def printone(u): source = parse.URL(u) if isinstance(u, url.URL) else parse.Stream(u) node = parse.tree(source, parse.Tidy(), parse.NS(html), parse.Node(base="", pool=xsc.Pool(html, xml))) if args.compact: node = node.normalized().compacted() node = node.pretty() print((node.string(encoding=sys.stdout.encoding)))
def test_htmlparse_base():
e = parse.tree(b"<a href='gurk.gif'/>",
parse.Tidy(),
parse.NS(html),
parse.Node(base="hurz/index.html"),
validate=True)
e = e.walknodes(html.a)[0]
assert str(e.attrs.href) == "hurz/gurk.gif"
def test_parse_tidy_attrs():
e = parse.tree(
b"<a xmlns:xl='http://www.w3.org/1999/xlink' xml:lang='de' xl:href='gurk.gif' href='gurk.gif'/>",
parse.Tidy(),
parse.NS(html),
parse.Node(pool=xsc.Pool(html, xml, xlink)),
validate=True)
a = e.walknodes(html.a)[0]
assert str(a.attrs["href"]) == "gurk.gif"
assert str(a.attrs[xml.Attrs.lang]) == "de"
assert str(a.attrs[xlink.Attrs.href]) == "gurk.gif"
def test_parse_tidy_empty():
e = parse.tree(b"", parse.Tidy(), parse.NS(), parse.Node(), validate=True)
assert not e