Moved tags into their own module
[sandbox] / serial / serial / binary.py
1 import collections
2 import functools
3 import io
4 import struct
5
6 from . import tag
7
8 TaggedObject = collections.namedtuple(
9     'TaggedObject',
10     [
11         'tag',
12         'instance',
13     ],
14 )
15
16 def _make_tag_only_serializer(tag, expected_value):
17     tag = bytes([tag])
18
19     def serializer(to):
20         assert to.instance == expected_value
21         return tag
22
23     return serializer
24
25 def _make_struct_serializer(fmt):
26     fmt = '!B' + fmt
27     packer = functools.partial(struct.pack, fmt)
28
29     def serializer(to):
30         return packer(to.tag, to.instance)
31
32     return serializer
33
34 def _make_string_serializer(encoder):
35     packer = functools.partial(struct.pack, '!BI')
36
37     def serializer(to):
38         encoded = encoder(to.instance)
39         return packer(to.tag, len(encoded)) + encoded
40
41     return serializer
42
43 def _serialize_tuple(to):
44     assert isinstance(to.instance, tuple)
45
46     payload = b''.join(serialize(item) for item in to.instance)
47
48     fmt = '!BI'
49
50     return struct.pack('!BI', tag.TUPLE, len(payload)) + payload
51
52
53 _TAGS_TO_SERIALIZERS = {
54     tag.NULL: _make_tag_only_serializer(tag.NULL, None),
55     tag.TRUE: _make_tag_only_serializer(tag.TRUE, True),
56     tag.FALSE: _make_tag_only_serializer(tag.FALSE, False),
57     tag.UINT8: _make_struct_serializer('B'),
58     tag.UINT16: _make_struct_serializer('H'),
59     tag.UINT32: _make_struct_serializer('I'),
60     tag.UINT64: _make_struct_serializer('Q'),
61     tag.INT8: _make_struct_serializer('b'),
62     tag.INT16: _make_struct_serializer('h'),
63     tag.INT32: _make_struct_serializer('i'),
64     tag.INT64: _make_struct_serializer('q'),
65     tag.BINARY: _make_string_serializer(lambda s: s),
66     tag.UTF8: _make_string_serializer(lambda s: s.encode('utf-8')),
67     tag.UTF16: _make_string_serializer(lambda s: s.encode('utf-16')),
68     tag.UTF32: _make_string_serializer(lambda s: s.encode('utf-32')),
69     tag.TUPLE: _serialize_tuple,
70 }
71
72 def serialize(to):
73     return _TAGS_TO_SERIALIZERS[to.tag](to)
74
75 def _make_tag_only_parser(tag, value):
76     def parser(b):
77         return 0, TaggedObject(tag = tag, instance = value)
78
79     return parser
80
81 def _make_struct_deserializer(tag, fmt):
82     fmt = '!' + fmt
83     size = struct.calcsize(fmt)
84     unpacker = functools.partial(struct.unpack, fmt)
85
86     def parser(b):
87         b = b.read(size)
88         assert len(b) == size
89         return size, TaggedObject(tag = tag, instance = unpacker(b)[0])
90
91     return parser
92
93 _LENGTH_FMT = '!I'
94 _LENGTH_FMT_SIZE = struct.calcsize(_LENGTH_FMT)
95
96 def _read_length_then_payload(b):
97     length_b = b.read(_LENGTH_FMT_SIZE)
98     assert len(length_b) == _LENGTH_FMT_SIZE
99     length = struct.unpack(_LENGTH_FMT, length_b)[0]
100
101     payload = b.read(length)
102     assert len(payload) == length
103     return _LENGTH_FMT_SIZE + length, payload
104
105 def _make_string_deserializer(tag, decoder):
106     fmt = '!I'
107     size = struct.calcsize(fmt)
108     unpacker = functools.partial(struct.unpack, fmt)
109
110     def parser(b):
111         bytes_read, payload = _read_length_then_payload(b)
112         return bytes_read, TaggedObject(tag = tag, instance = decoder(payload))
113
114     return parser
115
116 def _deserialize_tuple(b):
117     bytes_read, payload = _read_length_then_payload(b)
118
119     payload_stream = io.BytesIO(payload)
120
121     total_bytes_read = 0
122     instance = []
123
124     while total_bytes_read < len(payload):
125         partial_bytes_read, item = _deserialize_partial(payload_stream)
126         total_bytes_read += partial_bytes_read
127         instance.append(item)
128
129     return bytes_read, TaggedObject(tag = tag.TUPLE, instance = tuple(instance))
130
131 _TAGS_TO_PARSERS = {
132     tag.NULL: _make_tag_only_parser(tag.NULL, None),
133     tag.TRUE: _make_tag_only_parser(tag.TRUE, True),
134     tag.FALSE: _make_tag_only_parser(tag.FALSE, False),
135     tag.UINT8: _make_struct_deserializer(tag.UINT8, 'B'),
136     tag.UINT16: _make_struct_deserializer(tag.UINT16, 'H'),
137     tag.UINT32: _make_struct_deserializer(tag.UINT32, 'I'),
138     tag.UINT64: _make_struct_deserializer(tag.UINT64, 'Q'),
139     tag.INT8: _make_struct_deserializer(tag.INT8, 'b'),
140     tag.INT16: _make_struct_deserializer(tag.INT16, 'h'),
141     tag.INT32: _make_struct_deserializer(tag.INT32, 'i'),
142     tag.INT64: _make_struct_deserializer(tag.INT64, 'q'),
143     tag.BINARY: _make_string_deserializer(tag.BINARY, lambda b: b),
144     tag.UTF8: _make_string_deserializer(tag.UTF8, lambda b: b.decode('utf-8')),
145     tag.UTF16: _make_string_deserializer(tag.UTF16, lambda b: b.decode('utf-16')),
146     tag.UTF32: _make_string_deserializer(tag.UTF32, lambda b: b.decode('utf-32')),
147     tag.TUPLE: _deserialize_tuple,
148 }
149
150 def _deserialize_partial(b):
151     tag = b.read(1)
152     assert len(tag) == 1
153     bytes_read, to = _TAGS_TO_PARSERS[tag[0]](b)
154     return bytes_read + 1, to
155
156 def deserialize(b):
157     if isinstance(b, bytes):
158         b = io.BytesIO(b)
159
160     bytes_read, result = _deserialize_partial(b)
161
162     remainder = b.read()
163
164     if len(remainder) == 0:
165         return result
166
167     raise Exception('Unable to parse remainder: {}'.format(remainder))