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 _nextSibling(self): 173 if self._node.next is not None: 174 return Node(self._node.next) 175 else: 176 return None 177 178 def hasAttributeNS(self, ns, localName): 179 return self.getAttributeNS(ns, localName) is not None 180 181 def hasAttribute(self, name): 182 return self.getAttribute(name) is not None 183 184 def getAttributeNS(self, ns, localName): 185 return self._node.nsProp(localName, ns) 186 187 def getAttribute(self, name): 188 return self._node.prop(name) 189 190 def getAttributeNodeNS(self, ns, localName): 191 return self.attributes[(ns, localName)] 192 193 def getAttributeNode(self, localName): 194 # NOTE: Needs verifying. 195 return self.attributes[(None, localName)] 196 197 def setAttributeNS(self, ns, name, value): 198 prefix, localName = _get_prefix_and_localName(name) 199 if prefix is not None: 200 self._node.setNsProp(self._node.newNs(ns, prefix), localName, value) 201 elif ns == self._node.ns().content: 202 self._node.setNsProp(self._node.ns().content, localName, value) 203 else: 204 # NOTE: Needs verifying: what should happen to the namespace? 205 self._node.setNsProp(None, localName, value) 206 207 def setAttribute(self, name, value): 208 self._node.setProp(name, value) 209 210 def setAttributeNodeNS(self, ns, name, node): 211 # NOTE: Not actually putting the node on the element. 212 self.setAttributeNS(ns, name, node.nodeValue) 213 214 def setAttributeNode(self, name, node): 215 # NOTE: Not actually putting the node on the element. 216 self.setAttribute(name, node.nodeValue) 217 218 def createElementNS(self, ns, name): 219 prefix, localName = _get_prefix_and_localName(name) 220 _node = libxml2.newNode(localName) 221 _ns = _node.newNs(ns, prefix) 222 _node.setNs(_ns) 223 return Node(_node) 224 225 def createElement(self, name): 226 _node = libxml2.newNode(localName) 227 return Node(_node) 228 229 def createAttributeNS(self, ns, name): 230 prefix, localName = _get_prefix_and_localName(name) 231 return TemporaryNode(ns, name, xml.dom.Node.ATTRIBUTE_NODE) 232 233 def createAttribute(self, name): 234 return TemporaryNode(ns, name, xml.dom.Node.ATTRIBUTE_NODE) 235 236 def createTextNode(self, value): 237 return Node(libxml2.newText(value)) 238 239 def _add_node(self, tmp): 240 if tmp.nodeType == xml.dom.Node.ATTRIBUTE_NODE: 241 if tmp.ns is not None: 242 _child = self._node.newNsProp(None, tmp.localName, None) 243 _ns = _child.newNs(tmp.ns, tmp.prefix) 244 _child.setNs(_ns) 245 else: 246 _child = self._node.newProp(None, tmp.name, None) 247 else: 248 _child = None 249 250 return _child 251 252 def importNode(self, node, deep): 253 254 if node.nodeType == xml.dom.Node.ELEMENT_NODE: 255 imported_element = self.ownerDocument.createElementNS(node.namespaceURI, node.tagName) 256 for value in node.attributes.values(): 257 imported_element.setAttributeNS(value.namespaceURI, value.nodeName, value.nodeValue) 258 259 if deep: 260 for child in node.childNodes: 261 imported_child = self.importNode(child, deep) 262 if imported_child: 263 imported_element.appendChild(imported_child) 264 265 return imported_element 266 267 elif node.nodeType == xml.dom.Node.TEXT_NODE: 268 return self.ownerDocument.createTextNode(node.nodeValue) 269 270 elif node.nodeType == xml.dom.Node.ATTRIBUTE_NODE: 271 return self.ownerDocument.createAttributeNS(node.namespaceURI, node.name) 272 273 raise ValueError, node.nodeType 274 275 def insertBefore(self, tmp, oldNode): 276 if tmp.nodeType == xml.dom.Node.TEXT_NODE: 277 _child = tmp._node 278 elif tmp.nodeType == xml.dom.Node.ELEMENT_NODE: 279 _child = tmp._node 280 else: 281 _child = self._add_node(tmp) 282 _child.unlinkNode() 283 return Node(oldNode._node.addPrevSibling(_child)) 284 285 def replaceChild(self, tmp, oldNode): 286 if tmp.nodeType == xml.dom.Node.TEXT_NODE: 287 _child = tmp._node 288 elif tmp.nodeType == xml.dom.Node.ELEMENT_NODE: 289 _child = tmp._node 290 else: 291 _child = self._add_node(tmp) 292 _child.unlinkNode() 293 return Node(oldNode._node.replaceNode(_child)) 294 295 def appendChild(self, tmp): 296 if tmp.nodeType == xml.dom.Node.TEXT_NODE: 297 _child = self._node.addChild(tmp._node) 298 elif tmp.nodeType == xml.dom.Node.ELEMENT_NODE: 299 _child = self._node.addChild(tmp._node) 300 else: 301 _child = self._add_node(tmp) 302 return Node(_child) 303 304 def removeChild(self, tmp): 305 tmp._node.unlinkNode() 306 307 #doctype defined in __init__ 308 #ownerElement defined in __init__ 309 ownerDocument = property(_ownerDocument) 310 childNodes = property(_childNodes) 311 value = data = nodeValue = property(_nodeValue) 312 name = nodeName = property(_nodeName) 313 tagName = property(_tagName) 314 namespaceURI = property(_namespaceURI) 315 prefix = property(_prefix) 316 localName = property(_localName) 317 parentNode = property(_parentNode) 318 nodeType = property(_nodeType) 319 attributes = property(_attributes) 320 nextSibling = property(_nextSibling) 321 322 def isSameNode(self, other): 323 return self._node.nodePath() == other._node.nodePath() 324 325 def __eq__(self, other): 326 return self._node.nodePath() == other._node.nodePath() 327 328 # Utility functions. 329 330 def createDocumentType(localName, publicId, systemId): 331 return None 332 333 def createDocument(namespaceURI, localName, doctype): 334 # NOTE: Fixed to use version 1.0 only. 335 d = Node(libxml2.newDoc("1.0"), doctype=doctype) 336 if localName is not None: 337 root = d.createElementNS(namespaceURI, localName) 338 d.appendChild(root) 339 return d 340 341 def parse(stream_or_string): 342 if hasattr(stream_or_string, "read"): 343 stream = stream_or_string 344 else: 345 stream = open(stream_or_string) 346 return parseString(stream.read()) 347 348 def parseString(s): 349 return Node(libxml2.parseDoc(s)) 350 351 def parseURI(uri): 352 return Node(libxml2.parseURI(uri)) 353 354 def toString(node): 355 return node.as_native_node().serialize() 356 357 def toStream(node, stream=None): 358 stream = stream or sys.stdout 359 stream.write(toString(node)) 360 361 # vim: tabstop=4 expandtab shiftwidth=4