| 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 |
28
|
aaronmk
|
import xml.dom.minidom
|
| 7 |
21
|
aaronmk
|
|
| 8 |
73
|
aaronmk
|
import strings
|
| 9 |
|
|
|
| 10 |
|
|
def escape(str_):
|
| 11 |
|
|
return strings.to_unicode(cgi.escape(str_, True)).encode('ascii',
|
| 12 |
|
|
'xmlcharrefreplace')
|
| 13 |
|
|
|
| 14 |
|
|
def unescape(str_): return HTMLParser().unescape(str_)
|
| 15 |
|
|
|
| 16 |
21
|
aaronmk
|
def name_of(node): return node.tagName.lower()
|
| 17 |
|
|
|
| 18 |
|
|
def get_id(node): return node.getAttribute('id')
|
| 19 |
|
|
|
| 20 |
|
|
def set_id(node, id_): node.setAttribute('id', id_)
|
| 21 |
|
|
|
| 22 |
|
|
class NodeElemIter:
|
| 23 |
|
|
def __init__(self, node): self.child = node.firstChild
|
| 24 |
|
|
|
| 25 |
|
|
def __iter__(self): return self
|
| 26 |
|
|
|
| 27 |
|
|
def curr(self):
|
| 28 |
|
|
while self.child != None:
|
| 29 |
|
|
if self.child.nodeType == Node.ELEMENT_NODE: return self.child
|
| 30 |
|
|
self.child = self.child.nextSibling
|
| 31 |
|
|
raise StopIteration
|
| 32 |
|
|
|
| 33 |
|
|
def next(self):
|
| 34 |
|
|
child = self.curr()
|
| 35 |
|
|
self.child = self.child.nextSibling
|
| 36 |
|
|
return child
|
| 37 |
|
|
|
| 38 |
|
|
def first_elem(node): return NodeElemIter(node).next()
|
| 39 |
|
|
|
| 40 |
|
|
class NodeElemReverseIter:
|
| 41 |
|
|
def __init__(self, node): self.child = node.lastChild
|
| 42 |
|
|
|
| 43 |
|
|
def __iter__(self): return self
|
| 44 |
|
|
|
| 45 |
|
|
def curr(self):
|
| 46 |
|
|
while self.child != None:
|
| 47 |
|
|
if self.child.nodeType == Node.ELEMENT_NODE: return self.child
|
| 48 |
|
|
self.child = self.child.previousSibling
|
| 49 |
|
|
raise StopIteration
|
| 50 |
|
|
|
| 51 |
|
|
def next(self):
|
| 52 |
|
|
child = self.curr()
|
| 53 |
|
|
self.child = self.child.previousSibling
|
| 54 |
|
|
return child
|
| 55 |
|
|
|
| 56 |
|
|
def last_elem(node): return NodeElemReverseIter(node).next()
|
| 57 |
|
|
|
| 58 |
|
|
class NodeParentIter:
|
| 59 |
|
|
def __init__(self, node): self.node = node
|
| 60 |
|
|
|
| 61 |
|
|
def __iter__(self): return self
|
| 62 |
|
|
|
| 63 |
|
|
def curr(self):
|
| 64 |
|
|
if self.node != None and self.node.nodeType == Node.ELEMENT_NODE:
|
| 65 |
|
|
return self.node
|
| 66 |
|
|
raise StopIteration
|
| 67 |
|
|
|
| 68 |
|
|
def next(self):
|
| 69 |
|
|
node = self.curr()
|
| 70 |
|
|
self.node = self.node.parentNode
|
| 71 |
|
|
return node
|
| 72 |
|
|
|
| 73 |
|
|
def is_text(node):
|
| 74 |
|
|
for child in NodeElemIter(node): return False # has an element node
|
| 75 |
|
|
return True
|
| 76 |
|
|
|
| 77 |
|
|
def value(node):
|
| 78 |
29
|
aaronmk
|
if node.firstChild != None: return node.firstChild.nodeValue
|
| 79 |
21
|
aaronmk
|
else: return node.nodeValue
|
| 80 |
|
|
|
| 81 |
22
|
aaronmk
|
def set_value(doc, node, value):
|
| 82 |
|
|
if node.nodeType == Node.ELEMENT_NODE:
|
| 83 |
|
|
node.appendChild(doc.createTextNode(value))
|
| 84 |
|
|
else: node.nodeValue = value
|
| 85 |
|
|
|
| 86 |
86
|
aaronmk
|
class NodeTextEntryIter:
|
| 87 |
|
|
def __init__(self, node): self.iter_ = NodeElemIter(node)
|
| 88 |
|
|
|
| 89 |
|
|
def __iter__(self): return self
|
| 90 |
|
|
|
| 91 |
|
|
def curr(self):
|
| 92 |
|
|
while True:
|
| 93 |
|
|
child = self.iter_.curr()
|
| 94 |
|
|
if is_text(child): return (name_of(child), value(child))
|
| 95 |
|
|
self.iter_.next()
|
| 96 |
|
|
|
| 97 |
|
|
def next(self):
|
| 98 |
|
|
entry = self.curr()
|
| 99 |
|
|
self.iter_.next()
|
| 100 |
|
|
return entry
|
| 101 |
|
|
|
| 102 |
|
|
def replace(old_node, new_node):
|
| 103 |
|
|
old_node.parentNode.replaceChild(new_node, old_node) # note order reversed
|
| 104 |
|
|
|
| 105 |
|
|
def replace_with_text(doc, node, str_): replace(node, doc.createTextNode(str_))
|
| 106 |
|
|
|
| 107 |
22
|
aaronmk
|
def by_tag_name(node, name, last_only=False):
|
| 108 |
23
|
aaronmk
|
# last_only optimization returns last (most recently inserted) matching node
|
| 109 |
22
|
aaronmk
|
children = []
|
| 110 |
21
|
aaronmk
|
for child in NodeElemReverseIter(node):
|
| 111 |
22
|
aaronmk
|
if child.tagName == name:
|
| 112 |
|
|
children.append(child)
|
| 113 |
|
|
if last_only: break
|
| 114 |
|
|
return children
|
| 115 |
28
|
aaronmk
|
|
| 116 |
133
|
aaronmk
|
def create_doc(root='_'):
|
| 117 |
|
|
return xml.dom.minidom.getDOMImplementation().createDocument(None, root,
|
| 118 |
|
|
None)
|
| 119 |
|
|
|
| 120 |
73
|
aaronmk
|
# xml.dom.minidom modifications
|
| 121 |
|
|
|
| 122 |
|
|
def _write_data(writer, data): writer.write(escape(data))
|
| 123 |
|
|
|
| 124 |
|
|
xml.dom.minidom._write_data = _write_data
|
| 125 |
|
|
|
| 126 |
28
|
aaronmk
|
_writexml_orig = xml.dom.minidom.Element.writexml
|
| 127 |
|
|
|
| 128 |
|
|
def _writexml(self, writer, indent="", addindent="", newl=""):
|
| 129 |
36
|
aaronmk
|
if self.firstChild != None and self.firstChild.nextSibling == None\
|
| 130 |
|
|
and self.firstChild.nodeType == Node.TEXT_NODE: # a single text node
|
| 131 |
28
|
aaronmk
|
writer.write(indent+'<'+self.tagName)
|
| 132 |
70
|
aaronmk
|
for attr_idx in xrange(self.attributes.length):
|
| 133 |
28
|
aaronmk
|
attr = self.attributes.item(attr_idx)
|
| 134 |
73
|
aaronmk
|
writer.write(' '+attr.name+'='+escape(attr.value))
|
| 135 |
|
|
writer.write('>'+escape(value(self))+'</'+self.tagName+'>'+newl)
|
| 136 |
28
|
aaronmk
|
else: _writexml_orig(self, writer, indent, addindent, newl)
|
| 137 |
|
|
|
| 138 |
|
|
xml.dom.minidom.Element.writexml = _writexml
|
