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.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.SYNDICATE_BANK_HEADER_REGEX)
file_content = bsr_utils.get_file_content(_file, password)
json_formatted_data = {
constants.TRANSACTIONS_STR: []
}
is_transaction_started = False
acc_details = ''
index = 0
if file_content == 'wrongpassword':
return 'wrongpassword'
elif file_content == 'pdfnotreadable':
return 'pdfnotreadable'
while index < len(file_content):
line = file_content[index]
if is_transaction_started:
process_transaction(json_formatted_data, line, constants.SYNDICATE_BANK_TRANSACTION_REGEX,
constants.FEDERAL_BANK_DESC_REGEX, constants.SYNDICATE_BANK_IGNORABLE_REGEXS)
elif header_pattern.match(line):
is_transaction_started = True
index += 1
bsr_utils.put_custum_acc_details(json_formatted_data, acc_details,
constants.SYNDICATE_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
index = index + 1
return json_formatted_data
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 extract(_file, password):
header_pattern = re.compile(constants.YES_BANK_HEADER_REGEX)
file_end_pattern = re.compile(constants.YES_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):
json_formatted_data.update({
constants.OPENING_BALANCE_STR:
file_end_pattern.match(line).group(
constants.OPENING_BALANCE_STR)
})
break
elif is_transaction_started:
process_transaction(json_formatted_data, line,
constants.YES_BANK_TRANSACTION_REGEX,
constants.YES_BANK_DESC_REGEX,
constants.YES_BANK_IGNORABLE_REGEXS)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(json_formatted_data, acc_details,
constants.YES_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
i += 1
return json_formatted_data
def extract(_file, password):
header_pattern = re.compile(constants.MAHARASHTRA_BANK_HEADER_REGEX)
file_end_pattern = re.compile(
constants.MAHARASHTRA_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 = ''
existing_desc = ''
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
if is_transaction_started:
existing_desc = process_transaction(
json_formatted_data, i, existing_desc,
constants.MAHARASHTRA_BANK_TRANSACTION_REGEX, file_content,
constants.MAHARASHTRA_BANK_DESC_REGEX,
constants.MAHARASHTRA_BANK_IGNORABLE_REGEXS)
elif header_pattern.search(line):
is_transaction_started = True
bsr_utils.put_acc_details(
json_formatted_data, acc_details,
constants.MAHARASHTRA_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
i += 1
return json_formatted_data
def extract(_file, password):
header_pattern = re.compile(constants.UNITED_BANK_HEADER_REGEX)
file_end_pattern = re.compile(constants.UNITED_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
if is_transaction_started:
process_transaction(json_formatted_data, line,
constants.UNITED_BANK_TRANSACTION_REGEX,
constants.FEDERAL_BANK_DESC_REGEX)
elif header_pattern.match(line):
is_transaction_started = True
i += 1
put_opening_balance(json_formatted_data, file_content[i])
bsr_utils.put_custum_acc_details(
json_formatted_data, acc_details,
constants.UNITED_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
i = i + 1
return json_formatted_data
def extract(_file, password):
header_pattern = re.compile(constants.SBI_BANK_HEADER_REGEX)
file_end_pattern = re.compile(constants.SBI_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.SBI_BANK_TRANSACTION_REGEX,
constants.SBI_BANK_DESC_REGEX,
constants.SBI_BANK_DESC_WITH_DATE_REGEX,
constants.SBI_BANK_IGNORABLE_REGEXS)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(json_formatted_data, acc_details,
constants.SBI_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
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.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 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:
process_transaction(json_formatted_data, line,
constants.BOB_BANK_TRANSACTION_REGEX)
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)
else:
acc_details += line + '\n'
return json_formatted_data
def extract(_file, password):
header_pattern = re.compile(business_constants.ICICI_BANK_HEADER_REGEX)
file_content = bsr_utils.get_file_content(_file, password)
json_formatted_data = {business_constants.TRANSACTIONS_STR: []}
is_transaction_started = False
acc_details = ''
i = 0
existing_desc = ''
if file_content == 'wrongpassword':
return 'wrongpassword'
elif file_content == 'pdfnotreadable':
return 'pdfnotreadable'
while i < len(file_content):
line = file_content[i]
if is_transaction_started:
existing_desc = process_transaction(
json_formatted_data, line, existing_desc,
business_constants.ICICI_BANK_TRANSACTION_REGEX,
business_constants.ICICI_BANK_DESC_REGEX,
business_constants.ICICI_BANK_IGNORABLE_REGEXS, True)
transaction_pattern = re.compile(
business_constants.ICICI_BANK_TRANSACTION_REGEX)
# there must be transaction and existing desc should be empty
if transaction_pattern and not existing_desc:
i = i + 1
line = file_content[i]
existing_desc = process_transaction(
json_formatted_data, line, existing_desc,
business_constants.ICICI_BANK_TRANSACTION_REGEX,
business_constants.ICICI_BANK_DESC_REGEX,
business_constants.ICICI_BANK_IGNORABLE_REGEXS, False)
elif header_pattern.match(line):
i += 2
is_transaction_started = True
bsr_utils.put_custum_acc_details(
json_formatted_data, acc_details,
business_constants.ICICI_BANK_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
i += 1
return json_formatted_data
def extract(_file, password):
header_pattern = re.compile(constants.INDIAN_OVERSEAS_BANK_HEADER_REGEX)
file_end_pattern = re.compile(constants.INDIAN_OVERSEAS_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 + 2
line = file_content[i]
if file_end_pattern.match(line):
break
elif is_transaction_started:
process_transaction(json_formatted_data, i, constants.INDIAN_OVERSEAS_BANK_TRANSACTION_REGEX, file_content,
constants.INDIAN_OVERSEAS_BANK_DESC_REGEX)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_custum_acc_details(json_formatted_data, acc_details,
constants.INDIAN_OVERSEAS_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 = ''
existing_desc = ''
if file_content == 'wrongpassword':
return 'wrongpassword'
elif file_content == 'pdfnotreadable':
return 'pdfnotreadable'
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):
pass
elif is_transaction_started:
existing_desc = process_transaction(json_formatted_data, line, existing_desc,
constants.HDFC_BANK_TWO_TRANSACTION_REGEX,
constants.HDFC_BANK_TWO_DESC_REGEX,
constants.HDFC_BANK_TWO_IGNORABLE_REGEXS)
elif header_pattern.match(line):
is_transaction_started = True
bsr_utils.put_acc_details(json_formatted_data, acc_details, constants.HDFC_BANK_TWO_ACCOUNT_DETAILS_REGEX)
else:
acc_details += line + '\n'
i += 1
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
