6 

     6 

     7 def get_text_nodes(soup):

     7 def get_text_nodes(soup):

     8     return soup(text=lambda text:not isinstance(text, Comment))

     8     return soup(text=lambda text:not isinstance(text, Comment))

     9 

     9 

    10 def is_real_text_node(textNode):

    10 def is_real_text_node(textNode):

    11     return not textNode.findParent('style') 

    13     return not textNode.findParent('style') 

    12 

    14 

    13 def get_the_soup(input):

    15 def get_the_soup(input):

    14     return BeautifulSoup(input, convertEntities=BeautifulSoup.ALL_ENTITIES)

    16     return BeautifulSoup(input, convertEntities=BeautifulSoup.ALL_ENTITIES)

    15      

    17