interscript 0.1.2 → 0.1.7

data/maps/din-hin-Deva-Latn-33904-2018.yaml

@@ -0,0 +1,100 @@
+---
+authority_id: din
+id: 33904-2018
+language: iso-639-2:hin
+source_script: Deva
+destination_script: Latn
+name: "DIN 33904:2018-10 Information and documentation - Romanization Devanagari, Hindi"
+url: https://www.beuth.de/en/standard/din-33904/293049024
+creation_date: 2018
+description: |
+  Hindi system.
+
+  This standard specifies the transliteration of all languages of the Devanāgarī characters into Latin letters firmly.
+  It is suitable for any application.
+
+notes:
+  - |
+    Vowels:
+    The Ardhacandra ॅ is often used to denote English vowels:
+    EXAMPLE
+    बॅट     bêṭa       (bat)
+    डॉक्टर  ḍôkṭara (doctor)
+  - |
+    Consonants:
+
+    Since the Devanāgarī script is a syllabic script, it becomes -a inherent in the consonant.
+    Always written during transliteration, also in the final part
+    EXAMPLE 1
+    पुस्तक   pustaka
+
+    A line under the consonant, the so-called Virāma, means in the Devanāgarī script that that inherent -a not applicable.
+    EXAMPLE 2
+    वाक्   vāk
+
+    The anusvāra is always represented with -ṃ-.
+    EXAMPLE 3
+    संसकरण saṃskaraṇa
+
+    Before k, kh, g, gh, c, ch, j, jh, ṭ, ṭh, ḍ, ḍh, t, th, d, dh, p, ph, b, bh, the anusvāra becomes the class nasal transformed.
+    EXAMPLE 4
+    तंघी   kaṃghī becomes kaṅghī
+    पंजाबी  paṃjābī becomes pañjābī
+    पंडित  paṃḍita becomes paṇḍita
+    संधी   saṃdhī becomes sandhī
+    दिसंबर  disaṃbara becomes disambara
+
+    Anunāsika or Candrabindu signals the nasalization of vowels
+
+    EXAMPLE 5
+    माँ māṁ
+
+tests:
+  - source: "गंभीर मरीजों के मामले में भारत दूसरे नंबर पर"
+    expected: "gambhīra marījoṃ ke māmale meṃ bhārata dūsare nambara para"
+  - source: "कोरोना अपडेट्स"
+    expected: "koronā apaḍeṭsa"
+  - source: "सीडीसी चीफ का बयान अहम"
+    expected: "sīḍīsī cīpha kā bayāna ahama"
+  - source: "गूगल प्ले स्टोर पर पेटीएम की वापसी"
+    expected: "gūgala ple sṭora para peṭīema kī vāpasī"
+  - source: "भारत में गैंबलिंग की इजाजत नहीं"
+    expected: "bhārata meṃ gaimbaliṅga kī ijājata nahīṃ"
+  - source: "कोरोना वैक्सीन मुद्दे पर घिरे राष्ट्रपति; जो बाइडेन बोले- मुझे और देश को वैज्ञानिकों पर भरोसा है, डोनाल्ड ट्रम्प पर नहीं"
+    expected: "koronā vaiksīna mudde para ghire rāṣṭrapati; jo bāiḍena bole- mujhe aura deśa ko vaijñānikoṃ para bharosā hai, ḍonālḍa ṭrampa para nahīṃ"
+  - source: "गूगल की कार्रवाई पर पेटीएम ने कहा था कि ऐप को अस्थायी तौर पर प्ले-स्टोर से हटाया गया है, आपके पैसे सुरक्षित हैं"
+    expected: "gūgala kī kārravāī para peṭīema ne kahā thā ki aipa ko asthāyī taura para ple-sṭora se haṭāyā gayā hai, āpake paise surakṣita haiṃ"
+  - source: "०१९८"
+    expected: "0198"
+
+
+map:
+
+  inherit: din-san-Deva-Latn-33904-2018
+
+  characters:
+
+    'ॲ': 'ê'
+    'ऑ': 'ô'
+
+    # Dotted variants
+    'क़': 'ḵa'
+    'ख़': 'ḵha'
+    'ग़': 'g̲a' # \u0067 + \u0332
+    'ज़': 'j̲a' # \u006a + \u0332
+    'ड़': 'ṙa'
+    'ढ़': 'ṙha'
+    'फ़': 'p̲ha' # \u0070 + \u0332
+    'स़': 's̲a' # \u0073 + \u0332
+    'ह़': 'h̲a' # \u0068 + \u0332
+
+    # Dotted variants Virāma
+    'क़्': 'ḵ'
+    'ख़्': 'ḵh'
+    'ग़्': 'g̲' # \u0067 + \u0332
+    'ज़्': 'j̲' # \u006a + \u0332
+    'ड़्': 'ṙ'
+    'ढ़्': 'ṙh'
+    'फ़्': 'p̲h' # \u0070 + \u0332
+    'स़्': 's̲' # \u0073 + \u0332
+    'ह़्': 'h̲' # \u0068 + \u0332

data/maps/din-kat-Geor-Latn-32707-2010.yaml

@@ -0,0 +1,145 @@
+---
+authority_id: din
+id: 32707-2010
+language: iso-639-2:kat
+source_script: Geor
+destination_script: Latn
+name: "DIN 32707:2010-01 Information und Dokumentation - Umschrift des georgischen Alphabets"
+url: https://www.beuth.de/de/norm/din-32707/122821999
+creation_date: 2010
+confirmation_date: 2010
+description: |
+  Information and documentation - Romanization of the Georgian alphabet.
+
+notes:
+
+tests:
+
+  - source: ხაოფსე
+    expected: xaop̕se
+
+  - source: ჭლოუ
+    expected: člou
+
+  - source: ჩოხულდი
+    expected: č̕oxuldi
+
+  - source: ქვემო ლინდა
+    expected: k̕vemo linda
+
+  - source: ტამკვაჩ იგვავერა
+    expected: tamkvač̕ igvavera
+
+  - source: სვანეთი
+    expected: svanet̕i
+
+  - source: საცხვარისი
+    expected: sac̕xvarisi
+
+  - source: მუხრან-თელეთი
+    expected: muxran-t̕elet̕i
+
+  - source: მუცდი
+    expected: muc̕di
+
+  - source: ლეჩხუმი
+    expected: leč̕xumi
+
+  - source: ვერხნაია მწარა
+    expected: verxnaia mcara
+
+  - source: ეგრისის ქედი
+    expected: egrisis k̕edi
+
+  - source: დოჩარიფშა
+    expected: doč̕arip̕ša
+
+  - source: ბოლოკო
+    expected: boloko
+
+  - source: აჭანდარა
+    expected: ačandara
+
+  - source: აუალიცა
+    expected: aualic̕a
+
+  - source: აკალამრა
+    expected: akalamra
+
+  - source: ლასილი
+    expected: lasili
+
+  - source: გუბაზეული
+    expected: gubazeuli
+
+  - source: ბაყაყი
+    expected: baqaqi
+
+  - source: ძროხა
+    expected: jroxa
+
+  - source: ჰაერი
+    expected: haeri
+
+  - source: ჟოლო
+    expected: žolo
+
+  - source: ჯართი
+    expected: ǰart̕i
+
+  - source: ღრმაღელე
+    expected: ġrmaġele
+
+map:
+  characters:
+    '\u10d0' : 'a' # ა
+    '\u10d1' : 'b' # ბ
+    '\u10d2' : 'g' # გ
+    '\u10d3' : 'd' # დ
+    '\u10d4' : 'e' # ე
+    '\u10d5' : 'v' # ვ
+    '\u10d6' : 'z' # ზ
+
+    '\u10f1' : 'ê' # ჱ
+
+    '\u10d7' : 't̕' # თ
+    '\u10d8' : 'i' # ი
+    '\u10d9' : 'k' # კ
+    '\u10da' : 'l' # ლ
+    '\u10db' : 'm' # მ
+    '\u10dc' : 'n' # ნ
+
+    '\u10f2' : 'y' # ჲ
+
+    '\u10dd' : 'o' # ო
+    '\u10de' : 'p' # პ
+    '\u10df' : 'ž' # ჟ
+    '\u10e0' : 'r' # რ
+    '\u10e1' : 's' # ს
+    '\u10e2' : 't' # ტ
+
+    '\u10f3' : 'w' # ჳ
+
+    '\u10e3' : 'u' # უ
+    '\u10e4' : 'p̕' # ფ
+    '\u10e5' : 'k̕' # ქ
+    '\u10e6' : 'ġ' # ღ
+    '\u10e7' : 'q' # ყ
+    '\u10e8' : 'š' # შ
+    '\u10e9' : 'č̕' # ჩ
+    '\u10ea' : 'c̕' # ც
+    '\u10eb' : 'j' # ძ
+    '\u10ec' : 'c' # წ
+    '\u10ed' : 'č' # ჭ
+    '\u10ee' : 'x' # ხ
+
+    '\u10f4' : 'q̕' # ჴ
+
+    '\u10ef' : 'ǰ' # ჯ
+    '\u10f0' : 'h' # ჰ
+
+    '\u10f5' : 'ô' # ჵ
+
+    '\u10f6' : 'f' # ჶ
+    '\u10f7' : 'ẹ' # ჷ
+    '\u10f8' : 'ʼ' # ჸ

data/maps/din-mar-Deva-Latn-33904-2018.yaml

@@ -0,0 +1,84 @@
+---
+authority_id: din
+id: 33904-2018
+language: iso-639-2:mar
+source_script: Deva
+destination_script: Latn
+name: "DIN 33904:2018-10 Information and documentation - Romanization Devanagari, Marathi"
+url: https://www.beuth.de/en/standard/din-33904/293049024
+creation_date: 2018
+description: |
+  Marathi system.
+
+  Information und Dokumentation - Transliteration Devanagari
+
+  This standard specifies the transliteration of all languages of the Devanāgarī characters into Latin letters firmly.
+  It is suitable for any application.
+
+notes:
+  - |
+    Vowels:
+    The Ardhacandra ॅ is often used to denote English vowels:
+    EXAMPLE
+    बॅट     bêṭa       (bat)
+    डॉक्टर  ḍôkṭara (doctor)
+  - |
+    Consonants:
+
+    Since the Devanāgarī script is a syllabic script, it becomes -a inherent in the consonant.
+    Always written during transliteration, also in the final part
+    EXAMPLE 1
+    पुस्तक   pustaka
+
+    A line under the consonant, the so-called Virāma, means in the Devanāgarī script that that inherent -a not applicable.
+    EXAMPLE 2
+    वाक्   vāk
+
+    The anusvāra is always represented with -ṃ-.
+    EXAMPLE 3
+    संसकरण saṃskaraṇa
+
+    Before k, kh, g, gh, c, ch, j, jh, ṭ, ṭh, ḍ, ḍh, t, th, d, dh, p, ph, b, bh, the anusvāra becomes the class nasal transformed.
+    EXAMPLE 4
+    तंघी   kaṃghī becomes kaṅghī
+    पंजाबी  paṃjābī becomes pañjābī
+    पंडित  paṃḍita becomes paṇḍita
+    संधी   saṃdhī becomes sandhī
+    दिसंबर  disaṃbara becomes disambara
+
+    Anunāsika or Candrabindu signals the nasalization of vowels
+
+    EXAMPLE 5
+    माँ māṁ
+
+tests:
+  - source: "ठाणे - जिल्ह्यात बुधवारी एक हजार रुग्णांची वाढ, तर जणांच्या मृत्यूची नोंद"
+    expected: "ṭhāṇe - jilhyāta budhavārī eka hajāra rugṇāñcī vāḍha, tara jaṇāñcyā mṛtyūcī nonda"
+  - source: "एकता कपूर पुन्हा अडकली वादात, वेबसीरिजमधल्या 'त्या' सीनमुळे जमावाची घरावर दगडफेक"
+    expected: "ekatā kapūra punhā aḍakalī vādāta, vebasīrijamadhalyā 'tyā' sīnamuḷe jamāvācī gharāvara dagaḍapheka"
+  - source: "जाणून घ्या, बीएमसीच्या अधिकाऱ्यांनी कंगना राणौतच्या ऑफिसमधले नक्की काय- काय तोडलं"
+    expected: "jāṇūna ghyā, bīemasīcyā adhikāऱ्yānnī kaṅganā rāṇautacyā ôphisamadhale nakkī kāya- kāya toḍalaṃ"
+  - source: "कंगना मुंबईत दाखल होण्यापूर्वी 'मातोश्री'वरून फर्मान सुटले; प्रवक्त्यांना सक्त आदेश"
+    expected: "kaṅganā mumbaīta dākhala hoṇyāpūrvī 'mātośrī'varūna pharmāna suṭale; pravaktyānnā sakta ādeśa"
+  - source: "मराठा आरक्षणास तात्पुरती स्थगिती; सर्वोच्च न्यायालयाचा निर्णय"
+    expected: "marāṭhā ārakṣaṇāsa tātpuratī sthagitī; sarvocca nyāyālayācā nirṇaya"
+  - source: "भारताच्या तिन्ही लशींचा पहिला टप्पा यशस्वी, वाचा कधी येणार बाजारात"
+    expected: "bhāratācyā tinhī laśīñcā pahilā ṭappā yaśasvī, vācā kadhī yeṇāra bājārāta"
+  - source: "रुग्णवाढीमुळे खाटांची चणचण"
+    expected: "rugṇavāḍhīmuḷe khāṭāñcī caṇacaṇa"
+  - source: "पीएम स्वनिधी कर्ज योजनेला मुंबईतून अल्प प्रतिसाद"
+    expected: "pīema svanidhī karja yojanelā mumbaītūna alpa pratisāda"
+  - source: "सांताक्रूझ-चेंबूर लिंक रोडवरील उन्नत मार्गाला स्थगिती"
+    expected: "sāntākrūjha-cembūra liṅka roḍavarīla unnata mārgālā sthagitī"
+  - source: "संपादक अर्णब गोस्वामी यांच्याविरूद्ध खडक पोलिस ठाण्यात तक्रार"
+    expected: "sampādaka arṇaba gosvāmī yāñcyāvirūddha khaḍaka polisa ṭhāṇyāta takrāra"
+
+
+map:
+
+  inherit: 'din-san-Deva-Latn-33904-2018'
+
+  characters:
+
+    'ॲ': 'ê'
+    'ऑ': 'ô'

data/maps/din-nep-Deva-Latn-33904-2018.yaml

@@ -0,0 +1,119 @@
+---
+authority_id: din
+id: 33904-2018
+language: iso-639-2:nep
+source_script: Deva
+destination_script: Latn
+name: "DIN 33904:2018-10 Information and documentation - Romanization Devanagari, Nepali"
+url: https://www.beuth.de/en/standard/din-33904/293049024
+creation_date: 2018
+description: |
+  Nepali system.
+
+  This standard specifies the transliteration of all languages of the Devanāgarī characters into Latin letters firmly.
+  It is suitable for any application.
+
+notes:
+  - |
+    Vowels:
+    The Ardhacandra ॅ is often used to denote English vowels:
+    EXAMPLE
+    बॅट     bêṭa       (bat)
+    डॉक्टर  ḍôkṭara (doctor)
+  - |
+    Consonants:
+
+    Since the Devanāgarī script is a syllabic script, it becomes -a inherent in the consonant.
+    Always written during transliteration, also in the final part
+    EXAMPLE 1
+    पुस्तक   pustaka
+
+    A line under the consonant, the so-called Virāma, means in the Devanāgarī script that that inherent -a not applicable.
+    EXAMPLE 2
+    वाक्   vāk
+
+    The anusvāra is always represented with -ṃ-.
+    EXAMPLE 3
+    संसकरण saṃskaraṇa
+
+    Before k, kh, g, gh, c, ch, j, jh, ṭ, ṭh, ḍ, ḍh, t, th, d, dh, p, ph, b, bh, the anusvāra becomes the class nasal transformed.
+    EXAMPLE 4
+    तंघी   kaṃghī becomes kaṅghī
+    पंजाबी  paṃjābī becomes pañjābī
+    पंडित  paṃḍita becomes paṇḍita
+    संधी   saṃdhī becomes sandhī
+    दिसंबर  disaṃbara becomes disambara
+
+    Anunāsika or Candrabindu signals the nasalization of vowels
+
+    EXAMPLE 5
+    माँ māṁ
+
+tests:
+  - source: "लेखन"
+    expected: "lekhana"
+  - source: "मुद्रा"
+    expected: "mudrā"
+  - source: "प्रशंसा"
+    expected: "praśaṃsā"
+  - source: "अंक"
+    expected: "aṅka"
+  - source: "नेकपाले स्थगित स्थायी कमिटीको बैठक भदौ गते बोलाउने भएको"
+    expected: "nekapāle sthagita sthāyī kamiṭīko baiṭhaka bhadau gate bolāune bhaeko"
+  - source: "न घर रह्यो, न परिवार"
+    expected: "na ghara rahyo, na parivāra"
+  - source: "ढोरपाटनमा भुजीखोला बाढीपहिरोले अभिभावक गुमाएका बालबालिकाको बिचल्ली"
+    expected: "ḍhorapāṭanamā bhujīkholā bāḍhīpahirole abhibhāvaka gumāekā bālabālikāko bicallī"
+  - source: "सुस्मिताका काका हेमबहादुर र काकीलाई पनि पहिरोले बगायो"
+    expected: "susmitākā kākā hemabahādura ra kākīlāī pani pahirole bagāyo"
+  - source: "संविधान जारी भएसँगै सार्वजनिक प्रशासनमा नयाँ उत्साह आउने अपेक्षा थियो"
+    expected: "saṃvidhāna jārī bhaesaṁgai sārvajanika praśāsanamā nayāṁ utsāha āune apekṣā thiyo"
+  - source: "देशमा कोरोना संक्रमित र मृतकको संख्या हरेक दिन बढ्दो छ"
+    expected: "deśamā koronā saṅkramita ra mṛtakako saṅkhyā hareka dina baḍhdo cha"
+  - source: "गाउँपालिकाका अध्यक्ष टिका गुरुङका अनुसार विष्णुदासलाई राजुले सुत्नका लागि बेलुका साथी लगेका थिए"
+    expected: "gāuṁpālikākā adhyakṣa ṭikā guruṅakā anusāra viṣṇudāsalāī rājule sutnakā lāgi belukā sāthī lagekā thie"
+  - source: "यो आयोजना गाउँपालिकाको केन्द्र तेल्लोकमा पर्छ"
+    expected: "yo āyojanā gāuṁpālikāko kendra tellokamā parcha"
+  - source: "सुस्मिताका काका हेमबहादुर र काकीलाई पनि पहिरोले बगायो"
+    expected: "susmitākā kākā hemabahādura ra kākīlāī pani pahirole bagāyo"
+  - source: "चैत पहिलो साता घर आएका उनी लकडाउन भएपछि यतै रोकिए"
+    expected: "caita pahilo sātā ghara āekā unī lakaḍāuna bhaepachi yatai rokie"
+  - source: "काम गर्न जानेको हकमा रोजगारदाता कम्पनीको पत्रसँगै वडा र जिल्ला प्रशासनको सिफारिस अनिवार्य गरिएको छ"
+    expected: "kāma garna jāneko hakamā rojagāradātā kampanīko patrasaṁgai vaḍā ra jillā praśāsanako siphārisa anivārya garieko cha"
+  - source: "दुःख"
+    expected: "duḥkha"
+
+
+map:
+
+  inherit: 'din-san-Deva-Latn-33904-2018'
+
+  characters:
+
+    'ॲ': 'ê'
+    'ऑ': 'ô'
+
+    # Dotted variants
+    'क़': 'ḵa'
+    'ख़': 'ḵha'
+    'ग़': 'g̲a' # \u0067 + \u0332
+    'ज़': 'j̲a' # \u006a + \u0332
+    'ड़': 'ṙa'
+    'ढ़': 'ṙha'
+    'फ़': 'p̲ha' # \u0070 + \u0332
+    'स़': 's̲a' # \u0073 + \u0332
+    'ह़': 'h̲a' # \u0068 + \u0332
+    'व़': 'v̲a' # \u0076 + \u0332
+
+    # Dotted variants Virāma
+    'क़्': 'ḵ'
+    'ख़्': 'ḵh'
+    'ग़्': 'g̲' # \u0067 + \u0332
+    'ज़्': 'j̲' # \u006a + \u0332
+    'ड़्': 'ṙ'
+    'ढ़्': 'ṙh'
+    'फ़्': 'p̲h' # \u0070 + \u0332
+    'स़्': 's̲' # \u0073 + \u0332
+    'ह़्': 'h̲' # \u0068 + \u0332
+
+