def test_streamsource():
# Stream objects are not reusable
expect = open("setup.py", "rb").read()
parsed = b"".join(
event[1] for event in parse.Stream(open("setup.py", "rb"), bufsize=32)
if event[0] == "bytes")
assert parsed == expect
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 xsc2txt(instream, outstream, title, width):
e = parse.tree(parse.Stream(instream), parse.SGMLOP(), parse.NS(doc),
parse.Node(pool=xsc.docpool()))
if title is None:
title = xsc.Null
else:
title = doc.title(title)
e = html.html(html.body(doc.section(title, e)))
e = e.conv()
outstream.write(html.astext(e, width=width))
def test_parsingmethods():
t = "abc\U00012345\u3042xyz"
s = f'<?xml version="1.0" encoding="utf-8"?><a title="{t}">{t}</a>'
b = s.encode("utf-8")
def check(*pipeline):
node = parse.tree(*pipeline, validate=True)
node = node.walknodes(a)[0]
assert str(node) == t
assert str(node["title"]) == t
prefixes = {None: a.xmlns}
pool = xsc.Pool(a)
check(b, parse.Expat(), parse.NS(a.xmlns), parse.Node(pool))
check(s, parse.Encoder(encoding="utf-8"), parse.Expat(), parse.NS(a.xmlns),
parse.Node(pool))
check(parse.Iter(b), parse.Expat(), parse.NS(a.xmlns),
parse.Node(pool)) # parse byte by byte
check(parse.Stream(io.BytesIO(b), bufsize=1), parse.Expat(),
parse.NS(a.xmlns), parse.Node(pool))
check(parse.ETree(cElementTree.fromstring(b), defaultxmlns=a.xmlns),
parse.Node(pool))