interscript 0.1.6 → 0.1.7

data/maps/un-nep-Deva-Latn-1972.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: ungegn
 id: 1972
-language: nep
+language: iso-639-2:nep
 source_script: Deva
 destination_script: Latn
 name: REPORT ON THE CURRENT STATUS OF UNITED NATIONS ROMANIZATION SYSTEMS FOR GEOGRAPHICAL NAMES -- Nepali Romanization, 1972
@@ -28,40 +28,227 @@ notes:
   - |
     Ligatures may be formed with r as the first component: ~ह rha.
 
+  ######################################## Additional Note ############################################################
+  #    It is recommended that the vowel अ (a) should always be romanized except when it ends a                        #
+  #    name. In the scheme we are not yet able to detect a name. Analysing language pattern and the example           #
+  #    we found that actually when a word ends with a consonant it should be transliterated without the vowel in it.  #
+  #    so we are applying this rule as if a word ends with a consonant, in this case we are removing the ending 'a'.  #
+  #####################################################################################################################
+
+
 tests:
+  - source: "नेपाल"
+    expected: "nepāl"
+  - source: "नेपाल काठ्माडौं"
+    expected: "nepāl kāṭhmāḍauṁ"
   - source: "लेखन"
-    expected: "laekhana"
+    expected: "lekhan"
   - source: "मुद्रा"
-    expected: "maudaaraā"
+    expected: "mudrā"
   - source: "प्रशंसा"
-    expected: "paarashaṁsaā"
+    expected: "prashaṁsā"
   - source: "अंक"
-    expected: "aṁka"
+    expected: "aṁk"
   - source: "नेकपाले स्थगित स्थायी कमिटीको बैठक भदौ गते बोलाउने भएको"
-    expected: "naekapaālae saathagaita saathaāyaī kamaiṭaīkao baaiṭhaka bhadaau gatae baolaāunae bhaekao"
+    expected: "nekapāle sthagit sthāyī kamiṭīko baiṭhak bhadau gate bolāune bhaeko"
   - source: "न घर रह्यो, न परिवार"
-    expected: "na ghara rahaayao, na paraivaāra"
+    expected: "n ghar rahyo, n parivār"
   - source: "ढोरपाटनमा भुजीखोला बाढीपहिरोले अभिभावक गुमाएका बालबालिकाको बिचल्ली"
-    expected: "ḍhaorapaāṭanamaā bhaujaīkhaolaā baāḍhaīpahairaolae abhaibhaāvaka gaumaāekaā baālabaālaikaākao baichalaalaī"
+    expected: "ḍhorapāṭanamā bhujīkholā bāḍhīpahirole abhibhāvak gumāekā bālabālikāko bichallī"
   - source: "सुस्मिताका काका हेमबहादुर र काकीलाई पनि पहिरोले बगायो"
-    expected: "sausaamaitaākaā kaākaā haemabahaādaura ra kaākaīlaāī panai pahairaolae bagaāyao"
+    expected: "susmitākā kākā hemabahādur r kākīlāī pani pahirole bagāyo"
   - source: "संविधान जारी भएसँगै सार्वजनिक प्रशासनमा नयाँ उत्साह आउने अपेक्षा थियो"
-    expected: "saṁvaidhaāna jaāraī bhaesam̐gaai saāraavajanaika paarashaāsanamaā nayaām̐ utaasaāha āunae apaekaaṣhaā thaiyao"
+    expected: "saṁvidhān jārī bhaesam̐gai sārvajanik prashāsanamā nayām̐ utsāh āune apekṣhā thiyo"
   - source: "देशमा कोरोना संक्रमित र मृतकको संख्या हरेक दिन बढ्दो छ"
-    expected: "daeshamaā kaoraonaā saṁkaaramaita ra maṛtakakao saṁkhaayaā haraeka daina baḍhaadao chha"
+    expected: "deshamā koronā saṁkramit r mṛtakako saṁkhyā harek din baḍhdo chh"
   - source: "गाउँपालिकाका अध्यक्ष टिका गुरुङका अनुसार विष्णुदासलाई राजुले सुत्नका लागि बेलुका साथी लगेका थिए"
-    expected: "gaāum̐paālaikaākaā adhaayakaaṣha ṭaikaā gaurauṅakaā anausaāra vaiṣhaaṇaudaāsalaāī raājaulae sautaanakaā laāgai baelaukaā saāthaī lagaekaā thaie"
+    expected: "gāum̐pālikākā adhyakṣh ṭikā guruṅakā anusār viṣhṇudāsalāī rājule sutnakā lāgi belukā sāthī lagekā thie"
   - source: "यो आयोजना गाउँपालिकाको केन्द्र तेल्लोकमा पर्छ"
-    expected: "yao āyaojanaā gaāum̐paālaikaākao kaenaadaara taelaalaokamaā paraachha"
+    expected: "yo āyojanā gāum̐pālikāko kendr tellokamā parchh"
   - source: "सुस्मिताका काका हेमबहादुर र काकीलाई पनि पहिरोले बगायो"
-    expected: "sausaamaitaākaā kaākaā haemabahaādaura ra kaākaīlaāī panai pahairaolae bagaāyao"
+    expected: "susmitākā kākā hemabahādur r kākīlāī pani pahirole bagāyo"
   - source: "चैत पहिलो साता घर आएका उनी लकडाउन भएपछि यतै रोकिए"
-    expected: "chaaita pahailao saātaā ghara āekaā unaī lakaḍaāuna bhaepachhai yataai raokaie"
+    expected: "chait pahilo sātā ghar āekā unī lakaḍāun bhaepachhi yatai rokie"
   - source: "काम गर्न जानेको हकमा रोजगारदाता कम्पनीको पत्रसँगै वडा र जिल्ला प्रशासनको सिफारिस अनिवार्य गरिएको छ"
-    expected: "kaāma garaana jaānaekao hakamaā raojagaāradaātaā kamaapanaīkao pataarasam̐gaai vaḍaā ra jailaalaā paarashaāsanakao saiphaāraisa anaivaāraaya garaiekao chha"
+    expected: "kām garn jāneko hakamā rojagāradātā kampanīko patrasam̐gai vaḍā r jillā prashāsanako siphāris anivāry garieko chh"
+  - source: "ऋण"
+    expected: "ṛṇ"
+  - source: "अर्पित"
+    expected: "arpit"
+  - source: "अरार्यते"
+    expected: "arāryate"
+  - source: "श्रीमान्"
+    expected: "shrīmān"
+  - source: "खाँचो बिरुवा बैंकको"
+    expected: "khām̐cho biruvā baiṁkako"
+
 
 map:
 
+  rules:
+    - pattern: ([क]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'k'
+    - pattern: ([ख]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'kh'
+    - pattern: ([ग]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'g'
+    - pattern: ([घ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'gh'
+    - pattern: ([ङ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ṅ'
+    - pattern: ([च]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ch'
+    - pattern: ([छ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'chh'
+    - pattern: ([ज]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'j'
+    - pattern: ([झ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'jh'
+    - pattern: ([ञ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ñ'
+    - pattern: ([ट]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ṭ'
+    - pattern: ([ठ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ṭh'
+    - pattern: ([ड]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ḍ'
+    - pattern: ([ढ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ḍh'
+    - pattern: ([ण]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ṇ'
+    - pattern: ([त]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  't'
+    - pattern: ([थ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'th'
+    - pattern: ([द]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'd'
+    - pattern: ([ध]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'dh'
+    - pattern: ([न]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'n'
+    - pattern: ([प]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'p'
+    - pattern: ([फ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ph'
+    - pattern: ([ब]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'b'
+    - pattern: ([भ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'bh'
+    - pattern: ([म]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'm'
+    - pattern: ([य]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'y'
+    - pattern: ([र]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'r'
+    - pattern: ([ल]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'l'
+    - pattern: ([व]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'v'
+    - pattern: ([श]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'sh'
+    - pattern: ([ष]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ṣh'
+    - pattern: ([स]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  's'
+    - pattern: ([क़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'q'
+    - pattern: ([ख़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ḳh'
+    - pattern: ([ग़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'g'
+    - pattern: ([ज़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'z'
+    - pattern: ([ड़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ṙ'
+    - pattern: ([ढ़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'ṙh'
+    - pattern: ([फ़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'f'
+    - pattern: ([ह]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+      result:  'h'
+
+
+    - pattern: ([क])(?=\b) 
+      result:  'k'                  
+    - pattern: ([ख])(?=\b) 
+      result:  'kh'                 
+    - pattern: ([ग])(?=\b) 
+      result:  'g'                  
+    - pattern: ([घ])(?=\b) 
+      result:  'gh'                 
+    - pattern: ([ङ])(?=\b) 
+      result:  'ṅ'                 
+    - pattern: ([च])(?=\b) 
+      result:  'ch'                 
+    - pattern: ([छ])(?=\b) 
+      result:  'chh'                
+    - pattern: ([ज])(?=\b) 
+      result:  'j'                  
+    - pattern: ([झ])(?=\b) 
+      result:  'jh'                 
+    - pattern: ([ञ])(?=\b) 
+      result:  'ñ'                 
+    - pattern: ([ट])(?=\b) 
+      result:  'ṭ'                 
+    - pattern: ([ठ])(?=\b) 
+      result:  'ṭh'                
+    - pattern: ([ड])(?=\b) 
+      result:  'ḍ'                 
+    - pattern: ([ढ])(?=\b) 
+      result:  'ḍh'                
+    - pattern: ([ण])(?=\b) 
+      result:  'ṇ'                 
+    - pattern: ([त])(?=\b) 
+      result:  't'                  
+    - pattern: ([थ])(?=\b) 
+      result:  'th'                 
+    - pattern: ([द])(?=\b) 
+      result:  'd'                  
+    - pattern: ([ध])(?=\b) 
+      result:  'dh'                 
+    - pattern: ([न])(?=\b) 
+      result:  'n'                  
+    - pattern: ([प])(?=\b) 
+      result:  'p'                  
+    - pattern: ([फ])(?=\b) 
+      result:  'ph'                 
+    - pattern: ([ब])(?=\b) 
+      result:  'b'                  
+    - pattern: ([भ])(?=\b) 
+      result:  'bh'                 
+    - pattern: ([म])(?=\b)
+      result:  'm'                  
+    - pattern: ([य])(?=\b) 
+      result:  'y'                  
+    - pattern: ([र])(?=\b) 
+      result:  'r'                  
+    - pattern: ([ल])(?=\b) 
+      result:  'l'                  
+    - pattern: ([व])(?=\b) 
+      result:  'v'                  
+    - pattern: ([श])(?=\b) 
+      result:  'sh'                 
+    - pattern: ([ष])(?=\b) 
+      result:  'ṣh'                 
+    - pattern: ([स])(?=\b) 
+      result:  's'                  
+    - pattern: ([क़])(?=\b) 
+      result:  'q'                  
+    - pattern: ([ख़])(?=\b) 
+      result:  'ḳh'                 
+    - pattern: ([ग़])(?=\b) 
+      result:  'g'                  
+    - pattern: ([ज़])(?=\b) 
+      result:  'z'                  
+    - pattern: ([ड़])(?=\b) 
+      result:  'ṙ'                  
+    - pattern: ([ढ़])(?=\b) 
+      result:  'ṙh'                 
+    - pattern: ([फ़])(?=\b) 
+      result:  'f'                  
+    - pattern: ([ह])(?=\b) 
+      result:  'h'                  
+
   characters:
 
     # Vowels and Diphthongs
@@ -160,4 +347,4 @@ map:
     'ँ': 'm̐'
 
     # halanta
-    '्': 'a'
+    '्': ''

data/maps/un-nep-Deva-Latn-2013.yaml

@@ -0,0 +1,74 @@
+---
+authority_id: ungegn
+id: 2013
+language: iso-639-2:nep
+source_script: Deva
+destination_script: Latn
+name: REPORT ON THE CURRENT STATUS OF UNITED NATIONS ROMANIZATION SYSTEMS FOR GEOGRAPHICAL NAMES -- Nepali Romanization, 1972
+url: https://www.eki.ee/wgrs/v3_0/rom1_ne.pdf
+creation_date: 1972
+confirmation_date: 2013
+description: |
+  The United Nations recommended system was approved in 1972 (II/11), based on a report
+  prepared by D. N. Sharma. The note on the system was published in volume II of the
+  conference report.
+
+  There is no evidence of the use of the system either in Nepal or in international cartographic
+  products. The resolution IV/17 (1982) recommended association, inter alia, with Nepal in
+  carrying out further studies on the system. In Nepal a system of romanization is employed by
+  the Nepal Survey Department (NSD). This system has been applied to names appearing on
+  national mapping, and it is also used in the Geographic Information Infrastructure Programme
+  (NGIIP).
+
+  Nepali (Nepālī) uses the alphasyllabic script Devanāgarī.
+
+notes:
+
+  - In normal spelling ṙ and ṙh are not marked, instead dotless ड ḍa, ढ ḍha are used to denote these sounds.
+  - |
+    Ligatures may be formed with r as the first component: ~ह rha.
+
+tests:
+  - source: "नेपाल"
+    expected: "nepāl"
+  - source: "नेपाल काठ्माडौं"
+    expected: "nepāl kāṭhmāḍauṁ"
+  - source: "लेखन"
+    expected: "lekhan"
+  - source: "मुद्रा"
+    expected: "mudrā"
+  - source: "प्रशंसा"
+    expected: "prashaṁsā"
+  - source: "अंक"
+    expected: "aṁk"
+  - source: "नेकपाले स्थगित स्थायी कमिटीको बैठक भदौ गते बोलाउने भएको"
+    expected: "nekapāle sthagit sthāyī kamiṭīko baiṭhak bhadau gate bolāune bhaeko"
+  - source: "न घर रह्यो, न परिवार"
+    expected: "n ghar rahyo, n parivār"
+  - source: "ढोरपाटनमा भुजीखोला बाढीपहिरोले अभिभावक गुमाएका बालबालिकाको बिचल्ली"
+    expected: "ḍhorapāṭanamā bhujīkholā bāḍhīpahirole abhibhāvak gumāekā bālabālikāko bichallī"
+  - source: "सुस्मिताका काका हेमबहादुर र काकीलाई पनि पहिरोले बगायो"
+    expected: "susmitākā kākā hemabahādur r kākīlāī pani pahirole bagāyo"
+  - source: "संविधान जारी भएसँगै सार्वजनिक प्रशासनमा नयाँ उत्साह आउने अपेक्षा थियो"
+    expected: "saṁvidhān jārī bhaesam̐gai sārvajanik prashāsanamā nayām̐ utsāh āune apekṣhā thiyo"
+  - source: "देशमा कोरोना संक्रमित र मृतकको संख्या हरेक दिन बढ्दो छ"
+    expected: "deshamā koronā saṁkramit r mṛtakako saṁkhyā harek din baḍhdo chh"
+  - source: "गाउँपालिकाका अध्यक्ष टिका गुरुङका अनुसार विष्णुदासलाई राजुले सुत्नका लागि बेलुका साथी लगेका थिए"
+    expected: "gāum̐pālikākā adhyakṣh ṭikā guruṅakā anusār viṣhṇudāsalāī rājule sutnakā lāgi belukā sāthī lagekā thie"
+  - source: "यो आयोजना गाउँपालिकाको केन्द्र तेल्लोकमा पर्छ"
+    expected: "yo āyojanā gāum̐pālikāko kendr tellokamā parchh"
+  - source: "सुस्मिताका काका हेमबहादुर र काकीलाई पनि पहिरोले बगायो"
+    expected: "susmitākā kākā hemabahādur r kākīlāī pani pahirole bagāyo"
+  - source: "चैत पहिलो साता घर आएका उनी लकडाउन भएपछि यतै रोकिए"
+    expected: "chait pahilo sātā ghar āekā unī lakaḍāun bhaepachhi yatai rokie"
+  - source: "काम गर्न जानेको हकमा रोजगारदाता कम्पनीको पत्रसँगै वडा र जिल्ला प्रशासनको सिफारिस अनिवार्य गरिएको छ"
+    expected: "kām garn jāneko hakamā rojagāradātā kampanīko patrasam̐gai vaḍā r jillā prashāsanako siphāris anivāry garieko chh"
+  - source: "म त पहाडको एउटा गाउँबाट उडेर तल झरेको सिमल भुवा हुँ"
+    expected: "m t pahāḍako euṭā gāum̐bāṭ uḍer tal jhareko simal bhuvā hum̐"
+  - source: "अचम्म, धेरैले चिन्ने र सहजै बुझ्ने खालको चलेकै कवि पो भएँ"
+    expected: "achamm, dheraile chinne r sahajai bujhne khālako chalekai kavi po bhaem̐"
+  - source: "कतिपय सन्दर्भ र राजनीतिक रूपमा फाइदा हुने कुरामा कवि घिमिरेले गर्ने गरेको टिप्पणी पनि नेपाली लेखकमाझ आलोचित हुन्थ्यो"
+    expected: "katipay sandarbh r rājanītik rūpamā phāidā hune kurāmā kavi ghimirele garne gareko ṭippaṇī pani nepālī lekhakamājh ālochit hunthyo"
+
+map:
+  inherit: un-nep-Deva-Latn-1972

data/maps/un-rus-Cyrl-Latn-1987.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: ungegn
 id: 1987
-language: rus
+language: iso-639-2:rus
 source_script: Cyrl
 destination_script: Latn
 name: REPORT ON THE CURRENT STATUS OF UNITED NATIONS ROMANIZATION SYSTEMS FOR GEOGRAPHICAL NAMES -- Russian Romanization system

data/maps/un-ukr-Cyrl-Latn-1998.yaml

@@ -1,7 +1,7 @@
 ---
 authority_id: un
 id: 1998
-language: ukr
+language: iso-639-2:ukr
 source_script: Cyrl
 destination_script: Latn
 name: ROMANIZATION SYSTEM FOR BELARUSIAN, RUSSIAN AND UKRAINIAN CYRILLIC
@@ -14,17 +14,40 @@ map:
   inherit: gost-rus-Cyrl-Latn-16876-71-1983
 
   characters:
-    '\u0490' : '?'  # Ґ
-    '\u0491' : '?'  # ґ
-
-    '\u0404' : "Je" # Є
-    '\u0454' : "je" # є
-
+    '\u0413' : 'H'  # Г
+    '\u0490' : 'G'  # Ґ
+    '\u0404' : 'Ye' # Є
+    '\u0416' : 'Zh' # Ж
+    '\u0418' : 'Y'  # И
+    '\u0419' : 'Y'  # Й
     '\u0406' : 'I'  # І
-    '\u0456' : 'i'  # і
+    '\u0407' : 'Yi' # Ї
+    '\u0425' : 'Kh' # Х
+    '\u0426' : 'Ts' # Ц
+    '\u0427' : 'Ch' # Ч
+    '\u0428' : 'Sh' # Ш
+    '\u0429' : 'Shch' # Щ
+    '\u042C' : '' # Ь
+    '\u042E' : 'yu' # Ю
+    '\u042F' : 'ya' # Я
+    '\u042D' : 'E'  # Э
 
-    '\u0407' : 'I'  # Ї
-    '\u0457' : 'i'  # ї
+    '\u0433' : 'h'  # г
+    '\u0491' : 'g'  # ґ
+    '\u0454' : 'ye' # є
+    '\u0436' : 'zh' # ж
+    '\u0438' : 'y'  # и
+    '\u0439' : 'y'  # й
+    '\u0456' : 'i'  # і
+    '\u0457' : 'yi' # ї
+    '\u0445' : 'kh' # х
+    '\u0446' : 'ts' # ц
+    '\u0447' : 'ch' # ч
+    '\u0448' : 'sh' # ш
+    '\u0449' : 'shch' # щ
+    '\u044C' : ''   # ь
+    '\u044E' : 'yu' # ю
+    '\u044F' : 'ya' # я
+    '\u044D' : 'e'  # э
 
-    '\u2019' : '?'
-    "'" : '?'
+    '\u0027' : '"'  # '

data/maps/un-ukr-Cyrl-Latn-2012.yaml

@@ -0,0 +1,162 @@
+---
+authority_id: un
+id: 2012
+language: iso-639-2:ukr
+source_script: Cyrl
+destination_script: Latn
+name: ROMANIZATION SYSTEM FOR BELARUSIAN, RUSSIAN AND UKRAINIAN CYRILLIC
+url: http://www.eki.ee/wgrs/rom1_uk.htm
+creation_date: 2010
+
+tests:
+  - source: Алушта
+    expected: Alushta
+  - source: Андрій
+    expected: Andrii
+  - source: Борщагівка
+    expected: Borshchahivka
+  - source: Борисенко
+    expected: Borysenko
+  - source: Вінниця
+    expected: Vinnytsia
+  - source: Володимир
+    expected: Volodymyr
+  - source: Гадяч
+    expected: Hadiach
+  - source: Богдан
+    expected: Bohdan
+  - source: Згурський
+    expected: Zghurskyi
+  - source: Ґалаґан
+    expected: Galagan
+  - source: Ґорґани
+    expected: Gorgany
+  - source: Донецьк
+    expected: Donetsk
+  - source: Дмитро
+    expected: Dmytro
+  - source: Рівне
+    expected: Rivne
+  - source: Олег
+    expected: Oleh
+  - source: Есмань
+    expected: Esman
+  - source: Єнакієве
+    expected: Yenakiieve
+  - source: Гаєвич
+    expected: Haievych
+  - source: Короп'є
+    expected: Koropie
+  - source: Житомир
+    expected: Zhytomyr
+  - source: Жанна
+    expected: Zhanna
+  - source: Жежелів
+    expected: Zhezheliv
+  - source: Закарпаття
+    expected: Zakarpattia
+  - source: Казимирчук
+    expected: Kazymyrchuk
+  - source: Медвин
+    expected: Medvyn
+  - source: Михайленко
+    expected: Mykhailenko
+  - source: Іванків
+    expected: Ivankiv
+  - source: Іващенко
+    expected: Ivashchenko
+  - source: Їжакевич
+    expected: Yizhakevych
+  - source: Кадиївка
+    expected: Kadyivka
+  - source: Мар'їне
+    expected: Marine
+  - source: Йосипівка
+    expected: Yosypivka
+  - source: Стрий
+    expected: Stryi
+  - source: Олексій
+    expected: Oleksii
+  - source: Київ
+    expected: Kyiv
+  - source: Коваленко
+    expected: Kovalenko
+  - source: Лебедин
+    expected: Lebedyn
+  - source: Леонід
+    expected: Leonid
+  - source: Миколаїв
+    expected: Mykolaiv
+  - source: Маринич
+    expected: Marynych
+  - source: Ніжин
+    expected: Nizhyn
+  - source: Наталія
+    expected: Nataliia
+  - source: Одеса
+    expected: Odesa
+  - source: Онищенко
+    expected: Onyshchenko
+  - source: Полтава
+    expected: Poltava
+  - source: Петро
+    expected: Petro
+  - source: Решетилівка
+    expected: Reshetylivka
+  - source: Рибчинський
+    expected: Rybchynskyi
+  - source: Суми
+    expected: Sumy
+  - source: Соломія
+    expected: Solomiia
+  - source: Тернопіль
+    expected: Ternopil
+  - source: Троць
+    expected: Trots
+  - source: Ужгород
+    expected: Uzhhorod
+  - source: Уляна
+    expected: Uliana
+  - source: Фастів
+    expected: Fastiv
+  - source: Філіпчук
+    expected: Filipchuk
+  - source: Харків
+    expected: Kharkiv
+  - source: Христина
+    expected: Khrystyna
+  - source: Біла Церква
+    expected: Bila Tserkva
+  - source: Стеценко
+    expected: Stetsenko
+  - source: Чернівці
+    expected: Chernivtsi
+  - source: Шевченко
+    expected: Shevchenko
+  - source: Шостка
+    expected: Shostka
+  - source: Кишеньки
+    expected: Kyshenky
+  - source: Щербухи
+    expected: Shcherbukhy
+  - source: Гоща
+    expected: Hoshcha
+  - source: Гаращенко
+    expected: Harashchenko
+  - source: Яготин
+    expected: Yahotyn
+  - source: Ярошенко
+    expected: Yaroshenko
+  - source: Костянтин
+    expected: Kostiantyn
+  - source: Знам'янка
+    expected: Znamianka
+  - source: Феодосія
+    expected: Feodosiia
+  - source: Згорани
+    expected: Zghorany
+  - source: Розгон
+    expected: Rozghon
+
+map:
+  inherit: ua-ukr-Cyrl-Latn-2010