interscript 0.1.3 → 0.1.9

data/maps/iso-tha-Thai-Latn-11940-1998.yaml

@@ -1,13 +1,13 @@
 ---
 authority_id: iso
 id: 11940-1998
-language: tha
+language: iso-639-2:tha
 source_script: Thai
 destination_script: Latn
 name: ISO 11940:1998 Information and documentation -- Transliteration of Thai
 url: https://www.iso.org/standard/20574.html
 creation_date: 1998
-adoption_date: 
+adoption_date:
 description:
 
 notes:

data/maps/kp-kor-Hang-Latn-2002.yaml

@@ -1,10 +1,14 @@
 ---
 authority_id: kp
 id: 2002
-language: kor
+language: iso-639-2:kor
 source_script: Hang
 destination_script: Latn
 name: Korean Democratic People's Republic of Korea Korean System (2002)
+alias:
+  ogc11122:
+    code: kor_Hang2Latn_GKN_2002
+    description: Korean Democratic People's Republic of Korea Korean 2002 System
 url: https://unstats.un.org/unsd/geoinfo/UNGEGN/docs/8th-uncsgn-docs/inf/8th_UNCSGN_econf.94_INF.72.pdf
 creation_date:
 adoption_date:
@@ -12,7 +16,7 @@ description:
 
 notes:
 
-  - Here is a list of features that are listed in the guideline but 
+  - Here is a list of features that are listed in the guideline but
     not unimplemented in this map.
 
   - Note 3.2
@@ -30,15 +34,15 @@ notes:
     Except for the n-g combination, this has not been implemented.
 
   - Note 4.4
-    Geographical names "may be transliterated or translated". In this map, 
-    all names will be transliterated, not translated. Numerals will not be 
+    Geographical names "may be transliterated or translated". In this map,
+    all names will be transliterated, not translated. Numerals will not be
     transliterated.
 
   - Note 4.5
     Spacing rule for personal names has not been implemented.
 
   - Note 4.7
-    Optional omission of diacritics and optional simplification of 
+    Optional omission of diacritics and optional simplification of
     KK, TT, PP, SS, JJ to single letter have not been implemented.
 
 tests:
@@ -135,11 +139,15 @@ tests:
   - source: "평양"
     expected: "Pyongyang"
 
+  # Fix a problem with a trailing comma
+  - source: "구현, 글꼴, 문자 배열, 다국어 컴퓨팅."
+    expected: "Kuhyŏn, Külkkol, Munja Paeyŏl, Tagugŏ Khŏmphyuthing."
+
 map:
   character_separator: ""
   word_separator: " "
   title_case: True
-  inherit: "nil-kor-Hang-Hang-jamo"
+  inherit: "var-kor-Hang-Hang-jamo"
 
   rules:
 
@@ -860,35 +868,35 @@ map:
       result: "wae" # HANGUL JUNGSEONG WAE
     - pattern: "ᅰ"
       result: "we" # HANGUL JUNGSEONG WE
-    - pattern: "ᆨ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆨ(?!\\p{Hangul})"
       result: "k" # HANGUL JONGSEONG KIYEOK
-    - pattern: "ᆫ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆫ(?!\\p{Hangul})"
       result: "n" # HANGUL JONGSEONG NIEUN
-    - pattern: "ᆮ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆮ(?!\\p{Hangul})"
       result: "t" # HANGUL JONGSEONG TIEUT
-    - pattern: "ᆯ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆯ(?!\\p{Hangul})"
       result: "l" # HANGUL JONGSEONG RIEUL
-    - pattern: "ᆷ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆷ(?!\\p{Hangul})"
       result: "m" # HANGUL JONGSEONG MIEUM
-    - pattern: "ᆸ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆸ(?!\\p{Hangul})"
       result: "p" # HANGUL JONGSEONG PIEUP
-    - pattern: "ᆺ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆺ(?!\\p{Hangul})"
       result: "t" # HANGUL JONGSEONG SIOS
-    - pattern: "ᆼ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆼ(?!\\p{Hangul})"
       result: "ng" # HANGUL JONGSEONG IEUNG
-    - pattern: "ᆽ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆽ(?!\\p{Hangul})"
       result: "t" # HANGUL JONGSEONG CIEUC
-    - pattern: "ᆾ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆾ(?!\\p{Hangul})"
       result: "t" # HANGUL JONGSEONG CHIEUCH
-    - pattern: "ᆿ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆿ(?!\\p{Hangul})"
       result: "k" # HANGUL JONGSEONG KHIEUKH
-    - pattern: "ᇀ(?=[ A-Za-z0-9-])"
+    - pattern: "ᇀ(?!\\p{Hangul})"
       result: "t" # HANGUL JONGSEONG THIEUTH
-    - pattern: "ᇁ(?=[ A-Za-z0-9-])"
+    - pattern: "ᇁ(?!\\p{Hangul})"
       result: "p" # HANGUL JONGSEONG PHIEUPH
-    - pattern: "ᆰ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆰ(?!\\p{Hangul})"
       result: "k" # HANGUL JONGSEONG RIEUL-KIYEOK
-    - pattern: "ᆲ(?=[ A-Za-z0-9-])"
+    - pattern: "ᆲ(?!\\p{Hangul})"
       result: "p" # HANGUL JONGSEONG RIEUL-PIEUP
 
     # Remove space added

data/maps/lshk-yue-Hani-Latn-jyutping-1993.yaml

@@ -1,10 +1,10 @@
 ---
 authority_id: lshk
 id: jyutping
-language: yue
+language: iso-639-3:yue
 source_script: Hani
 destination_script: Latn
-name: LSHK Jyutping 
+name: LSHK Jyutping
 url: https://lshk.org/jyutping
 creation_date: 1993-12
 description:
@@ -39,7 +39,7 @@ map:
   #  - pattern: \u044c(?=[ЕеЁёИиОоЭэ]) # ь (before Е, Ё, И, O, Э)
   #    result: y
 
-  # Although the map is called "characters", phrases are added before individual characters 
+  # Although the map is called "characters", phrases are added before individual characters
   # in order to improve transliteration accuracy
 
   characters:

data/maps/masm-mon-Cyrl-Latn-5217-2012.yaml

@@ -0,0 +1,163 @@
+---
+authority_id: masm
+id: 2012
+language: iso-639-2:mon
+source_script: Cyrl
+destination_script: Latn
+name: Transliteration of Mongolian Cyrillic characters into Latin
+url: http://estandard.gov.mn/standard/reader/4635
+creation_date:
+confirmation_date: 2012-02
+description: |
+  The standard MNS 5217:2012 is a standard of the Mongolian Agency for Standardization and Metrology
+  for transliteration from Cyrillic Mongolian in Latin alphabet. It repeats and supplements the previous
+  standard dating from 2003 (MNS 5217:2003).
+
+  This standard differs from the ISO 9 system for Cyrillic adopted by Russia, Armenia, Azerbaijan,
+  Belarus, Kazakhstan, Kyrgyzstan, Uzbekistan, Tajikistan and Turkmenistan.
+
+tests:
+  - source: Аварга, халбага, аав
+    expected: Avarga, khalbaga, aav
+  - source: Бага, самбар
+    expected: Baga, sambar
+  - source: Аварга, вагон, сав
+    expected: Avarga, vagon, sav
+  - source: Газар, гэрээ, хэрэг
+    expected: Gazar, geree, khereg
+  - source: Дадлага, ахмад
+    expected: Dadlaga, akhmad
+  - source: Еэвэн, ерөөл
+    expected: Yeeven, yerööl
+  - source: Ёроол, оёдол
+    expected: Yorool, oyodol
+  - source: Жуулчин, ажил, Жон
+    expected: Juulchin, ajil, Jon
+  - source: Зам, азарга, бааз
+    expected: Zam, azarga, baaz
+  - source: Кино, километр, академи
+    expected: Kino, kilomyetr, akadyemi # Kino, kilometr, akademi - Provided sample is wrong
+  - source: Ишиг, бичиг, хань
+    expected: Ishig, bichig, khani
+  - source: Ийм, ээжийн
+    expected: Iim, eejiin
+  - source: Лам, алаг, мал
+    expected: Lam, alag, mal
+  - source: Мал, хамар, нам
+    expected: Mal, khamar, nam
+  - source: Нар, хана, үнэн
+    expected: Nar, khana, ünen
+  - source: Орон, боловсрол, тооно
+    expected: Oron, bolovsrol, toono
+  - source: Өдөр, өнөөдөр, өөрөөсөө
+    expected: Ödör, önöödör, ööröösöö
+  - source: Пуужин, апарат
+    expected: Puujin, aparat
+  - source: Рашаан, радио, сар
+    expected: Rashaan, radio, sar
+  - source: Сар, асар, эцэс
+    expected: Sar, asar, etses
+  - source: Тамга, татлага
+    expected: Tamga, tatlaga
+  - source: Уран, нуруу
+    expected: Uran, nuruu
+  - source: Үнэн, түргэн, тэргүүн
+    expected: Ünen, türgen, tergüün
+  - source: Фото, фонд
+    expected: Foto, fond
+  - source: Хавар, нөхөр, эх
+    expected: Khavar, nökhör, ekh
+  - source: Цацаг, цэцэг
+    expected: Tsatsag, tsetseg
+  - source: Чимэг, чадал, ач
+    expected: Chimeg, chadal, ach
+  - source: Шашин, ааш
+    expected: Shashin, aash
+  - source: Щедрин, щорс
+    expected: Shyedrin, shors # Shedrin, Shors - Provided sample is wrong
+  - source: Оръё, суръя, гаръя
+    expected: Oriyo, suriya, gariya
+  - source: Хааны, ахын
+    expected: Khaany, akhyn
+  - source: Харь, барь
+    expected: Khari, bari
+  - source: Эзэн, энэ, эмээл
+    expected: Ezen, ene, emeel
+  - source: Юм, юүдэн
+    expected: Yum, yuüden
+  - source: Ямар, ядуу, ая
+    expected: Yamar, yaduu, aya
+
+map:
+
+  characters:
+    "А": "A"      # \u0410
+    "Б": "B"      # \u0411
+    "В": "V"      # \u0412
+    "Г": "G"      # \u0413
+    "Д": "D"      # \u0414
+    "Е": "Ye"     # \u0415
+    "Ё": "Yo"     # \u0401
+    "Ж": "J"      # \u0416
+    "З": "Z"      # \u0417
+    "И": "I"      # \u0418
+    "Й": "I"      # \u0419
+    "К": "K"      # \u041a
+    "Л": "L"      # \u041b
+    "М": "M"      # \u041c
+    "Н": "N"      # \u041d
+    "О": "O"      # \u041e
+    "Ө": "Ö"      # \u04e8
+    "П": "P"      # \u041f
+    "Р": "R"      # \u0420
+    "С": "S"      # \u0421
+    "Т": "T"      # \u0422
+    "У": "U"      # \u0423
+    "Ү": "Ü"      # \u04ae
+    "Ф": "F"      # \u0424
+    "Х": "Kh"     # \u0425
+    "Ц": "Ts"     # \u0426
+    "Ч": "Ch"     # \u0427
+    "Ш": "Sh"     # \u0428
+    "Щ": "Sh"     # \u0429
+    "Ъ": "I"      # \u042a
+    "Ы": "Y"      # \u042b
+    "Ь": "I"      # \u042c
+    "Э": "E"      # \u042d
+    "Ю": "Yu"     # \u042e
+    "Я": "Ya"     # \u042f
+    "а": "a"      # \u0430
+    "б": "b"      # \u0431
+    "в": "v"      # \u0432
+    "г": "g"      # \u0433
+    "д": "d"      # \u0434
+    "е": "ye"     # \u0435
+    "ё": "yo"     # \u0451
+    "ж": "j"      # \u0436
+    "з": "z"      # \u0437
+    "и": "i"      # \u0438
+    "й": "i"      # \u0439
+    "к": "k"      # \u043a
+    "л": "l"      # \u043b
+    "м": "m"      # \u043c
+    "н": "n"      # \u043d
+    "о": "o"      # \u043e
+    "ө": "ö"      # \u04e9
+    "п": "p"      # \u043f
+    "р": "r"      # \u0440
+    "с": "s"      # \u0441
+    "т": "t"      # \u0442
+    "у": "u"      # \u0443
+    "ү": "ü"      # \u04af
+    "ф": "f"      # \u0444
+    "х": "kh"     # \u0445
+    "ц": "ts"     # \u0446
+    "ч": "ch"     # \u0447
+    "ш": "sh"     # \u0448
+    "щ": "sh"     # \u0449
+    "ъ": "i"      # \u044a
+    "ы": "y"      # \u044b
+    "ь": "i"      # \u044c
+    "э": "e"      # \u044d
+    "ю": "yu"     # \u044e
+    "я": "ya"     # \u044f

data/maps/masm-mon-Latn-Cyrl-5217-2012.yaml

@@ -0,0 +1,200 @@
+---
+authority_id: masm
+id: 2012
+language: iso-639-2:mon
+source_script: Latn
+destination_script: Cyrl
+name: Transliteration of Latin characters into Mongolian Cyrillic
+url: https://fr.wikipedia.org/wiki/MNS_5217:2012
+creation_date:
+confirmation_date:
+description: |
+  Latin characters into Mongolian Cyrillic is the opposite transliteration of based on MNS 5217:2012.
+  The standard MNS 5217:2012 is a standard of the Mongolian Agency for Standardization and Metrology
+  for transliteration from Cyrillic Mongolian in Latin alphabet.
+
+  The grammer of Khalkha Mongolian is required in order to make accurate translation of Latin characters
+  into Mongolian Cyrillic because of no distinguish between /и, й, ъ, ь/ for "i" and /ш, щ/ for "sh" in
+  the standard.
+
+  In Mongolia, the Khalkha dialect, currently written in both Cyrillic and traditional Mongolian script
+  (and at times in Latin for social networking), is predominant, while in Inner Mongolia, the language
+  is dialectally more diverse and is written in the traditional Mongolian script.
+
+notes: |
+  - Khalkha also has four diphthongs: /ui, ʊi, ɔi, ai/. Vowels can also combine to form a further
+    three diphthongs, and so seven in total: ia (иа), ʊa (уа) ei (эй). For example: ай in далай (eyes),
+    иа in амиараа (individually), ой in нохой (dog), уа in хуаран (barracks), уй in уйлах (to cry),
+    үй in үйлдвэр (factory), эй in хэрэгтэй (necessary).
+  - Custom mapping sort orders made as following by: /й, ъ, ь, и/ for "i" and /щ, ш/ for "sh".
+
+tests:
+  - source: Avarga, khalbaga, aav
+    expected: Аварга, халбага, аав
+  - source: Baga, sambar
+    expected: Бага, самбар
+  - source: Avarga, vagon, sav
+    expected: Аварга, вагон, сав
+  - source: Gazar, geree, khereg
+    expected: Газар, гэрээ, хэрэг
+  - source: Dadlaga, akhmad
+    expected: Дадлага, ахмад
+  - source: Yeeven, yerööl
+    expected: Еэвэн, ерөөл
+  - source: Yorool, oyodol
+    expected: Ёроол, оёдол
+  - source: Juulchin, ajil, Jon
+    expected: Жуулчин, ажил, Жон
+  - source: Zam, azarga, baaz
+    expected: Зам, азарга, бааз
+  - source: Kino, kilomyetr, akadyemi
+    expected: Кино, километр, академи
+  - source: Ishig, bichig, khani
+    expected: Ишиг, бичиг, хани # Correct expectation: Ишиг, бичиг, хань - Hard transliteration required for ь
+  - source: Iim, eejiin
+    expected: Ийм, ээжийн
+  - source: Lam, alag, mal
+    expected: Лам, алаг, мал
+  - source: Mal, khamar, nam
+    expected: Мал, хамар, нам
+  - source: Nar, khana, ünen
+    expected: Нар, хана, үнэн
+  - source: Oron, bolovsrol, toono
+    expected: Орон, боловсрол, тооно
+  - source: Ödör, önöödör, ööröösöö
+    expected: Өдөр, өнөөдөр, өөрөөсөө
+  - source: Puujin, aparat
+    expected: Пуужин, апарат
+  - source: Rashaan, radio, sar
+    expected: Рашаан, радио, сар
+  - source: Sar, asar, etses
+    expected: Сар, асар, эцэс
+  - source: Tamga, tatlaga
+    expected: Тамга, татлага
+  - source: Uran, nuruu
+    expected: Уран, нуруу
+  - source: Ünen, türgen, tergüün
+    expected: Үнэн, түргэн, тэргүүн
+  - source: Foto, fond
+    expected: Фото, фонд
+  - source: Khavar, nökhör, ekh
+    expected: Хавар, нөхөр, эх
+  - source: Tsatsag, tsetseg
+    expected: Цацаг, цэцэг
+  - source: Chimeg, chadal, ach
+    expected: Чимэг, чадал, ач
+  - source: Shashin, aash
+    expected: Шашин, ааш
+  - source: Shyedrin, shors
+    expected: Шедрин, шорс # Correct expectation: Щедрин, щорс - Hard transliteration required for щ usage
+  - source: Oriyo, suriya, gariya
+    expected: Ориё, сурия, гария # Correct expectation: Оръё, суръя, гаръя - Hard transliteration required for ъ
+  - source: Khaany, akhyn
+    expected: Хааны, ахын
+  - source: Khari, bari
+    expected: Хари, бари # Correct expectation: Харь, барь - Hard transliteration required for ь
+  - source: Ezen, ene, emeel
+    expected: Эзэн, энэ, эмээл
+  - source: Yum, yuüden
+    expected: Юм, юүдэн
+  - source: Yamar, yaduu, aya
+    expected: Ямар, ядуу, ая
+  - source: ii, ia, ua, ai, ei, oi, üi, Ii, Ai, Ei, Oi, Üi
+    expected: ий, иа, уа, ай, эй, ой, үй, Ий, Ай, Эй, Ой, Үй
+  - source: uu, üü, yuu, yuü
+    expected: уу, үү, юу, юү
+
+map:
+  rules:
+    - pattern: "ii"
+      result: "\u0438\u0439"
+    - pattern: "Ii"
+      result: "\u0418\u0439"
+    - pattern: "ai"
+      result: "\u0430\u0439"
+    - pattern: "Ai"
+      result: "\u0410\u0439"
+    - pattern: "ei"
+      result: "\u044d\u0439"
+    - pattern: "Ei"
+      result: "\u042d\u0439"
+    - pattern: "oi"
+      result: "\u043e\u0439"
+    - pattern: "Oi"
+      result: "\u041e\u0439"
+    - pattern: "üi"
+      result: "\u04af\u0439"
+    - pattern: "Üi"
+      result: "\u04ae\u0439"
+
+  characters:
+    "A": "А"     # \u0410
+    "B": "Б"     # \u0411
+    "V": "В"     # \u0412
+    "G": "Г"     # \u0413
+    "D": "Д"     # \u0414
+    "Ye": "Е"    # \u0415
+    "Yo": "Ё"    # \u0401
+    "J": "Ж"     # \u0416
+    "Z": "З"     # \u0417
+    "I": "Й"     # \u0419
+    "K": "К"     # \u041a
+    "L": "Л"     # \u041b
+    "M": "М"     # \u041c
+    "N": "Н"     # \u041d
+    "O": "О"     # \u041e
+    "Ö": "Ө"     # \u04e8
+    "P": "П"     # \u041f
+    "R": "Р"     # \u0420
+    "S": "С"     # \u0421
+    "T": "Т"     # \u0422
+    "U": "У"     # \u0423
+    "Ü": "Ү"     # \u04ae
+    "F": "Ф"     # \u0424
+    "Kh": "Х"    # \u0425
+    "Ts": "Ц"    # \u0426
+    "Ch": "Ч"    # \u0427
+    "Sh": "Щ"    # \u0429
+    "Sh": "Ш"    # \u0428
+    "I": "Ъ"     # \u042a
+    "Y": "Ы"     # \u042b
+    "I": "Ь"     # \u042c
+    "I": "И"     # \u0418
+    "E": "Э"     # \u042d
+    "Yu": "Ю"    # \u042e
+    "Ya": "Я"    # \u042f
+    "a": "а"     # \u0430
+    "b": "б"     # \u0431
+    "v": "в"     # \u0432
+    "g": "г"     # \u0433
+    "d": "д"     # \u0434
+    "ye": "е"    # \u0435
+    "yo": "ё"    # \u0451
+    "j": "ж"     # \u0436
+    "z": "з"     # \u0437
+    "i": "й"     # \u0439
+    "k": "к"     # \u043a
+    "l": "л"     # \u043b
+    "m": "м"     # \u043c
+    "n": "н"     # \u043d
+    "o": "о"     # \u043e
+    "ö": "ө"     # \u04e9
+    "p": "п"     # \u043f
+    "r": "р"     # \u0440
+    "s": "с"     # \u0441
+    "t": "т"     # \u0442
+    "u": "у"     # \u0443
+    "ü": "ү"     # \u04af
+    "f": "ф"     # \u0444
+    "kh": "х"    # \u0445
+    "ts": "ц"    # \u0446
+    "ch": "ч"    # \u0447
+    "sh": "щ"    # \u0449
+    "sh": "ш"    # \u0448
+    "i": "ъ"     # \u044a
+    "y": "ы"     # \u044b
+    "i": "ь"     # \u044c
+    "i": "и"     # \u0438
+    "e": "э"     # \u044d
+    "yu": "ю"    # \u044e
+    "ya": "я"    # \u044f