projects / mirror_edk2.git / blame
| Commit | Line | Data |
|---|---|---|
| 3257aa99 DM |
1 | import xml.sax\r |
| 2 | import xml.sax.handler\r | |
| 3 | import types\r | |
| 4 | \r | |
| 5 | try:\r | |
| 6 | _StringTypes = [types.StringType, types.UnicodeType]\r | |
| 7 | except AttributeError:\r | |
| 8 | _StringTypes = [types.StringType]\r | |
| 9 | \r | |
| 10 | START_ELEMENT = "START_ELEMENT"\r | |
| 11 | END_ELEMENT = "END_ELEMENT"\r | |
| 12 | COMMENT = "COMMENT"\r | |
| 13 | START_DOCUMENT = "START_DOCUMENT"\r | |
| 14 | END_DOCUMENT = "END_DOCUMENT"\r | |
| 15 | PROCESSING_INSTRUCTION = "PROCESSING_INSTRUCTION"\r | |
| 16 | IGNORABLE_WHITESPACE = "IGNORABLE_WHITESPACE"\r | |
| 17 | CHARACTERS = "CHARACTERS"\r | |
| 18 | \r | |
| 19 | class PullDOM(xml.sax.ContentHandler):\r | |
| 20 | _locator = None\r | |
| 21 | document = None\r | |
| 22 | \r | |
| 23 | def __init__(self, documentFactory=None):\r | |
| 24 | from xml.dom import XML_NAMESPACE\r | |
| 25 | self.documentFactory = documentFactory\r | |
| 26 | self.firstEvent = [None, None]\r | |
| 27 | self.lastEvent = self.firstEvent\r | |
| 28 | self.elementStack = []\r | |
| 29 | self.push = self.elementStack.append\r | |
| 30 | try:\r | |
| 31 | self.pop = self.elementStack.pop\r | |
| 32 | except AttributeError:\r | |
| 33 | # use class' pop instead\r | |
| 34 | pass\r | |
| 35 | self._ns_contexts = [{XML_NAMESPACE:'xml'}] # contains uri -> prefix dicts\r | |
| 36 | self._current_context = self._ns_contexts[-1]\r | |
| 37 | self.pending_events = []\r | |
| 38 | \r | |
| 39 | def pop(self):\r | |
| 40 | result = self.elementStack[-1]\r | |
| 41 | del self.elementStack[-1]\r | |
| 42 | return result\r | |
| 43 | \r | |
| 44 | def setDocumentLocator(self, locator):\r | |
| 45 | self._locator = locator\r | |
| 46 | \r | |
| 47 | def startPrefixMapping(self, prefix, uri):\r | |
| 48 | if not hasattr(self, '_xmlns_attrs'):\r | |
| 49 | self._xmlns_attrs = []\r | |
| 50 | self._xmlns_attrs.append((prefix or 'xmlns', uri))\r | |
| 51 | self._ns_contexts.append(self._current_context.copy())\r | |
| 52 | self._current_context[uri] = prefix or None\r | |
| 53 | \r | |
| 54 | def endPrefixMapping(self, prefix):\r | |
| 55 | self._current_context = self._ns_contexts.pop()\r | |
| 56 | \r | |
| 57 | def startElementNS(self, name, tagName , attrs):\r | |
| 58 | # Retrieve xml namespace declaration attributes.\r | |
| 59 | xmlns_uri = 'http://www.w3.org/2000/xmlns/'\r | |
| 60 | xmlns_attrs = getattr(self, '_xmlns_attrs', None)\r | |
| 61 | if xmlns_attrs is not None:\r | |
| 62 | for aname, value in xmlns_attrs:\r | |
| 63 | attrs._attrs[(xmlns_uri, aname)] = value\r | |
| 64 | self._xmlns_attrs = []\r | |
| 65 | uri, localname = name\r | |
| 66 | if uri:\r | |
| 67 | # When using namespaces, the reader may or may not\r | |
| 68 | # provide us with the original name. If not, create\r | |
| 69 | # *a* valid tagName from the current context.\r | |
| 70 | if tagName is None:\r | |
| 71 | prefix = self._current_context[uri]\r | |
| 72 | if prefix:\r | |
| 73 | tagName = prefix + ":" + localname\r | |
| 74 | else:\r | |
| 75 | tagName = localname\r | |
| 76 | if self.document:\r | |
| 77 | node = self.document.createElementNS(uri, tagName)\r | |
| 78 | else:\r | |
| 79 | node = self.buildDocument(uri, tagName)\r | |
| 80 | else:\r | |
| 81 | # When the tagname is not prefixed, it just appears as\r | |
| 82 | # localname\r | |
| 83 | if self.document:\r | |
| 84 | node = self.document.createElement(localname)\r | |
| 85 | else:\r | |
| 86 | node = self.buildDocument(None, localname)\r | |
| 87 | \r | |
| 88 | for aname,value in attrs.items():\r | |
| 89 | a_uri, a_localname = aname\r | |
| 90 | if a_uri == xmlns_uri:\r | |
| 91 | if a_localname == 'xmlns':\r | |
| 92 | qname = a_localname\r | |
| 93 | else:\r | |
| 94 | qname = 'xmlns:' + a_localname\r | |
| 95 | attr = self.document.createAttributeNS(a_uri, qname)\r | |
| 96 | node.setAttributeNodeNS(attr)\r | |
| 97 | elif a_uri:\r | |
| 98 | prefix = self._current_context[a_uri]\r | |
| 99 | if prefix:\r | |
| 100 | qname = prefix + ":" + a_localname\r | |
| 101 | else:\r | |
| 102 | qname = a_localname\r | |
| 103 | attr = self.document.createAttributeNS(a_uri, qname)\r | |
| 104 | node.setAttributeNodeNS(attr)\r | |
| 105 | else:\r | |
| 106 | attr = self.document.createAttribute(a_localname)\r | |
| 107 | node.setAttributeNode(attr)\r | |
| 108 | attr.value = value\r | |
| 109 | \r | |
| 110 | self.lastEvent[1] = [(START_ELEMENT, node), None]\r | |
| 111 | self.lastEvent = self.lastEvent[1]\r | |
| 112 | self.push(node)\r | |
| 113 | \r | |
| 114 | def endElementNS(self, name, tagName):\r | |
| 115 | self.lastEvent[1] = [(END_ELEMENT, self.pop()), None]\r | |
| 116 | self.lastEvent = self.lastEvent[1]\r | |
| 117 | \r | |
| 118 | def startElement(self, name, attrs):\r | |
| 119 | if self.document:\r | |
| 120 | node = self.document.createElement(name)\r | |
| 121 | else:\r | |
| 122 | node = self.buildDocument(None, name)\r | |
| 123 | \r | |
| 124 | for aname,value in attrs.items():\r | |
| 125 | attr = self.document.createAttribute(aname)\r | |
| 126 | attr.value = value\r | |
| 127 | node.setAttributeNode(attr)\r | |
| 128 | \r | |
| 129 | self.lastEvent[1] = [(START_ELEMENT, node), None]\r | |
| 130 | self.lastEvent = self.lastEvent[1]\r | |
| 131 | self.push(node)\r | |
| 132 | \r | |
| 133 | def endElement(self, name):\r | |
| 134 | self.lastEvent[1] = [(END_ELEMENT, self.pop()), None]\r | |
| 135 | self.lastEvent = self.lastEvent[1]\r | |
| 136 | \r | |
| 137 | def comment(self, s):\r | |
| 138 | if self.document:\r | |
| 139 | node = self.document.createComment(s)\r | |
| 140 | self.lastEvent[1] = [(COMMENT, node), None]\r | |
| 141 | self.lastEvent = self.lastEvent[1]\r | |
| 142 | else:\r | |
| 143 | event = [(COMMENT, s), None]\r | |
| 144 | self.pending_events.append(event)\r | |
| 145 | \r | |
| 146 | def processingInstruction(self, target, data):\r | |
| 147 | if self.document:\r | |
| 148 | node = self.document.createProcessingInstruction(target, data)\r | |
| 149 | self.lastEvent[1] = [(PROCESSING_INSTRUCTION, node), None]\r | |
| 150 | self.lastEvent = self.lastEvent[1]\r | |
| 151 | else:\r | |
| 152 | event = [(PROCESSING_INSTRUCTION, target, data), None]\r | |
| 153 | self.pending_events.append(event)\r | |
| 154 | \r | |
| 155 | def ignorableWhitespace(self, chars):\r | |
| 156 | node = self.document.createTextNode(chars)\r | |
| 157 | self.lastEvent[1] = [(IGNORABLE_WHITESPACE, node), None]\r | |
| 158 | self.lastEvent = self.lastEvent[1]\r | |
| 159 | \r | |
| 160 | def characters(self, chars):\r | |
| 161 | node = self.document.createTextNode(chars)\r | |
| 162 | self.lastEvent[1] = [(CHARACTERS, node), None]\r | |
| 163 | self.lastEvent = self.lastEvent[1]\r | |
| 164 | \r | |
| 165 | def startDocument(self):\r | |
| 166 | if self.documentFactory is None:\r | |
| 167 | import xml.dom.minidom\r | |
| 168 | self.documentFactory = xml.dom.minidom.Document.implementation\r | |
| 169 | \r | |
| 170 | def buildDocument(self, uri, tagname):\r | |
| 171 | # Can't do that in startDocument, since we need the tagname\r | |
| 172 | # XXX: obtain DocumentType\r | |
| 173 | node = self.documentFactory.createDocument(uri, tagname, None)\r | |
| 174 | self.document = node\r | |
| 175 | self.lastEvent[1] = [(START_DOCUMENT, node), None]\r | |
| 176 | self.lastEvent = self.lastEvent[1]\r | |
| 177 | self.push(node)\r | |
| 178 | # Put everything we have seen so far into the document\r | |
| 179 | for e in self.pending_events:\r | |
| 180 | if e[0][0] == PROCESSING_INSTRUCTION:\r | |
| 181 | _,target,data = e[0]\r | |
| 182 | n = self.document.createProcessingInstruction(target, data)\r | |
| 183 | e[0] = (PROCESSING_INSTRUCTION, n)\r | |
| 184 | elif e[0][0] == COMMENT:\r | |
| 185 | n = self.document.createComment(e[0][1])\r | |
| 186 | e[0] = (COMMENT, n)\r | |
| 187 | else:\r | |
| 188 | raise AssertionError("Unknown pending event ",e[0][0])\r | |
| 189 | self.lastEvent[1] = e\r | |
| 190 | self.lastEvent = e\r | |
| 191 | self.pending_events = None\r | |
| 192 | return node.firstChild\r | |
| 193 | \r | |
| 194 | def endDocument(self):\r | |
| 195 | self.lastEvent[1] = [(END_DOCUMENT, self.document), None]\r | |
| 196 | self.pop()\r | |
| 197 | \r | |
| 198 | def clear(self):\r | |
| 199 | "clear(): Explicitly release parsing structures"\r | |
| 200 | self.document = None\r | |
| 201 | \r | |
| 202 | class ErrorHandler:\r | |
| 203 | def warning(self, exception):\r | |
| 204 | print exception\r | |
| 205 | def error(self, exception):\r | |
| 206 | raise exception\r | |
| 207 | def fatalError(self, exception):\r | |
| 208 | raise exception\r | |
| 209 | \r | |
| 210 | class DOMEventStream:\r | |
| 211 | def __init__(self, stream, parser, bufsize):\r | |
| 212 | self.stream = stream\r | |
| 213 | self.parser = parser\r | |
| 214 | self.bufsize = bufsize\r | |
| 215 | if not hasattr(self.parser, 'feed'):\r | |
| 216 | self.getEvent = self._slurp\r | |
| 217 | self.reset()\r | |
| 218 | \r | |
| 219 | def reset(self):\r | |
| 220 | self.pulldom = PullDOM()\r | |
| 221 | # This content handler relies on namespace support\r | |
| 222 | self.parser.setFeature(xml.sax.handler.feature_namespaces, 1)\r | |
| 223 | self.parser.setContentHandler(self.pulldom)\r | |
| 224 | \r | |
| 225 | def __getitem__(self, pos):\r | |
| 226 | rc = self.getEvent()\r | |
| 227 | if rc:\r | |
| 228 | return rc\r | |
| 229 | raise IndexError\r | |
| 230 | \r | |
| 231 | def next(self):\r | |
| 232 | rc = self.getEvent()\r | |
| 233 | if rc:\r | |
| 234 | return rc\r | |
| 235 | raise StopIteration\r | |
| 236 | \r | |
| 237 | def __iter__(self):\r | |
| 238 | return self\r | |
| 239 | \r | |
| 240 | def expandNode(self, node):\r | |
| 241 | event = self.getEvent()\r | |
| 242 | parents = [node]\r | |
| 243 | while event:\r | |
| 244 | token, cur_node = event\r | |
| 245 | if cur_node is node:\r | |
| 246 | return\r | |
| 247 | if token != END_ELEMENT:\r | |
| 248 | parents[-1].appendChild(cur_node)\r | |
| 249 | if token == START_ELEMENT:\r | |
| 250 | parents.append(cur_node)\r | |
| 251 | elif token == END_ELEMENT:\r | |
| 252 | del parents[-1]\r | |
| 253 | event = self.getEvent()\r | |
| 254 | \r | |
| 255 | def getEvent(self):\r | |
| 256 | # use IncrementalParser interface, so we get the desired\r | |
| 257 | # pull effect\r | |
| 258 | if not self.pulldom.firstEvent[1]:\r | |
| 259 | self.pulldom.lastEvent = self.pulldom.firstEvent\r | |
| 260 | while not self.pulldom.firstEvent[1]:\r | |
| 261 | buf = self.stream.read(self.bufsize)\r | |
| 262 | if not buf:\r | |
| 263 | self.parser.close()\r | |
| 264 | return None\r | |
| 265 | self.parser.feed(buf)\r | |
| 266 | rc = self.pulldom.firstEvent[1][0]\r | |
| 267 | self.pulldom.firstEvent[1] = self.pulldom.firstEvent[1][1]\r | |
| 268 | return rc\r | |
| 269 | \r | |
| 270 | def _slurp(self):\r | |
| 271 | """ Fallback replacement for getEvent() using the\r | |
| 272 | standard SAX2 interface, which means we slurp the\r | |
| 273 | SAX events into memory (no performance gain, but\r | |
| 274 | we are compatible to all SAX parsers).\r | |
| 275 | """\r | |
| 276 | self.parser.parse(self.stream)\r | |
| 277 | self.getEvent = self._emit\r | |
| 278 | return self._emit()\r | |
| 279 | \r | |
| 280 | def _emit(self):\r | |
| 281 | """ Fallback replacement for getEvent() that emits\r | |
| 282 | the events that _slurp() read previously.\r | |
| 283 | """\r | |
| 284 | rc = self.pulldom.firstEvent[1][0]\r | |
| 285 | self.pulldom.firstEvent[1] = self.pulldom.firstEvent[1][1]\r | |
| 286 | return rc\r | |
| 287 | \r | |
| 288 | def clear(self):\r | |
| 289 | """clear(): Explicitly release parsing objects"""\r | |
| 290 | self.pulldom.clear()\r | |
| 291 | del self.pulldom\r | |
| 292 | self.parser = None\r | |
| 293 | self.stream = None\r | |
| 294 | \r | |
| 295 | class SAX2DOM(PullDOM):\r | |
| 296 | \r | |
| 297 | def startElementNS(self, name, tagName , attrs):\r | |
| 298 | PullDOM.startElementNS(self, name, tagName, attrs)\r | |
| 299 | curNode = self.elementStack[-1]\r | |
| 300 | parentNode = self.elementStack[-2]\r | |
| 301 | parentNode.appendChild(curNode)\r | |
| 302 | \r | |
| 303 | def startElement(self, name, attrs):\r | |
| 304 | PullDOM.startElement(self, name, attrs)\r | |
| 305 | curNode = self.elementStack[-1]\r | |
| 306 | parentNode = self.elementStack[-2]\r | |
| 307 | parentNode.appendChild(curNode)\r | |
| 308 | \r | |
| 309 | def processingInstruction(self, target, data):\r | |
| 310 | PullDOM.processingInstruction(self, target, data)\r | |
| 311 | node = self.lastEvent[0][1]\r | |
| 312 | parentNode = self.elementStack[-1]\r | |
| 313 | parentNode.appendChild(node)\r | |
| 314 | \r | |
| 315 | def ignorableWhitespace(self, chars):\r | |
| 316 | PullDOM.ignorableWhitespace(self, chars)\r | |
| 317 | node = self.lastEvent[0][1]\r | |
| 318 | parentNode = self.elementStack[-1]\r | |
| 319 | parentNode.appendChild(node)\r | |
| 320 | \r | |
| 321 | def characters(self, chars):\r | |
| 322 | PullDOM.characters(self, chars)\r | |
| 323 | node = self.lastEvent[0][1]\r | |
| 324 | parentNode = self.elementStack[-1]\r | |
| 325 | parentNode.appendChild(node)\r | |
| 326 | \r | |
| 327 | \r | |
| 328 | default_bufsize = (2 ** 14) - 20\r | |
| 329 | \r | |
| 330 | def parse(stream_or_string, parser=None, bufsize=None):\r | |
| 331 | if bufsize is None:\r | |
| 332 | bufsize = default_bufsize\r | |
| 333 | if type(stream_or_string) in _StringTypes:\r | |
| 334 | stream = open(stream_or_string)\r | |
| 335 | else:\r | |
| 336 | stream = stream_or_string\r | |
| 337 | if not parser:\r | |
| 338 | parser = xml.sax.make_parser()\r | |
| 339 | return DOMEventStream(stream, parser, bufsize)\r | |
| 340 | \r | |
| 341 | def parseString(string, parser=None):\r | |
| 342 | try:\r | |
| 343 | from cStringIO import StringIO\r | |
| 344 | except ImportError:\r | |
| 345 | from StringIO import StringIO\r | |
| 346 | \r | |
| 347 | bufsize = len(string)\r | |
| 348 | buf = StringIO(string)\r | |
| 349 | if not parser:\r | |
| 350 | parser = xml.sax.make_parser()\r | |
| 351 | return DOMEventStream(buf, parser, bufsize)\r |