interscript 0.1.7 → 0.1.9

data/maps/un-urd-Arab-Latn-1972.yaml

@@ -0,0 +1,405 @@
+---
+authority_id: ungegn
+id: 2017
+language: iso-639-2:ara
+source_script: Arab
+destination_script: Latn
+name: ROMANIZATION OF ARABIC -- UNGEGN 2017 System
+url: https://www.eki.ee/wgrs/rom1_ur.htm
+creation_date: 2017
+confirmation date: 2018-06
+description: |
+  The United Nations recommended system was approved in 1972 (
+  II/11) and amended in 1977 (III/12), based on a report
+  prepared by D. N. Sharma. The tables and their corrections
+  were published in volume II of the conference reports1,2.
+
+  There is no evidence of the use of the system either in
+  Pakistan, India or in international cartographic products.
+  Instead, in Pakistan the Hunterian system is officially
+  used3. The resolutions III/12 (1977) and IV/17 (1982)
+  recommended association, inter alia, with Pakistan, in
+  carrying out further studies on the system.
+
+  Urdu (Urdū) uses the Perso-Arabic script which is written
+  from right to left. In the script vowel points are usually
+  omitted which makes it difficult to obtain uniform
+  romanizations. Some of the Arabic consonants are
+  undifferentiated in romanization which means that the
+  system is not fully reversible.
+notes:
+  - A If preceded by short a, it is romanized ‘ā, e.g. مَعمُل M‘āmul.
+  - B When و is imperceptible, e.g. in a few words of Persian origin when preceded by خ (ḳh).
+  - C Word-finally after a short vowel.
+  - D Marks aspiration of consonants.
+  - E The character ے is used only word-finally.
+tests:
+  - source: بوغدِی
+    expected: Bvghdī
+
+  - source: مَعمُل
+    expected: M‘āmul
+
+  - source: پَالِير
+    expected: Pālīr
+
+  - source: بیزوت كَلے
+    expected: Byzvt Kale
+
+  - source: عَمَل كوٹ
+    expected: ‘Amal Kvṭ
+
+  - source: ثَابِر
+    expected: Sābir
+
+  - source: شَاه نَثَار ميلة
+    expected: Shāh Nasār Mylah
+
+  - source: چَپرِی
+    expected: Chaprī
+
+  - source: أَحمَد خَان كَلے
+    expected: Ahmad Ḳhān Kale
+
+  - source: دُرَانِي
+    expected: Durānī
+
+  - source: ڈَنگِیلا
+    expected: Ḍangīlā
+
+  - source: ذَرَانِی
+    expected: Zarānī
+
+  - source: بُركِي
+    expected: Burkī
+
+  - source: گِیدَڑَه
+    expected: Gīdaṙah
+
+  - source: عَلِي زَائِي
+    expected: ‘Alī Zā-ī
+
+  - source: ژوب
+    expected: Ỵvb
+
+  - source: بِسَاتُو
+    expected: Bisātū
+
+  - source: أَحمَدِي شَامَا
+    expected: Ahmadī Shāmā
+
+  - source: اَصَالَت كَلے
+    expected: Asālat Kale
+
+  - source: خَضَر خَان
+    expected: Ḳhazar Ḳhān
+
+  - source: سُلْطَان
+    expected: Sultān
+
+  - source: عَزَم سَيِّد نُور كَلے
+    expected: ‘Azam Sayyid Nūr Kale
+
+  - source: بغَاكِي
+    expected: Bghākī
+
+  - source: حَقدَرَه
+    expected: Haqdarah
+
+  - source: کَچکِینَہ
+    expected: Kachkīnaḥ
+
+  - source: بَاگَن
+    expected: Bāgan
+
+  - source: بُلبَلَک
+    expected: Bulbalak
+
+  - source: بِلیَامِین
+    expected: Bilyāmīn
+
+  - source: نَہر
+    expected: Nahr
+
+  - source: اَرَوْالِی
+    expected: Arawālī
+
+  - source: مَہردِی
+    expected: Mahrdī
+
+  - source: بَڑھ
+    expected: Baṙh
+
+  - source:  یَاردَا کَلے
+    expected: Yārdā Kale
+
+  - source: بهَائِي خَان
+    expected: Bhā-ī Ḳhān
+
+  - source: پھاشک
+    expected: Phāshk
+
+  - source: تھَلّ
+    expected: Thall
+
+  - source: پَٹھان ريَا
+    expected: Paṭhān Ryā
+
+  - source: جھِیل
+    expected: Jhīl
+
+  - source:  غَزْنِي سْپِين
+    expected: Ghaznī Spīn
+
+  - source: بَادشَاه چھُم
+    expected: Bādshāh Chhum
+
+  - source:  سِندھ
+    expected: Sindh
+
+  - source: ڈھَنڈ
+    expected: Ḍhanḍ
+
+  - source: خَان گھَڑِی
+    expected: Ḳhān Ghaṙī
+
+  - source: غُلَامَک كَلے
+    expected: Ghulāmak Kale
+
+  - source: خَپیَنگا
+    expected: Ḳhapyangā
+
+  - source: گَندَه كَلے
+    expected: Gandah Kale
+
+  - source: مَورپِتھِی
+    expected: Maurpithī
+
+  - source: درے پلارِی
+    expected: Dre Plārī
+
+  - source: آگرَہ
+    expected: Āgraḥ
+
+  - source: ڈَنڈَر
+    expected: Ḍanḍar
+
+  - source: گُبازانَہ
+    expected: Gubāzānaḥ
+
+  - source: حَےدَر عَلِی كَلے
+    expected: Haidar ‘Alī Kale
+
+  - source: تَودَہ چِینَہ
+    expected: Taudaḥ Chīnaḥ
+
+  - source: مُوسى خَان كَلے
+    expected: Mūsá Ḳhān Kale
+
+  - source: مُلَّا بَاغ
+    expected: Mullā Bāgh
+
+map:
+  postrules:
+    - pattern: (?<=\b)(?<!\b[‘|’|'|-])[\u0061-\uFFFF]
+      result: "upcase"
+    # don't capitalize defined article in the middle of a sentence
+    - pattern : ' At T'  # الت
+      result: ' at T'
+    - pattern : ' As̄ S̄'  # الث
+      result: ' as̄ S̄'
+    - pattern : ' Ad D' # الد
+      result: ' ad D'
+    - pattern : ' Az Z'  # الذ
+      result: ' az Z'
+    - pattern : ' Ar R' # الر
+      result: ' ar R'
+    - pattern : ' Az Z'  # الز
+      result: ' az Z'
+    - pattern : ' As S' # الس
+      result: ' as S'
+    - pattern : ' Ash Sh' # الش
+      result: ' ash Sh'
+    - pattern : ' As S'  # الص
+      result: ' as S'
+    - pattern : ' Az Z'  # الض
+      result: ' az Z'
+    - pattern : ' At T'  # الط
+      result: ' at T'
+    - pattern : ' Az Z' # الظ
+      result: ' az Z'
+    - pattern : ' Al L'  # الل
+      result: ' al L'
+    - pattern : ' An N'  # الن
+      result: ' an N'
+    - pattern: " Al " # ال
+      result: " al "
+  characters:
+    # special rules
+
+    '\s(?=\u0622\u0628\u064E\u0627\u062F)': '' # space followed by abad is removed
+    '\ufdf2': 'Allāh' # See note 5
+
+    # Vowels, Diphthongs, and Diacritical Marks
+    '\u064e' : 'a' # َ fatha
+    '\u064e\u0627' : 'ā'  # ـَا fatha followed by ا
+    '\u0627' : 'ā'  # ا
+    '\u0649\u0670' : 'ā' # ىٰ
+    '\u06D2\u0670' : 'ā' # ےٰ
+    '\u0622' : 'ā'  # آ
+    '\b\u0627' : ''  # ا
+    '\u064e(?=\u0629)' : '' # َ fatha followed by ta' marboota
+    '\u064e(?=a[h|t])' : '' # َ fatha followed by ta' marboota, handling different order of conversion
+
+    '\u0652' : ''  # ْ sokoon
+    '\u0659': 'ê'
+
+    '\u0650' : 'i' # karsra
+    '\u0650[\u064a|\u06cc]' : 'ī' # ـِي kasra followed by ي
+    '\u0650\u06d2\u0652' : 'e' # ـے
+    '\u0650\u06d2' : 'e' # ـے
+    '\u06d2' : 'e' # ـے
+
+    '\u064f' : 'u' # ُ damma
+    '\u064f\u0648' : 'ū'  # ـُو damma followed by و
+    '\u064f\u0648\u0652' : 'o'  # ـَوْ
+
+
+    '\u064e\u06d2' : 'ai' # ـے
+    '\u064e\u0648' : 'au'  # ـَو
+    '\u0670': 'á' # ىٰ
+    '\u0649': 'á' # ىٰ
+
+    # shadda
+    '\u0628\u0651' : 'bb'  # ب
+    '\u062a\u0651' : 'tt'  # ت
+    '\u062b\u0651' : 'ss'  # ث
+    '\u062c\u0651' : 'jj'  # ج
+    '\u062d\u0651' : 'hh' # ح
+    '\u062e\u0651' : 'ḳhḳh'  # خ
+    '\u062f\u0651' : 'dd'  # د
+    '\u0630\u0651' : 'zz'  # ذ
+    '\u0631\u0651' : 'rr'  # ر
+    '\u0632\u0651' : 'zz'  # ز
+    '\u0633\u0651' : 'ss'  # س
+    '\u0634\u0651' : 'sh'  # ش
+    '\u0635\u0651' : 'ss'  # ص
+    '\u0636\u0651' : 'ḏḏ'  # ض
+    '\u0637\u0651' : 'tt'  # ط
+    '\u0638\u0651' : 'zz'  # ظ
+    '\u063a\u0651' : 'ghgh'  # غ
+    '\u0641\u0651' : 'ff'  # ف
+    '\u0642\u0651' : 'qq'  # ق
+    '\u0643\u0651' : 'kk'  # ك
+    '\u0644\u0651' : 'll'  # ل
+    '\u0645\u0651' : 'mm'  # م
+    '\u0646\u0651' : 'nn'  # ن
+    '\u0647\u0651' : 'hh'  # ه
+    '\u0648\u0651' : 'vv'  # و
+    '[\u064a|\u06cc]\u0651' : 'yy'  # ي
+
+    # NOTE 1
+    '\u0650\b' : '-e' # ِ kasra
+    '\u0674' : '-e' # ٴ
+    '\u0654' : '-e' #  ٔ
+
+    '\u0650\u064a\u0651\u064e' : 'īy' # ـِيَّ
+    '\u0650\u064a(?=\u064e|u064f)' : 'iy' # ـِي kasra followed by ي
+    '\u064e\u0649' : 'ay'  # ـَى fatha followed by ى which is ا not ي
+    '\u064e\u0648\u0652' : 'aw'  # ـَوْ
+    '\u064e\u064a\u0652' : 'ay'  # ـَيْ
+    '\u0650\u06cc\u0651\u064e' : 'īy' # ـِيَّ
+    '\u064e\u064a' : 'aī'  # ـَي
+    '\u064e\u06cc' : 'aī'  # ـَي
+      # - '-ye'
+
+
+    # ta' marboota
+    '\u0629' : 'at' # ة in the middle of the sentence
+    '\u0629$' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{2})\u0629' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{3})\u0629' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{4})\u0629' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{5})\u0629' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{6})\u0629' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{7})\u0629' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{8})\u0629' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{9})\u0629' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{10})\u0629' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{11})\u0629' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{12})\u0629' : 'ah'
+    '(?<=\b\u0627\u0644[\u0600-\u06ff]{13})\u0629' : 'ah'
+
+
+
+    '\u0621' : '-' # ء
+    '\u0624' : '-' # ؤ
+    '\u0626' : '-' # ئ
+
+    '\u0623' : ''  # أ
+    '\u0625' : ''    # إ
+    # See note B
+    '\b\u0627\u0644' : 'al '  # ال
+    # '\uFE8E' : ''  # ﺎ
+
+    # Sun letters
+    '\b\u0627\u0644\u062a' : 'at t'  # الت
+    '\b\u0627\u0644\u062b' : 'as s'  # الث
+    '\b\u0627\u0644\u062f' : 'ad d' # الد
+    '\b\u0627\u0644\u0630' : 'az z'  # الذ
+    '\b\u0627\u0644\u0631' : 'ar r' # الر
+    '\b\u0627\u0644\u0632' : 'az z'  # الز
+    '\b\u0627\u0644\u0633' : 'as s' # الس
+    '\b\u0627\u0644\u0634' : 'ash sh' # الش
+    '\b\u0627\u0644\u0635' : 'as s'  # الص
+    '\b\u0627\u0644\u0636' : 'az z'  # الض
+    '\b\u0627\u0644\u0637' : 'at t'  # الط
+    '\b\u0627\u0644\u0638' : 'az z' # الظ
+    '\b\u0627\u0644\u0644' : 'al l'  # الل
+    '\b\u0627\u0644\u0646' : 'an n'  # الن
+
+
+    # consonant characters
+
+    '\u0628' : 'b'  # ب
+    '\u067E' : 'p' # پ
+    '\u062a' : 't'  # ت
+    '\u0679' : 'ṭ'  # ٹ
+    '\u062B' : 's'   # ث
+    '\u062c' : 'j'  # ج
+    '\u0686' : 'ch'  # ‫چ‬
+    '\u062d' : 'h' # ح
+    '\u062e' : 'ḳh'  # خ
+    '\u062f' : 'd'  # د
+    '\u0688' : 'ḍ' # ‫ڈ
+    '\u0630' : 'z'  # ذ
+    '\u0631' : 'r'  # ر
+    '\u0691' : 'ṙ'   # ڑ
+    '\u0632' : 'z'  # ز
+    '\u0698' : 'ỵ'  # ‫ژ‬
+    '\u0633' : 's'  # س
+    '\u0634' : 'sh'  # ش
+    '\u0635' : 's'  # ص
+    '\u0636' : 'z'  # ض
+    '\u0637' : 't'  # ط
+    '\u0638' : 'z'  # ظ
+    '\u0639' : '‘'  # ع
+    '\u064e\u0639' : '‘ā'  # ع NOTE A
+    '\u063a' : 'gh'  # غ
+    '\u0641' : 'f'  # ف
+    '\u0642' : 'q'  # ق
+    '\u0643' : 'k'  # ك
+    '\u06A9' : 'k' # ک
+    '\u06AF' : 'g' # ‫گ‬
+    '\u0644' : 'l'  # ل
+    '\u0645' : 'm'  # م
+    '[\u06BA|\u0646]' : 'n'  # ن, ں
+    '[\ufba9|\u06c1]' : 'h' # ہ , ﮩ
+    '(?<=[\u064e|\u0650|\u064f])[\ufba9|\u06c1]\b' : 'ḥ' # ہ , ﮩ NOTE C
+    '[\u0647|\u06be]' : 'h'  # ه, ھ
+    '\u0648' : 'v'  # و
+    '(?<=\u062e)\u0648' : 'ẉ'  # و NOTE B
+    '[\u064a|\u06cc]' : 'y'  # ي
+    # '\u0649' : 'y'  # ي
+    '\u06D0' : 'ē'   # ې
+    '\u06CD' : 'êy'  # ‫ۍ‬

data/maps/var-amh-Ethi-Latn-eae-2003.yaml

@@ -0,0 +1,466 @@
+---
+authority_id: var
+id: 2003
+language: amh
+source_script: Ethi
+destination_script: Latn
+name: Encyclopaedia Aethiopica Amharic transliteration system (2003)
+url: https://brill.com/view/book/edcoll/9789004419582/front-11.xml?language=en
+creation_date: 2003
+description: |
+  The Encyclopaedia Aethiopica (EAe) is a basic encyclopedia for Ethiopian and Eritrean studies. It employs an in-house form of romanization of Geez, Amharic, and other languages, which varies greatly from standard formats, such as BGN/PCGN: the emperor Menelek II's name, for example, is written as "Mənilək II".
+
+tests:
+  - source: የዜግነት ክብር በ ኢትዮጵያችን ጸንቶ
+    expected: yäzegənätə kəbərə bä ʾitəyoṗəyačənə ṣänəto
+  - source: ታየ ሕዝባዊነት ዳር እስከዳር በርቶ
+    expected: tayä ḥəzəbawinätə darə ʾəsəkädarə bärəto
+  - source: ለሰላም ለፍትህ ለሕዝቦች ነጻነት
+    expected: läsälamə läfətəhə läḥəzəbočə näṣanätə
+  - source: በእኩልነት በፍቅር ቆመናል ባንድነት
+    expected: bäʾəkulənätə bäfəqərə qomänalə banədənätə
+  - source: መሠረተ ፅኑ ሰብዕናን ያልሻርን
+    expected: mäśärätä ṣ̓ənu säbəʿənanə yaləšarənə
+  - source: ሕዝቦች ነን ለሥራ በሥራ የኖርን
+    expected: ḥəzəbočə nänə läśəra bäśəra yänorənə
+  - source: ድንቅ የባህል መድረክ ያኩሪ ቅርስ ባለቤት
+    expected: dənəqə yäbahələ mädəräkə yakuri qərəsə baläbetə
+  - source: የተፈጥሮ ጸጋ የጀግና ሕዝብ እናት
+    expected: yätäfäṭəro ṣäga yäǧägəna ḥəzəbə ʾənatə
+  - source: እንጠብቅሻለን አለብን አደራ
+    expected: ʾənəṭäbəqəšalänə ʾaläbənə ʾadära
+  - source: ኢትዮጵያችን ኑሪ እኛም ባንቺ እንኩራ
+    expected: ʾitəyoṗəyačənə nuri ʾəñamə banəči ʾənəkura
+  - source: ቋንቋ የድምጽ፣ የምልክት ወይም የምስል ቅንብር ሆኖ
+    expected: qʷanəqʷa yädəməṣə፣ yämələkətə wäyəmə yäməsələ qənəbərə hono
+  - source: ለማሰብ ወይም የታሰበን ሃሳብ ለሌላ ለማስተላለፍ የሚረዳ መሳሪያ ነው
+    expected: lämasäbə wäyəmə yätasäbänə hasabə lälela lämasətälaläfə yämiräda mäsariya näwə
+  - source: በአጭሩ ቋንቋ የምልክቶች ስርዓትና እኒህን ምልክቶች ለማቀናበር
+    expected: bäʾač̣əru qʷanəqʷa yämələkətočə sərəʿatəna ʾənihənə mələkətočə lämaqänabärə
+  - source: የሚያስፈልጉ ህጎች ጥንቅር ነው። ቋንቋወችን ለመፈረጅ እንዲሁም
+    expected: yämiyasəfäləgu həጎčə ṭənəqərə näwə። qʷanəqʷawäčənə lämäfäräǧə ʾənədihumə
+  - source: ለምክፈል የሚያስችሉ መስፈርቶችን ለማስቀመጥ ባለው ችግር
+    expected: läməkəfälə yämiyasəčəlu mäsəfärətočənə lämasəqämäṭə baläwə čəgərə
+  - source: ምክንያት በአሁኑ ሰዓት በርግጠኝነት ስንት ቋንቋ በዓለም ላይ
+    expected: məkənəyatə bäʾahunu säʿatə bärəgəṭäñənätə sənətə qʷanəqʷa bäʿalämə layə
+  - source: እንዳለ ማወቅ አስቸጋሪ ነው
+    expected: ʾənədalä mawäqə ʾasəčägari näwə
+  - source: አሰላ
+    expected: ʾasäla
+  - source: አሶሳ
+    expected: ʾasosa
+  - source: አንኮበር
+    expected: ʾanəkobärə
+  - source: አክሱም
+    expected: ʾakəsumə
+  - source: አዋሳ
+    expected: ʾawasa
+  - source: አዲስ ዘመን (ከተማ)
+    expected: ʾadisə zämänə (kätäma)
+  - source: አዲግራት
+    expected: ʾadigəratə
+  - source: አዳማ
+    expected: ʾadama
+  - source: ደምበጫ
+    expected: däməbäč̣a
+  - source: ደርባ
+    expected: därəba
+  - source: ደብረ ማርቆስ
+    expected: däbərä marəqosə
+  - source: ደብረ ብርሃን
+    expected: däbərä bərəhanə
+  - source: ደብረ ታቦር (ከተማ)
+    expected: däbərä taborə (kätäma)
+  - source: ደብረ ዘይት
+    expected: däbərä zäyətə
+  - source: ደገሃቡር
+    expected: dägähaburə
+  - source: ወልቂጤ
+    expected: wäləqiṭe
+  - source: ወልወል
+    expected: wäləwälə
+  - source: ወልደያ
+    expected: wälədäya
+  - source: ናይሎ ሳህራን
+    expected: nayəlo sahəranə
+  - source: አኙዋክኛ
+    expected: ʾañuwakəña
+  - source: ኡዱክኛ
+    expected: ʾudukəña
+  - source: ኦፓኛ
+    expected: ʾopaña
+  - source: ጉምዝኛ
+    expected: guməzəña
+  - source: አፋርኛ
+    expected: ʾafarəña
+  - source: አላባኛ
+    expected: ʾalabaña
+  - source: አርቦርኛ
+    expected: ʾarəborəña
+  - source: ባይሶኛ
+    expected: bayəsoña
+  - source: ቡሳኛ
+    expected: busaña
+  - source: ሁለተኛ ጥፋት ከገበያ ማንቀላፋት
+    expected: hulätäña ṭəfatə kägäbäya manəqälafatə
+  - source: ሁሉም ከልኩ አያልፍም
+    expected: hulumə käləku ʾayaləfəmə
+  - source: አልሞት ባይ ተጋዳይ
+    expected: ʾaləmotə bayə tägadayə
+  - source: ውርድ ከራሴ
+    expected: wərədə kärase
+  - source: ፀጉር መሰንጠቅ
+    expected: ṣ̓ägurə mäsänəṭäqə
+  - source: ግንትር ፀሐይ
+    expected: gənətərə ṣ̓äḥayə
+  - source: በሬ ወለደ
+    expected: bäre wälädä
+  - source: ራስ ሳይጠና ጉተና
+    expected: rasə sayəṭäna gutäna
+  - source: ለሆዴ ጠግቤ በልብሴ አንግቤ
+    expected: lähode ṭägəbe bäləbəse ʾanəgəbe
+  - source: ለልጅ ከሳቁለት ለውሻ ከሮጡለት
+    expected: läləǧə käsaqulätə läwəša käroṭulätə
+  - source: መልካም ባል መጥፎ ሴት ይገራል
+    expected: mäləkamə balə mäṭəfo setə yəgäralə
+  - source: ሆድና ግንባር አይሸሸግም
+    expected: hodəna gənəbarə ʾayəšäšägəmə
+  - source: ቀሊል አማት ሲሶ በትር አላት
+    expected: qälilə ʾamatə siso bätərə ʾalatə
+  - source: ጨው ለራስህ ብለህ ጣፍጥ አለበለዚያ ድንጋይ ነው ብለው ይወረውሩሀል
+    expected: č̣äwə lärasəhə bəlähə ṭafəṭə ʾaläbäläziya dənəgayə näwə bəläwə yəwäräwəruhalə
+  - source: ጀምሮ ይጨርሳል አልሞ ይተኩሳል
+    expected: ǧäməro yəč̣ärəsalə ʾaləmo yətäkusalə
+
+map:
+  characters:
+    '\u1200': "ha" # ሀ
+    '\u1201': "hu" # ሁ
+    '\u1202': "hi" # ሂ
+    '\u1203': "ha" # ሃ
+    '\u1204': "he" # ሄ
+    '\u1205': "hə" # ህ
+    '\u1206': "ho" # ሆ
+
+    '\u1208': "lä" # ለ
+    '\u1209': "lu" # ሉ
+    '\u120A': "li" # ሊ
+    '\u120B': "la" # ላ
+    '\u120C': "le" # ሌ
+    '\u120D': "lə" # ል
+    '\u120E': "lo" # ሎ
+
+    '\u1210': "ḥa" # ሐ
+    '\u1211': "ḥu" # ሑ
+    '\u1212': "ḥi" # ሒ
+    '\u1213': "ḥa" # ሓ
+    '\u1214': "ḥe" # ሔ
+    '\u1215': "ḥə" # ሕ
+    '\u1216': "ḥo" # ሖ
+
+    '\u1218': "mä" # መ
+    '\u1219': "mu" # ሙ
+    '\u121A': "mi" # ሚ
+    '\u121B': "ma" # ማ
+    '\u121C': "me" # ሜ
+    '\u121D': "mə" # ም
+    '\u121E': "mo" # ሞ
+
+    '\u1220': "śä" # ሠ
+    '\u1221': "śu" # ሡ
+    '\u1222': "śi" # ሢ
+    '\u1223': "śa" # ሣ
+    '\u1224': "śe" # ሤ
+    '\u1225': "śə" # ሥ
+    '\u1226': "śo" # ሦ
+
+    '\u1228': "rä" # ረ
+    '\u1229': "ru" # ሩ
+    '\u122A': "ri" # ሪ
+    '\u122B': "ra" # ራ
+    '\u122C': "re" # ሬ
+    '\u122D': "rə" # ር
+    '\u122E': "ro" # ሮ
+
+    '\u1230': "sä" # ሰ
+    '\u1231': "su" # ሱ
+    '\u1232': "si" # ሲ
+    '\u1233': "sa" # ሳ
+    '\u1234': "se" # ሴ
+    '\u1235': "sə" # ስ
+    '\u1236': "so" # ሶ
+
+    '\u1238': "šä" # ሸ
+    '\u1239': "šu" # ሹ
+    '\u123A': "ši" # ሺ
+    '\u123B': "ša" # ሻ
+    '\u123C': "še" # ሼ
+    '\u123D': "šə" # ሽ
+    '\u123E': "šo" # ሾ
+
+    '\u1240': "qä" # ቀ
+    '\u1241': "qu" # ቁ
+    '\u1242': "qi" # ቂ
+    '\u1243': "qa" # ቃ
+    '\u1244': "qe" # ቄ
+    '\u1245': "qə" # ቅ
+    '\u1246': "qo" # ቆ
+
+    '\u1260': "bä" # በ
+    '\u1261': "bu" # ቡ
+    '\u1262': "bi" # ቢ
+    '\u1263': "ba" # ባ
+    '\u1264': "be" # ቤ
+    '\u1265': "bə" # ብ
+    '\u1266': "bo" # ቦ
+
+    '\u1270': "tä" # ተ
+    '\u1271': "tu" # ቱ
+    '\u1272': "ti" # ቲ
+    '\u1273': "ta" # ታ
+    '\u1274': "te" # ቴ
+    '\u1275': "tə" # ት
+    '\u1276': "to" # ቶ
+
+    '\u1278': "čä" # ቸ
+    '\u1279': "ču" # ቹ
+    '\u127A': "či" # ቺ
+    '\u127B': "ča" # ቻ
+    '\u127C': "če" # ቼ
+    '\u127D': "čə" # ች
+    '\u127E': "čo" # ቾ
+
+    '\u1280': "ḫa" # ኀ
+    '\u1281': "ḫu" # ኁ
+    '\u1282': "ḫi" # ኂ
+    '\u1283': "ḫa" # ኃ
+    '\u1284': "ḫe" # ኄ
+    '\u1285': "ḫə" # ኅ
+    '\u1286': "ḫo" # ኆ
+
+    '\u1290': "nä" # ነ
+    '\u1291': "nu" # ኑ
+    '\u1292': "ni" # ኒ
+    '\u1293': "na" # ና
+    '\u1294': "ne" # ኔ
+    '\u1295': "nə" # ን
+    '\u1296': "no" # ኖ
+
+    '\u1298': "ñä" # ኘ
+    '\u1299': "ñu" # ኙ
+    '\u129A': "ñi" # ኚ
+    '\u129B': "ña" # ኛ
+    '\u129C': "ñe" # ኜ
+    '\u129D': "ñə" # ኝ
+    '\u129E': "ño" # ኞ
+
+    '\u12A0': "ʾa" # አ
+    '\u12A1': "ʾu" # ኡ
+    '\u12A2': "ʾi" # ኢ
+    '\u12A3': "ʾa" # ኣ
+    '\u12A4': "ʾe" # ኤ
+    '\u12A5': "ʾə" # እ
+    '\u12A6': "ʾo" # ኦ
+
+    '\u12A8': "kä" # ከ
+    '\u12A9': "ku" # ኩ
+    '\u12AA': "ki" # ኪ
+    '\u12AB': "ka" # ካ
+    '\u12AC': "ke" # ኬ
+    '\u12AD': "kə" # ክ
+    '\u12AE': "ko" # ኮ
+
+    '\u12B8': "ḵä" # ኸ
+    '\u12B9': "ḵu" # ኹ
+    '\u12BA': "ḵi" # ኺ
+    '\u12BB': "ḵa" # ኻ
+    '\u12BC': "ḵe" # ኼ
+    '\u12BD': "ḵə" # ኽ
+    '\u12BE': "ḵo" # ኾ
+
+    '\u12C8': "wä" # ወ
+    '\u12C9': "wu" # ዉ
+    '\u12CA': "wi" # ዊ
+    '\u12CB': "wa" # ዋ
+    '\u12CC': "we" # ዌ
+    '\u12CD': "wə" # ው
+    '\u12CE': "wo" # ዎ
+
+    '\u12D0': "ʿa" # ዐ
+    '\u12D1': "ʿu" # ዑ
+    '\u12D2': "ʿi" # ዒ
+    '\u12D3': "ʿa" # ዓ
+    '\u12D4': "ʿe" # ዔ
+    '\u12D5': "ʿə" # ዕ
+    '\u12D6': "ʿo" # ዖ
+
+    '\u12D8': "zä" # ዘ
+    '\u12D9': "zu" # ዙ
+    '\u12DA': "zi" # ዚ
+    '\u12DB': "za" # ዛ
+    '\u12DC': "ze" # ዜ
+    '\u12DD': "zə" # ዝ
+    '\u12DE': "zo" # ዞ
+
+    '\u12E0': "žä" # ዠ
+    '\u12E1': "žu" # ዡ
+    '\u12E2': "ži" # ዢ
+    '\u12E3': "ža" # ዣ
+    '\u12E4': "že" # ዤ
+    '\u12E5': "žə" # ዥ
+    '\u12E6': "žo" # ዦ
+
+    '\u12E8': "yä" # የ
+    '\u12E9': "yu" # ዩ
+    '\u12EA': "yi" # ዪ
+    '\u12EB': "ya" # ያ
+    '\u12EC': "ye" # ዬ
+    '\u12ED': "yə" # ይ
+    '\u12EE': "yo" # ዮ
+
+    '\u12F0': "dä" # ደ
+    '\u12F1': "du" # ዱ
+    '\u12F2': "di" # ዲ
+    '\u12F3': "da" # ዳ
+    '\u12F4': "de" # ዴ
+    '\u12F5': "də" # ድ
+    '\u12F6': "do" # ዶ
+
+    '\u1300': "ǧä" # ጀ
+    '\u1301': "ǧu" # ጁ
+    '\u1302': "ǧi" # ጂ
+    '\u1303': "ǧa" # ጃ
+    '\u1304': "ǧe" # ጄ
+    '\u1305': "ǧə" # ጅ
+    '\u1306': "ǧo" # ጆ
+
+    '\u1308': "gä" # ገ
+    '\u1309': "gu" # ጉ
+    '\u130A': "gi" # ጊ
+    '\u130B': "ga" # ጋ
+    '\u130C': "ge" # ጌ
+    '\u130D': "gə" # ግ
+
+    '\u1320': "ṭä" # ጠ
+    '\u1321': "ṭu" # ጡ
+    '\u1322': "ṭi" # ጢ
+    '\u1323': "ṭa" # ጣ
+    '\u1324': "ṭe" # ጤ
+    '\u1325': "ṭə" # ጥ
+    '\u1326': "ṭo" # ጦ
+
+    '\u1328': "č̣ä" # ጨ
+    '\u1329': "č̣u" # ጩ
+    '\u132A': "č̣i" # ጪ
+    '\u132B': "č̣a" # ጫ
+    '\u132C': "č̣e" # ጬ
+    '\u132D': "č̣ə" # ጭ
+    '\u132E': "č̣o" # ጮ
+
+    '\u1330': "ṗä" # ጰ
+    '\u1331': "ṗu" # ጱ
+    '\u1332': "ṗi" # ጲ
+    '\u1333': "ṗa" # ጳ
+    '\u1334': "ṗe" # ጴ
+    '\u1335': "ṗə" # ጵ
+    '\u1336': "ṗo" # ጶ
+
+    '\u1338': "ṣä" # ጸ
+    '\u1339': "ṣu" # ጹ
+    '\u133A': "ṣi" # ጺ
+    '\u133B': "ṣa" # ጻ
+    '\u133C': "ṣe" # ጼ
+    '\u133D': "ṣə" # ጽ
+    '\u133E': "ṣo" # ጾ
+
+    '\u1340': "ṣ̓ä" # ፀ
+    '\u1341': "ṣ̓u" # ፁ
+    '\u1342': "ṣ̓i" # ፂ
+    '\u1343': "ṣ̓a" # ፃ
+    '\u1344': "ṣ̓e" # ፄ
+    '\u1345': "ṣ̓ə" # ፅ
+    '\u1346': "ṣ̓o" # ፆ
+
+    '\u1348': "fä" # ፈ
+    '\u1349': "fu" # ፉ
+    '\u134A': "fi" # ፊ
+    '\u134B': "fa" # ፋ
+    '\u134C': "fe" # ፌ
+    '\u134D': "fə" # ፍ
+    '\u134E': "fo" # ፎ
+
+    '\u1350': "pä" # ፐ
+    '\u1351': "pu" # ፑ
+    '\u1352': "pi" # ፒ
+    '\u1353': "pa" # ፓ
+    '\u1354': "pe" # ፔ
+    '\u1355': "pə" # ፕ
+    '\u1356': "po" # ፖ
+
+    '\u1268': "vä" # ቨ
+    '\u1269': "vu" # ቩ
+    '\u126A': "vi" # ቪ
+    '\u126B': "va" # ቫ
+    '\u126C': "ve" # ቬ
+    '\u126D': "və" # ቭ
+    '\u126E': "vo" # ቮ
+
+    # Labiovelars
+    '\u1248': "qʷä" # ቈ
+    '\u124a': "qʷi" # ቊ
+    '\u124b': "qʷa" # ቋ
+    '\u124c': "qʷe" # ቌ
+    '\u124d': "qʷə" # ቍ
+
+    '\u1288': "ḫʷä" # ኈ
+    '\u128a': "ḫʷi" # ኊ
+    '\u128b': "ḫʷa" # ኋ
+    '\u128c': "ḫʷe" # ኌ
+    '\u128d': "ḫʷə" # ኍ
+
+    '\u12b0': "kʷä" # ኰ
+    '\u12b2': "kʷi" # ኲ
+    '\u12b3': "kʷa" # ኳ
+    '\u12b4': "kʷe" # ኴ
+    '\u12b5': "kʷə" # ኵ
+
+    '\u1310': "gʷä" # ጐ
+    '\u1312': "gʷi" # ጒ
+    '\u1313': "gʷa" # ጓ
+    '\u1314': "gʷe" # ጔ
+    '\u1315': "gʷə" # ጕ
+
+    '\u12c0': "ḵʷä" # ዀ
+    '\u12c2': "ḵʷi" # ዂ
+    '\u12c3': "ḵʷa" # ዃ
+    '\u12c4': "ḵʷe" # ዄ
+    '\u12c5': "ḵʷə" # ዅ
+
+    '\u1258': "q̱ʷä" # ቘ
+    '\u125a': "q̱ʷi" # ቚ
+    '\u125b': "q̱ʷa" # ቛ
+    '\u125c': "q̱ʷe" # ቜ
+    '\u125d': "q̱ʷə" # ቝ
+
+    # Numbers
+
+    '\u1369': "1"    # ፩
+    '\u136A': "2"    # ፪
+    '\u136B': "3"    # ፫
+    '\u136C': "4"    # ፬
+    '\u136D': "5"    # ፭
+    '\u136E': "6"    # ፮
+    '\u136F': "7"    # ፯
+    '\u1370': "8"    # ፰
+    '\u1371': "9"    # ፱
+    '\u1372': "10"   # ፲
+    '\u1373': "20"   # ፳
+    '\u1374': "30"   # ፴
+    '\u1375': "40"   # ፵
+    '\u1376': "50"   # ፶
+    '\u1377': "60"   # ፷
+    '\u1379': "80"   # ፹
+    '\u137A': "90"   # ፺
+    '\u137B': "100"  # ፻
+    '\u1372\u137B': "1000"  # ፲፻
+    '\u137B\u137B': "10000"  # ፻፻