def parse(xml):
tokens = tokenize(xml)
#skip anything before root node
start_i = 0
while start_i < len(tokens) and not isinstance(tokens[start_i], StartTag):
start_i += 1
end_i = findEnd(tokens, start_i)
root = Node()
root.tagname = tokens[start_i].tagname
root.attributes = tokens[start_i].attributes
children = parse_node(tokens, start_i+1, end_i-1)
root.childNodes = children
doc = XMLDocument(root)
if isinstance(tokens[0], Prolog):
doc.prolog = tokens[0]
return doc
def parse_node(elements, start_i, end_i):
"""
start_i end_i
<tag>................</tag>
"""
nodes = []
i = start_i
while i <= end_i:
if isinstance(elements[i], TextNode):
nodes.append(elements[i])
i += 1
elif isinstance(elements[i], StartTag):
end_j = findEnd(elements, i)
n = Node()
n.tagname = elements[i].tagname
n.attributes = elements[i].attributes
children = parse_node(elements, i+1, end_j-1)
n.childNodes = children
nodes.append(n)
i = end_j+1
elif isinstance(elements[i], SelfClosingTag):
n = Node(True)
n.tagname = elements[i].tagname
n.attributes = elements[i].attributes
nodes.append(n)
i += 1
elif isinstance(elements[i], EndTag):
i += 1
else:
raise ValueError("unknown element", elements[i])
return nodes