interscript 0.1.2 → 0.1.7

data/maps/din-tam-Taml-Latn-33903-2016.yaml

@@ -0,0 +1,213 @@
+---
+authority_id: din
+id: 33903-2016
+language: iso-639-2:tam
+source_script: Taml
+destination_script: Latn
+name: "DIN 33903:2016-02: Information und Dokumentation - Umschrift Tamil"
+url: https://www.beuth.de/de/norm/din-33903/247150527
+creation_date: 2018
+description: |
+  This standard specifies the transcription of the letters of the
+  Tamil alphabet in Latin letters. It is based on international
+  standards as well as on the preprint of the alphabetical tables,
+  which were published in 1976 with the introduction of the rules for
+  alphabetical cataloging. It can be used in all areas, but in
+  particular the aim is to standardize the library and publication
+  area.
+
+notes:
+  - |
+    Since the Tamil script is a syllabic script, the consonant -a inherent in the transliteration is always written, even in the final.
+
+  - |
+    A point above the consonant, the so-called "pulli", means in Tamil that the inherent -a is omitted:
+      கார்த்திகேயன் kārttikēyaṉ
+  - |
+    Tamil has no anusvara; in contrast to other South Asian scripts, the class nasal is always written.
+  - |
+    Tamil also has no ligatures.
+
+tests:
+  - source: "கார்த்திகேயன்"
+    expected: "kārttikēyaṉ"
+  - source: "௲"
+    expected: "1000"
+  - source: "இளைஞர்களின் உறுதியான மனநிலையை பிரதிபலிக்கிறது: நீட் தேர்வில் ௮௫-௯௦ சதவீத மாணவர்கள் பங்கேற்பு - ரமேஷ் பொக்ரியால்"
+    expected: "iḷaiñarkaḷiṉ uṟutiyāṉa maṉanilaiyai piratipalikkiṟatu: nīṭ tērvil 85-90 catavīta māṇavarkaḷ paṅkēṟpu - ramēṣ pokriyāl"
+  - source: "நாடாளுமன்றத்தில் 4 மசோதாக்களை எதிர்க்க காங்கிரஸ் முடிவு - ஜெயராம் ரமேஷ்"
+    expected: "nāṭāḷumaṉṟattil 4 macōtākkaḷai etirkka kāṅkiras muṭivu - jeyarām ramēṣ"
+  - source: "கர்நாடகாவில் மேலும் 9,894 பேருக்கு கொரோனா தொற்று உறுதி"
+    expected: "karnāṭakāvil mēlum 9,894 pērukku korōṉā toṟṟu uṟuti"
+  - source: "ஐதராபாத்துக்கு கைகொடுக்குமா அதிரடி?"
+    expected: "aitarāpāttukku kaikoṭukkumā atiraṭi?"
+  - source: "அமெரிக்க ஓபன் டென்னிஸ்: இறுதிப்போட்டியில் டொமினிக்-ஸ்வெரேவ்"
+    expected: "amerikka ōpaṉ ṭeṉṉis: iṟutippōṭṭiyil ṭomiṉik-sverēv"
+  - source: "ஐ.பி.எல். கிரிக்கெட்டில் களம் இறங்கும் அமெரிக்க வீரர்"
+    expected: "ai.pi.el. kirikkeṭṭil kaḷam iṟaṅkum amerikka vīrar"
+  - source: "அமெரிக்க ஓபன் டென்னிஸ்; நவோமி ஒசாகா சாம்பியன் பட்டம் வென்றார்"
+    expected: "amerikka ōpaṉ ṭeṉṉis; navōmi ocākā cāmpiyaṉ paṭṭam veṉṟār"
+  - source: "புதிய கல்விக்கொள்கைக்கு எதிர்ப்பு: முன்னாள் துணைவேந்தர்கள் 20 பேர் பிரதமருக்கு கடிதம்"
+    expected: "putiya kalvikkoḷkaikku etirppu: muṉṉāḷ tuṇaivēntarkaḷ 20 pēr piratamarukku kaṭitam"
+  - source: "இந்த ஆண்டு ஐ.பி.எல். கோப்பையை எந்த அணி வெல்லும்? - கெவின் பீட்டர்சன் கணிப்பு"
+    expected: "inta āṇṭu ai.pi.el. kōppaiyai enta aṇi vellum? - keviṉ pīṭṭarcaṉ kaṇippu"
+  - source: "இந்திய எண்ணெய் கப்பலில் தீ: விபத்து குறித்த எச்சரிக்கையை கப்பல் அதிகாரிகள் புறக்கணித்தனர் - இலங்கை கோர்ட்டு தகவல்"
+    expected: "intiya eṇṇey kappalil tī: vipattu kuṟitta eccarikkaiyai kappal atikārikaḷ puṟakkaṇittaṉar - ilaṅkai kōrṭṭu takaval"
+
+map:
+
+  rules:
+    - pattern: ([க]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'k'
+    - pattern: ([ங]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'ṅ'
+    - pattern: ([ச]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'c'
+    - pattern: ([ஞ]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'ñ'
+    - pattern: ([ட]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'ṭ'
+    - pattern: ([ண]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'ṇ'
+    - pattern: ([த]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 't'
+    - pattern: ([ந]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'n'
+    - pattern: ([ப]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'p'
+    - pattern: ([ம]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'm'
+    - pattern: ([ய]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'y'
+    - pattern: ([ர]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'r'
+    - pattern: ([ல]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'l'
+    - pattern: ([ள]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'ḷ'
+    - pattern: ([ழ]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'ḻ'
+    - pattern: ([வ]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'v'
+    - pattern: ([ற]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'ṟ'
+    - pattern: ([ன]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'ṉ'
+    - pattern: ([ஜ]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'j'
+    - pattern: ([ஶ]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'ś'
+    - pattern: ([ஷ]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'ṣ'
+    - pattern: ([ஸ]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 's'
+    - pattern: ([ஹ]=?)(?=[\u0BBE\u0BBF\u0BC0\u0BC1\u0BC2\u0BC6\u0BC7\u0BC8\u0BCA\u0BCB\u0BCC\u094B\u094C])
+      result: 'h'
+
+  characters:
+    'அ': 'a'
+    'ஆ': 'ā'
+    'ா': 'ā'
+
+    'இ': 'i'
+    'ி': 'i'
+
+    'ஈ': 'ī'
+    'ீ': 'ī'
+
+    'உ': 'u'
+    'ு': 'u'
+
+    'ஊ': 'ū'
+    'ூ': 'ū'
+
+    'ெ': "e"
+    'எ': 'e'
+
+    'ே': "ē"
+    'ஏ': 'ē'
+
+    'ஐ': 'ai'
+    'ை': "ai"
+
+    'ஒ': 'o'
+    'ொ': 'o'
+
+    'ோ': 'ō'
+    'ஓ': 'ō'
+
+    'ஔ': 'au'
+    'ௌ': 'au'
+
+    'க்': 'k'
+    'ங்': 'ṅ'
+    'ச்': 'c'
+    'ஞ்': 'ñ'
+    'ட்':  'ṭ'
+    'ண்':  'ṇ'
+    'த்': 't'
+    'ந்':  'n'
+    'ப்': 'p'
+    'ம்': 'm'
+    'ய்': 'y'
+    'ர்': 'r'
+    'ல்': 'l'
+    'ள்': 'ḷ'
+    'ழ்': 'ḻ'
+    'வ்': 'v'
+    'ற்': 'ṟ'
+    'ன்': 'ṉ'
+    'ஜ்': 'j'
+
+    'ஶ்': 'ś'
+    'ஷ்': 'ṣ'
+    'ஸ்': 's'
+    'ஹ்': 'h'
+
+
+    'க': 'ka'
+    'ங': 'ṅa'
+    'ஃ': 'ḵ'
+    'ச': 'ca'
+    'ஞ': 'ña'
+    'ட': 'ṭa'
+    'ண': 'ṇa'
+    'த': 'ta'
+    'ந': 'na'
+    'ப': 'pa'
+    'ம': 'ma'
+    'ய': 'ya'
+    'ர': 'ra'
+    'ல': 'la'
+    'ள': 'ḷa'
+    'ழ': 'ḻa'
+    'வ': 'va'
+    'ற': 'ṟa'
+    'ன': 'ṉa'
+    'ஜ': 'ja'
+    'ஶ': 'śa'
+    'ஷ': 'ṣa'
+    'ஸ': 'sa'
+    'ஹ': 'ha'
+
+    # Digits
+    '௧': '1'
+    '௨': '2'
+    '௩': '3'
+    '௪': '4'
+    '௫': '5'
+    '௬': '6'
+    '௭': '7'
+    '௮': '8'
+    '௯': '9'
+    '௦': '0'
+    '௰': '10'
+    '௱': '100'
+    '௲': '1000'
+
+
+
+
+    "‍": '' # no need for zero with joiner
+    "‌": '' # no need for zero with non joiner
+

data/maps/dos-nep-Deva-Latn-1997.yaml

@@ -0,0 +1,47 @@
+---
+authority_id: bgnpcgn
+id: 1997
+language: iso-639-2:nep
+source_script: Deva
+destination_script: Latn
+name: Nepali Romanization, 1997
+url: http://nationalgeoportal.gov.np/old/pdf/translation2.pdf
+creation_date: 1997
+description: |
+  Survey Department, Ministry of Land Management, Cooperatives and Poverty Alleviation, Government of Nepal.
+
+notes:
+  - |
+    ं (anusvara) is rendered by
+    ṅ before क, ख, ग, and घ
+    ñ before च, छ, ज, and झ
+    ṇ before ट, ठ, ड, and ढ
+    n before त, थ, द, and ध
+    ṁ before य, र, ल, व, श, ष, स
+
+tests:
+  - source: "दुःख"
+    expected: "duhkh"
+  - source: "पूरा भइसकेका विषयलाई माग बनाएर दबाब नदिनुस्"
+    expected: "pūrā bhiskekā viṣylāī māg bnāer dbāb ndinusa"
+  - source: "जाँदै छ कता नेपाली संगीत"
+    expected: "jā~dai chh ktā nepālī sṅgīt"
+  - source: "३५ मिनेटको यो डकुमेन्ट्री फिल्मले प्रथम पुरस्कारस्वरूप ग्रान्ड पिक्स अवार्ड पाएको हो"
+    expected: "35 mineṭko yo ḍkumenaṭarī philamle parthm pursakārsavrūp garānaḍ pikas avāraḍ pāeko ho"
+  - source: "विक्षनरी"
+    expected: "vikṣnrī"
+  - source: "रुसमा उत्कृष्ट"
+    expected: "rusmā utakṛiṣaṭ"
+  - source: "वाणिज्य"
+    expected: "vāṇijay"
+  - source: "अंक विद्या"
+    expected: "aṅk vidayā"
+
+map:
+
+  inherit: "bgnpcgn-nep-Deva-Latn-2011"
+
+  characters:
+
+    # Bisarga
+    'ः': 'h'

data/maps/elot-ell-Grek-Latn-743-1982-tl.yaml

@@ -0,0 +1,684 @@
+---
+authority_id: elot
+id: 1982-tl
+language: iso-639-2:ell
+source_script: Grek
+destination_script: Latn
+name: ELOT 743:1982 (transliteration)
+url:
+creation_date: 1982
+description: |
+  ELOT 743:1982 transliteration table for Greek
+
+note:
+  - This was the intended system in ELOT 743:1982 for transliteration, as is apparent from un-ell-Grek-Latn-1987-tl and iso-ell-Grek-Latn-843-1997-t2, and ELOT 743:2001, Annex B; but the reversible character diacritics are missing from the currently available document, and the document does not indicate what there was supposed to be.
+  - "\"Transliteration\" as used in this specification and its inheritors refers to reversible transliteration, as distinct from \"transcription\", which is a simpler transliteration and which is intended for default use."
+
+tests:
+  - source: |
+        Ένα πράμα μόνον με παρακίνησε κι εμένα να γράψω ότι τούτην την πατρίδα την έχομεν όλοι μαζί, και σοφοί κι αμαθείς και πλούσιοι και φτωχοί και πολιτικοί και στρατιωτικοί και οι πλέον μικρότεροι άνθρωποι· όσοι αγωνιστήκαμεν, αναλόγως ο καθείς, έχομεν να ζήσομεν εδώ. Το λοιπόν δουλέψαμεν όλοι μαζί, να την φυλάμεν κι όλοι μαζί και να μην λέγει ούτε ο δυνατός «εγώ» ούτε ο αδύνατος. Ξέρετε πότε να λέγει ο καθείς «εγώ»; Όταν αγωνιστεί μόνος του και φκιάσει ή χαλάσει, να λέγει «εγώ»· όταν όμως αγωνίζονται πολλοί και φκιάνουν, τότε να λένε «εμείς». Είμαστε εις το «εμείς» κι όχι εις το «εγώ». Και εις το εξής να μάθομεν γνώση, αν θέλομεν να φκιάσομεν χωριόν, να ζήσομεν όλοι μαζί.
+
+        Γιάννης Μακρυγιάννης.
+
+    expected: |
+        Éna práma mónon me parakíni̱se ki eména na grápso̱ óti toúti̱n ti̱n patrída ti̱n échomen óloi mazí, kai sofoí ki amatheís kai ploúsioi kai fto̱choí kai politikoí kai stratio̱tikoí kai oi pléon mikróteroi ánthro̱poi; ósoi ago̱nistí̱kamen, analógo̱s o katheís, échomen na zí̱somen edó̱. To loipón doulépsamen óloi mazí, na ti̱n fylámen ki óloi mazí kai na mi̱n légei oúte o dynatós «egó̱» oúte o adýnatos. Xérete póte na légei o katheís «egó̱»? Ótan ago̱nisteí mónos tou kai fkiásei í̱ chalásei, na légei «egó̱»; ótan ómo̱s ago̱nízontai polloí kai fkiánoun, tóte na léne «emeís». Eímaste eis to «emeís» ki óchi eis to «egó̱». Kai eis to exí̱s na máthomen gnó̱si̱, an thélomen na fkiásomen cho̱rión, na zí̱somen óloi mazí.
+
+        Giánni̱s Makrygiánni̱s.
+
+
+  - source: ΑΘΗΝΑ
+    expected: ATHI̱NA
+  - source: μπαμπάκι
+    expected: bampáki
+  - source: νταντά
+    expected: ntantá
+  - source: γκέγκε
+    expected: gkégke
+  - source: Γκαμπόν
+    expected: Gkampón
+  - source: Μάγχη
+    expected: Máṉchi̱
+  - source: κογξ
+    expected: koṉx
+  - source: υιός
+    expected: yiós
+  - source: Υιός
+    expected: Yiós
+  - source: νεράντζι
+    expected: nerántzi
+  - source: Γοίθιος
+    expected: Goíthios
+  - source: μπέικον
+    expected: béïkon
+  - source: μπέϊκον
+    expected: béïkon
+  - source: βόλεϊ
+    expected: vóleï
+  - source: αθεΐα
+    expected: atheḯa
+  - source: Εϊγιαφιάτλαγιοκουτλ
+    expected: Eïgiafiátlagiokoutl
+  - source: Εΐτζι
+    expected: Eḯtzi
+  - source: Μυρτώο
+    expected: Myrtó̱o
+  - source: αέρας
+    expected: aéras
+  - source: γαυ γαυ
+    expected: gaf̱ gaf̱
+  - source: Ταΰγετος
+    expected: Taÿ́getos
+  - source: σπρέυ
+    expected: spréy
+
+  - source: Αθήνα
+    expected: Athí̱na
+  - source: Άγιον Όρος
+    expected: Ágion Óros
+  - source: Άγραφα
+    expected: Ágrafa
+  - source: Αγρίνιο
+    expected: Agrínio
+  - source: Αίγινα
+    expected: Aígina
+  - source: Αίγιο
+    expected: Aígio
+  - source: Αλεξανδρούπολη
+    expected: Alexandroúpoli̱
+  - source: Αλεποχώρι
+    expected: Alepochó̱ri
+  - source: Αμοργός
+    expected: Amorgós
+  - source: Άμφισσα
+    expected: Ámfissa
+  - source: Αράχωβα
+    expected: Arácho̱va
+  - source: Άργος
+    expected: Árgos
+  - source: Αρκαδία
+    expected: Arkadía
+  - source: Άρτα
+    expected: Árta
+  - source: Βελούχι
+    expected: Veloúchi
+  - source: Βέροια
+    expected: Véroia
+  - source: Βοιωτία
+    expected: Voio̱tía
+  - source: Βόλος
+    expected: Vólos
+  - source: Βόνιτσα
+    expected: Vónitsa
+  - source: Γαλαξίδι
+    expected: Galaxídi
+  - source: Γαλάτσι
+    expected: Galátsi
+  - source: Γιαννιτσά
+    expected: Giannitsá
+  - source: Γλυφάδα
+    expected: Glyfáda
+  - source: Γρανίτσα
+    expected: Granítsa
+  - source: Γρεβενά
+    expected: Grevená
+  - source: Γύθειο
+    expected: Gýtheio
+  - source: Διόνυσος
+    expected: Diónysos
+  - source: Δίστομο
+    expected: Dístomo
+  - source: Δολιανά
+    expected: Dolianá
+  - source: Δράμα
+    expected: Dráma
+  - source: Δωδεκάνησα
+    expected: Do̱dekáni̱sa
+  - source: Έδεσσα
+    expected: Édessa
+  - source: Ελευσίνα
+    expected: Elef̱sína
+  - source: Επίδαυρος
+    expected: Epídav̱ros
+  - source: Επτάνησα
+    expected: Eptáni̱sa
+  - source: Ερμούπολη
+    expected: Ermoúpoli̱
+  - source: Εύβοια
+    expected: Év̱voia
+  - source: Ζάκυνθος
+    expected: Zákynthos
+  - source: Ήπειρος
+    expected: Í̱peiros
+  - source: Ηράκλειο
+    expected: I̱rákleio
+  - source: Θάσος
+    expected: Thásos
+  - source: Θεσσαλονίκη
+    expected: Thessaloníki̱
+  - source: Θεσσαλία
+    expected: Thessalía
+  - source: Θεσπρωτία
+    expected: Thespro̱tía
+  - source: Θήβα
+    expected: Thí̱va
+  - source: Θράκη
+    expected: Thráki̱
+  - source: Ιθάκη
+    expected: Itháki̱
+  - source: Ίος
+    expected: Íos
+  - source: Ιωάννινα
+    expected: Io̱ánnina
+  - source: Καβάλα
+    expected: Kavála
+  - source: Καλάβρυτα
+    expected: Kalávryta
+  - source: Καλαμάτα
+    expected: Kalamáta
+  - source: Καλαμπάκα
+    expected: Kalampáka
+  - source: Καλύβια
+    expected: Kalývia
+  - source: Κάλυμνος
+    expected: Kálymnos
+  - source: Καρδίτσα
+    expected: Kardítsa
+  - source: Καρπενήσι
+    expected: Karpení̱si
+  - source: Κάρυστος
+    expected: Kárystos
+  - source: Καστελλόριζο
+    expected: Kastellórizo
+  - source: Καστοριά
+    expected: Kastoriá
+  - source: Κατερίνη
+    expected: Kateríni̱
+  - source: Κάτω Αχαΐα
+    expected: Káto̱ Achaḯa
+  - source: Κερατέα
+    expected: Keratéa
+  - source: Κέρκυρα
+    expected: Kérkyra
+  - source: Κεφαλλονιά
+    expected: Kefalloniá
+  - source: Κηφισιά
+    expected: Ki̱fisiá
+  - source: Κιλκίς
+    expected: Kilkís
+  - source: Κοζάνη
+    expected: Kozáni̱
+  - source: Κολωνός
+    expected: Kolo̱nós
+  - source: Κομοτηνή
+    expected: Komoti̱ní̱
+  - source: Κόρινθος
+    expected: Kórinthos
+  - source: Κορώνη
+    expected: Koró̱ni̱
+  - source: Κρανίδι
+    expected: Kranídi
+  - source: Κρέστενα
+    expected: Kréstena
+  - source: Κρήτη
+    expected: Krí̱ti̱
+  - source: Κύθηρα
+    expected: Kýthi̱ra
+  - source: Κυκλάδες
+    expected: Kykládes
+  - source: Κύμη
+    expected: Kými̱
+  - source: Κυψέλη
+    expected: Kypséli̱
+  - source: Κως
+    expected: Ko̱s
+  - source: Λαγκαδάς
+    expected: Lagkadás
+  - source: Λαμία
+    expected: Lamía
+  - source: Λάρισα
+    expected: Lárisa
+  - source: Λαύριο
+    expected: Láv̱rio
+  - source: Λέρος
+    expected: Léros
+  - source: Λέσβος
+    expected: Lésvos
+  - source: Λευκάδα
+    expected: Lef̱káda
+  - source: Λήμνος
+    expected: Lí̱mnos
+  - source: Λιβαδειά
+    expected: Livadeiá
+  - source: Μακεδονία
+    expected: Makedonía
+  - source: Μάνη
+    expected: Máni̱
+  - source: Μαραθώνας
+    expected: Marathó̱nas
+  - source: Μαρκόπουλο
+    expected: Markópoulo
+  - source: Μαρούσι
+    expected: Maroúsi
+  - source: Μέγαρα
+    expected: Mégara
+  - source: Μεσολόγγι
+    expected: Mesolóṉgi
+  - source: Μεταξουργείο
+    expected: Metaxourgeío
+  - source: Μέτσοβο
+    expected: Métsovo
+  - source: Μήλος
+    expected: Mí̱los
+  - source: Μύκονος
+    expected: Mýkonos
+  - source: Μυστράς
+    expected: Mystrás
+  - source: Μυτιλήνη
+    expected: Mytilí̱ni̱
+  - source: Νάξος
+    expected: Náxos
+  - source: Νάουσα
+    expected: Náousa
+  - source: Ναύπακτος
+    expected: Náf̱paktos
+  - source: Ναύπλιο
+    expected: Náf̱plio
+  - source: Νέα Σμύρνη
+    expected: Néa Smýrni̱
+  - source: Νίσυρος
+    expected: Nísyros
+  - source: Ξάνθη
+    expected: Xánthi̱
+  - source: Όλυμπος
+    expected: Ólympos
+  - source: Παγκράτι
+    expected: Pagkráti
+  - source: Παπάγου
+    expected: Papágou
+  - source: Πάρος
+    expected: Páros
+  - source: Πασαλιμάνι
+    expected: Pasalimáni
+  - source: Πατήσια
+    expected: Patí̱sia
+  - source: Πάτμος
+    expected: Pátmos
+  - source: Πάτρα
+    expected: Pátra
+  - source: Πειραιάς
+    expected: Peiraiás
+  - source: Πελοπόννησος
+    expected: Pelopónni̱sos
+  - source: Περιστέρι
+    expected: Peristéri
+  - source: Πεύκη
+    expected: Péf̱ki̱
+  - source: Πήλιο
+    expected: Pí̱lio
+  - source: Πολύγυρος
+    expected: Polýgyros
+  - source: Πόρος
+    expected: Póros
+  - source: Πρέβεζα
+    expected: Préveza
+  - source: Πτολεμαΐδα
+    expected: Ptolemaḯda
+  - source: Πύλος
+    expected: Pýlos
+  - source: Πύργος
+    expected: Pýrgos
+  - source: Ρέθυμνο
+    expected: Réthymno
+  - source: Ρόδος
+    expected: Ródos
+  - source: Ρούμελη
+    expected: Roúmeli̱
+  - source: Σαλαμίνα
+    expected: Salamína
+  - source: Σαμοθράκη
+    expected: Samothráki̱
+  - source: Σάμος
+    expected: Sámos
+  - source: Σαντορίνη
+    expected: Santoríni̱
+  - source: Σέρρες
+    expected: Sérres
+  - source: Σίκινος
+    expected: Síkinos
+  - source: Σίφνος
+    expected: Sífnos
+  - source: Σκιάθος
+    expected: Skiáthos
+  - source: Σκόπελος
+    expected: Skópelos
+  - source: Σούλι
+    expected: Soúli
+  - source: Σπάρτη
+    expected: Spárti̱
+  - source: Στερεά Ελλάδα
+    expected: Stereá Elláda
+  - source: Στύρα
+    expected: Stýra
+  - source: Σύμη
+    expected: Sými̱
+  - source: Σύρος
+    expected: Sýros
+  - source: Σφακιά
+    expected: Sfakiá
+  - source: Τήλος
+    expected: Tí̱los
+  - source: Τήνος
+    expected: Tí̱nos
+  - source: Τρίκαλα
+    expected: Tríkala
+  - source: Τρίπολη
+    expected: Trípoli̱
+  - source: Τσακωνιά
+    expected: Tsako̱niá
+  - source: Ύδρα
+    expected: Ýdra
+  - source: Φάληρο
+    expected: Fáli̱ro
+  - source: Φλώρινα
+    expected: Fló̱rina
+  - source: Φολέγανδρος
+    expected: Folégandros
+  - source: Χάλκη
+    expected: Chálki̱
+  - source: Χαλκίδα
+    expected: Chalkída
+  - source: Χαλάνδρι
+    expected: Chalándri
+  - source: Χαλκιδική
+    expected: Chalkidikí̱
+  - source: Χανιά
+    expected: Chaniá
+  - source: Χίος
+    expected: Chíos
+  - source: Ψαρά
+    expected: Psará
+  - source: Αβάνα
+    expected: Avána
+  - source: Αγγλία
+    expected: Aṉglía
+  - source: Αϊβαλί
+    expected: Aïvalí
+  - source: Αλεξάνδρεια
+    expected: Alexándreia
+  - source: Άμστερνταμ
+    expected: Ámsterntam
+  - source: Βαυαρία
+    expected: Vav̱aría
+  - source: Βενετία
+    expected: Venetía
+  - source: Βερολίνο
+    expected: Verolíno
+  - source: Βερόνα
+    expected: Veróna
+  - source: Βιέννη
+    expected: Viénni̱
+  - source: Γένοβα
+    expected: Génova
+  - source: Δουβλίνο
+    expected: Douvlíno
+  - source: Καλαβρία
+    expected: Kalavría
+  - source: Καλιφόρνια
+    expected: Kalifórnia
+  - source: Καύκασος
+    expected: Káf̱kasos
+  - source: Κονγκό
+    expected: Kongkó
+  - source: Κορσική
+    expected: Korsikí̱
+  - source: Κουρδιστάν
+    expected: Kourdistán
+  - source: Κωνσταντινούπολη
+    expected: Ko̱nstantinoúpoli̱
+  - source: Κατεχόμενη Κύπρος
+    expected: Katechómeni̱ Kýpros
+  - source: Λαπωνία
+    expected: Lapo̱nía
+  - source: Λευκωσία
+    expected: Lef̱ko̱sía
+  - source: Λιβόρνο
+    expected: Livórno
+  - source: Λονδίνο
+    expected: Londíno
+  - source: Λυών
+    expected: Lyó̱n
+  - source: Μάλαγα
+    expected: Málaga
+  - source: Μασσαλία
+    expected: Massalía
+  - source: Μικρονησία
+    expected: Mikroni̱sía
+  - source: Μιλάνο
+    expected: Miláno
+  - source: Μόσχα
+    expected: Móscha
+  - source: Μπολόνια
+    expected: Bolónia
+  - source: Νάπολη
+    expected: Nápoli̱
+  - source: Νταγκεστάν
+    expected: Ntagkestán
+  - source: Νέα Υόρκη
+    expected: Néa Yórki̱
+  - source: Οξφόρδη
+    expected: Oxfórdi̱
+  - source: Ουαλία
+    expected: Oualía
+  - source: Παρίσι
+    expected: Parísi
+  - source: Πάφος
+    expected: Páfos
+  - source: Πολυνησία
+    expected: Polyni̱sía
+  - source: Ρώμη
+    expected: Ró̱mi̱
+  - source: Σαμάρεια
+    expected: Samáreia
+  - source: Σικελία
+    expected: Sikelía
+  - source: Σκανδιναβία
+    expected: Skandinavía
+  - source: Σκόπια
+    expected: Skópia
+  - source: Σκωτία
+    expected: Sko̱tía
+  - source: Σμύρνη
+    expected: Smýrni̱
+  - source: Ταϊτή
+    expected: Taïtí̱
+  - source: Ταταρστάν
+    expected: Tatarstán
+  - source: Τζαμάικα
+    expected: Tzamáika
+  - source: Τηλλυρία
+    expected: Ti̱llyría
+  - source: Τιρόλο
+    expected: Tirólo
+  - source: Τορίνο
+    expected: Toríno
+  - source: Φανάρι
+    expected: Fanári
+  - source: Φλωρεντία
+    expected: Flo̱rentía
+  - source: Χαβάη
+    expected: Chavái̱
+  - source: Χονγκ Κονγκ
+    expected: Chongk Kongk
+
+map:
+  # https://en.wikipedia.org/wiki/Romanization_of_Greek
+  rules:
+    - pattern: (?<=[ΑαΕεΗη])\u03A5(?=[θΘκΚξΞπΠσΣςτΤφΦχΧψΨ])    # Υ (after Α, Ε, Η)
+      result: F̱
+    - pattern: (?<=[ΑαΕεΗη])\u03A5(?=\b)    # Υ (after Α, Ε, Η)
+      result: F̱
+    - pattern: (?<=[ΑαΕεΗη])\u03C5(?=[θΘκΚξΞπΠσΣςτΤφΦχΧψΨ])    # υ (after Α, Ε, Η)
+      result: f̱
+    - pattern: (?<=[ΑαΕεΗη])\u03C5(?=\b)    # υ (after Α, Ε, Η)
+      result: f̱
+    - pattern: (?<=[ΑαΕεΗη])\u03A5(?![θΘκΚξΞπΠσΣςτΤφΦχΧψΨ]|\b)    # Υ (after Α, Ε, Η)
+      result: V̱
+    - pattern: (?<=[ΑαΕεΗη])\u03C5(?![θΘκΚξΞπΠσΣςτΤφΦχΧψΨ]|\b)    # υ (after Α, Ε, Η)
+      result: v̱
+    - pattern: \u0391\u03CD(?=[θΘκΚξΞπΠσΣςτΤφΦχΧψΨ])    # Αύ
+      result: Áf̱
+    - pattern: \u0391\u03CD(?=\b)    # Αύ
+      result: Áf̱
+    - pattern: \u0391\u03CD(?![θΘκΚξΞπΠσΣςτΤφΦχΧψΨ]|\b)    # Αύ
+      result: Áv̱
+    - pattern: \u03B1\u03CD(?=[θΘκΚξΞπΠσΣςτΤφΦχΧψΨ])    # Αύ
+      result: áf̱
+    - pattern: \u03B1\u03CD(?=\b)    # Αύ
+      result: áf̱
+    - pattern: \u03B1\u03CD(?![θΘκΚξΞπΠσΣςτΤφΦχΧψΨ]|\b)    # Αύ
+      result: áv̱
+    - pattern: \u0395\u03CD(?=[θΘκΚξΞπΠσΣςτΤφΦχΧψΨ])    # Εύ
+      result: Éf̱
+    - pattern: \u0395\u03CD(?=\b)    # Εύ
+      result: Éf̱
+    - pattern: \u0395\u03CD(?![θΘκΚξΞπΠσΣςτΤφΦχΧψΨ]|\b)    # Εύ
+      result: Év̱
+    - pattern: \u03B5\u03CD(?=[θΘκΚξΞπΠσΣςτΤφΦχΧψΨ])    # Εύ
+      result: éf̱
+    - pattern: \u03B5\u03CD(?=\b)    # Εύ
+      result: éf̱
+    - pattern: \u03B5\u03CD(?![θΘκΚξΞπΠσΣςτΤφΦχΧψΨ]|\b)    # Εύ
+      result: év̱
+    - pattern: \u0397\u03CD(?=[θΘκΚξΞπΠσΣςτΤφΦχΧψΨ])    # Εύ
+      result: Í̱f̱
+    - pattern: \u0397\u03CD(?=\b)    # Εύ
+      result: Í̱f̱
+    - pattern: \u0397\u03CD(?![θΘκΚξΞπΠσΣςτΤφΦχΧψΨ]|\b)    # Εύ
+      result: Í̱v̱
+    - pattern: \u03B7\u03CD(?=[θΘκΚξΞπΠσΣςτΤφΦχΧψΨ])    # Εύ
+      result: í̱f̱
+    - pattern: \u03B7\u03CD(?=\b)    # Εύ
+      result: í̱f̱
+    - pattern: \u03B7\u03CD(?![θΘκΚξΞπΠσΣςτΤφΦχΧψΨ]|\b)    # Εύ
+      result: í̱v̱
+    - pattern: (?<=[Οο])\u03A5    # Υ (after Ο)
+      result: U
+    - pattern: (?<=[Οο])\u03C5    # υ (after Ο)
+      result: u
+    - pattern: (?<=[Οο])\u03CD    # ύ (after Ο)
+      result: ú
+    - pattern: (?<=[ΆάΈέΉήΌό])\u03A5    # Άυ, Έυ, Ήυ, Όυ
+      result: Υ
+    - pattern: (?<=[ΆάΈέΉήΌό])\u03C5    # Άυ, Έυ, Ήυ, Όυ
+      result: y
+    - pattern: \u0393(?=[ξΞχΧ])    # Γ (before Γ, Ξ, Χ)
+      result: Ṉ
+    - pattern: \u03B3(?=[ξΞχΧ])    # γ (before Γ, Ξ, Χ)
+      result: ṉ
+    - pattern: \u0393[Γγ]    # Γ (before Γ)
+      result: Ṉg
+    - pattern: \u03B3\u03B3    # γ (before Γ)
+      result: ṉg
+    - pattern: \u0393[Κκ]    # Γ (before Κ)
+      result: Gk
+    - pattern: \u03B3[Κκ]    # γ (before Κ)
+      result: gk
+    - pattern: (?<=\b)\u039C[πΠ]    # ΜΠ (initially)
+      result: B
+    - pattern: (?<=\b)\u03BC[πΠ]    # μπ (initially)
+      result: b
+    - pattern: \u039C[πΠ](?=\b)    # ΜΠ (finally)
+      result: B
+    - pattern: \u03BC[πΠ](?=\b)    # μπ (finally)
+      result: b
+    - pattern: (?<!\b)\u039C[πΠ]    # ΜΠ (medially)
+      result: Mp
+    - pattern: (?<!\b)\u03BC[πΠ]    # μπ (medially)
+      result: mp
+    - pattern: \u037E   # ;
+      result: "?"
+    - pattern: \u003B   # ;
+      result: "?"
+
+  characters:
+    "\u0027": ""
+    "\u0386": "Á"  # Ά
+    "\u0391": "A"  # Α
+    "\u0392": "V"  # Β
+    "\u0393": "G"  # Γ
+    "\u0394": "D"  # Δ
+    "\u0395": "E"  # Ε
+    "\u0395\u03AA": "Εï"  # ΕΪ
+    "\u0395\u03CA": "Εï"  # Εϊ
+    "\u0388\u03CA": "Éï"  # Έϊ
+    "\u0388\u03CA": "Éï"  # Έι
+    "\u0396": "Z"  # Ζ
+    "\u0397": "I̱"  # Η
+    "\u0398": "Th" # Θ
+    "\u0399": "I"  # Ι
+    "\u039A": "K"  # Κ
+    "\u039B": "L"  # Λ
+    "\u039C": "M"  # Μ
+    "\u039D": "N"  # Ν
+    "\u039E": "X"  # Ξ
+    "\u039F": "O"  # Ο
+    "\u03A0": "P"  # Π
+    "\u03A1": "R"  # Ρ
+    "\u03A3": "S"  # Σ
+    "\u03A4": "T"  # Τ
+    "\u03A5": "Y"  # Υ
+    "\u03A6": "F" # Φ
+    "\u03A7": "Ch" # Χ
+    "\u03A8": "Ps" # Ψ
+    "\u03A9": "O̱"  # Ω
+    "\u0388": "É"  # Έ
+    "\u0389": "Í̱"  # Ή
+    "\u038A": "Í"  # Ί
+    "\u038C": "Ó"  # Ό
+    "\u038E": "Ý"  # Ύ
+    "\u038F": "Ó̱"  # Ώ
+    "\u03AA": "Ï"  # Ϊ
+    "\u03AB": "Ÿ"  # Ϋ
+
+    "\u03AC": "á"  # ά
+    "\u03B1": "a"  # α
+    "\u03B2": "v"  # β
+    "\u03B3": "g"  # γ
+    "\u03B4": "d"  # δ
+    "\u03B5": "e"  # ε
+    "\u03B5\u03CA": "eï"  # εϊ
+    "\u03AD\u03CA": "éï"  # έϊ
+    "\u03AD\u03B9": "éï"  # έι
+    "\u03B6": "z"  # ζ
+    "\u03B7": "i̱"  # η
+    "\u03B8": "th" # θ
+    "\u03B9": "i"  # ι
+    "\u03BA": "k"  # κ
+    "\u03BB": "l"  # λ
+    "\u03BC": "m"  # μ
+    "\u03BD": "n"  # ν
+    "\u03BE": "x"  # ξ
+    "\u03BF": "o"  # ο
+    "\u03C0": "p"  # π
+    "\u03C1": "r"  # ρ
+    "\u03C3": "s"  # σ
+    "\u03C2": "s"  # ς
+    "\u03C4": "t"  # τ
+    "\u03C5": "y"  # υ
+    "\u03C6": "f" # φ
+    "\u03C7": "ch" # χ
+    "\u03C8": "ps" # ψ
+    "\u03C9": "o̱"  # ω
+    "\u03AD": "é"  # έ
+    "\u03AE": "í̱"  # ή
+    "\u03AF": "í"  # ί
+    "\u03CC": "ó"  # ό
+    "\u03CD": "ý"  # ύ
+    "\u03CE": "ó̱"  # ώ
+    "\u03CA": "ï"  # ϊ
+    "\u03CB": "ÿ"  # ϋ
+    "\u0390": "ḯ"  # ΐ
+    "\u03B0": "ÿ́"  # ΰ
+
+    "\u0387": ";"   # ·
+    "\u00B7": ";"   # ·