interscript 0.1.1 → 0.1.6

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

@@ -0,0 +1,155 @@
+---
+authority_id: bis
+id: 1991
+language: tml
+source_script: Taml
+destination_script: Latn
+name: Indian script code for information interchange - ISCII - Tamil Romanization
+#url:
+creation_date: 1991
+description: |
+  IS 13194 (1991): Indian script code for information
+  interchange - ISCII [LITD 20: Indian Language Technologies
+  and Products]
+
+notes:
+  - |
+    Exception: Anusvāra is transliterated by:
+
+    a) ṅ before gutturals,
+    b) ñ before palatals,
+    c) ṇ before cerebrals,
+    d) n before dentals, and
+    e) m before labials.
+
+tests:
+  - source: "இளைஞர்களின் உறுதியான மனநிலையை பிரதிபலிக்கிறது: நீட் தேர்வில் 85-90 சதவீத மாணவர்கள் பங்கேற்பு - ரமேஷ் பொக்ரியால்"
+    expected: "iḷaiñrkḷiṉ uṟutiyāṉ mṉnilaiyai pirtiplikkiṟtu: nīṭ tērvil 85-90 ctvīt māṇvrkḷ pṅkēṟpu - rmēṣ pokriyāl"
+  - source: "நாடாளுமன்றத்தில் 4 மசோதாக்களை எதிர்க்க காங்கிரஸ் முடிவு - ஜெயராம் ரமேஷ்"
+    expected: "nāṭāḷumṉṟttil 4 mcōtākkḷai etirkk kāṅkirs muṭivu - jeyrām rmēṣ"
+  - source: "கர்நாடகாவில் மேலும் 9,894 பேருக்கு கொரோனா தொற்று உறுதி"
+    expected: "krnāṭkāvil mēlum 9,894 pērukku korōṉā toṟṟu uṟuti"
+  - source: "ஐதராபாத்துக்கு கைகொடுக்குமா அதிரடி?"
+    expected: "aitrāpāttukku kaikoṭukkumā atirṭi?"
+  - source: "அமெரிக்க ஓபன் டென்னிஸ்: இறுதிப்போட்டியில் டொமினிக்-ஸ்வெரேவ்"
+    expected: "amerikk ŏpṉ ṭeṉṉis: iṟutippōṭṭiyil ṭomiṉik-sverēv"
+  - source: "ஐ.பி.எல். கிரிக்கெட்டில் களம் இறங்கும் அமெரிக்க வீரர்"
+    expected: "ai.pi.el. kirikkeṭṭil kḷm iṟṅkum amerikk vīrr"
+  - source: "அமெரிக்க ஓபன் டென்னிஸ்; நவோமி ஒசாகா சாம்பியன் பட்டம் வென்றார்"
+    expected: "amerikk ŏpṉ ṭeṉṉis; nvōmi ocākā cāmpiyṉ pṭṭm veṉṟār"
+  - source: "புதிய கல்விக்கொள்கைக்கு எதிர்ப்பு: முன்னாள் துணைவேந்தர்கள் 20 பேர் பிரதமருக்கு கடிதம்"
+    expected: "putiy klvikkoḷkaikku etirppu: muṉṉāḷ tuṇaivēntrkḷ 20 pēr pirtmrukku kṭitm"
+  - source: "இந்த ஆண்டு ஐ.பி.எல். கோப்பையை எந்த அணி வெல்லும்? - கெவின் பீட்டர்சன் கணிப்பு"
+    expected: "int āṇṭu ai.pi.el. kōppaiyai ent aṇi vellum? - keviṉ pīṭṭrcṉ kṇippu"
+  - source: "இந்திய எண்ணெய் கப்பலில் தீ: விபத்து குறித்த எச்சரிக்கையை கப்பல் அதிகாரிகள் புறக்கணித்தனர் - இலங்கை கோர்ட்டு தகவல்"
+    expected: "intiy eṇṇey kpplil tī: vipttu kuṟitt eccrikkaiyai kppl atikārikḷ puṟkkṇittṉr - ilṅkai kōrṭṭu tkvl"
+
+map:
+
+  rules:
+    # note
+    - pattern: \u0B82(?=[கங])
+      result: ṅ
+    - pattern: \u0B82(?=[சஜஞ])
+      result: ñ
+    - pattern: \u0B82(?=[டண])
+      result: ṇ
+    - pattern: \u0B82(?=[தநன])
+      result: n
+    - pattern: \u0B82(?=[பம])
+      result: m
+
+  characters:
+    'அ': 'a'
+    'ஆ': 'ā'
+    'இ': 'i'
+    'ஈ': 'ī'
+    'உ': 'u'
+    'ஊ': 'ū'
+
+    'எ': 'e'
+    'ஏ': 'ē'
+    'ஐ': 'ai'
+
+    'ஒ': 'o'
+    'ஓ': 'ŏ'
+    'ஔ': 'au'
+
+    # II. Consonants (see Note 2)
+    # Gutturals
+    'க': 'k'
+    'ங': 'ṅ'
+
+    # Palatals
+    'ச': 'c'
+    'ஜ': 'j'
+    'ஞ': 'ñ'
+
+    # Cerebrals
+    'ட':  'ṭ'
+    'ண':  'ṇ'
+
+    # Dentals
+    'த': 't'
+    'ந':  'n'
+    'ன': 'ṉ'
+
+    # Labials
+    'ப': 'p'
+    'ம': 'm'
+
+    # Semivowels
+    'ய': 'y'
+    'ர': 'r'
+    'ற': 'ṟ'
+    'ல': 'l'
+    'ள': 'ḷ'
+    'ழ': 'ẕ'
+
+
+    # Sibilants
+    'வ': 'v'
+    'ஶ': 'ś'
+    'ஷ': 'ṣ'
+    'ஸ': 's'
+
+
+    # Aspirate
+    'ஹ': 'h'
+
+
+    # Bisarga
+    'ஃ': 'ḥ'
+
+    # Anusvāra
+    'ஂ': 'ṃ'
+
+    # Medials # Needed for connecting constants
+
+    'ா': 'ā'
+    'ி': 'i'
+    'ீ': 'ī'
+    'ு': 'u'
+    'ூ': 'ū'
+
+    'ൃ': "ṛ"
+    'ൄ': "ṝ"
+
+
+    'ெ': "e"
+    'ே': "ē"
+    'ை': "ai"
+    'ொ': 'o'
+    'ோ': 'ō'
+    'ௌ': 'au'
+
+
+    'ൺ': 'n'
+    'ൻ': 'ṇ'
+
+    '्': ''
+    '്': ''
+    '்': ''
+    'ൗ': ''
+    "‍": '' # no need for zero with joiner
+    "‌": '' # no need for zero with non joiner

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

@@ -0,0 +1,168 @@
+---
+authority_id: un
+id: 1998
+language: 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
+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
+  - source: Аршанскi
+    expected: Aršanski
+  - source: Бешанковічы
+    expected: Biešankovičy
+  - source: Віцебск
+    expected: Viciebsk
+  - source: Гомель
+    expected: Homieĺ
+  - source: Гаўя
+    expected: Haŭja
+  - source: Добруш
+    expected: Dobruš
+  - source: Ельск
+    expected: Jeĺsk
+  - source: Бабаедава
+    expected: Babajedava
+  - source: Венцавічы
+    expected: Viencavičy
+  - source: Ёды
+    expected: Jody
+  - source: Вераб'ёвічы
+    expected: Vierabjovičy
+  - source: Мёры
+    expected: Miory
+  - source: Зэльва
+    expected: Zeĺva
+  - source: Iванава
+    expected: Ivanava
+  - source: Iўе
+    expected: Iŭje
+  - source: Лагойск
+    expected: Lahojsk
+  - source: Круглае
+    expected: Kruhlaje
+  - source: Лошыца
+    expected: Lošyca
+  - source: Любань
+    expected: Liubań
+  - source: Магілёў
+    expected: Mahilioŭ
+  - source: Нясвіж
+    expected: Niasviž
+  - source: Орша
+    expected: Orša
+  - source: Паставы
+    expected: Pastavy
+  - source: Рагачоў
+    expected: Rahačoŭ
+  - source: Смаргонь
+    expected: Smarhoń
+  - source: Талачын
+    expected: Talačyn
+  - source: Узда
+    expected: Uzda
+  - source: Шаркаўшчына
+    expected: Šarkaŭščyna
+  - source: Фаніпаль
+    expected: Fanipaĺ
+  - source: Хоцімск
+    expected: Chocimsk
+  - source: Цёмны Лес
+    expected: Ciomny Lies
+  - source: Чавусы
+    expected: Čavusy
+  - source: Шумілiна
+    expected: Šumilina
+  - source: Чыгірынка
+    expected: Čyhirynka
+  - source: Чэрвень
+    expected: Červień
+  - source: Друць
+    expected: Druć
+  - source: Чачэрск
+    expected: Čačersk
+  - source: Юхнаўка
+    expected: Juchnaŭka
+  - source: Гаюціна
+    expected: Hajucina
+  - source: Цюрлі
+    expected: Ciurli
+  - source: Любонічы
+    expected: Liuboničy
+  - source: Ямнае
+    expected: Jamnaje
+  - source: Баяры
+    expected: Bajary
+  - source: Валяр'яны
+    expected: Valiarjany
+  - source: Вязынка
+    expected: Viazynka
+
+map:
+  inherit: gost-rus-Cyrl-Latn-16876-71-1983
+
+  rules:
+    - pattern: (?<=[БбВвГгДдЖжЗзЙйКкЛлМмНнПпРрСсТтФфХхЦцЧчШш])\u0415 # Е after consonants
+      result: IE
+    - pattern: (?<=[БбВвГгДдЖжЗзЙйКкЛлМмНнПпРрСсТтФфХхЦцЧчШш])\u0435 # е after consonants
+      result: ie
+    - pattern: (?<=[БбВвГгДдЖжЗзЙйКкЛлМмНнПпРрСсТтФфХхЦцЧчШш])\u0401 # Ё after consonants
+      result: IO
+    - pattern: (?<=[БбВвГгДдЖжЗзЙйКкЛлМмНнПпРрСсТтФфХхЦцЧчШш])\u0451 # ё after consonants
+      result: io
+    - pattern: (?<=[БбВвГгДдЖжЗзЙйКкЛлМмНнПпРрСсТтФфХхЦцЧчШш])\u042E # Ю after consonants
+      result: IU
+    - pattern: (?<=[БбВвГгДдЖжЗзЙйКкЛлМмНнПпРрСсТтФфХхЦцЧчШш])\u044E # ю after consonants
+      result: iu
+    - pattern: (?<=[БбВвГгДдЖжЗзЙйКкЛлМмНнПпРрСсТтФфХхЦцЧчШш])\u042F # Я after consonants
+      result: IA
+    - pattern: (?<=[БбВвГгДдЖжЗзЙйКкЛлМмНнПпРрСсТтФфХхЦцЧчШш])\u044F # я after consonants
+      result: ia
+    - pattern: \u0417\u042C # ЗЬ
+      result: "\u0179"
+    - pattern: \u0437\u044C # зь
+      result: "\u017A"
+    - pattern: \u041B\u042C # ЛЬ
+      result: "\u0139"
+    - pattern: \u043B\u044C # ль
+      result: "\u013A"
+    - pattern: \u0421\u042C # СЬ
+      result: "\u015A"
+    - pattern: \u0441\u044C # сь
+      result: "\u015B"
+    - pattern: \u0426\u042C # ЦЬ
+      result: "\u0106"
+    - pattern: \u0446\u044C # ць
+      result: "\u0107"
+    - pattern: \u041D\u042C # НЬ
+      result: "\u0143"
+    - pattern: \u043D\u044C # нь
+      result: "\u0144"
+  characters:
+    '\u0406' : 'I'  # І
+    '\u0413' : 'H'  # Г
+    '\u0415' : 'Je' # Е
+    '\u0401' : 'Jo' # Ё
+    '\U040E' : 'Ŭ'  # Ў
+    '\u0425' : 'Ch' # Х
+    '\u042C' : ''   # Ь
+    '\u042D' : 'E'  # Э
+
+    '\u0433' : 'h'  # г
+    '\u0456' : 'i'  # і
+    '\u0435' : 'je' # е
+    '\u0451' : 'jo' # ё
+    '\u045E' : 'ŭ'  # ў
+    '\u0445' : 'ch' # х
+    '\u044C' : ''   # ь
+    '\u044D' : 'e'  # э
+
+    '\u0027' : ''   # '

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

@@ -0,0 +1,115 @@
+---
+authority_id: by
+id: 2007
+language: bel
+source_script: Cyrl
+destination_script: Latn
+name: REGULATORY LEGAL ACTS OF THE NATIONAL BANK, NATIONAL ACADEMY OF SCIENCES OF BELARUS, MINISTRIES, OTHER REPUBLICAN GOVERNMENT BODIES 8/16668
+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
+
+  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 Belarusdated July 29, 2006
+  No. 958 "Issues of the State Committee on Property of the Republic of Belarus"
+
+tests:
+  - source: Аршанскi
+    expected: Aršanski
+  - source: Бешанковічы
+    expected: Biešankovičy
+  - source: Віцебск
+    expected: Viciebsk
+  - source: Гомель
+    expected: Homieĺ
+  - source: Гаўя
+    expected: Haŭja
+  - source: Добруш
+    expected: Dobruš
+  - source: Ельск
+    expected: Jeĺsk
+  - source: Бабаедава
+    expected: Babajedava
+  - source: Венцавічы
+    expected: Viencavičy
+  - source: Ёды
+    expected: Jody
+  - source: Вераб'ёвічы
+    expected: Vierabjovičy
+  - source: Мёры
+    expected: Miory
+  - source: Зэльва
+    expected: Zeĺva
+  - source: Iванава
+    expected: Ivanava
+  - source: Iўе
+    expected: Iŭje
+  - source: Лагойск
+    expected: Lahojsk
+  - source: Круглае
+    expected: Kruhlaje
+  - source: Лошыца
+    expected: Lošyca
+  - source: Любань
+    expected: Liubań
+  - source: Магілёў
+    expected: Mahilioŭ
+  - source: Нясвіж
+    expected: Niasviž
+  - source: Орша
+    expected: Orša
+  - source: Паставы
+    expected: Pastavy
+  - source: Рагачоў
+    expected: Rahačoŭ
+  - source: Смаргонь
+    expected: Smarhoń
+  - source: Талачын
+    expected: Talačyn
+  - source: Узда
+    expected: Uzda
+  - source: Шаркаўшчына
+    expected: Šarkaŭščyna
+  - source: Фаніпаль
+    expected: Fanipaĺ
+  - source: Хоцімск
+    expected: Chocimsk
+  - source: Цёмны Лес
+    expected: Ciomny Lies
+  - source: Чавусы
+    expected: Čavusy
+  - source: Шумілiна
+    expected: Šumilina
+  - source: Чыгірынка
+    expected: Čyhirynka
+  - source: Чэрвень
+    expected: Červień
+  - source: Друць
+    expected: Druć
+  - source: Чачэрск
+    expected: Čačersk
+  - source: Юхнаўка
+    expected: Juchnaŭka
+  - source: Гаюціна
+    expected: Hajucina
+  - source: Цюрлі
+    expected: Ciurli
+  - source: Любонічы
+    expected: Liuboničy
+  - source: Ямнае
+    expected: Jamnaje
+  - source: Баяры
+    expected: Bajary
+  - source: Валяр'яны
+    expected: Valiarjany
+  - source: Вязынка
+    expected: Viazynka
+
+map:
+  inherit: by-bel-Cyrl-Latn-1998

data/maps/dos-nep-Deva-Latn-1997.yaml

@@ -0,0 +1,33 @@
+---
+authority_id: bgnpcgn
+id: 1997
+language: nep
+source_script: Deva
+destination_script: Latn
+name: Nepali Romanization, 1997
+url: http://nationalgeoportal.gov.np/old/pdf/translation2.pdf
+creation_date: 1997
+description: |
+  Survey Department, Ministry of Land Management, Cooperatives and Poverty Alleviation, Government of Nepal.
+
+notes:
+  - |
+    ं (anusvara) is rendered by
+    ṅ before क, ख, ग, and घ
+    ñ before च, छ, ज, and झ
+    ṇ before ट, ठ, ड, and ढ
+    n before त, थ, द, and ध
+    ṁ before य, र, ल, व, श, ष, स
+
+tests:
+  - source: "दुःख"
+    expected: "duhkh"
+
+map:
+
+  inherit: "bgnpcgn-nep-Deva-Latn-2011"
+
+  characters:
+
+    # Bisarga
+    'ः': 'h'