interscript-maps 2.0.5

data/maps/bis-guj-Gujr-Latn-13194-1991.imp

@@ -0,0 +1,172 @@
+metadata {
+  authority_id: bis
+  id: 1991
+  language: iso-639-2:guj
+  source_script: Gujr
+  destination_script: Latn
+  name: Indian script code for information interchange - ISCII - Gujrati Romanization
+
+  #url:
+  creation_date: 1991
+  description: |
+    IS 13194 (1991): Indian script code for information
+    interchange - ISCII [LITD 20: Indian Language Technologies
+    and Products]
+
+  notes:
+    - |
+      Exception: Anusvāra is transliterated by:
+
+      a) ṅ before gutturals,
+      b) ñ before palatals,
+      c) ṇ before cerebrals,
+      d) n before dentals, and
+      e) m before labials.
+}
+
+tests {
+  test "અમિત શાહનો કોરોના રિપોર્ટ 2 ઓગસ્ટે પોઝિટિવ આવ્યો હતો, ત્યારથી તેમનું સ્વાસ્થ્ય સારું નથી", "amit śāhnŏ kŏrŏnā ripŏrṭ 2 ŏgsṭē pŏjhiṭiv āvyŏ htŏ, tyārthī tēmnuṃ svāsthy sāruṃ nthī"
+  test "મેદાંતા હોસ્પિટલમાં તેમનો ઇલાજ ચાલી રહ્યો હતો", "mēdāntā hŏspiṭlmāṃ tēmnŏ ilāj cālī rhyŏ htŏ"
+  test "ભારતના વિશ્વનાથન આનંદે શેનયાનમાં પહેલો ફિડે શતરંજ વિશ્વ કપ જીત્યો", "bhārtnā viśvnāthn ānndē śēnyānmāṃ phēlŏ phiḍē śtrñj viśv kp jītyŏ"
+  test "ભારતીય વડા પ્રધાન જવાહરલાલ નેહરુએ 40 લાખ હિન્દુઓ અને મુસલમાનોના પારસ્પરિક સ્થાનાંતરણનું સૂચન આપ્યું", "bhārtīy vḍā prdhān jvāhrlāl nēhruē 40 lākh hinduŏ anē muslmānŏnā pārsprik sthānāntrṇnuṃ sūcn āpyuṃ"
+  test "લિબિયાના એલ અજિજિયામાં ધરતી પર સૌથી વધુ તાપમાન નોંધાયું. એ વખતે છાયામાં નોંધવામાં આવેલું તાપમાન 58 ડિગ્રી સેલ્સિયસ હતું.", "libiyānā ēl ajijiyāmāṃ dhrtī pr sauthī vdhu tāpmān nŏndhāyuṃ. ē vkhtē chāyāmāṃ nŏndhvāmāṃ āvēluṃ tāpmān 58 ḍigrī sēlsiys htuṃ."
+  test "પ્રથમ વિશ્વયુદ્ધઃ જર્મની અને ફ્રાન્સ વચ્ચે એસ્નેની લડાઈ શરૂ થઈ હતી", "prthm viśvyuddhḥ jrmnī anē phrāns vccē ēsnēnī lḍāī śrū thī htī"
+  test "એન્ગ્લો-મિસ્ત્ર યુદ્ધઃ તેલ અલ કેબિરનું યુદ્ધ લડવામાં આવ્યું હતું.", "ēnglŏ-mistr yuddhḥ tēl al kēbirnuṃ yuddh lḍvāmāṃ āvyuṃ htuṃ."
+  test "પુરાવા ન હતા, એ જ કારણે કેસ ચાલ્યો નહીં, પણ તેમને નજરકેદ રાખવામાં આવ્યા", "purāvā n htā, ē j kārṇē kēs cālyŏ nhīṃ, pṇ tēmnē njrkēd rākhvāmāṃ āvyā"
+  test "સરદાર પટેલે નક્કી કર્યું હતું કે કાશ્મીર ભારતનો હિસ્સો બનશે; 91 વર્ષ પહેલાં લાહોર જેલમાં ભૂખહડતાળ દરમિયાન શહીદ થયા હતા જતીન દાસ", "srdār pṭēlē nkkī kryuṃ htuṃ kē kāśmīr bhārtnŏ hissŏ bnśē; 91 vrṣ phēlāṃ lāhŏr jēlmāṃ bhūkhhḍtāḷ drmiyān śhīd thyā htā jtīn dās"
+  test "કોરોના પ્રોટોકોલ વચ્ચે આજે મેડિકલ પ્રવેશ પરીક્ષા લેવાશેઃ એન્ટ્રી ટચ ફ્રી રહેશે, એડમિટ કાર્ડ બાર કોડથી ચેક થશે", "kŏrŏnā prŏṭŏkŏl vccē ājē mēḍikl prvēś prīkṣā lēvāśēḥ ēnṭrī ṭc phrī rhēśē, ēḍmiṭ kārḍ bār kŏḍthī cēk thśē"
+  test "૮૪૬૬૫૪૧૬૪૬૫૧", "846654164651"
+}
+
+stage {
+
+  # RULES
+  # note
+  sub "\u0A82", "ṅ", after: any("કખગઘઙ")
+  sub "\u0A82", "ñ", after: any("ચછજઝઞ")
+  sub "\u0A82", "ṇ", after: any("ટઠડઢણ")
+  sub "\u0A82", "n", after: any("તથદધન")
+  sub "\u0A82", "m", after: any("પફબભમ")
+
+
+
+  # CHARACTERS
+  parallel {
+    sub "અ", "a"
+    sub "આ", "ā"
+    sub "ઇ", "i"
+    sub "ઈ", "ī"
+    sub "ઉ", "u"
+    sub "ઊ", "ū"
+    # 'ઋ': 'ṛ'
+    sub "ઌ", "ḻ"
+
+    sub "ઍ", "e"
+    sub "એ", "ē"
+    sub "ઐ", "ai"
+
+    sub "ઑ", "o"
+    sub "ઓ", "ŏ"
+    sub "ઔ", "au"
+
+    # II. Consonants (see Note 2)
+    # Gutturals
+    sub "ક", "k"
+    sub "ખ", "kh"
+    sub "ગ", "g"
+    sub "ઘ", "gh"
+    sub "ઙ", "ṅ"
+
+    # Palatals
+    sub "ચ", "c"
+    sub "છ", "ch"
+    sub "જ", "j"
+    sub "ઝ", "jh"
+    sub "ઞ", "ñ"
+
+    # Cerebrals
+    sub "ટ", "ṭ"
+    sub "ઠ", "ṭh"
+    sub "ડ", "ḍ"
+    sub "ઢ", "ḍh"
+    sub "ણ", "ṇ"
+
+    # Dentals
+    sub "ત", "t"
+    sub "થ", "th"
+    sub "દ", "d"
+    sub "ધ", "dh"
+    sub "ન", "n"
+
+    # Labials
+    sub "પ", "p"
+    sub "ફ", "ph"
+    sub "બ", "b"
+    sub "ભ", "bh"
+    sub "મ", "m"
+
+    # Semivowels
+    sub "ય", "y"
+    sub "ર", "r"
+    sub "લ", "l"
+    sub "ળ", "ḷ"
+    sub "વ", "v"
+
+    # Sibilants
+    sub "શ", "ś"
+    sub "ષ", "ṣ"
+    sub "સ", "s"
+
+
+    # Aspirate
+    sub "હ", "h"
+
+    # Chandrabindu
+    sub "ઁ", "m"
+
+    # Bisarga
+    sub "ઃ", "ḥ"
+
+    # Anusvāra
+    sub "ં", "ṃ"
+
+    # Medials # Needed for connecting constants
+
+    sub "ા", "ā"
+    sub "િ", "i"
+    sub "ી", "ī"
+    sub "ુ", "u"
+    sub "ૂ", "ū"
+    sub "ૃ", "ṛ"
+    #'ૄ': 'ṛr'
+    sub "ૅ", "e"
+    sub "ે", "ē"
+    sub "ૈ", "ai"
+    sub "ૉ", "o"
+    sub "ો", "ŏ"
+    sub "ૌ", "au"
+
+
+
+    sub "\u09CD", "" # Used for joining
+    sub "્", ""
+    sub "઼", ""
+    sub "।", "."
+    sub "‍", "" # Used for joining
+
+    # digits
+
+    sub "૦", "0"
+    sub "૧", "1"
+    sub "૨", "2"
+    sub "૩", "3"
+    sub "૪", "4"
+    sub "૫", "5"
+    sub "૬", "6"
+    sub "૭", "7"
+    sub "૮", "8"
+    sub "૯", "9"
+  }
+
+  compose
+}

data/maps/bis-kan-Kana-Latn-13194-1991.imp

@@ -0,0 +1,166 @@
+metadata {
+  authority_id: bis
+  id: 1991
+  language: iso-639-2:kan
+  source_script: Kana
+  destination_script: Latn
+  name: Indian script code for information interchange - ISCII - Kannada Romanization
+
+  #url:
+  creation_date: 1991
+  description: |
+    IS 13194 (1991): Indian script code for information
+    interchange - ISCII [LITD 20: Indian Language Technologies
+    and Products]
+
+  notes:
+    - |
+      Exception: Anusvāra is transliterated by:
+
+      a) ṅ before gutturals,
+      b) ñ before palatals,
+      c) ṇ before cerebrals,
+      d) n before dentals, and
+      e) m before labials.
+}
+
+tests {
+  test "ಈಗ ವೈರಲ್ ಆಗುತ್ತಿದೆ ಕಂಗನಾ ರಣಾವುತ್ ಹಳೇಯ ವಿಡಿಯೋ", "īg vairl āguttide kṅgnā rṇāvut hḷēy viḍiyō"
+  test "ಸಂಕಷ್ಟ ಎದುರಾದರೆ ಬಿಎಸ್‌ವೈ ಬೆನ್ನಿಗೆ ಎಚ್‌ಡಿಕೆ ?", "sṅkṣṭ edurādre biesvai bennige ecḍike ?"
+  test "ಶಾಸಕರಿಂದಲೂ ಒತ್ತಡ?", "śāskrindlū ottḍ?"
+  test "ಏಕೆಂದರೆ, ಇವರ ಹೆಸರೇ ಕೊರೊನಾ!", "ēkendre, ivr hesrē keūronā!"
+  test "ಕೊರೊನಾಕ್ಕಿಂತಲೂ ಮುಂಚೆಯೇ ಅವರು ಕೊರೊನಾ ಆಗಿದ್ದವರು!", "keūronākkintlū muñceyē avru keūronā āgiddvru!"
+  test "ಕೇರಳದ ಕೊಟ್ಟಾಯಂನ ಮಹಿಳೆಯೊಬ್ಬರು ಈಗ ತಮ್ಮ ಹೆಸರು ಹೇಳಲು ಮುಜುಗರ ಪಡುವಂತಾಗಿದೆ", "kērḷd keūṭṭāynn mhiḷeyobbru īg tmm hesru hēḷlu mujugr pḍuvntāgide"
+  test "ಬೇರೆ ಬೆಳವಣಿಗೆಗೆ ಸಾಕ್ಷಿ ಸಾಧ್ಯತೆ", "bēre beḷvṇigege sākṣi sādhyte"
+  test "ಗುರು ಶನಿ ಗ್ರಹಗಳ ನಡುವೆ 3 ಜನರ ಪ್ರಯಾಣ", "guru śni grhgḷ nḍuve 3 jnr pryāṇ"
+  test "ಕೊರೊನಾ ಬಿಕ್ಕಟ್ಟಿನ ಕಾಲದಲ್ಲಿ “ಮಿಸೆಸ್‌ ಕೊರೊನಾ’ಗೆ ಸಮಸ್ಯೆ!", "keūronā bikkṭṭin kāldlli “mises keūronā’ge smsye!"
+  test "ಕೆಲವು ತಿಂಗಳಿಂದ ರಷ್ಯಾ ದೇಶದ ಏನಾಟೊಲಿ ಇವ್ಯಾನಿಶಿನ್‌ ಮತ್ತು ಇವಾನ್‌ ವ್ಯಾಗನರ್‌ ಹಾಗೂ ಅಮೆರಿಕಾದ ಕ್ರಿಸ್‌ ಕ್ಯಾಸಿಡಿ ಈ ಉಪಗ್ರಹದಲ್ಲಿ ವಾಸಿಸುತ್ತಿದ್ದಾರೆ", "kelvu tiṅgḷind rṣyā dēśd ēnāṭeūli ivyāniśin mttu ivān vyāgnr hāgū amerikād kris kyāsiḍi ī upgrhdlli vāsisuttiddāre"
+}
+
+stage {
+
+  # RULES
+  # note
+  sub "\u0C82", "ṅ", after: any("ಕಖಗಘಙ")
+  sub "\u0C82", "ñ", after: any("ಚಛಜಝಞ")
+  sub "\u0C82", "ṇ", after: any("ಟಠಡಢಣ")
+  sub "\u0C82", "n", after: any("ತಥದಧನ")
+  sub "\u0C82", "m", after: any("ಪಫಬಭಮ")
+
+
+  # CHARACTERS
+  parallel {
+    sub "ಅ", "a"
+    sub "ಆ", "ā"
+    sub "ಇ", "i"
+    sub "ಈ", "ī"
+    sub "ಉ", "u"
+    sub "ಊ", "ū"
+    sub "ಋ", "ṛ"
+    sub "ಌ", "ḻ"
+
+    sub "ಎ", "e"
+    sub "ಏ", "ē"
+    sub "ಐ", "ai"
+
+    sub "ಒ", "o"
+    sub "ಓ", "ŏ"
+    sub "ಔ", "au"
+
+    # II. Consonants (see Note 2)
+    # Gutturals
+    sub "ಕ", "k"
+    sub "ಖ", "kh"
+    sub "ಗ", "g"
+    sub "ಘ", "gh"
+    sub "ಙ", "ṅ"
+
+    # Palatals
+    sub "ಚ", "c"
+    sub "ಛ", "ch"
+    sub "ಜ", "j"
+    sub "ಝ", "jh"
+    sub "ಞ", "ñ"
+
+    # Cerebrals
+    sub "ಟ", "ṭ"
+    sub "ಠ", "ṭh"
+    sub "ಡ", "ḍ"
+    sub "ಢ", "ḍh"
+    sub "ಣ", "ṇ"
+
+    # Dentals
+    sub "ತ", "t"
+    sub "ಥ", "th"
+    sub "ದ", "d"
+    sub "ಧ", "dh"
+    sub "ನ", "n"
+
+    # Labials
+    sub "ಪ", "p"
+    sub "ಫ", "ph"
+    sub "ಬ", "b"
+    sub "ಭ", "bh"
+    sub "ಮ", "m"
+
+    # Semivowels
+    sub "ಯ", "y"
+    sub "ರ", "r"
+    sub "ಱ", "ṟ"
+    sub "ಲ", "l"
+    sub "ಳ", "ḷ"
+    #   'ऴ': 'ẕ'
+
+
+    sub "ವ", "v"
+
+    # Sibilants
+    sub "ಶ", "ś"
+    sub "ಷ", "ṣ"
+    sub "ಸ", "s"
+
+
+    # Aspirate
+    sub "ಹ", "h"
+
+
+    # Chandrabindu
+    sub "\u0C80", "m"
+    sub "\u0C81", "m"
+
+    # Bisarga
+    sub "ಃ", "ḥ"
+
+    # Anusvāra
+    sub "ಂ", "ṃ"
+
+    sub "$಼", "" #nukta
+
+    # Medials # Needed for connecting constants
+    sub "ಾ", "ā"
+    sub "ಿ", "i"
+    sub "ೀ", "ī"
+    sub "ು", "u"
+    sub "ೂ", "ū"
+    sub "ೃ", "ṛ"
+    sub "ೄ", "ṛr"
+
+
+    sub "ೆ", "e"
+    sub "ೇ", "ē"
+    sub "ೈ", "ai"
+
+
+    sub "ೊ", "o"
+    sub "ೋ", "ō"
+    sub "ೌ", "au"
+
+    sub "्", ""
+    sub "़", ""
+    sub "್", ""
+    sub "‍", "" # no need for zero with joiner
+    sub "‌", "" # no need for zero with non joiner
+  }
+
+  compose
+}

data/maps/bis-mlm-Mlym-Latn-13194-1991.imp

@@ -0,0 +1,170 @@
+metadata {
+  authority_id: bis
+  id: 1991
+  language: iso-639-2:mlm
+  source_script: Mlym
+  destination_script: Latn
+  name: Indian script code for information interchange - ISCII - Malayalam Romanization
+
+  #url:
+  creation_date: 1991
+  description: |
+    IS 13194 (1991): Indian script code for information
+    interchange - ISCII [LITD 20: Indian Language Technologies
+    and Products]
+
+  notes:
+    - |
+      Exception: Anusvāra is transliterated by:
+
+      a) ṅ before gutturals,
+      b) ñ before palatals,
+      c) ṇ before cerebrals,
+      d) n before dentals, and
+      e) m before labials.
+}
+
+tests {
+  test "സ്വപ്നയ്ക്കൊപ്പം ഹോട്ടലിൽ മന്ത്രിപുത്രൻ, ചിത്രങ്ങൾ; 4 കോടി കമ്മിഷനിലും പങ്കുപറ്റി", "svpnykkeāppṃ hōṭṭlil mntriputrṇ, citrṅṅḷ; 4 kōṭi kmmiṣniluṃ pṅkupṟṟi"
+  test "വിവാദങ്ങളിൽ മാപ്പില്ല, ആദ്യമായി ഐപിഎൽ കമന്ററിക്കില്ലാതെ മ‍ഞ്ജരേക്കര്‍; പുറത്ത് തന്നെ", "vivādṅṅḷil māppill, ādymāyi aipiel kmnṟṟikkillāte mñjrēkkr; puṟtt tnne"
+  test "പരമാവധി ഊറ്റിയെടുത്തു; എല്ലാം കഴിഞ്ഞ് ഉപേക്ഷിച്ചു: വിങ്ങലോടെ റംസിയുടെ സഹോദരി", "prmāvdhi ūṟṟiyeṭuttu; ellāṃ kẕiññ upēkṣiccu: viṅṅlōṭe ṟṃsiyuṭe shōdri"
+  test "വഴിനീളെ രോഷം; യൂത്ത്‌ കോണ്‍ഗ്രസുകാരന്റെ  കയ്യൊടിഞ്ഞു, കൈവീശികാട്ടി ജലീൽ", "vẕinīḷe rōṣṃ; yūtt kōṇgrsukārnṟe  kyyoṭiññu, kaivīśikāṭṭi jlīl"
+  test "‘വികൃതിപ്പയ്യനാ’യിരുന്ന കോലി മിന്നും താരമായത് ഇന്ത്യൻ ക്രിക്കറ്റിന്റെ ഗുണം: അക്തർ", "‘vikṛtippyynā’yirunn kōli minnuṃ tārmāyt intyṇ krikkṟṟinṟe guṇṃ: aktr"
+  test "ലോകത്തിനു വാക്സീൻ വേണമെങ്കിൽ ഈ നഗരം കനിയണം; തലയുയർത്തി ഇന്ത്യ", "lōkttinu vāksīṇ vēṇmeṅkil ī ngrṃ kniyṇṃ; tlyuyrtti inty"
+  test "‘അദ്ദേഹം ഒരു മകളെപ്പോലെ എന്നെ കേട്ടു’: ഗവർണറെ കണ്ട് കങ്കണ റനൗട്ട്", "‘addēhṃ oru mkḷeppōle enne kēṭṭu’: gvrṇṟe kṇṭ kṅkṇ ṟnṭṭ"
+  test "‘എല്ലാം ഫെയ്‌സ്ബുക്കില്‍ പറയുമെന്നു ജലീല്‍; കനത്ത സുരക്ഷയില്‍ യാത്ര, കരിങ്കൊടി", "‘ellāṃ pheysbukkil pṟyumennu jlīl; kntt surkṣyil yātr, kriṅkoṭi"
+  test "ഏറ്റവും ചെറുപ്പം ജോയി; ജയലക്ഷ്മി, ദീപ്തി, ജ്യോതി; പട്ടികയിലെ നിര ഇങ്ങനെ‌", "ēṟṟvuṃ ceṟuppṃ jōyi; jylkṣmi, dīpti, jyōti; pṭṭikyile nir iṅṅne"
+  test "പരിശോധന കുറച്ച് കേരളം; കോവിഡ് ടെസ്റ്റ് പോസിറ്റിവിറ്റി നിരക്ക് എറ്റവും ഉയർന്ന്; ആശങ്ക‌", "priśōdhn kuṟcc kērḷṃ; kōvid̂ ṭesṟṟ pōsiṟṟiviṟṟi nirkk eṟṟvuṃ uyrnn; āśṅk"
+}
+
+stage {
+
+  # RULES
+  # note
+  sub "\u0D02", "ṅ", after: any("കൿഖഗഘങ")
+  sub "\u0D02", "ñ", after: any("ചഛജഝഞ")
+  sub "\u0D02", "ṇ", after: any("ടഠഡഢണ")
+  sub "\u0D02", "n", after: any("തഥദധന")
+  sub "\u0D02", "m", after: any("പഫബഭമ")
+  # Those aren't tested
+
+
+  # CHARACTERS
+  parallel {
+    sub "അ", "a"
+    sub "ആ", "ā"
+    sub "ഇ", "i"
+    sub "ഈ", "ī"
+    sub "ഉ", "u"
+    sub "ഊ", "ū"
+    sub "ഋ", "ṛ"
+    sub "ഌ", "ḻ"
+
+
+    sub "എ", "e"
+    sub "ഏ", "ē"
+    sub "ഐ", "ai"
+
+    sub "ഒ", "o"
+    sub "ഓ", "ŏ"
+    sub "ഔ", "au"
+
+    # II. Consonants (see Note 2)
+    # Gutturals
+    sub "ക", "k"
+    sub "ൿ", "k"
+    sub "ഖ", "kh"
+    sub "ഗ", "g"
+    sub "ഘ", "gh"
+    sub "ങ", "ṅ"
+
+    # Palatals
+    sub "ച", "c"
+    sub "ഛ", "ch"
+    sub "ജ", "j"
+    sub "ഝ", "jh"
+    sub "ഞ", "ñ"
+
+    # Cerebrals
+    sub "ട", "ṭ"
+    sub "ഠ", "ṭh"
+    sub "ഡ", "d̂"
+    sub "ഢ", "ḍh"
+    sub "ണ", "ṇ"
+
+    # Dentals
+    sub "ത", "t"
+    sub "ഥ", "th"
+    sub "ദ", "d"
+    sub "ധ", "dh"
+    sub "ന", "n"
+
+    # Labials
+    sub "പ", "p"
+    sub "ഫ", "ph"
+    sub "ബ", "b"
+    sub "ഭ", "bh"
+    sub "മ", "m"
+
+    # Semivowels
+    sub "യ", "y"
+    sub "ര", "r"
+    sub "ർ", "r"
+    sub "റ", "ṟ"
+    sub "ൽ", "l"
+    sub "ല", "l"
+    sub "ള", "ḷ"
+    sub "ൾ", "ḷ"
+    sub "ഴ", "ẕ"
+    # Sibilants
+    sub "വ", "v"
+    sub "ശ", "ś"
+    sub "ഷ", "ṣ"
+    sub "സ", "s"
+
+
+    # Aspirate
+    sub "ഹ", "h"
+
+    # Chandrabindu
+    sub "\u0D01", "m"
+
+    # Bisarga
+    sub "ഃ", "ḥ"
+
+    # Anusvāra
+    sub "ം", "ṃ"
+
+    # Medials # Needed for connecting constants
+
+    sub "ാ", "ā"
+    sub "ി", "i"
+    sub "ീ", "ī"
+    sub "ു", "u"
+    sub "ൂ", "ū"
+
+    sub "ൃ", "ṛ"
+    sub "ൄ", "ṝ"
+
+
+    sub "െ", "e"
+    sub "േ", "ē"
+    sub "ൈ", "ai"
+    sub "ൊ", "o"
+    sub "ോ", "ō"
+    sub "ൌ", "au"
+
+
+    sub "ൺ", "n"
+    sub "ൻ", "ṇ"
+
+    sub "्", ""
+    sub "്", ""
+    sub "़", ""
+    sub "ൗ", ""
+    sub "‍", "" # no need for zero with joiner
+    sub "‌", "" # no need for zero with non joiner
+  }
+
+  compose
+}