def _getNextText(element):
while element is not None:
if lxmlutil.isVisibleElement(element) and element.text:
return element.text.strip()
elif element.tail:
return element.tail.strip()
element = lxmlutil.getFullNext(element)
return None
def _getChildTextLength(element):
result = 0
for item in element.getchildren():
if lxmlutil.isVisibleElement(item):
result += _getParagraphLengthByLink(item)
if item.tail:
result += len(item.tail.strip())
return result
def _getMaxChildTag(element):
result = {}
for item in element.getchildren():
textlen = 0
if lxmlutil.isVisibleElement(item):
textlen += _getParagraphLengthByLink(item)
if item.tail:
textlen += len(item.tail.strip())
if item.tag in result:
result[item.tag] += textlen
else:
result[item.tag] = textlen
return max(result.iterkeys(), key=(lambda key: result[key]))