def process_transaction(date, json_formatted_data, line, transaction_regex,
date_change_regex, desc_regex, ignorable_regexes):
transaction_pattern = re.compile(transaction_regex)
date_change_pattern = re.compile(date_change_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 date_change_pattern.match(line):
m = date_change_pattern.match(line)
date = m.group(constants.DATE_STR)
if transaction_pattern.match(line):
m = 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(date),
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 constants.OPENING_BALANCE_STR not in json_formatted_data:
json_formatted_data[
constants.OPENING_BALANCE_STR] = bsr_utils.pretty_format(
m.group(constants.AMOUNT_STR))
elif 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(date),
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)
# print json_formatted_data
return date
def process_transaction(json_formatted_data, line, transaction_regex,
desc_regex):
transaction_pattern = re.compile(transaction_regex)
desc_pattern = re.compile(desc_regex)
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 desc_pattern.match(line):
process_desc(json_formatted_data, desc_pattern, line)
def extract(_file, password):
header_pattern = re.compile(constants.BANDHAN_BANK_HEADER_REGEX)
file_content = bsr_utils.get_file_content(_file, password)
json_formatted_data = {
constants.TRANSACTIONS_STR: []
}
is_transaction_started = False
acc_details = ''
i = 0
existing_desc = ''
while i < len(file_content):
line = file_content[i]
if is_transaction_started:
existing_desc = process_transaction(json_formatted_data, line, existing_desc,
constants.BANDHAN_BANK_TRANSACTION_REGEX_TWO,
constants.BANDHAN_BANK_DESC_REGEX)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(json_formatted_data, acc_details,
constants.BANDHAN_BANK_ACCOUNT_DETAILS_REGEX_TWO)
else:
acc_details += line + '\n'
i = i + 1
j = len(json_formatted_data[constants.TRANSACTIONS_STR]) - 1
while j >= 0:
opening_balance = bsr_utils.get_rev_opening_balance(j, json_formatted_data)
transaction_type = bsr_utils.get_transaction_type(opening_balance,
json_formatted_data[constants.TRANSACTIONS_STR][j][
constants.CLOSING_BALANCE_STR])
json_formatted_data[constants.TRANSACTIONS_STR][j][constants.TRANSACTION_TYPE_STR] = transaction_type
j -= 1
return json_formatted_data
def process_transaction(json_formatted_data, line, existing_desc,
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:
existing_desc +
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))
})
existing_desc = ''
elif bsr_utils.is_ignorable(ignorable_patterns, line):
pass
elif desc_pattern.match(line):
existing_desc = process_desc(desc_pattern, line, existing_desc)
return existing_desc
def process_transaction(json_formatted_data, line, existing_desc, 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)
withdraw = float(str(bsr_utils.pretty_format(m.group(constants.WITHDRAW_AMOUNT_STR))).replace(',', ''))
deposit = float(str(bsr_utils.pretty_format(m.group(constants.DEPOSIT_AMOUNT_STR)).replace(',', '')))
amount = withdraw + deposit
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: existing_desc + 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))
})
existing_desc = ''
elif bsr_utils.is_ignorable(ignorable_patterns, line):
pass
elif desc_pattern.match(line):
existing_desc = process_desc_custum(desc_pattern, line, existing_desc)
return existing_desc
def extract(_file, password):
header_pattern = re.compile(constants.CANARA_BANK_HEADER_REGEX)
file_content = bsr_utils.get_file_content(_file, password)
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 is_transaction_started:
process_transaction(json_formatted_data, line,
constants.CANARA_BANK_TRANSACTION_REGEX,
constants.CANARA_BANK_DESC_REGEX,
constants.CANARA_BANK_IGNORABLE_REGEXS)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(
json_formatted_data, acc_details,
constants.CANARA_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
j = len(json_formatted_data[constants.TRANSACTIONS_STR]) - 1
while j >= 0:
opening_balance = bsr_utils.get_rev_opening_balance(
j, json_formatted_data)
transaction_type = bsr_utils.get_transaction_type(
opening_balance, json_formatted_data[constants.TRANSACTIONS_STR][j]
[constants.CLOSING_BALANCE_STR])
json_formatted_data[constants.TRANSACTIONS_STR][j][
constants.TRANSACTION_TYPE_STR] = transaction_type
j -= 1
return json_formatted_data
def extract(_file, password):
header_pattern = re.compile(constants.KOTAK_BANK_HEADER_REGEX_TWO)
file_end_pattern = re.compile(constants.KOTAK_BANK_STATEMENT_END_REGEX_TWO)
file_content = bsr_utils.get_file_content(_file, password)
json_formatted_data = {constants.TRANSACTIONS_STR: []}
is_transaction_started = False
acc_details = ''
line_index = len(file_content) - 1
while line_index > 0:
line = file_content[line_index]
if file_end_pattern.match(line):
line = file_content[line_index]
json_formatted_data.update(
{constants.OPENING_BALANCE_STR: line.split()[-1]})
break
line_index -= 1
line_index = 0
if file_content == 'wrongpassword':
return 'wrongpassword'
elif file_content == 'pdfnotreadable':
return 'pdfnotreadable'
while line_index < len(file_content):
line = file_content[line_index]
if file_end_pattern.match(line):
break
elif is_transaction_started:
process_transaction(json_formatted_data, line,
constants.KOTAK_BANK_TRANSACTION_REGEX_TWO,
constants.KOTAK_BANK_DESC_REGEX,
constants.KOTAK_BANK_IGNORABLE_REGEXS_TWO)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(
json_formatted_data, acc_details,
constants.KOTAK_BANK_ACCOUNT_DETAILS_REGEX_TWO)
line_index -= 1
else:
acc_details += line + '\n'
line_index += 1
j = len(json_formatted_data[constants.TRANSACTIONS_STR]) - 1
while j >= 0:
opening_balance = bsr_utils.get_rev_opening_balance(
j, json_formatted_data)
transaction_type = bsr_utils.get_transaction_type(
opening_balance, json_formatted_data[constants.TRANSACTIONS_STR][j]
[constants.CLOSING_BALANCE_STR])
json_formatted_data[constants.TRANSACTIONS_STR][j][
constants.TRANSACTION_TYPE_STR] = transaction_type
j -= 1
return json_formatted_data
def extract(_file, password):
header_pattern = re.compile(constants.RBL_BANK_HEADER_REGEX)
file_end_pattern = re.compile(constants.RBL_BANK_STATEMENT_END_REGEX)
file_content = bsr_utils.get_file_content(_file, password)
json_formatted_data = {constants.TRANSACTIONS_STR: []}
is_transaction_started = False
acc_details = ''
i = len(file_content) - 1
while i > 0:
if 'Opening Balance:' in file_content[i]:
json_formatted_data[
constants.OPENING_BALANCE_STR] = file_content[i].split()[3]
break
i -= 1
i = 0
if file_content == 'wrongpassword':
return 'wrongpassword'
elif file_content == 'pdfnotreadable':
return 'pdfnotreadable'
while i < len(file_content):
line = file_content[i]
if file_end_pattern.match(line):
break
elif is_transaction_started:
process_transaction(json_formatted_data, i, file_content,
constants.RBL_BANK_TRANSACTION_REGEX_1,
constants.RBL_BANK_TRANSACTION_REGEX_2,
constants.RBL_BANK_DESC_REGEX,
constants.RBL_BANK_IGNORABLE_REGEXS)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(json_formatted_data, acc_details,
constants.RBL_BANK_ACCOUNT_DETAILS_REGEX)
if "name" not in json_formatted_data:
bsr_utils.put_acc_details(
json_formatted_data, acc_details,
constants.RBL_BANK_ACCOUNT_DETAILS_REGEX_TWO)
else:
acc_details += line + '\n'
i += 1
j = len(json_formatted_data[constants.TRANSACTIONS_STR]) - 1
while j >= 0:
opening_balance = bsr_utils.get_rev_opening_balance(
j, json_formatted_data)
transaction_type = bsr_utils.get_transaction_type(
opening_balance, json_formatted_data[constants.TRANSACTIONS_STR][j]
[constants.CLOSING_BALANCE_STR])
json_formatted_data[constants.TRANSACTIONS_STR][j][
constants.TRANSACTION_TYPE_STR] = transaction_type
j -= 1
return json_formatted_data
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.PNB_BANK_HEADER_REGEX)
file_end_pattern = re.compile(constants.PNB_BANK_STATEMENT_END_REGEX)
file_content = bsr_utils.get_file_content(_file, password)
json_formatted_data = {
constants.TRANSACTIONS_STR: []
}
is_transaction_started = False
acc_details = ''
i = 0
if file_content == 'wrongpassword':
return 'wrongpassword'
elif file_content == 'pdfnotreadable':
return 'pdfnotreadable'
while i < len(file_content):
line = file_content[i]
if file_end_pattern.match(line):
break
elif is_transaction_started:
process_transaction(json_formatted_data, i, constants.PNB_BANK_TRANSACTION_REGEX, file_content)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(json_formatted_data, acc_details, constants.PNB_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
i += 1
j = len(json_formatted_data[constants.TRANSACTIONS_STR]) - 2
while j >= 0:
opening_balance = bsr_utils.get_rev_opening_balance(j, json_formatted_data)
transaction_type = bsr_utils.get_transaction_type(opening_balance,
json_formatted_data[constants.TRANSACTIONS_STR][j][
constants.CLOSING_BALANCE_STR])
json_formatted_data[constants.TRANSACTIONS_STR][j][constants.TRANSACTION_TYPE_STR] = transaction_type
j -= 1
json_formatted_data[constants.NAME_STR] = json_formatted_data[constants.NAME_STR].split(' SO ')[0]
return json_formatted_data
