Types coerced to string

[GPHemsley-RELX]

utils.parse_type() coerces certain types to string, losing the ability to manipulate or reformat native Python types.

access_parser/access_parser/utils.py

Lines 141 to 196 in 7b73391

	def parse_type(data_type, buffer, length=None, version=3, props=None):
	parsed = ""
	# Bool or int8
	if data_type == TYPE_INT8:
	parsed = struct.unpack_from("b", buffer)[0]
	elif data_type == TYPE_INT16:
	parsed = struct.unpack_from("h", buffer)[0]
	elif data_type == TYPE_INT32 or data_type == TYPE_COMPLEX:
	parsed = struct.unpack_from("i", buffer)[0]
	elif data_type == TYPE_MONEY:
	parsed = struct.unpack_from("q", buffer)[0]
	if props and "Format" in props:
	prop_format = props['Format']
	if parsed == 0:
	parsed = [y for x, y in FORMAT_TO_DEFAULT_VALUE.items() if prop_format.startswith(x)]
	if not parsed:
	LOGGER.warning(f"parse_type got unknown format while parsing money field {prop_format}")
	else:
	parsed = parsed[0]
	else:
	parsed = parse_money_type(parsed, prop_format)
	elif data_type == TYPE_FLOAT32:
	parsed = struct.unpack_from("f", buffer)[0]
	elif data_type == TYPE_FLOAT64:
	parsed = struct.unpack_from("d", buffer)[0]
	elif data_type == TYPE_DATETIME:
	double_datetime = struct.unpack_from("q", buffer)[0]
	parsed = mdb_date_to_readable(double_datetime)
	elif data_type == TYPE_BINARY:
	parsed = buffer[:length]
	offset = length
	elif data_type == TYPE_OLE:
	parsed = buffer
	elif data_type == TYPE_GUID:
	parsed = buffer[:16]
	guid = uuid.UUID(parsed.hex())
	parsed = str(guid)
	elif data_type == TYPE_96_bit_17_BYTES:
	parsed = buffer[:17]
	elif data_type == TYPE_TEXT:
	if version > 3:
	# Looks like if BOM is present text is already decoded
	if buffer.startswith(b"\xfe\xff") or buffer.startswith(b"\xff\xfe"):
	buff = buffer[2:]
	parsed = get_decoded_text(buff)
	else:
	parsed = buffer.decode("utf-16", errors='ignore')
	else:
	parsed = get_decoded_text(buffer)
	if "\x00" in parsed:
	LOGGER.debug(f"Parsed string contains NUL (0x00) characters: {parsed}")
	parsed = parsed.replace("\x00", "")
	else:
	LOGGER.debug(f"parse_type - unsupported data type: {data_type}")
	return parsed

This especially affects datetime, which is further processed by utils.mdb_date_to_readable():

access_parser/access_parser/utils.py

Lines 59 to 73 in 7b73391

	# https://stackoverflow.com/questions/45560782
	def mdb_date_to_readable(double_time):
	try:
	dtime_bytes = struct.pack("Q", double_time)
	dtime_double = struct.unpack('<d', dtime_bytes)[0]
	dtime_frac, dtime_whole = math.modf(dtime_double)
	dtime = (ACCESS_EPOCH + timedelta(days=dtime_whole) + timedelta(days=dtime_frac))
	if dtime == ACCESS_EPOCH:
	return "(Empty Date)"
	return str(dtime)
	except OverflowError:
	return "(Invalid Date)"
	except struct.error:
	return "(Invalid Date)"