blackwinter-cmess 0.2.0

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 (49) hide show
  1. data/COPYING +676 -0
  2. data/ChangeLog +54 -0
  3. data/README +63 -0
  4. data/Rakefile +51 -0
  5. data/bin/bconv +130 -0
  6. data/bin/cinderella +190 -0
  7. data/bin/decode_entities +106 -0
  8. data/bin/guess_encoding +223 -0
  9. data/data/chartab.yaml +26724 -0
  10. data/data/csets/iso_8859-1.yaml +195 -0
  11. data/data/csets/iso_8859-15.yaml +204 -0
  12. data/data/csets/latin1.yaml +195 -0
  13. data/data/csets/unicode/basic_latin.yaml +97 -0
  14. data/data/csets/unicode/cyrillic-supplement.yaml +17 -0
  15. data/data/csets/unicode/cyrillic.yaml +256 -0
  16. data/data/csets/unicode/greek.yaml +129 -0
  17. data/data/csets/unicode/ipa_extensions.yaml +97 -0
  18. data/data/csets/unicode/latin-extended-c.yaml +18 -0
  19. data/data/csets/unicode/latin-extended-d.yaml +3 -0
  20. data/data/csets/unicode/latin_1_supplement.yaml +128 -0
  21. data/data/csets/unicode/latin_extended_a.yaml +129 -0
  22. data/data/csets/unicode/latin_extended_additional.yaml +247 -0
  23. data/data/csets/unicode/latin_extended_b.yaml +209 -0
  24. data/data/csets/unicode/letterlike_symbols.yaml +80 -0
  25. data/data/csets/unicode/spacing_modifier_letters.yaml +81 -0
  26. data/data/csets/utf-8.yaml +1504 -0
  27. data/data/csets/utf8.yaml +1504 -0
  28. data/data/test_chars.yaml +14 -0
  29. data/example/cinderella/crop +127 -0
  30. data/example/cinderella/crop_repaired +127 -0
  31. data/example/cinderella/empty6-slash.txt +1495 -0
  32. data/example/cinderella/empty6-slash_repaired.txt +1495 -0
  33. data/example/cinderella/pot +1368 -0
  34. data/example/guess_encoding/check_results +60 -0
  35. data/example/guess_encoding/de.utf-8.txt +10030 -0
  36. data/example/guess_encoding/en.utf-8.txt +10030 -0
  37. data/example/guess_encoding/fr.utf-8.txt +10030 -0
  38. data/example/guess_encoding/it.utf-8.txt +10030 -0
  39. data/lib/cmess/bconv.rb +169 -0
  40. data/lib/cmess/cinderella.rb +66 -0
  41. data/lib/cmess/cli.rb +120 -0
  42. data/lib/cmess/decode_entities.rb +69 -0
  43. data/lib/cmess/guess_encoding/automatic.rb +343 -0
  44. data/lib/cmess/guess_encoding/encoding.rb +78 -0
  45. data/lib/cmess/guess_encoding/manual.rb +108 -0
  46. data/lib/cmess/guess_encoding.rb +61 -0
  47. data/lib/cmess/version.rb +51 -0
  48. data/lib/cmess.rb +49 -0
  49. metadata +136 -0
data/data/csets/unicode/cyrillic-supplement.yaml ADDED
@@ -0,0 +1,17 @@
1
+ ---
2
+ - "Ԁ" # CYRILLIC CAPITAL LETTER KOMI DE
3
+ - "ԁ" # CYRILLIC SMALL LETTER KOMI DE
4
+ - "Ԃ" # CYRILLIC CAPITAL LETTER KOMI DJE
5
+ - "ԃ" # CYRILLIC SMALL LETTER KOMI DJE
6
+ - "Ԅ" # CYRILLIC CAPITAL LETTER KOMI ZJE
7
+ - "ԅ" # CYRILLIC SMALL LETTER KOMI ZJE
8
+ - "Ԇ" # CYRILLIC CAPITAL LETTER KOMI DZJE
9
+ - "ԇ" # CYRILLIC SMALL LETTER KOMI DZJE
10
+ - "Ԉ" # CYRILLIC CAPITAL LETTER KOMI LJE
11
+ - "ԉ" # CYRILLIC SMALL LETTER KOMI LJE
12
+ - "Ԋ" # CYRILLIC CAPITAL LETTER KOMI NJE
13
+ - "ԋ" # CYRILLIC SMALL LETTER KOMI NJE
14
+ - "Ԍ" # CYRILLIC CAPITAL LETTER KOMI SJE
15
+ - "ԍ" # CYRILLIC SMALL LETTER KOMI SJE
16
+ - "Ԏ" # CYRILLIC CAPITAL LETTER KOMI TJE
17
+ - "ԏ" # CYRILLIC SMALL LETTER KOMI TJE
data/data/csets/unicode/cyrillic.yaml ADDED
@@ -0,0 +1,256 @@
1
+ ---
2
+ - "Ѐ" # CYRILLIC CAPITAL LETTER IE WITH GRAVE
3
+ - "Ё" # CYRILLIC CAPITAL LETTER IO (present in WGL4)
4
+ - "Ђ" # CYRILLIC CAPITAL LETTER DJE (present in WGL4)
5
+ - "Ѓ" # CYRILLIC CAPITAL LETTER GJE (present in WGL4)
6
+ - "Є" # CYRILLIC CAPITAL LETTER UKRAINIAN IE (present in WGL4)
7
+ - "Ѕ" # CYRILLIC CAPITAL LETTER DZE (present in WGL4)
8
+ - "І" # CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I (present in WGL4)
9
+ - "Ї" # CYRILLIC CAPITAL LETTER YI (present in WGL4)
10
+ - "Ј" # CYRILLIC CAPITAL LETTER JE (present in WGL4)
11
+ - "Љ" # CYRILLIC CAPITAL LETTER LJE (present in WGL4)
12
+ - "Њ" # CYRILLIC CAPITAL LETTER NJE (present in WGL4)
13
+ - "Ћ" # CYRILLIC CAPITAL LETTER TSHE (present in WGL4)
14
+ - "Ќ" # CYRILLIC CAPITAL LETTER KJE (present in WGL4)
15
+ - "Ѝ" # CYRILLIC CAPITAL LETTER I WITH GRAVE (present in WGL4)
16
+ - "Ў" # CYRILLIC CAPITAL LETTER SHORT U (present in WGL4)
17
+ - "Џ" # CYRILLIC CAPITAL LETTER DZHE (present in WGL4)
18
+ - "А" # CYRILLIC CAPITAL LETTER A (present in WGL4)
19
+ - "Б" # CYRILLIC CAPITAL LETTER BE (present in WGL4)
20
+ - "В" # CYRILLIC CAPITAL LETTER VE (present in WGL4)
21
+ - "Г" # CYRILLIC CAPITAL LETTER GHE (present in WGL4)
22
+ - "Д" # CYRILLIC CAPITAL LETTER DE (present in WGL4)
23
+ - "Е" # CYRILLIC CAPITAL LETTER IE (present in WGL4)
24
+ - "Ж" # CYRILLIC CAPITAL LETTER ZHE (present in WGL4)
25
+ - "З" # CYRILLIC CAPITAL LETTER ZE (present in WGL4)
26
+ - "И" # CYRILLIC CAPITAL LETTER I (present in WGL4)
27
+ - "Й" # CYRILLIC CAPITAL LETTER SHORT I (present in WGL4)
28
+ - "К" # CYRILLIC CAPITAL LETTER KA (present in WGL4)
29
+ - "Л" # CYRILLIC CAPITAL LETTER EL (present in WGL4)
30
+ - "М" # CYRILLIC CAPITAL LETTER EM (present in WGL4)
31
+ - "Н" # CYRILLIC CAPITAL LETTER EN (present in WGL4)
32
+ - "О" # CYRILLIC CAPITAL LETTER O (present in WGL4)
33
+ - "П" # CYRILLIC CAPITAL LETTER PE (present in WGL4)
34
+ - "Р" # CYRILLIC CAPITAL LETTER ER (present in WGL4)
35
+ - "С" # CYRILLIC CAPITAL LETTER ES (present in WGL4)
36
+ - "Т" # CYRILLIC CAPITAL LETTER TE (present in WGL4)
37
+ - "У" # CYRILLIC CAPITAL LETTER U (present in WGL4)
38
+ - "Ф" # CYRILLIC CAPITAL LETTER EF (present in WGL4)
39
+ - "Х" # CYRILLIC CAPITAL LETTER HA (present in WGL4)
40
+ - "Ц" # CYRILLIC CAPITAL LETTER TSE (present in WGL4)
41
+ - "Ч" # CYRILLIC CAPITAL LETTER CHE (present in WGL4)
42
+ - "Ш" # CYRILLIC CAPITAL LETTER SHA (present in WGL4)
43
+ - "Щ" # CYRILLIC CAPITAL LETTER SHCHA (present in WGL4)
44
+ - "Ъ" # CYRILLIC CAPITAL LETTER HARD SIGN (present in WGL4)
45
+ - "Ы" # CYRILLIC CAPITAL LETTER YERU (present in WGL4)
46
+ - "Ь" # CYRILLIC CAPITAL LETTER SOFT SIGN (present in WGL4)
47
+ - "Э" # CYRILLIC CAPITAL LETTER E (present in WGL4)
48
+ - "Ю" # CYRILLIC CAPITAL LETTER YU (present in WGL4)
49
+ - "Я" # CYRILLIC CAPITAL LETTER YA (present in WGL4)
50
+ - "а" # CYRILLIC SMALL LETTER A (present in WGL4)
51
+ - "б" # CYRILLIC SMALL LETTER BE (present in WGL4)
52
+ - "в" # CYRILLIC SMALL LETTER VE (present in WGL4)
53
+ - "г" # CYRILLIC SMALL LETTER GHE (present in WGL4)
54
+ - "д" # CYRILLIC SMALL LETTER DE (present in WGL4)
55
+ - "е" # CYRILLIC SMALL LETTER IE (present in WGL4)
56
+ - "ж" # CYRILLIC SMALL LETTER ZHE (present in WGL4)
57
+ - "з" # CYRILLIC SMALL LETTER ZE (present in WGL4)
58
+ - "и" # CYRILLIC SMALL LETTER I (present in WGL4)
59
+ - "й" # CYRILLIC SMALL LETTER SHORT I (present in WGL4)
60
+ - "к" # CYRILLIC SMALL LETTER KA (present in WGL4)
61
+ - "л" # CYRILLIC SMALL LETTER EL (present in WGL4)
62
+ - "м" # CYRILLIC SMALL LETTER EM (present in WGL4)
63
+ - "н" # CYRILLIC SMALL LETTER EN (present in WGL4)
64
+ - "о" # CYRILLIC SMALL LETTER O (present in WGL4)
65
+ - "п" # CYRILLIC SMALL LETTER PE (present in WGL4)
66
+ - "р" # CYRILLIC SMALL LETTER ER (present in WGL4)
67
+ - "с" # CYRILLIC SMALL LETTER ES (present in WGL4)
68
+ - "т" # CYRILLIC SMALL LETTER TE (present in WGL4)
69
+ - "у" # CYRILLIC SMALL LETTER U (present in WGL4)
70
+ - "ф" # CYRILLIC SMALL LETTER EF (present in WGL4)
71
+ - "х" # CYRILLIC SMALL LETTER HA (present in WGL4)
72
+ - "ц" # CYRILLIC SMALL LETTER TSE (present in WGL4)
73
+ - "ч" # CYRILLIC SMALL LETTER CHE (present in WGL4)
74
+ - "ш" # CYRILLIC SMALL LETTER SHA (present in WGL4)
75
+ - "щ" # CYRILLIC SMALL LETTER SHCHA (present in WGL4)
76
+ - "ъ" # CYRILLIC SMALL LETTER HARD SIGN (present in WGL4)
77
+ - "ы" # CYRILLIC SMALL LETTER YERU (present in WGL4)
78
+ - "ь" # CYRILLIC SMALL LETTER SOFT SIGN (present in WGL4)
79
+ - "э" # CYRILLIC SMALL LETTER E (present in WGL4)
80
+ - "ю" # CYRILLIC SMALL LETTER YU (present in WGL4)
81
+ - "я" # CYRILLIC SMALL LETTER YA (present in WGL4)
82
+ - "ѐ" # CYRILLIC SMALL LETTER IE WITH GRAVE (present in WGL4)
83
+ - "ё" # CYRILLIC SMALL LETTER IO (present in WGL4)
84
+ - "ђ" # CYRILLIC SMALL LETTER DJE (present in WGL4)
85
+ - "ѓ" # CYRILLIC SMALL LETTER GJE (present in WGL4)
86
+ - "є" # CYRILLIC SMALL LETTER UKRAINIAN IE (present in WGL4)
87
+ - "ѕ" # CYRILLIC SMALL LETTER DZE (present in WGL4)
88
+ - "і" # CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I (present in WGL4)
89
+ - "ї" # CYRILLIC SMALL LETTER YI (present in WGL4)
90
+ - "ј" # CYRILLIC SMALL LETTER JE (present in WGL4)
91
+ - "љ" # CYRILLIC SMALL LETTER LJE (present in WGL4)
92
+ - "њ" # CYRILLIC SMALL LETTER NJE (present in WGL4)
93
+ - "ћ" # CYRILLIC SMALL LETTER TSHE (present in WGL4)
94
+ - "ќ" # CYRILLIC SMALL LETTER KJE (present in WGL4)
95
+ - "ѝ" # CYRILLIC SMALL LETTER I WITH GRAVE (present in WGL4)
96
+ - "ў" # CYRILLIC SMALL LETTER SHORT U (present in WGL4)
97
+ - "џ" # CYRILLIC SMALL LETTER DZHE
98
+ - "Ѡ" # CYRILLIC CAPITAL LETTER OMEGA
99
+ - "ѡ" # CYRILLIC SMALL LETTER OMEGA
100
+ - "Ѣ" # CYRILLIC CAPITAL LETTER YAT
101
+ - "ѣ" # CYRILLIC SMALL LETTER YAT
102
+ - "Ѥ" # CYRILLIC CAPITAL LETTER IOTIFIED E
103
+ - "ѥ" # CYRILLIC SMALL LETTER IOTIFIED E
104
+ - "Ѧ" # CYRILLIC CAPITAL LETTER LITTLE YUS
105
+ - "ѧ" # CYRILLIC SMALL LETTER LITTLE YUS
106
+ - "Ѩ" # CYRILLIC CAPITAL LETTER IOTIFIED LITTLE YUS
107
+ - "ѩ" # CYRILLIC SMALL LETTER IOTIFIED LITTLE YUS
108
+ - "Ѫ" # CYRILLIC CAPITAL LETTER BIG YUS
109
+ - "ѫ" # CYRILLIC SMALL LETTER BIG YUS
110
+ - "Ѭ" # CYRILLIC CAPITAL LETTER IOTIFIED BIG YUS
111
+ - "ѭ" # CYRILLIC SMALL LETTER IOTIFIED BIG YUS
112
+ - "Ѯ" # CYRILLIC CAPITAL LETTER KSI
113
+ - "ѯ" # CYRILLIC SMALL LETTER KSI
114
+ - "Ѱ" # CYRILLIC CAPITAL LETTER PSI
115
+ - "ѱ" # CYRILLIC SMALL LETTER PSI
116
+ - "Ѳ" # CYRILLIC CAPITAL LETTER FITA
117
+ - "ѳ" # CYRILLIC SMALL LETTER FITA
118
+ - "Ѵ" # CYRILLIC CAPITAL LETTER IZHITSA
119
+ - "ѵ" # CYRILLIC SMALL LETTER IZHITSA
120
+ - "Ѷ" # CYRILLIC CAPITAL LETTER IZHITSA WITH DOUBLE GRAVE ACCENT
121
+ - "ѷ" # CYRILLIC SMALL LETTER IZHITSA WITH DOUBLE GRAVE ACCENT
122
+ - "Ѹ" # CYRILLIC CAPITAL LETTER UK
123
+ - "ѹ" # CYRILLIC SMALL LETTER UK
124
+ - "Ѻ" # CYRILLIC CAPITAL LETTER ROUND OMEGA
125
+ - "ѻ" # CYRILLIC SMALL LETTER ROUND OMEGA
126
+ - "Ѽ" # CYRILLIC CAPITAL LETTER OMEGA WITH TITLO
127
+ - "ѽ" # CYRILLIC SMALL LETTER OMEGA WITH TITLO
128
+ - "Ѿ" # CYRILLIC CAPITAL LETTER OT
129
+ - "ѿ" # CYRILLIC SMALL LETTER OT
130
+ - "Ҁ" # CYRILLIC CAPITAL LETTER KOPPA
131
+ - "ҁ" # CYRILLIC SMALL LETTER KOPPA
132
+ - "҂" # CYRILLIC THOUSANDS SIGN
133
+ - "о҃" # COMBINING CYRILLIC TITLO (combined with о)
134
+ - "о҄" # COMBINING CYRILLIC PALATALIZATION (combined with о)
135
+ - "о҅" # COMBINING CYRILLIC DASIA PNEUMATA (combined with о)
136
+ - "о҆" # COMBINING CYRILLIC PSILI PNEUMATA (combined with о)
137
+ - "о҈" # COMBINING CYRILLIC HUNDRED THOUSANDS SIGN (combined with о)
138
+ - "о҉" # COMBINING CYRILLIC MILLIONS SIGN (combined with о)
139
+ - "Ҋ" # CYRILLIC CAPITAL LETTER SHORT I WITH TAIL
140
+ - "ҋ" # CYRILLIC SMALL LETTER SHORT I WITH TAIL
141
+ - "Ҍ" # CYRILLIC CAPITAL LETTER SEMISOFT SIGN
142
+ - "ҍ" # CYRILLIC SMALL LETTER SEMISOFT SIGN
143
+ - "Ҏ" # CYRILLIC CAPITAL LETTER ER WITH TICK
144
+ - "ҏ" # CYRILLIC SMALL LETTER ER WITH TICK
145
+ - "Ґ" # CYRILLIC CAPITAL LETTER GHE WITH UPTURN (present in WGL4)
146
+ - "ґ" # CYRILLIC SMALL LETTER GHE WITH UPTURN (present in WGL4)
147
+ - "Ғ" # CYRILLIC CAPITAL LETTER GHE WITH STROKE
148
+ - "ғ" # CYRILLIC SMALL LETTER GHE WITH STROKE
149
+ - "Ҕ" # CYRILLIC CAPITAL LETTER GHE WITH MIDDLE HOOK
150
+ - "ҕ" # CYRILLIC SMALL LETTER GHE WITH MIDDLE HOOK
151
+ - "Җ" # CYRILLIC CAPITAL LETTER ZHE WITH DESCENDER
152
+ - "җ" # CYRILLIC SMALL LETTER ZHE WITH DESCENDER
153
+ - "Ҙ" # CYRILLIC CAPITAL LETTER ZE WITH DESCENDER
154
+ - "ҙ" # CYRILLIC SMALL LETTER ZE WITH DESCENDER
155
+ - "Қ" # CYRILLIC CAPITAL LETTER KA WITH DESCENDER
156
+ - "қ" # CYRILLIC SMALL LETTER KA WITH DESCENDER
157
+ - "Ҝ" # CYRILLIC CAPITAL LETTER KA WITH VERTICAL STROKE
158
+ - "ҝ" # CYRILLIC SMALL LETTER KA WITH VERTICAL STROKE
159
+ - "Ҟ" # CYRILLIC CAPITAL LETTER KA WITH STROKE
160
+ - "ҟ" # CYRILLIC SMALL LETTER KA WITH STROKE
161
+ - "Ҡ" # CYRILLIC CAPITAL LETTER BASHKIR KA
162
+ - "ҡ" # CYRILLIC SMALL LETTER BASHKIR KA
163
+ - "Ң" # CYRILLIC CAPITAL LETTER EN WITH DESCENDER
164
+ - "ң" # CYRILLIC SMALL LETTER EN WITH DESCENDER
165
+ - "Ҥ" # CYRILLIC CAPITAL LIGATURE EN GHE
166
+ - "ҥ" # CYRILLIC SMALL LIGATURE EN GHE
167
+ - "Ҧ" # CYRILLIC CAPITAL LETTER PE WITH MIDDLE HOOK
168
+ - "ҧ" # CYRILLIC SMALL LETTER PE WITH MIDDLE HOOK
169
+ - "Ҩ" # CYRILLIC CAPITAL LETTER ABKHASIAN HA
170
+ - "ҩ" # CYRILLIC SMALL LETTER ABKHASIAN HA
171
+ - "Ҫ" # CYRILLIC CAPITAL LETTER ES WITH DESCENDER
172
+ - "ҫ" # CYRILLIC SMALL LETTER ES WITH DESCENDER
173
+ - "Ҭ" # CYRILLIC CAPITAL LETTER TE WITH DESCENDER
174
+ - "ҭ" # CYRILLIC SMALL LETTER TE WITH DESCENDER
175
+ - "Ү" # CYRILLIC CAPITAL LETTER STRAIGHT U
176
+ - "ү" # CYRILLIC SMALL LETTER STRAIGHT U
177
+ - "Ұ" # CYRILLIC CAPITAL LETTER STRAIGHT U WITH STROKE
178
+ - "ұ" # CYRILLIC SMALL LETTER STRAIGHT U WITH STROKE
179
+ - "Ҳ" # CYRILLIC CAPITAL LETTER HA WITH DESCENDER
180
+ - "ҳ" # CYRILLIC SMALL LETTER HA WITH DESCENDER
181
+ - "Ҵ" # CYRILLIC CAPITAL LIGATURE TE TSE
182
+ - "ҵ" # CYRILLIC SMALL LIGATURE TE TSE
183
+ - "Ҷ" # CYRILLIC CAPITAL LETTER CHE WITH DESCENDER
184
+ - "ҷ" # CYRILLIC SMALL LETTER CHE WITH DESCENDER
185
+ - "Ҹ" # CYRILLIC CAPITAL LETTER CHE WITH VERTICAL STROKE
186
+ - "ҹ" # CYRILLIC SMALL LETTER CHE WITH VERTICAL STROKE
187
+ - "Һ" # CYRILLIC CAPITAL LETTER SHHA
188
+ - "һ" # CYRILLIC SMALL LETTER SHHA
189
+ - "Ҽ" # CYRILLIC CAPITAL LETTER ABKHASIAN CHE
190
+ - "ҽ" # CYRILLIC SMALL LETTER ABKHASIAN CHE
191
+ - "Ҿ" # CYRILLIC CAPITAL LETTER ABKHASIAN CHE WITH DESCENDER
192
+ - "ҿ" # CYRILLIC SMALL LETTER ABKHASIAN CHE WITH DESCENDER
193
+ - "Ӏ" # CYRILLIC LETTER PALOCHKA
194
+ - "Ӂ" # CYRILLIC CAPITAL LETTER ZHE WITH BREVE
195
+ - "ӂ" # CYRILLIC SMALL LETTER ZHE WITH BREVE
196
+ - "Ӄ" # CYRILLIC CAPITAL LETTER KA WITH HOOK
197
+ - "ӄ" # CYRILLIC SMALL LETTER KA WITH HOOK
198
+ - "Ӆ" # CYRILLIC CAPITAL LETTER EL WITH TAIL
199
+ - "ӆ" # CYRILLIC SMALL LETTER EL WITH TAIL
200
+ - "Ӈ" # CYRILLIC CAPITAL LETTER EN WITH HOOK
201
+ - "ӈ" # CYRILLIC SMALL LETTER EN WITH HOOK
202
+ - "Ӊ" # CYRILLIC CAPITAL LETTER EN WITH TAIL
203
+ - "ӊ" # CYRILLIC SMALL LETTER EN WITH TAIL
204
+ - "Ӌ" # CYRILLIC CAPITAL LETTER KHAKASSIAN CHE
205
+ - "ӌ" # CYRILLIC SMALL LETTER KHAKASSIAN CHE
206
+ - "Ӎ" # CYRILLIC CAPITAL LETTER EM WITH TAIL
207
+ - "ӎ" # CYRILLIC SMALL LETTER EM WITH TAIL
208
+ - "ӏ" # CYRILLIC SMALL LETTER PALOCHKA
209
+ - "Ӑ" # CYRILLIC CAPITAL LETTER A WITH BREVE
210
+ - "ӑ" # CYRILLIC SMALL LETTER A WITH BREVE
211
+ - "Ӓ" # CYRILLIC CAPITAL LETTER A WITH DIAERESIS
212
+ - "ӓ" # CYRILLIC SMALL LETTER A WITH DIAERESIS
213
+ - "Ӕ" # CYRILLIC CAPITAL LIGATURE A IE
214
+ - "ӕ" # CYRILLIC SMALL LIGATURE A IE
215
+ - "Ӗ" # CYRILLIC CAPITAL LETTER IE WITH BREVE
216
+ - "ӗ" # CYRILLIC SMALL LETTER IE WITH BREVE
217
+ - "Ә" # CYRILLIC CAPITAL LETTER SCHWA
218
+ - "ә" # CYRILLIC SMALL LETTER SCHWA
219
+ - "Ӛ" # CYRILLIC CAPITAL LETTER SCHWA WITH DIAERESIS
220
+ - "ӛ" # CYRILLIC SMALL LETTER SCHWA WITH DIAERESIS
221
+ - "Ӝ" # CYRILLIC CAPITAL LETTER ZHE WITH DIAERESIS
222
+ - "ӝ" # CYRILLIC SMALL LETTER ZHE WITH DIAERESIS
223
+ - "Ӟ" # CYRILLIC CAPITAL LETTER ZE WITH DIAERESIS
224
+ - "ӟ" # CYRILLIC SMALL LETTER ZE WITH DIAERESIS
225
+ - "Ӡ" # CYRILLIC CAPITAL LETTER ABKHASIAN DZE
226
+ - "ӡ" # CYRILLIC SMALL LETTER ABKHASIAN DZE
227
+ - "Ӣ" # CYRILLIC CAPITAL LETTER I WITH MACRON
228
+ - "ӣ" # CYRILLIC SMALL LETTER I WITH MACRON
229
+ - "Ӥ" # CYRILLIC CAPITAL LETTER I WITH DIAERESIS
230
+ - "ӥ" # CYRILLIC SMALL LETTER I WITH DIAERESIS
231
+ - "Ӧ" # CYRILLIC CAPITAL LETTER O WITH DIAERESIS
232
+ - "ӧ" # CYRILLIC SMALL LETTER O WITH DIAERESIS
233
+ - "Ө" # CYRILLIC CAPITAL LETTER BARRED O
234
+ - "ө" # CYRILLIC SMALL LETTER BARRED O
235
+ - "Ӫ" # CYRILLIC CAPITAL LETTER BARRED O WITH DIAERESIS
236
+ - "ӫ" # CYRILLIC SMALL LETTER BARRED O WITH DIAERESIS
237
+ - "Ӭ" # CYRILLIC CAPITAL LETTER E WITH DIAERESIS
238
+ - "ӭ" # CYRILLIC SMALL LETTER E WITH DIAERESIS
239
+ - "Ӯ" # CYRILLIC CAPITAL LETTER U WITH MACRON
240
+ - "ӯ" # CYRILLIC SMALL LETTER U WITH MACRON
241
+ - "Ӱ" # CYRILLIC CAPITAL LETTER U WITH DIAERESIS
242
+ - "ӱ" # CYRILLIC SMALL LETTER U WITH DIAERESIS
243
+ - "Ӳ" # CYRILLIC CAPITAL LETTER U WITH DOUBLE ACUTE
244
+ - "ӳ" # CYRILLIC SMALL LETTER U WITH DOUBLE ACUTE
245
+ - "Ӵ" # CYRILLIC CAPITAL LETTER CHE WITH DIAERESIS
246
+ - "ӵ" # CYRILLIC SMALL LETTER CHE WITH DIAERESIS
247
+ - "Ӷ" # CYRILLIC CAPITAL LETTER GHE WITH DESCENDER
248
+ - "ӷ" # CYRILLIC SMALL LETTER GHE WITH DESCENDER
249
+ - "Ӹ" # CYRILLIC CAPITAL LETTER YERU WITH DIAERESIS
250
+ - "ӹ" # CYRILLIC SMALL LETTER YERU WITH DIAERESIS
251
+ - "Ӻ" # CYRILLIC CAPITAL LETTER GHE WITH STROKE AND HOOK
252
+ - "ӻ" # CYRILLIC SMALL LETTER GHE WITH STROKE AND HOOK
253
+ - "Ӽ" # CYRILLIC CAPITAL LETTER HA WITH HOOK
254
+ - "ӽ" # CYRILLIC SMALL LETTER HA WITH HOOK
255
+ - "Ӿ" # CYRILLIC CAPITAL LETTER HA WITH STROKE
256
+ - "ӿ" # CYRILLIC SMALL LETTER HA WITH STROKE
data/data/csets/unicode/greek.yaml ADDED
@@ -0,0 +1,129 @@
1
+ ---
2
+ - "ʹ" # GREEK NUMERAL SIGN
3
+ - "͵" # GREEK LOWER NUMERAL SIGN
4
+ - "ͺ" # GREEK YPOGEGRAMMENI
5
+ - "ͻ" # GREEK SMALL REVERSED LUNATE SIGMA SYMBOL
6
+ - "ͼ" # GREEK SMALL DOTTED LUNATE SIGMA SYMBOL
7
+ - "ͽ" # GREEK SMALL REVERSED DOTTED LUNATE SIGMA SYMBOL
8
+ - ";" # GREEK QUESTION MARK
9
+ - "΄" # GREEK TONOS (present in WGL4)
10
+ - "΅" # GREEK DIALYTIKA TONOS (present in WGL4)
11
+ - "Ά" # GREEK CAPITAL LETTER ALPHA WITH TONOS (present in WGL4)
12
+ - "·" # GREEK ANO TELEIA (present in WGL4)
13
+ - "Έ" # GREEK CAPITAL LETTER EPSILON WITH TONOS (present in WGL4)
14
+ - "Ή" # GREEK CAPITAL LETTER ETA WITH TONOS (present in WGL4)
15
+ - "Ί" # GREEK CAPITAL LETTER IOTA WITH TONOS (present in WGL4)
16
+ - "Ό" # GREEK CAPITAL LETTER OMICRON WITH TONOS (present in WGL4)
17
+ - "Ύ" # GREEK CAPITAL LETTER UPSILON WITH TONOS (present in WGL4)
18
+ - "Ώ" # GREEK CAPITAL LETTER OMEGA WITH TONOS (present in WGL4)
19
+ - "ΐ" # GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS (present in WGL4)
20
+ - "Α" # GREEK CAPITAL LETTER ALPHA (present in WGL4 and in Symbol font)
21
+ - "Β" # GREEK CAPITAL LETTER BETA (present in WGL4 and in Symbol font)
22
+ - "Γ" # GREEK CAPITAL LETTER GAMMA (present in WGL4 and in Symbol font)
23
+ - "Δ" # GREEK CAPITAL LETTER DELTA (present in WGL4 and in Symbol font)
24
+ - "Ε" # GREEK CAPITAL LETTER EPSILON (present in WGL4 and in Symbol font)
25
+ - "Ζ" # GREEK CAPITAL LETTER ZETA (present in WGL4 and in Symbol font)
26
+ - "Η" # GREEK CAPITAL LETTER ETA (present in WGL4 and in Symbol font)
27
+ - "Θ" # GREEK CAPITAL LETTER THETA (present in WGL4 and in Symbol font)
28
+ - "Ι" # GREEK CAPITAL LETTER IOTA (present in WGL4 and in Symbol font)
29
+ - "Κ" # GREEK CAPITAL LETTER KAPPA (present in WGL4 and in Symbol font)
30
+ - "Λ" # GREEK CAPITAL LETTER LAMDA (present in WGL4 and in Symbol font)
31
+ - "Μ" # GREEK CAPITAL LETTER MU (present in WGL4 and in Symbol font)
32
+ - "Ν" # GREEK CAPITAL LETTER NU (present in WGL4 and in Symbol font)
33
+ - "Ξ" # GREEK CAPITAL LETTER XI (present in WGL4 and in Symbol font)
34
+ - "Ο" # GREEK CAPITAL LETTER OMICRON (present in WGL4 and in Symbol font)
35
+ - "Π" # GREEK CAPITAL LETTER PI (present in WGL4 and in Symbol font)
36
+ - "Ρ" # GREEK CAPITAL LETTER RHO (present in WGL4 and in Symbol font)
37
+ - "Σ" # GREEK CAPITAL LETTER SIGMA (present in WGL4 and in Symbol font)
38
+ - "Τ" # GREEK CAPITAL LETTER TAU (present in WGL4 and in Symbol font)
39
+ - "Υ" # GREEK CAPITAL LETTER UPSILON (present in WGL4 and in Symbol font)
40
+ - "Φ" # GREEK CAPITAL LETTER PHI (present in WGL4 and in Symbol font)
41
+ - "Χ" # GREEK CAPITAL LETTER CHI (present in WGL4 and in Symbol font)
42
+ - "Ψ" # GREEK CAPITAL LETTER PSI (present in WGL4 and in Symbol font)
43
+ - "Ω" # GREEK CAPITAL LETTER OMEGA (present in WGL4 and MacRoman, and in Symbol font)
44
+ - "Ϊ" # GREEK CAPITAL LETTER IOTA WITH DIALYTIKA
45
+ - "Ϋ" # GREEK CAPITAL LETTER UPSILON WITH DIALYTIKA
46
+ - "ά" # GREEK SMALL LETTER ALPHA WITH TONOS
47
+ - "έ" # GREEK SMALL LETTER EPSILON WITH TONOS
48
+ - "ή" # GREEK SMALL LETTER ETA WITH TONOS
49
+ - "ί" # GREEK SMALL LETTER IOTA WITH TONOS
50
+ - "ΰ" # GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
51
+ - "α" # GREEK SMALL LETTER ALPHA (present in WGL4 and in Symbol font)
52
+ - "β" # GREEK SMALL LETTER BETA (present in WGL4 and in Symbol font)
53
+ - "γ" # GREEK SMALL LETTER GAMMA (present in WGL4 and in Symbol font)
54
+ - "δ" # GREEK SMALL LETTER DELTA (present in WGL4 and in Symbol font)
55
+ - "ε" # GREEK SMALL LETTER EPSILON (present in WGL4 and in Symbol font)
56
+ - "ζ" # GREEK SMALL LETTER ZETA (present in WGL4 and in Symbol font)
57
+ - "η" # GREEK SMALL LETTER ETA (present in WGL4 and in Symbol font)
58
+ - "θ" # GREEK SMALL LETTER THETA (present in WGL4 and in Symbol font)
59
+ - "ι" # GREEK SMALL LETTER IOTA (present in WGL4 and in Symbol font)
60
+ - "κ" # GREEK SMALL LETTER KAPPA (present in WGL4 and in Symbol font)
61
+ - "λ" # GREEK SMALL LETTER LAMDA (present in WGL4 and in Symbol font)
62
+ - "μ" # GREEK SMALL LETTER MU (present in WGL4 and in Symbol font)
63
+ - "ν" # GREEK SMALL LETTER NU (present in WGL4 and in Symbol font)
64
+ - "ξ" # GREEK SMALL LETTER XI (present in WGL4 and in Symbol font)
65
+ - "ο" # GREEK SMALL LETTER OMICRON (present in WGL4 and in Symbol font)
66
+ - "π" # GREEK SMALL LETTER PI (present in WGL4 and MacRoman, and in Symbol font)
67
+ - "ρ" # GREEK SMALL LETTER RHO (present in WGL4 and in Symbol font)
68
+ - "ς" # GREEK SMALL LETTER FINAL SIGMA (present in WGL4 and in Symbol font)
69
+ - "σ" # GREEK SMALL LETTER SIGMA (present in WGL4 and in Symbol font)
70
+ - "τ" # GREEK SMALL LETTER TAU (present in WGL4 and in Symbol font)
71
+ - "υ" # GREEK SMALL LETTER UPSILON (present in WGL4 and in Symbol font)
72
+ - "φ" # GREEK SMALL LETTER PHI (present in WGL4 and in Symbol font)
73
+ - "χ" # GREEK SMALL LETTER CHI (present in WGL4 and in Symbol font)
74
+ - "ψ" # GREEK SMALL LETTER PSI (present in WGL4 and in Symbol font)
75
+ - "ω" # GREEK SMALL LETTER OMEGA (present in WGL4 and in Symbol font)
76
+ - "ϊ" # GREEK SMALL LETTER IOTA WITH DIALYTIKA (present in WGL4)
77
+ - "ϋ" # GREEK SMALL LETTER UPSILON WITH DIALYTIKA (present in WGL4)
78
+ - "ό" # GREEK SMALL LETTER OMICRON WITH TONOS (present in WGL4)
79
+ - "ύ" # GREEK SMALL LETTER UPSILON WITH TONOS (present in WGL4)
80
+ - "ώ" # GREEK SMALL LETTER OMEGA WITH TONOS (present in WGL4)
81
+ - "ϐ" # GREEK BETA SYMBOL
82
+ - "ϑ" # GREEK THETA SYMBOL (present in Symbol font)
83
+ - "ϒ" # GREEK UPSILON WITH HOOK SYMBOL (present in Symbol font)
84
+ - "ϓ" # GREEK UPSILON WITH ACUTE AND HOOK SYMBOL
85
+ - "ϔ" # GREEK UPSILON WITH DIAERESIS AND HOOK SYMBOL
86
+ - "ϕ" # GREEK PHI SYMBOL (present in Symbol font)
87
+ - "ϖ" # GREEK PI SYMBOL (present in Symbol font)
88
+ - "ϗ" # GREEK KAI SYMBOL
89
+ - "Ϙ" # GREEK LETTER ARCHAIC KOPPA
90
+ - "ϙ" # GREEK SMALL LETTER ARCHAIC KOPPA
91
+ - "Ϛ" # GREEK LETTER STIGMA
92
+ - "ϛ" # GREEK SMALL LETTER STIGMA
93
+ - "Ϝ" # GREEK LETTER DIGAMMA
94
+ - "ϝ" # GREEK SMALL LETTER DIGAMMA
95
+ - "Ϟ" # GREEK LETTER KOPPA
96
+ - "ϟ" # GREEK SMALL LETTER KOPPA
97
+ - "Ϡ" # GREEK LETTER SAMPI
98
+ - "ϡ" # GREEK SMALL LETTER SAMPI
99
+ - "Ϣ" # COPTIC CAPITAL LETTER SHEI
100
+ - "ϣ" # COPTIC SMALL LETTER SHEI
101
+ - "Ϥ" # COPTIC CAPITAL LETTER FEI
102
+ - "ϥ" # COPTIC SMALL LETTER FEI
103
+ - "Ϧ" # COPTIC CAPITAL LETTER KHEI
104
+ - "ϧ" # COPTIC SMALL LETTER KHEI
105
+ - "Ϩ" # COPTIC CAPITAL LETTER HORI
106
+ - "ϩ" # COPTIC SMALL LETTER HORI
107
+ - "Ϫ" # COPTIC CAPITAL LETTER GANGIA
108
+ - "ϫ" # COPTIC SMALL LETTER GANGIA
109
+ - "Ϭ" # COPTIC CAPITAL LETTER SHIMA
110
+ - "ϭ" # COPTIC SMALL LETTER SHIMA
111
+ - "Ϯ" # COPTIC CAPITAL LETTER DEI
112
+ - "ϯ" # COPTIC SMALL LETTER DEI
113
+ - "ϰ" # GREEK KAPPA SYMBOL
114
+ - "ϱ" # GREEK RHO SYMBOL
115
+ - "ϲ" # GREEK LUNATE SIGMA SYMBOL
116
+ - "ϳ" # GREEK LETTER YOT
117
+ - "ϴ" # GREEK CAPITAL THETA SYMBOL
118
+ - "ϵ" # GREEK LUNATE EPSILON SYMBOL
119
+ - "϶" # GREEK REVERSED LUNATE EPSILON SYMBOL
120
+ - "Ϸ" # GREEK CAPITAL LETTER SHO
121
+ - "ϸ" # GREEK SMALL LETTER SHO
122
+ - "Ϲ" # GREEK CAPITAL LUNATE SIGMA SYMBOL
123
+ - "Ϻ" # GREEK CAPITAL LETTER SAN
124
+ - "ϻ" # GREEK SMALL LETTER SAN
125
+ - "ϼ" # GREEK RHO WITH STROKE SYMBOL
126
+ - "Ͻ" # GREEK CAPITAL REVERSED LUNATE SIGMA SYMBOL
127
+ - "Ͼ" # GREEK CAPITAL DOTTED LUNATE SIGMA SYMBOL
128
+ - "Ͽ" # GREEK CAPITAL REVERSED DOTTED LUNATE SIGMA SYMBOL
129
+ - "₯" # DRACHMA SIGN
data/data/csets/unicode/ipa_extensions.yaml ADDED
@@ -0,0 +1,97 @@
1
+ ---
2
+ - "ɐ" # LATIN SMALL LETTER TURNED A
3
+ - "ɑ" # LATIN SMALL LETTER ALPHA
4
+ - "ɒ" # LATIN SMALL LETTER TURNED ALPHA
5
+ - "ɓ" # LATIN SMALL LETTER B WITH HOOK
6
+ - "ɔ" # LATIN SMALL LETTER OPEN O
7
+ - "ɕ" # LATIN SMALL LETTER C WITH CURL
8
+ - "ɖ" # LATIN SMALL LETTER D WITH TAIL
9
+ - "ɗ" # LATIN SMALL LETTER D WITH HOOK
10
+ - "ɘ" # LATIN SMALL LETTER REVERSED E
11
+ - "ə" # LATIN SMALL LETTER SCHWA
12
+ - "ɚ" # LATIN SMALL LETTER SCHWA WITH HOOK
13
+ - "ɛ" # LATIN SMALL LETTER OPEN E
14
+ - "ɜ" # LATIN SMALL LETTER REVERSED OPEN E
15
+ - "ɝ" # LATIN SMALL LETTER REVERSED OPEN E WITH HOOK
16
+ - "ɞ" # LATIN SMALL LETTER CLOSED REVERSED OPEN E
17
+ - "ɟ" # LATIN SMALL LETTER DOTLESS J WITH STROKE
18
+ - "ɠ" # LATIN SMALL LETTER G WITH HOOK
19
+ - "ɡ" # LATIN SMALL LETTER SCRIPT G
20
+ - "ɢ" # LATIN LETTER SMALL CAPITAL G
21
+ - "ɣ" # LATIN SMALL LETTER GAMMA
22
+ - "ɤ" # LATIN SMALL LETTER RAMS HORN
23
+ - "ɥ" # LATIN SMALL LETTER TURNED H
24
+ - "ɦ" # LATIN SMALL LETTER H WITH HOOK
25
+ - "ɧ" # LATIN SMALL LETTER HENG WITH HOOK
26
+ - "ɨ" # LATIN SMALL LETTER I WITH STROKE
27
+ - "ɩ" # LATIN SMALL LETTER IOTA
28
+ - "ɪ" # LATIN LETTER SMALL CAPITAL I
29
+ - "ɫ" # LATIN SMALL LETTER L WITH MIDDLE TILDE
30
+ - "ɬ" # LATIN SMALL LETTER L WITH BELT
31
+ - "ɭ" # LATIN SMALL LETTER L WITH RETROFLEX HOOK
32
+ - "ɮ" # LATIN SMALL LETTER LEZH
33
+ - "ɯ" # LATIN SMALL LETTER TURNED M
34
+ - "ɰ" # LATIN SMALL LETTER TURNED M WITH LONG LEG
35
+ - "ɱ" # LATIN SMALL LETTER M WITH HOOK
36
+ - "ɲ" # LATIN SMALL LETTER N WITH LEFT HOOK
37
+ - "ɳ" # LATIN SMALL LETTER N WITH RETROFLEX HOOK
38
+ - "ɴ" # LATIN LETTER SMALL CAPITAL N
39
+ - "ɵ" # LATIN SMALL LETTER BARRED O
40
+ - "ɶ" # LATIN LETTER SMALL CAPITAL OE
41
+ - "ɷ" # LATIN SMALL LETTER CLOSED OMEGA
42
+ - "ɸ" # LATIN SMALL LETTER PHI
43
+ - "ɹ" # LATIN SMALL LETTER TURNED R
44
+ - "ɺ" # LATIN SMALL LETTER TURNED R WITH LONG LEG
45
+ - "ɻ" # LATIN SMALL LETTER TURNED R WITH HOOK
46
+ - "ɼ" # LATIN SMALL LETTER R WITH LONG LEG
47
+ - "ɽ" # LATIN SMALL LETTER R WITH TAIL
48
+ - "ɾ" # LATIN SMALL LETTER R WITH FISHHOOK
49
+ - "ɿ" # LATIN SMALL LETTER REVERSED R WITH FISHHOOK
50
+ - "ʀ" # LATIN LETTER SMALL CAPITAL R
51
+ - "ʁ" # LATIN LETTER SMALL CAPITAL INVERTED R
52
+ - "ʂ" # LATIN SMALL LETTER S WITH HOOK
53
+ - "ʃ" # LATIN SMALL LETTER ESH
54
+ - "ʄ" # LATIN SMALL LETTER DOTLESS J WITH STROKE AND HOOK
55
+ - "ʅ" # LATIN SMALL LETTER SQUAT REVERSED ESH
56
+ - "ʆ" # LATIN SMALL LETTER ESH WITH CURL
57
+ - "ʇ" # LATIN SMALL LETTER TURNED T
58
+ - "ʈ" # LATIN SMALL LETTER T WITH RETROFLEX HOOK
59
+ - "ʉ" # LATIN SMALL LETTER U BAR
60
+ - "ʊ" # LATIN SMALL LETTER UPSILON
61
+ - "ʋ" # LATIN SMALL LETTER V WITH HOOK
62
+ - "ʌ" # LATIN SMALL LETTER TURNED V
63
+ - "ʍ" # LATIN SMALL LETTER TURNED W
64
+ - "ʎ" # LATIN SMALL LETTER TURNED Y
65
+ - "ʏ" # LATIN LETTER SMALL CAPITAL Y
66
+ - "ʐ" # LATIN SMALL LETTER Z WITH RETROFLEX HOOK
67
+ - "ʑ" # LATIN SMALL LETTER Z WITH CURL
68
+ - "ʒ" # LATIN SMALL LETTER EZH
69
+ - "ʓ" # LATIN SMALL LETTER EZH WITH CURL
70
+ - "ʔ" # LATIN LETTER GLOTTAL STOP
71
+ - "ʕ" # LATIN LETTER PHARYNGEAL VOICED FRICATIVE
72
+ - "ʖ" # LATIN LETTER INVERTED GLOTTAL STOP
73
+ - "ʗ" # LATIN LETTER STRETCHED C
74
+ - "ʘ" # LATIN LETTER BILABIAL CLICK
75
+ - "ʙ" # LATIN LETTER SMALL CAPITAL B
76
+ - "ʚ" # LATIN SMALL LETTER CLOSED OPEN E
77
+ - "ʛ" # LATIN LETTER SMALL CAPITAL G WITH HOOK
78
+ - "ʜ" # LATIN LETTER SMALL CAPITAL H
79
+ - "ʝ" # LATIN SMALL LETTER J WITH CROSSED-TAIL
80
+ - "ʞ" # LATIN SMALL LETTER TURNED K
81
+ - "ʟ" # LATIN LETTER SMALL CAPITAL L
82
+ - "ʠ" # LATIN SMALL LETTER Q WITH HOOK
83
+ - "ʡ" # LATIN LETTER GLOTTAL STOP WITH STROKE
84
+ - "ʢ" # LATIN LETTER REVERSED GLOTTAL STOP WITH STROKE
85
+ - "ʣ" # LATIN SMALL LETTER DZ DIGRAPH
86
+ - "ʤ" # LATIN SMALL LETTER DEZH DIGRAPH
87
+ - "ʥ" # LATIN SMALL LETTER DZ DIGRAPH WITH CURL
88
+ - "ʦ" # LATIN SMALL LETTER TS DIGRAPH
89
+ - "ʧ" # LATIN SMALL LETTER TESH DIGRAPH
90
+ - "ʨ" # LATIN SMALL LETTER TC DIGRAPH WITH CURL
91
+ - "ʩ" # LATIN SMALL LETTER FENG DIGRAPH
92
+ - "ʪ" # LATIN SMALL LETTER LS DIGRAPH
93
+ - "ʫ" # LATIN SMALL LETTER LZ DIGRAPH
94
+ - "ʬ" # LATIN SMALL LETTER BILABIAL PERCUSSIVE
95
+ - "ʭ" # LATIN SMALL LETTER BIDENTAL PERCUSSIVE
96
+ - "ʮ" # LATIN SMALL LETTER TURNED H WITH FISHHOOK
97
+ - "ʯ" # LATIN SMALL LETTER TURNED H WITH FISHHOOK AND TAIL
data/data/csets/unicode/latin-extended-c.yaml ADDED
@@ -0,0 +1,18 @@
1
+ ---
2
+ - "Ⱡ" # LATIN CAPITAL LETTER L WITH DOUBLE BAR
3
+ - "ⱡ" # LATIN SMALL LETTER L WITH DOUBLE BAR
4
+ - "Ɫ" # LATIN CAPITAL LETTER L WITH MIDDLE TILDE
5
+ - "Ᵽ" # LATIN CAPITAL LETTER P WITH STROKE
6
+ - "Ɽ" # LATIN CAPITAL LETTER R WITH TAIL
7
+ - "ⱥ" # LATIN SMALL LETTER A WITH STROKE
8
+ - "ⱦ" # LATIN SMALL LETTER T WITH DIAGONAL STROKE
9
+ - "Ⱨ" # LATIN CAPITAL LETTER H WITH DESCENDER
10
+ - "ⱨ" # LATIN SMALL LETTER H WITH DESCENDER
11
+ - "Ⱪ" # LATIN CAPITAL LETTER K WITH DESCENDER
12
+ - "ⱪ" # LATIN SMALL LETTER K WITH DESCENDER
13
+ - "Ⱬ" # LATIN CAPITAL LETTER Z WITH DESCENDER
14
+ - "ⱬ" # LATIN SMALL LETTER Z WITH DESCENDER
15
+ - "ⱴ" # LATIN SMALL LETTER V WITH CURL
16
+ - "Ⱶ" # LATIN CAPITAL LETTER HALF H
17
+ - "ⱶ" # LATIN SMALL LETTER HALF H
18
+ - "ⱷ" # LATIN SMALL LETTER TAILLESS PHI
data/data/csets/unicode/latin-extended-d.yaml ADDED
@@ -0,0 +1,3 @@
1
+ ---
2
+ - "꜠" # MODIFIER LETTER STRESS AND HIGH TONE
3
+ - "꜡" # MODIFIER LETTER STRESS AND LOW TONE
data/data/csets/unicode/latin_1_supplement.yaml ADDED
@@ -0,0 +1,128 @@
1
+ ---
2
+ - "€" # [ANSI - EURO SIGN; MacRoman - LATIN CAPITAL LETTER A WITH DIAERESIS]
3
+ - "" # [MacRoman - LATIN CAPITAL LETTER A WITH RING ABOVE]
4
+ - "‚" # [ANSI - SINGLE LOW-9 QUOTATION MARK; MacRoman - LATIN CAPITAL LETTER C WITH CEDILLA]
5
+ - "ƒ" # [ANSI - LATIN SMALL LETTER F WITH HOOK; MacRoman - LATIN CAPITAL LETTER E WITH ACUTE]
6
+ - "„" # [ANSI - DOUBLE LOW-9 QUOTATION MARK; MacRoman - LATIN CAPITAL LETTER N WITH TILDE]
7
+ - "…" # [ANSI - HORIZONTAL ELLIPSIS; MacRoman - LATIN CAPITAL LETTER O WITH DIAERESIS]
8
+ - "†" # [ANSI - DAGGER; MacRoman - LATIN CAPITAL LETTER U WITH DIAERESIS]
9
+ - "‡" # [ANSI - DOUBLE DAGGER; MacRoman - LATIN SMALL LETTER A WITH ACUTE]
10
+ - "ˆ" # [ANSI - MODIFIER LETTER CIRCUMFLEX ACCENT; MacRoman - LATIN SMALL LETTER A WITH GRAVE]
11
+ - "‰" # [ANSI - PER MILLE SIGN; MacRoman - LATIN SMALL LETTER A WITH CIRCUMFLEX]
12
+ - "Š" # [ANSI - LATIN CAPITAL LETTER S WITH CARON; MacRoman - LATIN SMALL LETTER A WITH DIAERESIS]
13
+ - "‹" # [ANSI - SINGLE LEFT-POINTING ANGLE QUOTATION MARK; MacRoman - LATIN SMALL LETTER A WITH TILDE]
14
+ - "Œ" # [ANSI - LATIN CAPITAL LETTER OE; MacRoman - LATIN SMALL LETTER A WITH RING ABOVE]
15
+ - "" # [MacRoman - LATIN SMALL LETTER C WITH CEDILLA]
16
+ - "Ž" # [ANSI - LATIN CAPITAL LETTER Z WITH CARON; MacRoman - LATIN SMALL LETTER E WITH ACUTE]
17
+ - "" # [MacRoman - LATIN SMALL LETTER E WITH GRAVE]
18
+ - "" # [MacRoman - LATIN SMALL LETTER E WITH CIRCUMFLEX]
19
+ - "‘" # [ANSI - LEFT SINGLE QUOTATION MARK; MacRoman - LATIN SMALL LETTER E WITH DIAERESIS]
20
+ - "’" # [ANSI - RIGHT SINGLE QUOTATION MARK; MacRoman - LATIN SMALL LETTER I WITH ACUTE]
21
+ - "“" # [ANSI - LEFT DOUBLE QUOTATION MARK; MacRoman - LATIN SMALL LETTER I WITH GRAVE]
22
+ - "”" # [ANSI - RIGHT DOUBLE QUOTATION MARK; MacRoman - LATIN SMALL LETTER I WITH CIRCUMFLEX]
23
+ - "•" # [ANSI - BULLET; MacRoman - LATIN SMALL LETTER I WITH DIAERESIS]
24
+ - "–" # [ANSI - EN DASH; MacRoman - LATIN SMALL LETTER N WITH TILDE]
25
+ - "—" # [ANSI - EM DASH; MacRoman - LATIN SMALL LETTER O WITH ACUTE]
26
+ - "˜" # [ANSI - SMALL TILDE; MacRoman - LATIN SMALL LETTER O WITH GRAVE]
27
+ - "™" # [ANSI - TRADE MARK SIGN; MacRoman - LATIN SMALL LETTER O WITH CIRCUMFLEX]
28
+ - "š" # [ANSI - LATIN SMALL LETTER S WITH CARON; MacRoman - LATIN SMALL LETTER O WITH DIAERESIS]
29
+ - "›" # [ANSI - SINGLE RIGHT-POINTING ANGLE QUOTATION MARK; MacRoman - LATIN SMALL LETTER O WITH TILDE]
30
+ - "œ" # [ANSI - LATIN SMALL LETTER OE; MacRoman - LATIN SMALL LETTER U WITH ACUTE]
31
+ - "" # [MacRoman - LATIN SMALL LETTER U WITH GRAVE]
32
+ - "ž" # [ANSI - LATIN SMALL LETTER Z WITH CARON; MacRoman - LATIN SMALL LETTER U WITH CIRCUMFLEX]
33
+ - "Ÿ" # [ANSI - LATIN CAPITAL LETTER Y WITH DIAERESIS; MacRoman - LATIN SMALL LETTER U WITH DIAERESIS]
34
+ - "¡" # INVERTED EXCLAMATION MARK
35
+ - "¢" # CENT SIGN
36
+ - "£" # POUND SIGN [lira sign is ₤]
37
+ - "¤" # CURRENCY SIGN
38
+ - "¥" # YEN SIGN
39
+ - "¦" # BROKEN BAR
40
+ - "§" # SECTION SIGN
41
+ - "¨" # DIAERESIS
42
+ - "©" # COPYRIGHT SIGN (present in Symbol font)
43
+ - "ª" # FEMININE ORDINAL INDICATOR
44
+ - "«" # LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
45
+ - "¬" # NOT SIGN (present in Symbol font)
46
+ - "­" # SOFT HYPHEN
47
+ - "®" # REGISTERED SIGN (present in Symbol font)
48
+ - "¯" # MACRON
49
+ - "°" # DEGREE SIGN
50
+ - "±" # PLUS-MINUS SIGN (present in Symbol font)
51
+ - "²" # SUPERSCRIPT TWO
52
+ - "³" # SUPERSCRIPT THREE
53
+ - "´" # ACUTE ACCENT
54
+ - "µ" # MICRO SIGN
55
+ - "¶" # PILCROW SIGN
56
+ - "·" # MIDDLE DOT
57
+ - "¸" # CEDILLA
58
+ - "¹" # SUPERSCRIPT ONE
59
+ - "º" # MASCULINE ORDINAL INDICATOR
60
+ - "»" # RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
61
+ - "¼" # VULGAR FRACTION ONE QUARTER
62
+ - "½" # VULGAR FRACTION ONE HALF
63
+ - "¾" # VULGAR FRACTION THREE QUARTERS
64
+ - "¿" # INVERTED QUESTION MARK
65
+ - "À" # LATIN CAPITAL LETTER A WITH GRAVE
66
+ - "Á" # LATIN CAPITAL LETTER A WITH ACUTE
67
+ - "Â" # LATIN CAPITAL LETTER A WITH CIRCUMFLEX
68
+ - "Ã" # LATIN CAPITAL LETTER A WITH TILDE
69
+ - "Ä" # LATIN CAPITAL LETTER A WITH DIAERESIS
70
+ - "Å" # LATIN CAPITAL LETTER A WITH RING ABOVE
71
+ - "Æ" # LATIN CAPITAL LETTER AE
72
+ - "Ç" # LATIN CAPITAL LETTER C WITH CEDILLA
73
+ - "È" # LATIN CAPITAL LETTER E WITH GRAVE
74
+ - "É" # LATIN CAPITAL LETTER E WITH ACUTE
75
+ - "Ê" # LATIN CAPITAL LETTER E WITH CIRCUMFLEX
76
+ - "Ë" # LATIN CAPITAL LETTER E WITH DIAERESIS
77
+ - "Ì" # LATIN CAPITAL LETTER I WITH GRAVE
78
+ - "Í" # LATIN CAPITAL LETTER I WITH ACUTE
79
+ - "Î" # LATIN CAPITAL LETTER I WITH CIRCUMFLEX
80
+ - "Ï" # LATIN CAPITAL LETTER I WITH DIAERESIS
81
+ - "Ð" # LATIN CAPITAL LETTER ETH
82
+ - "Ñ" # LATIN CAPITAL LETTER N WITH TILDE
83
+ - "Ò" # LATIN CAPITAL LETTER O WITH GRAVE
84
+ - "Ó" # LATIN CAPITAL LETTER O WITH ACUTE
85
+ - "Ô" # LATIN CAPITAL LETTER O WITH CIRCUMFLEX
86
+ - "Õ" # LATIN CAPITAL LETTER O WITH TILDE
87
+ - "Ö" # LATIN CAPITAL LETTER O WITH DIAERESIS
88
+ - "×" # MULTIPLICATION SIGN (present in Symbol font)
89
+ - "Ø" # LATIN CAPITAL LETTER O WITH STROKE
90
+ - "Ù" # LATIN CAPITAL LETTER U WITH GRAVE
91
+ - "Ú" # LATIN CAPITAL LETTER U WITH ACUTE
92
+ - "Û" # LATIN CAPITAL LETTER U WITH CIRCUMFLEX
93
+ - "Ü" # LATIN CAPITAL LETTER U WITH DIAERESIS
94
+ - "Ý" # LATIN CAPITAL LETTER Y WITH ACUTE
95
+ - "Þ" # LATIN CAPITAL LETTER THORN
96
+ - "ß" # LATIN SMALL LETTER SHARP S
97
+ - "à" # LATIN SMALL LETTER A WITH GRAVE
98
+ - "á" # LATIN SMALL LETTER A WITH ACUTE
99
+ - "â" # LATIN SMALL LETTER A WITH CIRCUMFLEX
100
+ - "ã" # LATIN SMALL LETTER A WITH TILDE
101
+ - "ä" # LATIN SMALL LETTER A WITH DIAERESIS
102
+ - "å" # LATIN SMALL LETTER A WITH RING ABOVE
103
+ - "æ" # LATIN SMALL LETTER AE
104
+ - "ç" # LATIN SMALL LETTER C WITH CEDILLA
105
+ - "è" # LATIN SMALL LETTER E WITH GRAVE
106
+ - "é" # LATIN SMALL LETTER E WITH ACUTE
107
+ - "ê" # LATIN SMALL LETTER E WITH CIRCUMFLEX
108
+ - "ë" # LATIN SMALL LETTER E WITH DIAERESIS
109
+ - "ì" # LATIN SMALL LETTER I WITH GRAVE
110
+ - "í" # LATIN SMALL LETTER I WITH ACUTE
111
+ - "î" # LATIN SMALL LETTER I WITH CIRCUMFLEX
112
+ - "ï" # LATIN SMALL LETTER I WITH DIAERESIS
113
+ - "ð" # LATIN SMALL LETTER ETH
114
+ - "ñ" # LATIN SMALL LETTER N WITH TILDE
115
+ - "ò" # LATIN SMALL LETTER O WITH GRAVE
116
+ - "ó" # LATIN SMALL LETTER O WITH ACUTE
117
+ - "ô" # LATIN SMALL LETTER O WITH CIRCUMFLEX
118
+ - "õ" # LATIN SMALL LETTER O WITH TILDE
119
+ - "ö" # LATIN SMALL LETTER O WITH DIAERESIS
120
+ - "÷" # DIVISION SIGN (present in Symbol font)
121
+ - "ø" # LATIN SMALL LETTER O WITH STROKE
122
+ - "ù" # LATIN SMALL LETTER U WITH GRAVE
123
+ - "ú" # LATIN SMALL LETTER U WITH ACUTE
124
+ - "û" # LATIN SMALL LETTER U WITH CIRCUMFLEX
125
+ - "ü" # LATIN SMALL LETTER U WITH DIAERESIS
126
+ - "ý" # LATIN SMALL LETTER Y WITH ACUTE
127
+ - "þ" # LATIN SMALL LETTER THORN
128
+ - "ÿ" # LATIN SMALL LETTER Y WITH DIAERESIS