cmess 0.0.4.136

Sign up to get free protection for your applications and to get access to all the features.
Files changed (37) hide show
  1. data/COPYING +676 -0
  2. data/ChangeLog +6 -0
  3. data/README +53 -0
  4. data/Rakefile +30 -0
  5. data/bin/cinderella +186 -0
  6. data/bin/decode_entities +101 -0
  7. data/bin/guess_encoding +183 -0
  8. data/data/csets/iso_8859-1.yaml +195 -0
  9. data/data/csets/iso_8859-15.yaml +204 -0
  10. data/data/csets/latin1.yaml +195 -0
  11. data/data/csets/unicode/basic_latin.yaml +97 -0
  12. data/data/csets/unicode/cyrillic-supplement.yaml +17 -0
  13. data/data/csets/unicode/cyrillic.yaml +256 -0
  14. data/data/csets/unicode/greek.yaml +129 -0
  15. data/data/csets/unicode/ipa_extensions.yaml +97 -0
  16. data/data/csets/unicode/latin-extended-c.yaml +18 -0
  17. data/data/csets/unicode/latin-extended-d.yaml +3 -0
  18. data/data/csets/unicode/latin_1_supplement.yaml +128 -0
  19. data/data/csets/unicode/latin_extended_a.yaml +129 -0
  20. data/data/csets/unicode/latin_extended_additional.yaml +247 -0
  21. data/data/csets/unicode/latin_extended_b.yaml +209 -0
  22. data/data/csets/unicode/letterlike_symbols.yaml +80 -0
  23. data/data/csets/unicode/spacing_modifier_letters.yaml +81 -0
  24. data/data/csets/utf-8.yaml +1504 -0
  25. data/data/csets/utf8.yaml +1504 -0
  26. data/example/crop +127 -0
  27. data/example/crop_repaired +127 -0
  28. data/example/empty6-slash.txt +1495 -0
  29. data/example/empty6-slash_repaired.txt +1495 -0
  30. data/example/pot +1368 -0
  31. data/lib/cmess.rb +44 -0
  32. data/lib/cmess/cinderella.rb +63 -0
  33. data/lib/cmess/cli.rb +79 -0
  34. data/lib/cmess/decode_entities.rb +68 -0
  35. data/lib/cmess/guess_encoding.rb +372 -0
  36. data/lib/cmess/version.rb +51 -0
  37. metadata +119 -0
data/data/csets/unicode/basic_latin.yaml ADDED
@@ -0,0 +1,97 @@
1
+ ---
2
+ - " " # SPACE
3
+ - "!" # EXCLAMATION MARK
4
+ - '"' # QUOTATION MARK
5
+ - "#" # NUMBER SIGN
6
+ - "$" # DOLLAR SIGN
7
+ - "%" # PERCENT SIGN
8
+ - "&" # AMPERSAND
9
+ - "'" # APOSTROPHE
10
+ - "(" # LEFT PARENTHESIS
11
+ - ")" # RIGHT PARENTHESIS
12
+ - "*" # ASTERISK
13
+ - "+" # PLUS SIGN
14
+ - "," # COMMA
15
+ - "-" # HYPHEN-MINUS
16
+ - "." # FULL STOP
17
+ - "/" # SOLIDUS
18
+ - "0" # DIGIT ZERO
19
+ - "1" # DIGIT ONE
20
+ - "2" # DIGIT TWO
21
+ - "3" # DIGIT THREE
22
+ - "4" # DIGIT FOUR
23
+ - "5" # DIGIT FIVE
24
+ - "6" # DIGIT SIX
25
+ - "7" # DIGIT SEVEN
26
+ - "8" # DIGIT EIGHT
27
+ - "9" # DIGIT NINE
28
+ - ":" # COLON
29
+ - ";" # SEMICOLON
30
+ - "<" # LESS-THAN SIGN
31
+ - "=" # EQUALS SIGN
32
+ - ">" # GREATER-THAN SIGN
33
+ - "?" # QUESTION MARK
34
+ - "@" # COMMERCIAL AT
35
+ - "A" # LATIN CAPITAL LETTER A
36
+ - "B" # LATIN CAPITAL LETTER B
37
+ - "C" # LATIN CAPITAL LETTER C
38
+ - "D" # LATIN CAPITAL LETTER D
39
+ - "E" # LATIN CAPITAL LETTER E
40
+ - "F" # LATIN CAPITAL LETTER F
41
+ - "G" # LATIN CAPITAL LETTER G
42
+ - "H" # LATIN CAPITAL LETTER H
43
+ - "I" # LATIN CAPITAL LETTER I
44
+ - "J" # LATIN CAPITAL LETTER J
45
+ - "K" # LATIN CAPITAL LETTER K
46
+ - "L" # LATIN CAPITAL LETTER L
47
+ - "M" # LATIN CAPITAL LETTER M
48
+ - "N" # LATIN CAPITAL LETTER N
49
+ - "O" # LATIN CAPITAL LETTER O
50
+ - "P" # LATIN CAPITAL LETTER P
51
+ - "Q" # LATIN CAPITAL LETTER Q
52
+ - "R" # LATIN CAPITAL LETTER R
53
+ - "S" # LATIN CAPITAL LETTER S
54
+ - "T" # LATIN CAPITAL LETTER T
55
+ - "U" # LATIN CAPITAL LETTER U
56
+ - "V" # LATIN CAPITAL LETTER V
57
+ - "W" # LATIN CAPITAL LETTER W
58
+ - "X" # LATIN CAPITAL LETTER X
59
+ - "Y" # LATIN CAPITAL LETTER Y
60
+ - "Z" # LATIN CAPITAL LETTER Z
61
+ - "[" # LEFT SQUARE BRACKET
62
+ - "\\" # REVERSE SOLIDUS
63
+ - "]" # RIGHT SQUARE BRACKET
64
+ - "^" # CIRCUMFLEX ACCENT
65
+ - "_" # LOW LINE
66
+ - "`" # GRAVE ACCENT
67
+ - "a" # LATIN SMALL LETTER A
68
+ - "b" # LATIN SMALL LETTER B
69
+ - "c" # LATIN SMALL LETTER C
70
+ - "d" # LATIN SMALL LETTER D
71
+ - "e" # LATIN SMALL LETTER E
72
+ - "f" # LATIN SMALL LETTER F
73
+ - "g" # LATIN SMALL LETTER G
74
+ - "h" # LATIN SMALL LETTER H
75
+ - "i" # LATIN SMALL LETTER I
76
+ - "j" # LATIN SMALL LETTER J
77
+ - "k" # LATIN SMALL LETTER K
78
+ - "l" # LATIN SMALL LETTER L
79
+ - "m" # LATIN SMALL LETTER M
80
+ - "n" # LATIN SMALL LETTER N
81
+ - "o" # LATIN SMALL LETTER O
82
+ - "p" # LATIN SMALL LETTER P
83
+ - "q" # LATIN SMALL LETTER Q
84
+ - "r" # LATIN SMALL LETTER R
85
+ - "s" # LATIN SMALL LETTER S
86
+ - "t" # LATIN SMALL LETTER T
87
+ - "u" # LATIN SMALL LETTER U
88
+ - "v" # LATIN SMALL LETTER V
89
+ - "w" # LATIN SMALL LETTER W
90
+ - "x" # LATIN SMALL LETTER X
91
+ - "y" # LATIN SMALL LETTER Y
92
+ - "z" # LATIN SMALL LETTER Z
93
+ - "{" # LEFT CURLY BRACKET
94
+ - "|" # VERTICAL LINE
95
+ - "}" # RIGHT CURLY BRACKET
96
+ - "~" # TILDE
97
+ - "" # DEL
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