Bug 418670 - Search not respecting UTF-8 charset
Summary: Search not respecting UTF-8 charset
Status: RESOLVED FIXED
Alias: None
Product: digikam
Classification: Applications
Component: Searches-Engine (show other bugs)
Version: 7.0.0
Platform: Appimage Linux
: NOR minor
Target Milestone: ---
Assignee: Digikam Developers
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2020-03-09 13:23 UTC by sse450
Modified: 2022-02-05 17:39 UTC (History)
2 users (show)

See Also:
Latest Commit:
Version Fixed In: 7.6.0


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description sse450 2020-03-09 13:23:00 UTC
SUMMARY
Just an example, we have "ö" as a character which is different than "o". When searching "okay", pictures with "gökay" tag are also coming up.

STEPS TO REPRODUCE
1. Tag one picture with "okay". Another picture with "gökay".
2. On the right hand side, select Search tab.
3. Enter "okay" as search term.
4. You will see the picture with tag "gökay" in the result.

OBSERVED RESULT
search for "okay" brings "gökay" as well.

EXPECTED RESULT
"ö" and "o" are two different letters. "gökay" should not come up in the results.

SOFTWARE/OS VERSIONS
Linux: Ubuntu 19.10

ADDITIONAL INFORMATION
My digiKam version: digikam-7.0.0-beta3-20200309T071114-x86-64.appimage

Similarly, "ü" and "u" are also different.
Comment 1 Maik Qualmann 2020-03-09 13:27:34 UTC
The problem cannot be reproduced here. It is probably related to the current AppImage locale problem. There are currently many other problems with this.

Maik
Comment 2 sse450 2020-03-09 13:30:42 UTC
OK, Maik.

Just wanted to inform you. Not a deal breaker for me.

Thank you.
Comment 3 Maik Qualmann 2020-03-09 13:33:12 UTC
SQLite or MySQL database?

Maik
Comment 4 caulier.gilles 2020-03-09 13:38:15 UTC
Maik,

I confirm : AppImage still do not use ICU for locale management under Linux and it use Posix method instead, which is a basic fail-back on Qt.

Gilles
Comment 5 sse450 2020-03-09 13:40:05 UTC
MySQL on the server.
Comment 6 Maik Qualmann 2020-03-10 20:51:04 UTC
I can reproduce the problem with MySQL too. However, this is even a DIN standard and fully intended for the "LIKE" command. Here is the answer from a German MySQL forum. Switching to "latin1_german2_ci" is not an option for an international program like digiKam.

https://forums.mysql.com/read.php?70,164894,164969#msg-164969

Maik
Comment 7 sse450 2020-03-11 12:39:34 UTC
Thank you, Maik.

Surely, if it needs latin*, it is better to keep current config. I am very happy with digiKam as it is.
Comment 8 caulier.gilles 2022-01-19 19:08:51 UTC
Git commit 00e4d5da2948ccdc101cdeeb23f7b05405812436 by Gilles Caulier.
Committed on 19/01/2022 at 19:04.
Pushed by cgilles into branch 'master'.

Great news : AppImage with Qt 5.15.2 compiled under Mageia 7.1 now support ICU
This require to recompile whole AppImage build system on Continuous Deployement server.
This will be done in a few days. I only tested here on my laptop, and i can confirm that ICU work fine now.
Related: bug 406583, bug 425168, bug 410980, bug 407506, bug 413842

M  +2    -3    project/bundles/3rdparty/ext_qt/5.15/CMakeLists.txt

https://invent.kde.org/graphics/digikam/commit/00e4d5da2948ccdc101cdeeb23f7b05405812436
Comment 9 caulier.gilles 2022-02-05 17:39:43 UTC
Hi,

With the ICU support introduced in next 7.6.0 release, this problem is now fixed.

You can test using 7.6.0 pre-release AppImage bundle available here : https://files.kde.org/digikam/

Best Regards

Gilles Caulier