Hi, I am new to Koha. I basically just set-up a fresh Koha 3.14 installation on Ubuntu 13.10 using the Debian packages from http://debian.koha-community.org/ We are a small library with mostly Ukrainian books. We want to catalogue the books in original language (Ukrainian) and therefore in Cyrillic script. After testing Koha after cataloging the first couple of books I experienced the following problem when searching in the Opac: Whenever I entered any arbitrary combination of Cyrillic characters (not even real words) in the search field Koha returned all records in the catalogue. Whether there was a match or not. If I did the same in Latin script the search only returned matching records only. Searching for authorities in the worked if I searched with the „is exactly“ option. If I had „starts with“ or „contains“ it worked only for search „in main entry“, but not for search „in keyword“ or „any heading“. Does anyone have an idea what could be wrong? Regards, Stephan
Same here*: when I search for a (key)word in Greek, using the "Search the catalog" prompt in the admin interface, I apparently get all the items with some info or another entered in Greek. Furthermore, if I search for a single Greek character, "ξ" for example, I get even more results: "N result(s) found for 'kw,wrdl: ξ'" some of them in Greek, some in Latin alphabets. I remembered reading quite some time ago that one had to explicitly set zebra searches to be utf-aware, and tried to find the relevant threads. Found the following: - Cyrillic Character Search w/ Zebra (from May 2010) - Can we use Unicode to enter records? (from January 2011) - Tune Koha to enable precise regional language searching (from October 2013) The odd thing is that although mention is made on the older threads, of a (git) patch to apply when the "out-of-the-box" Unicode functionality didn't work, the 2013 thread repeats the same instructions, as if the patch hadn't made it to the current release in the two years that ensued. If indeed some tuning is needed for koha/zebra to correctly update and query zebra indexes, I wonder if it would be best set-up in the administrative interface, in a page where the admin could indicate/check the languages used in his/her installation's bibliographical entries. Perhaps the administrative interface could even search through the biblio entries and propose the languages mentioned in them (in USMARC 041, 008/35-37 or even 008/15-17 for example). kind regards, Manos PETRIDIS * my installation: v3.14 on Debian, freshly installed from packages. USMARC QueryFuzzy and QueryWeightFields are enabled, however disabling them at search-time makes no difference in search results. -----Original Message----- From: koha-bounces@lists.katipo.co.nz [mailto:koha-bounces@lists.katipo.co.nz] On Behalf Of Witoszynskyj Stephan Sent: Monday, January 06, 2014 9:43 AM To: koha@lists.katipo.co.nz Subject: [Koha] Search problems Hi, I am new to Koha. I basically just set-up a fresh Koha 3.14 installation on Ubuntu 13.10 using the Debian packages from http://debian.koha-community.org/ We are a small library with mostly Ukrainian books. We want to catalogue the books in original language (Ukrainian) and therefore in Cyrillic script. After testing Koha after cataloging the first couple of books I experienced the following problem when searching in the Opac: Whenever I entered any arbitrary combination of Cyrillic characters (not even real words) in the search field Koha returned all records in the catalogue. Whether there was a match or not. If I did the same in Latin script the search only returned matching records only. Searching for authorities in the worked if I searched with the "is exactly" option. If I had "starts with" or "contains" it worked only for search "in main entry", but not for search "in keyword" or "any heading". Does anyone have an idea what could be wrong? Regards, Stephan _______________________________________________ Koha mailing list http://koha-community.org Koha@lists.katipo.co.nz http://lists.katipo.co.nz/mailman/listinfo/koha
Hi Stephan, in order to search non latin scripts with Koha correctly you need to make sure that you are using ICU with Zebra. Then the search should work nicely. Some notes on how to activate ICU indexing can be found on the wiki - it should also work for Ukrainian: http://wiki.koha-community.org/wiki/Correcting_Search_of_Arabic_records Hope this helps, Katrin -----Ursprüngliche Nachricht----- Von: koha-bounces@lists.katipo.co.nz im Auftrag von Witoszynskyj Stephan Gesendet: Mo 06.01.2014 08:42 An: koha@lists.katipo.co.nz Betreff: [Koha] Search problems Hi, I am new to Koha. I basically just set-up a fresh Koha 3.14 installation on Ubuntu 13.10 using the Debian packages from http://debian.koha-community.org/ We are a small library with mostly Ukrainian books. We want to catalogue the books in original language (Ukrainian) and therefore in Cyrillic script. After testing Koha after cataloging the first couple of books I experienced the following problem when searching in the Opac: Whenever I entered any arbitrary combination of Cyrillic characters (not even real words) in the search field Koha returned all records in the catalogue. Whether there was a match or not. If I did the same in Latin script the search only returned matching records only. Searching for authorities in the worked if I searched with the "is exactly" option. If I had "starts with" or "contains" it worked only for search "in main entry", but not for search "in keyword" or "any heading". Does anyone have an idea what could be wrong? Regards, Stephan _______________________________________________ Koha mailing list http://koha-community.org Koha@lists.katipo.co.nz http://lists.katipo.co.nz/mailman/listinfo/koha
Hi Katrin, thank’s a lot! This solved my problem. Apparently, the only thing that was necessary was to switch to yaz-icu and to modify zebradb/etc/default.idx accordingly. I did not create a transliteration table. Thank you very much again! Stephan Am 06.01.2014 um 11:10 schrieb Fischer, Katrin <Katrin.Fischer@bsz-bw.de>:
Hi Stephan,
in order to search non latin scripts with Koha correctly you need to make sure that you are using ICU with Zebra. Then the search should work nicely. Some notes on how to activate ICU indexing can be found on the wiki - it should also work for Ukrainian:
http://wiki.koha-community.org/wiki/Correcting_Search_of_Arabic_records
Hope this helps,
Katrin
-----Ursprüngliche Nachricht----- Von: koha-bounces@lists.katipo.co.nz im Auftrag von Witoszynskyj Stephan Gesendet: Mo 06.01.2014 08:42 An: koha@lists.katipo.co.nz Betreff: [Koha] Search problems
Hi,
I am new to Koha. I basically just set-up a fresh Koha 3.14 installation on Ubuntu 13.10 using the Debian packages from http://debian.koha-community.org/
We are a small library with mostly Ukrainian books. We want to catalogue the books in original language (Ukrainian) and therefore in Cyrillic script. After testing Koha after cataloging the first couple of books I experienced the following problem when searching in the Opac:
Whenever I entered any arbitrary combination of Cyrillic characters (not even real words) in the search field Koha returned all records in the catalogue. Whether there was a match or not.
If I did the same in Latin script the search only returned matching records only.
Searching for authorities in the worked if I searched with the "is exactly" option. If I had "starts with" or "contains" it worked only for search "in main entry", but not for search "in keyword" or "any heading".
Does anyone have an idea what could be wrong?
Regards, Stephan
_______________________________________________ Koha mailing list http://koha-community.org Koha@lists.katipo.co.nz http://lists.katipo.co.nz/mailman/listinfo/koha
Hello All, I tried to apply the below fix, but everytime I execute the zebra rebuild function, I get the following error: "11:14:23-13/02 zebraidx(2157) [warn] default.idx:13: Failed to load ICU chain words-icu.xml" Any suggestions of what might be wrong? Your help is very much appreciated. Regards. Layale Bassil Software Engineer – Business Analyst IT Solutions Delivery Department American University of Beirut Office of Information Technology P.O.Box 11-0236 Riad El-Solh, Beirut 1107 2020, Lebanon T: +961 (1) 350000 x 3649 | M: +961 (03) 929876 E: lb26@aub.edu.lb W . Fb . Fl . T . Y . L . IT Think Green... Keep it on the Screen… Save a Tree... -----Original Message----- From: koha-bounces@lists.katipo.co.nz [mailto:koha-bounces@lists.katipo.co.nz] On Behalf Of Witoszynskyj Stephan Sent: Monday, January 06, 2014 4:40 PM To: koha@lists.katipo.co.nz Subject: Re: [Koha] Search problems Hi Katrin, thank’s a lot! This solved my problem. Apparently, the only thing that was necessary was to switch to yaz-icu and to modify zebradb/etc/default.idx accordingly. I did not create a transliteration table. Thank you very much again! Stephan Am 06.01.2014 um 11:10 schrieb Fischer, Katrin <Katrin.Fischer@bsz-bw.de>:
Hi Stephan,
in order to search non latin scripts with Koha correctly you need to make sure that you are using ICU with Zebra. Then the search should work nicely. Some notes on how to activate ICU indexing can be found on the wiki - it should also work for Ukrainian:
http://wiki.koha-community.org/wiki/Correcting_Search_of_Arabic_record s
Hope this helps,
Katrin
-----Ursprüngliche Nachricht----- Von: koha-bounces@lists.katipo.co.nz im Auftrag von Witoszynskyj Stephan Gesendet: Mo 06.01.2014 08:42 An: koha@lists.katipo.co.nz Betreff: [Koha] Search problems
Hi,
I am new to Koha. I basically just set-up a fresh Koha 3.14 installation on Ubuntu 13.10 using the Debian packages from http://debian.koha-community.org/
We are a small library with mostly Ukrainian books. We want to catalogue the books in original language (Ukrainian) and therefore in Cyrillic script. After testing Koha after cataloging the first couple of books I experienced the following problem when searching in the Opac:
Whenever I entered any arbitrary combination of Cyrillic characters (not even real words) in the search field Koha returned all records in the catalogue. Whether there was a match or not.
If I did the same in Latin script the search only returned matching records only.
Searching for authorities in the worked if I searched with the "is exactly" option. If I had "starts with" or "contains" it worked only for search "in main entry", but not for search "in keyword" or "any heading".
Does anyone have an idea what could be wrong?
Regards, Stephan
_______________________________________________ Koha mailing list http://koha-community.org Koha@lists.katipo.co.nz http://lists.katipo.co.nz/mailman/listinfo/koha
_______________________________________________ Koha mailing list http://koha-community.org Koha@lists.katipo.co.nz http://lists.katipo.co.nz/mailman/listinfo/koha
Hello to all, Il 13/02/2014 15:18, Layale Bassil ha scritto:
I tried to apply the below fix, but everytime I execute the zebra rebuild function, I get the following error:
"11:14:23-13/02 zebraidx(2157) [warn] default.idx:13: Failed to load ICU chain words-icu.xml"
Any suggestions of what might be wrong? Your help is very much appreciated. Regards.
I think you want to use arabic alphabet with Koha. The help page is this: http://wiki.koha-community.org/wiki/Correcting_Search_of_Arabic_records Your problem is probabli connect with position of the file words-icu.xml. The file words-icu.xml must to be same folder of default.idx. Bye Zeno Tajoli -- Dr. Zeno Tajoli Dipartimento Gestione delle Informazioni e della Conoscenza z.tajoli@cineca.it fax +39 02 2135520 CINECA - Sede operativa di Segrate
Thank you for your reply. Yes I am trying to use Arabic alphabet with Koha. I tried the procedure in the link below but when I rebuild Zebra I get the error. The file words-icu.xml is in the same folder as default.idx. Any other suggestions? Thanks. Layale Bassil Software Engineer – Business Analyst IT Solutions Delivery Department American University of Beirut Office of Information Technology P.O.Box 11-0236 Riad El-Solh, Beirut 1107 2020, Lebanon T: +961 (1) 350000 x 3649 | M: +961 (03) 929876 E: lb26@aub.edu.lb W . Fb . Fl . T . Y . L . IT Think Green... Keep it on the Screen… Save a Tree... -----Original Message----- From: koha-bounces@lists.katipo.co.nz [mailto:koha-bounces@lists.katipo.co.nz] On Behalf Of Zeno Tajoli Sent: Thursday, February 13, 2014 4:27 PM To: koha@lists.katipo.co.nz Subject: Re: [Koha] Search problems Hello to all, Il 13/02/2014 15:18, Layale Bassil ha scritto:
I tried to apply the below fix, but everytime I execute the zebra rebuild function, I get the following error:
"11:14:23-13/02 zebraidx(2157) [warn] default.idx:13: Failed to load ICU chain words-icu.xml"
Any suggestions of what might be wrong? Your help is very much appreciated. Regards.
I think you want to use arabic alphabet with Koha. The help page is this: http://wiki.koha-community.org/wiki/Correcting_Search_of_Arabic_records Your problem is probabli connect with position of the file words-icu.xml. The file words-icu.xml must to be same folder of default.idx. Bye Zeno Tajoli -- Dr. Zeno Tajoli Dipartimento Gestione delle Informazioni e della Conoscenza z.tajoli@cineca.it fax +39 02 2135520 CINECA - Sede operativa di Segrate _______________________________________________ Koha mailing list http://koha-community.org Koha@lists.katipo.co.nz http://lists.katipo.co.nz/mailman/listinfo/koha
Hi, Il 13/02/2014 15:43, Layale Bassil ha scritto:
Thank you for your reply.
Yes I am trying to use Arabic alphabet with Koha. I tried the procedure in the link below but when I rebuild Zebra I get the error.
The file words-icu.xml is in the same folder as default.idx. Any other suggestions?
the only other suggestions that I can do are: -- check the owner of default.idx and words-icu.xml files. Is it the same ? -- check the permissions of default.idx and words-icu.xml files. -- check that words-icu.xml is a Unix file, without dos end of lines and end of file. I don't know more options. Bye Zeno Tajoli -- Dr. Zeno Tajoli Dipartimento Gestione delle Informazioni e della Conoscenza z.tajoli@cineca.it fax +39 02 2135520 CINECA - Sede operativa di Segrate
Thanks a lot again for your reply. I checked all the below and they were correct. I just noticed that the error is coming from the Arabic characters that define the rules inside the words-icu.xml file. For example: <transliterate rule="{ الا > ا "/> Is there anything related to installing the Arabic support on the Koha server? When I remove all the Arabic characters from the file, everything works fine. Thanks again, Regards. Layale Bassil Software Engineer – Business Analyst IT Solutions Delivery Department American University of Beirut Office of Information Technology P.O.Box 11-0236 Riad El-Solh, Beirut 1107 2020, Lebanon T: +961 (1) 350000 x 3649 | M: +961 (03) 929876 E: lb26@aub.edu.lb W . Fb . Fl . T . Y . L . IT Think Green... Keep it on the Screen… Save a Tree... -----Original Message----- From: koha-bounces@lists.katipo.co.nz [mailto:koha-bounces@lists.katipo.co.nz] On Behalf Of Zeno Tajoli Sent: Thursday, February 13, 2014 4:53 PM To: koha@lists.katipo.co.nz Subject: Re: [Koha] Search problems Hi, Il 13/02/2014 15:43, Layale Bassil ha scritto:
Thank you for your reply.
Yes I am trying to use Arabic alphabet with Koha. I tried the procedure in the link below but when I rebuild Zebra I get the error.
The file words-icu.xml is in the same folder as default.idx. Any other suggestions?
the only other suggestions that I can do are: -- check the owner of default.idx and words-icu.xml files. Is it the same ? -- check the permissions of default.idx and words-icu.xml files. -- check that words-icu.xml is a Unix file, without dos end of lines and end of file. I don't know more options. Bye Zeno Tajoli -- Dr. Zeno Tajoli Dipartimento Gestione delle Informazioni e della Conoscenza z.tajoli@cineca.it fax +39 02 2135520 CINECA - Sede operativa di Segrate _______________________________________________ Koha mailing list http://koha-community.org Koha@lists.katipo.co.nz http://lists.katipo.co.nz/mailman/listinfo/koha
Hi to all, Il 13/02/2014 16:14, Layale Bassil ha scritto:
Thanks a lot again for your reply. I checked all the below and they were correct.
I just noticed that the error is coming from the Arabic characters that define the rules inside the words-icu.xml file. For example: <transliterate rule="{ الا > ا "/>
Is there anything related to installing the Arabic support on the Koha server?
When I remove all the Arabic characters from the file, everything works fine.
I'm sorry but I' don't have experience with Arabic support in deep. Bye Zeno Tajoli -- Dr. Zeno Tajoli Dipartimento Gestione delle Informazioni e della Conoscenza z.tajoli@cineca.it fax +39 02 2135520 CINECA - Sede operativa di Segrate
participants (5)
-
Fischer, Katrin -
Layale Bassil -
Manos PETRIDIS -
Witoszynskyj Stephan -
Zeno Tajoli