diff options
author | Leonard Richardson <leonard.richardson@canonical.com> | 2009-04-10 15:20:58 -0400 |
---|---|---|
committer | Leonard Richardson <leonard.richardson@canonical.com> | 2009-04-10 15:20:58 -0400 |
commit | 522c72ef59e2b544a30a6bfbc3001dc1e84fed45 (patch) | |
tree | 32f1d105d930207d2dd9b822e4051e2cb8f987d6 /src/beautifulsoup/builder/lxml_builder.py | |
parent | 07874f2172a7402306582ba9c5f5ceb92f0f5f44 (diff) |
Added a simple compatibility test suite for tree builders.
Diffstat (limited to 'src/beautifulsoup/builder/lxml_builder.py')
-rw-r--r-- | src/beautifulsoup/builder/lxml_builder.py | 35 |
1 files changed, 35 insertions, 0 deletions
diff --git a/src/beautifulsoup/builder/lxml_builder.py b/src/beautifulsoup/builder/lxml_builder.py new file mode 100644 index 0000000..2963a02 --- /dev/null +++ b/src/beautifulsoup/builder/lxml_builder.py @@ -0,0 +1,35 @@ +from lxml import etree +from beautifulsoup.element import Comment +from beautifulsoup.builder import TreeBuilder + +class LXMLTreeBuilder(TreeBuilder): + + def __init__(self, parser_class=etree.HTMLParser, self_closing_tags=[]): + self.parser = parser_class(target=self) + self.self_closing_tags = self_closing_tags + self.soup = None + + def isSelfClosingTag(self, name): + return name in self.self_closing_tags + + 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 comment(self, content): + "Handle comments as Comment objects." + self.soup.endData() + self.soup.handle_data(content) + self.soup.endData(Comment) |