code.kerkeslager.com Git - ton/blob - don/string.py

   1 import binascii
   2 import collections
   3 import functools
   4 import re
   5
   6 from don import tags, _shared
   7
   8 def _integer_size_to_string_serializer(integer_size):
   9     minimum = -(2 ** (integer_size - 1))
  10     maximum = 2 ** (integer_size - 1) - 1
  11
  12     def serializer(integer):
  13         assert minimum <= integer and integer <= maximum
  14         return '{}i{}'.format(integer, integer_size)
  15
  16     return serializer
  17
  18 def _serialize_float(f):
  19     return '{}f'.format(f)
  20
  21 def _serialize_double(d):
  22     return '{}d'.format(d)
  23
  24 def _serialize_binary(b):
  25     return '"{}"b'.format(binascii.hexlify(b).decode('ascii'))
  26
  27 def _utf_encoding_to_serializer(utf_encoding):
  28     def serializer(s):
  29         return '"{}"{}'.format(s, utf_encoding)
  30
  31     return serializer
  32
  33 def _string_serialize_list(l):
  34     return '[{}]'.format(', '.join(map(serialize, l)))
  35
  36 def _string_serialize_dictionary(d):
  37     def serialize_kvp(kvp):
  38         return serialize(kvp[0]) + ': ' + serialize(kvp[1])
  39     return '{ ' + ', '.join(map(serialize_kvp, d.items())) + ' }'
  40
  41 _STRING_SERIALIZERS = {
  42     tags.VOID: lambda o: 'null',
  43     tags.TRUE: lambda o: 'true',
  44     tags.FALSE: lambda o: 'false',
  45     tags.INT8: _integer_size_to_string_serializer(8),
  46     tags.INT16: _integer_size_to_string_serializer(16),
  47     tags.INT32: _integer_size_to_string_serializer(32),
  48     tags.INT64: _integer_size_to_string_serializer(64),
  49     tags.FLOAT: _serialize_float,
  50     tags.DOUBLE: _serialize_double,
  51     tags.BINARY: _serialize_binary,
  52     tags.UTF8: _utf_encoding_to_serializer('utf8'),
  53     tags.UTF16: _utf_encoding_to_serializer('utf16'),
  54     tags.UTF32: _utf_encoding_to_serializer('utf32'),
  55     tags.LIST: _string_serialize_list,
  56     tags.DICTIONARY: _string_serialize_dictionary,
  57 }
  58
  59 def serialize(o):
  60     o = tags._tag(o)
  61
  62     return _STRING_SERIALIZERS[o.tag](o.value)
  63
  64 def _consume_leading_whitespace(wrapped_parser):
  65     @functools.wraps(wrapped_parser)
  66     def parser(s):
  67         s = s.lstrip()
  68         return wrapped_parser(s)
  69
  70     return parser
  71
  72 def _make_constant_parser(constant, value):
  73     @_consume_leading_whitespace
  74     def constant_parser(s):
  75         if s.startswith(constant):
  76             result = _shared.ParseResult(
  77                 success = True,
  78                 value = value,
  79                 remaining = s[len(constant):],
  80             )
  81             return result
  82
  83         return _shared._FAILED_PARSE_RESULT
  84
  85     return constant_parser
  86
  87 def _make_integer_parser(width):
  88     matcher = re.compile(r'(-?\d+)i' + str(width))
  89
  90     @_consume_leading_whitespace
  91     def integer_parser(s):
  92         match = matcher.match(s)
  93
  94         if match:
  95             # TODO Validate that the integer is in range
  96             return _shared.ParseResult(
  97                 success = True,
  98                 value = int(match.group(1)),
  99                 remaining = s[match.end():],
 100             )
 101
 102         return _shared._FAILED_PARSE_RESULT
 103
 104     return integer_parser
 105
 106 _BINARY32_MATCHER = re.compile(r'(-?\d+\.\d+)f')
 107 _BINARY64_MATCHER = re.compile(r'(-?\d+\.\d+)d')
 108
 109 @_consume_leading_whitespace
 110 def _binary32_parser(s):
 111     match = _BINARY32_MATCHER.match(s)
 112
 113     if match:
 114         # TODO Validate that the float is in range
 115         return _shared.ParseResult(
 116             success = True,
 117             value = float(match.group(1)),
 118             remaining = s[match.end():],
 119         )
 120
 121     return _shared._FAILED_PARSE_RESULT
 122
 123 @_consume_leading_whitespace
 124 def _binary64_parser(s):
 125     match = _BINARY64_MATCHER.match(s)
 126
 127     if match:
 128         # TODO Validate that the double is in range
 129         return _shared.ParseResult(
 130             success = True,
 131             value = float(match.group(1)),
 132             remaining = s[match.end():],
 133         )
 134
 135     return _shared._FAILED_PARSE_RESULT
 136
 137 _BINARY_MATCHER = re.compile(r'"([\da-f]*)"b')
 138
 139 @_consume_leading_whitespace
 140 def _binary_parser(s):
 141     match = _BINARY_MATCHER.match(s)
 142
 143     if match:
 144         return _shared.ParseResult(
 145             success = True,
 146             value = binascii.unhexlify(match.group(1)),
 147             remaining = s[match.end():],
 148         )
 149
 150     return _shared._FAILED_PARSE_RESULT
 151
 152 def _make_utf_parser(encoding):
 153     matcher = re.compile(r'"(.*?)"' + encoding)
 154
 155     @_consume_leading_whitespace
 156     def utf_parser(s):
 157         match = matcher.match(s)
 158
 159         if match:
 160             return _shared.ParseResult(
 161                 success = True,
 162                 value = match.group(1),
 163                 remaining = s[match.end():],
 164             )
 165
 166         return _shared._FAILED_PARSE_RESULT
 167
 168     return utf_parser
 169
 170 def _prefix_with_comma(parser):
 171     def wrapped(s):
 172         if s.startswith(','):
 173             s = s[1:]
 174
 175             result = parser(s)
 176             if not result.success:
 177                 raise Exception('Trailing comma before "{}"'.format(s))
 178
 179             return result
 180
 181         return _shared._FAILED_PARSE_RESULT
 182
 183     return wrapped
 184
 185 def _comma_separate_and_wrap(wrapped_parser, start_wrap, end_wrap, typecaster):
 186     parser_prefixed_with_comma = _prefix_with_comma(wrapped_parser)
 187
 188     def parser(s):
 189         if s.startswith(start_wrap):
 190             s = s[1:]
 191         else:
 192             return _shared._FAILED_PARSE_RESULT
 193
 194         value = []
 195         first = True
 196
 197         parse_result = wrapped_parser(s)
 198
 199         while parse_result.success:
 200             value.append(parse_result.value)
 201             s = parse_result.remaining
 202             parse_result = parser_prefixed_with_comma(s)
 203
 204         if s.startswith(end_wrap):
 205             s = s[1:]
 206         else:
 207             return _shared._FAILED_PARSE_RESULT
 208
 209         return _shared.ParseResult(
 210             success = True,
 211             value = typecaster(value),
 212             remaining = s,
 213         )
 214
 215     return parser
 216
 217 # This uses _PARSERS which has not been defined yet, but is defined here so it can be used in
 218 # the definition of _list_parser
 219 def _object_parser(source):
 220     for parser in _PARSERS:
 221         result = parser(source)
 222
 223         if result.success:
 224             return result
 225
 226     return _shared._FAILED_PARSE_RESULT
 227
 228 _list_parser = _comma_separate_and_wrap(_object_parser, '[', ']', list)
 229
 230 def _kvp_parser(s):
 231     key_parse_result = _object_parser(s)
 232     if key_parse_result.success:
 233         s = key_parse_result.remaining
 234     else:
 235         return _shared._FAILED_PARSE_RESULT
 236
 237     if s.startswith(':'):
 238         s = s[1:]
 239     else:
 240         return _shared._FAILED_PARSE_RESULT
 241
 242     value_parse_result = _object_parser(s)
 243     if value_parse_result.success:
 244         s = value_parse_result.remaining
 245     else:
 246         return _shared._FAILED_PARSE_RESULT
 247
 248     return _shared.ParseResult(
 249         success = True,
 250         value = (key_parse_result.value, value_parse_result.value),
 251         remaining = s,
 252     )
 253
 254 _dictionary_parser = _comma_separate_and_wrap(_kvp_parser, '{', '}', collections.OrderedDict)
 255
 256
 257 _PARSERS = [
 258     _make_constant_parser('null', None),
 259     _make_constant_parser('true', True),
 260     _make_constant_parser('false', False),
 261     _make_integer_parser(8),
 262     _make_integer_parser(16),
 263     _make_integer_parser(32),
 264     _make_integer_parser(64),
 265     _binary32_parser,
 266     _binary64_parser,
 267     _binary_parser,
 268     _make_utf_parser('utf8'),
 269     _make_utf_parser('utf16'),
 270     _make_utf_parser('utf32'),
 271     _list_parser,
 272     _dictionary_parser,
 273 ]
 274
 275 def _parse(parser, source):
 276     result = parser(source)
 277
 278     if result.success:
 279         if result.remaining.strip() == '':
 280             return result.value
 281
 282         raise Exception('Unparsed trailing characters: "{}"'.format(result.remaining))
 283
 284     raise Exception('Unable to parse: "{}"'.format(source))
 285
 286 def deserialize(s):
 287     return _parse(_object_parser, s)