from xml.dom import pulldom
import sys

def getInnerText(node):
    """
    Get all the inner text of a DOM node (recursively).
    """
    # inspired by http://mail.python.org/pipermail/xml-sig/2005-March/011022.html
    inner_text = []
    for child in node.childNodes:
        if child.nodeType == child.TEXT_NODE or child.nodeType == child.CDATA_SECTION_NODE:
            inner_text.append(child.data)
        elif child.nodeType == child.ELEMENT_NODE:
            inner_text.extend(getInnerText(child))
        else:
           pass
    print repr(inner_text)
    return u"".join(inner_text)

class Deserializer(object):
    def __init__(self, stream):
        self.stream = stream
        self.event_stream = pulldom.parse(self.stream)

    def __iter__(self):
        return self

    def next(self):
        for event, node in self.event_stream:
            if event == "START_ELEMENT" and node.nodeName == "object":
                self.event_stream.expandNode(node)
                for field_node in node.getElementsByTagName("field"):
                    getInnerText(field_node)
        raise StopIteration

fixture = file('unicode.xml', 'r')
for elements in Deserializer(fixture):
    pass