blob: e674385307e51b0beb85c489454a42d1d15c7a19 [file] [log] [blame]
Fred Drakec16adce2000-12-14 18:00:18 +00001import xml.sax
2import xml.sax.handler
Fred Drake55c38192000-06-29 19:39:57 +00003
Fred Drake1f549022000-09-24 05:21:58 +00004START_ELEMENT = "START_ELEMENT"
5END_ELEMENT = "END_ELEMENT"
6COMMENT = "COMMENT"
7START_DOCUMENT = "START_DOCUMENT"
8END_DOCUMENT = "END_DOCUMENT"
9PROCESSING_INSTRUCTION = "PROCESSING_INSTRUCTION"
10IGNORABLE_WHITESPACE = "IGNORABLE_WHITESPACE"
11CHARACTERS = "CHARACTERS"
Fred Drake55c38192000-06-29 19:39:57 +000012
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +000013class PullDOM(xml.sax.ContentHandler):
Fred Drakec16adce2000-12-14 18:00:18 +000014 _locator = None
15 document = None
16
17 def __init__(self, documentFactory=None):
18 self.documentFactory = documentFactory
Fred Drake1f549022000-09-24 05:21:58 +000019 self.firstEvent = [None, None]
20 self.lastEvent = self.firstEvent
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +000021 self._ns_contexts = [{}] # contains uri -> prefix dicts
22 self._current_context = self._ns_contexts[-1]
Fred Drake55c38192000-06-29 19:39:57 +000023
Fred Drakec16adce2000-12-14 18:00:18 +000024 def setDocumentLocator(self, locator):
25 self._locator = locator
Fred Drake55c38192000-06-29 19:39:57 +000026
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +000027 def startPrefixMapping(self, prefix, uri):
28 self._ns_contexts.append(self._current_context.copy())
Fred Drakec16adce2000-12-14 18:00:18 +000029 self._current_context[uri] = prefix or ''
Fred Drake55c38192000-06-29 19:39:57 +000030
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +000031 def endPrefixMapping(self, prefix):
Fred Drakec16adce2000-12-14 18:00:18 +000032 self._current_context = self._ns_contexts.pop()
Fred Drake1f549022000-09-24 05:21:58 +000033
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +000034 def startElementNS(self, name, tagName , attrs):
Fred Drakec16adce2000-12-14 18:00:18 +000035 uri, localname = name
Martin v. Löwis2c8a89c2000-10-06 22:36:03 +000036 if uri:
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +000037 # When using namespaces, the reader may or may not
38 # provide us with the original name. If not, create
39 # *a* valid tagName from the current context.
40 if tagName is None:
Martin v. Löwis2c8a89c2000-10-06 22:36:03 +000041 tagName = self._current_context[uri] + ":" + localname
42 node = self.document.createElementNS(uri, tagName)
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +000043 else:
44 # When the tagname is not prefixed, it just appears as
Martin v. Löwis2c8a89c2000-10-06 22:36:03 +000045 # localname
46 node = self.document.createElement(localname)
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +000047
48 for aname,value in attrs.items():
Martin v. Löwis2c8a89c2000-10-06 22:36:03 +000049 a_uri, a_localname = aname
50 if a_uri:
51 qname = self._current_context[a_uri] + ":" + a_localname
52 attr = self.document.createAttributeNS(a_uri, qname)
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +000053 else:
Martin v. Löwis2c8a89c2000-10-06 22:36:03 +000054 attr = self.document.createAttribute(a_localname)
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +000055 attr.value = value
Martin v. Löwis2c8a89c2000-10-06 22:36:03 +000056 node.setAttributeNode(attr)
Fred Drake16f63292000-10-23 18:09:50 +000057
Fred Drakec16adce2000-12-14 18:00:18 +000058 node.parentNode = self.curNode
Fred Drake55c38192000-06-29 19:39:57 +000059 self.curNode = node
Lars Gustäbele84bf752000-09-24 18:31:37 +000060
Fred Drake1f549022000-09-24 05:21:58 +000061 self.lastEvent[1] = [(START_ELEMENT, node), None]
62 self.lastEvent = self.lastEvent[1]
63 #self.events.append((START_ELEMENT, node))
Fred Drake55c38192000-06-29 19:39:57 +000064
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +000065 def endElementNS(self, name, tagName):
Fred Drake55c38192000-06-29 19:39:57 +000066 node = self.curNode
Fred Drake1f549022000-09-24 05:21:58 +000067 self.lastEvent[1] = [(END_ELEMENT, node), None]
68 self.lastEvent = self.lastEvent[1]
69 #self.events.append((END_ELEMENT, node))
Fred Drakec16adce2000-12-14 18:00:18 +000070 self.curNode = self.curNode.parentNode
Fred Drake55c38192000-06-29 19:39:57 +000071
Lars Gustäbeld178ba62000-10-11 22:34:04 +000072 def startElement(self, name, attrs):
73 node = self.document.createElement(name)
74
75 for aname,value in attrs.items():
76 attr = self.document.createAttribute(aname)
77 attr.value = value
78 node.setAttributeNode(attr)
Fred Drake16f63292000-10-23 18:09:50 +000079
Fred Drakec16adce2000-12-14 18:00:18 +000080 node.parentNode = self.curNode
Lars Gustäbeld178ba62000-10-11 22:34:04 +000081 self.curNode = node
82
83 self.lastEvent[1] = [(START_ELEMENT, node), None]
84 self.lastEvent = self.lastEvent[1]
85 #self.events.append((START_ELEMENT, node))
86
87 def endElement(self, name):
88 node = self.curNode
89 self.lastEvent[1] = [(END_ELEMENT, node), None]
90 self.lastEvent = self.lastEvent[1]
91 #self.events.append((END_ELEMENT, node))
92 self.curNode = node.parentNode
Fred Drake16f63292000-10-23 18:09:50 +000093
Fred Drake1f549022000-09-24 05:21:58 +000094 def comment(self, s):
95 node = self.document.createComment(s)
96 parent = self.curNode
97 node.parentNode = parent
Fred Drake1f549022000-09-24 05:21:58 +000098 self.lastEvent[1] = [(COMMENT, node), None]
99 self.lastEvent = self.lastEvent[1]
100 #self.events.append((COMMENT, node))
Fred Drake55c38192000-06-29 19:39:57 +0000101
Fred Drake1f549022000-09-24 05:21:58 +0000102 def processingInstruction(self, target, data):
103 node = self.document.createProcessingInstruction(target, data)
Fred Drake16f63292000-10-23 18:09:50 +0000104
Fred Drake1f549022000-09-24 05:21:58 +0000105 parent = self.curNode
106 node.parentNode = parent
Fred Drake1f549022000-09-24 05:21:58 +0000107 self.lastEvent[1] = [(PROCESSING_INSTRUCTION, node), None]
108 self.lastEvent = self.lastEvent[1]
109 #self.events.append((PROCESSING_INSTRUCTION, node))
Fred Drake55c38192000-06-29 19:39:57 +0000110
Fred Drake1f549022000-09-24 05:21:58 +0000111 def ignorableWhitespace(self, chars):
Fred Drakec16adce2000-12-14 18:00:18 +0000112 node = self.document.createTextNode(chars)
Fred Drake1f549022000-09-24 05:21:58 +0000113 parent = self.curNode
114 node.parentNode = parent
Fred Drake1f549022000-09-24 05:21:58 +0000115 self.lastEvent[1] = [(IGNORABLE_WHITESPACE, node), None]
116 self.lastEvent = self.lastEvent[1]
117 #self.events.append((IGNORABLE_WHITESPACE, node))
Fred Drake55c38192000-06-29 19:39:57 +0000118
Fred Drake1f549022000-09-24 05:21:58 +0000119 def characters(self, chars):
120 node = self.document.createTextNode(chars)
Lars Gustäbeld178ba62000-10-11 22:34:04 +0000121 parent = self.curNode
Lars Gustäbeld178ba62000-10-11 22:34:04 +0000122 node.parentNode = parent
Fred Drake1f549022000-09-24 05:21:58 +0000123 self.lastEvent[1] = [(CHARACTERS, node), None]
124 self.lastEvent = self.lastEvent[1]
Fred Drake55c38192000-06-29 19:39:57 +0000125
Fred Drake1f549022000-09-24 05:21:58 +0000126 def startDocument(self):
Fred Drakec16adce2000-12-14 18:00:18 +0000127 publicId = systemId = None
128 if self._locator:
129 publicId = self._locator.getPublicId()
130 systemId = self._locator.getSystemId()
131 if self.documentFactory is None:
132 import xml.dom.minidom
133 self.documentFactory = xml.dom.minidom.Document.implementation
134 node = self.documentFactory.createDocument(None, publicId, systemId)
135 self.curNode = self.document = node
Fred Drake1f549022000-09-24 05:21:58 +0000136 self.lastEvent[1] = [(START_DOCUMENT, node), None]
137 self.lastEvent = self.lastEvent[1]
138 #self.events.append((START_DOCUMENT, node))
Fred Drake55c38192000-06-29 19:39:57 +0000139
Fred Drake1f549022000-09-24 05:21:58 +0000140 def endDocument(self):
Fred Drakec16adce2000-12-14 18:00:18 +0000141 assert self.curNode.parentNode is None, \
142 "not all elements have been properly closed"
143 assert self.curNode.documentElement is not None, \
144 "document does not contain a root element"
145 node = self.curNode.documentElement
Fred Drake1f549022000-09-24 05:21:58 +0000146 self.lastEvent[1] = [(END_DOCUMENT, node), None]
147 #self.events.append((END_DOCUMENT, self.curNode))
Fred Drake55c38192000-06-29 19:39:57 +0000148
149class ErrorHandler:
Fred Drake1f549022000-09-24 05:21:58 +0000150 def warning(self, exception):
Fred Drake55c38192000-06-29 19:39:57 +0000151 print exception
Fred Drake1f549022000-09-24 05:21:58 +0000152 def error(self, exception):
Fred Drake16f63292000-10-23 18:09:50 +0000153 raise exception
Fred Drake1f549022000-09-24 05:21:58 +0000154 def fatalError(self, exception):
Fred Drake16f63292000-10-23 18:09:50 +0000155 raise exception
Fred Drake55c38192000-06-29 19:39:57 +0000156
157class DOMEventStream:
Fred Drake1f549022000-09-24 05:21:58 +0000158 def __init__(self, stream, parser, bufsize):
159 self.stream = stream
160 self.parser = parser
161 self.bufsize = bufsize
Fred Drake55c38192000-06-29 19:39:57 +0000162 self.reset()
163
Fred Drake1f549022000-09-24 05:21:58 +0000164 def reset(self):
Fred Drake55c38192000-06-29 19:39:57 +0000165 self.pulldom = PullDOM()
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +0000166 # This content handler relies on namespace support
Fred Drakec16adce2000-12-14 18:00:18 +0000167 self.parser.setFeature(xml.sax.handler.feature_namespaces, 1)
Fred Drake1f549022000-09-24 05:21:58 +0000168 self.parser.setContentHandler(self.pulldom)
Fred Drake55c38192000-06-29 19:39:57 +0000169
Fred Drake1f549022000-09-24 05:21:58 +0000170 def __getitem__(self, pos):
171 rc = self.getEvent()
172 if rc:
173 return rc
Fred Drake55c38192000-06-29 19:39:57 +0000174 raise IndexError
175
Fred Drake1f549022000-09-24 05:21:58 +0000176 def expandNode(self, node):
177 event = self.getEvent()
Fred Drake55c38192000-06-29 19:39:57 +0000178 while event:
Fred Drake1f549022000-09-24 05:21:58 +0000179 token, cur_node = event
180 if cur_node is node:
181 return
Lars Gustäbelec964d52000-10-13 20:53:27 +0000182 if token != END_ELEMENT:
183 cur_node.parentNode.appendChild(cur_node)
Fred Drake1f549022000-09-24 05:21:58 +0000184 event = self.getEvent()
Fred Drake55c38192000-06-29 19:39:57 +0000185
Fred Drake1f549022000-09-24 05:21:58 +0000186 def getEvent(self):
Fred Drake55c38192000-06-29 19:39:57 +0000187 if not self.pulldom.firstEvent[1]:
Fred Drake1f549022000-09-24 05:21:58 +0000188 self.pulldom.lastEvent = self.pulldom.firstEvent
Fred Drake55c38192000-06-29 19:39:57 +0000189 while not self.pulldom.firstEvent[1]:
Fred Drakec16adce2000-12-14 18:00:18 +0000190 buf = self.stream.read(self.bufsize)
Fred Drake55c38192000-06-29 19:39:57 +0000191 if not buf:
192 #FIXME: why doesn't Expat close work?
193 #self.parser.close()
194 return None
Fred Drake1f549022000-09-24 05:21:58 +0000195 self.parser.feed(buf)
196 rc = self.pulldom.firstEvent[1][0]
197 self.pulldom.firstEvent[1] = self.pulldom.firstEvent[1][1]
Fred Drake55c38192000-06-29 19:39:57 +0000198 return rc
199
Lars Gustäbelec964d52000-10-13 20:53:27 +0000200class SAX2DOM(PullDOM):
201
202 def startElementNS(self, name, tagName , attrs):
203 PullDOM.startElementNS(self, name, tagName, attrs)
204 self.curNode.parentNode.appendChild(self.curNode)
205
206 def startElement(self, name, attrs):
207 PullDOM.startElement(self, name, attrs)
208 self.curNode.parentNode.appendChild(self.curNode)
209
210 def processingInstruction(self, target, data):
211 PullDOM.processingInstruction(self, target, data)
212 node = self.lastEvent[0][1]
Fred Drake16f63292000-10-23 18:09:50 +0000213 node.parentNode.appendChild(node)
Lars Gustäbelec964d52000-10-13 20:53:27 +0000214
215 def ignorableWhitespace(self, chars):
216 PullDOM.ignorableWhitespace(self, chars)
217 node = self.lastEvent[0][1]
Fred Drake16f63292000-10-23 18:09:50 +0000218 node.parentNode.appendChild(node)
Lars Gustäbelec964d52000-10-13 20:53:27 +0000219
220 def characters(self, chars):
221 PullDOM.characters(self, chars)
222 node = self.lastEvent[0][1]
Fred Drake16f63292000-10-23 18:09:50 +0000223 node.parentNode.appendChild(node)
224
Fred Drakec16adce2000-12-14 18:00:18 +0000225
Fred Drake1f549022000-09-24 05:21:58 +0000226default_bufsize = (2 ** 14) - 20
227
Fred Drakec16adce2000-12-14 18:00:18 +0000228def parse(stream_or_string, parser=None, bufsize=None):
229 if bufsize is None:
230 bufsize = default_bufsize
231 if type(stream_or_string) in [type(""), type(u"")]:
Fred Drake1f549022000-09-24 05:21:58 +0000232 stream = open(stream_or_string)
Fred Drake55c38192000-06-29 19:39:57 +0000233 else:
Fred Drake1f549022000-09-24 05:21:58 +0000234 stream = stream_or_string
Fred Drake16f63292000-10-23 18:09:50 +0000235 if not parser:
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +0000236 parser = xml.sax.make_parser()
Fred Drake1f549022000-09-24 05:21:58 +0000237 return DOMEventStream(stream, parser, bufsize)
Fred Drake55c38192000-06-29 19:39:57 +0000238
Fred Drake1f549022000-09-24 05:21:58 +0000239def parseString(string, parser=None):
Fred Drake55c38192000-06-29 19:39:57 +0000240 try:
Fred Drake1f549022000-09-24 05:21:58 +0000241 from cStringIO import StringIO
Fred Drake55c38192000-06-29 19:39:57 +0000242 except ImportError:
Fred Drake1f549022000-09-24 05:21:58 +0000243 from StringIO import StringIO
Fred Drake16f63292000-10-23 18:09:50 +0000244
Fred Drake1f549022000-09-24 05:21:58 +0000245 bufsize = len(string)
246 buf = StringIO(string)
Martin v. Löwisa13a9dc2000-09-24 21:54:14 +0000247 if not parser:
248 parser = xml.sax.make_parser()
Fred Drake1f549022000-09-24 05:21:58 +0000249 return DOMEventStream(buf, parser, bufsize)