1 #!/usr/bin/env python 2 3 """ 4 DOM wrapper around libxml2, specifically the libxml2mod Python extension module. 5 """ 6 7 __version__ = "0.2" 8 9 import libxml2 10 from libxml2dom.macrolib import * 11 from libxml2dom.macrolib import \ 12 createDocument as Node_createDocument, \ 13 parseString as Node_parseString, parseURI as Node_parseURI, \ 14 parseFile as Node_parseFile, \ 15 toString as Node_toString, toStream as Node_toStream, \ 16 toFile as Node_toFile 17 import weakref 18 19 # Attribute and node list wrappers. 20 21 class NamedNodeMap(object): 22 23 """ 24 A wrapper around Node objects providing DOM and dictionary convenience 25 methods. 26 """ 27 28 def __init__(self, node): 29 self.node = node 30 31 def getNamedItem(self, name): 32 return self.node.getAttributeNode(name) 33 34 def getNamedItemNS(self, ns, localName): 35 return self.node.getAttributeNodeNS(ns, localName) 36 37 def setNamedItem(self, node): 38 self.node.setAttributeNode(node) 39 # NOTE: Non-standard return of attribute node, since the actual node is 40 # NOTE: not set on the target node. 41 return self.getNamedItem(node.name) 42 43 def setNamedItemNS(self, node): 44 self.node.setAttributeNodeNS(node) 45 # NOTE: Non-standard return of attribute node, since the actual node is 46 # NOTE: not set on the target node. 47 return self.getNamedItemNS(node.namespaceURI, node.name) 48 49 # Dictionary emulation methods. 50 51 def __getitem__(self, name): 52 return self.getNamedItem(name) 53 54 def __setitem__(self, name, node): 55 if name == node.nodeName: 56 self.setNamedItem(node) 57 else: 58 raise KeyError, name 59 60 def __delitem__(self, name): 61 # NOTE: To be implemented. 62 pass 63 64 def values(self): 65 return [Node(_node, self.node.ownerDocument) for _node in Node_attributes(self.node.as_native_node()).values()] 66 67 def keys(self): 68 return [(attr.namespaceURI, attr.localName) for attr in self.values()] 69 70 def items(self): 71 return [((attr.namespaceURI, attr.localName), attr) for attr in self.values()] 72 73 def __repr__(self): 74 return str(self) 75 76 def __str__(self): 77 return "{%s}" % ",\n".join(["%s : %s" % (repr(key), repr(value)) for key, value in self.items()]) 78 79 class NodeList(list): 80 81 "A wrapper around node lists." 82 83 def item(self, index): 84 return self[index] 85 86 def length(self): 87 return len(self) 88 89 # Node classes. 90 91 class Node(object): 92 93 """ 94 A DOM-style wrapper around libxml2mod objects. 95 """ 96 97 ATTRIBUTE_NODE = xml.dom.Node.ATTRIBUTE_NODE 98 COMMENT_NODE = xml.dom.Node.COMMENT_NODE 99 DOCUMENT_NODE = xml.dom.Node.DOCUMENT_NODE 100 DOCUMENT_TYPE_NODE = xml.dom.Node.DOCUMENT_TYPE_NODE 101 ELEMENT_NODE = xml.dom.Node.ELEMENT_NODE 102 ENTITY_NODE = xml.dom.Node.ENTITY_NODE 103 ENTITY_REFERENCE_NODE = xml.dom.Node.ENTITY_REFERENCE_NODE 104 NOTATION_NODE = xml.dom.Node.NOTATION_NODE 105 PROCESSING_INSTRUCTION_NODE = xml.dom.Node.PROCESSING_INSTRUCTION_NODE 106 TEXT_NODE = xml.dom.Node.TEXT_NODE 107 108 def __init__(self, node, ownerDocument=None): 109 self._node = node 110 self.ownerDocument = ownerDocument 111 112 def as_native_node(self): 113 return self._node 114 115 def _nodeType(self): 116 return Node_nodeType(self._node) 117 118 def _childNodes(self): 119 120 # NOTE: Consider a generator instead. 121 122 return NodeList([Node(_node, self.ownerDocument) for _node in Node_childNodes(self._node)]) 123 124 def _attributes(self): 125 return NamedNodeMap(self) 126 127 def _namespaceURI(self): 128 return Node_namespaceURI(self._node) 129 130 def _nodeValue(self): 131 return Node_nodeValue(self._node) 132 133 def _setNodeValue(self, value): 134 Node_setNodeValue(self._node, value) 135 136 def _prefix(self): 137 return Node_prefix(self._node) 138 139 def _nodeName(self): 140 return Node_nodeName(self._node) 141 142 def _tagName(self): 143 return Node_tagName(self._node) 144 145 def _localName(self): 146 return Node_localName(self._node) 147 148 def _parentNode(self): 149 return Node(Node_parentNode(self._node), self.ownerDocument) 150 151 def _previousSibling(self): 152 return Node(Node_previousSibling(self._node), self.ownerDocument) 153 154 def _nextSibling(self): 155 return Node(Node_nextSibling(self._node), self.ownerDocument) 156 157 def hasAttributeNS(self, ns, localName): 158 return Node_hasAttributeNS(self._node, ns, localName) 159 160 def hasAttribute(self, name): 161 return Node_hasAttribute(self._node, name) 162 163 def getAttributeNS(self, ns, localName): 164 return Node_getAttributeNS(self._node, ns, localName) 165 166 def getAttribute(self, name): 167 return Node_getAttribute(self._node, name) 168 169 def getAttributeNodeNS(self, ns, localName): 170 return Node(Node_getAttributeNodeNS(self._node, ns, localName), self.ownerDocument) 171 172 def getAttributeNode(self, localName): 173 return Node(Node_getAttributeNode(self._node, localName), self.ownerDocument) 174 175 def setAttributeNS(self, ns, name, value): 176 Node_setAttributeNS(self._node, ns, name, value) 177 178 def setAttribute(self, name, value): 179 Node_setAttribute(self._node, name, value) 180 181 def setAttributeNodeNS(self, node): 182 Node_setAttributeNodeNS(self._node, node) 183 184 def setAttributeNode(self, node): 185 Node_setAttributeNode(self._node, node) 186 187 def createElementNS(self, ns, name): 188 return Node(Node_createElementNS(self._node, ns, name), self.ownerDocument) 189 190 def createElement(self, name): 191 return Node(Node_createElement(self._node, name), self.ownerDocument) 192 193 #def createAttributeNS(self, ns, name): 194 # return Node(Node_createAttributeNS(self._node, ns, name)) 195 196 #def createAttribute(self, name): 197 # return Node(Node_createAttribute(self._node, name)) 198 199 def createTextNode(self, value): 200 return Node(Node_createTextNode(self._node, value), self.ownerDocument) 201 202 def createComment(self, value): 203 return Node(Node_createComment(self._node, value), self.ownerDocument) 204 205 def importNode(self, node, deep): 206 if hasattr(node, "as_native_node"): 207 return Node(Node_importNode(self._node, node.as_native_node(), deep), self.ownerDocument) 208 else: 209 return Node(Node_importNode_DOM(self._node, node, deep), self.ownerDocument) 210 211 def insertBefore(self, tmp, oldNode): 212 if hasattr(tmp, "as_native_node"): 213 return Node(Node_insertBefore(self._node, tmp.as_native_node(), oldNode.as_native_node()), self.ownerDocument) 214 else: 215 return Node(Node_insertBefore(self._node, tmp, oldNode.as_native_node()), self.ownerDocument) 216 217 def replaceChild(self, tmp, oldNode): 218 if hasattr(tmp, "as_native_node"): 219 return Node(Node_replaceChild(self._node, tmp.as_native_node(), oldNode.as_native_node()), self.ownerDocument) 220 else: 221 return Node(Node_replaceChild(self._node, tmp, oldNode.as_native_node()), self.ownerDocument) 222 223 def appendChild(self, tmp): 224 if hasattr(tmp, "as_native_node"): 225 return Node(Node_appendChild(self._node, tmp.as_native_node()), self.ownerDocument) 226 else: 227 return Node(Node_appendChild(self._node, tmp), self.ownerDocument) 228 229 def removeChild(self, tmp): 230 if hasattr(tmp, "as_native_node"): 231 Node_removeChild(self._node, tmp.as_native_node()) 232 else: 233 Node_removeChild(self._node, tmp) 234 235 childNodes = property(_childNodes) 236 value = data = nodeValue = property(_nodeValue, _setNodeValue) 237 name = nodeName = property(_nodeName) 238 tagName = property(_tagName) 239 namespaceURI = property(_namespaceURI) 240 prefix = property(_prefix) 241 localName = property(_localName) 242 ownerElement = parentNode = property(_parentNode) 243 nodeType = property(_nodeType) 244 attributes = property(_attributes) 245 previousSibling = property(_previousSibling) 246 nextSibling = property(_nextSibling) 247 248 #def isSameNode(self, other): 249 # return self._node.nodePath() == other._node.nodePath() 250 251 #def __eq__(self, other): 252 # return self._node.nodePath() == other._node.nodePath() 253 254 def getElementsByTagName(self, tagName): 255 return self.xpath("//" + tagName) 256 257 # 4DOM extensions to the usual PyXML API. 258 # NOTE: To be finished. 259 260 def xpath(self, expr, variables=None, namespaces=None): 261 result = Node_xpath(self._node, expr, variables, namespaces) 262 if hasattr(result, "__len__"): 263 return NodeList([Node(_node, self.ownerDocument) for _node in result]) 264 else: 265 return result 266 267 # Convenience methods. 268 269 def toString(self, encoding=None): 270 return toString(self, encoding) 271 272 def toStream(self, stream, encoding=None): 273 toStream(self, stream, encoding) 274 275 def toFile(self, f, encoding=None): 276 toFile(self, f, encoding) 277 278 # Document housekeeping mechanisms. 279 280 class Document(Node): 281 282 "A class providing document-level housekeeping." 283 284 def __init__(self, node): 285 self._node = node 286 self.weakref_ownerDocument = weakref.ref(self) 287 288 def _ownerDocument(self): 289 return self.weakref_ownerDocument() 290 291 def __del__(self): 292 #print "Freeing document", self._node 293 libxml2mod.xmlFreeDoc(self._node) 294 295 ownerDocument = property(_ownerDocument) 296 297 # Utility functions. 298 299 def createDocumentType(localName, publicId, systemId): 300 return None 301 302 def createDocument(namespaceURI, localName, doctype): 303 return Document(Node_createDocument(namespaceURI, localName, doctype)) 304 305 def parse(stream_or_string): 306 if hasattr(stream_or_string, "read"): 307 stream = stream_or_string 308 return parseString(stream.read()) 309 else: 310 return parseFile(stream_or_string) 311 312 def parseFile(s): 313 return Document(Node_parseFile(s)) 314 315 def parseString(s): 316 return Document(Node_parseString(s)) 317 318 def parseURI(uri): 319 return Document(Node_parseURI(uri)) 320 321 def toString(node, encoding=None): 322 return Node_toString(node.as_native_node(), encoding) 323 324 def toStream(node, stream, encoding=None): 325 Node_toStream(node.as_native_node(), stream, encoding) 326 327 def toFile(node, f, encoding=None): 328 Node_toFile(node.as_native_node(), f, encoding) 329 330 def adoptNodes(nodes): 331 if len(nodes) == 0: 332 return [] 333 doc = Document(libxml2mod.doc(nodes[0])) 334 results = [] 335 for node in nodes: 336 results.append(Node(node, doc)) 337 return results 338 339 # vim: tabstop=4 expandtab shiftwidth=4