Codec lookup failing under turkish locale #46138
Comments
|
When switching to a turkish locale, the codecs registry fails on a codec This happens when the codec name contains an uppercase 'I'. What Replacing the tolower() call with this made the lookup work: int my_tolower(char c)
{
if ('A' <= c && c <= 'Z')
c += 32;
return c;
}PS: If the turkish locale is not supported, this here will enable it to a) sudo cp /usr/share/i18n/SUPPORTED /var/lib/locales/supported.d/local |
arnimar
mannequin
added
interpreter-core
arnimar
mannequin
added
stdlib
|
I can confirm this on SVN trunk on a Mandriva system. |
|
There is more to this bug than appears. I'm guessing that the name See this example: #!/usr/bin/python2.5 import localeprint 'TR', locale.normalize('tr') print locale.setlocale(locale.LC_ALL, ('tr_TR', 'ISO8859-9')) # first issue, not quite the same coming out, as came in # and this fails First, the value returned from getlocale is ('tr_TR', 'so8859-9'), not |
|
The C library's tolower() and toupper() are used in a handful of source Modules/_sre.c: return ((ch) < 256 ? (unsigned int)tolower((ch)) : ch); |
|
Even if we don't fix all uses of (?to)(lower|upper) in the source tree, Perhaps also the str type should grow ascii_lower() and ascii_upper() |
|
I agree that it's a bit unfortunate that the 8-bit string APIs in Python .lower() and .upper() for 8-bit strings were always locale dependent and In Python 3k the problem will probably go away, since .lower() and Perhaps we should just convert a few of the cases you found to using |
|
Marc-Andre: How should we proceed with this bug? Discuss on python-dev |
|
Sean: I'd suggest to discuss this on python-dev. Note that even if we do use Unicode for the cases in question, the |
|
Does anyone know if this was discussed on python-dev? I've tried searching the archives and didn't find anything, but that's not to say it isn't there. |
|
There is also a locale normalization function in unicodeobject.c: normalize_encoding(). This function uses "if (ISUPPER(*e)) *l++ = TOLOWER(e++);" which uses the Python, *locale-independent, implementation of ctype. We should maybe use the ISUPPER / TOLOWER in codecs.c. Anyway, a function should be fixed, but I don't know which one :-) |
|
We've included this patch in Gentoo for about two years now. Can we get some discussion going on doing something like this? |
|
Looking at this again, I think we should change the codec registry C code to use Py_TOLOWER() and the encoding search function code to use the .translate() approach that Antoine suggested. |
|
The decimal module has been fixed in Python 2.7, 3.2 and 3.3 for Turkish local: issue bpo-11830. |
|
New changeset 92d02de91cc9 by Antoine Pitrou in branch '3.2': New changeset a77a4df54b95 by Antoine Pitrou in branch '3.2': New changeset fe0caf8c48d2 by Antoine Pitrou in branch 'default': |
|
New changeset 739958134fe5 by Antoine Pitrou in branch '2.7': |
|
Finally fixed in 2.7, 3.2, 3.3! |
|
The Fedora bot fails because here ... locale.setlocale(locale.LC_CTYPE, loc)
loc = ('tr_TR', 'ISO8859-9'), and apparently setlocale can only
handle "tr_TR", but not "tr_TR.ISO8859-9":144 if (locale) { |
|
Stefan Krah <report@bugs.python.org> wrote:
Perhaps this is a bug in Fedora's setlocale that can't handle the turkish 'I' |
Perhaps indeed. Maybe you should try to report it. |
|
Yes, it's a bug. This works: #include <stdio.h>
#include <locale.h>
int
main(void)
{
char *s;
printf("%s\n", setlocale(LC_CTYPE, "tr_TR.ISO8859-9"));
printf("%s\n", setlocale(LC_CTYPE, NULL));
s = setlocale(LC_CTYPE, "tr_TR.ISO8859-9");
printf("%s\n", s ? s : "null");
return 0;
}But when I change the first setlocale call to "tr_TR", the result of |
|
I'm seeing this test failure in Gentoo, as well. |
|
Fedora bug report: |
|
Unrelated to the Fedora issue: The test is currently skipped on the diff -r 0b52b6f1bfab Lib/test/test_locale.py
--- a/Lib/test/test_locale.py Tue Aug 02 10:16:45 2011 +0200
+++ b/Lib/test/test_locale.py Tue Aug 02 11:37:39 2011 +0200
@@ -399,7 +399,7 @@
oldlocale = locale.setlocale(locale.LC_CTYPE)
self.addCleanup(locale.setlocale, locale.LC_CTYPE, oldlocale)
try:
- locale.setlocale(locale.LC_CTYPE, 'tr_TR')
+ locale.setlocale(locale.LC_CTYPE, 'tr_TR.UTF-8')
except locale.Error:
# Unsupported locale on this system
self.skipTest('test needs Turkish locale') |
|
As I wrote on python-dev, this test also fails on Debian lenny, which has So, indeed the test should be skipped on a multitude of platforms. |
|
On Tue, 02 Aug 2011 12:12:37 +0200, Stefan Krah <stefan@bytereef.org> wrote:
This is true for my Gentoo buildbots. Once we've figured out the When I run the C test program I get null as the final output of that This is with glibc-2.13-r2 (the r2 is Gentoo's mod number). As someone pointed out on python-dev, if this isn't fixable then it should be an expected failure, not a skip. One question is, is there any platform on which the turkish locale is installed where this test actually works? |
|
[Re-opening to fix the skips] Yes, the test works on: Ubuntu Lucid (libc-2.11.1), OpenSUSE (libc-2.11.1), FreeBSD-8.2 Failure: Fedora 14 (libc-2.13), Debian lenny (libc-2.7), Gentoo (libc-2.13-r2) So perhaps this test should be marked as expected failure on Linux |
The Python bug is fixed, the problem is apparently some libcs have the
Well, it works here (Mageia). |
|
https://bugzilla.redhat.com/show_bug.cgi?id=726536 claims that the I suspect the only way of running the test case reliably is whitelisting |
|
New changeset a55ffb6c1993 by Stefan Krah in branch '3.2': New changeset 4244e4348362 by Stefan Krah in branch 'default': New changeset 0b8917fc6db5 by Stefan Krah in branch '2.7': |
|
I've upgraded the Fedora buildbot to Fedora-16. The specific glibc So the test will now fail again on all systems that a) have the bug |
Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.
Show more details
GitHub fields:
bugs.python.org fields:
The text was updated successfully, but these errors were encountered: