interscript-maps 2.1.0 → 2.3.1a

data/maps/un-tam-Taml-Latn-1972.imp

@@ -4,7 +4,7 @@ metadata {
   language: iso-639-2:tam
   source_script: Gujr
   destination_script: Taml
-  name: REPORT ON THE CURRENT STATUS OF UNITED NATIONS ROMANIZATION SYSTEMS FOR GEOGRAPHICAL NAMES --Tamil Romanization Version 4.0
+  name: Romanization of Tamil -- UNGEGN 4.0 (1972)
   url: https://www.eki.ee/wgrs/rom1_ta.htm
   creation_date: 1972
   confirmation_date: 2016

data/maps/un-tel-Telu-Latn-1972.imp

@@ -4,7 +4,7 @@ metadata {
   language: iso-639-2:tel
   source_script: Telu
   destination_script: Latn
-  name: REPORT ON THE CURRENT STATUS OF UNITED NATIONS ROMANIZATION SYSTEMS FOR GEOGRAPHICAL NAMES -- Telugu Romanization, 1972
+  name: Romanization of Telugu -- UNGEGN 4.0 (1972)
   url: https://www.eki.ee/wgrs/rom1_te.htm
   creation_date: 1972
   confirmation_date: 2016

data/maps/un-ukr-Cyrl-Latn-1998.imp

@@ -4,7 +4,7 @@ metadata {
   language: iso-639-2:ukr
   source_script: Cyrl
   destination_script: Latn
-  name: ROMANIZATION SYSTEM FOR BELARUSIAN, RUSSIAN AND UKRAINIAN CYRILLIC
+  name: Romanization of Belarusian, Russian And Ukrainian Cyrillic (1998)
   url: https://unstats.un.org/unsd/geoinfo/UNGEGN/docs/7th-uncsgn-docs/econf/7th_UNCSGN_econf.91_3_Add1.pdf
   # page 320
   creation_date: 1998

data/maps/un-ukr-Cyrl-Latn-2012.imp

@@ -4,7 +4,7 @@ metadata {
   language: iso-639-2:ukr
   source_script: Cyrl
   destination_script: Latn
-  name: ROMANIZATION SYSTEM FOR BELARUSIAN, RUSSIAN AND UKRAINIAN CYRILLIC
+  name: Romanization of Belarusian, Russian And Ukrainian Cyrillic (2012)
   url: http://www.eki.ee/wgrs/rom1_uk.htm
   creation_date: 2010
 }

data/maps/un-urd-Arab-Latn-1972.imp

@@ -1,10 +1,10 @@
 metadata {
-  authority_id: ungegn
+  authority_id: un
   id: 2017
   language: iso-639-2:ara
   source_script: Arab
   destination_script: Latn
-  name: ROMANIZATION OF ARABIC -- UNGEGN 2017 System
+  name: Romanization of Arabic (2017)
   url: https://www.eki.ee/wgrs/rom1_ur.htm
   creation_date: 2017
   confirmation_date: 2018-06
@@ -207,20 +207,20 @@ stage {
     # '\uFE8E' : ''  # ﺎ
 
     # Sun letters
-    sub boundary + "\u0627\u0644\u062a", "at t" # الت
-    sub boundary + "\u0627\u0644\u062b", "as s" # الث
-    sub boundary + "\u0627\u0644\u062f", "ad d" # الد
-    sub boundary + "\u0627\u0644\u0630", "az z" # الذ
-    sub boundary + "\u0627\u0644\u0631", "ar r" # الر
-    sub boundary + "\u0627\u0644\u0632", "az z" # الز
-    sub boundary + "\u0627\u0644\u0633", "as s" # الس
-    sub boundary + "\u0627\u0644\u0634", "ash sh" # الش
-    sub boundary + "\u0627\u0644\u0635", "as s" # الص
-    sub boundary + "\u0627\u0644\u0636", "az z" # الض
-    sub boundary + "\u0627\u0644\u0637", "at t" # الط
-    sub boundary + "\u0627\u0644\u0638", "az z" # الظ
-    sub boundary + "\u0627\u0644\u0644", "al l" # الل
-    sub boundary + "\u0627\u0644\u0646", "an n" # الن
+    sub boundary + "\u0627\u0644\u062a" + maybe("\u0651"), "at t" # الت
+    sub boundary + "\u0627\u0644\u062b" + maybe("\u0651"), "as s" # الث
+    sub boundary + "\u0627\u0644\u062f" + maybe("\u0651"), "ad d" # الد
+    sub boundary + "\u0627\u0644\u0630" + maybe("\u0651"), "az z" # الذ
+    sub boundary + "\u0627\u0644\u0631" + maybe("\u0651"), "ar r" # الر
+    sub boundary + "\u0627\u0644\u0632" + maybe("\u0651"), "az z" # الز
+    sub boundary + "\u0627\u0644\u0633" + maybe("\u0651"), "as s" # الس
+    sub boundary + "\u0627\u0644\u0634" + maybe("\u0651"), "ash sh" # الش
+    sub boundary + "\u0627\u0644\u0635" + maybe("\u0651"), "as s" # الص
+    sub boundary + "\u0627\u0644\u0636" + maybe("\u0651"), "az z" # الض
+    sub boundary + "\u0627\u0644\u0637" + maybe("\u0651"), "at t" # الط
+    sub boundary + "\u0627\u0644\u0638" + maybe("\u0651"), "az z" # الظ
+    sub boundary + "\u0627\u0644\u0644" + maybe("\u0651"), "al l" # الل
+    sub boundary + "\u0627\u0644\u0646" + maybe("\u0651"), "an n" # الن
 
 
     # consonant characters

data/maps/var-amh-Ethi-Latn-eae-2003.imp

@@ -8,7 +8,11 @@ metadata {
   url: https://brill.com/view/book/edcoll/9789004419582/front-11.xml?language=en
   creation_date: 2003
   description: |
-    The Encyclopaedia Aethiopica (EAe) is a basic encyclopedia for Ethiopian and Eritrean studies. It employs an in-house form of romanization of Geez, Amharic, and other languages, which varies greatly from standard formats, such as BGN/PCGN: the emperor Menelek II's name, for example, is written as "Mənilək II".
+    The Encyclopaedia Aethiopica (EAe) is a basic encyclopedia for Ethiopian and
+    Eritrean studies. It employs an in-house form of romanization of Geez,
+    Amharic, and other languages, which varies greatly from standard formats,
+    such as BGN/PCGN: the emperor Menelek II's name, for example, is written as
+    "Mənilək II".
 }
 
 tests {
@@ -25,7 +29,7 @@ tests {
   test "ቋንቋ የድምጽ፣ የምልክት ወይም የምስል ቅንብር ሆኖ", "qʷanəqʷa yädəməṣə፣ yämələkətə wäyəmə yäməsələ qənəbərə hono"
   test "ለማሰብ ወይም የታሰበን ሃሳብ ለሌላ ለማስተላለፍ የሚረዳ መሳሪያ ነው", "lämasäbə wäyəmə yätasäbänə hasabə lälela lämasətälaläfə yämiräda mäsariya näwə"
   test "በአጭሩ ቋንቋ የምልክቶች ስርዓትና እኒህን ምልክቶች ለማቀናበር", "bäʾač̣əru qʷanəqʷa yämələkətočə sərəʿatəna ʾənihənə mələkətočə lämaqänabärə"
-  test "የሚያስፈልጉ ህጎች ጥንቅር ነው። ቋንቋወችን ለመፈረጅ እንዲሁም", "yämiyasəfäləgu həጎčə ṭənəqərə näwə። qʷanəqʷawäčənə lämäfäräǧə ʾənədihumə"
+  test "የሚያስፈልጉ ህጎች ጥንቅር ነው። ቋንቋወችን ለመፈረጅ እንዲሁም", "yämiyasəfäləgu həgočə ṭənəqərə näwə። qʷanəqʷawäčənə lämäfäräǧə ʾənədihumə"
   test "ለምክፈል የሚያስችሉ መስፈርቶችን ለማስቀመጥ ባለው ችግር", "läməkəfälə yämiyasəčəlu mäsəfärətočənə lämasəqämäṭə baläwə čəgərə"
   test "ምክንያት በአሁኑ ሰዓት በርግጠኝነት ስንት ቋንቋ በዓለም ላይ", "məkənəyatə bäʾahunu säʿatə bärəgəṭäñənätə sənətə qʷanəqʷa bäʿalämə layə"
   test "እንዳለ ማወቅ አስቸጋሪ ነው", "ʾənədalä mawäqə ʾasəčägari näwə"
@@ -85,7 +89,7 @@ stage {
     sub "\u1204", "he" # ሄ
     sub "\u1205", "hə" # ህ
     sub "\u1206", "ho" # ሆ
-    
+
     sub "\u1208", "lä" # ለ
     sub "\u1209", "lu" # ሉ
     sub "\u120A", "li" # ሊ
@@ -93,7 +97,7 @@ stage {
     sub "\u120C", "le" # ሌ
     sub "\u120D", "lə" # ል
     sub "\u120E", "lo" # ሎ
-    
+
     sub "\u1210", "ḥa" # ሐ
     sub "\u1211", "ḥu" # ሑ
     sub "\u1212", "ḥi" # ሒ
@@ -101,7 +105,7 @@ stage {
     sub "\u1214", "ḥe" # ሔ
     sub "\u1215", "ḥə" # ሕ
     sub "\u1216", "ḥo" # ሖ
-    
+
     sub "\u1218", "mä" # መ
     sub "\u1219", "mu" # ሙ
     sub "\u121A", "mi" # ሚ
@@ -109,7 +113,7 @@ stage {
     sub "\u121C", "me" # ሜ
     sub "\u121D", "mə" # ም
     sub "\u121E", "mo" # ሞ
-    
+
     sub "\u1220", "śä" # ሠ
     sub "\u1221", "śu" # ሡ
     sub "\u1222", "śi" # ሢ
@@ -117,7 +121,7 @@ stage {
     sub "\u1224", "śe" # ሤ
     sub "\u1225", "śə" # ሥ
     sub "\u1226", "śo" # ሦ
-    
+
     sub "\u1228", "rä" # ረ
     sub "\u1229", "ru" # ሩ
     sub "\u122A", "ri" # ሪ
@@ -125,7 +129,7 @@ stage {
     sub "\u122C", "re" # ሬ
     sub "\u122D", "rə" # ር
     sub "\u122E", "ro" # ሮ
-    
+
     sub "\u1230", "sä" # ሰ
     sub "\u1231", "su" # ሱ
     sub "\u1232", "si" # ሲ
@@ -133,7 +137,7 @@ stage {
     sub "\u1234", "se" # ሴ
     sub "\u1235", "sə" # ስ
     sub "\u1236", "so" # ሶ
-    
+
     sub "\u1238", "šä" # ሸ
     sub "\u1239", "šu" # ሹ
     sub "\u123A", "ši" # ሺ
@@ -141,7 +145,7 @@ stage {
     sub "\u123C", "še" # ሼ
     sub "\u123D", "šə" # ሽ
     sub "\u123E", "šo" # ሾ
-    
+
     sub "\u1240", "qä" # ቀ
     sub "\u1241", "qu" # ቁ
     sub "\u1242", "qi" # ቂ
@@ -149,7 +153,7 @@ stage {
     sub "\u1244", "qe" # ቄ
     sub "\u1245", "qə" # ቅ
     sub "\u1246", "qo" # ቆ
-    
+
     sub "\u1260", "bä" # በ
     sub "\u1261", "bu" # ቡ
     sub "\u1262", "bi" # ቢ
@@ -157,7 +161,7 @@ stage {
     sub "\u1264", "be" # ቤ
     sub "\u1265", "bə" # ብ
     sub "\u1266", "bo" # ቦ
-    
+
     sub "\u1270", "tä" # ተ
     sub "\u1271", "tu" # ቱ
     sub "\u1272", "ti" # ቲ
@@ -165,7 +169,7 @@ stage {
     sub "\u1274", "te" # ቴ
     sub "\u1275", "tə" # ት
     sub "\u1276", "to" # ቶ
-    
+
     sub "\u1278", "čä" # ቸ
     sub "\u1279", "ču" # ቹ
     sub "\u127A", "či" # ቺ
@@ -173,7 +177,7 @@ stage {
     sub "\u127C", "če" # ቼ
     sub "\u127D", "čə" # ች
     sub "\u127E", "čo" # ቾ
-    
+
     sub "\u1280", "ḫa" # ኀ
     sub "\u1281", "ḫu" # ኁ
     sub "\u1282", "ḫi" # ኂ
@@ -181,7 +185,7 @@ stage {
     sub "\u1284", "ḫe" # ኄ
     sub "\u1285", "ḫə" # ኅ
     sub "\u1286", "ḫo" # ኆ
-    
+
     sub "\u1290", "nä" # ነ
     sub "\u1291", "nu" # ኑ
     sub "\u1292", "ni" # ኒ
@@ -189,7 +193,7 @@ stage {
     sub "\u1294", "ne" # ኔ
     sub "\u1295", "nə" # ን
     sub "\u1296", "no" # ኖ
-    
+
     sub "\u1298", "ñä" # ኘ
     sub "\u1299", "ñu" # ኙ
     sub "\u129A", "ñi" # ኚ
@@ -197,7 +201,7 @@ stage {
     sub "\u129C", "ñe" # ኜ
     sub "\u129D", "ñə" # ኝ
     sub "\u129E", "ño" # ኞ
-    
+
     sub "\u12A0", "ʾa" # አ
     sub "\u12A1", "ʾu" # ኡ
     sub "\u12A2", "ʾi" # ኢ
@@ -205,7 +209,7 @@ stage {
     sub "\u12A4", "ʾe" # ኤ
     sub "\u12A5", "ʾə" # እ
     sub "\u12A6", "ʾo" # ኦ
-    
+
     sub "\u12A8", "kä" # ከ
     sub "\u12A9", "ku" # ኩ
     sub "\u12AA", "ki" # ኪ
@@ -213,7 +217,7 @@ stage {
     sub "\u12AC", "ke" # ኬ
     sub "\u12AD", "kə" # ክ
     sub "\u12AE", "ko" # ኮ
-    
+
     sub "\u12B8", "ḵä" # ኸ
     sub "\u12B9", "ḵu" # ኹ
     sub "\u12BA", "ḵi" # ኺ
@@ -221,7 +225,7 @@ stage {
     sub "\u12BC", "ḵe" # ኼ
     sub "\u12BD", "ḵə" # ኽ
     sub "\u12BE", "ḵo" # ኾ
-    
+
     sub "\u12C8", "wä" # ወ
     sub "\u12C9", "wu" # ዉ
     sub "\u12CA", "wi" # ዊ
@@ -229,7 +233,7 @@ stage {
     sub "\u12CC", "we" # ዌ
     sub "\u12CD", "wə" # ው
     sub "\u12CE", "wo" # ዎ
-    
+
     sub "\u12D0", "ʿa" # ዐ
     sub "\u12D1", "ʿu" # ዑ
     sub "\u12D2", "ʿi" # ዒ
@@ -237,7 +241,7 @@ stage {
     sub "\u12D4", "ʿe" # ዔ
     sub "\u12D5", "ʿə" # ዕ
     sub "\u12D6", "ʿo" # ዖ
-    
+
     sub "\u12D8", "zä" # ዘ
     sub "\u12D9", "zu" # ዙ
     sub "\u12DA", "zi" # ዚ
@@ -245,7 +249,7 @@ stage {
     sub "\u12DC", "ze" # ዜ
     sub "\u12DD", "zə" # ዝ
     sub "\u12DE", "zo" # ዞ
-    
+
     sub "\u12E0", "žä" # ዠ
     sub "\u12E1", "žu" # ዡ
     sub "\u12E2", "ži" # ዢ
@@ -253,7 +257,7 @@ stage {
     sub "\u12E4", "že" # ዤ
     sub "\u12E5", "žə" # ዥ
     sub "\u12E6", "žo" # ዦ
-    
+
     sub "\u12E8", "yä" # የ
     sub "\u12E9", "yu" # ዩ
     sub "\u12EA", "yi" # ዪ
@@ -261,7 +265,7 @@ stage {
     sub "\u12EC", "ye" # ዬ
     sub "\u12ED", "yə" # ይ
     sub "\u12EE", "yo" # ዮ
-    
+
     sub "\u12F0", "dä" # ደ
     sub "\u12F1", "du" # ዱ
     sub "\u12F2", "di" # ዲ
@@ -269,7 +273,7 @@ stage {
     sub "\u12F4", "de" # ዴ
     sub "\u12F5", "də" # ድ
     sub "\u12F6", "do" # ዶ
-    
+
     sub "\u1300", "ǧä" # ጀ
     sub "\u1301", "ǧu" # ጁ
     sub "\u1302", "ǧi" # ጂ
@@ -277,14 +281,15 @@ stage {
     sub "\u1304", "ǧe" # ጄ
     sub "\u1305", "ǧə" # ጅ
     sub "\u1306", "ǧo" # ጆ
-    
+
     sub "\u1308", "gä" # ገ
     sub "\u1309", "gu" # ጉ
     sub "\u130A", "gi" # ጊ
     sub "\u130B", "ga" # ጋ
     sub "\u130C", "ge" # ጌ
     sub "\u130D", "gə" # ግ
-    
+    sub "\u130E", "go" # ጎ
+
     sub "\u1320", "ṭä" # ጠ
     sub "\u1321", "ṭu" # ጡ
     sub "\u1322", "ṭi" # ጢ
@@ -292,7 +297,7 @@ stage {
     sub "\u1324", "ṭe" # ጤ
     sub "\u1325", "ṭə" # ጥ
     sub "\u1326", "ṭo" # ጦ
-    
+
     sub "\u1328", "č̣ä" # ጨ
     sub "\u1329", "č̣u" # ጩ
     sub "\u132A", "č̣i" # ጪ
@@ -300,7 +305,7 @@ stage {
     sub "\u132C", "č̣e" # ጬ
     sub "\u132D", "č̣ə" # ጭ
     sub "\u132E", "č̣o" # ጮ
-    
+
     sub "\u1330", "ṗä" # ጰ
     sub "\u1331", "ṗu" # ጱ
     sub "\u1332", "ṗi" # ጲ
@@ -308,7 +313,7 @@ stage {
     sub "\u1334", "ṗe" # ጴ
     sub "\u1335", "ṗə" # ጵ
     sub "\u1336", "ṗo" # ጶ
-    
+
     sub "\u1338", "ṣä" # ጸ
     sub "\u1339", "ṣu" # ጹ
     sub "\u133A", "ṣi" # ጺ
@@ -316,7 +321,7 @@ stage {
     sub "\u133C", "ṣe" # ጼ
     sub "\u133D", "ṣə" # ጽ
     sub "\u133E", "ṣo" # ጾ
-    
+
     sub "\u1340", "ṣ̓ä" # ፀ
     sub "\u1341", "ṣ̓u" # ፁ
     sub "\u1342", "ṣ̓i" # ፂ
@@ -324,7 +329,7 @@ stage {
     sub "\u1344", "ṣ̓e" # ፄ
     sub "\u1345", "ṣ̓ə" # ፅ
     sub "\u1346", "ṣ̓o" # ፆ
-    
+
     sub "\u1348", "fä" # ፈ
     sub "\u1349", "fu" # ፉ
     sub "\u134A", "fi" # ፊ
@@ -332,7 +337,7 @@ stage {
     sub "\u134C", "fe" # ፌ
     sub "\u134D", "fə" # ፍ
     sub "\u134E", "fo" # ፎ
-    
+
     sub "\u1350", "pä" # ፐ
     sub "\u1351", "pu" # ፑ
     sub "\u1352", "pi" # ፒ
@@ -340,7 +345,7 @@ stage {
     sub "\u1354", "pe" # ፔ
     sub "\u1355", "pə" # ፕ
     sub "\u1356", "po" # ፖ
-    
+
     sub "\u1268", "vä" # ቨ
     sub "\u1269", "vu" # ቩ
     sub "\u126A", "vi" # ቪ
@@ -348,46 +353,46 @@ stage {
     sub "\u126C", "ve" # ቬ
     sub "\u126D", "və" # ቭ
     sub "\u126E", "vo" # ቮ
-    
+
     # Labiovelars
     sub "\u1248", "qʷä" # ቈ
     sub "\u124a", "qʷi" # ቊ
     sub "\u124b", "qʷa" # ቋ
     sub "\u124c", "qʷe" # ቌ
     sub "\u124d", "qʷə" # ቍ
-    
+
     sub "\u1288", "ḫʷä" # ኈ
     sub "\u128a", "ḫʷi" # ኊ
     sub "\u128b", "ḫʷa" # ኋ
     sub "\u128c", "ḫʷe" # ኌ
     sub "\u128d", "ḫʷə" # ኍ
-    
+
     sub "\u12b0", "kʷä" # ኰ
     sub "\u12b2", "kʷi" # ኲ
     sub "\u12b3", "kʷa" # ኳ
     sub "\u12b4", "kʷe" # ኴ
     sub "\u12b5", "kʷə" # ኵ
-    
+
     sub "\u1310", "gʷä" # ጐ
     sub "\u1312", "gʷi" # ጒ
     sub "\u1313", "gʷa" # ጓ
     sub "\u1314", "gʷe" # ጔ
     sub "\u1315", "gʷə" # ጕ
-    
+
     sub "\u12c0", "ḵʷä" # ዀ
     sub "\u12c2", "ḵʷi" # ዂ
     sub "\u12c3", "ḵʷa" # ዃ
     sub "\u12c4", "ḵʷe" # ዄ
     sub "\u12c5", "ḵʷə" # ዅ
-    
+
     sub "\u1258", "q̱ʷä" # ቘ
     sub "\u125a", "q̱ʷi" # ቚ
     sub "\u125b", "q̱ʷa" # ቛ
     sub "\u125c", "q̱ʷe" # ቜ
     sub "\u125d", "q̱ʷə" # ቝ
-    
+
     # Numbers
-    
+
     sub "\u1369", "1" # ፩
     sub "\u136A", "2" # ፪
     sub "\u136B", "3" # ፫

data/maps/var-ara-Arab-Arab-rababa.imp

@@ -0,0 +1,25 @@
+metadata {
+  authority_id: var
+  id: rababa
+  language: iso-639-2:ara
+  source_script: Arab
+  destination_script: Latn
+  name: RABABA the Arabic Diacritization Library
+  url: https://github.com/interscript/rababa/
+  description: |
+    Arabic diacritization is useful for several practical business cases
+    like text to speech or Romanization of Arabic texts or scripts.
+
+    This map in particular uses the RABABA library to automatically detect
+    and add diacritics using machine learning.
+}
+
+tests {
+  test 'قطر' ,'قِطْرَ'
+  test 'abc', 'abc'
+  test '‘Iz. Ibrāhīm as-Sa‘danī', '‘Iz. Ibrāhīm as-Sa‘danī'
+}
+
+stage {
+  rababa config: "200"
+}