Mercurial > traipse_dev
annotate orpg/pulldom.py @ 39:93a27aea09ce traipse_dev
Going to post some daily builds as advancements happen. Repo List now updates.
author | sirebral |
---|---|
date | Wed, 05 Aug 2009 08:16:32 -0500 |
parents | 551cd440acce |
children |
rev | line source |
---|---|
0
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
1 import minidom |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
2 import xml.sax,xml.sax.handler |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
3 import types |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
4 |
23
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
5 try: from cStringIO import StringIO |
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
6 except ImportError: from StringIO import StringIO |
0
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
7 |
23
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
8 try: _StringTypes = [types.StringType, types.UnicodeType] |
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
9 except AttributeError: _StringTypes = [types.StringType] |
0
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
10 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
11 START_ELEMENT = "START_ELEMENT" |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
12 END_ELEMENT = "END_ELEMENT" |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
13 COMMENT = "COMMENT" |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
14 START_DOCUMENT = "START_DOCUMENT" |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
15 END_DOCUMENT = "END_DOCUMENT" |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
16 PROCESSING_INSTRUCTION = "PROCESSING_INSTRUCTION" |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
17 IGNORABLE_WHITESPACE = "IGNORABLE_WHITESPACE" |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
18 CHARACTERS = "CHARACTERS" |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
19 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
20 class PullDOM(xml.sax.ContentHandler): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
21 def __init__(self): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
22 self.firstEvent = [None, None] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
23 self.lastEvent = self.firstEvent |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
24 self._ns_contexts = [{}] # contains uri -> prefix dicts |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
25 self._current_context = self._ns_contexts[-1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
26 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
27 def setDocumentLocator(self, locator): pass |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
28 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
29 def startPrefixMapping(self, prefix, uri): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
30 self._ns_contexts.append(self._current_context.copy()) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
31 self._current_context[uri] = prefix |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
32 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
33 def endPrefixMapping(self, prefix): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
34 del self._ns_contexts[-1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
35 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
36 def startElementNS(self, name, tagName , attrs): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
37 uri,localname = name |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
38 if uri: |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
39 # When using namespaces, the reader may or may not |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
40 # provide us with the original name. If not, create |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
41 # *a* valid tagName from the current context. |
23
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
42 if tagName is None: tagName = self._current_context[uri] + ":" + localname |
0
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
43 node = self.document.createElementNS(uri, tagName) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
44 else: |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
45 # When the tagname is not prefixed, it just appears as |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
46 # localname |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
47 node = self.document.createElement(localname) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
48 for aname,value in attrs.items(): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
49 a_uri, a_localname = aname |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
50 if a_uri: |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
51 qname = self._current_context[a_uri] + ":" + a_localname |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
52 attr = self.document.createAttributeNS(a_uri, qname) |
23
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
53 else: attr = self.document.createAttribute(a_localname) |
0
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
54 attr.value = value |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
55 node.setAttributeNode(attr) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
56 parent = self.curNode |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
57 node.parentNode = parent |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
58 self.curNode = node |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
59 self.lastEvent[1] = [(START_ELEMENT, node), None] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
60 self.lastEvent = self.lastEvent[1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
61 #self.events.append((START_ELEMENT, node)) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
62 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
63 def endElementNS(self, name, tagName): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
64 node = self.curNode |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
65 self.lastEvent[1] = [(END_ELEMENT, node), None] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
66 self.lastEvent = self.lastEvent[1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
67 #self.events.append((END_ELEMENT, node)) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
68 self.curNode = node.parentNode |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
69 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
70 def startElement(self, name, attrs): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
71 node = self.document.createElement(name) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
72 for aname,value in attrs.items(): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
73 attr = self.document.createAttribute(aname) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
74 attr.value = value |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
75 node.setAttributeNode(attr) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
76 parent = self.curNode |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
77 node.parentNode = parent |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
78 self.curNode = node |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
79 self.lastEvent[1] = [(START_ELEMENT, node), None] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
80 self.lastEvent = self.lastEvent[1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
81 #self.events.append((START_ELEMENT, node)) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
82 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
83 def endElement(self, name): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
84 node = self.curNode |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
85 self.lastEvent[1] = [(END_ELEMENT, node), None] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
86 self.lastEvent = self.lastEvent[1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
87 #self.events.append((END_ELEMENT, node)) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
88 self.curNode = node.parentNode |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
89 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
90 def comment(self, s): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
91 node = self.document.createComment(s) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
92 parent = self.curNode |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
93 node.parentNode = parent |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
94 self.lastEvent[1] = [(COMMENT, node), None] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
95 self.lastEvent = self.lastEvent[1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
96 #self.events.append((COMMENT, node)) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
97 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
98 def processingInstruction(self, target, data): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
99 node = self.document.createProcessingInstruction(target, data) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
100 parent = self.curNode |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
101 node.parentNode = parent |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
102 self.lastEvent[1] = [(PROCESSING_INSTRUCTION, node), None] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
103 self.lastEvent = self.lastEvent[1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
104 #self.events.append((PROCESSING_INSTRUCTION, node)) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
105 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
106 def ignorableWhitespace(self, chars): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
107 node = self.document.createTextNode(chars[start:start + length]) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
108 parent = self.curNode |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
109 node.parentNode = parent |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
110 self.lastEvent[1] = [(IGNORABLE_WHITESPACE, node), None] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
111 self.lastEvent = self.lastEvent[1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
112 #self.events.append((IGNORABLE_WHITESPACE, node)) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
113 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
114 def characters(self, chars): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
115 node = self.document.createTextNode(chars) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
116 parent = self.curNode |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
117 node.parentNode = parent |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
118 self.lastEvent[1] = [(CHARACTERS, node), None] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
119 self.lastEvent = self.lastEvent[1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
120 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
121 def startDocument(self): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
122 node = self.curNode = self.document = minidom.Document() |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
123 node.parentNode = None |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
124 self.lastEvent[1] = [(START_DOCUMENT, node), None] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
125 self.lastEvent = self.lastEvent[1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
126 #self.events.append((START_DOCUMENT, node)) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
127 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
128 def endDocument(self): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
129 assert not self.curNode.parentNode |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
130 for node in self.curNode.childNodes: |
23
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
131 if node.nodeType == node.ELEMENT_NODE: self.document.documentElement = node |
0
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
132 #if not self.document.documentElement: |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
133 # raise Error, "No document element" |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
134 self.lastEvent[1] = [(END_DOCUMENT, node), None] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
135 #self.events.append((END_DOCUMENT, self.curNode)) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
136 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
137 class ErrorHandler: |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
138 def warning(self, exception): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
139 print exception |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
140 def error(self, exception): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
141 raise exception |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
142 def fatalError(self, exception): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
143 raise exception |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
144 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
145 class DOMEventStream: |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
146 def __init__(self, stream, parser, bufsize): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
147 self.stream = stream |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
148 self.parser = parser |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
149 self.bufsize = bufsize |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
150 self.reset() |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
151 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
152 def reset(self): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
153 self.pulldom = PullDOM() |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
154 # This content handler relies on namespace support |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
155 self.parser.setFeature(xml.sax.handler.feature_namespaces,1) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
156 self.parser.setContentHandler(self.pulldom) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
157 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
158 def __getitem__(self, pos): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
159 rc = self.getEvent() |
23
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
160 if rc: return rc |
0
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
161 raise IndexError |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
162 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
163 def expandNode(self, node): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
164 event = self.getEvent() |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
165 while event: |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
166 token, cur_node = event |
23
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
167 if cur_node is node: return |
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
168 if token != END_ELEMENT: cur_node.parentNode.appendChild(cur_node) |
0
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
169 event = self.getEvent() |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
170 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
171 def getEvent(self): |
23
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
172 if not self.pulldom.firstEvent[1]: self.pulldom.lastEvent = self.pulldom.firstEvent |
0
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
173 while not self.pulldom.firstEvent[1]: |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
174 buf=self.stream.read(self.bufsize) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
175 if not buf: |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
176 #FIXME: why doesn't Expat close work? |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
177 #self.parser.close() |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
178 return None |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
179 self.parser.feed(buf) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
180 rc = self.pulldom.firstEvent[1][0] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
181 self.pulldom.firstEvent[1] = self.pulldom.firstEvent[1][1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
182 return rc |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
183 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
184 class SAX2DOM(PullDOM): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
185 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
186 def startElementNS(self, name, tagName , attrs): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
187 PullDOM.startElementNS(self, name, tagName, attrs) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
188 self.curNode.parentNode.appendChild(self.curNode) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
189 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
190 def startElement(self, name, attrs): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
191 PullDOM.startElement(self, name, attrs) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
192 self.curNode.parentNode.appendChild(self.curNode) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
193 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
194 def processingInstruction(self, target, data): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
195 PullDOM.processingInstruction(self, target, data) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
196 node = self.lastEvent[0][1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
197 node.parentNode.appendChild(node) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
198 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
199 def ignorableWhitespace(self, chars): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
200 PullDOM.ignorableWhitespace(self, chars) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
201 node = self.lastEvent[0][1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
202 node.parentNode.appendChild(node) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
203 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
204 def characters(self, chars): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
205 PullDOM.characters(self, chars) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
206 node = self.lastEvent[0][1] |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
207 node.parentNode.appendChild(node) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
208 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
209 default_bufsize = (2 ** 14) - 20 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
210 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
211 def parse(stream_or_string, parser=None, bufsize=default_bufsize): |
23
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
212 if type(stream_or_string) in _StringTypes: stream = open(stream_or_string) |
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
213 else: stream = stream_or_string |
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
214 if not parser: parser = xml.sax.make_parser() |
0
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
215 return DOMEventStream(stream, parser, bufsize) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
216 |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
217 def parseString(string, parser=None): |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
218 bufsize = len(string) |
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
219 buf = StringIO(string) |
23
551cd440acce
Final Dev build of the Day. Should reduce RAM usage more.
sirebral
parents:
0
diff
changeset
|
220 if not parser: parser = xml.sax.make_parser() |
0
4385a7d0efd1
Deleted and repushed it with the 'grumpy-goblin' branch. I forgot a y
sirebral
parents:
diff
changeset
|
221 return DOMEventStream(buf, parser, bufsize) |