def test_itersource():
expect = b"hinz & kunz"
source = parse.Iter([b"hinz", b" & ", b"kunz"])
for i in range(3):
parsed = b"".join(data for (evtype, data) in source
if evtype == "bytes")
assert parsed == expect
def test_sgmlop_no_multiple_text_events():
# Test that we don't get consecutive text events with sgmlop
i = parse.events(
parse.Iter(b"<a>gurk & hurz & hinz & kunz</a>"),
parse.SGMLOP())
assert next(i) == ("url", url.URL("ITER"))
assert next(i) == ("enterstarttag", "a")
assert next(i) == ("leavestarttag", "a")
assert next(i) == ("text", "gurk & hurz & hinz & kunz")
assert next(i) == ("endtag", "a")
with pytest.raises(StopIteration):
next(i)
def test_itertree_skip():
def xml():
yield f"<ul xmlns='{html.xmlns}'>".encode("utf-8")
for i in range(10):
yield f"<li>{i}</li>".encode("utf-8")
yield "</ul>".encode("utf-8")
for c in parse.itertree(parse.Iter(xml()),
parse.Expat(ns=True),
parse.Node(),
enterelementnode=True,
validate=True):
if isinstance(c.node, html.ul):
c.entercontent = False
assert not isinstance(c.node, html.li)
def test_itertree_large():
def xml():
yield f"<ul xmlns='{html.xmlns}'>".encode("utf-8")
for i in range(1000):
yield f"<li>{i}</li>".encode("utf-8")
yield "</ul>".encode("utf-8")
for (i, c) in enumerate(
parse.itertree(parse.Iter(xml()),
parse.Expat(ns=True),
parse.Node(),
selector=html.li,
validate=True)):
assert int(str(c.node)) == i
c.path[-2].content.clear()
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))
