interscript 0.1.6 → 0.1.7

data/maps/odni-rus-Cyrl-Latn-2015.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: odni
 id: 2015
-language: rus
+language: iso-639-2:rus
 source_script: Cyrl
 destination_script: Latn
 name: Office of the Director Of National Intelligence Russian Personal Names 2015, ICS-630-01 Annex K

data/maps/odni-srp-Cyrl-Latn-2015.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: odni
 id: 2015
-language: srp
+language: iso-639-2:srp
 source_script: Cyrl
 destination_script: Latn
 name: Office of the Director Of National Intelligence Serbian Personal Names 2015, ICS-630-01 Annex L

data/maps/odni-tat-Cyrl-Latn-2015.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: odni
 id: 2015
-language: tat
+language: iso-639-2:tat
 source_script: Cyrl
 destination_script: Latn
 name: Standards for the transliteration of tatar personal names in written reports and products

data/maps/odni-tgk-Cyrl-Latn-2015.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: odni
 id: 2015
-language: tgk
+language: iso-639-2:tgk
 source_script: Cyrl
 destination_script: Latn
 name: Standards for the transliteration of tajik personal names in written reports and products

data/maps/odni-uig-Cyrl-Latn-2015.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: odni
 id: 2015
-language: uig
+language: iso-639-2:uig
 source_script: Cyrl
 destination_script: Latn
 name: Standards for the transliteration of uyghur personal names in written reports and products

data/maps/odni-ukr-Cyrl-Latn-2015.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: odni
 id: 2015
-language: ukr
+language: iso-639-2:ukr
 source_script: Cyrl
 destination_script: Latn
 name: Office of the Director Of National Intelligence Ukrainian Personal Names 2004 System, ICS 630-01 Annex M

data/maps/odni-urd-Arab-Latn-2015.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: odni
 id: 2015
-language: urd
+language: iso-639-2:urd
 source_script: Arab
 destination_script: Latn
 name: Office of the Director Of National Intelligence Hindi Urdu Personal Names 2015 System, ICS-630-01 Annex F

data/maps/odni-uzb-Cyrl-Latn-2015.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: odni
 id: 2015
-language: uzb
+language: iso-639-2:uzb
 source_script: Cyrl
 destination_script: Latn
 name: Office of the Director Of National Intelligence Uzbek Personal Names 2015, ICS-630-01 Annex V

data/maps/royin-tha-Thai-Latn-1939-generic.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: royin
 id: 1939-generic
-language: tha
+language: iso-639-2:tha
 source_script: Thai
 destination_script: Latn
 name: Royal Thai General System of Transcription (1939) Generic

data/maps/royin-tha-Thai-Latn-1968.yaml

@@ -1,10 +1,14 @@
 ---
 authority_id: royin
 id: 1968-chained
-language: tha
+language: iso-639-2:tha
 source_script: Thai
 destination_script: Latn
 name: Royal Thai General System of Transcription (1968)
+alias:
+  ogc11122:
+    code: tha_Thai2Latn_RIT_1968
+    description: Thai Royal Institute of Thailand 1968 System (out of date)
 url: http://www.royin.go.th/wp-content/uploads/royin-ebook/276/FileUpload/758_6484.pdf
 creation_date: 1968
 adoption_date:

data/maps/royin-tha-Thai-Latn-1999-chained.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: royin
 id: 1999-chained
-language: tha
+language: iso-639-2:tha
 source_script: Thai
 destination_script: Latn
 name: Royal Thai General System of Transcription (1999)

data/maps/royin-tha-Thai-Latn-1999.yaml

@@ -1,10 +1,14 @@
 ---
 authority_id: royin
 id: 1999
-language: tha
+language: iso-639-2:tha
 source_script: Thai
 destination_script: Latn
 name: Royal Thai General System of Transcription (1999)
+alias:
+  ogc11122:
+    code: tir_Thai2Latn_RIT_2000
+    description: Thai Royal Institute of Thailand 2000 System
 url: http://www.royin.go.th/wp-content/uploads/royin-ebook/276/FileUpload/758_6484.pdf
 creation_date: 1999
 adoption_date:

data/maps/sac-zho-Hans-Latn-1979.yaml

@@ -1,10 +1,14 @@
 ---
 authority_id: sac
 id: 1979
-language: chn
+language: iso-639-2:chn
 source_script: Hans
 destination_script: Latn
 name: Pinyin
+alias:
+  ogc11122:
+    code: zho_Hani2Latn_GCH_1979
+    description: Chinese Hanyu Pinyin 1979 System
 url: ftp://ftp.cuhk.hk/pub/chinese/ifcss/software/data/Uni2Pinyin.gz
 creation_date: 1996-04-17
 description: |

data/maps/sasm-mon-Mong-Latn-general-1978.yaml

@@ -0,0 +1,389 @@
+---
+authority_id: sasm
+id: general-1978
+language: iso-639-2:mon
+source_script: Mong
+destination_script: Latn
+name: Transliteration of Place Names in Minority Languages in Chinese Pinyin (Mongolian, general)
+url: http://www.moe.gov.cn/ewebeditor/uploadfile/2015/01/12/20150112161436994.pdf
+creation_date: 1976-06
+description: |
+  Article 1
+
+  The main purpose of the transliteration method of Chinese pinyin
+  letter transliteration of place names in minority languages:
+
+  (1) As a standard for spelling place names in minority languages ​​
+  with Chinese phonetic alphabet;
+  (2) As a phonetic tool for surveying and recording ethnic place
+  names in map surveying and mapping work;
+  (3) As the main basis for the pronunciation and selection of place
+  names in ethnic minority languages ​​in Chinese character
+  transliteration;
+  (4) Provide convenient conditions for the unified compilation of my
+  country's place-name information and index in alphabetical order.
+
+  Article 2
+
+  The transliteration method is limited to the 26 letters in the
+  "Hanyu Pinyin Plan", two letters with additional symbols and a
+  sound-proof symbol. In order to make the transcription and phonetic
+  more accurate, the syllable structure can not be restricted by the
+  syllable form of Mandarin Chinese. Soundproof symbols can be used
+  in various occasions that are easy to be confused. When recording,
+  additional symbols can be added to specific letters to represent
+  special sounds.
+
+  The transliteration and transliteration of ethnic minority scripts
+  in Latin script shall be based on the script. Letters with the same
+  or similar pronunciation and usage as in the "Hanyu Pinyin Plan"
+  shall be written according to them; different or dissimilar letters
+  shall be transcribed separately. If the text does not use Latin
+  letters, the corresponding Chinese phonetic letters shall be used
+  according to the pronunciation of the text. If there is no text,
+  according to the general voice mark.
+
+  Article 3
+
+  Special place names are handled with reference to the following
+  methods:
+
+  (1) If the commonly used Chinese name is partly transliterated and
+  partly translated, the transliteration part is spelled according to
+  the transliteration method, and the free translation part is
+  spelled according to the pronunciation of the Chinese character.
+
+  (2) If the commonly used Chinese character translation is a section
+  translation, the actual situation may be considered. Some are
+  transliterated according to the full name of the original name, and
+  some are spelled according to the Hanyu pronunciation of the
+  section translation.
+
+  (3) If the translated name of a Chinese character originally came
+  from a minority language, but later became a Chinese form and has
+  been commonly used, it can be spelled according to the
+  pronunciation of the Chinese character, including the original name
+  transliterated if necessary.
+
+  (4) Other special circumstances shall be dealt with according to
+  specific discretion.
+
+original_description: |
+  第一条
+
+  少数民族语地名汉语拼音字母音译转写法的主要用途：
+
+  （1） 作为用汉语拼音字母拼写少数民族语地名的标准；
+  （2）作为地图测绘工作中调査记录少数民族语地名的记音工具；
+  （3） 作为汉字音译少数民族语地名定音和选字的主要依据；
+  （4）为按照字母颐序统一编排我国地名资料和索引提供便利条件。
+
+  第二条
+
+  音译转写法限用《汉语拼音方案》中的26个字母，两个有附加符号的字母和一
+  个隔音符号，为了使转写和记音比较准确，音节结构可以不受汉语普通话音节
+  形式的限制，隔音符号可以在各种
+  容易混淆的场合应用，记音的时候附加符号可以加在特定的字母上面代表特殊
+  语音。
+
+
+  少数民族文字用拉丁字母的，音译转写以其文字为依据。跟《汉语拼音方案》
+  中读音和用法相同或相近的字母，一律照写；不同或不相近的字母分别规定转
+  写方式，文字不用拉丁字母的，根据文字的读音采用相应的汉语拼音字母表示
+  。没有文字的，根据通用语音标记。
+
+  第三条
+
+  特殊的地名参照下列办法处理：
+
+
+  (1）惯用的汉字译名如果是一部分音译，一部分意译，其音译部分根据音译转
+  写法拼写，意译部分按照汉字读音拼写。
+
+
+  (2）惯用的汉字译名如果是节译，可以斟酌具体体情况，有的按照原名全称音
+  译转写，有的按照节译的汉宇读音拼写。
+
+  (3) 汉字译名如果原先来自少数民族语，后来变成汉语形式并且已经通用，可以按
+  照汉字读音拼写，必要时括注音译转写的原名。
+
+  (4) 其他特殊情况具体斟酌处理。
+
+notes:
+  - The transliteration of Mongolian place names is based on the
+    customary pronunciation combining written and spoken languages.
+  - The length of Mongolian vowels are generally spelled regardless
+    of their spelling. Long vowels in place names are represented by
+    double letters.
+  - The Chinese phonetic alphabets transliterated by transliteration
+    are juxtaposed with or without symbols and adjuncts. Normally, the
+    spelling letters are without symbols, and the place names are
+    recorded by the acronym.
+
+original_notes:
+  (1) 蒙古语地名的音译转写，以书面语和口语通用相结合的习惯读法为依据。
+  (2) 蒙古语元音的长短，一般拼写不分，地名记音中长元音用双字母表示。
+  (3) 音译转写的汉语拼音字母有无符和加符并列的，一般拼写用无符字母，地名记音用加符宇母。
+
+implementation_notes:
+  - A separated final form of vowels a or e is common, and can appear at the end
+    of a word, word stem, or suffix. This form requires a final-shaped preceding
+    consonant and an inter-word gap in between. The vowels themselves appear as ᠎ᠠ,
+    and with consonants as ‍ᠬ᠎ᠠ q‑a, ‍ᠷ᠎ᠠ r‑a/r‑e, etc. This gap can be transliterated
+    with a hyphen. In digital typesetting, these forms are triggered by inserting
+    a U+180E ᠎ MONGOLIAN VOWEL SEPARATOR (HTML ᠎ · MVS) between the consonant
+    and vowel.
+  - All case suffixes, as well as any plural suffixes consisting of one or two
+    syllables are likewise separated by a preceding and hyphen-transliterated gap.
+    In digital typesetting, this gap is represented by a U+202F   NARROW NO-BREAK SPACE
+    (HTML   · NNBSP). A maximum of two case suffixes can be added to a stem.
+    Single-letter vowel suffixes appear with the final-shaped forms of a/e, i, or u/ü,
+    as in ᠭᠠᠵᠠᠷ ᠠ γaǰar‑a 'to the country' and ᠡᠳᠦᠷ ᠡ edür‑e 'on the day', or ᠤᠯᠤᠰ ᠢ ulus‑i
+    'the state' etc. Multi-letter suffixes most often start with an initial- (consonants),
+    medial- (vowels), or variant-shaped form.
+
+tests:
+  - source: ᠣᠷᠳᠣᠰ ᠬᠣᠲᠠ # Ордос хот
+    expected: ordos hota
+  - source: ᠪᠠᠶᠠᠨᠨᠠᠭᠤᠷ ᠬᠣᠲᠠ # Баяннуур хот
+    expected: bayannagur hota
+  - source: ᠤᠯᠠᠭᠠᠨᠬᠣᠲᠠ # Улаан хот
+    expected: ulaganhota
+  - source: ᠬᠣᠣᠯᠢᠠ ᠭᠣᠤᠯ ᠬᠣᠲᠠ # Хоолингол хот
+    expected: hoolia goul hota
+  - source: ᠡᠷᠢᠶᠡᠨ ᠬᠣᠲᠠ #  Эрээн хот
+    expected: eriyen hota
+  - source: ᠷᠠᠰᠢᠶᠠᠨ ᠬᠣᠲᠠ # Рашаан хот
+    expected: rasiyan hota
+  - source: ᠪᠠᠭᠠᠷᠢᠨ ᠪᠠᠷᠠᠭᠤᠨ ᠬᠣᠰᠢᠭᠤ # Баарин баруун хошуу
+    expected: bagarin baragun hosigu
+  - source: ᠪᠠᠭᠠᠷᠢᠨ ᠵᠡᠭᠦᠨ ᠬᠣᠰᠢᠭᠤ # Баарин зүүн хошуу
+    expected: bagarin jegun hosigu
+  - source: ᠲᠦᠩᠯᠢᠶᠣᠤ ᠬᠣᠲᠠ # Байшинт хот
+    expected: tungliyou hota
+  - source: ᠰᠢᠯᠢ ᠶᠢᠨ ᠬᠣᠲᠠ # Шилийн хот
+    expected: sili-yin hota
+  - source: ᠬᠣᠷᠢᠨ ᠭᠡᠷ ᠰᠢᠶᠠᠨ # Horinger County
+    expected: horin ger siyan
+  - source: ᠤᠳᠠ ᠲᠣᠭᠣᠷᠢᠭ # Уда тойрог
+    expected: uda togorig
+  - source: ᠦᠬᠠᠢ ᠬᠣᠲᠠ # Үхай хот
+    expected: uhai hota
+  - source: ᠬᠠᠶᠢᠨᠠᠨ ᠲᠣᠭᠣᠷᠢᠭ # Hainan District
+    expected: hayinan togorig
+  - source: ᠬᠠᠶᠢᠷᠤᠪ ᠤᠨ ᠲᠣᠬᠣᠢ ᠲᠣᠭᠣᠷᠢᠭ # Haibowan District
+    expected: hayirub-un tohoi togorig
+  - source: ᠤᠯᠠᠭᠠᠨᠴᠠᠪ ᠬᠣᠲᠠ # Улаанцав хот
+    expected: ulaganqab hota
+  - source: ᠵᠢᠨᠢᠩ ᠲᠣᠭᠣᠷᠢᠭ # Жинин тойрог
+    expected: jining togorig
+  - source: ᠹᠸᠩᠵᠸᠡ ᠬᠣᠲᠠ # Фенжень хот
+    expected: fwngjwe hota
+  - source: ᠵᠦᠸᠧᠽᠢ ᠰᠢᠶᠠᠨ # Zhuozi County
+    expected: juwezi siyan
+  - source: ᠬᠤᠸᠠᠳᠧ ᠰᠢᠶᠠᠨ # Huade County
+    expected: huwade siyan
+  - source: ᠱᠠᠩᠳᠤ ᠰᠢᠶᠠᠨ # Шанду шянь
+    expected: xangdu siyan
+  - source: ᠰᠢᠩᠾᠧ ᠰᠢᠶᠠᠨ # Шинхэ шянь
+    expected: singhe siyan
+  - source: ᠯᠢᠶᠠᠩᠴᠠᠩ ᠰᠢᠶᠠᠨ # Liangcheng County
+    expected: liyangqang siyan
+  - source: ᠴᠠᠬᠠᠷ ᠪᠠᠷᠠᠭᠤᠨ ᠭᠠᠷᠤᠨ ᠡᠮᠦᠨᠡᠳᠦ ᠬᠣᠰᠢᠭᠤ # Цахар баруун гарын өмнөд хошуу
+    expected: qahar baragun garun emunedu hosigu
+  - source: ᠴᠠᠬᠠᠷ ᠪᠠᠷᠠᠭᠤᠨ ᠭᠠᠷᠤᠨ ᠳᠤᠮᠳᠠᠳᠤ ᠬᠣᠰᠢᠭᠤ # Цахар баруун гарын дундад хошуу
+    expected: qahar baragun garun dumdadu hosigu
+  - source: ᠴᠠᠬᠠᠷ ᠪᠠᠷᠠᠭᠤᠨ ᠭᠠᠷᠤᠨ ᠬᠣᠶᠢᠲᠤ ᠬᠣᠰᠢᠭᠤ # Цахар баруун гарын хойд хошуу
+    expected: qahar baragun garun hoyitu hosigu
+  - source: ᠳᠥᠷᠪᠡᠳ ᠬᠣᠰᠢᠭᠤ # Дөрвөд хошуу
+    expected: dorbed hosigu
+  - source: ᠪᠤᠭᠤᠲᠤ ᠬᠣᠲᠠ # Бугaт хот
+    expected: bugutu hota
+  - source: ᠬᠥᠨᠳᠡᠯᠡᠨ ᠲᠣᠭᠣᠷᠢᠭ # Hondlon District
+    expected: hondelen togorig
+  - source: ᠳᠦᠩᠾᠧ ᠲᠣᠭᠣᠷᠢᠭ # Donghe District
+    expected: dunghe togorig
+  - source: ᠴᠢᠩᠱᠠᠨ ᠲᠣᠭᠣᠷᠢᠭ # Qingshan District
+    expected: qingxan togorig
+  - source: ᠰᠢᠭᠤᠶᠢᠲᠤ ᠲᠣᠭᠣᠷᠢᠭ # Шигуай тойрог
+    expected: siguyitu togorig
+  - source: ᠵᠢᠦ ᠶᠤᠸᠠᠨ ᠲᠣᠭᠣᠷᠢᠭ # Jiuyuan District
+    expected: jiu yuwan togorig
+  - source: ᠭᠦᠶᠠᠩ ᠰᠢᠶᠠᠨ # Guyang County
+    expected: guyang siyan
+  - source: ᠲᠦᠮᠡᠳ ᠵᠡᠭᠦᠨ ᠬᠣᠰᠢᠭᠤ # Түмэд Зүүн хошуу
+    expected: tumed jegun hosigu
+  - source: ᠲᠦᠮᠡᠳ ᠪᠠᠷᠠᠭᠤᠨ ᠬᠣᠰᠢᠭᠤ # Түмэд Баруун хошуу
+    expected: tumed baragun hosigu
+  - source: ᠳᠠᠷᠬᠠᠨ ᠮᠤᠤᠮᠢᠩᠭ᠋ᠠᠨ ᠬᠣᠯᠪᠣᠭᠠᠲᠤ ᠬᠣᠰᠢᠭᠤ # Darhan Muminggan United Banner
+    expected: darhan muumingg᠋an holbogatu hosigu
+  - source: ᠬᠡᠰᠢᠭᠲᠡᠨ ᠬᠣᠰᠢᠭᠤ # Hexigten Banner
+    expected: hesigten hosigu
+  - source: ᠰᠢᠨ᠎ᠡ ᠪᠠᠷᠭᠤ ᠵᠡᠭᠦᠨ ᠬᠣᠰᠢᠭᠤ # New Barag Left Banner
+    expected: sin-e bargu jegun hosigu
+  - source: ᠰᠢᠨ᠎ᠡ ᠪᠠᠷᠭᠤ ᠪᠠᠷᠠᠭᠤᠨ ᠬᠣᠰᠢᠭᠤ # New Barag Right Banner
+    expected: sin-e bargu baragun hosigu
+  - source: ᠣᠲᠣᠭ ᠬᠣᠰᠢᠭᠤ # Otog Banner
+    expected: otog hosigu
+  - source: ᠳ᠋ᠦᠩᠱᠧᠩ ᠲᠣᠭᠣᠷᠢᠭ # Dongsheng District
+    expected: d᠋ungxeng togorig
+  - source: ᠳᠠᠯᠠᠳ ᠬᠣᠰᠢᠭᠤ # Dalad Banner
+    expected: dalad hosigu
+  - source: ᠵᠡᠭᠦᠨᠭᠠᠷ ᠬᠣᠰᠢᠭᠤ # Jungar Banner
+    expected: jegungar hosigu
+  - source: ᠣᠲᠣᠭ ᠤᠨ ᠡᠮᠦᠨᠡᠳᠦ ᠬᠣᠰᠢᠭᠤ # Otog Front Banner
+    expected: otog-un emunedu hosigu
+  - source: ᠬᠠᠩᠭᠢᠨ ᠬᠣᠰᠢᠭᠤ # Hanggin Banner
+    expected:	hanggin hosigu
+  - source: ᠦᠦᠰᠢᠨ ᠬᠣᠰᠢᠭᠤ # Uxin Banner
+    expected: uusin hosigu
+  - source: ᠡᠵᠢᠨ ᠬᠣᠷᠣᠭ᠎ᠠ ᠬᠣᠰᠢᠭᠤ # Эзэн Хороо хошуу
+    expected: ejin horog-a hosigu
+  - source: ᠵᠠᠯᠠᠨ ᠠᠢ᠌ᠯ ᠬᠣᠲᠠ # Жалан-Айл хот
+    expected: jalan ai᠌l hota
+  - source: ᠶᠠᠭᠰᠢ ᠬᠣᠲᠠ # Ягши хот
+    expected: yagsi hota
+  - source: ᠮᠠᠨᠵᠤᠤᠷ ᠬᠣᠲᠠ # Манжуур хот
+    expected: manjuur hota
+  - source: ᠬᠠᠶᠢᠯᠠᠷ ᠲᠣᠭᠣᠷᠢᠭ # Хайлаар тойрог
+    expected: hayilar togorig
+  - source: ᠬᠣᠷᠴᠢᠨ ᠲᠣᠭᠣᠷᠢᠭ # Horqin District
+    expected: horqin togorig
+  - source: ᠺᠠᠶᠢᠯᠦ ᠰᠢᠶᠠᠨ # Kailu County
+    expected: kayilu siyan
+  - source: ᠬᠦᠷᠢᠶ᠎ᠡ ᠬᠣᠰᠢᠭᠤ # Hure Banner
+    expected: huriy-e hosigu
+  - source: ᠨᠠᠢᠮᠠᠨ ᠬᠣᠰᠢᠭᠤ # Naiman Banner
+    expected: naiman hosigu
+  - source: ᠵᠠᠷᠤᠳ ᠬᠣᠰᠢᠭᠤ # Jarud Banner
+    expected: jarud hosigu
+  - source: ᠬᠣᠷᠴᠢᠨ ᠪᠠᠷᠠᠭᠤᠨ ᠭᠠᠷᠤᠨ ᠳᠤᠮᠳᠠᠳᠤ ᠬᠣᠰᠢᠭᠤ # Хорчин Баруун Гарын Дундад Хошуу
+    expected: horqin baragun garun dumdadu hosigu
+  - source: ᠬᠣᠷᠴᠢᠨ ᠵᠡᠭᠦᠨ ᠭᠠᠷᠤᠨ ᠳᠤᠮᠳᠠᠳᠤ ᠬᠣᠰᠢᠭᠤ # Horqin Left Middle Banner
+    expected: horqin jegun garun dumdadu hosigu
+  - source: ᠬᠣᠷᠴᠢᠨ ᠵᠡᠭᠦᠨ ᠭᠠᠷᠤᠨ ᠬᠣᠶᠢᠲᠤ ᠬᠣᠰᠢᠭᠤ # Horqin Left Rear Banner
+    expected: horqin jegun garun hoyitu hosigu
+  - source: ᠤᠯᠠᠭᠠᠨᠬᠠᠳᠠ ᠬᠣᠲᠠ # Улаанхад хот
+    expected: ulaganhada hota
+  - source: ᠰᠡᠷᠡᠴᠢ # Salaqi
+    expected: sereqi
+  - source: ᠭᠡᠭᠡᠨ ᠭᠣᠤᠯ ᠬᠣᠲᠠ # Гэгээнгол хот
+    expected: gegen goul hota
+  - source: ᠠᠯᠠᠱᠠ ᠵᠡᠭᠦᠨ ᠬᠣᠰᠢᠭᠤ # Alxa Left Banner
+    expected: alaxa jegun hosigu
+  - source: ᠠᠯᠠᠱᠠ ᠪᠠᠷᠠᠭᠤᠨ ᠬᠣᠰᠢᠭᠤ # Alxa Right Banner
+    expected: alaxa baragun hosigu
+  - source: ᠡᠵᠡᠨ᠎ᠡ ᠬᠣᠰᠢᠭᠤ # Ejin Banner
+    expected: ejen-e hosigu
+  - source: ᠬᠥᠪᠡᠭᠡᠲᠦ ᠰᠢᠷ᠎ᠠ ᠬᠣᠰᠢᠭᠤ # Хөвөөт Шар хошуу
+    expected: hobegetu sir-a hosigu
+  - source: ᠦᠶᠤᠸᠠᠨ ᠰᠢᠶᠠᠨ # Үюань шянь
+    expected: uyuwan siyan
+  - source: ᠦᠴᠤᠸᠠᠨ ᠰᠢᠶᠠᠨ # Wuchuan County
+    expected: uquwan siyan
+  - source: ᠲᠦᠴᠢᠤᠸᠠᠨ ᠰᠢᠶᠠᠨ # Тучуань шянь
+    expected: tuqiuwan siyan
+  - source: ᠨᠢᠩᠴᠧᠩ ᠰᠢᠶᠠᠨ # Ningcheng County
+    expected: ningqeng siyan
+  - source: ᠯᠢᠨᠰᠢ ᠰᠢᠶᠠᠨ # Linxi County
+    expected: linsi siyan
+  - source: ᠯᠢᠨᠾᠧ ᠲᠣᠭᠣᠷᠢᠭ # Linhe
+    expected: linhe togorig
+  - source: ᠬᠠᠷᠠᠴᠢᠨ ᠬᠣᠰᠢᠭᠤ # Harqin Banner
+    expected: haraqin hosigu
+  - source: ᠳ᠋ᠧᠩᠺᠧᠦ ᠰᠢᠶᠠᠨ # Dengkou County
+    expected: d᠋engkeu siyan
+  - source: ᠠᠤᠬᠠᠨ ᠬᠣᠰᠢᠭᠤ # Aohan Banner
+    expected: auhan hosigu
+  - source: ᠴᠢᠩ ᠱᠦᠢ ᠾᠧ ᠰᠢᠶᠠᠨ # Qingshuihe County
+    expected: qing xui he siyan
+  - source: ᠬᠣᠳᠣᠩ ᠠᠷᠠᠳ ᠤᠨ ᠲᠣᠭᠣᠷᠢᠭ # Huimin District
+    expected: hodong arad-un togorig
+  - source: ᠰᠠᠶᠢᠬᠠᠨ ᠲᠣᠭᠣᠷᠢᠭ # Saihan District
+    expected: sayihan togorig
+  - source: ᠰᠢᠨ᠎ᠡ ᠬᠣᠲᠠ ᠲᠣᠭᠣᠷᠢᠭ # Xincheng District
+    expected: sin-e hota togorig
+  - source: ᠬᠠᠶᠢᠷᠤᠪ ᠤᠨ ᠲᠣᠬᠣᠢ ᠲᠣᠭᠣᠷᠢᠭ # Haibowan (Hairibin Tohoi) District
+    expected: hayirub-un tohoi togorig
+  - source: ᠰᠦᠩ ᠱᠠᠨ ᠲᠣᠭᠣᠷᠢᠭ # Songshan District
+    expected: sung xan togorig
+  - source: ᠬᠦᠩ ᠱᠠᠨ ᠲᠣᠭᠣᠷᠢᠭ # Hongshan District
+    expected: hung xan togorig
+  - source: ᠠᠷᠤᠨ ᠬᠣᠰᠢᠭᠤ # Арун хошуу
+    expected: arun hosigu
+  - source: ᠶᠤᠸᠠᠨ ᠪᠣᠣ ᠱᠠᠨ ᠲᠣᠭᠣᠷᠢᠭ # Юаньбаошань тойрог
+    expected: yuwan boo xan togorig
+  - source: ᠰᠢᠯᠤᠭᠤᠨ ᠬᠥᠪᠡᠭᠡᠲᠦ ᠴᠠᠭᠠᠨ ᠬᠣᠰᠢᠭᠤ # Шулуун Хөвөөт Цагаан хошуу
+    expected: silugun hobegetu qagan hosigu
+  - source: ᠰᠢᠯᠤᠭᠤᠨ ᠬᠥᠬᠡ ᠬᠣᠰᠢᠭᠤ # Шулуун хөх хошуу
+    expected: silugun hohe hosigu
+  - source: ᠤᠷᠠᠳ ᠤᠨ ᠳᠤᠮᠳᠠᠳᠤ ᠬᠣᠰᠢᠭᠤ # Урадын Дундад Хошуу
+    expected: urad-un dumdadu hosigu
+  - source: ᠤᠷᠠᠳ ᠤᠨ ᠡᠮᠦᠨᠡᠳᠦ ᠬᠣᠰᠢᠭᠤ # Урадын Өмнөд Хошуу
+    expected: urad-un emunedu hosigu
+  - source: ᠲᠣᠭᠲᠠᠬᠤ ᠰᠢᠶᠠᠨ # Тогтох шянь
+    expected: togtahu siyan
+  - source: ᠲᠠᠶᠢᠫᠤᠰᠧ ᠬᠣᠰᠢᠭᠤ # Тайвас хошуу
+    expected: tayipuse hosigu
+  - source: ᠰᠥᠨᠡᠳ ᠵᠡᠭᠦᠨ ᠬᠣᠰᠢᠭᠤ # Sonid Left Banner
+    expected: soned jegun hosigu
+  - source: ᠰᠥᠨᠡᠳ ᠪᠠᠷᠠᠭᠤᠨ ᠬᠣᠰᠢᠭᠤ # Sonid Right Banner
+    expected: soned baragun hosigu
+  - source: ᠣᠩᠨᠢᠭᠤᠳ ᠬᠣᠰᠢᠭᠤ # Ongniud Banner
+    expected: ongnigud hosigu
+  - source: ᠵᠠᠯᠠᠢᠳ ᠬᠣᠰᠢᠭᠤ # Jalaid Banner
+    expected: jalaid hosigu
+  - source: ᠬᠠᠩᠭᠢᠨ ᠬᠣᠶᠢᠲᠤ ᠬᠣᠰᠢᠭᠤ # Hanggin Rear Banner
+    expected: hanggin hoyitu hosigu
+  - source: ᠵᠡᠭᠦᠨ ᠤᠵᠤᠮᠤᠴᠢᠨ ᠬᠣᠰᠢᠭᠤ # East Ujimqin Banner
+    expected: jegun ujumuqin hosigu
+  - source: ᠠᠷᠤ ᠬᠣᠷᠴᠢᠨ ᠬᠣᠰᠢᠭᠤ # Ar Horqin Banner
+    expected: aru horqin hosigu
+  - source: ᠠᠪᠠᠭ᠎ᠠ ᠬᠣᠰᠢᠭᠤ # Abag Banner
+    expected: abag-a hosigu
+  - source: ᠪᠤᠷᠢᠶᠠᠳ # Буриад
+    expected: buriyad
+  - source: ᠤᠯᠠᠭᠠᠨᠪᠠᠭᠠᠲᠤᠷ # Улаанбаатар
+    expected: ulaganbagatur
+
+map:
+
+  characters:
+    "ᠠ": "a"      # \u1820
+    "ᠪ": "b"      # \u182a
+    "ᠼ": "c"      # \u183c
+    "ᠲ": "d"      # \u1832
+    "ᠳ": "d"      # \u1833
+    "ᠡ": "e"      # \u1821
+    "ᠹ": "f"      # \u1839
+    "ᠭ": "g"      # \u182d
+    "ᠺ": "g"      # \u183a
+    "ᠬ": "h"      # \u182c
+    "ᠾ": "h"     # \u183e
+    "ᠢ": "i"     # \u1822
+    "ᠵ": "j"      # \u1835
+    "ᠺ": "k"      # \u183a
+    "ᠯ": "l"      # \u182f
+    "ᠮ": "m"      # \u182e
+    "ᠨ": "n"      # \u1828
+    "ᠥ": "o"     # \u1825
+    "ᠫ": "p"      # \u182b
+    "ᠴ": "q"      # \u1834
+    "ᠷ": "r"      # \u1837
+    "ᠰ": "s"      # \u1830
+    "ᠲ": "t"      # \u1832
+    "ᠦ": "u"     # \u1826
+    "ᠸ": "w"      # \u1838
+    "ᠱ": "x"      # \u1831
+    "ᠶ": "y"      # \u1836
+    "ᠽ": "z"      # \u183d
+    "ᠣ": "o"      # \u1823
+      # - "o"     # General use
+      # - "ô"     # For place names
+    "ᠤ": "u"      # \u1824
+      # - "u"     # General use
+      # - "û"     # For place names
+    "ᠧ": "e"     # \u1827
+    "ᠩ": "ng"     # \u1829
+    "\u180e": "-" # MVS
+    "\u202f": "-" # NNBSP