data, desc_pattern, line) return existing_desc