mediawiki_dump_tools/Mediawiki-Utilities/mw/xml_dump/iteration/text.py

   1 from ...types import serializable
   2 from ...util import none_or
   3
   4
   5 class Text(str, serializable.Type):
   6     """
   7     Revision text content.  This class behaves identically to
   8     :class:`str` except that it takes and stores an additional set of parameters.
   9
  10     **deleted**
  11         Was the text deleted? : `bool`
  12     **xml_space**
  13         What to do with extra whitespace? : `str`
  14     **id**
  15         TODO: ??? : `int` | `None`
  16     **bytes**
  17         TODO: ??? : `int` | `None`
  18
  19     >>> from mw.xml_dump import Text
  20     >>>
  21     >>> t = Text("foo")
  22     >>> t == "foo"
  23     True
  24     >>> t.deleted
  25     False
  26     >>> t.xml_space
  27     'preserve'
  28     """
  29
  30     def __new__(cls, string_or_text="", deleted=False, xml_space="preserve", id=None, bytes=None):
  31         if isinstance(string_or_text, cls):
  32             return string_or_text
  33         else:
  34             inst = super().__new__(cls, string_or_text)
  35             inst.initialize(string_or_text, deleted, xml_space, id, bytes)
  36             return inst
  37
  38     def initialize(self, string, deleted, xml_space, id, bytes):
  39         self.deleted = bool(deleted)
  40         self.xml_space = none_or(xml_space, str)
  41         self.id = none_or(id, int)
  42         self.bytes = none_or(bytes, int)
  43
  44     def __str__(self):
  45         return str.__str__(self)
  46
  47     def __repr__(self):
  48         return "{0}({1})".format(
  49             self.__class__.__name__,
  50             ", ".join([
  51                 str.__repr__(self),
  52                 "deleted={0}".format(repr(self.deleted))
  53             ])
  54         )
  55
  56     def serialize(self):
  57         return {
  58             "string_or_text": str(self),
  59             "deleted": self.deleted,
  60             "xml_space": self.xml_space,
  61             "id": self.id,
  62             "bytes": self.bytes
  63         }
  64
  65     @classmethod
  66     def from_element(cls, e):
  67         content = e.text or ""
  68         return cls(
  69             content,
  70             deleted=e.attr('deleted', False),
  71             xml_space=e.attr('xml:space'),
  72             id=e.attr('id'),
  73             bytes=e.attr('bytes')
  74         )