interscript-maps 2.0.5

data/maps/alalc-tel-Telu-Latn-2011.imp

@@ -0,0 +1,58 @@
+metadata {
+  authority_id: alalc
+  id: 2011
+  language: iso-639-2:tel
+  source_script: Telu
+  destination_script: Latn
+  name: Telugu Romanization, 2011
+  url: http://catdir.loc.gov/catdir/cpso/romanization/telugu.pdf
+  creation_date: 2011
+  description: |
+    ALA-LC Romanization table for Telugu
+
+  notes:
+
+    - |
+      Only the vowel forms that appear at the beginning of a syllable are listed; the forms used for
+      vowels following a consonant can be found in grammars; no distinction between the two is
+      made in transliteration.
+    - |
+      The vowel a is implicit after all consonants and consonant clusters and is supplied in
+      transliteration, with the following exceptions:
+      a) when another vowel is indicated by its appropriate sign; and
+      b) when the absence of any vowel is indicated by the superscript sign ( ౬ ) called valapalagilaka.
+    - |
+      Exception: Sunna is transliterated by:
+      a) ṅ before gutturals,
+      b) ñ before palatals,
+      c) ṇ before cerebrals,
+      d) n before dentals, and
+      e) m before labials.
+    - |
+      Ardhasunna before gutturals and palatal, cerebral, and dental occlusives is transliterated n̐.
+      Before labials, sibilants, semivowels, the aspirate, vowels, and in final position it is
+      transliterated m
+}
+
+tests {
+  test "తమిళనాడు", "tamiḷanāḍu"
+  test "దేవాస్‌కు ౮౯౩౯ కోట్లివ్వండి", "dēvāsaku 8939 kŏṭalivavaṃḍi"
+  test "యూరప్, అమెరికాకు కోవిడ్‌ దడ", "yūrapa, amerikāku kŏviḍa daḍa"
+  test "జనవరి నాటికి అమెరికాలో టీకా", "janavari nāṭiki amerikālŏ ṭīkā"
+  test "ఫ్రాన్స్‌ను ముస్లింలు శిక్షించవచ్చు", "pharānasanu musaliṃlu śikaṣiñcavacacu"
+  test "క్లాస్‌ రూంలో ఉపాధ్యాయుడి వికృత చేష్టలు", "kalāsa rūṃlŏ upādhayāyuḍi vikṛta cēṣaṭalu"
+  test "భారీ భూకంపం; భయంకరమైన అనుభవాలు", "bhārī bhūkampaṃ; bhayaṅkaramaina anubhavālu"
+  test "నిట్ట నిలువునా కూలిన అపార్ట్‌మెంట్‌", "niṭaṭa niluvunā kūlina apāraṭameṇṭa"
+  test "పిచ్చి ప్రయోగాలకు పోతే జరిగేది ఇదే", "picaci parayŏgālaku pŏtē jarigēdi idē"
+  test "కరోనాపై సీడీసీ వైఫల్యం ఎందుకు?", "karŏnāpai sīḍīsī vaiphalayaṃ eṃduku?"
+  test "అత్యంత అరుదైన పులి పిల్లలు ఇవే!", "atayanta arudaina puli pilalalu ivē!"
+}
+
+dependency "alalc-tel-Telu-Latn-1997", as: telulatn
+
+stage {
+
+  run map.telulatn.stage.main
+
+}
+

data/maps/alalc-ukr-Cyrl-Latn-1997.imp

@@ -0,0 +1,123 @@
+metadata {
+  authority_id: alalc
+  id: 1997
+  language: iso-639-2:ukr
+  source_script: Cyrl
+  destination_script: Latn
+  name: ALA-LC Romanization System 1997
+  url: http://catdir.loc.gov/catdir/cpso/romanization/ukrainia.pdf
+  creation_date: 1997
+  description: |
+    The ALA-LC Romanization Tables, published by the American Library Association (1885)
+    and Library of Congress (1905). Used to represent bibliographic information by US and
+    Canadian libraries, by the British Library since 1975,[3] and in North American publications.
+    The latest 1997 revision is very similar to the 1905 version.
+
+  notes:
+    - The z͡h ligature is necessary to distinguish ж from the combination зг.
+    - The t͡s ligature is necessary to distinguish ц from the combination тс.
+}
+
+tests {
+  test "Автономна Республіка Крим", "Avtonomna Respublika Krym"
+  test "Висунь", "Vysunʹ"
+  test "Долинське", "Dolynsʹke"
+  test "Дубище", "Dubyshche"
+  test "Єнакієве", "I͡enakii͡eve"
+  test "Зупиночний Пункт Мокіївці", "Zupynochnyĭ Punkt Mokiïvt͡si"
+  test "Іванівщина", "Ivanivshchyna"
+  test "Киликиїв", "Kylykyïv"
+  test "Кожанка", "Koz͡hanka"
+  test "Краснянка", "Krasni͡anka"
+  test "Краснівка", "Krasnivka"
+  test "Мале Микільське", "Male Mykilʹsʹke"
+  test "Маломиколаївка", "Malomykolaïvka"
+  test "Нове Село", "Nove Selo"
+  test "Новопавлівка", "Novopavlivka"
+  test "Новошичі", "Novoshychi"
+  test "Новоєфремівка", "Novoi͡efremivka"
+  test "Одеська Область", "Odesʹka Oblastʹ"
+  test "Орлівське", "Orlivsʹke"
+  test "Раневичі", "Ranevychi"
+  test "Рокувата", "Rokuvata"
+  test "Рудаєве", "Rudai͡eve"
+  test "Сахнівці", "Sakhnivt͡si"
+  test "Тернівка", "Ternivka"
+  test "Турбівка", "Turbivka"
+  test "Херсонська Область", "Khersonsʹka Oblastʹ"
+}
+
+stage {
+
+  # CHARACTERS
+  parallel {
+    sub "А", "A" # А
+    sub "Б", "B" # Б
+    sub "В", "V" # В
+    sub "Г", "H" # Г
+    sub "Ґ", "G" # Ґ
+    sub "Д", "D" # Д
+    sub "Е", "E" # Е
+    sub "Є", "I͡e" # Є -> I͡e
+    sub "Ж", "Z͡h" # Ж -> Z͡h
+    sub "З", "Z" # З
+    sub "И", "Y" # И
+    sub "І", "I" # І
+    sub "Ї", "Ï" # Ї -> Ï
+    sub "Й", "Ĭ" # Й -> Ĭ
+    sub "К", "K" # К
+    sub "Л", "L" # Л
+    sub "М", "M" # М
+    sub "Н", "N" # Н
+    sub "О", "O" # О
+    sub "П", "P" # П
+    sub "Р", "R" # Р
+    sub "С", "S" # С
+    sub "Т", "T" # Т
+    sub "У", "U" # У
+    sub "Ф", "F" # Ф
+    sub "Х", "Kh" # Х
+    sub "Ц", "T͡s" # Ц -> T͡s
+    sub "Ч", "Ch" # Ч
+    sub "Ш", "Sh" # Ш
+    sub "Щ", "Shch" # Щ
+    sub "Ю", "I͡u" # Ю -> I͡u
+    sub "Я", "I͡a" # Я -> I͡a
+    sub "Ь", "ʹ" # Ь -> ʹ
+    sub "а", "a" # а
+    sub "б", "b" # б
+    sub "в", "v" # в
+    sub "г", "h" # г
+    sub "ґ", "g" # ґ
+    sub "д", "d" # д
+    sub "е", "e" # е
+    sub "є", "i͡e" # є -> i͡e
+    sub "ж", "z͡h" # ж -> z͡h
+    sub "з", "z" # з
+    sub "и", "y" # и
+    sub "і", "i" # і
+    sub "ї", "ï" # ї -> ï
+    sub "й", "ĭ" # й -> ĭ
+    sub "к", "k" # к
+    sub "л", "l" # л
+    sub "м", "m" # м
+    sub "н", "n" # н
+    sub "о", "o" # о
+    sub "п", "p" # п
+    sub "р", "r" # р
+    sub "с", "s" # с
+    sub "т", "t" # т
+    sub "у", "u" # у
+    sub "ф", "f" # ф
+    sub "х", "kh" # х
+    sub "ц", "t͡s" # ц -> t͡s
+    sub "ч", "ch" # ч
+    sub "ш", "sh" # ш
+    sub "щ", "shch" # щ
+    sub "ю", "i͡u" # ю -> i͡u
+    sub "я", "i͡a" # я -> i͡a
+    sub "ь", "ʹ" # Ь -> ʹ
+  }
+
+  compose
+}

data/maps/alalc-ukr-Cyrl-Latn-2011.imp

@@ -0,0 +1,32 @@
+metadata {
+  authority_id: alalc
+  id: 2011
+  language: iso-639-2:ukr
+  source_script: Cyrl
+  destination_script: Latn
+  name: ALA-LC Romanization System 1997
+  url: https://www.loc.gov/catdir/cpso/romanization/ukrainia.pdf
+  creation_date: 2011
+
+  notes:
+    - The z͡h ligature is necessary to distinguish ж from the combination зг.
+    - The t͡s ligature is necessary to distinguish ц from the combination тс.
+}
+
+# This map has been partially converted by the bin/maps_v1_to_v2 script
+# The section below requires human attention. Remember to remove this
+# comment and move the converted map to 'maps/' directory. Please also
+# take note that the maps-staging directory will be cleaned up whenever
+# you run the bin/maps_v1_to_v2 script. You should particularly be
+# concerned about any regular expressions found in this file and about
+# advanced expressions in parallel {} parts, and also about the order
+# of particular parts of the stage.
+
+dependency "alalc-ukr-Cyrl-Latn-1997", as: cyrllatn
+
+stage {
+
+  run map.cyrllatn.stage.main
+
+}
+

data/maps/apcbg-bul-Cyrl-Latn-1995.imp

@@ -0,0 +1,194 @@
+metadata {
+  authority_id: apcbg
+  id: 1995
+  language: iso-639-2:bul
+  source_script: Cyrl
+  destination_script: Latn
+  name: Toponymic Guidelines for Antarctica, Lyubomir Ivanov. Antarctic Place-names Commission of Bulgaria, Sofia, 1995.
+  url: http://www.apcbg.org/Toponymic-Guidelines.htm
+  creation_date: 1995
+  description: |
+    In Antarctica geographical names are important elements of identification, orientation, localization and navigation,
+    providing an essential reference system for logistic operations, including search and rescue measures, and for
+    international scientific research.  They facilitate information exchange in the field, in scientific publications and in
+    administrative measures under the Antarctic Treaty System. Geographical names also reflect the history of Antarctic
+    exploration.
+}
+
+tests {
+  test "нунатак Абрит", "nunatak Abrit"
+  test "връх Академия", "vrah Akademiya"
+  test "връх Ами Буе", "vrah Ami Bue"
+  test "нос Айтос", "nos Aytos"
+  test "залив Баба Тонка", "zaliv Baba Tonka"
+  test "Балабански камък", "Balabanski kamak"
+  test "Бедечки поток", "Bedechki potok"
+  test "нос Бяга", "nos Byaga"
+  test "остров Качо", "ostrov Kacho" # wiki ostrov Cacho
+  test "Чакъров остров", "Chakarov ostrov"
+  test "връх Дъбник", "vrah Dabnik"
+  test "залив Десислава", "zaliv Desislava"
+  test "ледник Джераси", "lednik Dzherasi"
+  test "Джегова скала", "Dzhegova skala"
+  test "Нунатак Едуард", "Nunatak Eduard"
+  test "Елховска седловина", "Elhovska sedlovina"
+  test "ледник Етър", "lednik Etar"
+  test "нунатак Филип Тотю", "nunatak Filip Totyu"
+  test "ледник Габаре", "lednik Gabare"
+  test "риф Гергини", "rif Gergini"
+  test "Гяуров връх", "Gyaurov vrah"
+  test "Гуцалски рид", "Gutsalski rid" # not found
+  test "Хараламбиев остров", "Haralambiev ostrov"
+  test "връх Ичера", "vrah Ichera"
+  test "полуостров Йоан Павел II", "poluostrov Yoan Pavel II"
+  test "нос Иван Александър", "nos Ivan Aleksandar"
+  test "нос Иречек", "nos Irechek"
+  test "нос Кърджали", "nos Kardzhali"
+  test "седловина Кърнаре", "sedlovina Karnare"
+  test "нунатак Керсеблепт", "nunatak Kerseblept"
+  test "Кондофрейски възвишения", "Kondofreyski vazvisheniya"
+  test "Костинбродски проход", "Kostinbrodski prohod"
+  test "връх Кожух", "vrah Kozhuh"
+  test "Кукерски нунатаци", "Kukerski nunatatsi"
+  test "залив Лазурен бряг", "zaliv Lazuren bryag"
+  test "връх Лудогорие", "vrah Ludogorie"
+  test "Лютибродски скали", "Lyutibrodski skali"
+  test "Масларов нунатак", "Maslarov nunatak"
+  test "Михневски връх", "Mihnevski vrah"
+  test "рид Митино", "rid Mitino"
+  test "езеро Наяда", "ezero Nayada"
+  test "нос Никюп", "nos Nikyup"
+  test "рид Оборище", "rid Oborishte"
+  test "залив Олуша", "zaliv Olusha"
+  test "Оряховски възвишения", "Oryahovski vazvisheniya"
+  test "нунатак Памидово", "nunatak Pamidovo"
+  test "връх Парангалица", "vrah Parangalitsa"
+  test "Първомайски провлак", "Parvomayski provlak"
+  test "ледник Патлейна", "lednik Patleyna"
+  test "полуостров Перник", "poluostrov Pernik"
+  test "връх Петко Войвода", "vrah Petko Voyvoda"
+  test "остров Фанагория", "ostrov Fanagoriya"
+  test "нос Плас", "nos Plas"
+  test "Пресиянов рид", "Presiyanov rid"
+  test "връх Принсипе де Астуриас", "vrah Prinsipe de Asturias" # wiki: vrah Príncipe de Asturias
+  test "нунатак Ръченица", "nunatak Rachenitsa"
+  test "връх Райна Княгиня", "vrah Rayna Knyaginya"
+  test "Рид Ръжана", "Rid Razhana"
+  test "връх Ригс", "vrah Rigs"
+  test "остров Рогулят", "ostrov Rogulyat"
+  test "ледник Сабазий", "lednik Sabaziy"
+  test "ледник Съединение", "lednik Saedinenie"
+  test "нунатак Сенокос", "nunatak Senokos"
+  test "Сейдолски камък", "Seydolski kamak"
+  test "ледник Щерна", "lednik Shterna"
+  test "връх Шишман", "vrah Shishman"
+  test "ледник Сигмен", "lednik Sigmen"
+  test "Седловина Синитово", "Sedlovina Sinitovo"
+  test "Ледник Скаплизо", "Lednik Skaplizo"
+  test "залив Слънчев бряг", "zaliv Slanchev bryag"
+  test "остров Соатрис", "ostrov Soatris"
+  test "планина Софийски Университет", "planina Sofiyski Universitet"
+  test "ледник Сребърна", "lednik Srebarna"
+  test "Средногорски възвишения", "Srednogorski vazvisheniya"
+  test "Св. Евтимиев камък", "Sv. Evtimiev kamak"
+  test "база Св. Климент Охридски", "baza Sv. Kliment Ohridski"
+  test "връх Стъргел", "vrah Stargel"
+  test "нунатак Сторгозия", "nunatak Storgoziya" # nunatak Storgozia according to wiki
+  test "нунатак Сурвакари", "nunatak Survakari"
+  test "ледник Световрачене", "lednik Svetovrachene"
+  test "остров Теменуга", "ostrov Temenuga"
+  test "Тракийски възвишения", "Trakiyski vazvisheniya"
+  test "хълм Цамблак", "halm Tsamblak"
+  test "ледник Урдовиза", "lednik Urdoviza"
+  test "остров Вълчедръм", "ostrov Valchedram"
+  test "острови Вардим", "ostrovi Vardim"
+  test "Владигеров проток", "Vladigerov protok"
+  test "ледник Ябланица", "lednik Yablanitsa"
+  test "залив Ямфорина", "zaliv Yamforina"
+  test "Йовков нос", "Yovkov nos"
+  test "рид Заберново", "rid Zabernovo"
+  test "ледник Збелсурд", "lednik Zbelsurd"
+  test "Жефарович камък", "Zhefarovich kamak"
+  test "връх Зиези", "vrah Ziezi"
+  test "залив Златни пясъци", "zaliv Zlatni pyasatsi"
+  test "ледник Злокучене", "lednik Zlokuchene"
+  test "проток Злогош", "protok Zlogosh"
+}
+
+stage {
+
+  # CHARACTERS
+  parallel {
+    sub "\u0410", "A"
+    sub "\u0411", "B"
+    sub "\u0412", "V"
+    sub "\u0413", "G"
+    sub "\u0414", "D"
+    sub "\u0415", "E"
+    sub "\u0416", "Zh"
+    sub "\u0417", "Z"
+    sub "\u0418", "I"
+    sub "\u0419", "Y"
+    sub "\u041a", "K"
+    sub "\u041b", "L"
+    sub "\u041c", "M"
+    sub "\u041d", "N"
+    sub "\u041e", "O"
+    sub "\u041f", "P"
+    sub "\u0420", "R"
+    sub "\u0421", "S"
+    sub "\u0422", "T"
+    sub "\u0423", "U"
+    sub "\u0424", "F"
+    sub "\u0425", "H"
+    sub "\u0426", "Ts"
+    sub "\u0427", "Ch"
+    sub "\u0428", "Sh"
+    sub "\u0429", "Sht"
+    sub "\u042a", "A"
+    sub "\u042c", "Y"
+    sub "\u042e", "Yu"
+    sub "\u042f", "Ya"
+    
+    sub "\u0430", "a"
+    sub "\u0431", "b"
+    sub "\u0432", "v"
+    sub "\u0433", "g"
+    sub "\u0434", "d"
+    sub "\u0435", "e"
+    sub "\u0436", "zh"
+    sub "\u0437", "z"
+    sub "\u0438", "i"
+    sub "\u0439", "y"
+    sub "\u043a", "k"
+    sub "\u043b", "l"
+    sub "\u043c", "m"
+    sub "\u043d", "n"
+    sub "\u043e", "o"
+    sub "\u043f", "p"
+    sub "\u0440", "r"
+    sub "\u0441", "s"
+    sub "\u0442", "t"
+    sub "\u0443", "u"
+    sub "\u0444", "f"
+    sub "\u0445", "h"
+    sub "\u0446", "ts"
+    sub "\u0447", "ch"
+    sub "\u0448", "sh"
+    sub "\u0449", "sht"
+    sub "\u044a", "a"
+    sub "\u044c", "y"
+    sub "\u044e", "yu"
+    sub "\u044f", "ya"
+    
+    # note 2
+    sub "\u046A", "Ŭ" # Ѫ
+    sub "\u046B", "ŭ" # ѫ
+    
+    # note[3]
+    sub "\u0462", "YE" # Ѣ
+    sub "\u0463", "ye" # ѣ
+  }
+
+}
+

data/maps/az-aze-Cyrl-Latn-1939.imp

@@ -0,0 +1,105 @@
+metadata {
+  authority_id: az
+  id: 1939
+  language: iso-639-2:aze
+  source_script: Cyrl
+  destination_script: Latn
+  url: https://omniglot.com/writing/azeri.htm
+  creation_date: 1939
+  description: |
+    In 1939 Joseph Stalin ordered the Cyrillic alphabet to be used by Azeri speakers in the Soviet Union.
+
+  notes:
+    # from Wikipedia
+    - The letter Ц, intended for the sound [ц] in borrowed words, was used in the Azerbaijani Cyrillic alphabet until 1951. In the Azeri language, the sound [c] usually becomes [s].
+}
+
+tests {
+  test "Юя", "Yuya"
+  # from internet
+  test "Азәрбайҹан әлифбасы", "Azərbaycan əlifbası"
+  test "Бүтүн инсанлар ләйагәт вә һүгугларына ҝөрә азад бәрабәр доғулурлар.\nОнларын шүурлары вә виҹданлары вар вә бир-бирләринә мүнасибәтдә гардашлыг руһунда давранмалыдырлар.", "Bütün insanlar ləyaqət və hüquqlarına görə azad bərabər doğulurlar.\nOnların şüurları və vicdanları var və bir-birlərinə münasibətdə qardaşlıq ruhunda davranmalıdırlar."
+}
+
+stage {
+
+  # CHARACTERS
+  parallel {
+    sub "А", "A" # А
+    sub "Б", "B" # Б
+    sub "В", "V" # В
+    sub "Г", "Q" # Г
+    sub "Ғ", "Ğ" # Ғ
+    sub "Д", "D" # Д
+    sub "Е", "E" # Е
+    sub "Ә", "Ə" # Ә
+    sub "Ж", "J" # Ж
+    sub "З", "Z" # З
+    sub "И", "İ" # И
+    sub "Й", "Y" # Й
+    sub "К", "K" # К
+    sub "Ҝ", "G" # Ҝ
+    sub "Л", "L" # Л
+    sub "М", "M" # М
+    sub "Н", "N" # Н
+    sub "О", "O" # О
+    sub "Ө", "Ö" # Ө
+    sub "П", "P" # П
+    sub "Р", "R" # Р
+    sub "С", "S" # С
+    sub "Т", "T" # Т
+    sub "У", "U" # У
+    sub "Ү", "Ü" # Ү
+    sub "Ф", "F" # Ф
+    sub "Х", "X" # Х
+    sub "Ц", "s" # Ц note[1]
+    sub "Һ", "H" # Һ
+    sub "Ч", "Ç" # Ч
+    sub "Ҹ", "C" # Ҹ
+    sub "Ш", "Ş" # Ш
+    sub "Ы", "I" # Ы
+    sub "Э", "E" # Э
+    sub "Ю", "Yu" # Ю
+    sub "Я", "Ya" # Я
+
+    sub "а", "a" # а
+    sub "б", "b" # б
+    sub "в", "v" # в
+    sub "г", "q" # г
+    sub "ғ", "ğ" # ғ
+    sub "д", "d" # д
+    sub "е", "e" # е
+    sub "ә", "ə" # ә
+    sub "ж", "j" # ж
+    sub "з", "z" # з
+    sub "и", "i" # и
+    sub "й", "y" # й
+    sub "к", "k" # к
+    sub "ҝ", "g" # ҝ
+    sub "л", "l" # л
+    sub "м", "m" # м
+    sub "н", "n" # н
+    sub "о", "o" # о
+    sub "ө", "ö" # ө
+    sub "п", "p" # п
+    sub "р", "r" # р
+    sub "с", "s" # с
+    sub "т", "t" # т
+    sub "у", "u" # у
+    sub "ү", "ü" # ү
+    sub "ф", "f" # ф
+    sub "х", "x" # х
+    sub "ц", "s" # ц note[1]
+    sub "һ", "h" # һ
+    sub "ч", "ç" # ч
+    sub "ҹ", "c" # ҹ
+    sub "ш", "ş" # ш
+    sub "ы", "ı" # ы
+    sub "э", "e" # э
+    sub "ю", "yu" # ю
+    sub "я", "ya" # я
+
+    sub "'", "" # '
+  }
+
+}