1 |
21
|
aaronmk
|
# XML DOM tree manipulation
|
2 |
|
|
|
3 |
73
|
aaronmk
|
import cgi
|
4 |
|
|
from HTMLParser import HTMLParser
|
5 |
21
|
aaronmk
|
from xml.dom import Node
|
6 |
299
|
aaronmk
|
import xml.dom.minidom as minidom
|
7 |
21
|
aaronmk
|
|
8 |
73
|
aaronmk
|
import strings
|
9 |
331
|
aaronmk
|
import util
|
10 |
73
|
aaronmk
|
|
11 |
|
|
def escape(str_):
|
12 |
|
|
return strings.to_unicode(cgi.escape(str_, True)).encode('ascii',
|
13 |
|
|
'xmlcharrefreplace')
|
14 |
|
|
|
15 |
|
|
def unescape(str_): return HTMLParser().unescape(str_)
|
16 |
|
|
|
17 |
21
|
aaronmk
|
def get_id(node): return node.getAttribute('id')
|
18 |
|
|
|
19 |
|
|
def set_id(node, id_): node.setAttribute('id', id_)
|
20 |
|
|
|
21 |
453
|
aaronmk
|
def is_completely_empty(node): return node.firstChild == None
|
22 |
135
|
aaronmk
|
|
23 |
301
|
aaronmk
|
def has_one_child(node):
|
24 |
|
|
return node.firstChild != None and node.firstChild.nextSibling == None
|
25 |
|
|
|
26 |
305
|
aaronmk
|
class NodeIter:
|
27 |
|
|
def __init__(self, node): self.child = node.firstChild
|
28 |
|
|
|
29 |
|
|
def __iter__(self): return self
|
30 |
|
|
|
31 |
|
|
def curr(self):
|
32 |
|
|
if self.child != None: return self.child
|
33 |
|
|
raise StopIteration
|
34 |
|
|
|
35 |
|
|
def next(self):
|
36 |
|
|
child = self.curr()
|
37 |
|
|
self.child = self.child.nextSibling
|
38 |
|
|
return child
|
39 |
|
|
|
40 |
453
|
aaronmk
|
def is_comment(node): return node.nodeType == Node.COMMENT_NODE
|
41 |
|
|
|
42 |
298
|
aaronmk
|
def is_elem(node): return node.nodeType == Node.ELEMENT_NODE
|
43 |
|
|
|
44 |
453
|
aaronmk
|
def is_empty(node):
|
45 |
|
|
for child in NodeIter(node):
|
46 |
|
|
if not is_comment(child): return False
|
47 |
|
|
return True
|
48 |
|
|
|
49 |
21
|
aaronmk
|
class NodeElemIter:
|
50 |
|
|
def __init__(self, node): self.child = node.firstChild
|
51 |
|
|
|
52 |
|
|
def __iter__(self): return self
|
53 |
|
|
|
54 |
|
|
def curr(self):
|
55 |
|
|
while self.child != None:
|
56 |
298
|
aaronmk
|
if is_elem(self.child): return self.child
|
57 |
21
|
aaronmk
|
self.child = self.child.nextSibling
|
58 |
|
|
raise StopIteration
|
59 |
|
|
|
60 |
|
|
def next(self):
|
61 |
|
|
child = self.curr()
|
62 |
|
|
self.child = self.child.nextSibling
|
63 |
|
|
return child
|
64 |
|
|
|
65 |
|
|
def first_elem(node): return NodeElemIter(node).next()
|
66 |
|
|
|
67 |
450
|
aaronmk
|
def has_elems(node):
|
68 |
|
|
try: first_elem(node); return True
|
69 |
|
|
except StopIteration: return False
|
70 |
|
|
|
71 |
21
|
aaronmk
|
class NodeElemReverseIter:
|
72 |
|
|
def __init__(self, node): self.child = node.lastChild
|
73 |
|
|
|
74 |
|
|
def __iter__(self): return self
|
75 |
|
|
|
76 |
|
|
def curr(self):
|
77 |
|
|
while self.child != None:
|
78 |
298
|
aaronmk
|
if is_elem(self.child): return self.child
|
79 |
21
|
aaronmk
|
self.child = self.child.previousSibling
|
80 |
|
|
raise StopIteration
|
81 |
|
|
|
82 |
|
|
def next(self):
|
83 |
|
|
child = self.curr()
|
84 |
|
|
self.child = self.child.previousSibling
|
85 |
|
|
return child
|
86 |
|
|
|
87 |
|
|
def last_elem(node): return NodeElemReverseIter(node).next()
|
88 |
|
|
|
89 |
|
|
class NodeParentIter:
|
90 |
|
|
def __init__(self, node): self.node = node
|
91 |
|
|
|
92 |
|
|
def __iter__(self): return self
|
93 |
|
|
|
94 |
|
|
def curr(self):
|
95 |
298
|
aaronmk
|
if self.node != None and is_elem(self.node): return self.node
|
96 |
21
|
aaronmk
|
raise StopIteration
|
97 |
|
|
|
98 |
|
|
def next(self):
|
99 |
|
|
node = self.curr()
|
100 |
|
|
self.node = self.node.parentNode
|
101 |
|
|
return node
|
102 |
|
|
|
103 |
298
|
aaronmk
|
def is_text_node(node): return node.nodeType == Node.TEXT_NODE
|
104 |
|
|
|
105 |
301
|
aaronmk
|
def is_text(node): return has_one_child(node) and is_text_node(node.firstChild)
|
106 |
21
|
aaronmk
|
|
107 |
|
|
def value(node):
|
108 |
29
|
aaronmk
|
if node.firstChild != None: return node.firstChild.nodeValue
|
109 |
21
|
aaronmk
|
else: return node.nodeValue
|
110 |
|
|
|
111 |
143
|
aaronmk
|
def set_value(node, value):
|
112 |
298
|
aaronmk
|
if is_elem(node): node.appendChild(node.ownerDocument.createTextNode(value))
|
113 |
22
|
aaronmk
|
else: node.nodeValue = value
|
114 |
|
|
|
115 |
86
|
aaronmk
|
class NodeTextEntryIter:
|
116 |
|
|
def __init__(self, node): self.iter_ = NodeElemIter(node)
|
117 |
|
|
|
118 |
|
|
def __iter__(self): return self
|
119 |
|
|
|
120 |
|
|
def curr(self):
|
121 |
|
|
while True:
|
122 |
|
|
child = self.iter_.curr()
|
123 |
139
|
aaronmk
|
if is_text(child): return (child.tagName, value(child))
|
124 |
86
|
aaronmk
|
self.iter_.next()
|
125 |
|
|
|
126 |
|
|
def next(self):
|
127 |
|
|
entry = self.curr()
|
128 |
|
|
self.iter_.next()
|
129 |
|
|
return entry
|
130 |
|
|
|
131 |
135
|
aaronmk
|
def set_child(node, name, value):
|
132 |
|
|
'''Note: does not remove any existing child of the same name'''
|
133 |
|
|
child = node.ownerDocument.createElement(name)
|
134 |
143
|
aaronmk
|
set_value(child, value)
|
135 |
135
|
aaronmk
|
node.appendChild(child)
|
136 |
|
|
|
137 |
435
|
aaronmk
|
def remove(node): node.parentNode.removeChild(node)
|
138 |
|
|
|
139 |
86
|
aaronmk
|
def replace(old_node, new_node):
|
140 |
|
|
old_node.parentNode.replaceChild(new_node, old_node) # note order reversed
|
141 |
|
|
|
142 |
142
|
aaronmk
|
def replace_with_text(node, str_):
|
143 |
|
|
replace(node, node.ownerDocument.createTextNode(str_))
|
144 |
86
|
aaronmk
|
|
145 |
22
|
aaronmk
|
def by_tag_name(node, name, last_only=False):
|
146 |
135
|
aaronmk
|
'''last_only optimization returns last matching node'''
|
147 |
22
|
aaronmk
|
children = []
|
148 |
21
|
aaronmk
|
for child in NodeElemReverseIter(node):
|
149 |
22
|
aaronmk
|
if child.tagName == name:
|
150 |
|
|
children.append(child)
|
151 |
|
|
if last_only: break
|
152 |
|
|
return children
|
153 |
28
|
aaronmk
|
|
154 |
133
|
aaronmk
|
def create_doc(root='_'):
|
155 |
303
|
aaronmk
|
return minidom.getDOMImplementation().createDocument(None, root, None)
|
156 |
133
|
aaronmk
|
|
157 |
304
|
aaronmk
|
prettyxml_config = dict(addindent=' ', newl='\n')
|
158 |
331
|
aaronmk
|
toprettyxml_config = prettyxml_config.copy()
|
159 |
|
|
util.rename_key(toprettyxml_config, 'addindent', 'indent')
|
160 |
304
|
aaronmk
|
|
161 |
299
|
aaronmk
|
# minidom modifications
|
162 |
73
|
aaronmk
|
|
163 |
299
|
aaronmk
|
minidom._write_data = lambda writer, data: writer.write(escape(data))
|
164 |
73
|
aaronmk
|
|
165 |
305
|
aaronmk
|
minidom.Node.__iter__ = lambda self: NodeIter(self)
|
166 |
|
|
|
167 |
331
|
aaronmk
|
minidom.Node.__str__ = lambda self: self.toprettyxml(**toprettyxml_config)
|
168 |
301
|
aaronmk
|
|
169 |
315
|
aaronmk
|
minidom.Node.pop = lambda self: self.removeChild(self.lastChild)
|
170 |
|
|
|
171 |
|
|
def __Node_clear(self):
|
172 |
|
|
while not is_empty(self): self.pop()
|
173 |
|
|
minidom.Node.clear = __Node_clear
|
174 |
|
|
|
175 |
301
|
aaronmk
|
def __Element_write_opening(self, writer, indent='', addindent='', newl=''):
|
176 |
298
|
aaronmk
|
writer.write(indent+'<'+self.tagName)
|
177 |
|
|
for attr_idx in xrange(self.attributes.length):
|
178 |
|
|
attr = self.attributes.item(attr_idx)
|
179 |
|
|
writer.write(' '+attr.name+'='+escape(attr.value))
|
180 |
|
|
writer.write('>'+newl)
|
181 |
301
|
aaronmk
|
minidom.Element.write_opening = __Element_write_opening
|
182 |
73
|
aaronmk
|
|
183 |
301
|
aaronmk
|
def __Element_write_closing(self, writer, indent='', addindent='', newl=''):
|
184 |
298
|
aaronmk
|
writer.write('</'+self.tagName+'>'+newl)
|
185 |
301
|
aaronmk
|
minidom.Element.write_closing = __Element_write_closing
|
186 |
298
|
aaronmk
|
|
187 |
299
|
aaronmk
|
_writexml_orig = minidom.Element.writexml
|
188 |
301
|
aaronmk
|
def __Element_writexml(self, writer, indent='', addindent='', newl=''):
|
189 |
306
|
aaronmk
|
if isinstance(indent, int): indent = addindent*indent
|
190 |
298
|
aaronmk
|
if is_text(self):
|
191 |
|
|
self.write_opening(writer, indent, addindent, '') # no newline
|
192 |
|
|
writer.write(escape(value(self)))
|
193 |
|
|
self.write_closing(writer, indent, addindent, newl)
|
194 |
28
|
aaronmk
|
else: _writexml_orig(self, writer, indent, addindent, newl)
|
195 |
301
|
aaronmk
|
minidom.Element.writexml = __Element_writexml
|
196 |
28
|
aaronmk
|
|
197 |
301
|
aaronmk
|
def __Document_write_opening(self, writer, indent='', addindent='', newl='',
|
198 |
|
|
encoding=None):
|
199 |
|
|
xmlDecl = '<?xml version="1.0" '
|
200 |
|
|
if encoding != None: xmlDecl += 'encoding="'+escape(encoding)+'"'
|
201 |
|
|
xmlDecl += '?>'+newl
|
202 |
|
|
writer.write(xmlDecl)
|
203 |
|
|
assert has_one_child(self)
|
204 |
|
|
assert is_elem(self.firstChild)
|
205 |
|
|
self.firstChild.write_opening(writer, indent, addindent, newl)
|
206 |
|
|
minidom.Document.write_opening = __Document_write_opening
|
207 |
|
|
|
208 |
|
|
def __Document_write_closing(self, writer, indent='', addindent='', newl=''):
|
209 |
|
|
self.firstChild.write_closing(writer, indent, addindent, newl)
|
210 |
|
|
minidom.Document.write_closing = __Document_write_closing
|