interscript 0.1.7 → 0.1.9

data/maps/un-mal-Mlym-Latn-1972.yaml

@@ -0,0 +1,251 @@
+---
+authority_id: un
+id: 1972
+language: iso-639-2:mal
+source_script: Mlym
+destination_script: Latn
+name: REPORT ON THE CURRENT STATUS OF UNITED NATIONS ROMANIZATION SYSTEMS FOR GEOGRAPHICAL NAMES --Malayalam Romanization Version 4.0
+url: https://www.eki.ee/wgrs/rom1_ml.htm
+creation_date: 1972
+confirmation_date: 2016
+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 reports.
+
+  There is no evidence of the use of the system either in India or in international cartographic products.
+
+  Malayalam (Malayāḷam) 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 but now
+  outdated because since the 1970's a new orthography has been introduced and the use of many character
+  combinations and ligatures has been simplified. Nevertheless the user of the romanization table would have to
+  recognize many ligatures not given in the original table. The system is mostly reversible but there exist some
+  ambiguities in the romanization of vowels (independent vs. abbreviated characters) and
+  consonants (ligatures vs. character sequences), particularly in the case of older orthography.
+
+
+notes:
+  - |
+    ടു For traditional script forms ഖു khu, മു mu, etc., exceptionally കു ku (ക്കു kku, ങ്കു ṅku), ഗു gu, ഛു chhu, ജു ju, ണു ṇu, തു tu, നു nu (ന്നു nnu), ഭു bhu, രു ru, ശു shu, ഹു hu
+  - |
+    ടൂ For traditional script forms ഖൂ khū, മൂ mū, etc., exceptionally കൂ kū, ഗൂ gū, ഛൂ chhū, ജൂ jū, ണൂ ṇū, തൂ tū, നൂ nū, ഭൂ bhū, രൂ rū, ശൂ shū, ഹൂ hū
+  - |
+    ടൃ Traditionally written in conjunction with the consonant character: കൃ kṛ .
+  - |
+    ടൗ In the older spelling: ൌ.
+  - |
+    ട്‌  Indicates absence of the inherent short [a] vowel, e.g. ക്‌ k , സ്‌ s. Special, co-called chillu-forms of consonants with that symbol: ണ്‍ ṇ, ന്‍ n, ര്‍ r, ല്‍ l, ള്‍ ḷ.
+  - |
+    റ  Special ligature: റ്റ ṭ̈.
+
+tests:
+  - source: "ചൈനയ്ക്കെതിരെ ലഡാക്കിൽ സദാസജ്ജം; യുഎസിൽനിന്ന് ൭൨,൫൦൦ സിഗ്–൧൬ റൈഫിൾ"
+    expected: "chainaykkĕtirĕ laḍākkil sadāsajjaṃ; yuĕsilninn 72,500 sig–16 ṟaiphiḷ"
+  - source: "സർഗഭൂമിക’യ്ക്കില്ല; ലളിതച്ചേച്ചി അങ്ങനെ പറഞ്ഞിട്ടുണ്ടാവില്ല: ആർഎൽവി രാമകൃഷ്ണൻ"
+    expected: "sargabhūmika’ykkilla; laḷitachchechchi aṅṅanĕ paṟaññiṭṭuṇṭāvilla: ārĕlvi rāmakṛṣhṇan"
+  - source: "സ്വർണക്കടത്ത്‌: ഫൈസൽ ഫരീദും റബിന്‍സും ദുബായിൽ അറസ്റ്റിലായെന്ന്‌ എന്‍ഐഎ"
+    expected: "svarṇakkaṭatt: phaisal pharīduṃ ṟabinsuṃ dubāyil aṟasṟṟilāyĕnn ĕnaiĕ"
+  - source: "വരുമോ ചൈനയുടെ വാക്സീൻ?; ആഗോള ഉപയോഗത്തിന് ഡബ്ല്യുഎച്ച്ഒയുമായി ചർച്ച"
+    expected: "varumo chainayuṭĕ vāksīn?; āgoḷa upayogattin ḍablyuĕchchŏyumāyi charchcha"
+  - source: "കുട്ടികളുടെ മാനസിക പിരിമുറുക്കം മാറ്റാൻ പരിശീലനം; ക്ലാസുമായി പോക്സോ പ്രതി"
+    expected: "kuṭṭikaḷuṭĕ mānasika pirimuṟukkaṃ māṟṟān parishīlanaṃ; klāsumāyi pokso prati"
+  - source: "ആദ്യം അമിത് ഷാ, ഇപ്പോൾ മോദി; ബിജെപിയെ പുണരാൻ ജഗൻ; ആന്ധ്രയിലെ കരുനീക്കങ്ങൾ"
+    expected: "ādyaṃ amit ṣhā, ippoḷ modi; bijĕpiyĕ puṇarān jagan; āndhrayilĕ karunīkkaṅṅaḷ"
+  - source: "ലഹരിമരുന്ന് കേസ്: ബിനീഷ് കോടിയേരിയെ ഇഡി 6 മണിക്കൂർ ചോദ്യം ചെയ്തു"
+    expected: "laharimarunn kes: binīṣh koṭiyeriyĕ iḍi 6 maṇikkūr chodyaṃ chĕytu"
+  - source: "ഈന്തപ്പഴം വിതരണം ചെയ്തത് ശിവശങ്കര്‍ പറഞ്ഞതു പ്രകാരം: ടി.വി അനുപമയുടെ മൊഴി"
+    expected: "īntappaḻaṃ vitaraṇaṃ chĕytat shivashaṅkar paṟaññatu prakāraṃ: ṭi.vi anupamayuṭĕ mŏḻi"
+  - source: "൫൦൦൦ മണിക്കൂർ കാത്തിരിക്കാൻ തയാറെന്ന് രാഹുൽ: ഒടുവിൽ വഴങ്ങി ഹരിയാന"
+    expected: "5000 maṇikkūr kāttirikkān tayāṟĕnn rāhul: ŏṭuvil vaḻaṅṅi hariyāna"
+  - source: "കാരണം ഷോര്‍ട്ട്‌സര്‍ക്യൂട്ടല്ല; കത്തിയത് ഫയല്‍ മാത്രം, സാനിറ്റൈസര്‍ ഉള്‍പ്പെടെ കത്തിയില്ല"
+    expected: "kāraṇaṃ ṣhorṭṭsarkyūṭṭalla; kattiyat phayal mātraṃ, sāniṟṟaisar uḷppĕṭĕ kattiyilla"
+  - source: "വിമൺ സയൻറിസ്റ്റ്സ് സ്കീം"
+    expected: "vimaṇ sayanṟisṟṟs skīṃ"
+
+map:
+
+  rules:
+    - pattern: ([ക]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'k'
+    - pattern: ([ഖ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'kh'
+    - pattern: ([ഗ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'g'
+    - pattern: ([ഘ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'gh'
+    - pattern: ([ങ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'ṅ'
+    - pattern: ([ച]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'ch'
+    - pattern: ([ഛ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'chh'
+    - pattern: ([ജ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'j'
+    - pattern: ([ഝ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'jh'
+    - pattern: ([ഞ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'ñ'
+    - pattern: ([ട]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result:  'ṭ'
+    - pattern: ([ഠ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result:  'ṭh'
+    - pattern: ([ഡ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'ḍ'
+    - pattern: ([ഢ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'ḍh'
+    - pattern: ([ണ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'ṇ'
+    - pattern: ([ത]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 't'
+    - pattern: ([ഥ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'th'
+    - pattern: ([ദ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'd'
+    - pattern: ([ധ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'dh'
+    - pattern: ([ന]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'n'
+    - pattern: ([പ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'p'
+    - pattern: ([ഫ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'ph'
+    - pattern: ([ബ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'b'
+    - pattern: ([ഭ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'bh'
+    - pattern: ([മ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'm'
+    - pattern: ([യ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'y'
+    - pattern: ([ര]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'r'
+    - pattern: ([ല]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'l'
+    - pattern: ([വ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'v'
+    - pattern: ([ശ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'sh'
+    - pattern: ([ഷ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'ṣh'
+    - pattern: ([സ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 's'
+    - pattern: ([ഹ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'h'
+    - pattern: ([ള]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'ḷ'
+    - pattern: ([ഴ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'ḻ'
+    - pattern: ([റ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'ṟ'
+    - pattern: ([ക്ഷ]=?)(?=[\u0d3e\u0d3f\u0d40\u0d41\u0d42\u0d43\u0d46\u0d47\u0d48\u0d4a\u0d4b\u0d4c\u0d4d])
+      result: 'kṣh'
+
+  characters:
+    # I. Independent vowel characters
+    'അ': 'a'
+    'ആ': 'ā'
+    'ഇ': 'i'
+    'ഈ': 'ī'
+    'ഉ': 'u'
+    'ഊ': 'ū'
+    'ഋ': "ṛ"
+    'എ': 'ĕ'
+    'ഏ': 'e'
+    'ഐ': 'ai'
+    'ഒ': 'ŏ'
+    'ഓ': 'o'
+    'ഔ': 'au'
+
+    # II. Abbreviated vowel characters
+    'ാ': 'ā'
+    'ി': 'i'
+    'ീ': 'ī'
+    'ു': 'u'
+    'ൂ': 'ū'
+    'ൃ': "ṛ"
+    'െ': "ĕ"
+    'േ': "e"
+    'ൈ': "ai"
+    'ൊ': 'ŏ'
+    'ോ': 'o'
+    'ൌ': 'au'
+
+    #III. Other symbols
+    'ഃ': 'ḥ'
+    'ം': 'ṃ'
+    '്': ''
+
+    #IV. Consonant characters
+    'ക': 'ka'
+    'ഖ': 'kha'
+    'ഗ': 'ga'
+    'ഘ': 'gha'
+    'ങ': 'ṅa'
+    'ച': 'cha'
+    'ഛ': 'chha'
+    'ജ': 'ja'
+    'ഝ': 'jha'
+    'ഞ': 'ña'
+    'ട':  'ṭa'
+    'ഠ':  'ṭha'
+    'ഡ': 'ḍa'
+    'ഢ': 'ḍha'
+    'ണ': 'ṇa'
+    'ത': 'ta'
+    'ഥ': 'tha'
+    'ദ': 'da'
+    'ധ': 'dha'
+    'ന': 'na'
+    'പ': 'pa'
+    'ഫ': 'pha'
+    'ബ': 'ba'
+    'ഭ': 'bha'
+    'മ': 'ma'
+    'യ': 'ya'
+    'ര': 'ra'
+    'ല': 'la'
+    'വ': 'va'
+    'ശ': 'sha'
+    'ഷ': 'ṣha'
+    'സ': 'sa'
+    'ഹ': 'ha'
+    'ക്ഷ': 'kṣha'
+    'ള': 'ḷa'
+    'ഴ': 'ḻa'
+    'റ': 'ṟa'
+    'റ്റ':  'ṭ̈'
+
+
+    '्': ''
+    '़': ''
+    'ൗ': ''
+    "‍": '' # no need for zero with joiner
+    "‌": '' # no need for zero with non joiner
+
+
+    # numbers
+    '൦': '0'
+    '൧': '1'
+    '൨': '2'
+    '൩': '3'
+    '൪': '4'
+    '൫': '5'
+    '൬': '6'
+    '൭': '7'
+    '൮': '8'
+    '൯': '9'
+    '൰': '10'
+    '൱': '100'
+    '൲': '1000'
+
+
+    # chillu-forms of consonants note 5
+    'ൿ': 'k'
+    'ൺ': 'ṇ'
+    'ൻ': 'n'
+    'ൽ': 'l'
+    'ൾ': 'ḷ'
+    'ർ': 'r'
+    'സ്': 's'
+

data/maps/un-mar-Deva-Latn-2016.yaml

@@ -31,33 +31,44 @@ notes:
   - |
     The following independent characters are sometimes used: अि i, अी ī, अु u, अू ū, अे e, अै ai.
 
+  ######################################## 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: "महाराष्ट्र"
+    expected: "mahārāṣhṭr"
+  - source: "मुंबई"
+    expected: "muṁbaī"
   - source: "परिपक्क"
-    expected: "paraipakka"
+    expected: "paripakk"
   - source: "ठाणे - जिल्ह्यात बुधवारी एक हजार रुग्णांची वाढ, तर जणांच्या मृत्यूची नोंद"
-    expected: "ṭhaāṇae - jailhayaāta baudhavaāraī eka hajaāra raugṇaāṁchaī vaāḍha, tara jaṇaāṁchyaā maṛtyaūchaī naoṁda"
+    expected: "ṭhāṇe - jilhyāt budhavārī ek hajār rugṇāṁchī vāḍh, tar jaṇāṁchyā mṛtyūchī noṁd"
   - source: "एकता कपूर पुन्हा अडकली वादात, वेबसीरिजमधल्या 'त्या' सीनमुळे जमावाची घरावर दगडफेक"
-    expected: "ekataā kapaūra paunhaā aḍakalaī vaādaāta, vaebasaīraijamadhalyaā 'tyaā' saīnamauḷae jamaāvaāchaī gharaāvara dagaḍaphaeka"
+    expected: "ekatā kapūr punhā aḍakalī vādāt, vebasīrijamadhalyā 'tyā' sīnamuḷae jamāvāchī gharāvar dagaḍaphek"
   - source: "जाणून घ्या, बीएमसीच्या अधिकाऱ्यांनी कंगना राणौतच्या ऑफिसमधले नक्की काय- काय तोडलं"
-    expected: "jaāṇaūna ghyaā, baīemasaīchyaā adhaikaāryaāṁnaī kaṁganaā raāṇaautachyaā ôphaisamadhalae nakkaī kaāya- kaāya taoḍalaṁ"
+    expected: "jāṇūn ghyā, bīemasīchyā adhikāryāṁnī kaṁganā rāṇautachyā ôphisamadhale nakkī kāy- kāy toḍalaṁ"
   - source: "कंगना मुंबईत दाखल होण्यापूर्वी 'मातोश्री'वरून फर्मान सुटले; प्रवक्त्यांना सक्त आदेश"
-    expected: "kaṁganaā mauṁbaīta daākhala haoṇyaāpaūrvaī 'maātaoshraī'varaūna pharmaāna sauṭalae; pravaktyaāṁnaā sakta ādaesha"
+    expected: "kaṁganā muṁbaīt dākhal hoṇyāpūrvī 'mātoshrī'varūn pharmān suṭale; pravaktyāṁnā sakt ādesh"
   - source: "मराठा आरक्षणास तात्पुरती स्थगिती; सर्वोच्च न्यायालयाचा निर्णय"
-    expected: "maraāṭhaā ārakṣhaṇaāsa taātpaurataī sthagaitaī; sarvaochcha nyaāyaālayaāchaā nairṇaya"
+    expected: "marāṭhā ārakṣhaṇās tātpuratī sthagitī; sarvochch nyāyālayāchā nirṇay"
   - source: "भारताच्या तिन्ही लशींचा पहिला टप्पा यशस्वी, वाचा कधी येणार बाजारात"
-    expected: "bhaārataāchyaā tainhaī lashaīṁchaā pahailaā ṭappaā yashasvaī, vaāchaā kadhaī yaeṇaāra baājaāraāta"
+    expected: "bhāratāchyā tinhī lashīṁchā pahilā ṭappā yashasvī, vāchā kadhī yeṇār bājārāt"
   - source: "रुग्णवाढीमुळे खाटांची चणचण"
-    expected: "raugṇavaāḍhaīmauḷae khaāṭaāṁchaī chaṇachaṇa"
+    expected: "rugṇavāḍhīmuḷae khāṭāṁchī chaṇachaṇ"
   - source: "पीएम स्वनिधी कर्ज योजनेला मुंबईतून अल्प प्रतिसाद"
-    expected: "paīema svanaidhaī karja yaojanaelaā mauṁbaītaūna alpa prataisaāda"
+    expected: "pīem svanidhī karj yojanelā muṁbaītūn alp pratisād"
   - source: "सांताक्रूझ-चेंबूर लिंक रोडवरील उन्नत मार्गाला स्थगिती"
-    expected: "saāṁtaākraūjha-chaeṁbaūra laiṁka raoḍavaraīla unnata maārgaālaā sthagaitaī"
+    expected: "sāṁtākrūjh-cheṁbūr liṁk roḍavarīl unnat mārgālā sthagitī"
   - source: "संपादक अर्णब गोस्वामी यांच्याविरूद्ध खडक पोलिस ठाण्यात तक्रार"
-    expected: "saṁpaādaka arṇaba gaosvaāmaī yaāṁchyaāvairaūddha khaḍaka paolaisa ṭhaāṇyaāta takraāra"
+    expected: "saṁpādak arṇab gosvāmī yāṁchyāvirūddh khaḍak polis ṭhāṇyāt takrār"
   - source: "करणाऱ्या मुलांना अनुक्रमे प्ले ग्रूप"
-    expected: "karaṇaāryaā maulaāṁnaā anaukramae plae graūpa"
+    expected: "karaṇāryā mulāṁnā anukrame ple grūp"
   - source: "राज्यातील शाळा दिवाळीपर्यंत बंद, मंत्र्यांच्या बैठकीत निर्णय, शिक्षकांची जबाबदारी वाढली"
-    expected: "raājyaātaīla shaāḷaā daivaāḷaīparyaṁta baṁda, maṁtryaāṁchyaā baaiṭhakaīta nairṇaya, shaikṣhakaāṁchaī jabaābadaāraī vaāḍhalaī"
+    expected: "rājyātīl shāḷaā divāḷaīparyaṁt baṁd, maṁtryāṁchyā baiṭhakīt nirṇay, shikṣhakāṁchī jabābadārī vāḍhalī"
 
 map:
   inherit: "un-hin-Deva-Latn-2016"

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

@@ -86,169 +86,88 @@ tests:
 map:
 
   rules:
-    - pattern: ([क]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - 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]) 
+    - pattern: (([ह]=?)(?=[\u093E\u093F\u0940\u0941\u0942\u0943\u0944\u0945\u0947\u0948\u0949\u094B\u094C\u0020\u094d]))|(([ह])(?=\b))
       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