class Node:
    COMMENT_NODE = 1
    PROCESSING_INSTRUCTION_NODE = 2
    TEXT_NODE = 3
    ATTRIBUTE_NODE = 4
    DOCUMENT_TYPE_NODE = 5
    NOTATION_NODE = 6
    ENTITY_NODE = 7
    ELEMENT_NODE = 8
    
    def __init__(self, nodeType, nodeName='', nodeValue=''):
        self.nodeType = nodeType
        self.nodeName = nodeName
        self.nodeValue = nodeValue
        self.childNodes = []
        if nodeType == Node.ELEMENT_NODE:
            self.localName = nodeName

trace = []
def _get_textContents(node, depth=0):
    for child in node.childNodes:
        nodeType = child.nodeType
        if (nodeType == child.COMMENT_NODE or
            nodeType == child.PROCESSING_INSTRUCTION_NODE):
            continue
        if nodeType == child.TEXT_NODE:
            trace.append("T:%i" % depth)
            trace.append("Y:%i" % depth)
            yield child.nodeValue
        elif (nodeType in [
            child.ATTRIBUTE_NODE,
            child.DOCUMENT_TYPE_NODE,
            child.NOTATION_NODE,
            child.ENTITY_NODE,
            ]):
            continue # or you might prefer, e.g, to look inside attributes
        else:
            # iterate through elements are left
            for text in _get_textContents(child, depth+1):
                trace.append("Y:%i" % depth)
                yield text
    trace.append("R:%i" % depth)
    
trace2 = []

def _get_textContents2(node, L_res, L_test, depth=0, ):
    for child in node.childNodes:
        nodeType = child.nodeType
        if (nodeType == child.COMMENT_NODE or
            nodeType == child.PROCESSING_INSTRUCTION_NODE):
            continue
        if nodeType == child.TEXT_NODE:
            trace2.append("T:%i" % depth)
            L_test.append(child.nodeValue)
            L_res.append(child.nodeValue)
        elif (nodeType in [
            child.ATTRIBUTE_NODE,
            child.DOCUMENT_TYPE_NODE,
            child.NOTATION_NODE,
            child.ENTITY_NODE,
            ]):
            continue # or you might prefer, e.g, to look inside attributes
        elif nodeType == child.ELEMENT_NODE:
            if child.localName == "p":
                P_text = list()
                _get_textContents2(child, P_text, L_test, depth+1, )
                L_res.append(''.join(P_text))
            else:
                _get_textContents2(child, L_res, L_test, depth+1, )
        else:
            _get_textContents2(child, L_res, L_test, depth+1, )
    trace2.append("R:%i" % depth)
   
def main():

    top = Node(Node.ELEMENT_NODE, 'top')
    p1 = Node(Node.ELEMENT_NODE, 'p')
    p2 = Node(Node.ELEMENT_NODE, 'p')
    top.childNodes.append(p1)
    top.childNodes.append(p2)
    p1.childNodes.append(Node(Node.TEXT_NODE, '#text', 'foo'))
    strong = Node(Node.ELEMENT_NODE, 'strong')
    strong.childNodes.append(Node(Node.TEXT_NODE, '#text', 'bar'))
    p1.childNodes.append(strong)
    p1.childNodes.append(Node(Node.TEXT_NODE, '#text', 'baz'))
    p2.childNodes.append(Node(Node.TEXT_NODE, '#text', 'qux'))

    qa = Node(Node.ELEMENT_NODE, 'l', '#list' )
    qb = Node(Node.ELEMENT_NODE, 'i', '#item')
    p3 = Node(Node.ELEMENT_NODE, 'p')
    top.childNodes.append(qa)
    qa.childNodes.append(qb)
    qb.childNodes.append(p3)
    one = Node(Node.TEXT_NODE, '#text', 'one ')
    two = Node(Node.ELEMENT_NODE, 'even')
    two.childNodes.append(Node(Node.TEXT_NODE, '#text', 'two'))
    three = Node(Node.ELEMENT_NODE, 'odd')
    three.childNodes.append(Node(Node.TEXT_NODE, '#text', 'three'))
    four = Node(Node.ELEMENT_NODE, 'even')
    four.childNodes.append(Node(Node.TEXT_NODE, '#text', 'four'))
    five = Node(Node.TEXT_NODE, '#text', ' five')
    p3.childNodes.append(one)
    p3.childNodes.append(two)
    p3.childNodes.append(Node(Node.TEXT_NODE, '#text', ' '))
    p3.childNodes.append(three)
    p3.childNodes.append(Node(Node.TEXT_NODE, '#text', ' '))
    p3.childNodes.append(four)
    p3.childNodes.append(five)

#    assert list(_get_textContents(top)) == ['foo', 'bar', 'baz', 'qux']
#    assert ' '.join(_get_textContents(top)) == 'foo bar baz qux'

    text1 = list(_get_textContents(top))
    print text1
    print filter(lambda s: not s.startswith("T"), trace)
#    print trace

    text = list()
    text2 = list()
    _get_textContents2(top, text, text2)
    print text2
    print text
    print filter(lambda s: not s.startswith("T"), trace2)
#    print trace2

    assert text1 == text2
    assert text == ['foobarbaz', 'qux', 'one two three four five']

if __name__ == '__main__':
    main()