xml dom accessor

gistfile1.py

from xml.dom.minidom import parseString
from xml.dom import Node

def parse(source):
  dom = parseString(source)
  return AccessProxy(dom)

class AccessProxy(object):
  def __init__(self, dom):
    self.dom = dom

  def __getattr__(self, attr):
    nodes = [AccessProxy(node)
        for node in self.dom.getElementsByTagName(attr)]
    if nodes:
      return self.reduceNodes(nodes)
    return getattr(self.dom, attr)

  def reduceNodes(self, nodes):
    if len(nodes) > 1:
      return nodes
    childs = nodes[0].childNodes
    if len(childs) == 1:
      if childs[0].nodeType == Node.TEXT_NODE:
        return childs[0].nodeValue
    return nodes[0]

  def __str__(self):
    return self.dom.toprettyxml()

def test():
  '''
  >>> source = """
  ... <members>
  ... <member><name>a</name></member>
  ... <member><name>b</name></member>
  ... </members>"""
  >>> members = parse(source)
  >>> len(members.member)
  2
  >>> members.member[0].name
  u'a'
  >>> members.member[1].name
  u'b'
  '''
  pass

if __name__ == '__main__':
  import doctest
  doctest.testmod()