interscript 0.1.5 → 0.1.6

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (116) hide show
  1. checksums.yaml +4 -4
  2. data/lib/__pycache__/g2pwrapper.cpython-38.pyc +0 -0
  3. data/lib/interscript.rb +5 -1
  4. data/lib/interscript/fs.rb +3 -1
  5. data/lib/interscript/mapping.rb +2 -2
  6. data/lib/interscript/opal.rb +5 -1
  7. data/lib/interscript/opal/maps.js.erb +7 -4
  8. data/lib/interscript/version.rb +1 -1
  9. data/maps/acadsin-zho-Hani-Latn-2002.yaml +1 -1
  10. data/maps/alalc-amh-Ethi-Latn-1997.yaml +509 -0
  11. data/maps/alalc-amh-Ethi-Latn-2011.yaml +138 -0
  12. data/maps/alalc-ara-Arab-Latn-1997.yaml +1283 -0
  13. data/maps/alalc-asm-Deva-Latn-1997.yaml +159 -0
  14. data/maps/alalc-aze-Cyrl-Latn-1997.yaml +1 -1
  15. data/maps/{alalc-bel-cyrl-latn-1997.yaml → alalc-bel-Cyrl-Latn-1997.yaml} +2 -2
  16. data/maps/alalc-ell-Grek-Latn-1997.yaml +2 -3
  17. data/maps/alalc-ell-Grek-Latn-2010.yaml +2 -3
  18. data/maps/alalc-hin-Deva-Latn-2020.yaml +159 -0
  19. data/maps/alalc-kat-Geok-Latn-1997.yaml +1 -2
  20. data/maps/alalc-kor-Hang-Latn-1997.yaml +1 -1
  21. data/maps/alalc-mar-Deva-Latn-1997.yaml +170 -0
  22. data/maps/{alalc-mkd-cyrl-latn-1997.yaml → alalc-mkd-Cyrl-Latn-1997.yaml} +0 -0
  23. data/maps/alalc-pan-Deva-Latn-1997.yaml +237 -0
  24. data/maps/alalc-rus-Cyrl-Latn-1997.yaml +1 -2
  25. data/maps/alalc-rus-Cyrl-Latn-2012.yaml +2 -2
  26. data/maps/{alalc-srp-cyrl-latn-2013.yaml → alalc-srp-Cyrl-Latn-2013.yaml} +0 -0
  27. data/maps/bas-rus-Cyrl-Latn-2017-bss.yaml +0 -1
  28. data/maps/bgn-jpn-Hrkt-Latn-1962.yaml +0 -2
  29. data/maps/bgn-kor-Hang-Latn-1943.yaml +1 -1
  30. data/maps/bgn-kor-Kore-Latn-1943.yaml +1 -1
  31. data/maps/bgnpcgn-amh-Ethi-Latn-1967.yaml +528 -0
  32. data/maps/bgnpcgn-ara-Arab-Latn-1956.yaml +592 -0
  33. data/maps/bgnpcgn-bak-Cyrl-Latn-2007.yaml +5 -5
  34. data/maps/{bgnpcgn-bel-cyrl-latn-1979.yaml → bgnpcgn-bel-Cyrl-Latn-1979.yaml} +0 -0
  35. data/maps/bgnpcgn-ell-Grek-Latn-1962.yaml +3 -4
  36. data/maps/bgnpcgn-ell-Grek-Latn-1996.yaml +0 -1
  37. data/maps/bgnpcgn-kat-Geor-Latn-2009.yaml +0 -1
  38. data/maps/bgnpcgn-kor-Hang-Latn-kn-1945.yaml +17 -17
  39. data/maps/bgnpcgn-kor-Hang-Latn-rok-2011.yaml +2 -2
  40. data/maps/bgnpcgn-kor-Kore-Latn-rok-2011.yaml +2 -2
  41. data/maps/bgnpcgn-mkd-Cyrl-Latn-2013.yaml +1 -1
  42. data/maps/bgnpcgn-nep-Deva-Latn-2011.yaml +200 -0
  43. data/maps/bgnpcgn-per-Arab-Latn-1956.yaml +0 -1
  44. data/maps/bgnpcgn-ukr-Cyrl-Latn-1965.yaml +0 -1
  45. data/maps/bis-asm-Beng-Latn-13194-1991.yaml +159 -0
  46. data/maps/bis-ben-Beng-Latn-13194-1991.yaml +156 -0
  47. data/maps/bis-dev-Deva-Latn-13194-1991.yaml +184 -0
  48. data/maps/bis-gjr-Gujr-Latn-13194-1991.yaml +166 -0
  49. data/maps/bis-knd-Knda-Latn-13194-1991.yaml +173 -0
  50. data/maps/bis-mlm-Mlym-Latn-13194-1991.yaml +176 -0
  51. data/maps/bis-ori-Orya-Latn-13194-1991.yaml +160 -0
  52. data/maps/bis-pnj-Guru-Latn-13194-1991.yaml +175 -0
  53. data/maps/bis-tel-Telu-Latn-13194-1991.yaml +170 -0
  54. data/maps/bis-tml-Taml-Latn-13194-1991.yaml +155 -0
  55. data/maps/by-bel-Cyrl-Latn-1998.yaml +4 -4
  56. data/maps/by-bel-Cyrl-Latn-2007.yaml +3 -3
  57. data/maps/dos-nep-Deva-Latn-1997.yaml +33 -0
  58. data/maps/elot-ell-Grek-Latn-743-1982-tl.yaml +4 -5
  59. data/maps/elot-ell-Grek-Latn-743-1982-ts.yaml +4 -5
  60. data/maps/elot-ell-Grek-Latn-743-2001-tl.yaml +0 -1
  61. data/maps/elot-ell-Grek-Latn-743-2001-ts.yaml +0 -1
  62. data/maps/ggg-kat-Geor-Latn-2002.yaml +0 -1
  63. data/maps/{gki-bel-cyrl-latn-1992.yaml → gki-bel-Cyrl-Latn-1992.yaml} +1 -1
  64. data/maps/{gki-bel-cyrl-latn-2000.yaml → gki-bel-Cyrl-Latn-2000.yaml} +1 -1
  65. data/maps/{gost-rus-cyrl-latn-16876-71-1983.yaml → gost-rus-Cyrl-Latn-16876-71-1983.yaml} +1 -1
  66. data/maps/icao-bel-Cyrl-Latn-9303.yaml +0 -5
  67. data/maps/icao-bul-Cyrl-Latn-9303.yaml +0 -4
  68. data/maps/icao-per-Arab-Latn-9303.yaml +0 -1
  69. data/maps/icao-rus-Cyrl-Latn-9303.yaml +0 -1
  70. data/maps/icao-ukr-Cyrl-Latn-9303.yaml +0 -1
  71. data/maps/iso-ara-Arab-Latn-233-1984.yaml +323 -0
  72. data/maps/iso-ell-Grek-Latn-843-1997-t1.yaml +4 -5
  73. data/maps/iso-ell-Grek-Latn-843-1997-t2.yaml +1 -2
  74. data/maps/iso-rus-Cyrl-Latn-9-1995.yaml +0 -1
  75. data/maps/iso-tha-Thai-Latn-11940-1998.yaml +1 -1
  76. data/maps/kp-kor-Hang-Latn-2002.yaml +4 -4
  77. data/maps/lshk-yue-Hani-Latn-jyutping-1993.yaml +2 -2
  78. data/maps/mvd-bel-Cyrl-Latn-2008.yaml +4 -4
  79. data/maps/mvd-bel-Cyrl-Latn-2010.yaml +4 -4
  80. data/maps/mvd-rus-Cyrl-Latn-2008.yaml +1 -2
  81. data/maps/mvd-rus-Cyrl-Latn-2010.yaml +4 -4
  82. data/maps/nil-kor-Hang-Hang-jamo.yaml +3 -3
  83. data/maps/odni-aze-Cyrl-Latn-2015.yaml +1 -1
  84. data/maps/odni-bel-Cyrl-Latn-2015.yaml +1 -1
  85. data/maps/odni-bul-Cyrl-Latn-2015.yaml +3 -3
  86. data/maps/odni-hin-Deva-Latn-2015.yaml +258 -0
  87. data/maps/odni-kat-Geor-Latn-2015.yaml +0 -1
  88. data/maps/{odni-mkd-cyrl-latn-2015.yaml → odni-mkd-Cyrl-Latn-2015.yaml} +0 -0
  89. data/maps/odni-rus-Cyrl-Latn-2015.yaml +1 -1
  90. data/maps/odni-srp-Cyrl-Latn-2015.yaml +2 -2
  91. data/maps/odni-urd-Arab-Latn-2015.yaml +221 -0
  92. data/maps/odni-uzb-Cyrl-Latn-2015.yaml +1 -2
  93. data/maps/royin-tha-Thai-Latn-1939-generic.yaml +4 -4
  94. data/maps/royin-tha-Thai-Latn-1968.yaml +4 -4
  95. data/maps/royin-tha-Thai-Latn-1999-chained.yaml +4 -4
  96. data/maps/royin-tha-Thai-Latn-1999.yaml +3 -3
  97. data/maps/{ses-ara-arab-latn-1930.yaml → ses-ara-Arab-Latn-1930.yaml} +7 -3
  98. data/maps/un-ara-Arab-Latn-1971.yaml +16 -4
  99. data/maps/un-ara-Arab-Latn-1972.yaml +14 -7
  100. data/maps/un-ara-Arab-Latn-2017.yaml +56 -19
  101. data/maps/un-bel-Cyrl-Latn-2007.yaml +3 -3
  102. data/maps/un-ell-Grek-Latn-1987-tl.yaml +1 -2
  103. data/maps/un-ell-Grek-Latn-1987-ts.yaml +0 -1
  104. data/maps/un-ell-Grek-Latn-phonetic-1987.yaml +42 -42
  105. data/maps/un-mon-Mong-Latn-2013.yaml +9 -3
  106. data/maps/un-nep-Deva-Latn-1972.yaml +163 -0
  107. data/maps/un-rus-Cyrl-Latn-1987.yaml +1 -1
  108. data/maps/{un-ukr-cyrl-latn-1998.yaml → un-ukr-Cyrl-Latn-1998.yaml} +1 -1
  109. data/maps/ungegn-amh-Ethi-Latn-2016.yaml +575 -0
  110. data/maps/var-kor-Hang-Latn-mr-1939.yaml +2 -2
  111. data/maps/var-kor-Kore-Hang-2013.yaml +1 -1
  112. data/maps/var-kor-Kore-Latn-mr-1939.yaml +1 -2
  113. data/maps/var-tha-Thai-Thai-phonemic.yaml +5 -5
  114. data/maps/var-tha-Thai-Zsym-ipa.yaml +12 -12
  115. data/maps/var-zho-Hani-Latn-1979.yaml +7 -7
  116. metadata +41 -15
checksums.yaml CHANGED
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  SHA256:
3
- metadata.gz: 3eb5c858ad7332995b5f8ed1c7766305f6be5fb5524e91de7df1d0d3d770a3ee
4
- data.tar.gz: 80be952635dd65cf1e23107f70c44557ec9711590759e610a042f2c746f67d4e
3
+ metadata.gz: 7557ea1ca381562c61be7dbbeab2ea4adc42ef57ce857ef86acd62a08e5ce588
4
+ data.tar.gz: 5099d4a7bf07817155d620db716af452607b8565f4e08d37aa805f486556e0e1
5
5
  SHA512:
6
- metadata.gz: bdb63b2ea8e6337a31879e7c1331f6e9a95dd205580bc8656cdb806e624b7cf946c17d300125470494c1643332b3f5f1a3d713a040bd6ee907dcba5d189db95e
7
- data.tar.gz: 1aa1b254b253f148b4a5cc316027d3e67ca1ba68e866ebedbc025dc9c5f5c78680b12c24bc52177d65b2c9a6dd262ee97e34f52b4d32f83dfc9c232dc2c2d0de
6
+ metadata.gz: f8a738a34aba269c0a01b4d123f01cc423a3c296541f302f65a6043bd2b618ef2c702f32d469866ee7c488624985b45ed2eee2ad9bbe2955f230b5b45472c364
7
+ data.tar.gz: a3a66af7fcb9d8c82bcf927b17ea68686298bccec2f6e8dfac25796e7013b36f68fa9f130bd2fb9e7bc214a528896cafdd41a2f464886f09c504f376c19065d8
@@ -109,7 +109,11 @@ module Interscript
109
109
  mapping.postrules.each do |r|
110
110
  next unless output
111
111
  re = mkregexp(r["pattern"])
112
- output = output.gsub(re, r["result"])
112
+ output = if r["result"] == "upcase"
113
+ output.gsub(re, &:upcase)
114
+ else
115
+ output.gsub(re, r["result"])
116
+ end
113
117
  end
114
118
 
115
119
  return unless output
@@ -1,3 +1,5 @@
1
+ require 'pathname'
2
+
1
3
  module Interscript
2
4
  module Fs
3
5
  ALPHA_REGEXP = '[[:alpha:]]'
@@ -66,4 +68,4 @@ module Interscript
66
68
  end
67
69
 
68
70
  end
69
- end
71
+ end
@@ -1,5 +1,5 @@
1
1
  require 'rambling-trie'
2
- require 'yaml'
2
+ require 'yaml' unless RUBY_ENGINE == 'opal'
3
3
  require 'json'
4
4
 
5
5
  module Interscript
@@ -139,4 +139,4 @@ module Interscript
139
139
  dictionary_trie.concat dictionary.keys
140
140
  end
141
141
  end
142
- end
142
+ end
@@ -19,5 +19,9 @@ module Interscript
19
19
  string
20
20
  end
21
21
 
22
+ def load_map_json(name, json)
23
+ `Opal.global.InterscriptMaps[#{name}] = #{json}`
24
+ end
25
+
22
26
  end
23
- end
27
+ end
@@ -1,7 +1,10 @@
1
+ <%# A placeholder for the maps to be loaded %>
1
2
  Opal.global.InterscriptMaps = {
2
3
  <% Dir['maps/*.yaml'].each do |yaml_file| %>
3
- "<%= File.basename(yaml_file, ".yaml") %>": <%=
4
- JSON.dump(Interscript::OpalMapTranslate.translate_regexp(JSON.dump(YAML.load(File.read(yaml_file)))))
5
- %>,
4
+ "<%= File.basename(yaml_file, ".yaml") %>": null,
6
5
  <% end %>
7
- }
6
+ }
7
+
8
+ <%#=
9
+ JSON.dump(Interscript::OpalMapTranslate.translate_regexp(JSON.dump(YAML.load(File.read(yaml_file)))))
10
+ %>
@@ -1,3 +1,3 @@
1
1
  module Interscript
2
- VERSION = "0.1.5"
2
+ VERSION = "0.1.6"
3
3
  end
@@ -5,7 +5,7 @@ language: zho
5
5
  source_script: Hani
6
6
  destination_script: Latn
7
7
  name: Chinese Tongyong Pinyin Academica Sinica 2002 System
8
- url:
8
+ url:
9
9
  description: Chinese Tongyong Pinyin Academica Sinica 2002 System
10
10
 
11
11
  tests:
@@ -0,0 +1,509 @@
1
+ ---
2
+ authority_id: alalc
3
+ id: 1997
4
+ language: amh
5
+ source_script: Ethi
6
+ destination_script: Latn
7
+ name: ALA-LC Romanization Table -- Amharic (1997)
8
+ url: http://catdir.loc.gov/catdir/cpso/romanization/amharic.pdf
9
+ creation_date: 1997
10
+ description: |
11
+ ALA-LC Romanization Tables: Transliteration Schemes for Non-Roman Scripts. Randal K. Berry (ed.). Library of Congress, 1997.
12
+
13
+ notes:
14
+ - 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.
15
+
16
+ tests:
17
+ - source: የዜግነት ክብር በ ኢትዮጵያችን ጸንቶ
18
+ expected: yazégenate kebere ba ʼiteyop̣eyāčene ṣaneto
19
+ - source: ታየ ሕዝባዊነት ዳር እስከዳር በርቶ
20
+ expected: tāya ḥezebāwinate dāre ʼesekadāre bareto
21
+ - source: ለሰላም ለፍትህ ለሕዝቦች ነጻነት
22
+ expected: lasalāme lafetehe laḥezeboče naṣānate
23
+ - source: በእኩልነት በፍቅር ቆመናል ባንድነት
24
+ expected: baʼekulenate bafeqere qomanāle bānedenate
25
+ - source: መሠረተ ፅኑ ሰብዕናን ያልሻርን
26
+ expected: maśarata ṡenu sabeʻenāne yālešārene
27
+ - source: ሕዝቦች ነን ለሥራ በሥራ የኖርን
28
+ expected: ḥezeboče nane laśerā baśerā yanorene
29
+ - source: ድንቅ የባህል መድረክ ያኩሪ ቅርስ ባለቤት
30
+ expected: deneqe yabāhele maderake yākuri qerese bālabéte
31
+ - source: የተፈጥሮ ጸጋ የጀግና ሕዝብ እናት
32
+ expected: yatafaṭero ṣagā yaǧagenā ḥezebe ʼenāte
33
+ - source: እንጠብቅሻለን አለብን አደራ
34
+ expected: ʼeneṭabeqešālane ʼalabene ʼadarā
35
+ - source: ኢትዮጵያችን ኑሪ እኛም ባንቺ እንኩራ
36
+ expected: ʼiteyop̣eyāčene nuri ʼeñāme bāneči ʼenekurā
37
+ - source: ቋንቋ የድምጽ፣ የምልክት ወይም የምስል ቅንብር ሆኖ
38
+ expected: qwāneqwā yademeṣe፣ yamelekete wayeme yamesele qenebere hono
39
+ - source: ለማሰብ ወይም የታሰበን ሃሳብ ለሌላ ለማስተላለፍ የሚረዳ መሳሪያ ነው
40
+ expected: lamāsabe wayeme yatāsabane hāsābe lalélā lamāsetalālafe yamiradā masāriyā nawe
41
+ - source: በአጭሩ ቋንቋ የምልክቶች ስርዓትና እኒህን ምልክቶች ለማቀናበር
42
+ expected: baʼaċeru qwāneqwā yameleketoče sereʻātenā ʼenihene meleketoče lamāqanābare
43
+ - source: የሚያስፈልጉ ህጎች ጥንቅር ነው። ቋንቋወችን ለመፈረጅ እንዲሁም
44
+ expected: yamiyāsefalegu hegoče ṭeneqere nawe። qwāneqwāwačene lamafaraǧe ʼenedihume
45
+ - source: ለምክፈል የሚያስችሉ መስፈርቶችን ለማስቀመጥ ባለው ችግር
46
+ expected: lamekefale yamiyāsečelu masefaretočene lamāseqamaṭe bālawe čegere
47
+ - source: ምክንያት በአሁኑ ሰዓት በርግጠኝነት ስንት ቋንቋ በዓለም ላይ
48
+ expected: mekeneyāte baʼahunu saʻāte baregeṭañenate senete qwāneqwā baʻālame lāye
49
+ - source: እንዳለ ማወቅ አስቸጋሪ ነው
50
+ expected: ʼenedāla māwaqe ʼasečagāri nawe
51
+ - source: አሰላ
52
+ expected: ʼasalā
53
+ - source: አሶሳ
54
+ expected: ʼasosā
55
+ - source: አንኮበር
56
+ expected: ʼanekobare
57
+ - source: አክሱም
58
+ expected: ʼakesume
59
+ - source: አዋሳ
60
+ expected: ʼawāsā
61
+ - source: አዲስ ዘመን (ከተማ)
62
+ expected: ʼadise zamane (katamā)
63
+ - source: አዲግራት
64
+ expected: ʼadigerāte
65
+ - source: አዳማ
66
+ expected: ʼadāmā
67
+ - source: ደምበጫ
68
+ expected: damebaċā
69
+ - source: ደርባ
70
+ expected: darebā
71
+ - source: ደብረ ማርቆስ
72
+ expected: dabera māreqose
73
+ - source: ደብረ ብርሃን
74
+ expected: dabera berehāne
75
+ - source: ደብረ ታቦር (ከተማ)
76
+ expected: dabera tābore (katamā)
77
+ - source: ደብረ ዘይት
78
+ expected: dabera zayete
79
+ - source: ደገሃቡር
80
+ expected: dagahābure
81
+ - source: ወልቂጤ
82
+ expected: waleqiṭé
83
+ - source: ወልወል
84
+ expected: walewale
85
+ - source: ወልደያ
86
+ expected: waledayā
87
+ - source: ናይሎ ሳህራን
88
+ expected: nāyelo sāherāne
89
+ - source: አኙዋክኛ
90
+ expected: ʼañuwākeñā
91
+ - source: ኡዱክኛ
92
+ expected: ʼudukeñā
93
+ - source: ኦፓኛ
94
+ expected: ʼopāñā
95
+ - source: ጉምዝኛ
96
+ expected: gumezeñā
97
+ - source: አፋርኛ
98
+ expected: ʼafāreñā
99
+ - source: አላባኛ
100
+ expected: ʼalābāñā
101
+ - source: አርቦርኛ
102
+ expected: ʼareboreñā
103
+ - source: ባይሶኛ
104
+ expected: bāyesoñā
105
+ - source: ቡሳኛ
106
+ expected: busāñā
107
+ - source: ራስ ዓሊ (ትልቁ) ፬
108
+ expected: rāse ʻāli (telequ) 4
109
+ - source: ራስ ዓሊጋዝ ፭
110
+ expected: rāse ʻāligāze 5
111
+ - source: ራስ ዐሥራትና ፮
112
+ expected: rāse ʻaśerātenā 6
113
+ - source: ራስ ጉግሣ ፳፮
114
+ expected: rāse gugeśā 206
115
+ - source: ራስ ይማም ፪
116
+ expected: rāse yemāme 2
117
+ - source: ራስ ማርዬ ፫
118
+ expected: rāse māreyé 3
119
+ - source: ራስ ዶሪ ፫ ወር
120
+ expected: rāse dori 3 ware
121
+ - source: ራስ ዓሊ (ትንሹ) ፳
122
+ expected: rāse ʻāli (tenešu) 20
123
+ - source: ዓፄ ቴዎድሮስ ፲፭
124
+ expected: ʻāṡé téwoderose 105
125
+ - source: ዳግማዊ ዓጼ ተክለ ጊዮርጊስ ፫
126
+ expected: dāgemāwi ʻāṣé takela giyoregise 3
127
+ - source: ዓፄ ዮሐንስ ፲፰
128
+ expected: ʻāṡé yoḥanese 108
129
+ - source: ዳግማዊ ዓጼ ምኒልክ ፳፬
130
+ expected: dāgemāwi ʻāṣé menileke 204
131
+ - source: ልጅ ኢያሱ ፫
132
+ expected: leǧe ʼiyāsu 3
133
+ - source: ንግሥት ዘውዲቱ ፲፫
134
+ expected: negeśete zaweditu 103
135
+ - source: ቀዳማዊ ኃይለ ሥላሴ
136
+ expected: qadāmāwi hāyela śelāsé
137
+
138
+
139
+ map:
140
+ characters:
141
+ "\u1200": "ha"
142
+ "\u1201": "hu"
143
+ "\u1202": "hi"
144
+ "\u1203": "hā"
145
+ "\u1204": "hé"
146
+ "\u1205":
147
+ - "he"
148
+ - "h"
149
+ "\u1206": "ho"
150
+ "\u1208": "la"
151
+ "\u1209": "lu"
152
+ "\u120A": "li"
153
+ "\u120B": "lā"
154
+ "\u120C": "lé"
155
+ "\u120D":
156
+ - "le"
157
+ - "l"
158
+ "\u120E": "lo"
159
+ "\u1210": "ḥa"
160
+ "\u1211": "ḥu"
161
+ "\u1212": "ḥi"
162
+ "\u1213": "ḥā"
163
+ "\u1214": "ḥé"
164
+ "\u1215":
165
+ - "ḥe"
166
+ - "ḥ"
167
+ "\u1216": "ḥo"
168
+ "\u1218": "ma"
169
+ "\u1219": "mu"
170
+ "\u121A": "mi"
171
+ "\u121B": "mā"
172
+ "\u121C": "mé"
173
+ "\u121D":
174
+ - "me"
175
+ - "m"
176
+ "\u121E": "mo"
177
+ "\u1220": "śa"
178
+ "\u1221": "śu"
179
+ "\u1222": "śi"
180
+ "\u1223": "śā"
181
+ "\u1224": "śé"
182
+ "\u1225":
183
+ - "śe"
184
+ - "ś"
185
+ "\u1226": "śo"
186
+ "\u1228": "ra"
187
+ "\u1229": "ru"
188
+ "\u122A": "ri"
189
+ "\u122B": "rā"
190
+ "\u122C": "ré"
191
+ "\u122D":
192
+ - "re"
193
+ - "r"
194
+ "\u122E": "ro"
195
+ "\u1230": "sa"
196
+ "\u1231": "su"
197
+ "\u1232": "si"
198
+ "\u1233": "sā"
199
+ "\u1234": "sé"
200
+ "\u1235":
201
+ - "se"
202
+ - "s"
203
+ "\u1236": "so"
204
+ "\u1238": "ša"
205
+ "\u1239": "šu"
206
+ "\u123A": "ši"
207
+ "\u123B": "šā"
208
+ "\u123C": "šé"
209
+ "\u123D":
210
+ - "še"
211
+ - "š"
212
+ "\u123E": "šo"
213
+ "\u1240": "qa"
214
+ "\u1241": "qu"
215
+ "\u1242": "qi"
216
+ "\u1243": "qā"
217
+ "\u1244": "qé"
218
+ "\u1245":
219
+ - "qe"
220
+ - "q"
221
+ "\u1246": "qo"
222
+ "\u1260": "ba"
223
+ "\u1261": "bu"
224
+ "\u1262": "bi"
225
+ "\u1263": "bā"
226
+ "\u1264": "bé"
227
+ "\u1265":
228
+ - "be"
229
+ - "b"
230
+ "\u1266": "bo"
231
+ "\u1270": "ta"
232
+ "\u1271": "tu"
233
+ "\u1272": "ti"
234
+ "\u1273": "tā"
235
+ "\u1274": "té"
236
+ "\u1275":
237
+ - "te"
238
+ - "t"
239
+ "\u1276": "to"
240
+ "\u1278": "ča"
241
+ "\u1279": "ču"
242
+ "\u127A": "či"
243
+ "\u127B": "čā"
244
+ "\u127C": "čé"
245
+ "\u127D":
246
+ - "če"
247
+ - "č"
248
+ "\u127E": "čo"
249
+ "\u1280": "ha"
250
+ "\u1281": "hu"
251
+ "\u1282": "hi"
252
+ "\u1283": "hā"
253
+ "\u1284": "hé"
254
+ "\u1285":
255
+ - "he"
256
+ - "h"
257
+ "\u1286": "ho"
258
+ "\u1290": "na"
259
+ "\u1291": "nu"
260
+ "\u1292": "ni"
261
+ "\u1293": "nā"
262
+ "\u1294": "né"
263
+ "\u1295":
264
+ - "ne"
265
+ - "n"
266
+ "\u1296": "no"
267
+ "\u1298": "ña"
268
+ "\u1299": "ñu"
269
+ "\u129A": "ñi"
270
+ "\u129B": "ñā"
271
+ "\u129C": "ñé"
272
+ "\u129D":
273
+ - "ñe"
274
+ - "ñ"
275
+ "\u129E": "ño"
276
+ "\u12A0": "ʼa"
277
+ "\u12A1": "ʼu"
278
+ "\u12A2": "ʼi"
279
+ "\u12A3": "ʼā"
280
+ "\u12A4": "ʼé"
281
+ "\u12A5": "ʼe"
282
+ "\u12A6": "ʼo"
283
+ "\u12A8": "ka"
284
+ "\u12A9": "ku"
285
+ "\u12AA": "ki"
286
+ "\u12AB": "kā"
287
+ "\u12AC": "ké"
288
+ "\u12AD":
289
+ - "ke"
290
+ - "k"
291
+ "\u12AE": "ko"
292
+ "\u12B8": "xa"
293
+ "\u12B9": "xu"
294
+ "\u12BA": "xi"
295
+ "\u12BB": "xā"
296
+ "\u12BC": "xé"
297
+ "\u12BD":
298
+ - "xe"
299
+ - "x"
300
+ "\u12BE": "xo"
301
+ "\u12C8": "wa"
302
+ "\u12C9": "wu"
303
+ "\u12CA": "wi"
304
+ "\u12CB": "wā"
305
+ "\u12CC": "wé"
306
+ "\u12CD":
307
+ - "we"
308
+ - "w"
309
+ "\u12CE": "wo"
310
+ "\u12D0": "ʻa"
311
+ "\u12D1": "ʻu"
312
+ "\u12D2": "ʻi"
313
+ "\u12D3": "ʻā"
314
+ "\u12D4": "ʻé"
315
+ "\u12D5": "ʻe"
316
+ "\u12D6": "ʻo"
317
+ "\u12D8": "za"
318
+ "\u12D9": "zu"
319
+ "\u12DA": "zi"
320
+ "\u12DB": "zā"
321
+ "\u12DC": "zé"
322
+ "\u12DD":
323
+ - "ze"
324
+ - "z"
325
+ "\u12DE": "zo"
326
+ "\u12E0": "ža"
327
+ "\u12E1": "žu"
328
+ "\u12E2": "ži"
329
+ "\u12E3": "žā"
330
+ "\u12E4": "žé"
331
+ "\u12E5":
332
+ - "že"
333
+ - "ž"
334
+ "\u12E6": "žo"
335
+ "\u12E8": "ya"
336
+ "\u12E9": "yu"
337
+ "\u12EA": "yi"
338
+ "\u12EB": "yā"
339
+ "\u12EC": "yé"
340
+ "\u12ED":
341
+ - "ye"
342
+ - "y"
343
+ "\u12EE": "yo"
344
+ "\u12F0": "da"
345
+ "\u12F1": "du"
346
+ "\u12F2": "di"
347
+ "\u12F3": "dā"
348
+ "\u12F4": "dé"
349
+ "\u12F5":
350
+ - "de"
351
+ - "d"
352
+ "\u12F6": "do"
353
+ "\u1300": "ǧa"
354
+ "\u1301": "ǧu"
355
+ "\u1302": "ǧi"
356
+ "\u1303": "ǧā"
357
+ "\u1304": "ǧé"
358
+ "\u1305":
359
+ - "ǧe"
360
+ - "ǧ"
361
+ "\u1306": "ǧo"
362
+ "\u1308": "ga"
363
+ "\u1309": "gu"
364
+ "\u130A": "gi"
365
+ "\u130B": "gā"
366
+ "\u130C": "gé"
367
+ "\u130D":
368
+ - "ge"
369
+ - "g"
370
+ "\u130E": "go"
371
+ "\u1320": "ṭa"
372
+ "\u1321": "ṭu"
373
+ "\u1322": "ṭi"
374
+ "\u1323": "ṭā"
375
+ "\u1324": "ṭé"
376
+ "\u1325":
377
+ - "ṭe"
378
+ - "ṭ"
379
+ "\u1326": "ṭo"
380
+ "\u1328": "ċa"
381
+ "\u1329": "ċu"
382
+ "\u132A": "ċi"
383
+ "\u132B": "ċā"
384
+ "\u132C": "ċé"
385
+ "\u132D":
386
+ - "ċe"
387
+ - "ċ"
388
+ "\u132E": "ċo"
389
+ "\u1330": "p̣a"
390
+ "\u1331": "p̣u"
391
+ "\u1332": "p̣i"
392
+ "\u1333": "p̣ā"
393
+ "\u1334": "p̣é"
394
+ "\u1335":
395
+ - "p̣e"
396
+ - s"p̣"
397
+ "\u1336": "p̣o"
398
+ "\u1338": "ṣa"
399
+ "\u1339": "ṣu"
400
+ "\u133A": "ṣi"
401
+ "\u133B": "ṣā"
402
+ "\u133C": "ṣé"
403
+ "\u133D":
404
+ - "ṣe"
405
+ - "ṣ"
406
+ "\u133E": "ṣo"
407
+ "\u1340": "ṡa"
408
+ "\u1341": "ṡu"
409
+ "\u1342": "ṡi"
410
+ "\u1343": "ṡā"
411
+ "\u1344": "ṡé"
412
+ "\u1345":
413
+ - "ṡe"
414
+ - "ṡ"
415
+ "\u1346": "ṡo"
416
+ "\u1348": "fa"
417
+ "\u1349": "fu"
418
+ "\u134A": "fi"
419
+ "\u134B": "fā"
420
+ "\u134C": "fé"
421
+ "\u134D":
422
+ - "fe"
423
+ - "f"
424
+ "\u134E": "fo"
425
+ "\u1350": "pa"
426
+ "\u1351": "pu"
427
+ "\u1352": "pi"
428
+ "\u1353": "pā"
429
+ "\u1354": "pé"
430
+ "\u1355":
431
+ - "pe"
432
+ - "p"
433
+ "\u1356": "po"
434
+ "\u1268": "va"
435
+ "\u1269": "vu"
436
+ "\u126A": "vi"
437
+ "\u126B": "vā"
438
+ "\u126C": "vé"
439
+ "\u126D":
440
+ - "ve"
441
+ - "v"
442
+ "\u126E": "vo"
443
+
444
+ "\u1248": "qwa"
445
+ "\u124A": "qwi"
446
+ "\u124B": "qwā"
447
+ "\u124C": "qwé"
448
+ "\u124D": "qwe"
449
+ "\u1288": "hwa"
450
+ "\u128A": "hwi"
451
+ "\u128B": "hwā"
452
+ "\u128C": "hwé"
453
+ "\u128D": "hwe"
454
+ "\u12B0": "kwa"
455
+ "\u12B2": "kwi"
456
+ "\u12B3": "kwā"
457
+ "\u12B4": "kwé"
458
+ "\u12B5": "kwe"
459
+ "\u1310": "gwa"
460
+ "\u1312": "gwi"
461
+ "\u1313": "gwā"
462
+ "\u1314": "gwé"
463
+ "\u1315": "gwe"
464
+
465
+ "\u120F": "lwa"
466
+ "\u1267": "bwa"
467
+ "\u12DF": "zwa"
468
+ "\u1327": "ṭwā"
469
+ "\u121F": "mwa"
470
+ "\u2D81": "mwa"
471
+ "\u1277": "twa"
472
+ "\u12E7": "žwa"
473
+ "\u132F": "ċwā"
474
+ "\u122F": "rwa"
475
+ "\u127F": "čwa"
476
+ "\u12EF": "ywa"
477
+ "\u133F": "ṣwā"
478
+ "\u1237": "swa"
479
+ "\u1297": "nwa"
480
+ "\u12F7": "dwa"
481
+ "\u134F": "fwā"
482
+ "\u123F": "šwa"
483
+ "\u129F": "ñwa"
484
+ "\u1307": "ǧwa"
485
+
486
+ "\u1359": "mya"
487
+ "\u135A": "fya"
488
+
489
+ "\u12A7": "ă"
490
+
491
+ "\u1369": "1"
492
+ "\u136A": "2"
493
+ "\u136B": "3"
494
+ "\u136C": "4"
495
+ "\u136D": "5"
496
+ "\u136E": "6"
497
+ "\u136F": "7"
498
+ "\u1370": "8"
499
+ "\u1371": "9"
500
+ "\u1372": "10"
501
+ "\u1373": "20"
502
+ "\u1374": "30"
503
+ "\u1375": "40"
504
+ "\u1376": "50"
505
+ "\u1377": "60"
506
+ "\u1378": "70"
507
+ "\u1379": "80"
508
+ "\u137A": "90"
509
+ "\u137B": "100"