#7311: fix html.parser to accept non-ASCII attribute values.

This commit is contained in:
Ezio Melotti 2011-04-07 22:03:31 +03:00
parent 9b5ac3efa6
commit 2e3607c1e7
3 changed files with 20 additions and 1 deletions

View file

@ -217,6 +217,23 @@ DOCTYPE html [
("starttag", "a", [("href", "mailto:xyz@example.com")]),
])
def test_attr_nonascii(self):
# see issue 7311
self._run_check("<img src=/foo/bar.png alt=\u4e2d\u6587>", [
("starttag", "img", [("src", "/foo/bar.png"),
("alt", "\u4e2d\u6587")]),
])
self._run_check("<a title='\u30c6\u30b9\u30c8' "
"href='\u30c6\u30b9\u30c8.html'>", [
("starttag", "a", [("title", "\u30c6\u30b9\u30c8"),
("href", "\u30c6\u30b9\u30c8.html")]),
])
self._run_check('<a title="\u30c6\u30b9\u30c8" '
'href="\u30c6\u30b9\u30c8.html">', [
("starttag", "a", [("title", "\u30c6\u30b9\u30c8"),
("href", "\u30c6\u30b9\u30c8.html")]),
])
def test_attr_entity_replacement(self):
self._run_check("""<a b='&amp;&gt;&lt;&quot;&apos;'>""", [
("starttag", "a", [("b", "&><\"'")]),