Fix (enable) transliteration of Kirghiz, Church Slavic and Tigrinya

They were not found since the transliterations were defined using the three-letter ISO 639-2 codes but the languages actually have two-letter ISO 639-1 codes.
libris · Oct 19, 2023 · 6504763 · 6504763
1 parent f6a271e
commit 6504763
Show file tree

Hide file tree

Showing 5 changed files with 8 additions and 8 deletions.
diff --git a/whelk-core/src/main/groovy/whelk/util/Romanizer.groovy b/whelk-core/src/main/groovy/whelk/util/Romanizer.groovy
@@ -39,13 +39,13 @@ class Romanizer {
             // TODO: investigate how well these handle case/capitalization
             auto('am-Ethi',  'am-Latn-t-am-Ethi-m0-alaloc', ['loc/am-Latn-t-am-Ethi-m0-alaloc.txt']),
             auto('az-Cyrl',  'az-Latn-t-az-Cyrl-m0-alaloc', ['loc/az-Latn-t-az-Cyrl-m0-alaloc.txt']),
-            auto('chu'    ,  'chu-Latn-t-chu-Cyrs-m0-alaloc', ['loc/chu-Latn-t-chu-Cyrs-m0-alaloc.txt']),
+            auto('cu'    ,   'cu-Latn-t-cu-Cyrs-m0-alaloc', ['loc/cu-Latn-t-cu-Cyrs-m0-alaloc.txt']),
             auto('ka'     ,  'ka-Latn-t-ka-m0-alaloc', ['loc/ka-Latn-t-ka-m0-alaloc.txt']),
             auto('hi-Deva',  'hi-Latn-t-hi-Deva-m0-alaloc', ['loc/hi-Latn-t-hi-Deva-m0-alaloc.txt']),
             auto('hy-Armn',  'hy-Latn-t-hy-Armn-m0-alaloc', ['loc/hy-Latn-t-hy-Armn-m0-alaloc.txt']),
-            auto('kir-Cyrl', 'kir-Latn-t-kir-Cyrl-m0-alaloc', ['loc/kir-Latn-t-kir-Cyrl-m0-alaloc.txt']),
+            auto('ky-Cyrl',  'ky-Latn-t-ky-Cyrl-m0-alaloc', ['loc/ky-Latn-t-ky-Cyrl-m0-alaloc.txt']),
             auto('mn-Mong',  'mn-Latn-t-mn-Mong-m0-alaloc', ['loc/mn-Latn-t-mn-Mong-m0-alaloc.txt']),
-            auto('tir-Ethi',  'tir-Latn-t-tir-Ethi-m0-alaloc', ['loc/am-Latn-t-am-Ethi-m0-alaloc.txt']),
+            auto('ti-Ethi',  'ti-Latn-t-ti-Ethi-m0-alaloc', ['loc/am-Latn-t-am-Ethi-m0-alaloc.txt']),
             auto('tt-Cyrl',  'tt-Latn-t-tt-Cyrl-m0-alaloc', ['loc/tt-Latn-t-tt-Cyrl-m0-alaloc.txt']),
             auto('tg-Cyrl',  'tg-Latn-t-tg-Cyrl-m0-alaloc', ['loc/tg-Latn-t-tg-Cyrl-m0-alaloc.txt']),
             auto('tk-Cyrl',  'tk-Latn-t-tk-Cyrl-m0-alaloc', ['loc/tk-Latn-t-tk-Cyrl-m0-alaloc.txt']),

diff --git a/whelk-core/src/main/groovy/whelk/util/ScriptShifterShifter.groovy b/whelk-core/src/main/groovy/whelk/util/ScriptShifterShifter.groovy
@@ -17,11 +17,11 @@ class ScriptShifterShifter {
             'armenian'               : 'hy-Latn-t-hy-Armn-m0-alaloc',
             'azerbaijani'            : 'az-Latn-t-az-Cyrl-m0-alaloc',
             'chinese'                : 'zh-Latn-t-zh-Hani-m0-alaloc',
-            'church_slavonic'        : 'chu-Latn-t-chu-Cyrs-m0-alaloc',
+            'church_slavonic'        : 'cu-Latn-t-cu-Cyrs-m0-alaloc',
             'ethiopic'               : 'am-Latn-t-am-Ethi-m0-alaloc',
             'georgian'               : 'ka-Latn-t-ka-m0-alaloc',
             'hindi'                  : 'hi-Latn-t-hi-Deva-m0-alaloc',
-            'kyrgyz'                 : 'kir-Latn-t-kir-Cyrl-m0-alaloc',
+            'kyrgyz'                 : 'ky-Latn-t-ky-Cyrl-m0-alaloc',
             'mongolian_mongol_bichig': 'mn-Latn-t-mn-Mong-m0-alaloc',
             'tajik'                  : 'tg-Latn-t-tg-Cyrl-m0-alaloc',
             'tatar'                  : 'tt-Latn-t-tt-Cyrl-m0-alaloc',

diff --git a/...zer/loc/chu-Latn-t-chu-Cyrs-m0-alaloc.txt → ...nizer/loc/cu-Latn-t-cu-Cyrs-m0-alaloc.txt b/...zer/loc/chu-Latn-t-chu-Cyrs-m0-alaloc.txt → ...nizer/loc/cu-Latn-t-cu-Cyrs-m0-alaloc.txt
@@ -2,7 +2,7 @@
 # Source: https://raw.githubusercontent.com/lcnetdev/transliterator/main/scriptshifter/tables/data/church_slavonic.yml
 # ICU transform rules
 
-# chu-Latn-t-chu-Cyrs-m0-alaloc
+# cu-Latn-t-cu-Cyrs-m0-alaloc
 
 # script_to_roman
 'Є' > 'Ē' ;

diff --git a/...zer/loc/kir-Latn-t-kir-Cyrl-m0-alaloc.txt → ...nizer/loc/ky-Latn-t-ky-Cyrl-m0-alaloc.txt b/...zer/loc/kir-Latn-t-kir-Cyrl-m0-alaloc.txt → ...nizer/loc/ky-Latn-t-ky-Cyrl-m0-alaloc.txt
@@ -2,7 +2,7 @@
 # Source: https://raw.githubusercontent.com/lcnetdev/transliterator/main/scriptshifter/tables/data/kyrgyz.yml
 # ICU transform rules
 
-# kir-Latn-t-kir-Cyrl-m0-alaloc
+# ky-Latn-t-ky-Cyrl-m0-alaloc
 
 # script_to_roman
 'Є' > 'I︠E︡' ;

diff --git a/whelk-core/src/test/groovy/whelk/util/RomanizerSpec.groovy b/whelk-core/src/test/groovy/whelk/util/RomanizerSpec.groovy
@@ -301,7 +301,7 @@ class RomanizerSpec extends Specification {
 
     def "Tigrinya with ALA-LOC"() {
         expect:
-        new Romanizer().romanize(source, 'tir')['tir-Latn-t-tir-Ethi-m0-alaloc'] == target
+        new Romanizer().romanize(source, 'ti')['ti-Latn-t-ti-Ethi-m0-alaloc'] == target
         where:
         source         || target
         // zh9lf7193q7n08w