1 #!/usr/bin/env python 2 3 """ 4 DOM wrapper around libxml2. 5 """ 6 7 import xml.dom 8 import libxml2 9 import sys 10 11 # NOTE: Consider a generator instead. 12 13 class NamedNodeMap(object): 14 15 def __init__(self, node): 16 self.node = node 17 18 def getNamedItem(self, name): 19 return self.node.getAttributeNode(name) 20 21 def getNamedItemNS(self, ns, localName): 22 return self.node.getAttributeNodeNS(ns, localName) 23 24 def setNamedItem(self, node): 25 self.node.setAttributeNode(node.name, node) 26 27 def setNamedItemNS(self, node): 28 self.node.setAttributeNodeNS(node.namespaceURI, node.localName, node) 29 30 def __getitem__(self, name): 31 return self.getNamedItem(name) 32 33 def __setitem__(self, name, node): 34 if name == node.nodeName: 35 self.setNamedItem(node) 36 else: 37 raise KeyError, name 38 39 def __delitem__(self, name): 40 # NOTE: To be implemented. 41 pass 42 43 def values(self): 44 attributes = [] 45 _attribute = self.node.as_native_node().properties 46 while _attribute is not None: 47 attributes.append(Node(_attribute, ownerElement=self.node)) 48 _attribute = _attribute.next 49 return attributes 50 51 def keys(self): 52 return [(attr.namespaceURI, attr.localName) for attr in self.values()] 53 54 def items(self): 55 return [((attr.namespaceURI, attr.localName), attr) for attr in self.values()] 56 57 def __repr__(self): 58 return str(self) 59 60 def __str__(self): 61 return "{%s}" % ",\n".join(["%s : %s" % (repr(key), repr(value)) for key, value in self.items()]) 62 63 def _get_prefix_and_localName(name): 64 t = name.split(":") 65 if len(t) == 1: 66 return None, name 67 elif len(t) == 2: 68 return t 69 else: 70 # NOTE: Should raise an exception. 71 return None, None 72 73 class TemporaryNode(object): 74 def __init__(self, ns, name, nodeType): 75 self.ns = ns 76 self.name = name 77 self.nodeType = nodeType 78 self.prefix, self.localName = _get_prefix_and_localName(self.name) 79 80 class Node(object): 81 82 _nodeTypes = { 83 "attribute" : xml.dom.Node.ATTRIBUTE_NODE, 84 "comment" : xml.dom.Node.COMMENT_NODE, 85 "document_xml" : xml.dom.Node.DOCUMENT_NODE, 86 "doctype" : xml.dom.Node.DOCUMENT_TYPE_NODE, 87 "dtd" : xml.dom.Node.DOCUMENT_TYPE_NODE, # NOTE: Needs verifying. 88 "element" : xml.dom.Node.ELEMENT_NODE, 89 "entity" : xml.dom.Node.ENTITY_NODE, 90 "entity_ref" : xml.dom.Node.ENTITY_REFERENCE_NODE, 91 "notation" : xml.dom.Node.NOTATION_NODE, 92 "pi" : xml.dom.Node.PROCESSING_INSTRUCTION_NODE, 93 "text" : xml.dom.Node.TEXT_NODE 94 } 95 96 def __init__(self, node, ownerElement=None, doctype=None): 97 self._node = node 98 self.ownerElement = ownerElement 99 self.doctype = doctype 100 101 def as_native_node(self): 102 return self._node 103 104 def _ownerDocument(self): 105 return Node(self._node.doc) 106 107 def _nodeType(self): 108 return self._nodeTypes[self._node.type] 109 110 def _childNodes(self): 111 112 # NOTE: Consider a generator instead. 113 114 child_nodes = [] 115 _node = self._node.children 116 while _node is not None: 117 child_nodes.append(Node(_node)) 118 _node = _node.next 119 return child_nodes 120 121 def _attributes(self): 122 return NamedNodeMap(self) 123 124 def _getNs(self): 125 126 "Internal namespace information retrieval." 127 128 try: 129 return self._node.ns() 130 except libxml2.treeError: 131 return None 132 133 def _namespaceURI(self): 134 ns = self._getNs() 135 if ns is not None: 136 return ns.content 137 else: 138 return None 139 140 def _nodeValue(self): 141 return self._node.content 142 143 def _prefix(self): 144 ns = self._getNs() 145 if ns is not None: 146 return ns.name 147 else: 148 return None 149 150 def _nodeName(self): 151 prefix = self._prefix() 152 if prefix is not None: 153 return prefix + ":" + self._localName() 154 else: 155 return self._localName() 156 157 def _tagName(self): 158 if self._node.type == "element": 159 return self._nodeName() 160 else: 161 return None 162 163 def _localName(self): 164 return self._node.name 165 166 def _parentNode(self): 167 if self.nodeType == xml.dom.Node.DOCUMENT_NODE: 168 return None 169 else: 170 return Node(self._node.parent) 171 172 def hasAttributeNS(self, ns, localName): 173 return self.getAttributeNS(ns, localName) is not None 174 175 def hasAttribute(self, name): 176 return self.getAttribute(name) is not None 177 178 def getAttributeNS(self, ns, localName): 179 return self._node.nsProp(localName, ns) 180 181 def getAttribute(self, name): 182 return self._node.prop(name) 183 184 def getAttributeNodeNS(self, ns, localName): 185 return self.attributes[(ns, localName)] 186 187 def getAttributeNode(self, localName): 188 # NOTE: Needs verifying. 189 return self.attributes[(None, localName)] 190 191 def setAttributeNS(self, ns, name, value): 192 prefix, localName = _get_prefix_and_localName(name) 193 if prefix is not None: 194 self._node.setNsProp(self._node.newNs(ns, prefix), localName, value) 195 elif ns == self._node.ns().content: 196 self._node.setNsProp(self._node.ns().content, localName, value) 197 else: 198 # NOTE: Needs verifying: what should happen to the namespace? 199 self._node.setNsProp(None, localName, value) 200 201 def setAttribute(self, name, value): 202 self._node.setProp(name, value) 203 204 def setAttributeNodeNS(self, ns, name, node): 205 # NOTE: Not actually putting the node on the element. 206 self.setAttributeNS(ns, name, node.nodeValue) 207 208 def setAttributeNode(self, name, node): 209 # NOTE: Not actually putting the node on the element. 210 self.setAttribute(name, node.nodeValue) 211 212 def createElementNS(self, ns, name): 213 prefix, localName = _get_prefix_and_localName(name) 214 _node = libxml2.newNode(localName) 215 _ns = _node.newNs(ns, prefix) 216 _node.setNs(_ns) 217 return Node(_node) 218 219 def createElement(self, name): 220 _node = libxml2.newNode(localName) 221 return Node(_node) 222 223 def createAttributeNS(self, ns, name): 224 prefix, localName = _get_prefix_and_localName(name) 225 return TemporaryNode(ns, name, xml.dom.Node.ATTRIBUTE_NODE) 226 227 def createAttribute(self, name): 228 return TemporaryNode(ns, name, xml.dom.Node.ATTRIBUTE_NODE) 229 230 def createTextNode(self, value): 231 return Node(libxml2.newText(value)) 232 233 def _add_node(self, tmp): 234 if tmp.nodeType == xml.dom.Node.ATTRIBUTE_NODE: 235 if tmp.ns is not None: 236 _child = self._node.newNsProp(None, tmp.localName, None) 237 _ns = _child.newNs(tmp.ns, tmp.prefix) 238 _child.setNs(_ns) 239 else: 240 _child = self._node.newProp(None, tmp.name, None) 241 else: 242 _child = None 243 244 return _child 245 246 def importNode(self, node, deep): 247 248 if node.nodeType == xml.dom.Node.ELEMENT_NODE: 249 imported_element = self.ownerDocument.createElementNS(node.namespaceURI, node.tagName) 250 for value in node.attributes.values(): 251 imported_element.setAttributeNS(value.namespaceURI, value.nodeName, value.nodeValue) 252 253 if deep: 254 for child in node.childNodes: 255 imported_child = self.importNode(child, deep) 256 if imported_child: 257 imported_element.appendChild(imported_child) 258 259 return imported_element 260 261 elif node.nodeType == xml.dom.Node.TEXT_NODE: 262 return self.ownerDocument.createTextNode(node.nodeValue) 263 264 elif node.nodeType == xml.dom.Node.ATTRIBUTE_NODE: 265 return self.ownerDocument.createAttributeNS(node.namespaceURI, node.name) 266 267 raise ValueError, node.nodeType 268 269 def insertBefore(self, tmp, oldNode): 270 if tmp.nodeType == xml.dom.Node.TEXT_NODE: 271 _child = tmp._node 272 elif tmp.nodeType == xml.dom.Node.ELEMENT_NODE: 273 _child = tmp._node 274 else: 275 _child = self._add_node(tmp) 276 _child.unlinkNode() 277 return Node(oldNode._node.addPrevSibling(_child)) 278 279 def replaceChild(self, tmp, oldNode): 280 if tmp.nodeType == xml.dom.Node.TEXT_NODE: 281 _child = tmp._node 282 elif tmp.nodeType == xml.dom.Node.ELEMENT_NODE: 283 _child = tmp._node 284 else: 285 _child = self._add_node(tmp) 286 _child.unlinkNode() 287 return Node(oldNode._node.replaceNode(_child)) 288 289 def appendChild(self, tmp): 290 if tmp.nodeType == xml.dom.Node.TEXT_NODE: 291 _child = self._node.addChild(tmp._node) 292 elif tmp.nodeType == xml.dom.Node.ELEMENT_NODE: 293 _child = self._node.addChild(tmp._node) 294 else: 295 _child = self._add_node(tmp) 296 return Node(_child) 297 298 def removeChild(self, tmp): 299 tmp._node.unlinkNode() 300 301 #doctype defined in __init__ 302 #ownerElement defined in __init__ 303 ownerDocument = property(_ownerDocument) 304 childNodes = property(_childNodes) 305 value = data = nodeValue = property(_nodeValue) 306 name = nodeName = property(_nodeName) 307 tagName = property(_tagName) 308 namespaceURI = property(_namespaceURI) 309 prefix = property(_prefix) 310 localName = property(_localName) 311 parentNode = property(_parentNode) 312 nodeType = property(_nodeType) 313 attributes = property(_attributes) 314 315 def isSameNode(self, other): 316 return self._node.nodePath() == other._node.nodePath() 317 318 def __eq__(self, other): 319 return self._node.nodePath() == other._node.nodePath() 320 321 # Utility functions. 322 323 def createDocumentType(localName, publicId, systemId): 324 return None 325 326 def createDocument(namespaceURI, localName, doctype): 327 # NOTE: Fixed to use version 1.0 only. 328 d = Node(libxml2.newDoc("1.0"), doctype=doctype) 329 if localName is not None: 330 root = d.createElementNS(namespaceURI, localName) 331 d.appendChild(root) 332 return d 333 334 def parse(stream_or_string): 335 if hasattr(stream_or_string, "read"): 336 stream = stream_or_string 337 else: 338 stream = open(stream_or_string) 339 return parseString(stream.read()) 340 341 def parseString(s): 342 return Node(libxml2.parseDoc(s)) 343 344 def parseURI(uri): 345 return Node(libxml2.parseURI(uri)) 346 347 def toString(node): 348 return node.as_native_node().serialize() 349 350 def toStream(node, stream=None): 351 stream = stream or sys.stdout 352 stream.write(toString(node)) 353 354 # vim: tabstop=4 expandtab shiftwidth=4