interscript-maps 2.4.3 → 2.5.0

data/maps/bis-tml-Taml-Latn-13194-1991.isc

@@ -0,0 +1,166 @@
+system "BIS:tml-Taml:Latn:13194-1991" {
+
+metadata {
+  authority_id bis
+  id 13194-1991
+  language iso-639-3:tml
+  source_script Taml
+  destination_script Latn
+  name Indian script code for information interchange (ISCII) -- Tamil Romanization
+
+  #
+  creation_date 1991
+  description {    IS 13194 (1991): Indian script code for information
+    interchange - ISCII [LITD 20: Indian Language Technologies
+    and Products]
+  }
+
+  notes {
+    note "      Exception: Anusvāra is transliterated by:\n\n      a) ṅ before gutturals,\n      b) ñ before palatals,\n      c) ṇ before cerebrals,\n      d) n before dentals, and\n      e) m before labials."
+  }
+}
+
+tests {
+   "இளைஞர்களின் உறுதியான மனநிலையை பிரதிபலிக்கிறது: நீட் தேர்வில் 85-90 சதவீத மாணவர்கள் பங்கேற்பு - ரமேஷ் பொக்ரியால்" ->  "iḷaiñrkḷiṉ uṟutiyāṉ mṉnilaiyai pirtiplikkiṟtu: nīṭ tērvil 85-90 ctvīt māṇvrkḷ pṅkēṟpu - rmēṣ pokriyāl"
+   "நாடாளுமன்றத்தில் 4 மசோதாக்களை எதிர்க்க காங்கிரஸ் முடிவு - ஜெயராம் ரமேஷ்" ->  "nāṭāḷumṉṟttil 4 mcōtākkḷai etirkk kāṅkirs muṭivu - jeyrām rmēṣ"
+   "கர்நாடகாவில் மேலும் 9,894 பேருக்கு கொரோனா தொற்று உறுதி" ->  "krnāṭkāvil mēlum 9,894 pērukku korōṉā toṟṟu uṟuti"
+   "ஐதராபாத்துக்கு கைகொடுக்குமா அதிரடி?" ->  "aitrāpāttukku kaikoṭukkumā atirṭi?"
+   "அமெரிக்க ஓபன் டென்னிஸ்: இறுதிப்போட்டியில் டொமினிக்-ஸ்வெரேவ்" ->  "amerikk ŏpṉ ṭeṉṉis: iṟutippōṭṭiyil ṭomiṉik-sverēv"
+   "ஐ.பி.எல். கிரிக்கெட்டில் களம் இறங்கும் அமெரிக்க வீரர்" ->  "ai.pi.el. kirikkeṭṭil kḷm iṟṅkum amerikk vīrr"
+   "அமெரிக்க ஓபன் டென்னிஸ்; நவோமி ஒசாகா சாம்பியன் பட்டம் வென்றார்" ->  "amerikk ŏpṉ ṭeṉṉis; nvōmi ocākā cāmpiyṉ pṭṭm veṉṟār"
+   "புதிய கல்விக்கொள்கைக்கு எதிர்ப்பு: முன்னாள் துணைவேந்தர்கள் 20 பேர் பிரதமருக்கு கடிதம்" ->  "putiy klvikkoḷkaikku etirppu: muṉṉāḷ tuṇaivēntrkḷ 20 pēr pirtmrukku kṭitm"
+   "இந்த ஆண்டு ஐ.பி.எல். கோப்பையை எந்த அணி வெல்லும்? - கெவின் பீட்டர்சன் கணிப்பு" ->  "int āṇṭu ai.pi.el. kōppaiyai ent aṇi vellum? - keviṉ pīṭṭrcṉ kṇippu"
+   "இந்திய எண்ணெய் கப்பலில் தீ: விபத்து குறித்த எச்சரிக்கையை கப்பல் அதிகாரிகள் புறக்கணித்தனர் - இலங்கை கோர்ட்டு தகவல்" ->  "intiy eṇṇey kpplil tī: vipttu kuṟitt eccrikkaiyai kppl atikārikḷ puṟkkṇittṉr - ilṅkai kōrṭṭu tkvl"
+}
+
+stage main {
+
+  # RULES
+  # note
+  sub {
+    from "\u0B82"
+    to "ṅ"
+    after any("கங")
+  }
+  sub {
+    from "\u0B82"
+    to "ñ"
+    after any("சஜஞ")
+  }
+  sub {
+    from "\u0B82"
+    to "ṇ"
+    after any("டண")
+  }
+  sub {
+    from "\u0B82"
+    to "n"
+    after any("தநன")
+  }
+  sub {
+    from "\u0B82"
+    to "m"
+    after any("பம")
+  }
+  # Those aren't tested
+
+  # CHARACTERS
+  parallel {
+    sub "அ" "a"
+    sub "ஆ" "ā"
+    sub "இ" "i"
+    sub "ஈ" "ī"
+    sub "உ" "u"
+    sub "ஊ" "ū"
+
+    sub "எ" "e"
+    sub "ஏ" "ē"
+    sub "ஐ" "ai"
+
+    sub "ஒ" "o"
+    sub "ஓ" "ŏ"
+    sub "ஔ" "au"
+
+    # II. Consonants (see Note 2)
+    # Gutturals
+    sub "க" "k"
+    sub "ங" "ṅ"
+
+    # Palatals
+    sub "ச" "c"
+    sub "ஜ" "j"
+    sub "ஞ" "ñ"
+
+    # Cerebrals
+    sub "ட" "ṭ"
+    sub "ண" "ṇ"
+
+    # Dentals
+    sub "த" "t"
+    sub "ந" "n"
+    sub "ன" "ṉ"
+
+    # Labials
+    sub "ப" "p"
+    sub "ம" "m"
+
+    # Semivowels
+    sub "ய" "y"
+    sub "ர" "r"
+    sub "ற" "ṟ"
+    sub "ல" "l"
+    sub "ள" "ḷ"
+    sub "ழ" "ẕ"
+
+
+    # Sibilants
+    sub "வ" "v"
+    sub "ஶ" "ś"
+    sub "ஷ" "ṣ"
+    sub "ஸ" "s"
+
+
+    # Aspirate
+    sub "ஹ" "h"
+
+
+    # Bisarga
+    sub "ஃ" "ḥ"
+
+    # Anusvāra
+    sub "ஂ" "ṃ"
+
+    # Medials # Needed for connecting constants
+
+    sub "ா" "ā"
+    sub "ி" "i"
+    sub "ீ" "ī"
+    sub "ு" "u"
+    sub "ூ" "ū"
+
+    sub "ൃ" "ṛ"
+    sub "ൄ" "ṝ"
+
+
+    sub "ெ" "e"
+    sub "ே" "ē"
+    sub "ை" "ai"
+    sub "ொ" "o"
+    sub "ோ" "ō"
+    sub "ௌ" "au"
+
+
+    sub "ൺ" "n"
+    sub "ൻ" "ṇ"
+
+    sub "्" ""
+    sub "്" ""
+    sub "்" ""
+    sub "ൗ" ""
+    sub "‍" ""
+    sub "‌" ""
+  }
+
+  compose
+}
+}

data/maps/by-bel-Cyrl-Latn-1998.isc

@@ -0,0 +1,158 @@
+system "BY:bel-Cyrl:Latn:1998" {
+
+metadata {
+  authority_id by
+  id 1998
+  language iso-639-2:bel
+  source_script Cyrl
+  destination_script Latn
+  name National System of Geographic Names Transmission into Roman Alphabet in Belarus
+  url https://unstats.un.org/unsd/geoinfo/UNGEGN/docs/20th-gegn-docs/20th_gegn_WP34.pdf
+  # page 48
+  creation_date 1998
+  description {    The national system of romanization for Belarusian was approved by the State Committee and Land Resources,
+    Geodesy and Cartography, Republic of Belarus, on 20 March, 1998. This scheme was also supported by
+    the Y. Kolas Institute of Linguistics and the Republic Committee on Toponymy at the Belarusian Academy of Sciences.
+    While the system is still based on GOST 1983, it takes more precisely into account the peculiarities of
+    the Belarusian orthography. The system is reversible though there may exist some ambiguous consonant combinations.
+  }
+}
+
+tests {
+  # the same as in by-bel-Cyrl-Latn-2007
+   "Аршанскi" ->  "Aršanski"
+   "Бешанковічы" ->  "Biešankovičy"
+   "Віцебск" ->  "Viciebsk"
+   "Гомель" ->  "Homieĺ"
+   "Гаўя" ->  "Haŭja"
+   "Добруш" ->  "Dobruš"
+   "Ельск" ->  "Jeĺsk"
+   "Бабаедава" ->  "Babajedava"
+   "Венцавічы" ->  "Viencavičy"
+   "Ёды" ->  "Jody"
+   "Вераб'ёвічы" ->  "Vierabjovičy"
+   "Мёры" ->  "Miory"
+   "Зэльва" ->  "Zeĺva"
+   "Iванава" ->  "Ivanava"
+   "Iўе" ->  "Iŭje"
+   "Лагойск" ->  "Lahojsk"
+   "Круглае" ->  "Kruhlaje"
+   "Лошыца" ->  "Lošyca"
+   "Любань" ->  "Liubań"
+   "Магілёў" ->  "Mahilioŭ"
+   "Нясвіж" ->  "Niasviž"
+   "Орша" ->  "Orša"
+   "Паставы" ->  "Pastavy"
+   "Рагачоў" ->  "Rahačoŭ"
+   "Смаргонь" ->  "Smarhoń"
+   "Талачын" ->  "Talačyn"
+   "Узда" ->  "Uzda"
+   "Шаркаўшчына" ->  "Šarkaŭščyna"
+   "Фаніпаль" ->  "Fanipaĺ"
+   "Хоцімск" ->  "Chocimsk"
+   "Цёмны Лес" ->  "Ciomny Lies"
+   "Чавусы" ->  "Čavusy"
+   "Шумілiна" ->  "Šumilina"
+   "Чыгірынка" ->  "Čyhirynka"
+   "Чэрвень" ->  "Červień"
+   "Друць" ->  "Druć"
+   "Чачэрск" ->  "Čačersk"
+   "Юхнаўка" ->  "Juchnaŭka"
+   "Гаюціна" ->  "Hajucina"
+   "Цюрлі" ->  "Ciurli"
+   "Любонічы" ->  "Liuboničy"
+   "Ямнае" ->  "Jamnaje"
+   "Баяры" ->  "Bajary"
+   "Валяр'яны" ->  "Valiarjany"
+   "Вязынка" ->  "Viazynka"
+}
+
+dependency "gost-rus-Cyrl-Latn-16876-71-1983" as cyrllatn
+dependency "var-Cyrl"
+
+stage main {
+  # RULES
+  sub {
+    from "\u0415"
+    to "IE"
+    before bel_consonant
+  }
+  sub {
+    from "\u0435"
+    to "ie"
+    before bel_consonant
+  }
+  sub {
+    from "\u0401"
+    to "IO"
+    before bel_consonant
+  }
+  sub {
+    from "\u0451"
+    to "io"
+    before bel_consonant
+  }
+  sub {
+    from "\u042E"
+    to "IU"
+    before bel_consonant
+  }
+  sub {
+    from "\u044E"
+    to "iu"
+    before bel_consonant
+  }
+  sub {
+    from "\u042F"
+    to "IA"
+    before bel_consonant
+  }
+  sub {
+    from "\u044F"
+    to "ia"
+    before bel_consonant
+  }
+
+  parallel {
+    sub "\u0417\u042C" "Ź"
+    sub "\u0437\u044C" "ź"
+    sub "\u041B\u042C" "Ĺ"
+    sub "\u043B\u044C" "ĺ"
+    sub "\u0421\u042C" "Ś"
+    sub "\u0441\u044C" "ś"
+    sub "\u0426\u042C" "Ć"
+    sub "\u0446\u044C" "ć"
+    sub "\u041D\u042C" "Ń"
+    sub "\u043D\u044C" "ń"
+  }
+
+  # CHARACTERS
+  parallel {
+    sub "\u0406" "I"
+    sub "\u0413" "H"
+    sub "\u0415" "Je"
+    sub "\u0401" "Jo"
+    sub "\u040E" "Ŭ"
+    sub "\u0425" "Ch"
+    sub "\u042C" ""
+    sub "\u042D" "E"
+    sub "\u042E" "Ju"
+    sub "\u042F" "Ja"
+
+    sub "\u0433" "h"
+    sub "\u0456" "i"
+    sub "\u0435" "je"
+    sub "\u0451" "jo"
+    sub "\u045E" "ŭ"
+    sub "\u0445" "ch"
+    sub "\u044C" ""
+    sub "\u044D" "e"
+    sub "\u044E" "ju"
+    sub "\u044F" "ja"
+
+    sub "\u0027" ""
+  }
+
+  run  map.cyrllatn.stage.main
+}
+}

data/maps/by-bel-Cyrl-Latn-2007.isc

@@ -0,0 +1,86 @@
+system "BY:bel-Cyrl:Latn:2007" {
+
+metadata {
+  authority_id by
+  id 2007
+  language iso-639-2:bel
+  source_script Cyrl
+  destination_script Latn
+  name Transliteration of geographical names of the Republic of Belarus in letters of the Latin alphabet (2007)
+  url http://www.pravo.by/pdf/2007-159/2007-159(027-028).pdf
+  creation_date 2007
+  description {    RESOLUTION OF THE STATE COMMITTEE
+    ON PROPERTY OF THE REPUBLIC OF BELARUS June 11, 2007 No. 38
+
+    Regulatory Legal Acts Of The National Bank, National Academy Of Sciences Of
+    Belarus, Ministries, Other Republican Government Bodies 8/16668 (06/18/2007)
+    On amendments and additions to the Instructions for the transliteration of
+    geographical names of the Republic of Belarus in letters of the Latin
+    alphabet
+
+    Based on the Regulation on the State Property Committee of the Republic of Belarus,
+    approved by the Decree of the Council of Ministers of the Republic of Belarus dated July 29, 2006
+    No. 958 "Issues of the State Committee on Property of the Republic of Belarus"
+  }
+}
+
+tests {
+   "Аршанскi" ->  "Aršanski"
+   "Бешанковічы" ->  "Biešankovičy"
+   "Віцебск" ->  "Viciebsk"
+   "Гомель" ->  "Homieĺ"
+   "Гаўя" ->  "Haŭja"
+   "Добруш" ->  "Dobruš"
+   "Ельск" ->  "Jeĺsk"
+   "Бабаедава" ->  "Babajedava"
+   "Венцавічы" ->  "Viencavičy"
+   "Ёды" ->  "Jody"
+   "Вераб'ёвічы" ->  "Vierabjovičy"
+   "Мёры" ->  "Miory"
+   "Зэльва" ->  "Zeĺva"
+   "Iванава" ->  "Ivanava"
+   "Iўе" ->  "Iŭje"
+   "Лагойск" ->  "Lahojsk"
+   "Круглае" ->  "Kruhlaje"
+   "Лошыца" ->  "Lošyca"
+   "Любань" ->  "Liubań"
+   "Магілёў" ->  "Mahilioŭ"
+   "Нясвіж" ->  "Niasviž"
+   "Орша" ->  "Orša"
+   "Паставы" ->  "Pastavy"
+   "Рагачоў" ->  "Rahačoŭ"
+   "Смаргонь" ->  "Smarhoń"
+   "Талачын" ->  "Talačyn"
+   "Узда" ->  "Uzda"
+   "Шаркаўшчына" ->  "Šarkaŭščyna"
+   "Фаніпаль" ->  "Fanipaĺ"
+   "Хоцімск" ->  "Chocimsk"
+   "Цёмны Лес" ->  "Ciomny Lies"
+   "Чавусы" ->  "Čavusy"
+   "Шумілiна" ->  "Šumilina"
+   "Чыгірынка" ->  "Čyhirynka"
+   "Чэрвень" ->  "Červień"
+   "Друць" ->  "Druć"
+   "Чачэрск" ->  "Čačersk"
+   "Юхнаўка" ->  "Juchnaŭka"
+   "Гаюціна" ->  "Hajucina"
+   "Цюрлі" ->  "Ciurli"
+   "Любонічы" ->  "Liuboničy"
+   "Ямнае" ->  "Jamnaje"
+   "Баяры" ->  "Bajary"
+   "Валяр'яны" ->  "Valiarjany"
+   "Вязынка" ->  "Viazynka"
+
+  # UN card tests
+   "Беларусь" ->  "Bielaruś"
+   "Минск" ->  "Minsk"
+}
+
+dependency "by-bel-Cyrl-Latn-1998" as cyrllatn
+
+stage main {
+
+  run  map.cyrllatn.stage.main
+
+}
+}