def extract(_file, password):
header_pattern = re.compile(constants.BOI_BANK_HEADER_REGEX)
file_end_pattern = re.compile(constants.BOI_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 = ''
if file_content == 'wrongpassword':
return 'wrongpassword'
elif file_content == 'pdfnotreadable':
return 'pdfnotreadable'
for line in file_content:
if file_end_pattern.match(line):
break
elif is_transaction_started:
process_transaction(json_formatted_data, line,
constants.BOI_BANK_TRANSACTION_REGEX,
constants.BOI_BANK_DESC_REGEX,
constants.BOI_BANK_IGNORABLE_REGEXS)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(json_formatted_data, acc_details,
constants.BOI_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
if "name" in json_formatted_data:
json_formatted_data[constants.NAME_STR] = json_formatted_data[
constants.NAME_STR].split(' S/O ')[0]
return json_formatted_data
def extract(_file, password):
header_pattern = re.compile(constants.CORPORATION_BANK_HEADER_REGEX)
file_end_pattern = re.compile(constants.CORPORATION_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, line, constants.CORPORATION_BANK_TRANSACTION_REGEX,
constants.CORPORATION_BANK_DESC_REGEX, constants.CORPORATION_BANK_IGNORABLE_REGEXS)
elif header_pattern.match(line):
is_transaction_started = True
i -= 1
bsr_utils.put_acc_details(json_formatted_data, acc_details,
constants.CORPORATION_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
i += 1
return json_formatted_data
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.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 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 extract(_file, password):
header_pattern = re.compile(constants.BOB_BANK_HEADER_REGEX)
file_content = bsr_utils.get_file_content(_file, password)
json_formatted_data = {constants.TRANSACTIONS_STR: []}
transaction_pattern = re.compile(constants.BOB_BANK_TRANSACTION_REGEX)
credit_transaction_pattern = re.compile(
constants.BOB_BANK_CREDIT_TRANSACTION_REGEX)
if file_content == 'wrongpassword':
return 'wrongpassword'
elif file_content == 'pdfnotreadable':
return 'pdfnotreadable'
for line in file_content:
if transaction_pattern.match(line):
m = transaction_pattern.match(line)
if credit_transaction_pattern.match(line):
json_formatted_data[
constants.OPENING_BALANCE_STR] = bsr_utils.get_opening_bal(
bsr_utils.pretty_format(
m.group(constants.CLOSING_BALANCE_STR)),
bsr_utils.pretty_format(m.group(constants.AMOUNT_STR)),
constants.DEPOSIT_TYPE)
else:
json_formatted_data[
constants.OPENING_BALANCE_STR] = bsr_utils.get_opening_bal(
bsr_utils.pretty_format(
m.group(constants.CLOSING_BALANCE_STR)),
bsr_utils.pretty_format(m.group(constants.AMOUNT_STR)),
constants.WITHDRAW_TYPE)
break
is_transaction_started = False
acc_details = ''
for line in file_content:
if is_transaction_started:
if "Opening Balance" in line:
put_opening_balance(json_formatted_data, line)
process_transaction(json_formatted_data, line,
constants.BOB_BANK_TRANSACTION_REGEX_TWO,
constants.BOB_BANK_IGNORABLE_REGEXS)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(
json_formatted_data, acc_details,
constants.BOB_BANK_ACCOUNT_DETAILS_REGEX_TWO)
else:
acc_details += line + '\n'
return json_formatted_data
def extract(_file, password):
header_pattern = re.compile(constants.CENTRAL_BANK_HEADER_REGEX)
file_end_pattern = re.compile(constants.CENTRAL_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):
if i == 0:
i = i + 1
line = file_content[i]
if file_end_pattern.match(line):
break
elif is_transaction_started:
process_transaction(json_formatted_data, i,
constants.CENTRAL_BANK_TRANSACTION_REGEX,
file_content,
constants.CENTRAL_BANK_DESC_REGEX)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(
json_formatted_data, acc_details,
constants.CENTRAL_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
i = 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_custom_transaction_type(
opening_balance, json_formatted_data[constants.TRANSACTIONS_STR][j]
[constants.CLOSING_BALANCE_STR])
json_formatted_data[constants.TRANSACTIONS_STR][j + 1][
constants.TRANSACTION_TYPE_STR] = transaction_type
j -= 1
return json_formatted_data
def extract(_file, password):
header_pattern = re.compile(constants.HDFC_BANK_HEADER_REGEX)
file_end_pattern = re.compile(constants.HDFC_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 = ''
if file_content == 'wrongpassword':
return 'wrongpassword'
elif file_content == 'pdfnotreadable':
return 'pdfnotreadable'
try:
i = len(file_content) - 1
while i > 0:
line = file_content[i]
if file_end_pattern.match(line):
line = file_content[i + 1]
json_formatted_data.update(
{constants.OPENING_BALANCE_STR: line.split()[0]})
break
i -= 1
i = 0
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, line,
constants.HDFC_BANK_TRANSACTION_REGEX,
constants.HDFC_BANK_DESC_REGEX,
constants.HDFC_BANK_IGNORABLE_REGEXS)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(
json_formatted_data, acc_details,
constants.HDFC_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
i += 1
except:
return 'pdfnotreadable'
return json_formatted_data
def extract(file, password):
header_pattern = re.compile(constants.BOI_BANK_HEADER_REGEX_TWO)
file_end_pattern = re.compile(constants.BOI_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 = ''
for line in file_content:
if file_end_pattern.match(line):
break
elif is_transaction_started:
process_transaction(json_formatted_data, line, constants.BOI_BANK_TRANSACTION_REGEX_TWO,
constants.BOI_BANK_DESC_REGEX)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(json_formatted_data, acc_details, constants.BOI_BANK_ACCOUNT_DETAILS_REGEX_TWO)
else:
acc_details += line + '\n'
return json_formatted_data
def extract(file, header_regex, file_end_regex, account_regex,
transaction_regex, desc_regex, ignorable_regexes):
header_pattern = re.compile(header_regex)
file_end_pattern = re.compile(file_end_regex)
file_content = bsr_utils.get_file_content(file)
json_formatted_data = {constants.TRANSACTIONS_STR: []}
is_transaction_started = False
acc_details = ''
for line in file_content:
if file_end_pattern.match(line):
break
elif is_transaction_started:
process_transaction(json_formatted_data, line,
transaction_regex, desc_regex,
ignorable_regexes)
elif header_pattern.match(line):
is_transaction_started = True
put_acc_details(json_formatted_data, acc_details,
account_regex)
else:
acc_details += line + '\n'
return json_formatted_data
