From 84860e56a587dd2fa518526de84a04b6de98053c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Olov=20Ylinenp=C3=A4=C3=A4?= <olov.ylinenpaa@kb.se>
Date: Thu, 19 Oct 2023 20:08:29 +0200
Subject: [PATCH 1/3] Add transcription of Bulgarian and Russian (KR76)

---
 .../main/groovy/whelk/util/Romanizer.groovy   |   2 +
 .../src/main/resources/romanizer/bg-kr76.txt  |  58 ++++++
 .../main/resources/romanizer/cyrl-kr76.txt    | 192 ++++++++++++++++++
 .../src/main/resources/romanizer/ru-kr76.txt  |  92 +++++++++
 .../groovy/whelk/util/RomanizerSpec.groovy    |  27 +++
 5 files changed, 371 insertions(+)
 create mode 100644 whelk-core/src/main/resources/romanizer/bg-kr76.txt
 create mode 100644 whelk-core/src/main/resources/romanizer/cyrl-kr76.txt
 create mode 100644 whelk-core/src/main/resources/romanizer/ru-kr76.txt

diff --git a/whelk-core/src/main/groovy/whelk/util/Romanizer.groovy b/whelk-core/src/main/groovy/whelk/util/Romanizer.groovy
index 60dcbb52c9..cd0e4f22c0 100644
--- a/whelk-core/src/main/groovy/whelk/util/Romanizer.groovy
+++ b/whelk-core/src/main/groovy/whelk/util/Romanizer.groovy
@@ -23,6 +23,7 @@ class Romanizer {
     private static final List<Transform> AUTO = [
             auto('be-Cyrl',  'be-Latn-t-be-Cyrl-m0-iso-1968', ['be-iso.txt', 'slavic-iso.txt']),
             auto('bg-Cyrl',  'bg-Latn-t-bg-Cyrl-m0-iso-1968', ['bg-iso.txt', 'slavic-iso.txt']),
+            auto('bg-Cyrl',  'bg-Latn-t-bg-Cyrl-x0-kr76', ['cyrl-kr76.txt', 'bg-kr76.txt']),
             auto('bs-Cyrl',  'bs-Latn-t-bs-Cyrl-m0-iso-1968', ['bs-sr-iso.txt', 'slavic-iso.txt']),
             auto('el'     ,  'el-Latn-t-el-Grek-x0-btj', ['el-btj.txt']),
             auto('grc'    ,  'grc-Latn-t-grc-Grek-x0-skr-1980', ['grc-skr.txt']),
@@ -32,6 +33,7 @@ class Romanizer {
             auto('mk-Cyrl',  'mk-Latn-t-mk-Cyrl-m0-iso-1968', ['mk-iso.txt', 'slavic-iso.txt']),
             auto('mn-Cyrl',  'mn-Latn-t-mn-Cyrl-x0-lessing', ['mn-lessing.txt']),
             auto('ru-Cyrl',  'ru-Latn-t-ru-Cyrl-m0-iso-1968', ['ru-iso.txt', 'slavic-iso.txt']),
+            auto('ru-Cyrl',  'ru-Latn-t-ru-Cyrl-x0-kr76', ['cyrl-kr76.txt', 'ru-kr76.txt']),
             auto('sr-Cyrl',  'sr-Latn-t-sr-Cyrl-m0-iso-1968', ['bs-sr-iso.txt', 'slavic-iso.txt']),
             auto('uk-Cyrl',  'uk-Latn-t-uk-Cyrl-m0-iso-1968', ['uk-iso.txt', 'slavic-iso.txt']),
 
diff --git a/whelk-core/src/main/resources/romanizer/bg-kr76.txt b/whelk-core/src/main/resources/romanizer/bg-kr76.txt
new file mode 100644
index 0000000000..1820ac5e76
--- /dev/null
+++ b/whelk-core/src/main/resources/romanizer/bg-kr76.txt
@@ -0,0 +1,58 @@
+# ICU transform rules
+
+# Bulgarian
+# See also: cyrl-kr76.txt
+
+# Е е e
+\u0415 > E ;
+\u0435 > e ;
+
+# NOTE not in table
+# Ё ё e
+\u0401 > E ;
+\u0451 > e ;
+
+# Ж ж zj
+\u0416 } [:LowercaseLetter:] > Zj ;
+\u0416 > ZJ ;
+\u0436 > zj ;
+
+# И и i
+\u0418 > I ;
+\u0438 > i ;
+
+# Х х ch
+\u0425 } [:LowercaseLetter:] <> Ch;
+\u0425 <> CH ;
+\u0445 <> ch ;
+
+# Ц ц ts
+\u0426 } [:LowercaseLetter:] <> Ts ;
+\u0426 <> TS ;
+\u0446 <> ts ;
+
+# Ч ч tj
+\u0427 } [:LowercaseLetter:] <> Tj ;
+\u0427 <> TJ ;
+\u0447 <> tj ;
+
+# Ш ш sj
+\u0428 } [:LowercaseLetter:] <> Sj ;
+\u0428 <> SJ ;
+\u0448 <> sj ;
+
+# Щ щ sjt
+\u0429 } [:LowercaseLetter:] <> Sjt ;
+\u0429 <> SJT ;
+\u0449 <> sjt ;
+
+# Ъ ъ ă
+# 5) återges ej då den enligt äldre ortografi står i ordslut
+\u042A } [:^Letter:] <> ;
+\u042A <> \u0102 ;
+\u044A } [:^Letter:] <> ;
+\u044A <> \u0103 ;
+
+# Ѣ ѣ e
+\u0462 > E ;
+\u0463 > e ;
\ No newline at end of file
diff --git a/whelk-core/src/main/resources/romanizer/cyrl-kr76.txt b/whelk-core/src/main/resources/romanizer/cyrl-kr76.txt
new file mode 100644
index 0000000000..291eaaeff4
--- /dev/null
+++ b/whelk-core/src/main/resources/romanizer/cyrl-kr76.txt
@@ -0,0 +1,192 @@
+# ICU transform rules
+
+# Common base for transcription of cyrillic script.
+# "KR76 (Kjellberg) - Transkribering av kyrilliska alfabetet (folkbibliotekens praxis)"
+# https://metadatabyran.kb.se/download/18.6945cdaa174b74a2c361604/1601918194487/Transkriberingsschema_Kyrilliska%20alfabet_2015.pdf
+# https://metadatabyran.kb.se/beskrivning/specialanvisningar/mangsprak/libris-romaniseringspraxis#h-Kyrilliskskrift
+
+$vowel = [а у о ы и э я ю ё е А У О Ы И Э Я Ю Ё Е] ;
+$s_t = [с т С Т] ;
+$soft_hard = [ь ъ Ь Ъ] ;
+
+# А а a
+\u0410 > A ;
+\u0430 > a ;
+
+# Б б b
+\u0411 > B ;
+\u0431 > b ;
+
+# В в v
+\u0412 > V ;
+\u0432 > v ;
+
+# Г г g
+\u0413 > G ;
+\u0433 > g ;
+
+# TODO not in table
+# Ѓ ѓ g
+\u0403 > G ;
+\u0453 > g ;
+
+# TODO not in table
+# Ґ ґ g
+\u0490 > G ;
+\u0491 > g ;
+
+# Д д d
+\u0414 > D ;
+\u0434 > d ;
+
+# TODO serbocroatian
+# Ђ ђ đ
+\u0402 > \u0110 ;
+\u0452 > \u0111 ;
+
+# З з z
+\u0417 > Z ;
+\u0437 > z ;
+
+# Й й j
+\u0419 > J ;
+\u0439 > j ;
+
+# І і і
+# 3) tillhör den äldre ortografin
+\u0406 > I ;
+\u0456 > i ;
+
+# TODO not in table
+# Ї ї ï
+\u0407 > \u00CF ;
+\u0457 > \u00EF ;
+
+# TODO no in russian table
+# J j j
+\u0408 > J ;
+\u0458 > j ;
+
+# К к k
+\u041A > K ;
+\u043A > k ;
+
+# TODO not in table
+# Ќ ќ ḱ
+\u040C > \u1E30 ;
+\u045C > \u1E31 ;
+
+# Л л l
+\u041B > L ;
+\u043B > l ;
+
+# TODO not in russian table
+# Љ љ lj
+\u0409 } [:LowercaseLetter:] > Lj;
+\u0409 > LJ ;
+\u0459 > lj ;
+
+# М м m
+\u041C > M ;
+\u043C > m ;
+
+# Н н n
+\u041D > N ;
+\u043D > n ;
+
+# TODO not in russian table
+# Њ њ nj
+\u040A } [:LowercaseLetter:] > Nj;
+\u040A > NJ ;
+\u045A > nj ;
+
+# О о o
+\u041E > O ;
+\u043E > o ;
+
+# П п p
+\u041F > P ;
+\u043F > p ;
+
+# Р р r
+\u0420 > R ;
+\u0440 > r ;
+
+# С с s
+\u0421 > S ;
+\u0441 > s ;
+
+# Т т t
+\u0422 > T ;
+\u0442 > t ;
+
+# TODO not in russian table
+# Ћ ћ ć
+\u040B <> \u0106 ;
+\u045B <> \u0107 ;
+
+# У у u
+\u0423 <> U ;
+\u0443 <> u ;
+
+# TODO not in table
+# Ў ў ŭ
+\u040E <> \u016C ;
+\u045E <> \u016D ;
+
+# Ф ф f
+\u0424 <> F ;
+\u0444 <> f ;
+
+# TODO not in russian table
+# Џ џ dž
+\u040F } [:LowercaseLetter:] <> D\u017E;
+\u040F <> D\u017D ;
+\u045F <> d\u017E ;
+
+# Ы ы y
+\u042B <> Y ;
+\u044B <> y ;
+
+# Ь ь
+# 4) återges ej; se dock not 1
+# 6) återges ej
+\u042C > ;
+\u044C > ;
+
+# Э э e
+\u042D > E ;
+\u044D > e ;
+
+# Ю ю ju
+# 8) iu efter с, т och з
+[с т з С Т З] { \u042E } [:LowercaseLetter:] <> Iu;
+[с т з С Т З] { \u042E <> IU ;
+[с т з С Т З] { \u044E <> iu ;
+\u042E } [:LowercaseLetter:] <> Ju;
+\u042E <> JU ;
+\u044E <> ju ;
+
+# Я я ja
+# 9) ia efter с, т och з
+[с т з С Т З] { \u042F } [:LowercaseLetter:] <> Ia;
+[с т з С Т З] { \u042F <> IA ;
+[с т з С Т З] { \u044F <> ia ;
+\u042F } [[:LowercaseLetter:][:^Letter:]] <> Ja;
+\u042F <> JA ;
+\u044F <> ja ;
+
+# Ө ө f
+# 3) tillhör den äldre ortografin
+\u04E8 > F ;
+\u04E9 > f ;
+
+# Ѵ ѵ i
+# 3) tillhör den äldre ortografin
+\u0474 > I ;
+\u0475 > i ;
+
+# Ѫ ѫ ă
+# 3) tillhör den äldre ortografin
+\u046A <> \u0102 ;
+\u046B <> \u0103 ;
\ No newline at end of file
diff --git a/whelk-core/src/main/resources/romanizer/ru-kr76.txt b/whelk-core/src/main/resources/romanizer/ru-kr76.txt
new file mode 100644
index 0000000000..435f540c47
--- /dev/null
+++ b/whelk-core/src/main/resources/romanizer/ru-kr76.txt
@@ -0,0 +1,92 @@
+# ICU transform rules
+
+# Russian
+# See also: cyrl-kr76.txt
+
+# Е е e
+# 1) efter vokal, ь, ъ och i början av ord je; dock ie om ь eller ъ föregås av с eller т
+$s_t $soft_hard { \u0415 } > IE;
+$s_t $soft_hard { \u0435 > ie;
+
+[$vowel $soft_hard] { \u0415 } [:LowercaseLetter:] > Je;
+[$vowel $soft_hard] { \u0415 > JE ;
+[$vowel $soft_hard] { \u0435 > je ;
+
+[:^Letter:] { \u0415 } [:LowercaseLetter:] > Je;
+[:^Letter:] { \u0415 } [:Letter:] > JE ;
+[:^Letter:] { \u0435 } [:Letter:] > je ;
+
+\u0415 > E ;
+\u0435 > e ;
+
+# Ё ё e
+# 1) efter vokal, ь, ъ och i början av ord je; dock ie om ь eller ъ föregås av с eller т
+$s_t $soft_hard { \u0401 > IE;
+$s_t $soft_hard { \u0451 > ie;
+
+[$vowel $soft_hard] { \u0401 } [:LowercaseLetter:] > Je;
+[$vowel $soft_hard] { \u0401 > JE ;
+[$vowel $soft_hard] { \u0451 > je ;
+
+[:^Letter:] { \u0401 } [:LowercaseLetter:] > Je;
+[:^Letter:] { \u0401 } [:Letter:] > JE ;
+[:^Letter:] { \u0451 } [:Letter:] > je ;
+
+\u0401 > E ;
+\u0451 > e ;
+
+# Ж ж zj
+\u0416 } [:LowercaseLetter:] > Zj ;
+\u0416 > ZJ ;
+\u0436 > zj ;
+
+# И и i
+# 2) efter ь ji
+[ь Ь] { \u0418 } [:LowercaseLetter:] > Ji ;
+[ь Ь] { \u0418 > JI ;
+[ь Ь] { \u0438 > ji ;
+\u0418 > I ;
+\u0438 > i ;
+
+# Х х ch
+\u0425 } [:LowercaseLetter:] <> Ch;
+\u0425 <> CH ;
+\u0445 <> ch ;
+
+# Ц ц ts
+\u0426 } [:LowercaseLetter:] <> Ts ;
+\u0426 <> TS ;
+\u0446 <> ts ;
+
+# Ч ч tj
+\u0427 } [:LowercaseLetter:] <> Tj ;
+\u0427 <> TJ ;
+\u0447 <> tj ;
+
+# Ш ш sj
+\u0428 } [:LowercaseLetter:] <> Sj ;
+\u0428 <> SJ ;
+\u0448 <> sj ;
+
+# Щ щ sjtj
+\u0429 } [:LowercaseLetter:] <> Sjtj ;
+\u0429 <> SJTJ ;
+\u0449 <> sjtj ;
+
+# Ъ ъ
+# 4) återges ej; se dock not 1
+\u042A > ;
+\u044A > ;
+
+# Ѣ ѣ e
+# 7) efter vokal och i början av ord je
+$vowel { \u0462 } [:LowercaseLetter:] > Je;
+$vowel [:^Letter:] { \u0462 > JE ;
+$vowel [:^Letter:] { \u0463 > je ;
+
+[:^Letter:] { \u0462 } [:LowercaseLetter:] > Je;
+[:^Letter:] { \u0462 } [:Letter:] > JE ;
+[:^Letter:] { \u0463 } [:Letter:]> je ;
+
+\u0462 > E ;
+\u0463 > e ;
\ No newline at end of file
diff --git a/whelk-core/src/test/groovy/whelk/util/RomanizerSpec.groovy b/whelk-core/src/test/groovy/whelk/util/RomanizerSpec.groovy
index 56063803e7..274117a540 100644
--- a/whelk-core/src/test/groovy/whelk/util/RomanizerSpec.groovy
+++ b/whelk-core/src/test/groovy/whelk/util/RomanizerSpec.groovy
@@ -116,6 +116,22 @@ class RomanizerSpec extends Specification {
         'Ё' || 'E'
     }
 
+    def "Russian with KR76"() {
+        expect:
+        new Romanizer().romanize(source, 'ru')['ru-Latn-t-ru-Cyrl-x0-kr76'] == target
+        where:
+        source || target
+        // https://libris.kb.se/dr3t6tk7btsm1xpm#it
+        'Маша и медведь'           || 'Masja i medved'
+        'Русские народные сказки' || 'Russkie narodnye skazki'
+        // Hard sign
+        'съесть' || 'sest'
+        'съ'     || 's'
+        // Yo
+        'сёрфингист' || 'serfingist'
+        'Ё' || 'E'
+    }
+
     def "Belarusian with ISO"() {
         expect:
         new Romanizer().romanize(source, 'be')['be-Latn-t-be-Cyrl-m0-iso-1968'] == target
@@ -154,6 +170,17 @@ class RomanizerSpec extends Specification {
         "Ще се удавят в сълзите на майките си" || "Šte se udavjat v sălzite na majkite si"
     }
 
+    def "Bulgarian with KR76"() {
+        expect:
+        new Romanizer().romanize(source, 'bg')['bg-Latn-t-bg-Cyrl-x0-kr76'] == target
+        where:
+        source || target
+        // https://libris.kb.se/fzr6pkkr2vnc152#it
+        'Баба праща поздрави и се извинява' || 'Baba prasjta pozdravi i se izvinjava'
+        // https://libris.kb.se/jxqb93w0gxhtbf7d#it
+        "Белия зъб" || "Belija zăb"
+    }
+
     def "Macedonian with ISO"() {
         expect:
         new Romanizer().romanize(source, 'mk')['mk-Latn-t-mk-Cyrl-m0-iso-1968'] == target

From 5f21be4df41d3b384e23f8918299faa7c1841df9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Olov=20Ylinenp=C3=A4=C3=A4?= <olov.ylinenpaa@kb.se>
Date: Thu, 19 Oct 2023 20:31:10 +0200
Subject: [PATCH 2/3] Add transcription of Bosnian and Serbian (KR76)

---
 .../main/groovy/whelk/util/Romanizer.groovy   |  2 +
 .../main/resources/romanizer/bs-sr-kr76.txt   | 46 +++++++++++++++++++
 .../main/resources/romanizer/cyrl-kr76.txt    |  9 ----
 .../groovy/whelk/util/RomanizerSpec.groovy    | 20 ++++++++
 4 files changed, 68 insertions(+), 9 deletions(-)
 create mode 100644 whelk-core/src/main/resources/romanizer/bs-sr-kr76.txt

diff --git a/whelk-core/src/main/groovy/whelk/util/Romanizer.groovy b/whelk-core/src/main/groovy/whelk/util/Romanizer.groovy
index cd0e4f22c0..b71223f99c 100644
--- a/whelk-core/src/main/groovy/whelk/util/Romanizer.groovy
+++ b/whelk-core/src/main/groovy/whelk/util/Romanizer.groovy
@@ -25,6 +25,7 @@ class Romanizer {
             auto('bg-Cyrl',  'bg-Latn-t-bg-Cyrl-m0-iso-1968', ['bg-iso.txt', 'slavic-iso.txt']),
             auto('bg-Cyrl',  'bg-Latn-t-bg-Cyrl-x0-kr76', ['cyrl-kr76.txt', 'bg-kr76.txt']),
             auto('bs-Cyrl',  'bs-Latn-t-bs-Cyrl-m0-iso-1968', ['bs-sr-iso.txt', 'slavic-iso.txt']),
+            auto('bs-Cyrl',  'bs-Latn-t-bs-Cyrl-x0-kr76', ['cyrl-kr76.txt', 'bs-sr-kr76.txt']),
             auto('el'     ,  'el-Latn-t-el-Grek-x0-btj', ['el-btj.txt']),
             auto('grc'    ,  'grc-Latn-t-grc-Grek-x0-skr-1980', ['grc-skr.txt']),
             auto('yi-Hebr',  'yi-Latn-t-yi-Hebr-x0-yivo', ['yi-yivo.txt']),
@@ -35,6 +36,7 @@ class Romanizer {
             auto('ru-Cyrl',  'ru-Latn-t-ru-Cyrl-m0-iso-1968', ['ru-iso.txt', 'slavic-iso.txt']),
             auto('ru-Cyrl',  'ru-Latn-t-ru-Cyrl-x0-kr76', ['cyrl-kr76.txt', 'ru-kr76.txt']),
             auto('sr-Cyrl',  'sr-Latn-t-sr-Cyrl-m0-iso-1968', ['bs-sr-iso.txt', 'slavic-iso.txt']),
+            auto('sr-Cyrl',  'sr-Latn-t-sr-Cyrl-x0-kr76', ['cyrl-kr76.txt', 'bs-sr-kr76.txt']),
             auto('uk-Cyrl',  'uk-Latn-t-uk-Cyrl-m0-iso-1968', ['uk-iso.txt', 'slavic-iso.txt']),
 
             // Converted from LOC mappings
diff --git a/whelk-core/src/main/resources/romanizer/bs-sr-kr76.txt b/whelk-core/src/main/resources/romanizer/bs-sr-kr76.txt
new file mode 100644
index 0000000000..da4963e55b
--- /dev/null
+++ b/whelk-core/src/main/resources/romanizer/bs-sr-kr76.txt
@@ -0,0 +1,46 @@
+# ICU transform rules
+
+# Bosnian & Serbian
+# See also: cyrl-kr76.txt
+
+# Е е e
+\u0415 <> E ;
+\u0435 <> e ;
+
+# NOTE not in table
+# Ё ё e
+\u0401 > E ;
+\u0451 > e ;
+
+# Ж ж ž
+\u0416 > \u017d ;
+\u0436 > \u017e ;
+
+# И и i
+\u0418 > I ;
+\u0438 > i ;
+
+# Х х h
+\u0425 <> H ;
+\u0445 <> h ;
+
+# Ц ц c
+\u0426 <> C ;
+\u0446 <> c ;
+
+# Ч ч č
+\u0427 <> \u010c ;
+\u0447 <> \u010d ;
+
+# Щ щ ž
+\u0429 <> \u0160\u010C ;
+\u0449 <> \u0161\u010D ;
+
+# Џ џ dž
+\u040F } [:LowercaseLetter:] <> D\u017E;
+\u040F <> D\u017D ;
+\u045F <> d\u017E ;
+
+# Ш ш š
+\u0428 <> \u0160 ;
+\u0448 <> \u0161 ;
\ No newline at end of file
diff --git a/whelk-core/src/main/resources/romanizer/cyrl-kr76.txt b/whelk-core/src/main/resources/romanizer/cyrl-kr76.txt
index 291eaaeff4..47526add0e 100644
--- a/whelk-core/src/main/resources/romanizer/cyrl-kr76.txt
+++ b/whelk-core/src/main/resources/romanizer/cyrl-kr76.txt
@@ -80,7 +80,6 @@ $soft_hard = [ь ъ Ь Ъ] ;
 \u041B > L ;
 \u043B > l ;
 
-# TODO not in russian table
 # Љ љ lj
 \u0409 } [:LowercaseLetter:] > Lj;
 \u0409 > LJ ;
@@ -94,7 +93,6 @@ $soft_hard = [ь ъ Ь Ъ] ;
 \u041D > N ;
 \u043D > n ;
 
-# TODO not in russian table
 # Њ њ nj
 \u040A } [:LowercaseLetter:] > Nj;
 \u040A > NJ ;
@@ -120,7 +118,6 @@ $soft_hard = [ь ъ Ь Ъ] ;
 \u0422 > T ;
 \u0442 > t ;
 
-# TODO not in russian table
 # Ћ ћ ć
 \u040B <> \u0106 ;
 \u045B <> \u0107 ;
@@ -138,12 +135,6 @@ $soft_hard = [ь ъ Ь Ъ] ;
 \u0424 <> F ;
 \u0444 <> f ;
 
-# TODO not in russian table
-# Џ џ dž
-\u040F } [:LowercaseLetter:] <> D\u017E;
-\u040F <> D\u017D ;
-\u045F <> d\u017E ;
-
 # Ы ы y
 \u042B <> Y ;
 \u044B <> y ;
diff --git a/whelk-core/src/test/groovy/whelk/util/RomanizerSpec.groovy b/whelk-core/src/test/groovy/whelk/util/RomanizerSpec.groovy
index 274117a540..4b514040d0 100644
--- a/whelk-core/src/test/groovy/whelk/util/RomanizerSpec.groovy
+++ b/whelk-core/src/test/groovy/whelk/util/RomanizerSpec.groovy
@@ -156,6 +156,15 @@ class RomanizerSpec extends Specification {
         // https://libris.kb.se/5m0smjhb3d66cj4c#it
         "Бајке за дјецу" || "Bajke za djecu" // "Bajke za djecu" in record
     }
+
+    def "Bosnian with KR76"() {
+        expect:
+        new Romanizer().romanize(source, 'bs')['bs-Latn-t-bs-Cyrl-x0-kr76'] == target
+        where:
+        source           || target
+        // https://libris.kb.se/5m0smjhb3d66cj4c#it
+        "Бајке за дјецу" || "Bajke za djecu" // "Bajke za djecu" in record
+    }
     
     def "Bulgarian with ISO"() {
         expect:
@@ -204,6 +213,17 @@ class RomanizerSpec extends Specification {
         'Тиги, хаjдемо у шетњу' || 'Tigi, hajdemo u šetnju'
     }
 
+    def "Serbian with KR76"() {
+        expect:
+        new Romanizer().romanize(source, 'sr')['sr-Latn-t-sr-Cyrl-x0-kr76'] == target
+        where:
+        source || target
+        // https://libris.kb.se/2dbbcc810dxjnk9l#it
+        "Узбуна у кући белих медведа" || "Uzbuna u kući belih medveda"
+        // https://libris.kb.se/dqnnpp1sbxm436wp#it
+        'Тиги, хаjдемо у шетњу' || 'Tigi, hajdemo u šetnju'
+    }
+
     def "Ukrainian with ISO"() {
         expect:
         new Romanizer().romanize(source, 'uk')['uk-Latn-t-uk-Cyrl-m0-iso-1968'] == target

From bae858b8c43c47f247b4242b96fafdef0620ae2c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Olov=20Ylinenp=C3=A4=C3=A4?= <olov.ylinenpaa@kb.se>
Date: Thu, 19 Oct 2023 20:49:23 +0200
Subject: [PATCH 3/3] Clean up

---
 whelk-core/src/main/resources/romanizer/bs-sr-kr76.txt | 4 ++++
 whelk-core/src/main/resources/romanizer/cyrl-kr76.txt  | 6 ------
 2 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/whelk-core/src/main/resources/romanizer/bs-sr-kr76.txt b/whelk-core/src/main/resources/romanizer/bs-sr-kr76.txt
index da4963e55b..0750746066 100644
--- a/whelk-core/src/main/resources/romanizer/bs-sr-kr76.txt
+++ b/whelk-core/src/main/resources/romanizer/bs-sr-kr76.txt
@@ -12,6 +12,10 @@
 \u0401 > E ;
 \u0451 > e ;
 
+# Ђ ђ đ
+\u0402 > \u0110 ;
+\u0452 > \u0111 ;
+
 # Ж ж ž
 \u0416 > \u017d ;
 \u0436 > \u017e ;
diff --git a/whelk-core/src/main/resources/romanizer/cyrl-kr76.txt b/whelk-core/src/main/resources/romanizer/cyrl-kr76.txt
index 47526add0e..dccc39dcd9 100644
--- a/whelk-core/src/main/resources/romanizer/cyrl-kr76.txt
+++ b/whelk-core/src/main/resources/romanizer/cyrl-kr76.txt
@@ -39,11 +39,6 @@ $soft_hard = [ь ъ Ь Ъ] ;
 \u0414 > D ;
 \u0434 > d ;
 
-# TODO serbocroatian
-# Ђ ђ đ
-\u0402 > \u0110 ;
-\u0452 > \u0111 ;
-
 # З з z
 \u0417 > Z ;
 \u0437 > z ;
@@ -62,7 +57,6 @@ $soft_hard = [ь ъ Ь Ъ] ;
 \u0407 > \u00CF ;
 \u0457 > \u00EF ;
 
-# TODO no in russian table
 # J j j
 \u0408 > J ;
 \u0458 > j ;