diff options
| author | Georg Brandl <georg@python.org> | 2009-03-31 22:11:53 (GMT) |
|---|---|---|
| committer | Georg Brandl <georg@python.org> | 2009-03-31 22:11:53 (GMT) |
| commit | 0c7b2c9c191a97b4aede17c15e2b5013e7edcfb5 (patch) | |
| tree | 1119329789eadc18b6f0bb2dec9cceb7fdedc5b3 | |
| parent | 95fafec73281b58ecea37b2a61a2a44d0ce98e3a (diff) | |
| download | cpython-0c7b2c9c191a97b4aede17c15e2b5013e7edcfb5.zip cpython-0c7b2c9c191a97b4aede17c15e2b5013e7edcfb5.tar.gz cpython-0c7b2c9c191a97b4aede17c15e2b5013e7edcfb5.tar.bz2 | |
#1651995: fix _convert_ref for non-ASCII characters.
| -rw-r--r-- | Lib/sgmllib.py | 2 | ||||
| -rw-r--r-- | Lib/test/test_sgmllib.py | 9 |
2 files changed, 10 insertions, 1 deletions
diff --git a/Lib/sgmllib.py b/Lib/sgmllib.py index e5fbd32..104b25f 100644 --- a/Lib/sgmllib.py +++ b/Lib/sgmllib.py @@ -396,7 +396,7 @@ class SGMLParser(markupbase.ParserBase): n = int(name) except ValueError: return - if not 0 <= n <= 255: + if not 0 <= n <= 127: return return self.convert_codepoint(n) diff --git a/Lib/test/test_sgmllib.py b/Lib/test/test_sgmllib.py index 34fd7f0..081e0e1 100644 --- a/Lib/test/test_sgmllib.py +++ b/Lib/test/test_sgmllib.py @@ -373,6 +373,15 @@ DOCTYPE html PUBLIC '-//W3C//DTD HTML 4.01//EN' if len(data) != CHUNK: break + def test_only_decode_ascii(self): + # SF bug #1651995, make sure non-ascii character references are not decoded + s = '<signs exclamation="!" copyright="©" quoteleft="‘">' + self.check_events(s, [ + ('starttag', 'signs', + [('exclamation', '!'), ('copyright', '©'), + ('quoteleft', '‘')]), + ]) + # XXX These tests have been disabled by prefixing their names with # an underscore. The first two exercise outstanding bugs in the # sgmllib module, and the third exhibits questionable behavior |