From d89c8878ea86a2575c87e9fad8081cfcd81e0bcd Mon Sep 17 00:00:00 2001 From: Leonard Richardson Date: Thu, 10 Feb 2011 16:41:10 -0500 Subject: Added some elementary doctype handling. --- beautifulsoup/builder/lxml_builder.py | 41 +++++++++++++++++++++++++++++++++++ 1 file changed, 41 insertions(+) create mode 100644 beautifulsoup/builder/lxml_builder.py (limited to 'beautifulsoup/builder/lxml_builder.py') diff --git a/beautifulsoup/builder/lxml_builder.py b/beautifulsoup/builder/lxml_builder.py new file mode 100644 index 0000000..8336ab4 --- /dev/null +++ b/beautifulsoup/builder/lxml_builder.py @@ -0,0 +1,41 @@ +from lxml import etree +from beautifulsoup.element import Comment, Doctype +from beautifulsoup.builder import HTMLTreeBuilder + +class LXMLTreeBuilder(HTMLTreeBuilder): + + def __init__(self, parser_class=etree.HTMLParser): + self.parser = parser_class(target=self) + self.soup = None + + def feed(self, markup): + self.parser.feed(markup) + self.parser.close() + + def close(self): + pass + + def start(self, name, attrs): + self.soup.handle_starttag(name, attrs) + + def end(self, name): + self.soup.handle_endtag(name) + + def data(self, content): + self.soup.handle_data(content) + + def doctype(self, name, pubid, system): + self.soup.endData() + self.soup.handle_data(name) + self.soup.endData(Doctype) + + def comment(self, content): + "Handle comments as Comment objects." + self.soup.endData() + self.soup.handle_data(content) + self.soup.endData(Comment) + + def test_fragment_to_document(self, fragment): + """See `TreeBuilder`.""" + return u'%s' % fragment + -- cgit v1.2.3