HTML2Text

HTML2Text.py

def HTML2Text(html):
    
    class _HTMLParser(HTMLParser):
        
        def __init__(self):
            HTMLParser.__init__(self)
            self._text = []
     
        def handle_data(self, data):
            append = True
            text = data.split()
            if text:
                tag = self.get_starttag_text()
                if tag:
                    tag = tag.lower()
                    append = not tag.startswith(('<script','<style'))
                if append:
                    self._text.extend(text)

    h = _HTMLParser()
    h.feed(html)
    return ' '.join(h._text)