interscript 0.1.3 → 0.1.9

data/lib/interscript/opal.rb

@@ -0,0 +1,196 @@
+require "onigmo"
+require "onigmo/core_ext"
+
+# Increase this if there are out-of-memory errors. This setting is
+# tested to be big enough to handle all the maps provided.
+Onigmo::FFI.library.memory.grow(128)
+
+module Interscript
+  module Opal
+    def mkregexp(regexpstring)
+      @cache ||= {}
+      if s = @cache[regexpstring]
+        s
+      else
+        # JS regexp is more performant than Onigmo. Let's use the JS
+        # regexp wherever possible, but use Onigmo where we must.
+        # Let's allow those characters to happen for the regexp to be
+        # considered compatible: ()|.*+?{} ** BUT NOT (? **.
+        if /[\\$^\[\]]|\(\?/.match?(regexpstring)
+          # Ruby caches its regexps internally. We can't GC. We could
+          # think about freeing them, but we really can't, because they
+          # may be in use.
+
+          # Uncomment those to keep track of Onigmo/JS regexp compilation.
+          # print '#'
+          @cache[regexpstring] = Onigmo::Regexp.new(regexpstring)
+        else
+          # print '.'
+          @cache[regexpstring] = Regexp.new(regexpstring)
+        end
+      end
+    end
+
+    def sub_replace(string, pos, size, repl)
+      string[0, pos] + repl + string[pos + size..-1]
+    end
+
+    def external_processing(mapping, string)
+      string
+    end
+
+    def load_map_json(_, json)
+      json = Hash.new(json) if native? json
+      json = JSON.load(json) if String === json
+      json.each do |k,v|
+        `Opal.global.InterscriptMaps[#{k}] = #{JSON.dump(v)}`
+      end
+    end
+
+    # Use #on_load_maps if possible. It will be available earlier.
+    # See lib/interscript/opal/entrypoint.rb
+    def load_maps(opts, &block)
+      # Convert arg
+      opts = Hash.new(opts) if native? opts
+
+      defaults = {
+        maps: [],
+        path: nil,
+        node_path: "./maps/",
+        ajax_path: "maps/",
+        loader: nil,
+        processor: proc { |i| i },
+      }
+
+      opts = defaults.merge opts
+      opts[:maps] = Array(opts[:maps])
+
+      %x{
+        var ajax_loader = function(map) {
+          return new Promise(function(ok, fail) {
+            var httpRequest = new XMLHttpRequest();
+            httpRequest.onreadystatechange = function() {
+              if (httpRequest.readyState === XMLHttpRequest.DONE) {
+                if (httpRequest.responseText) {
+                  ok(JSON.parse(httpRequest.responseText));
+                }
+                else {
+                  if (is_local) {
+                    console.log(httpRequest.responseText);
+                    fail("Ajax failed load: "+map+". Status: "+httpRequest.statusText+". "+
+                      "Are you running this locally? Try adding: "+
+                      "--allow-file-access-from-files to your Chromium command line.")
+                  }
+                  else fail("Ajax failed load: "+map+". Status: "+httpRequest.statusText);
+                }
+              }
+            };
+            httpRequest.open('GET', #{opts[:path] || opts[:ajax_path]}+map+".json", true);
+            httpRequest.send();
+          });
+        };
+
+        var fetch_loader = function(map) {
+          return fetch(#{opts[:path] || opts[:ajax_path]}+map+".json").then(function(response) {
+            return response.json();
+          });
+        };
+
+        var node_loader = function(map) {
+          var resolve = null, error = null;
+          var prom = new Promise(function(ok, fail) {
+            resolve = ok;
+            error = fail;
+          });
+          try {
+            var node_require = eval("require");
+            var data = node_require(#{opts[:path] || opts[:node_path]}+map+'.json');
+            resolve(data);
+          }
+          catch(e) {
+            error("Node failed load: "+map+". Error: "+e);
+          }
+          return prom;
+        };
+
+        var is_local = false;
+        if (typeof document !== "undefined" &&
+            typeof document.location !== "undefined" &&
+            typeof document.location.protocol !== "undefined") {
+              is_local = document.location.protocol == "file:";
+            }
+
+        var loader = function(map) {
+          if (#{opts[:loader] != nil}) {
+            return #{opts[:loader]}(#{opts[:path]}+map+'.json').then(#{opts[:processor]});
+          }
+          else if (typeof window !== "undefined") {
+            return ajax_loader(map);
+          }
+          else if (typeof global !== "undefined") {
+            return node_loader(map);
+          }
+          else if (!is_local && typeof fetch === "function") {
+            return fetch_loader(map);
+          }
+          else {
+            #{raise StandardError, "We couldn't find a good way to load a map"}
+          }
+        };
+      }
+
+      prom = `new Promise(function(ok, fail) {
+        #{
+          maps = opts[:maps]
+          maps = maps.map { |i| map_resolve i }
+          maps = maps.reject { |i| map_loaded? i }
+          #p ["Loading:", maps]
+          maps = maps.map do |i|
+            `loader(#{i})`.JS.then do |map|
+              load_map_json(nil, map)
+
+              m = Native(map)
+              inherits = []
+              m.each do |mapname, mapvalue|
+                inherits += Array(Native(mapvalue)[:map][:inherit])
+                inherits += Array(Native(mapvalue)[:chain])
+              end
+              inherits = inherits.uniq
+              inherits = inherits.reject { |i| map_loaded? i }
+
+              load_maps(opts.merge({maps: inherits})) unless inherits.empty?
+            end.JS.catch do |response|
+              `fail(#{response})`
+            end
+          end
+        }
+        Promise.all(#{maps}).then(ok).catch(fail);
+      })`
+
+      if block_given?
+        prom.JS.then(block)
+      else
+        prom
+      end
+    end
+
+    def aliases
+      @aliases ||= Hash.new(`Opal.global.InterscriptMapAliases`)
+    end
+
+    def map_exist?(map)
+      `typeof(Opal.global.InterscriptMaps[#{map}]) !== 'undefined'`
+    end
+
+    def map_loaded?(map)
+      `!!Opal.global.InterscriptMaps[#{map}]`
+    end
+  end
+end
+
+class String
+  # Opal has a wrong implementation of String#unicode_normalize
+  def unicode_normalize
+    self.JS.normalize
+  end
+end

data/lib/interscript/opal/entrypoint.rb

@@ -0,0 +1,20 @@
+require "opal"
+require "interscript/opal/maps"
+require "onigmo/onigmo-wasm"
+
+module Interscript
+  def self.on_load(&block)
+    WebAssembly.wait_for("onigmo/onigmo-wasm", &block)
+  end
+
+  # on_load + load_maps
+  def self.on_load_maps(arg, &block)
+    self.on_load.JS.then do
+      self.load_maps(arg, &block)
+    end
+  end
+end
+
+Interscript.on_load do
+  require "interscript"
+end

data/lib/interscript/opal/exports.rb

@@ -0,0 +1,11 @@
+# Exports - the methods used publically. Listed so that
+# opal-optimizer will know the methods it needs to preserve.
+
+Interscript.on_load
+Interscript.on_load_maps
+Interscript.load_maps
+Interscript.load_map_json
+Interscript.transliterate
+
+# TODO: Fix handling of "ord".$to_proc() in opal-optimizer
+"a".ord

data/lib/interscript/opal/maps.js.erb

@@ -0,0 +1,8 @@
+<%# A placeholder for the maps to be loaded %>
+Opal.global.InterscriptMaps = {
+  <% Dir['maps/*.yaml'].each do |yaml_file| %>
+    "<%= File.basename(yaml_file, ".yaml") %>": null,
+  <% end %>
+};
+
+Opal.global.InterscriptMapAliases = <%= File.read('aliases.json') %>;

data/lib/interscript/version.rb

@@ -1,3 +1,3 @@
 module Interscript
-  VERSION = "0.1.3"
+  VERSION = "0.1.9"
 end

data/maps/acadsin-zho-Hani-Latn-2002.yaml

@@ -1,11 +1,15 @@
 ---
 authority_id: acadsin
 id: 2002
-language: zho
+language: iso-639-2:zho
 source_script: Hani
 destination_script: Latn
 name: Chinese Tongyong Pinyin Academica Sinica 2002 System
-url: 
+alias:
+  ogc11122:
+    code: zho_Hani2Latn_AcadSin_2002
+    description: Chinese Tongyong Pinyin Academica Sinica 2002 System
+url:
 description: Chinese Tongyong Pinyin Academica Sinica 2002 System
 
 tests:

data/maps/alalc-amh-Ethi-Latn-1997.yaml

@@ -0,0 +1,513 @@
+---
+authority_id: alalc
+id: 1997
+language: iso-639-2:amh
+source_script: Ethi
+destination_script: Latn
+name: ALA-LC Romanization Table -- Amharic (1997)
+alias:
+  ogc11122:
+    code: amh_Ethi2Latn_ALA_1997
+    description: Amharic ALA-Library of Congress 1997 System
+url: http://catdir.loc.gov/catdir/cpso/romanization/amharic.pdf
+creation_date: 1997
+description: |
+  ALA-LC Romanization Tables: Transliteration Schemes for Non-Roman Scripts. Randal K. Berry (ed.). Library of Congress, 1997.
+
+notes:
+  - The Ethiopic script used for Amharic is also used for other languages, including Ge’ez, Argobba, Gurage, and Tigre. Ge’ez, which is chiefly a liturgical language, uses only 26 basic letter forms from this table.
+
+tests:
+  - source: የዜግነት ክብር በ ኢትዮጵያችን ጸንቶ
+    expected: yazégenate kebere ba ʼiteyop̣eyāčene ṣaneto
+  - source: ታየ ሕዝባዊነት ዳር እስከዳር በርቶ
+    expected: tāya ḥezebāwinate dāre ʼesekadāre bareto
+  - source: ለሰላም ለፍትህ ለሕዝቦች ነጻነት
+    expected: lasalāme lafetehe laḥezeboče naṣānate
+  - source: በእኩልነት በፍቅር ቆመናል ባንድነት
+    expected: baʼekulenate bafeqere qomanāle bānedenate
+  - source: መሠረተ ፅኑ ሰብዕናን ያልሻርን
+    expected: maśarata ṡenu sabeʻenāne yālešārene
+  - source: ሕዝቦች ነን ለሥራ በሥራ የኖርን
+    expected: ḥezeboče nane laśerā baśerā yanorene
+  - source: ድንቅ የባህል መድረክ ያኩሪ ቅርስ ባለቤት
+    expected: deneqe yabāhele maderake yākuri qerese bālabéte
+  - source: የተፈጥሮ ጸጋ የጀግና ሕዝብ እናት
+    expected: yatafaṭero ṣagā yaǧagenā ḥezebe ʼenāte
+  - source: እንጠብቅሻለን አለብን አደራ
+    expected: ʼeneṭabeqešālane ʼalabene ʼadarā
+  - source: ኢትዮጵያችን ኑሪ እኛም ባንቺ እንኩራ
+    expected: ʼiteyop̣eyāčene nuri ʼeñāme bāneči ʼenekurā
+  - source: ቋንቋ የድምጽ፣ የምልክት ወይም የምስል ቅንብር ሆኖ
+    expected: qwāneqwā yademeṣe፣ yamelekete wayeme yamesele qenebere hono
+  - source: ለማሰብ ወይም የታሰበን ሃሳብ ለሌላ ለማስተላለፍ የሚረዳ መሳሪያ ነው
+    expected: lamāsabe wayeme yatāsabane hāsābe lalélā lamāsetalālafe yamiradā masāriyā nawe
+  - source: በአጭሩ ቋንቋ የምልክቶች ስርዓትና እኒህን ምልክቶች ለማቀናበር
+    expected: baʼaċeru qwāneqwā yameleketoče sereʻātenā ʼenihene meleketoče lamāqanābare
+  - source: የሚያስፈልጉ ህጎች ጥንቅር ነው። ቋንቋወችን ለመፈረጅ እንዲሁም
+    expected: yamiyāsefalegu hegoče ṭeneqere nawe። qwāneqwāwačene lamafaraǧe ʼenedihume
+  - source: ለምክፈል የሚያስችሉ መስፈርቶችን ለማስቀመጥ ባለው ችግር
+    expected: lamekefale yamiyāsečelu masefaretočene lamāseqamaṭe bālawe čegere
+  - source: ምክንያት በአሁኑ ሰዓት በርግጠኝነት ስንት ቋንቋ በዓለም ላይ
+    expected: mekeneyāte baʼahunu saʻāte baregeṭañenate senete qwāneqwā baʻālame lāye
+  - source: እንዳለ ማወቅ አስቸጋሪ ነው
+    expected: ʼenedāla māwaqe ʼasečagāri nawe
+  - source: አሰላ
+    expected: ʼasalā
+  - source: አሶሳ
+    expected: ʼasosā
+  - source: አንኮበር
+    expected: ʼanekobare
+  - source: አክሱም
+    expected: ʼakesume
+  - source: አዋሳ
+    expected: ʼawāsā
+  - source: አዲስ ዘመን (ከተማ)
+    expected: ʼadise zamane (katamā)
+  - source: አዲግራት
+    expected: ʼadigerāte
+  - source: አዳማ
+    expected: ʼadāmā
+  - source: ደምበጫ
+    expected: damebaċā
+  - source: ደርባ
+    expected: darebā
+  - source: ደብረ ማርቆስ
+    expected: dabera māreqose
+  - source: ደብረ ብርሃን
+    expected: dabera berehāne
+  - source: ደብረ ታቦር (ከተማ)
+    expected: dabera tābore (katamā)
+  - source: ደብረ ዘይት
+    expected: dabera zayete
+  - source: ደገሃቡር
+    expected: dagahābure
+  - source: ወልቂጤ
+    expected: waleqiṭé
+  - source: ወልወል
+    expected: walewale
+  - source: ወልደያ
+    expected: waledayā
+  - source: ናይሎ ሳህራን
+    expected: nāyelo sāherāne
+  - source: አኙዋክኛ
+    expected: ʼañuwākeñā
+  - source: ኡዱክኛ
+    expected: ʼudukeñā
+  - source: ኦፓኛ
+    expected: ʼopāñā
+  - source: ጉምዝኛ
+    expected: gumezeñā
+  - source: አፋርኛ
+    expected: ʼafāreñā
+  - source: አላባኛ
+    expected: ʼalābāñā
+  - source: አርቦርኛ
+    expected: ʼareboreñā
+  - source: ባይሶኛ
+    expected: bāyesoñā
+  - source: ቡሳኛ
+    expected: busāñā
+  - source: ራስ ዓሊ (ትልቁ) ፬
+    expected: rāse ʻāli (telequ) 4
+  - source: ራስ ዓሊጋዝ ፭
+    expected: rāse ʻāligāze 5
+  - source: ራስ ዐሥራትና ፮
+    expected: rāse ʻaśerātenā 6
+  - source: ራስ ጉግሣ ፳፮
+    expected: rāse gugeśā 206
+  - source: ራስ ይማም ፪
+    expected: rāse yemāme 2
+  - source: ራስ ማርዬ ፫
+    expected: rāse māreyé 3
+  - source: ራስ ዶሪ ፫ ወር
+    expected: rāse dori 3 ware
+  - source: ራስ ዓሊ (ትንሹ) ፳
+    expected: rāse ʻāli (tenešu) 20
+  - source: ዓፄ ቴዎድሮስ ፲፭
+    expected: ʻāṡé téwoderose 105
+  - source: ዳግማዊ ዓጼ ተክለ ጊዮርጊስ ፫
+    expected: dāgemāwi ʻāṣé takela giyoregise 3
+  - source: ዓፄ ዮሐንስ ፲፰
+    expected: ʻāṡé yoḥanese 108
+  - source: ዳግማዊ ዓጼ ምኒልክ ፳፬
+    expected: dāgemāwi ʻāṣé menileke 204
+  - source: ልጅ ኢያሱ ፫
+    expected: leǧe ʼiyāsu 3
+  - source: ንግሥት ዘውዲቱ ፲፫
+    expected: negeśete zaweditu 103
+  - source: ቀዳማዊ ኃይለ ሥላሴ
+    expected: qadāmāwi hāyela śelāsé
+
+
+map:
+  characters:
+    "\u1200": "ha"
+    "\u1201": "hu"
+    "\u1202": "hi"
+    "\u1203": "hā"
+    "\u1204": "hé"
+    "\u1205":
+      - "he"
+      - "h"
+    "\u1206": "ho"
+    "\u1208": "la"
+    "\u1209": "lu"
+    "\u120A": "li"
+    "\u120B": "lā"
+    "\u120C": "lé"
+    "\u120D":
+      - "le"
+      - "l"
+    "\u120E": "lo"
+    "\u1210": "ḥa"
+    "\u1211": "ḥu"
+    "\u1212": "ḥi"
+    "\u1213": "ḥā"
+    "\u1214": "ḥé"
+    "\u1215":
+      - "ḥe"
+      - "ḥ"
+    "\u1216": "ḥo"
+    "\u1218": "ma"
+    "\u1219": "mu"
+    "\u121A": "mi"
+    "\u121B": "mā"
+    "\u121C": "mé"
+    "\u121D":
+      - "me"
+      - "m"
+    "\u121E": "mo"
+    "\u1220": "śa"
+    "\u1221": "śu"
+    "\u1222": "śi"
+    "\u1223": "śā"
+    "\u1224": "śé"
+    "\u1225":
+      - "śe"
+      - "ś"
+    "\u1226": "śo"
+    "\u1228": "ra"
+    "\u1229": "ru"
+    "\u122A": "ri"
+    "\u122B": "rā"
+    "\u122C": "ré"
+    "\u122D":
+      - "re"
+      - "r"
+    "\u122E": "ro"
+    "\u1230": "sa"
+    "\u1231": "su"
+    "\u1232": "si"
+    "\u1233": "sā"
+    "\u1234": "sé"
+    "\u1235":
+      - "se"
+      - "s"
+    "\u1236": "so"
+    "\u1238": "ša"
+    "\u1239": "šu"
+    "\u123A": "ši"
+    "\u123B": "šā"
+    "\u123C": "šé"
+    "\u123D":
+      - "še"
+      - "š"
+    "\u123E": "šo"
+    "\u1240": "qa"
+    "\u1241": "qu"
+    "\u1242": "qi"
+    "\u1243": "qā"
+    "\u1244": "qé"
+    "\u1245":
+      - "qe"
+      - "q"
+    "\u1246": "qo"
+    "\u1260": "ba"
+    "\u1261": "bu"
+    "\u1262": "bi"
+    "\u1263": "bā"
+    "\u1264": "bé"
+    "\u1265":
+      - "be"
+      - "b"
+    "\u1266": "bo"
+    "\u1270": "ta"
+    "\u1271": "tu"
+    "\u1272": "ti"
+    "\u1273": "tā"
+    "\u1274": "té"
+    "\u1275":
+      - "te"
+      - "t"
+    "\u1276": "to"
+    "\u1278": "ča"
+    "\u1279": "ču"
+    "\u127A": "či"
+    "\u127B": "čā"
+    "\u127C": "čé"
+    "\u127D":
+      - "če"
+      - "č"
+    "\u127E": "čo"
+    "\u1280": "ha"
+    "\u1281": "hu"
+    "\u1282": "hi"
+    "\u1283": "hā"
+    "\u1284": "hé"
+    "\u1285":
+      - "he"
+      - "h"
+    "\u1286": "ho"
+    "\u1290": "na"
+    "\u1291": "nu"
+    "\u1292": "ni"
+    "\u1293": "nā"
+    "\u1294": "né"
+    "\u1295":
+      - "ne"
+      - "n"
+    "\u1296": "no"
+    "\u1298": "ña"
+    "\u1299": "ñu"
+    "\u129A": "ñi"
+    "\u129B": "ñā"
+    "\u129C": "ñé"
+    "\u129D":
+      - "ñe"
+      - "ñ"
+    "\u129E": "ño"
+    "\u12A0": "ʼa"
+    "\u12A1": "ʼu"
+    "\u12A2": "ʼi"
+    "\u12A3": "ʼā"
+    "\u12A4": "ʼé"
+    "\u12A5": "ʼe"
+    "\u12A6": "ʼo"
+    "\u12A8": "ka"
+    "\u12A9": "ku"
+    "\u12AA": "ki"
+    "\u12AB": "kā"
+    "\u12AC": "ké"
+    "\u12AD":
+      - "ke"
+      - "k"
+    "\u12AE": "ko"
+    "\u12B8": "xa"
+    "\u12B9": "xu"
+    "\u12BA": "xi"
+    "\u12BB": "xā"
+    "\u12BC": "xé"
+    "\u12BD":
+      - "xe"
+      - "x"
+    "\u12BE": "xo"
+    "\u12C8": "wa"
+    "\u12C9": "wu"
+    "\u12CA": "wi"
+    "\u12CB": "wā"
+    "\u12CC": "wé"
+    "\u12CD":
+      - "we"
+      - "w"
+    "\u12CE": "wo"
+    "\u12D0": "ʻa"
+    "\u12D1": "ʻu"
+    "\u12D2": "ʻi"
+    "\u12D3": "ʻā"
+    "\u12D4": "ʻé"
+    "\u12D5": "ʻe"
+    "\u12D6": "ʻo"
+    "\u12D8": "za"
+    "\u12D9": "zu"
+    "\u12DA": "zi"
+    "\u12DB": "zā"
+    "\u12DC": "zé"
+    "\u12DD":
+      - "ze"
+      - "z"
+    "\u12DE": "zo"
+    "\u12E0": "ža"
+    "\u12E1": "žu"
+    "\u12E2": "ži"
+    "\u12E3": "žā"
+    "\u12E4": "žé"
+    "\u12E5":
+      - "že"
+      - "ž"
+    "\u12E6": "žo"
+    "\u12E8": "ya"
+    "\u12E9": "yu"
+    "\u12EA": "yi"
+    "\u12EB": "yā"
+    "\u12EC": "yé"
+    "\u12ED":
+      - "ye"
+      - "y"
+    "\u12EE": "yo"
+    "\u12F0": "da"
+    "\u12F1": "du"
+    "\u12F2": "di"
+    "\u12F3": "dā"
+    "\u12F4": "dé"
+    "\u12F5":
+      - "de"
+      - "d"
+    "\u12F6": "do"
+    "\u1300": "ǧa"
+    "\u1301": "ǧu"
+    "\u1302": "ǧi"
+    "\u1303": "ǧā"
+    "\u1304": "ǧé"
+    "\u1305":
+      - "ǧe"
+      - "ǧ"
+    "\u1306": "ǧo"
+    "\u1308": "ga"
+    "\u1309": "gu"
+    "\u130A": "gi"
+    "\u130B": "gā"
+    "\u130C": "gé"
+    "\u130D":
+      - "ge"
+      - "g"
+    "\u130E": "go"
+    "\u1320": "ṭa"
+    "\u1321": "ṭu"
+    "\u1322": "ṭi"
+    "\u1323": "ṭā"
+    "\u1324": "ṭé"
+    "\u1325":
+      - "ṭe"
+      - "ṭ"
+    "\u1326": "ṭo"
+    "\u1328": "ċa"
+    "\u1329": "ċu"
+    "\u132A": "ċi"
+    "\u132B": "ċā"
+    "\u132C": "ċé"
+    "\u132D":
+      - "ċe"
+      - "ċ"
+    "\u132E": "ċo"
+    "\u1330": "p̣a"
+    "\u1331": "p̣u"
+    "\u1332": "p̣i"
+    "\u1333": "p̣ā"
+    "\u1334": "p̣é"
+    "\u1335":
+      - "p̣e"
+      - s"p̣"
+    "\u1336": "p̣o"
+    "\u1338": "ṣa"
+    "\u1339": "ṣu"
+    "\u133A": "ṣi"
+    "\u133B": "ṣā"
+    "\u133C": "ṣé"
+    "\u133D":
+      - "ṣe"
+      - "ṣ"
+    "\u133E": "ṣo"
+    "\u1340": "ṡa"
+    "\u1341": "ṡu"
+    "\u1342": "ṡi"
+    "\u1343": "ṡā"
+    "\u1344": "ṡé"
+    "\u1345":
+      - "ṡe"
+      - "ṡ"
+    "\u1346": "ṡo"
+    "\u1348": "fa"
+    "\u1349": "fu"
+    "\u134A": "fi"
+    "\u134B": "fā"
+    "\u134C": "fé"
+    "\u134D":
+      - "fe"
+      - "f"
+    "\u134E": "fo"
+    "\u1350": "pa"
+    "\u1351": "pu"
+    "\u1352": "pi"
+    "\u1353": "pā"
+    "\u1354": "pé"
+    "\u1355":
+      - "pe"
+      - "p"
+    "\u1356": "po"
+    "\u1268": "va"
+    "\u1269": "vu"
+    "\u126A": "vi"
+    "\u126B": "vā"
+    "\u126C": "vé"
+    "\u126D":
+      - "ve"
+      - "v"
+    "\u126E": "vo"
+
+    "\u1248": "qwa"
+    "\u124A": "qwi"
+    "\u124B": "qwā"
+    "\u124C": "qwé"
+    "\u124D": "qwe"
+    "\u1288": "hwa"
+    "\u128A": "hwi"
+    "\u128B": "hwā"
+    "\u128C": "hwé"
+    "\u128D": "hwe"
+    "\u12B0": "kwa"
+    "\u12B2": "kwi"
+    "\u12B3": "kwā"
+    "\u12B4": "kwé"
+    "\u12B5": "kwe"
+    "\u1310": "gwa"
+    "\u1312": "gwi"
+    "\u1313": "gwā"
+    "\u1314": "gwé"
+    "\u1315": "gwe"
+
+    "\u120F": "lwa"
+    "\u1267": "bwa"
+    "\u12DF": "zwa"
+    "\u1327": "ṭwā"
+    "\u121F": "mwa"
+    "\u2D81": "mwa"
+    "\u1277": "twa"
+    "\u12E7": "žwa"
+    "\u132F": "ċwā"
+    "\u122F": "rwa"
+    "\u127F": "čwa"
+    "\u12EF": "ywa"
+    "\u133F": "ṣwā"
+    "\u1237": "swa"
+    "\u1297": "nwa"
+    "\u12F7": "dwa"
+    "\u134F": "fwā"
+    "\u123F": "šwa"
+    "\u129F": "ñwa"
+    "\u1307": "ǧwa"
+
+    "\u1359": "mya"
+    "\u135A": "fya"
+
+    "\u12A7": "ă"
+
+    "\u1369": "1"
+    "\u136A": "2"
+    "\u136B": "3"
+    "\u136C": "4"
+    "\u136D": "5"
+    "\u136E": "6"
+    "\u136F": "7"
+    "\u1370": "8"
+    "\u1371": "9"
+    "\u1372": "10"
+    "\u1373": "20"
+    "\u1374": "30"
+    "\u1375": "40"
+    "\u1376": "50"
+    "\u1377": "60"
+    "\u1378": "70"
+    "\u1379": "80"
+    "\u137A": "90"
+    "\u137B": "100"