diff options
| author | Leonard Richardson <leonard.richardson@canonical.com> | 2011-02-11 09:10:56 -0500 |
|---|---|---|
| committer | Leonard Richardson <leonard.richardson@canonical.com> | 2011-02-11 09:10:56 -0500 |
| commit | d0531c4204a67a4289025bf7108a922f680fa057 (patch) | |
| tree | cdad3f97812e658d84a611b6017b7198fd97d818 /beautifulsoup | |
| parent | 3366ad67dc2dfdd508267efc87dfc851b612fb0d (diff) | |
| parent | d89c8878ea86a2575c87e9fad8081cfcd81e0bcd (diff) | |
Ported some more tests, fixed an encoding problem, and added rudimentary doctype handling.
Diffstat (limited to 'beautifulsoup')
| -rw-r--r-- | beautifulsoup/builder/lxml_builder.py | 41 | ||||
| -rw-r--r-- | beautifulsoup/element.py | 8 |
2 files changed, 46 insertions, 3 deletions
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'<html><body>%s</body></html>' % fragment + diff --git a/beautifulsoup/element.py b/beautifulsoup/element.py index bd9bcbf..b2e0e12 100644 --- a/beautifulsoup/element.py +++ b/beautifulsoup/element.py @@ -346,9 +346,6 @@ class NavigableString(unicode, PageElement): else: raise AttributeError, "'%s' object has no attribute '%s'" % (self.__class__.__name__, attr) - def encode(self, encoding=DEFAULT_OUTPUT_ENCODING): - return self.decode().encode(encoding) - def decodeGivenEventualEncoding(self, eventualEncoding): return self @@ -373,6 +370,11 @@ class Declaration(NavigableString): def decodeGivenEventualEncoding(self, eventualEncoding): return u'<!' + self + u'>' +class Doctype(NavigableString): + + def decodeGivenEventualEncoding(self, eventualEncoding): + return u'<!DOCTYPE ' + self + u'>' + class Tag(PageElement, Entities): """Represents a found HTML tag with its attributes and contents.""" |