lbry-sdk/lbry/dht/serialization/bencoding.py

76 lines
2.8 KiB
Python
Raw Normal View History

2019-01-22 18:49:43 +01:00
import typing
2019-06-21 02:55:47 +02:00
from lbry.dht.error import DecodeError
2017-03-31 19:32:43 +02:00
2016-12-14 00:08:29 +01:00
2019-01-22 18:49:43 +01:00
def _bencode(data: typing.Union[int, bytes, bytearray, str, list, tuple, dict]) -> bytes:
if isinstance(data, int):
return b'i%de' % data
elif isinstance(data, (bytes, bytearray)):
return b'%d:%s' % (len(data), data)
elif isinstance(data, str):
return b'%d:%s' % (len(data), data.encode())
elif isinstance(data, (list, tuple)):
encoded_list_items = b''
for item in data:
2019-01-22 18:49:43 +01:00
encoded_list_items += _bencode(item)
return b'l%se' % encoded_list_items
elif isinstance(data, dict):
encoded_dict_items = b''
keys = data.keys()
for key in sorted(keys):
2019-01-22 18:49:43 +01:00
encoded_dict_items += _bencode(key)
encoded_dict_items += _bencode(data[key])
return b'd%se' % encoded_dict_items
else:
2019-01-22 18:49:43 +01:00
raise TypeError(f"Cannot bencode {type(data)}")
2019-01-22 18:49:43 +01:00
def _bdecode(data: bytes, start_index: int = 0) -> typing.Tuple[typing.Union[int, bytes, list, tuple, dict], int]:
if data[start_index] == ord('i'):
end_pos = data[start_index:].find(b'e') + start_index
return int(data[start_index + 1:end_pos]), end_pos + 1
elif data[start_index] == ord('l'):
start_index += 1
decoded_list = []
while data[start_index] != ord('e'):
2019-01-22 18:49:43 +01:00
list_data, start_index = _bdecode(data, start_index)
decoded_list.append(list_data)
return decoded_list, start_index + 1
elif data[start_index] == ord('d'):
start_index += 1
decoded_dict = {}
while data[start_index] != ord('e'):
2019-01-22 18:49:43 +01:00
key, start_index = _bdecode(data, start_index)
value, start_index = _bdecode(data, start_index)
decoded_dict[key] = value
return decoded_dict, start_index
else:
split_pos = data[start_index:].find(b':') + start_index
try:
length = int(data[start_index:split_pos])
2019-01-22 18:49:43 +01:00
except (ValueError, TypeError) as err:
raise DecodeError(err)
start_index = split_pos + 1
end_pos = start_index + length
2020-01-03 05:16:44 +01:00
return data[start_index:end_pos], end_pos
2019-01-22 18:49:43 +01:00
def bencode(data: typing.Dict) -> bytes:
if not isinstance(data, dict):
raise TypeError()
return _bencode(data)
def bdecode(data: bytes, allow_non_dict_return: typing.Optional[bool] = False) -> typing.Dict:
assert isinstance(data, bytes), DecodeError(f"invalid data type: {str(type(data))}")
2019-01-22 18:49:43 +01:00
if len(data) == 0:
raise DecodeError('Cannot decode empty string')
try:
result = _bdecode(data)[0]
if not allow_non_dict_return and not isinstance(result, dict):
raise ValueError(f'expected dict, got {type(result)}')
return result
except (ValueError, TypeError) as err:
raise DecodeError(err)