interscript 0.1.2 → 0.1.7

data/maps/bgn-jpn-Hrkt-Latn-1962.yaml

@@ -0,0 +1,292 @@
+---
+authority_id: bgn
+id: 1962
+language: iso-639-2:jpn
+source_script: Hrkt
+destination_script: Latn
+name: BGN (Modified Hepburn) System
+url:
+creation_date: 1930
+adoption_date: 1962
+description: |
+  The BGN (Modified Hepburn) System for the transliteration of Japanese
+  has been in use by the Board on Geographic Names since about 1930 and
+  has been extensively employed in the systematic standardsization of
+  thousands of geographic names of Japan in romanized form.
+
+notes: |
+
+  1. The "tsu" forms (ツ/つ) are also used to indicate a double consonant and
+  are generally (but not alwyas) written in smaller script or type
+  slightly to the right of or below the regular line. These characters
+  are trasnliterated as k before k; s before s or sh; t before t, ts, or
+  ch; and p before p. Occasionally, when a "ku" (ク/く) or "ki" (キ/き) form
+  precedes k, the u in ku or the i in ki is dropped.
+
+  2. The transliterations in parentheses are used in specific cases when
+  the kana symbol is known to be so pronounced.
+
+  3. The transliteration m is used before b, p, and m.
+
+  4. This letter has been added for the use in transliterating foreign
+  words.
+
+  5. The asterisk (*) indicates standard combined forms. Those combined
+  forms not so marked are rarely used.
+
+  ----
+
+  Implementation Notes:
+
+  a. Despite the mentioning of the term "Modified Hepburn" in the
+  specification, the handling of ん/ン in this standard is different from
+  Modified Hepburn. It follows the Traditional Hepburn in that the
+  letter m is used before b, m, p.
+
+  b. This document includes obsolete (pre-reform) combinations.
+  Pre-reform combinations will clash with modern Japanese transliteration.
+
+  c. There is no discussion on how cross-morpheme vowel sounds should be
+  handled.
+
+  d. There is no mentioning of separation mark between n and another vowel.
+
+  e. Everything not explicitly stated in the specification will be
+  assumed to be inherited from var-jpn-Hrkt-Latn-hepburn-1954.
+
+  f. Obsolete combinations can be handled by post rules, and are
+  included for the sake of completeness only. They have been commented
+  out, since they are rarely used and follow different rules than modern
+  Japanese.
+
+tests:
+  # Note: these test cases follow the pre-reform standard.
+  # They are commented out for now.
+  #
+  # - source: "けふ"
+  #   expected: "kyō"
+  # - source: "ぎうにう"
+  #   expected: "gyūnyū"
+  # - source: "きふ"   # きふ should always be kifu in Modern Japanese
+  #   expected: "kyū"
+  # - source: "ちう"
+  #   expected: "chū"
+  # - source: "けう"
+  #   expected: "kyō"
+
+  # Modern Japanese test cases
+  - source: "しんばし"
+    expected: "shimbashi"
+  - source: "とうきょう"
+    expected: "tōkyō"
+  - source: "しんじゅく"
+    expected: "shinjuku"
+  - source: かんおう
+    expected: kan’ō
+  - source: かのう
+    expected: kanō
+  - source: きんゆう
+    expected: kin’yū
+  - source: とうきょう
+    expected: tōkyō
+  - source: かごっま
+    expected: kagomma
+  - source: ぽっぽっや
+    expected: poppoyya
+  - source: てっら
+    expected: terra
+  - source: にゃっほー
+    expected: nyahhō
+
+
+map:
+
+  inherit: var-jpn-Hrkt-Latn-hepburn-1954
+
+  rules:
+    # Convert ん into m before b, m, p
+    - pattern: "[んン](?=[ばびぶべぼまみむめもぱぴぷぺぽバビブベボマミムメモパピプペポ])"
+      result: "m"
+  postrules:
+    # Handle obsolete forms
+    # Note that these forms are present in the rules, but will break
+    # if used with Modern Japanese. They are commented out for now.
+    #
+    # - pattern: "ef?[uo]|iyau"
+    #   result: "yō"
+    # - pattern: "if?u"
+    #   result: "yū"
+    # - pattern: "[ao]f?[uo]"
+    #   result: "ō"
+    # - pattern: "iy"
+    #   result: "y"
+    # - pattern: "ty"
+    #   result: "ch"
+    # - pattern: "dy"
+    #   result: "j"
+    # - pattern: "[jz]y"
+    #   result: "j"
+    # - pattern: "(?<=[sc])hy"
+    #   result: "h"
+    # - pattern: "sy"
+    #   result: "sh"
+
+  characters:
+    # ke
+    # These are listed as alternative pronunciation, but in fact this usage of ヶ
+    # as the archaic possessive marker is not found in Kana only texts.
+    # Also it is always typed using the smaller form. (ヶ U+30F6)
+    "け": ["ke", "ga", "ka", "ko"]
+    "ケ": ["ke", ga", "ka", "ko"]
+    "ヶ": ["ga", "ka", "ko"]
+
+
+    # The Ha-column
+    # は is still pronounced as wa when used as a particle,
+    # the alternative pronunciations for the other four kana's are obsolete.
+    "は": ["ha", "wa"]
+    "ひ": ["hi", "i"]
+    "ふ": ["fu", "u", "o"]
+    "へ": ["he", "e"]
+    "ほ": ["ho", "o"]
+    "ハ": ["ha", "wa"]
+    "ヒ": ["hi", "i"]
+    "フ": ["fu", "u", "o"]
+    "ヘ": ["he", "e"]
+    "ホ": ["ho", "o"]
+
+
+    # The Wa-column
+    # These two kanas below are only used in pre-reform texts.
+    "ゐ" : "i"
+    "ゑ" : "e"
+    "ヰ" : "i"
+    "ヱ" : "e"
+
+    # Combined forms
+    # These are obsolete forms. See Note 5.
+    # They can be handled by post-rules if ever needed.
+    # "あう": "ō"
+    # "あふ": "ō"
+    # "いふ": "yū"
+    # "えう": "yō"
+    # "えふ": "yō"
+    # "おふ": "ō"
+    # "かう": "kō"
+    # "かふ": "kō"
+    # "がう": "gō"
+    # "がふ": "gō"
+    # "きう": "kyū"
+    # "きふ": "kyū"
+    # "きやう": "kyō"
+    # "ぎう": "gyū"
+    # "ぎふ": "gyū"
+    # "ぎやう": "gyō"
+    "くわ": "ka"
+    "くわう": "kō"
+    "ぐわ": "ga"
+    "ぐわう": "gō"
+    "クワ": "ka"
+    "クワウ": "kō"
+    "グワ": "ga"
+    "グワウ": "gō"
+    # "けう": "kyō"
+    # "けふ": "kyō"
+    # "げう": "gyō"
+    # "げふ": "gyō"
+    # "こふ": "kō"
+    # "ごふ": "gō"
+    # "さう": "sō"
+    # "さふ": "sō"
+    # "ざう": "zō"
+    # "ざふ": "zō"
+    # "しう": "shū"
+    # "しふ": "shū"
+    # "しやう": "shō"
+    # "じう": "jū"
+    # "じふ": "jū"
+    # "じやう": "jō"
+    # "せう": "shō"
+    # "せふ": "shō"
+    # "ぜう": "jō"
+    # "ぜふ": "jō"
+    # "そふ": "sō"
+    # "ぞふ": "zō"
+    # "たう": "tō"
+    # "たふ": "tō"
+    # "だう": "dō"
+    # "だふ": "dō"
+    # "ちう": "chū"
+    # "ちふ": "chū"
+    # "ちやう": "chō"
+    # "ぢう": "jū"
+    # "ぢふ": "jū"
+    # "ぢや": "ja"
+    # "ぢやう": "jō"
+    # "ぢゆ": "ju"
+    # "ぢよ": "jo"
+    # "ぢよう": "jō"
+    # "てう": "chō"
+    # "てふ": "chō"
+    # "でう": "jō"
+    # "でふ": "jō"
+    # "とふ": "tō"
+    # "どふ": "dō"
+    # "なう": "nō"
+    # "なふ": "nō"
+    # "にう": "nyū"
+    # "にふ": "nyū"
+    # "にやう": "nyō"
+    # "ねう": "nyō"
+    # "ねふ": "nyō"
+    # "のふ": "nō"
+    # "はう": ["hō","ō"]
+    # "はふ": "hō"
+    # "ばふ": "bō"
+    # "ばう": "bō"
+    # "ぱう": "pō"
+    # "ぱふ": "pō"
+    # "ひう": "hyū"
+    # "ひふ": "hyū"
+    # "ひやう": "hyō"
+    # "びう": "byū"
+    # "びふ": "byū"
+    # "びやう": "byō"
+    # "ぴう": "pyū"
+    # "ぴふ": "pyū"
+    # "ぴやう": "pyō"
+    # "へう": "hyō"
+    # "へふ": "hyō"
+    # "べう": "byō"
+    # "べふ": "byō"
+    # "ぺう": "pyō"
+    # "ぺふ": "pyō"
+    # "ほふ": "hō"
+    # "ぼふ": "bō"
+    # "ぽふ": "pō"
+    # "まう": "mō"
+    # "まふ": "mō"
+    # "まを": "mō"
+    # "みやう": "myō"
+    # "みう": "myū"
+    # "みふ": "myū"
+    # "めう": "myō"
+    # "めふ": "myō"
+    # "めを": "myō"
+    # "もふ": "mō"
+    # "やう": "yō"
+    # "やふ": "yō"
+    # "よふ": "yō"
+    # "らう": "rō"
+    # "らふ": "rō"
+    # "りう": "ryū"
+    # "りふ": "ryū"
+    # "りやう": "ryō"
+    # "れう": "ryō"
+    # "れふ": "ryō"
+    # "ろふ": "rō"
+    # "わう": "wō"
+    # "わふ": "wō"
+    # "ゑふ": "yō"
+    # "をう": "ō"
+    # "をふ": "ō"

data/maps/bgn-kor-Hang-Latn-1943.yaml

@@ -0,0 +1,35 @@
+---
+authority_id: bgn
+id: 1943
+language: iso-639-2:kor
+source_script: Hang
+destination_script: Latn
+name: BGN Agreement -- Korean McCune-Reischauer System (1943)
+alias:
+  ogc11122:
+    code: kor_Hang2Latn_MR_1939
+    description: Korean McCune-Reischauer System, 1943 US Board on Geographic Names (BGN) Agreement
+url:
+creation_date: 1943
+adoption_date:
+description:
+
+notes:
+
+tests:
+  - source: 서울
+    expected: "Sŏul"
+  - source: 평양
+    expected: "P’yŏngyang"
+  - source: 부산
+    expected: "Pusan"
+  - source: 대구
+    expected: "Taegu"
+  - source: 경상남도
+    expected: "Kyŏngsangnamdo"
+
+map:
+  character_separator: ""
+  word_separator: " "
+  title_case: True
+  inherit: var-kor-Hang-Latn-mr-1939

data/maps/bgn-kor-Kore-Latn-1943.yaml

@@ -0,0 +1,31 @@
+---
+authority_id: bgn
+id: 1943
+language: iso-639-2:kor
+source_script: Kore
+destination_script: Latn
+name: BGN Agreement -- Korean McCune-Reischauer System (1943)
+url:
+creation_date: 1943
+adoption_date:
+description:
+
+notes:
+
+tests:
+  - source: 서울
+    expected: "Sŏul"
+  - source: 平壤
+    expected: "P’yŏngyang"
+  - source: 釜山
+    expected: "Pusan"
+  - source: 大邱
+    expected: "Taegu"
+  - source: 慶尚南道
+    expected: "Kyŏngsangnamdo"
+
+map:
+  character_separator: ""
+  word_separator: " "
+  title_case: True
+  inherit: [var-kor-Kore-Hang-2013, bgn-kor-Hang-Latn-1943]

data/maps/bgna-bul-Cyrl-Latn-2006.yaml

@@ -0,0 +1,208 @@
+---
+authority_id: bgna
+id: 2006
+language: iso-639-2:bul
+source_script: Cyrl
+destination_script: Latn
+name: Republic of Bulgaria system for transliteration of Bulgarian geographical names with Roman letters
+creation_date: 2006
+adoption_date: 2006-10-26
+
+tests:
+  - source: нунатак Абрит
+    expected: nunatak Abrit
+  - source: връх Академия
+    expected: vrah Akademiya
+  - source: връх Ами Буе
+    expected: vrah Ami Bue
+  - source: нос Айтос
+    expected: nos Aytos
+  - source: залив Баба Тонка
+    expected: zaliv Baba Tonka
+  - source: Балабански камък
+    expected: Balabanski kamak
+  - source: Бедечки поток
+    expected: Bedechki potok
+  - source: нос Бяга
+    expected: nos Byaga
+  - source: остров Качо
+    expected: ostrov Kacho # wiki ostrov Cacho
+  - source: Чакъров остров
+    expected: Chakarov ostrov
+  - source: връх Дъбник
+    expected: vrah Dabnik
+  - source: залив Десислава
+    expected: zaliv Desislava
+  - source: ледник Джераси
+    expected: lednik Dzherasi
+  - source: Джегова скала
+    expected: Dzhegova skala
+  - source: Нунатак Едуард
+    expected: Nunatak Eduard
+  - source: Елховска седловина
+    expected: Elhovska sedlovina
+  - source: ледник Етър
+    expected: lednik Etar
+  - source: нунатак Филип Тотю
+    expected: nunatak Filip Totyu
+  - source: ледник Габаре
+    expected: lednik Gabare
+  - source: риф Гергини
+    expected: rif Gergini
+  - source: Гяуров връх
+    expected: Gyaurov vrah
+  - source: Гуцалски рид
+    expected: Gutsalski rid # not found
+  - source: Хараламбиев остров
+    expected: Haralambiev ostrov
+  - source: връх Ичера
+    expected: vrah Ichera
+  - source: полуостров Йоан Павел II
+    expected: poluostrov Yoan Pavel II
+  - source: нос Иван Александър
+    expected: nos Ivan Aleksandar
+  - source: нос Иречек
+    expected: nos Irechek
+  - source: нос Кърджали
+    expected: nos Kardzhali
+  - source: седловина Кърнаре
+    expected: sedlovina Karnare
+  - source: нунатак Керсеблепт
+    expected: nunatak Kerseblept
+  - source: Кондофрейски възвишения
+    expected: Kondofreyski vazvisheniya
+  - source: Костинбродски проход
+    expected: Kostinbrodski prohod
+  - source: връх Кожух
+    expected: vrah Kozhuh
+  - source: Кукерски нунатаци
+    expected: Kukerski nunatatsi
+  - source: залив Лазурен бряг
+    expected: zaliv Lazuren bryag
+  - source: връх Лудогорие
+    expected: vrah Ludogorie
+  - source: Лютибродски скали
+    expected: Lyutibrodski skali
+  - source: Масларов нунатак
+    expected: Maslarov nunatak
+  - source: Михневски връх
+    expected: Mihnevski vrah
+  - source: рид Митино
+    expected: rid Mitino
+  - source: езеро Наяда
+    expected: ezero Nayada
+  - source: нос Никюп
+    expected: nos Nikyup
+  - source: рид Оборище
+    expected: rid Oborishte
+  - source: залив Олуша
+    expected: zaliv Olusha
+  - source: Оряховски възвишения
+    expected: Oryahovski vazvisheniya
+  - source: нунатак Памидово
+    expected: nunatak Pamidovo
+  - source: връх Парангалица
+    expected: vrah Parangalitsa
+  - source: Първомайски провлак
+    expected: Parvomayski provlak
+  - source: ледник Патлейна
+    expected: lednik Patleyna
+  - source: полуостров Перник
+    expected: poluostrov Pernik
+  - source: връх Петко Войвода
+    expected: vrah Petko Voyvoda
+  - source: остров Фанагория
+    expected: ostrov Fanagoriya
+  - source: нос Плас
+    expected: nos Plas
+  - source: Пресиянов рид
+    expected: Presiyanov rid
+  - source: връх Принсипе де Астуриас
+    expected: vrah Prinsipe de Asturias # wiki: vrah Príncipe de Asturias
+  - source: нунатак Ръченица
+    expected: nunatak Rachenitsa
+  - source: връх Райна Княгиня
+    expected: vrah Rayna Knyaginya
+  - source: Рид Ръжана
+    expected: Rid Razhana
+  - source: връх Ригс
+    expected: vrah Rigs
+  - source: остров Рогулят
+    expected: ostrov Rogulyat
+  - source: ледник Сабазий
+    expected: lednik Sabaziy
+  - source: ледник Съединение
+    expected: lednik Saedinenie
+  - source: нунатак Сенокос
+    expected: nunatak Senokos
+  - source: Сейдолски камък
+    expected: Seydolski kamak
+  - source: ледник Щерна
+    expected: lednik Shterna
+  - source: връх Шишман
+    expected: vrah Shishman
+  - source: ледник Сигмен
+    expected: lednik Sigmen
+  - source: Седловина Синитово
+    expected: Sedlovina Sinitovo
+  - source: Ледник Скаплизо
+    expected: Lednik Skaplizo
+  - source: залив Слънчев бряг
+    expected: zaliv Slanchev bryag
+  - source: остров Соатрис
+    expected: ostrov Soatris
+  - source: планина Софийски Университет
+    expected: planina Sofiyski Universitet
+  - source: ледник Сребърна
+    expected: lednik Srebarna
+  - source: Средногорски възвишения
+    expected: Srednogorski vazvisheniya
+  - source: Св. Евтимиев камък
+    expected: Sv. Evtimiev kamak
+  - source: база Св. Климент Охридски
+    expected: baza Sv. Kliment Ohridski
+  - source: връх Стъргел
+    expected: vrah Stargel
+  - source: нунатак Сторгозия
+    expected: nunatak Storgoziya # nunatak Storgozia according to wiki
+  - source: нунатак Сурвакари
+    expected: nunatak Survakari
+  - source: ледник Световрачене
+    expected: lednik Svetovrachene
+  - source: остров Теменуга
+    expected: ostrov Temenuga
+  - source: Тракийски възвишения
+    expected: Trakiyski vazvisheniya
+  - source: хълм Цамблак
+    expected: halm Tsamblak
+  - source: ледник Урдовиза
+    expected: lednik Urdoviza
+  - source: остров Вълчедръм
+    expected: ostrov Valchedram
+  - source: острови Вардим
+    expected: ostrovi Vardim
+  - source: Владигеров проток
+    expected: Vladigerov protok
+  - source: ледник Ябланица
+    expected: lednik Yablanitsa
+  - source: залив Ямфорина
+    expected: zaliv Yamforina
+  - source: Йовков нос
+    expected: Yovkov nos
+  - source: рид Заберново
+    expected: rid Zabernovo
+  - source: ледник Збелсурд
+    expected: lednik Zbelsurd
+  - source: Жефарович камък
+    expected: Zhefarovich kamak
+  - source: връх Зиези
+    expected: vrah Ziezi
+  - source: залив Златни пясъци
+    expected: zaliv Zlatni pyasatsi
+  - source: ледник Злокучене
+    expected: lednik Zlokuchene
+  - source: проток Злогош
+    expected: protok Zlogosh
+
+map:
+  inherit: apcbg-bul-Cyrl-Latn-1995