interscript 0.1.3 → 0.1.4

data/maps/bgnpcgn-bak-Cyrl-Latn-2007.yaml

@@ -0,0 +1,184 @@
+---
+authority_id: bgnpcgn
+id: 2007
+language: rus
+source_script: Cyrl
+destination_script: Latn
+name: BASHKIR TABLE OF CORRESPONDENCES CYRILLIC-ROMAN BGN/PCGN 2007 Agreement
+url: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/829203/TABLE_OF_CORRESPONDENCES__FOR_BASHKIR.pdf
+creation_date: 2007
+confirmation_date: 2019
+description: |
+  Bashkir is an official language within Respublika Bashkortostan, one of the
+  republics of the Russian Federation. It will normally be encountered in Cyrillic script, in
+  which case it should be romanized by means of the Cyrillic-Roman table of
+  correspondences given below
+
+notes:
+  - The letter w is used word initially and before a vowel. # 'and' or 'or' ? 
+  - The letter sequence ye is used word initially and before a vowel. # 'and' or 'or' ? 
+  - The letter w is used between or after vowels.
+  - The letter w is used after e, u, ö and ə.
+  - |
+    An inventory of letter-diacritic combinations, with their Unicode encoding,
+    in addition to the unmodified letters of the basic Roman script is:
+      Ğ (U+011E) ğ (U+011F)
+      Ź (U+0179) ź (U+017A)
+      Ë (U+00CB) ë (U+00EB)
+      Ñ (U+00D1) ñ (U+00F1)
+      Ö (U+00D6) ö (U+00F6)
+      Ś (U+015A) ś (U+015B)
+      Ü (U+00DC) ü (U+00FC)
+      Ç (U+00C7) ç (U+00E7)
+      Ş (U+015E) ş (U+015F)
+      Ə (U+018F) ə (U+0259)
+  - |
+    The Roman-script columns show only lowercase forms but, when applying the table,
+    uppercase and lowercase Roman letters as appropriate should be used.
+
+tests:
+  # adopted http://www.eki.ee/knab/lat/kblba.pdf
+  - source: Васйылға
+    expected: Wasyılğa
+  - source: Еҙем
+    expected: Yeźem
+  - source: Раевка
+    expected: Raevka
+  - source: Сәйетҡол
+    expected: Səyetqol
+  - source: Ауырғазы
+    expected: Awırğazı
+  - source: Бурһыҡтау
+    expected: Burhıqtaw
+  - source: Мәләүез
+    expected: Mələwez
+  - source: Ҡыҙылъяр
+    expected: Qıźılyar
+  # adopted https://en.wikipedia.org/wiki/Bashkir_language#Grammar
+  - source: кемдең 
+    expected: kemdeñ
+  - source: кем
+    expected: kem
+  - source: был
+    expected: bıl
+  - source: ошо
+    expected: oşo
+  - source: быларҙың
+    expected: bılarźıñ
+  - source: һеҙҙән 
+    expected: heźźən
+  - source: һин
+    expected: hin
+  - source: һеҙҙең
+    expected: heźźeñ
+
+map:
+  rules:
+    # note[1]
+    - pattern: \b\u0412(?=[АаЕеЁёИиОоӨөУуҮЫыЭэӘәЮюЯя])
+      result: "W"
+    - pattern: \b\u0432(?=[АаЕеЁёИиОоӨөУуҮЫыЭэӘәЮюЯя])
+      result: "w"
+    # note[2]
+    - pattern: \b\u0415
+      result: "Ye"
+    - pattern: \b\u0435
+      result: "ye"
+    - pattern: (?=\b)\u0415(?<=[АаЕеЁёИиОоӨөУуҮЫыЭэӘәЮюЯя])
+      result: "Ye"
+    - pattern: (?=\b)\u0435(?<=[АаЕеЁёИиОоӨөУуҮЫыЭэӘәЮюЯя])
+      result: "ye"
+
+    # note[3] # note[4]
+    - pattern: (?<=[АаЕеЁёИиОоӨөУуҮЫыЭэӘәЮюЯя])[\u0423\u04AE]
+      result: W
+    - pattern: (?<=[АаЕеЁёИиОоӨөУуҮЫыЭэӘәЮюЯя])[\u0443\u04AF]
+      result: w
+
+
+  characters:
+    '\u0410': 'A'      # А
+    '\u0411': 'B'      # Б note[1]
+    '\u0412': 'V'      # В
+    '\u0413': 'G'      # Г
+    '\u0492': "\u011E" # Ғ
+    '\u0414': 'D'      # Д
+    '\u0498': "\u0179" # Ҙ
+    '\u0415': 'E'      # Е note[2]
+    '\u0401': 'Ë'      # Ё
+    '\u0416': 'J'      # Ж
+    '\u0417': 'Z'      # З
+    '\u0418': 'I'      # И
+    '\u0419': 'Y'      # Й
+    '\u041A': 'K'      # К
+    '\u04A0': 'Q'      # Ҡ
+    '\u041B': 'L'      # Л
+    '\u041C': 'M'      # М
+    '\u041D': 'N'      # Н
+    '\u04A2': 'Ñ'      # Ң
+    '\u041E': 'O'      # О
+    '\u04E8': "ö"      # Ө
+    '\u041F': 'P'      # П
+    '\u0420': 'R'      # Р
+    '\u0421': 'S'      # С
+    '\u04AA': 'Ś'      # Ҫ
+    '\u0422': 'T'      # Т
+    '\u0423': 'U'      # У
+    '\u04AE': 'Ü'      # Ү note[3]
+    '\u0424': 'F'      # Ф
+    '\u0425': 'X'      # Х
+    '\u04BA': 'H'      # Һ
+    '\u0426': 'Ts'     # Ц
+    '\u0427': 'Ç'      # Ч
+    '\u0428': 'Ş'      # Ш
+    '\u0429': 'ŞÇ'     # Щ
+    '\u042A': ''       # Ъ
+    '\u042B': 'I'      # Ы
+    '\u042C': ''       # Ь
+    '\u042D': 'E'      # Э
+    '\u04D8': "\u018F" # Ә 
+    '\u042E': 'Yu'     # Ю
+    '\u042F': 'Ya'     # Я
+
+    '\u0430': 'a'      # а
+    '\u0431': 'b'      # б
+    '\u0432': 'v'      # в note[1]
+    '\u0433': 'g'      # г
+    '\u0493': "\u011F" # ғ
+    '\u0434': 'd'      # д
+    '\u0499': 'ź'      # ҙ
+    '\u0435': 'e'      # e note[2]
+    '\u0451': 'yo'     # ё
+    '\u0436': 'j'      # ж
+    '\u0437': 'z'      # з
+    '\u0438': 'i'      # и
+    '\u0439': 'y'      # й
+    '\u043A': 'k'      # к
+    '\u04A1': 'q'      # ҡ
+    '\u043B': 'l'      # л
+    '\u043C': 'm'      # м
+    '\u043D': 'n'      # н
+    '\u04A3': 'ñ'      # ң
+    '\u043E': 'o'      # о
+    '\u04E9': "\u00F6" # ө
+    '\u043F': 'p'      # п
+    '\u0440': 'r'      # р
+    '\u0441': 's'      # с
+    '\u04AB': 'ś'      # ҫ
+    '\u0442': 't'      # т
+    '\u0443': 'u'      # у
+    "\u04AF": 'ü'      # ү note[3]
+    '\u0444': 'f'      # ф
+    '\u0445': 'x'      # х
+    '\u04BB': 'h'      # һ
+    '\u0446': 'ts'     # ц
+    '\u0447': 'ç'      # ч
+    '\u0448': 'ş'      # ш
+    '\u0449': 'şç'     # щ
+    '\u044A': ''       # ъ
+    '\u044B': "\u0131" # ы
+    '\u044C': ''       # ь
+    '\u044D': 'e'      # э
+    '\u04D9': "\u0259" # ә
+    '\u044E': 'yu'     # ю
+    '\u044F': 'ya'     # я

data/maps/bgnpcgn-bul-Cyrl-Latn-1952.yaml

@@ -51,7 +51,7 @@ map:
     '\u0413': 'G'
     '\u0414': 'D'
     '\u0415': 'E'
-    '\u0416': 'ZH'
+    '\u0416': 'Zh'
     '\u0417': 'Z'
     '\u0418': 'I'
     '\u0419': 'Y'
@@ -66,15 +66,15 @@ map:
     '\u0422': 'T'
     '\u0423': 'U'
     '\u0424': 'F'
-    '\u0425': 'KH'
-    '\u0426': 'TS'
-    '\u0427': 'CH'
-    '\u0428': 'SH'
-    '\u0429': 'SHT'
+    '\u0425': 'Kh'
+    '\u0426': 'Ts'
+    '\u0427': 'Ch'
+    '\u0428': 'Sh'
+    '\u0429': 'St'
     '\u042a': "U\u0306"
     '\u042c': "\\'"
-    '\u042e': 'YU'
-    '\u042f': 'YA'
+    '\u042e': 'Yu'
+    '\u042f': 'Ya'
     '\u0430': 'a'
     '\u0431': 'b'
     '\u0432': 'v'
@@ -111,5 +111,5 @@ map:
     '\u046B': "u\u0306" # ѫ
 
     # note[3]
-    '\u0462': "YE" # Ѣ
+    '\u0462': "Ye" # Ѣ
     '\u0463': "ye" # ѣ

data/maps/mvd-bel-Cyrl-Latn-2008.yaml

@@ -0,0 +1,225 @@
+---
+authority_id: mvd
+id: 19678
+language: bel
+source_script: Cyrl
+destination_script: Latn
+name: About approval of the Instructions for transliteration of surnames and proper names of citizens of the Republic of Belarus when their personal data is included in the population register
+url: https://www.icao.int/publications/pages/publication.aspx?docnum=9303
+creation_date: 2008
+
+description: |
+  RESOLUTION OF THE MINISTRY OF INTERNAL AFFAIRS OF THE REPUBLIC OF BELARUS
+  October 9, 2008, No. 288
+  8/19678 (10.23.2008)
+  About 8/19678 approval of the Instructions for transliteration of surnames and proper names of citizens of
+  the Republic of Belarus when their personal data is included in the population register
+
+notes:
+  # Original notes
+  # - | 
+  #   Инструкция по транслитерации фамилий и собственных имен граждан Республики Беларусь при включении
+  #   их персональных данных в регистр населения устанавливает правила передачи с белорусской либо русской формы
+  #   написания на латиницу при включении фамилий и собственных имен граждан Республики Беларусь в регистр населения.
+  # - |
+  #   Передача фамилий и собственных имен граждан Республики Беларусь на латиницу осуществляется с их правильного
+  #   написания на белорусском либо русском языке, за свидетельствованного документами, удостоверяющими личность.
+  # - Передача фамилий и собственных имен граждан Республики Беларусь осуществляется путем транслитерации
+  #   литер (букв, знаков) белорусского либорусского написания соответствующими литерами латиницы.
+  # - Транслитерацией достигается общность и унифицированность системы латинизированного написания,
+  #   позволяющей пользоваться ею во всех латинопишущих государствах.
+  # - Смягчение согласной буквы, обозначенное мягким знаком, в белорусской латинице следует показывать
+  #   диакритическим знаком (́), который располагается над соответствующей буквой: дзь – dź, зь – ź, ль – ĺ,
+  #   нь – ń, сь – ś, ць – ć.
+  # - Правила транслитерации букв белорусского и русского алфавитов соответствующими буквами латиницы
+  #   приведены в таблице транслитерации букв белорусского и русского алфавитов буквами латиницы соглас
+  #   но приложению к настоящей Инструкции.
+  # - Сложные и составные фамилии и собственные имена, пишущиеся слитно, раздельно или через дефис,
+  #   сохраняют слитное, раздельное или дефисное написание и в латинице.
+  - | # 1
+    Instructions for transliterating the names and first names of citizens of the Republic of Belarus 
+    when including their personal data in the population register sets the rules for transferring 
+    from the Belarusian or Russian form of writing in Latin when including the names and first names
+    of citizens of the Republic of Belarus in the population register.
+  - | # 2
+    Transfer of surnames and proper names of citizens of the Republic of Belarus to the Latin alphabet
+    is carried out with their correct spelling in Belarusian or Russian, for evidence of identity documents.
+  - | # 3
+    The transfer of surnames and proper names of citizens of the Republic of Belarus is carried out by
+    transliteration of the letters (letters, signs) of the Belarusian or Russian spelling in the corresponding
+    Latin letters.
+  - | # 4
+    Transliteration achieves the generality and unification of the system of Latinized writing,
+    which allows it to be used in all Latin-writing countries.
+  - | # 5
+    The softening of the consonant, indicated by a soft sign, in the Belarusian Latin should be shown
+    with a diacritic mark (́), which is located above the corresponding letter:
+      дзь - dź,
+      зь - ź,
+      ль - ĺ,
+      нь - ń,
+      сь - ś,
+      ць - ć.
+  # 6
+  - The rules for transliterating letters of the Belarusian and Russian alphabets with the corresponding
+    letters of the Latin alphabet are given in the table of transliteration of letters of the Belarusian
+    and Russian alphabets with the Latin letters according to the appendix to this Instruction.
+  # 7
+  - Compound and compound surnames and proper names, spelled together, separately or through a hyphen,
+    keep a single, separate or hyphen spelling in Latin.
+
+tests:
+  - source: Ева
+    expected: Jeva
+  - source: Васiльева
+    expected: Vasiĺjeva
+  - source: Васiлёнак
+    expected: Vasilionak
+  - source: Ёрш
+    expected: Jorsh
+  - source: Вераб’ёў
+    expected: Vierabjow
+  - source: Салаўёва
+    expected: Salawjova
+  - source: Любоў
+    expected: Liubow
+  - source: В’юноў
+    expected: Vjunow
+  - source: Чарняк
+    expected: Charniak
+  - source: Лябецкая
+    expected: Liabietskaja # in reference doc it's Liabetskaja CAMOBAP waiting confirmation from officials
+  - source: Дар’я
+    expected: Darja
+
+map:
+  rules:
+    - pattern: (\u2019\u0415) # Е
+      result: Je
+    - pattern: (\u2019\u0435) # е
+      result: je
+    - pattern: (\u2019\u0401) # Ë
+      result: Jo
+    - pattern: (\u2019\u0451) # ё
+      result: jo
+    - pattern: (\u2019\u042E) # Ю
+      result: Ju
+    - pattern: (\u2019\u044E) # ю
+      result: ju
+    - pattern: (\u2019\u042F) # Я
+      result: Ja
+    - pattern: (\u2019\u044F) # я
+      result: ja
+
+    - pattern: (?<=[АаЕеЁёИиОоУуЭэЮюЯяЬьЎў])\u0415 # Е after vowels
+      result: Je
+    - pattern: (?<=[АаЕеЁёИиОоУуЭэЮюЯяЬьЎў])\u0435 # е after vowels
+      result: je
+    - pattern: (?<=[АаЕеЁёИиОоУуЭэЮюЯяЬьЎў])\u0401 # Ё after vowels
+      result: Jo
+    - pattern: (?<=[АаЕеЁёИиОоУуЭэЮюЯяЬьЎў])\u0451 # ё after vowels
+      result: jo
+    - pattern: (?<=[АаЕеЁёИиОоУуЭэЮюЯяЬьЎў])\u042E # Ю after vowels
+      result: Ju
+    - pattern: (?<=[АаЕеЁёИиОоУуЭэЮюЯяЬьЎў])\u044E # ю after vowels
+      result: ju
+    - pattern: (?<=[АаЕеЁёИиОоУуЭэЮюЯяЬьЎў])\u042F # Я after vowels
+      result: Ja
+    - pattern: (?<=[АаЕеЁёИиОоУуЭэЮюЯяЬьЎў])\u044F # я after vowels
+      result: ja
+
+    # note[5]
+    - pattern: (?<=[ЗзЛлНнСсЦц])\u044C # ь after consonants
+      result: "\\1\u0301"
+    - pattern: (?<=[ЗзЛлНнСсЦц])\u02B9 # Ь after consonants
+      result: "\\1\u0301"
+
+    # vowels initially
+    - pattern: \b\u0415 # Е
+      result: Je
+    - pattern: \b\u0435 # е
+      result: je
+    - pattern: \b\u0401 # Ё
+      result: Jo
+    - pattern: \b\u0451 # ё
+      result: jo
+    - pattern: \b\u042E # Ю
+      result: Ju
+    - pattern: \b\u044E # ю
+      result: ju
+    - pattern: \b\u042F # Я
+      result: Ja
+    - pattern: \b\u044F # я
+      result: ja
+
+  postrules:
+    - pattern: \u2019
+      result: j
+
+  characters:
+
+    '\u0410' : 'A'  # А
+    '\u0411' : 'B'  # Б
+    '\u0412' : 'V'  # B
+    '\u0413' : 'G'  # Г
+    '\u0414' : 'D'  # Д
+    '\u0415' : 'Ie' # Е or JE TODO add rule
+    '\u0401' : 'Io' # Ё or JO TODO add rule
+    '\u0416' : 'Zh' # Ж 
+    '\u0417' : 'Z'  # З
+    '\u0406' : 'I'  # І
+    '\u0419' : "J"  # Й
+    '\u041A' : 'K'  # К
+    '\u041B' : 'L'  # Л
+    '\u041C' : 'M'  # М
+    '\u041D' : 'N'  # Н
+    '\u041E' : 'O'  # О
+    '\u041F' : 'P'  # П
+    '\u0420' : 'R'  # Р
+    '\u0421' : 'S'  # С
+    '\u0422' : 'T'  # Т
+    '\u0423' : 'U'  # У
+    '\U040E' : 'W'  # Ў
+    '\u0424' : 'F'  # Ф
+    '\u0425' : 'Kh' # Х
+    '\u0426' : 'Ts' # Ц
+    '\u0427' : 'Ch' # Ч
+    '\u0428' : 'Sh' # Ш
+    '\u0429' : 'Shch' # Щ
+    '\u042B' : 'Y'  # Ы
+    '\u042D' : 'E'  # Э
+    '\u042E' : "Iu" # Ю
+    '\u042F' : "Ia" # Я
+
+    '\u0430' : 'a'  # а
+    '\u0431' : 'b'  # б
+    '\u0432' : 'v'  # в
+    '\u0433' : 'g'  # г
+    '\u0434' : 'd'  # д
+    '\u0435' : 'ie' # е
+    '\u0451' : 'io' # ё
+    '\u0436' : 'zh' # ж
+    '\u0437' : 'z'  # з
+    '\u0456' : 'i'  # і
+    '\u0439' : 'j'  # й
+    '\u043A' : 'k'  # к
+    '\u043B' : 'l'  # л
+    '\u043C' : 'm'  # м
+    '\u043D' : 'n'  # н
+    '\u043E' : 'o'  # о
+    '\u043F' : 'p'  # п
+    '\u0440' : 'r'  # р
+    '\u0441' : 's'  # с
+    '\u0442' : 't'  # т
+    '\u0443' : 'u'  # у
+    '\u045E' : 'w'  # ў
+    '\u0444' : 'f'  # ф
+    '\u0445' : 'kh' # х
+    '\u0446' : 'ts' # Ц
+    '\u0447' : 'ch' # ч
+    '\u0448' : 'sh' # ш
+    '\u0449' : 'shch' # щ
+    '\u044B' : 'y'  # ы
+    '\u044D' : 'e'  # э
+    '\u044E' : "iu" # ю
+    '\u044F' : "ia" # я

data/maps/mvd-bel-Cyrl-Latn-2010.yaml

@@ -0,0 +1,63 @@
+---
+authority_id: mvd
+id: 22721
+language: bel
+source_script: Cyrl
+destination_script: Latn
+name: |
+  About approval of the Instructions on the organization of work of units of citizenship 
+  and migration of internal affairs bodies on the issuance, registration, exchange, 
+  invalidation, seizure, storage and destruction of a passport of a citizen of the Republic of Belarus
+url: https://pravo.by/document/?guid=3871&p0=W21022721
+creation_date: 2010
+
+description: |
+  RESOLUTION OF THE MINISTRY OF INTERNAL AFFAIRS OF THE REPUBLIC OF BELARUS
+  June 28, 2010 No. 200
+  On approval of the Instructions on the organization of work of units of citizenship 
+  and migration of internal affairs bodies on the issuance, registration, exchange, 
+  invalidation, seizure, storage and destruction of a passport of a citizen of the Republic of Belarus
+
+notes:
+  - |
+    It is not allowed to use in the spelling of the surname, own name superscripts, punctuation, except
+    for the use of the apostrophe in Belarusian and Latin spelling.
+  - The Belarusian letter "Г" is written as the Latin "H"
+
+tests:
+  - source: Бабрыковіч Аляксандр
+    expected: Babrykovich Aliaksandr
+  - source: Міховіч Марыя
+    expected: Mikhovich Maryia
+  - source: Максім
+    expected: Maksim
+  - source: Іван
+    expected: Ivan
+  - source: СВЯТЛАНА
+    expected: SVIATLANA
+  - source: Ігар
+    expected: Ihar
+  - source: МІХАІЛ
+    expected: MIKHAIL
+
+map:
+  inherit: "mvd-bel-Cyrl-Latn-2008"
+
+  rules:
+    # note[5]
+    - pattern: (?<=[ЗзЛлНнСсЦц])\u044C # ь after consonants
+      result: "\\1"
+    - pattern: (?<=[ЗзЛлНнСсЦц])\u02B9 # Ь after consonants
+      result: "\\1"
+    # Й at end
+    - pattern: (?<=[ЕеЁёЫыЮюЯя])\u0419$ # Я after vowels
+      result: ""
+    - pattern: (?<=[ЕеЁёЫыЮюЯя])\u0439$ # я after vowels
+      result: ""
+
+  characters:
+    '\u0413' : 'H'  # Г
+    '\u0433' : 'h'  # г
+
+    '\u042C' : ''  # Ь
+    '\u044C' : ''  # ь