0 $ˆl–Ãa¾”d*j"TÚ�zã@¸Ø”Å£Ä „e·\wÿÁ?÷"""Facility to use the Expat parser to load a minidom instance from a string or file. This avoids all the overhead of SAX and pulldom to gain performance. """ # Warning! # # This module is tightly bound to the implementation details of the # minidom DOM and can't be used with other DOM implementations. This # is due, in part, to a lack of appropriate methods in the DOM (there is # no way to create Entity and Notation nodes via the DOM Level 2 # interface), and for performance. The latter is the cause of some fairly # cryptic code. # # Performance hacks: # # - .character_data_handler() has an extra case in which continuing # data is appended to an existing Text node; this can be a # speedup since pyexpat can break up character data into multiple # callbacks even though we set the buffer_text attribute on the # parser. This also gives us the advantage that we don't need a # separate normalization pass. # # - Determining that a node exists is done using an identity comparison # with None rather than a truth test; this avoids searching for and # calling any methods on the node object if it exists. (A rather # nice speedup is achieved this way as well!) from xml.dom import xmlbuilder, minidom, Node from xml.dom import EMPTY_NAMESPACE, EMPTY_PREFIX, XMLNS_NAMESPACE from xml.parsers import expat from xml.dom.minidom import _append_child, _set_attribute_node from xml.dom.NodeFilter import NodeFilter TEXT_NODE = Node.TEXT_NODE CDATA_SECTION_NODE = Node.CDATA_SECTION_NODE DOCUMENT_NODE = Node.DOCUMENT_NODE FILTER_ACCEPT = xmlbuilder.DOMBuilderFilter.FILTER_ACCEPT FILTER_REJECT = xmlbuilder.DOMBuilderFilter.FILTER_REJECT FILTER_SKIP = xmlbuilder.DOMBuilderFilter.FILTER_SKIP FILTER_INTERRUPT = xmlbuilder.DOMBuilderFilter.FILTER_INTERRUPT theDOMImplementation = minidom.getDOMImplementation() # Expat typename -> TypeInfo _typeinfo_map = { "CDATA": minidom.TypeInfo(None, "cdata"), "ENUM": minidom.TypeInfo(None, "enumeration"), "ENTITY": minidom.TypeInfo(None, "entity"), "ENTITIES": minidom.TypeInfo(None, "entities"), "ID": minidom.TypeInfo(None, "id"), "IDREF": minidom.TypeInfo(None, "idref"), "IDREFS": minidom.TypeInfo(None, "idrefs"), "NMTOKEN": minidom.TypeInfo(None, "nmtoken"), "NMTOKENS": minidom.TypeInfo(None, "nmtokens"), } class ElementInfo(object): __slots__ = '_attr_info', '_model', 'tagName' def __init__(self, tagName, model=None): self.tagName = tagName self._attr_info = [] self._model = model def __getstate__(self): return self._attr_info, self._model, self.tagName def __setstate__(self, state): self._attr_info, self._model, self.tagName = state def getAttributeType(self, aname): for info in self._attr_info: if info[1] == aname: t = info[-2] if t[0] == "(": return _typeinfo_map["ENUM"] else: return _typeinfo_map[info[-2]] return minidom._no_type def getAttributeTypeNS(self, namespaceURI, localName): return minidom._no_type def isElementContent(self): if self._model: type = self._model[0] return type not in (expat.model.XML_CTYPE_ANY, expat.model.XML_CTYPE_MIXED) else: return False def isEmpty(self): if self._model: return self._model[0] == expat.model.XML_CTYPE_EMPTY else: return False def isId(self, aname): for info in self._attr_info: if info[1] == aname: return info[-2] == "ID" return False def isIdNS(self, euri, ename, auri, aname): # not sure this is meaningful return self.isId((auri, aname)) def _intern(builder, s): return builder._intern_setdefault(s, s) def _parse_ns_name(builder, name): assert ' ' in name parts = name.split(' ') intern