interscript 0.1.9 → 2.0.5

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

@@ -1,269 +0,0 @@
----
-authority_id: ungegn
-id: 1972
-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: 2010
-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.
-
-  ######################################## 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: "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: "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]))|(([क])(?=\b))
-      result:  'k'
-    - pattern: (([ख]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ख])(?=\b))
-      result:  'kh'
-    - pattern: (([ग]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ग])(?=\b))
-      result:  'g'
-    - pattern: (([घ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([घ])(?=\b))
-      result:  'gh'
-    - pattern: (([ङ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ङ])(?=\b))
-      result:  'ṅ'
-    - pattern: (([च]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([च])(?=\b))
-      result:  'ch'
-    - pattern: (([छ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([छ])(?=\b))
-      result:  'chh'
-    - pattern: (([ज]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ज])(?=\b))
-      result:  'j'
-    - pattern: (([झ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([झ])(?=\b))
-      result:  'jh'
-    - pattern: (([ञ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ञ])(?=\b))
-      result:  'ñ'
-    - pattern: (([ट]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ट])(?=\b))
-      result:  'ṭ'
-    - pattern: (([ठ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ठ])(?=\b))
-      result:  'ṭh'
-    - pattern: (([ड]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ड])(?=\b))
-      result:  'ḍ'
-    - pattern: (([ढ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ढ])(?=\b))
-      result:  'ḍh'
-    - pattern: (([ण]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ण])(?=\b))
-      result:  'ṇ'
-    - pattern: (([त]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([त])(?=\b))
-      result:  't'
-    - pattern: (([थ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([थ])(?=\b))
-      result:  'th'
-    - pattern: (([द]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([द])(?=\b))
-      result:  'd'
-    - pattern: (([ध]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ध])(?=\b))
-      result:  'dh'
-    - pattern: (([न]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([न])(?=\b))
-      result:  'n'
-    - pattern: (([प]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([प])(?=\b))
-      result:  'p'
-    - pattern: (([फ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([फ])(?=\b))
-      result:  'ph'
-    - pattern: (([ब]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ब])(?=\b))
-      result:  'b'
-    - pattern: (([भ]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([भ])(?=\b))
-      result:  'bh'
-    - pattern: (([म]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([म])(?=\b))
-      result:  'm'
-    - pattern: (([य]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([य])(?=\b))
-      result:  'y'
-    - pattern: (([र]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([र])(?=\b))
-      result:  'r'
-    - pattern: (([ल]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ल])(?=\b))
-      result:  'l'
-    - pattern: (([व]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([व])(?=\b))
-      result:  'v'
-    - pattern: (([श]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([श])(?=\b))
-      result:  'sh'
-    - pattern: (([ष]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ष])(?=\b))
-      result:  'ṣh'
-    - pattern: (([स]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([स])(?=\b))
-      result:  's'
-    - pattern: (([क़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([क़])(?=\b))
-      result:  'q'
-    - pattern: (([ख़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ख़])(?=\b))
-      result:  'ḳh'
-    - pattern: (([ग़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ग़])(?=\b))
-      result:  'g'
-    - pattern: (([ज़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ज़])(?=\b))
-      result:  'z'
-    - pattern: (([ड़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ड़])(?=\b))
-      result:  'ṙ'
-    - pattern: (([ढ़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ढ़])(?=\b))
-      result:  'ṙh'
-    - pattern: (([फ़]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([फ़])(?=\b))
-      result:  'f'
-    - pattern: (([ह]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ह])(?=\b))
-      result:  'h'
-
-
-  characters:
-
-    # Vowels and Diphthongs
-    'अ': 'a'
-    'आ': 'ā'
-    'इ': 'i'
-    'ई': 'ī'
-    'उ': 'u'
-    'ऊ': 'ū'
-    'ऋ': 'ṛ'
-    'ॠ': 'ṝ'
-    'ऌ': 'l̤'
-    'ए': 'e'
-    'ऐ': 'ai'
-    'ओ': 'o'
-    'औ': 'au'
-
-    # Medials # Needed for connecting constants
-
-    'ा': "ā"
-    'ि': "i"
-    'ी': "ī"
-    'ु': "u"
-    'ू': "ū"
-    'ृ': "ṛ"
-    'े': "e"
-    'ै': "ai"
-    'ो': "o"
-    'ौ': "au"
-
-    # Consonants (see Note 1)
-
-    # Gutturals
-    'क': 'ka'
-    'ख': 'kha'
-    'ग': 'ga'
-    'घ': 'gha'
-    'ङ': 'ṅa'
-
-    # Palatals
-    'च': 'cha'
-    'छ': 'chha'
-    'ज': 'ja'
-    'झ': 'jha'
-    'ञ': 'ña'
-
-    # Cerebrals
-    'ट': 'ṭa'
-    'ठ': 'ṭha'
-    'ड': 'ḍa'
-    'ढ': 'ḍha'
-    'ण': 'ṇa'
-
-    # Dentals
-    'त': 'ta'
-    'थ': 'tha'
-    'द': 'da'
-    'ध': 'dha'
-    'न': 'na'
-
-    # Labials
-    'प': 'pa'
-    'फ': 'pha'
-    'ब': 'ba'
-    'भ': 'bha'
-    'म': 'ma'
-
-    # Semivowels
-    'य': 'ya'
-    'र': 'ra'
-    'ल': 'la'
-    'व': 'va' # or wa [Note#3]
-
-    # Sibilants
-    'श': 'sha'
-    'ष': 'ṣha'
-    'स': 'sa'
-
-    # Dotted variants
-    'क़': qa
-    'ख़': ḳha
-    'ग़': ga
-    'ज़': za
-    'ड़': ṙa
-    'ढ़': ṙha
-    'फ़': fa
-
-
-    # Aspirate
-    'ह': 'ha'
-
-    # Anusvāra
-    'ं': 'ṁ'
-
-    # Anunāsika
-    'ँ': 'm̐'
-
-    # halanta
-    '्': ''

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

@@ -1,74 +0,0 @@
----
-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-ori-Orya-Latn-1972.yaml

@@ -1,247 +0,0 @@
----
-authority_id: ungegn
-id: 1972
-language: iso-639-2:ori
-source_script: Orya
-destination_script: Latn
-name: REPORT ON THE CURRENT STATUS OF UNITED NATIONS ROMANIZATION SYSTEMS FOR GEOGRAPHICAL NAMES -- Oriya Romanization, 1972
-url: http://www.eki.ee/wgrs/v2_2/rom1_or.pdf
-creation_date: 1972
-confirmation_date: 2003
-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 reports.
-
-  There is no evidence of the use of the system either in India or in international cartographic
-  products.
-
-  Oriya uses an alphasyllabic script whereby each character represents a syllable rather than one sound.
-  Vowels and diphthongs are marked in two ways: as independent characters (used syllable-initially) and in an
-  abbreviated form, to denote vowels after consonants. The romanization table is unambiguous. The system is mostly
-  reversible but there may exist some ambiguities in the romanization of vowels (independent vs. abbreviated characters)
-  and consonants (combinations with subscript consonants vs. character sequences).
-
-notes:
-  - Combinations with r as the first component are written with a special superscript symbol, e.g. ର୍କ rka.
-
-tests:
-  - source: "ର୍କ"
-    expected: "rka"
-  - source: "ଓଡ଼ିଆ"
-    expected: "oṙiā"
-  - source: "ଓଡ଼ିଶା"
-    expected: "oṙishā"
-  - source: "ଭୁବନେଶ୍ୱର"
-    expected: "bhubaneshvara"
-  - source: "ଆଇପିଏଲ୍‌-୧୩: ଦିଲ୍ଲୀ କ୍ୟାପିଟାଲ୍ସକୁ ୮୮ ରନ୍‌ ପରାସ୍ତ କଲା ସନରାଇଜର୍ସ ହାଇଦ୍ରାବାଦ"
-    expected: "āipiel-13: dillī kyāpiṭālsaku 88 ran parāsta kalā sanarāijarsa hāidrābāda"
-  - source: "ପ୍ରେମ ସମ୍ପର୍କରେ ଭଟ୍ଟା: ରାଗରେ ପ୍ରେମିକାର ତଣ୍ଟି କାଟି ନିଜେ ବିଷ ପିଇଲା ପ୍ରେମିକ"
-    expected: "prema samparkare bhaṭṭā: rāgare premikāra taṇṭi kāṭi nije biṣha piilā premika"
-  - source: "ପ୍ରେମ ସମ୍ପର୍କରେ ଭଟ୍ଟା: ରାଗରେ ପ୍ରେମିକାର ତଣ୍ଟି କାଟି ନିଜେ ବିଷ ପିଇଲା ପ୍ରେମିକ"
-    expected: "prema samparkare bhaṭṭā: rāgare premikāra taṇṭi kāṭi nije biṣha piilā premika"
-  - source: "ହୋଟେଲ, ଲଜ୍‌ରେ ରୁମ୍‌ ମିଳୁନି: ନେତା‌ଙ୍କ ନାଁରେ ଆଗୁଆ ହୋଇଯାଇଛି ବୁକିଂ"
-    expected: "heāṭela, lajre rum miḷuni: netāṅka nāmre āguā heāiỵāichhhi bukiṃ"
-  - source: "ପର୍ଯ୍ୟଟକମାନଙ୍କ ନିମନ୍ତେ ନଭେମ୍ବର ୧ରୁ ଖୋଲିବ ଶିମିଳିପାଳ ଅଭୟାରଣ୍ୟ"
-    expected: "parỵyaṭakamānaṅka nimante nabhembara 1ru kholiba shimiḷipāḷa abhayāraṇya"
-  - source: "ପାରିବାରିକ ଅଶାନ୍ତିର କରୁଣ ପରିଣତି: କୂଅକୁ ଡେଇଁଲେ ମା’-ଝିଅ, ଝିଅ ମୃତ"
-    expected: "pāribārika ashāntira karuṇa pariṇati: kūaku ḍeimle mā’-jhia, jhia mṛta"
-  - source: "‘ଭ୍ରଷ୍ଟାଚାରର ବଂଶବାଦ’ ଏବେ ସାଜିଛି ଦେଶ ପାଇଁ ନୂଆ ସମସ୍ୟା; ପ୍ରଧାନମନ୍ତ୍ରୀ ମୋଦୀ"
-    expected: "‘bhraṣhṭāchārara baṃshabāda’ ebe sājichhhi desha pāim nūā samasyā; pradhānamantrī modī"
-  - source: "ପାହାଡ଼ି ଇଲାକାବାସୀଙ୍କ ଆଶାର ବତୀ ‘ପାର୍ବତୀ’"
-    expected: "pāhāṙi ilākābāsīṅka āshāra batī ‘pārbatī’"
-
-
-map:
-
-  rules:
-    - pattern: ([କ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'k'
-    - pattern: ([ଖ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'kh'
-    - pattern: ([ଗ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'g'
-    - pattern: ([ଘ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'gh'
-    - pattern: ([ଙ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ṅ'
-    - pattern: ([ଚ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ch'
-    - pattern: ([ଛ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'chhh'
-    - pattern: ([ଜ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'j'
-    - pattern: ([ଝ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'jh'
-    - pattern: ([ଞ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ñ'
-    - pattern: ([ଟ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ṭ'
-    - pattern: ([ଠ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ṭh'
-    - pattern: ([ଡ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ḍ'
-    - pattern: ([ଡ଼]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ṙ'
-    - pattern: ([ଢ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ḍh'
-    - pattern: ([ଢ଼]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ṙh'
-    - pattern: ([ଣ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ṇ'
-    - pattern: ([ତ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   't'
-    - pattern: ([ଥ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'th'
-    - pattern: ([ଦ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'd'
-    - pattern: ([ଧ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'dh'
-    - pattern: ([ନ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'n'
-    - pattern: ([ପ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'p'
-    - pattern: ([ଫ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ph'
-    - pattern: ([ବ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'b'
-    - pattern: ([ଭ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'bh'
-    - pattern: ([ମ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'm'
-    - pattern: ([ଯ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ỵ'
-    - pattern: ([ୟ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'y'
-    - pattern: ([ର]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'r'
-    - pattern: ([ଲ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'l'
-    - pattern: ([ଳ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ḷ'
-    - pattern: ([ଶ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'sh'
-    - pattern: ([ଷ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'ṣh'
-    - pattern: ([ସ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   's'
-    - pattern: ([ହ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'h'
-    - pattern: ([କ୍ଷ]=?)(?=[\u0b4d\u0b3e\u0b3f\u0b40\u0b41\u0b42\u0b43\u0b47\u0b48\u0b4b\u0b4c])
-      result:   'kṣh'
-
-  characters:
-    'ଅ': 'a'
-    'ଆ': 'ā'
-    'ଇ': 'i'
-    'ଈ': 'ī'
-    'ଉ': 'u'
-    'ଊ': 'ū'
-    'ଋ': 'ṛ'
-    'ୠ': 'ṝ'
-    'ଌ': 'ḻ'
-    'ଏ': 'e'
-    'ଐ': 'ai'
-    'ଓ': 'o'
-    'ୱ': 'va'
-    'ଔ': 'au'
-
-    # II. Consonants
-    # Gutturals
-    'କ': 'ka'
-    'ଖ': 'kha'
-    'ଗ': 'ga'
-    'ଘ': 'gha'
-    'ଙ': 'ṅa'
-
-    # Palatals
-    'ଚ': 'cha'
-    'ଛ': 'chha'
-    'ଜ': 'ja'
-    'ଝ': 'jha'
-    'ଞ': 'ña'
-
-    # Cerebrals
-    'ଟ':  'ṭa'
-    'ଠ':  'ṭha'
-    'ଡ':  'ḍa'
-    'ଡ଼':  'ṙa'
-    'ଢ':  'ḍha'
-    'ଢ଼':  'ṙha'
-    'ଣ':  'ṇa'
-
-    # Dentals
-    'ତ': 'ta'
-    'ଥ': 'tha'
-    'ଦ': 'da'
-    'ଧ': 'dha'
-    'ନ': 'na'
-
-    # Labials
-    'ପ': 'pa'
-    'ଫ': 'pha'
-    'ବ': 'ba'
-    'ଭ': 'bha'
-    'ମ': 'ma'
-
-    # Semivowels
-    'ଯ': 'ỵa'
-    'ୟ': 'ya'
-    'ର': 'ra'
-    'ଲ': 'la'
-    'ଳ': 'ḷa'
-
-    # Sibilants
-    'ଶ': 'sha'
-    'ଷ': 'ṣha'
-    'ସ': 'sa'
-
-
-    # Aspirate
-    'ହ': 'ha'
-
-    'କ୍ଷ': 'kṣha'
-
-    # Chandrabindu
-    'ଁ': 'm'
-
-    # Bisarga
-    'ଃ': 'ḥ'
-
-    # Anusvāra
-    'ଂ': 'ṃ'
-
-    # Medials # Needed for connecting constants
-
-    'ା': 'ā'
-    'ି': 'i'
-    'ୀ': 'ī'
-    'ୁ': 'u'
-    'ୂ': 'ū'
-    'ୃ': 'ṛ'
-    'େ': 'e'
-    'ୈ': 'ai'
-    'ୋ': 'o'
-    'ୌ': 'au'
-
-    '्': ''
-    '୍': ''
-    '़': ''
-    '଼': ''
-    '।': '.'
-    "‍": ''# Used for joining
-    "‌": ''# Used for non joining
-
-    # Numbers
-
-    '୦': '0'
-    '୧': '1'
-    '୨': '2'
-    '୩': '3'
-    '୪': '4'
-    '୫': '5'
-    '୬': '6'
-    '୭': '7'
-    '୮': '8'
-    '୯': '9'
-