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 def _binary32_parser(s):
 110     match = _BINARY32_MATCHER.match(s)
 111
 112     if match:
 113         # TODO Validate that the float is in range
 114         return _shared.ParseResult(
 115             success = True,
 116             value = float(match.group(1)),
 117             remaining = s[match.end():],
 118         )
 119
 120     return _shared._FAILED_PARSE_RESULT
 121
 122 def _binary64_parser(s):
 123     match = _BINARY64_MATCHER.match(s)
 124
 125     if match:
 126         # TODO Validate that the double is in range
 127         return _shared.ParseResult(
 128             success = True,
 129             value = float(match.group(1)),
 130             remaining = s[match.end():],
 131         )
 132
 133     return _shared._FAILED_PARSE_RESULT
 134
 135 _BINARY_MATCHER = re.compile(r'"([\da-f]*)"b')
 136
 137 def _binary_parser(s):
 138     match = _BINARY_MATCHER.match(s)
 139
 140     if match:
 141         return _shared.ParseResult(
 142             success = True,
 143             value = binascii.unhexlify(match.group(1)),
 144             remaining = s[match.end():],
 145         )
 146
 147     return _shared._FAILED_PARSE_RESULT
 148
 149 def _make_utf_parser(encoding):
 150     matcher = re.compile(r'"(.*?)"' + encoding)
 151
 152     def parser(s):
 153         match = matcher.match(s)
 154
 155         if match:
 156             return _shared.ParseResult(
 157                 success = True,
 158                 value = match.group(1),
 159                 remaining = s[match.end():],
 160             )
 161
 162         return _shared._FAILED_PARSE_RESULT
 163
 164     return parser
 165
 166 def _prefix_with_comma(parser):
 167     def wrapped(s):
 168         if s.startswith(','):
 169             s = s[1:]
 170
 171             result = parser(s)
 172             if not result.success:
 173                 raise Exception('Trailing comma before "{}"'.format(s))
 174
 175             return result
 176
 177         return _shared._FAILED_PARSE_RESULT
 178
 179     return wrapped
 180
 181 def _comma_separate_and_wrap(wrapped_parser, start_wrap, end_wrap, typecaster):
 182     parser_prefixed_with_comma = _prefix_with_comma(wrapped_parser)
 183
 184     def parser(s):
 185         if s.startswith(start_wrap):
 186             s = s[1:]
 187         else:
 188             return _shared._FAILED_PARSE_RESULT
 189
 190         value = []
 191         first = True
 192
 193         parse_result = wrapped_parser(s)
 194
 195         while parse_result.success:
 196             value.append(parse_result.value)
 197             s = parse_result.remaining
 198             parse_result = parser_prefixed_with_comma(s)
 199
 200         if s.startswith(end_wrap):
 201             s = s[1:]
 202         else:
 203             return _shared._FAILED_PARSE_RESULT
 204
 205         return _shared.ParseResult(
 206             success = True,
 207             value = typecaster(value),
 208             remaining = s,
 209         )
 210
 211     return parser
 212
 213 # This uses _PARSERS which has not been defined yet, but is defined here so it can be used in
 214 # the definition of _list_parser
 215 def _object_parser(source):
 216     for parser in _PARSERS:
 217         result = parser(source)
 218
 219         if result.success:
 220             return result
 221
 222     return _shared._FAILED_PARSE_RESULT
 223
 224 _list_parser = _comma_separate_and_wrap(_object_parser, '[', ']', list)
 225
 226 def _kvp_parser(s):
 227     key_parse_result = _object_parser(s)
 228     if key_parse_result.success:
 229         s = key_parse_result.remaining
 230     else:
 231         return _shared._FAILED_PARSE_RESULT
 232
 233     if s.startswith(':'):
 234         s = s[1:]
 235     else:
 236         return _shared._FAILED_PARSE_RESULT
 237
 238     value_parse_result = _object_parser(s)
 239     if value_parse_result.success:
 240         s = value_parse_result.remaining
 241     else:
 242         return _shared._FAILED_PARSE_RESULT
 243
 244     return _shared.ParseResult(
 245         success = True,
 246         value = (key_parse_result.value, value_parse_result.value),
 247         remaining = s,
 248     )
 249
 250 _dictionary_parser = _comma_separate_and_wrap(_kvp_parser, '{', '}', collections.OrderedDict)
 251
 252
 253 _PARSERS = [
 254     _make_constant_parser('null', None),
 255     _make_constant_parser('true', True),
 256     _make_constant_parser('false', False),
 257     _make_integer_parser(8),
 258     _make_integer_parser(16),
 259     _make_integer_parser(32),
 260     _make_integer_parser(64),
 261     _binary32_parser,
 262     _binary64_parser,
 263     _binary_parser,
 264     _make_utf_parser('utf8'),
 265     _make_utf_parser('utf16'),
 266     _make_utf_parser('utf32'),
 267     _list_parser,
 268     _dictionary_parser,
 269 ]
 270
 271 def _parse(parser, source):
 272     result = parser(source)
 273
 274     if result.success:
 275         if result.remaining.strip() == '':
 276             return result.value
 277
 278         raise Exception('Unparsed trailing characters: "{}"'.format(result.remaining))
 279
 280     raise Exception('Unable to parse: "{}"'.format(source))
 281
 282 def deserialize(s):
 283     return _parse(_object_parser, s)