interscript 0.1.2 → 0.1.7

data/maps/iso-ben-Beng-Latn-15919-2001.yaml

@@ -0,0 +1,175 @@
+---
+authority_id: iso
+id: 15919-2001
+language: iso-639-2:ben
+source_script: Beng
+destination_script: Latn
+name: "Information and documentation — Transliteration of Devanagari and related Indic scripts into Latin characters"
+url: https://www.chatranjali.fr/Scripts/Standards/ISO15919.pdf
+creation_date: 2001
+adoption_date: 2001
+description: |
+  Script conversion is often required for documents such as historical and literary texts, geographical texts (including
+  maps and atlases), bibliographies, catalogues, lists and passports (and other identification documents).
+
+  Text in Devanagari script or other Indic scripts sometimes needs to be shown in Latin script, where users, or
+  equipment that they are using, cannot read or write the text
+
+  This International Standard applies to transliteration of Devanagari, and to Indic scripts related to Devanagari,
+  independent of the period in which it is or was used.
+
+notes:
+
+  - All transliterations made using this International Standard shall be case-insensitive.
+  - Inherent a with a consonant shall always be transliterated.
+  - anusvara (including Vedic anusvara) shall be transliterated as ṁ
+  - candrabindu shall be transliterated as m̐
+  - When m̐, ṃ or ṁ are associated with a vowel, they shall be placed after the vowel. When m̐ is associated
+    with a semivowel, it shall be placed before the semivowel.
+  - Latin punctuation signs and Hindu-Arabic numerals shall remain unchanged in transliteration.
+    Indic punctuation is outside the scope of this International Standard.
+  - The Vedic accent Udatta shall be transliterated as an acute accent over the transliterated vowel, and the
+    independent Svarita as a grave accent over the transliterated vowel. In the case of the digraphs ai, au, the accent
+    shall be attached to the second vowel.
+  - |
+    A colon: before a Latin character shall be used to resolve ambiguity. Some normative cases are as
+    follows.
+     -  :’ for avagraha in modern text. (The apostrophe in modern text remains unchanged in accordance with previous rule.
+     -  Vowel hiatus, not digraph transliteration of diphthongs; as in Bengali ba:i (not bai), “book”.
+     -  Unexpected use of secondary ya/ẏa , as in Bengali r:yāpāra , English “wrapper” (using full r); a:yā (not ayā ).
+  - If a character in an Indic script is defined in such a way as to be equivalent to another character in any
+    script, where the second character has a transliteration in this International Standard, then the first character shall
+    be transliterated in the same way as the second character.
+  - Where it is desired to show the Vedic accent Anudatta, it should be transliterated as an underscore. In the case of
+    the digraphs ai, au, both Latin vowels should be underscored.
+    Where word boundaries are not shown in the original text (as happens commonly in Sanskrit) and a word ends in a
+    consonant, the transliteration should show word division by a space; but when phonological processes result in two
+    words sharing a common vowel, no attempt should be made to separate them. This will require a good knowledge
+    of the language in question.
+
+
+tests:
+  - source: "আগুয়েরোর হাঁটুর চোট"
+    expected: "āgauẏaeraora haām̐ṭaura caoṭa"
+  - source: "তঅস্ত্রোপচারের আশ্রয় নিতে হবে এ তারকাকে, তাই দলে জায়গা পাননি"
+    expected: "taasataraopacaāraera āśaraẏa naitae habae e taārakaākae, taāi dalae jaāẏagaā paānanai"
+  - source: "ওদিকে সদ্য করোনা থেকে ফিরে এসে দাপটের সঙ্গে পিএসজির হয়ে খেলে যাওয়া দি মারিয়া কেন ডাক পাননি, কেউ জানে না"
+    expected: "odaikae sadaya karaonaā thaekae phairae esae daāpaṭaera saṅagae paiesajaira haẏae khaelae yaāoẏaā dai maāraiẏaā kaena ḍaāka paānanai, kaeu jaānae naā"
+  - source: "তিন বিভাগে শনিবার থেকে ভারী বৃষ্টি, কমবে তাপমাত্রা"
+    expected: "taina baibhaāgae śanaibaāra thaekae bhaāraī baṛṣaṭai, kamabae taāpamaātaraā"
+  - source: "ফেসবুকে সময় নষ্ট না করে মায়ের সঙ্গে আড্ডা ভালোবাসেন তিনি"
+    expected: "phaesabaukae samaẏa naṣaṭa naā karae maāẏaera saṅagae āḍaḍaā bhaālaobaāsaena tainai"
+  - source: "এখনই আর্জেন্টিনার জার্সি খুলো না—দি মারিয়াকে কোচ"
+    expected: "ekhanai ārajaenaṭainaāra jaārasai khaulao naā—dai maāraiẏaākae kaoca"
+  - source: "প্রিয় তারকার আওয়াজ শুনে কোমা থেকে জেগে উঠলেন তিনি"
+    expected: "paraiẏa taārakaāra āoẏaāja śaunae kaomaā thaekae jaegae uṭhalaena tainai"
+  - source: "সুয়ারেজকে ‘বের করে দেওয়া’য় বার্সাকে ধুয়ে দিলেন মেসি"
+    expected: "sauẏaāraejakae ‘baera karae daeoẏaā’ẏa baārasaākae dhauẏae dailaena maesai"
+  - source: "আমি না থাকলে আর্জেন্টিনা দলে মেসিরও থাকার যোগ্যতা নেই"
+    expected: "āmai naā thaākalae ārajaenaṭainaā dalae maesairao thaākaāra yaogayataā naei"
+  - source: "তবে, নিয়মিত মুখগুলোর মধ্যে দলে নেই মেসির দুই ‘প্রিয় বন্ধু’ পিএসজির মিডফিল্ডার আনহেল দি মারিয়া ও ম্যানচেস্টার সিটির সার্জিও আগুয়েরো"
+    expected: "tabae, naiẏamaita maukhagaulaora madhayae dalae naei maesaira daui ‘paraiẏa banadhau’ paiesajaira maiḍaphailaḍaāra ānahaela dai maāraiẏaā o mayaānacaesaṭaāra saiṭaira saārajaio āgauẏaerao"
+
+map:
+  characters:
+
+    #vowel characters
+
+    'অ': 'a'
+    'আ': 'ā'
+    'ই': 'i'
+    'ঈ': 'ī'
+    'উ': 'u'
+    'ঊ': 'ū'
+    'ঋ': 'ṛ'
+    'ৠ': 'ṝ'
+    'ঌ': 'ḻ'
+    'ৡ': 'ḹ'
+    'এ': 'e'
+    'ঐ': 'ai'
+    'ও': 'o'
+    'ঔ': 'au'
+
+    'া': 'ā'
+    'ি': 'i'
+    'ী': 'ī'
+    'ু': 'u'
+    'ূ': 'ū'
+    'ৃ': 'ṛ'
+    'ৄ': 'ṝ'
+    'ৢ': 'ḻ'
+    'ৣ': 'ḹ'
+    'ে': 'e'
+    'ৈ': 'ai'
+    'ো': 'o'
+    'ৌ': 'au'
+
+
+    'ং':	'ṁ'
+    'ঁ':	'm̐'
+    'ঃ':	'ḥ'
+
+    # Consonant characters
+
+    'ক': 'ka'
+    'খ': 'kha'
+    'গ': 'ga'
+    'ঘ': 'gha'
+    'ঙ': 'ṅa'
+    'চ': 'ca'
+    'ছ': 'cha'
+    'জ': 'ja'
+    'ঝ': 'jha'
+    'ঞ': 'ña'
+    'ট': 'ṭa'
+    'ঠ': 'ṭha'
+    'ড': 'ḍa'
+    'ঢ': 'ḍha'
+    'ণ': 'ṇa'
+    'ত': 'ta'
+    'থ': 'tha'
+    'দ': 'da'
+    'ধ': 'dha'
+    'ন': 'na'
+    'প': 'pa'
+    'ফ': 'pha'
+    'ব': 'ba'
+    'ভ': 'bha'
+    'ম': 'ma'
+    'য': 'ya'
+    'র': 'ra'
+    'ল': 'la'
+    'শ': 'śa'
+    'ষ': 'ṣa'
+    'স': 'sa'
+    'হ': 'ha'
+
+    'ড়': 'ṙa'
+    'ঢ়': 'ṙha'
+    'য়': 'ẏa'
+
+
+    # for semivowel rule no. 5
+
+    'যঁ': 'm̐ya'
+    'রঁ': 'm̐ra'
+    'লঁ': 'm̐la'
+    'বঁ': 'm̐va'
+
+    'ৎ': 't'
+    '্': ''
+
+
+   # numbers
+
+    '১': '1' 
+    '২': '2'
+    '৩': '3'
+    '৪': '4'
+    '৫': '5'
+    '৬': '6'
+    '৭': '7'
+    '৮': '8'
+    '৯': '9'
+    '০': '0'
+

data/maps/iso-ell-Grek-Latn-843-1997-t1.yaml

@@ -0,0 +1,613 @@
+---
+authority_id: iso
+id: 1997
+language: iso-639-2:ell
+source_script: Grek
+destination_script: Latn
+name: "ISO 843:1997 Information and documentation — Conversion of Greek characters into Latin characters"
+url: https://www.iso.org/standard/5215.html
+creation_date: 1997
+description: |
+  Establishes a system for the transliteration and/or transcription 
+  of Greek characters into Latin characters. This system applies to
+  the characters of the Greek script, independent of the period in
+  which it is or was used, i.e. scripts from all periods of Classic
+  or Modern Greek. Replaces ISO/R 843.
+
+note:
+  - Transliteration of Greek into Latin: Type 1, Clause 3 Table 1
+  - Equivalent to elot-ell-Grek-Latn-743-2001-ts, the transliteration table of ELOT 743:2001
+  - Assuming that ou, au, eu transliterations are only intended for historical diphthongs /u, av, ev/, and that όυ, άυ, έυ are not to be transliterated as ou, au, eu
+  - Introduced casing to digamma and lunate sigma. (Casing was late introduction to character sets for those characters)
+
+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 parakínīse ki eména na grápsō óti toútīn tīn patrída tīn échomen óloi mazí, kai sofoí ki amatheís kai ploúsioi kai ftōchoí kai politikoí kai stratiōtikoí kai oi pléon mikróteroi ánthrōpoi; ósoi agōnistī́kamen, analógōs o katheís, échomen na zī́somen edṓ. To loipón doulépsamen óloi mazí, na tīn fylámen ki óloi mazí kai na mīn légei oúte o dynatós «egṓ» oúte o adýnatos. Xérete póte na légei o katheís «egṓ»? Ótan agōnisteí mónos tou kai fkiásei ī́ chalásei, na légei «egṓ»; ótan ómōs agōnízontai polloí kai fkiánoun, tóte na léne «emeís». Eímaste eis to «emeís» ki óchi eis to «egṓ». Kai eis to exī́s na máthomen gnṓsī, an thélomen na fkiásomen chōrión, na zī́somen óloi mazí.
+
+        Giánnīs Makrygiánnīs.
+
+
+  - source: ΑΘΗΝΑ
+    expected: ATHĪNA
+  - source: μπαμπάκι
+    expected: mpampáki
+  - source: νταντά
+    expected: ntantá
+  - source: γκέγκε
+    expected: gkégke
+  - source: Γκαμπόν
+    expected: Gkampón
+  - source: Μάγχη
+    expected: Mágchī
+  - source: κογξ
+    expected: kogx
+  - source: υιός
+    expected: yiós
+  - source: Υιός
+    expected: Yiós
+  - source: νεράντζι
+    expected: nerántzi
+  - source: Γοίθιος
+    expected: Goíthios
+  - source: μπέικον
+    expected: mpéikon
+  - source: μπέϊκον
+    expected: mpéïkon
+  - source: βόλεϊ
+    expected: vóleï
+  - source: αθεΐα
+    expected: atheḯa
+  - source: Εϊγιαφιάτλαγιοκουτλ
+    expected: Eïgiafiátlagiokoutl
+  - source: Εΐτζι
+    expected: Eḯtzi
+  - source: Μυρτώο
+    expected: Myrtṓo
+  - source: αέρας
+    expected: aéras
+  - source: γαυ γαυ
+    expected: gau gau
+  - source: Ταΰγετος
+    expected: Taÿ́getos
+  - source: σπρέυ
+    expected: spréy
+
+  - source: Αθήνα
+    expected: Athī́na
+  - 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: Alexandroúpolī
+  - source: Αλεποχώρι
+    expected: Alepochṓri
+  - source: Αμοργός
+    expected: Amorgós
+  - source: Άμφισσα
+    expected: Ámfissa
+  - source: Αράχωβα
+    expected: Aráchōva
+  - source: Άργος
+    expected: Árgos
+  - source: Αρκαδία
+    expected: Arkadía
+  - source: Άρτα
+    expected: Árta
+  - source: Βελούχι
+    expected: Veloúchi
+  - source: Βέροια
+    expected: Véroia
+  - source: Βοιωτία
+    expected: Voiōtía
+  - 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: Dōdekánīsa
+  - source: Έδεσσα
+    expected: Édessa
+  - source: Ελευσίνα
+    expected: Eleusína
+  - source: Επίδαυρος
+    expected: Epídauros
+  - source: Επτάνησα
+    expected: Eptánīsa
+  - source: Ερμούπολη
+    expected: Ermoúpolī
+  - source: Εύβοια
+    expected: Eúvoia
+  - source: Ζάκυνθος
+    expected: Zákynthos
+  - source: Ήπειρος
+    expected: Ī́peiros
+  - source: Ηράκλειο
+    expected: Īrákleio
+  - source: Θάσος
+    expected: Thásos
+  - source: Θεσσαλονίκη
+    expected: Thessaloníkī
+  - source: Θεσσαλία
+    expected: Thessalía
+  - source: Θεσπρωτία
+    expected: Thesprōtía
+  - source: Θήβα
+    expected: Thī́va
+  - source: Θράκη
+    expected: Thrákī
+  - source: Ιθάκη
+    expected: Ithákī
+  - source: Ίος
+    expected: Íos
+  - source: Ιωάννινα
+    expected: Iōánnina
+  - 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: Karpenī́si
+  - source: Κάρυστος
+    expected: Kárystos
+  - source: Καστελλόριζο
+    expected: Kastellórizo
+  - source: Καστοριά
+    expected: Kastoriá
+  - source: Κατερίνη
+    expected: Katerínī
+  - source: Κάτω Αχαΐα
+    expected: Kátō Achaḯa
+  - source: Κερατέα
+    expected: Keratéa
+  - source: Κέρκυρα
+    expected: Kérkyra
+  - source: Κεφαλλονιά
+    expected: Kefalloniá
+  - source: Κηφισιά
+    expected: Kīfisiá
+  - source: Κιλκίς
+    expected: Kilkís
+  - source: Κοζάνη
+    expected: Kozánī
+  - source: Κολωνός
+    expected: Kolōnós
+  - source: Κομοτηνή
+    expected: Komotīnī́
+  - source: Κόρινθος
+    expected: Kórinthos
+  - source: Κορώνη
+    expected: Korṓnī
+  - source: Κρανίδι
+    expected: Kranídi
+  - source: Κρέστενα
+    expected: Kréstena
+  - source: Κρήτη
+    expected: Krī́tī
+  - source: Κύθηρα
+    expected: Kýthīra
+  - source: Κυκλάδες
+    expected: Kykládes
+  - source: Κύμη
+    expected: Kýmī
+  - source: Κυψέλη
+    expected: Kypsélī
+  - source: Κως
+    expected: Kōs
+  - source: Λαγκαδάς
+    expected: Lagkadás
+  - source: Λαμία
+    expected: Lamía
+  - source: Λάρισα
+    expected: Lárisa
+  - source: Λαύριο
+    expected: Laúrio
+  - source: Λέρος
+    expected: Léros
+  - source: Λέσβος
+    expected: Lésvos
+  - source: Λευκάδα
+    expected: Leukáda
+  - source: Λήμνος
+    expected: Lī́mnos
+  - source: Λιβαδειά
+    expected: Livadeiá
+  - source: Μακεδονία
+    expected: Makedonía
+  - source: Μάνη
+    expected: Mánī
+  - source: Μαραθώνας
+    expected: Marathṓnas
+  - source: Μαρκόπουλο
+    expected: Markópoulo
+  - source: Μαρούσι
+    expected: Maroúsi
+  - source: Μέγαρα
+    expected: Mégara
+  - source: Μεσολόγγι
+    expected: Mesológgi
+  - source: Μεταξουργείο
+    expected: Metaxourgeío
+  - source: Μέτσοβο
+    expected: Métsovo
+  - source: Μήλος
+    expected: Mī́los
+  - source: Μύκονος
+    expected: Mýkonos
+  - source: Μυστράς
+    expected: Mystrás
+  - source: Μυτιλήνη
+    expected: Mytilī́nī
+  - source: Νάξος
+    expected: Náxos
+  - source: Νάουσα
+    expected: Náousa
+  - source: Ναύπακτος
+    expected: Naúpaktos
+  - source: Ναύπλιο
+    expected: Naúplio
+  - source: Νέα Σμύρνη
+    expected: Néa Smýrnī
+  - source: Νίσυρος
+    expected: Nísyros
+  - source: Ξάνθη
+    expected: Xánthī
+  - source: Όλυμπος
+    expected: Ólympos
+  - source: Παγκράτι
+    expected: Pagkráti
+  - source: Παπάγου
+    expected: Papágou
+  - source: Πάρος
+    expected: Páros
+  - source: Πασαλιμάνι
+    expected: Pasalimáni
+  - source: Πατήσια
+    expected: Patī́sia
+  - source: Πάτμος
+    expected: Pátmos
+  - source: Πάτρα
+    expected: Pátra
+  - source: Πειραιάς
+    expected: Peiraiás
+  - source: Πελοπόννησος
+    expected: Pelopónnīsos
+  - source: Περιστέρι
+    expected: Peristéri
+  - source: Πεύκη
+    expected: Peúkī
+  - source: Πήλιο
+    expected: Pī́lio
+  - 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: Roúmelī
+  - source: Σαλαμίνα
+    expected: Salamína
+  - source: Σαμοθράκη
+    expected: Samothrákī
+  - source: Σάμος
+    expected: Sámos
+  - source: Σαντορίνη
+    expected: Santorínī
+  - 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: Spártī
+  - source: Στερεά Ελλάδα
+    expected: Stereá Elláda
+  - source: Στύρα
+    expected: Stýra
+  - source: Σύμη
+    expected: Sýmī
+  - source: Σύρος
+    expected: Sýros
+  - source: Σφακιά
+    expected: Sfakiá
+  - source: Τήλος
+    expected: Tī́los
+  - source: Τήνος
+    expected: Tī́nos
+  - source: Τρίκαλα
+    expected: Tríkala
+  - source: Τρίπολη
+    expected: Trípolī
+  - source: Τσακωνιά
+    expected: Tsakōniá
+  - source: Ύδρα
+    expected: Ýdra
+  - source: Φάληρο
+    expected: Fálīro
+  - source: Φλώρινα
+    expected: Flṓrina
+  - source: Φολέγανδρος
+    expected: Folégandros
+  - source: Χάλκη
+    expected: Chálkī
+  - source: Χαλκίδα
+    expected: Chalkída
+  - source: Χαλάνδρι
+    expected: Chalándri
+  - source: Χαλκιδική
+    expected: Chalkidikī́
+  - source: Χανιά
+    expected: Chaniá
+  - source: Χίος
+    expected: Chíos
+  - source: Ψαρά
+    expected: Psará
+  - source: Αβάνα
+    expected: Avána
+  - source: Αγγλία
+    expected: Agglía
+  - source: Αϊβαλί
+    expected: Aïvalí
+  - source: Αλεξάνδρεια
+    expected: Alexándreia
+  - source: Άμστερνταμ
+    expected: Ámsterntam
+  - source: Βαυαρία
+    expected: Vauaría
+  - source: Βενετία
+    expected: Venetía
+  - source: Βερολίνο
+    expected: Verolíno
+  - source: Βερόνα
+    expected: Veróna
+  - source: Βιέννη
+    expected: Viénnī
+  - source: Γένοβα
+    expected: Génova
+  - source: Δουβλίνο
+    expected: Douvlíno
+  - source: Καλαβρία
+    expected: Kalavría
+  - source: Καλιφόρνια
+    expected: Kalifórnia
+  - source: Καύκασος
+    expected: Kaúkasos
+  - source: Κονγκό
+    expected: Kongkó
+  - source: Κορσική
+    expected: Korsikī́
+  - source: Κουρδιστάν
+    expected: Kourdistán
+  - source: Κωνσταντινούπολη
+    expected: Kōnstantinoúpolī
+  - source: Κατεχόμενη Κύπρος
+    expected: Katechómenī Kýpros
+  - source: Λαπωνία
+    expected: Lapōnía
+  - source: Λευκωσία
+    expected: Leukōsía
+  - source: Λιβόρνο
+    expected: Livórno
+  - source: Λονδίνο
+    expected: Londíno
+  - source: Λυών
+    expected: Lyṓn
+  - source: Μάλαγα
+    expected: Málaga
+  - source: Μασσαλία
+    expected: Massalía
+  - source: Μικρονησία
+    expected: Mikronīsía
+  - source: Μιλάνο
+    expected: Miláno
+  - source: Μόσχα
+    expected: Móscha
+  - source: Μπολόνια
+    expected: Mpolónia
+  - source: Νάπολη
+    expected: Nápolī
+  - source: Νταγκεστάν
+    expected: Ntagkestán
+  - source: Νέα Υόρκη
+    expected: Néa Yórkī
+  - source: Οξφόρδη
+    expected: Oxfórdī
+  - source: Ουαλία
+    expected: Oualía
+  - source: Παρίσι
+    expected: Parísi
+  - source: Πάφος
+    expected: Páfos
+  - source: Πολυνησία
+    expected: Polynīsía
+  - source: Ρώμη
+    expected: Rṓmī
+  - source: Σαμάρεια
+    expected: Samáreia
+  - source: Σικελία
+    expected: Sikelía
+  - source: Σκανδιναβία
+    expected: Skandinavía
+  - source: Σκόπια
+    expected: Skópia
+  - source: Σκωτία
+    expected: Skōtía
+  - source: Σμύρνη
+    expected: Smýrnī
+  - source: Ταϊτή
+    expected: Taïtī́
+  - source: Ταταρστάν
+    expected: Tatarstán
+  - source: Τζαμάικα
+    expected: Tzamáika
+  - source: Τηλλυρία
+    expected: Tīllyría
+  - source: Τιρόλο
+    expected: Tirólo
+  - source: Τορίνο
+    expected: Toríno
+  - source: Φανάρι
+    expected: Fanári
+  - source: Φλωρεντία
+    expected: Flōrentía
+  - source: Χαβάη
+    expected: Chaváī
+  - source: Χονγκ Κονγκ
+    expected: Chongk Kongk
+
+map:
+  # https://en.wikipedia.org/wiki/Romanization_of_Greek
+  rules:
+    - pattern: (?<=[ΑαΕεΟο])\u03A5    # Υ (after Α, Ε, Η)
+      result: U
+    - pattern: (?<=[ΑαΕεΟο])\u03C5    # υ (after Α, Ε, Η)
+      result: u
+    - pattern: (?<=[ΑαΕεΟο])\u03CD    # ύ (after Ο)
+      result: ú
+    - 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
+    "\u0396": "Z"  # Ζ
+    "\u0397": "Ī"  # 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": "Ō"  # Ω
+    "\u0388": "É"  # Έ
+    "\u0389": "Ī́"  # Ή
+    "\u038A": "Í"  # Ί
+    "\u038C": "Ó"  # Ό
+    "\u038E": "Ý"  # Ύ
+    "\u038F": "Ṓ"  # 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
+    "\u03B6": "z"  # ζ
+    "\u03B7": "ī"  # 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": "ō"  # ω
+    "\u03AD": "é"  # έ
+    "\u03AE": "ī́"  # ή
+    "\u03AF": "í"  # ί
+    "\u03CC": "ó"  # ό
+    "\u03CD": "ý"  # ύ
+    "\u03CE": "ṓ"  # ώ
+    "\u03CA": "ï"  # ϊ
+    "\u03CB": "ÿ"  # ϋ
+    "\u0390": "ḯ"  # ΐ
+    "\u03B0": "ÿ́"  # ΰ
+
+    "\u03DC": "W"   # Ϝ
+    "\u03DD": "w"   # ϝ
+    "\u03F2": "s"   # ϲ
+    "\u03F9": "S"   # Ϲ
+
+    "\u0387": ";"   # ·
+    "\u00B7": ";"   # ·