def test_char_count(self): # {{{ from calibre.srv.render_book import get_length from calibre.ebooks.oeb.parse_utils import html5_parse root = html5_parse('<p>a b\nc\td\re') self.ae(get_length(root), 5) root = html5_parse('<script>xyz</script>a<iMg>b') self.ae(get_length(root), 1002) root = html5_parse('<p><!-- abc -->m') self.ae(get_length(root), 1)
def t(html, body_children, nsmap=('http://www.w3.org/1999/xhtml', )): root = html5_parse(html) raw = html_as_json(root) # print(raw.decode('utf-8')) data = json.loads(raw) self.ae(data['version'], 1) self.ae(tuple(data['ns_map']), nsmap) bc = data['tree']['c'][1]['c'] self.ae(bc, body_children)