util.py 文件源码

def get_tree(page):
    page = page.replace(" ", " ")  # otherwise starts-with for lxml doesn't work
    try:
        tree = html.fromstring(page)
    except (etree.XMLSyntaxError, etree.ParserError) as e:
        print u"not parsing, beause etree error in get_tree: {}".format(e)
        tree = None
    return tree