lib/yaml/representer.py

   1
   2 __all__ = ['BaseRepresenter', 'SafeRepresenter', 'Representer',
   3     'RepresenterError']
   4
   5 from error import *
   6 from nodes import *
   7
   8 try:
   9     import datetime
  10     datetime_available = True
  11 except ImportError:
  12     datetime_available = False
  13
  14 try:
  15     set
  16 except NameError:
  17     from sets import Set as set
  18
  19 import sys, copy_reg
  20
  21 class RepresenterError(YAMLError):
  22     pass
  23
  24 class BaseRepresenter:
  25
  26     yaml_representers = {}
  27     yaml_multi_representers = {}
  28
  29     def __init__(self, default_style=None, default_flow_style=None):
  30         self.default_style = default_style
  31         self.default_flow_style = default_flow_style
  32         self.represented_objects = {}
  33
  34     def represent(self, data):
  35         node = self.represent_data(data)
  36         self.serialize(node)
  37         self.represented_objects = {}
  38
  39     class C: pass
  40     c = C()
  41     def f(): pass
  42     classobj_type = type(C)
  43     instance_type = type(c)
  44     function_type = type(f)
  45     builtin_function_type = type(abs)
  46     module_type = type(sys)
  47     del C, c, f
  48
  49     def get_classobj_bases(self, cls):
  50         bases = [cls]
  51         for base in cls.__bases__:
  52             bases.extend(self.get_classobj_bases(base))
  53         return bases
  54
  55     def represent_data(self, data):
  56         if self.ignore_aliases(data):
  57             alias_key = None
  58         else:
  59             alias_key = id(data)
  60         if alias_key is not None:
  61             if alias_key in self.represented_objects:
  62                 node = self.represented_objects[alias_key]
  63                 if node is None:
  64                     raise RepresenterError("recursive objects are not allowed: %r" % data)
  65                 return node
  66             self.represented_objects[alias_key] = None
  67         data_types = type(data).__mro__
  68         if type(data) is self.instance_type:
  69             data_types = self.get_classobj_bases(data.__class__)+list(data_types)
  70         if data_types[0] in self.yaml_representers:
  71             node = self.yaml_representers[data_types[0]](self, data)
  72         else:
  73             for data_type in data_types:
  74                 if data_type in self.yaml_multi_representers:
  75                     node = self.yaml_multi_representers[data_type](self, data)
  76                     break
  77             else:
  78                 if None in self.yaml_multi_representers:
  79                     node = self.yaml_multi_representers[None](self, data)
  80                 elif None in self.yaml_representers:
  81                     node = self.yaml_representers[None](self, data)
  82                 else:
  83                     node = ScalarNode(None, unicode(data))
  84         if alias_key is not None:
  85             self.represented_objects[alias_key] = node
  86         return node
  87
  88     def add_representer(cls, data_type, representer):
  89         if not 'yaml_representers' in cls.__dict__:
  90             cls.yaml_representers = cls.yaml_representers.copy()
  91         cls.yaml_representers[data_type] = representer
  92     add_representer = classmethod(add_representer)
  93
  94     def add_multi_representer(cls, data_type, representer):
  95         if not 'yaml_multi_representers' in cls.__dict__:
  96             cls.yaml_multi_representers = cls.yaml_multi_representers.copy()
  97         cls.yaml_multi_representers[data_type] = representer
  98     add_multi_representer = classmethod(add_multi_representer)
  99
 100     def represent_scalar(self, tag, value, style=None):
 101         if style is None:
 102             style = self.default_style
 103         return ScalarNode(tag, value, style=style)
 104
 105     def represent_sequence(self, tag, sequence, flow_style=None):
 106         best_style = True
 107         value = []
 108         for item in sequence:
 109             node_item = self.represent_data(item)
 110             if not (isinstance(node_item, ScalarNode) and not node_item.style):
 111                 best_style = False
 112             value.append(self.represent_data(item))
 113         if flow_style is None:
 114             flow_style = self.default_flow_style
 115         if flow_style is None:
 116             flow_style = best_style
 117         return SequenceNode(tag, value, flow_style=flow_style)
 118
 119     def represent_mapping(self, tag, mapping, flow_style=None):
 120         best_style = True
 121         if hasattr(mapping, 'keys'):
 122             value = {}
 123             for item_key in mapping.keys():
 124                 item_value = mapping[item_key]
 125                 node_key = self.represent_data(item_key)
 126                 node_value = self.represent_data(item_value)
 127                 if not (isinstance(node_key, ScalarNode) and not node_key.style):
 128                     best_style = False
 129                 if not (isinstance(node_value, ScalarNode) and not node_value.style):
 130                     best_style = False
 131                 value[node_key] = node_value
 132         else:
 133             value = []
 134             for item_key, item_value in mapping:
 135                 node_key = self.represent_data(item_key)
 136                 node_value = self.represent_data(item_value)
 137                 if not (isinstance(node_key, ScalarNode) and not node_key.style):
 138                     best_style = False
 139                 if not (isinstance(node_value, ScalarNode) and not node_value.style):
 140                     best_style = False
 141                 value.append((node_key, node_value))
 142         if flow_style is None:
 143             flow_style = self.default_flow_style
 144         if flow_style is None:
 145             flow_style = best_style
 146         return MappingNode(tag, value, flow_style=flow_style)
 147
 148     def ignore_aliases(self, data):
 149         return False
 150
 151 class SafeRepresenter(BaseRepresenter):
 152
 153     def ignore_aliases(self, data):
 154         if data in [None, ()]:
 155             return True
 156         if isinstance(data, (str, unicode, bool, int, float)):
 157             return True
 158
 159     def represent_none(self, data):
 160         return self.represent_scalar(u'tag:yaml.org,2002:null',
 161                 u'null')
 162
 163     def represent_str(self, data):
 164         tag = None
 165         style = None
 166         try:
 167             data = unicode(data, 'ascii')
 168             tag = u'tag:yaml.org,2002:str'
 169         except UnicodeDecodeError:
 170             try:
 171                 data = unicode(data, 'utf-8')
 172                 tag = u'tag:yaml.org,2002:str'
 173             except UnicodeDecodeError:
 174                 data = data.encode('base64')
 175                 tag = u'tag:yaml.org,2002:binary'
 176                 style = '|'
 177         return self.represent_scalar(tag, data, style=style)
 178
 179     def represent_unicode(self, data):
 180         return self.represent_scalar(u'tag:yaml.org,2002:str', data)
 181
 182     def represent_bool(self, data):
 183         if data:
 184             value = u'true'
 185         else:
 186             value = u'false'
 187         return self.represent_scalar(u'tag:yaml.org,2002:bool', value)
 188
 189     def represent_int(self, data):
 190         return self.represent_scalar(u'tag:yaml.org,2002:int', unicode(data))
 191
 192     def represent_long(self, data):
 193         return self.represent_scalar(u'tag:yaml.org,2002:int', unicode(data))
 194
 195     repr_pos_inf = repr(1e300000)
 196     repr_neg_inf = repr(-1e300000)
 197     repr_nan = repr(1e300000/1e300000)
 198
 199     def represent_float(self, data):
 200         repr_data = repr(data)
 201         if repr_data == self.repr_pos_inf:
 202             value = u'.inf'
 203         elif repr_data == self.repr_neg_inf:
 204             value = u'-.inf'
 205         elif repr_data == self.repr_nan:
 206             value = u'.nan'
 207         else:
 208             value = unicode(repr_data)
 209         return self.represent_scalar(u'tag:yaml.org,2002:float', value)
 210
 211     def represent_list(self, data):
 212         pairs = (len(data) > 0 and isinstance(data, list))
 213         if pairs:
 214             for item in data:
 215                 if not isinstance(item, tuple) or len(item) != 2:
 216                     pairs = False
 217                     break
 218         if not pairs:
 219             return self.represent_sequence(u'tag:yaml.org,2002:seq', data)
 220         value = []
 221         for item_key, item_value in data:
 222             value.append(self.represent_mapping(u'tag:yaml.org,2002:map',
 223                 [(item_key, item_value)]))
 224         return SequenceNode(u'tag:yaml.org,2002:pairs', value)
 225
 226     def represent_dict(self, data):
 227         return self.represent_mapping(u'tag:yaml.org,2002:map', data)
 228
 229     def represent_set(self, data):
 230         value = {}
 231         for key in data:
 232             value[key] = None
 233         return self.represent_mapping(u'tag:yaml.org,2002:set', value)
 234
 235     def represent_date(self, data):
 236         value = u'%04d-%02d-%02d' % (data.year, data.month, data.day)
 237         return self.represent_scalar(u'tag:yaml.org,2002:timestamp', value)
 238
 239     def represent_datetime(self, data):
 240         value = u'%04d-%02d-%02d %02d:%02d:%02d' \
 241                 % (data.year, data.month, data.day,
 242                     data.hour, data.minute, data.second)
 243         if data.microsecond:
 244             value += u'.' + unicode(data.microsecond/1000000.0).split(u'.')[1]
 245         if data.utcoffset():
 246             value += unicode(data.utcoffset())
 247         return self.represent_scalar(u'tag:yaml.org,2002:timestamp', value)
 248
 249     def represent_yaml_object(self, tag, data, cls, flow_style=None):
 250         if hasattr(data, '__getstate__'):
 251             state = data.__getstate__()
 252         else:
 253             state = data.__dict__.copy()
 254         if isinstance(state, dict):
 255             state = state.items()
 256             state.sort()
 257         return self.represent_mapping(tag, state, flow_style=flow_style)
 258
 259     def represent_undefined(self, data):
 260         raise RepresenterError("cannot represent an object: %s" % data)
 261
 262 SafeRepresenter.add_representer(type(None),
 263         SafeRepresenter.represent_none)
 264
 265 SafeRepresenter.add_representer(str,
 266         SafeRepresenter.represent_str)
 267
 268 SafeRepresenter.add_representer(unicode,
 269         SafeRepresenter.represent_unicode)
 270
 271 SafeRepresenter.add_representer(bool,
 272         SafeRepresenter.represent_bool)
 273
 274 SafeRepresenter.add_representer(int,
 275         SafeRepresenter.represent_int)
 276
 277 SafeRepresenter.add_representer(long,
 278         SafeRepresenter.represent_long)
 279
 280 SafeRepresenter.add_representer(float,
 281         SafeRepresenter.represent_float)
 282
 283 SafeRepresenter.add_representer(list,
 284         SafeRepresenter.represent_list)
 285
 286 SafeRepresenter.add_representer(tuple,
 287         SafeRepresenter.represent_list)
 288
 289 SafeRepresenter.add_representer(dict,
 290         SafeRepresenter.represent_dict)
 291
 292 SafeRepresenter.add_representer(set,
 293         SafeRepresenter.represent_set)
 294
 295 if datetime_available:
 296     SafeRepresenter.add_representer(datetime.date,
 297             SafeRepresenter.represent_date)
 298     SafeRepresenter.add_representer(datetime.datetime,
 299             SafeRepresenter.represent_datetime)
 300
 301 SafeRepresenter.add_representer(None,
 302         SafeRepresenter.represent_undefined)
 303
 304 class Representer(SafeRepresenter):
 305
 306     def represent_str(self, data):
 307         tag = None
 308         style = None
 309         try:
 310             data = unicode(data, 'ascii')
 311             tag = u'tag:yaml.org,2002:str'
 312         except UnicodeDecodeError:
 313             try:
 314                 data = unicode(data, 'utf-8')
 315                 tag = u'tag:yaml.org,2002:python/str'
 316             except UnicodeDecodeError:
 317                 data = data.encode('base64')
 318                 tag = u'tag:yaml.org,2002:binary'
 319                 style = '|'
 320         return self.represent_scalar(tag, data, style=style)
 321
 322     def represent_unicode(self, data):
 323         tag = None
 324         try:
 325             data.encode('ascii')
 326             tag = u'tag:yaml.org,2002:python/unicode'
 327         except UnicodeEncodeError:
 328             tag = u'tag:yaml.org,2002:str'
 329         return self.represent_scalar(tag, data)
 330
 331     def represent_long(self, data):
 332         tag = u'tag:yaml.org,2002:int'
 333         if int(data) is not data:
 334             tag = u'tag:yaml.org,2002:python/long'
 335         return self.represent_scalar(tag, unicode(data))
 336
 337     def represent_complex(self, data):
 338         if data.imag == 0.0:
 339             data = u'%r' % data.real
 340         elif data.real == 0.0:
 341             data = u'%rj' % data.imag
 342         elif data.imag > 0:
 343             data = u'%r+%rj' % (data.real, data.imag)
 344         else:
 345             data = u'%r%rj' % (data.real, data.imag)
 346         return self.represent_scalar(u'tag:yaml.org,2002:python/complex', data)
 347
 348     def represent_tuple(self, data):
 349         return self.represent_sequence(u'tag:yaml.org,2002:python/tuple', data)
 350
 351     def represent_name(self, data):
 352         name = u'%s.%s' % (data.__module__, data.__name__)
 353         return self.represent_scalar(u'tag:yaml.org,2002:python/name:'+name, u'')
 354
 355     def represent_module(self, data):
 356         return self.represent_scalar(
 357                 u'tag:yaml.org,2002:python/module:'+data.__name__, u'')
 358
 359     def represent_instance(self, data):
 360         # For instances of classic classes, we use __getinitargs__ and
 361         # __getstate__ to serialize the data.
 362
 363         # If data.__getinitargs__ exists, the object must be reconstructed by
 364         # calling cls(**args), where args is a tuple returned by
 365         # __getinitargs__. Otherwise, the cls.__init__ method should never be
 366         # called and the class instance is created by instantiating a trivial
 367         # class and assigning to the instance's __class__ variable.
 368
 369         # If data.__getstate__ exists, it returns the state of the object.
 370         # Otherwise, the state of the object is data.__dict__.
 371
 372         # We produce either a !!python/object or !!python/object/new node.
 373         # If data.__getinitargs__ does not exist and state is a dictionary, we
 374         # produce a !!python/object node . Otherwise we produce a
 375         # !!python/object/new node.
 376
 377         cls = data.__class__
 378         class_name = u'%s.%s' % (cls.__module__, cls.__name__)
 379         args = None
 380         state = None
 381         if hasattr(data, '__getinitargs__'):
 382             args = list(data.__getinitargs__())
 383         if hasattr(data, '__getstate__'):
 384             state = data.__getstate__()
 385         else:
 386             state = data.__dict__
 387         if args is None and isinstance(state, dict):
 388             state = state.items()
 389             state.sort()
 390             return self.represent_mapping(
 391                     u'tag:yaml.org,2002:python/object:'+class_name, state)
 392         if isinstance(state, dict) and not state:
 393             return self.represent_sequence(
 394                     u'tag:yaml.org,2002:python/object/new:'+class_name, args)
 395         value = {}
 396         if args:
 397             value['args'] = args
 398         value['state'] = state
 399         return self.represent_mapping(
 400                 u'tag:yaml.org,2002:python/object/new:'+class_name, value)
 401
 402     def represent_object(self, data):
 403         # We use __reduce__ API to save the data. data.__reduce__ returns
 404         # a tuple of length 2-5:
 405         #   (function, args, state, listitems, dictitems)
 406
 407         # For reconstructing, we calls function(*args), then set its state,
 408         # listitems, and dictitems if they are not None.
 409
 410         # A special case is when function.__name__ == '__newobj__'. In this
 411         # case we create the object with args[0].__new__(*args).
 412
 413         # Another special case is when __reduce__ returns a string - we don't
 414         # support it.
 415
 416         # We produce a !!python/object, !!python/object/new or
 417         # !!python/object/apply node.
 418
 419         cls = type(data)
 420         if cls in copy_reg.dispatch_table:
 421             reduce = copy_reg.dispatch_table[cls]
 422         elif hasattr(data, '__reduce_ex__'):
 423             reduce = data.__reduce_ex__(2)
 424         elif hasattr(data, '__reduce__'):
 425             reduce = data.__reduce__()
 426         else:
 427             raise RepresenterError("cannot represent object: %r" % data)
 428         reduce = (list(reduce)+[None]*5)[:5]
 429         function, args, state, listitems, dictitems = reduce
 430         args = list(args)
 431         if state is None:
 432             state = {}
 433         if listitems is not None:
 434             listitems = list(listitems)
 435         if dictitems is not None:
 436             dictitems = dict(dictitems)
 437         if function.__name__ == '__newobj__':
 438             function = args[0]
 439             args = args[1:]
 440             tag = u'tag:yaml.org,2002:python/object/new:'
 441             newobj = True
 442         else:
 443             tag = u'tag:yaml.org,2002:python/object/apply:'
 444             newobj = False
 445         function_name = u'%s.%s' % (function.__module__, function.__name__)
 446         if not args and not listitems and not dictitems \
 447                 and isinstance(state, dict) and newobj:
 448             state = state.items()
 449             state.sort()
 450             return self.represent_mapping(
 451                     u'tag:yaml.org,2002:python/object:'+function_name, state)
 452         if not listitems and not dictitems  \
 453                 and isinstance(state, dict) and not state:
 454             return self.represent_sequence(tag+function_name, args)
 455         value = {}
 456         if args:
 457             value['args'] = args
 458         if state or not isinstance(state, dict):
 459             value['state'] = state
 460         if listitems:
 461             value['listitems'] = listitems
 462         if dictitems:
 463             value['dictitems'] = dictitems
 464         return self.represent_mapping(tag+function_name, value)
 465
 466 Representer.add_representer(str,
 467         Representer.represent_str)
 468
 469 Representer.add_representer(unicode,
 470         Representer.represent_unicode)
 471
 472 Representer.add_representer(long,
 473         Representer.represent_long)
 474
 475 Representer.add_representer(complex,
 476         Representer.represent_complex)
 477
 478 Representer.add_representer(tuple,
 479         Representer.represent_tuple)
 480
 481 Representer.add_representer(type,
 482         Representer.represent_name)
 483
 484 Representer.add_representer(Representer.classobj_type,
 485         Representer.represent_name)
 486
 487 Representer.add_representer(Representer.function_type,
 488         Representer.represent_name)
 489
 490 Representer.add_representer(Representer.builtin_function_type,
 491         Representer.represent_name)
 492
 493 Representer.add_representer(Representer.module_type,
 494         Representer.represent_module)
 495
 496 Representer.add_multi_representer(Representer.instance_type,
 497         Representer.represent_instance)
 498
 499 Representer.add_multi_representer(object,
 500         Representer.represent_object)
 501