regexp_parser 2.1.1 → 2.5.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +94 -6
- data/Gemfile +2 -1
- data/LICENSE +1 -1
- data/README.md +40 -30
- data/Rakefile +6 -70
- data/lib/regexp_parser/error.rb +1 -1
- data/lib/regexp_parser/expression/base.rb +75 -0
- data/lib/regexp_parser/expression/classes/anchor.rb +0 -2
- data/lib/regexp_parser/expression/classes/{backref.rb → backreference.rb} +1 -0
- data/lib/regexp_parser/expression/classes/{set → character_set}/intersection.rb +0 -0
- data/lib/regexp_parser/expression/classes/{set → character_set}/range.rb +2 -2
- data/lib/regexp_parser/expression/classes/{set.rb → character_set.rb} +2 -2
- data/lib/regexp_parser/expression/classes/{type.rb → character_type.rb} +0 -2
- data/lib/regexp_parser/expression/classes/conditional.rb +2 -2
- data/lib/regexp_parser/expression/classes/{escape.rb → escape_sequence.rb} +13 -7
- data/lib/regexp_parser/expression/classes/free_space.rb +1 -3
- data/lib/regexp_parser/expression/classes/group.rb +6 -6
- data/lib/regexp_parser/expression/classes/keep.rb +2 -0
- data/lib/regexp_parser/expression/classes/literal.rb +1 -5
- data/lib/regexp_parser/expression/classes/root.rb +3 -6
- data/lib/regexp_parser/expression/classes/{property.rb → unicode_property.rb} +1 -2
- data/lib/regexp_parser/expression/methods/construct.rb +43 -0
- data/lib/regexp_parser/expression/methods/match_length.rb +1 -1
- data/lib/regexp_parser/expression/methods/strfregexp.rb +1 -1
- data/lib/regexp_parser/expression/methods/tests.rb +10 -1
- data/lib/regexp_parser/expression/quantifier.rb +41 -23
- data/lib/regexp_parser/expression/sequence.rb +9 -24
- data/lib/regexp_parser/expression/sequence_operation.rb +2 -2
- data/lib/regexp_parser/expression/shared.rb +85 -0
- data/lib/regexp_parser/expression/subexpression.rb +11 -8
- data/lib/regexp_parser/expression.rb +10 -132
- data/lib/regexp_parser/lexer.rb +8 -6
- data/lib/regexp_parser/parser.rb +21 -72
- data/lib/regexp_parser/scanner/properties/long.csv +622 -0
- data/lib/regexp_parser/scanner/properties/short.csv +246 -0
- data/lib/regexp_parser/scanner/property.rl +1 -1
- data/lib/regexp_parser/scanner/scanner.rl +48 -35
- data/lib/regexp_parser/scanner.rb +735 -801
- data/lib/regexp_parser/syntax/any.rb +2 -7
- data/lib/regexp_parser/syntax/base.rb +91 -66
- data/lib/regexp_parser/syntax/token/anchor.rb +15 -0
- data/lib/regexp_parser/syntax/{tokens → token}/assertion.rb +2 -2
- data/lib/regexp_parser/syntax/token/backreference.rb +30 -0
- data/lib/regexp_parser/syntax/{tokens → token}/character_set.rb +2 -2
- data/lib/regexp_parser/syntax/{tokens → token}/character_type.rb +3 -3
- data/lib/regexp_parser/syntax/{tokens → token}/conditional.rb +3 -3
- data/lib/regexp_parser/syntax/token/escape.rb +31 -0
- data/lib/regexp_parser/syntax/{tokens → token}/group.rb +7 -7
- data/lib/regexp_parser/syntax/{tokens → token}/keep.rb +1 -1
- data/lib/regexp_parser/syntax/{tokens → token}/meta.rb +2 -2
- data/lib/regexp_parser/syntax/{tokens → token}/posix_class.rb +3 -3
- data/lib/regexp_parser/syntax/token/quantifier.rb +35 -0
- data/lib/regexp_parser/syntax/token/unicode_property.rb +717 -0
- data/lib/regexp_parser/syntax/token.rb +45 -0
- data/lib/regexp_parser/syntax/version_lookup.rb +20 -29
- data/lib/regexp_parser/syntax/versions/1.8.6.rb +13 -20
- data/lib/regexp_parser/syntax/versions/1.9.1.rb +10 -17
- data/lib/regexp_parser/syntax/versions/1.9.3.rb +3 -10
- data/lib/regexp_parser/syntax/versions/2.0.0.rb +8 -15
- data/lib/regexp_parser/syntax/versions/2.2.0.rb +3 -9
- data/lib/regexp_parser/syntax/versions/2.3.0.rb +3 -9
- data/lib/regexp_parser/syntax/versions/2.4.0.rb +3 -9
- data/lib/regexp_parser/syntax/versions/2.4.1.rb +2 -8
- data/lib/regexp_parser/syntax/versions/2.5.0.rb +3 -9
- data/lib/regexp_parser/syntax/versions/2.6.0.rb +3 -9
- data/lib/regexp_parser/syntax/versions/2.6.2.rb +3 -9
- data/lib/regexp_parser/syntax/versions/2.6.3.rb +3 -9
- data/lib/regexp_parser/syntax/versions/3.1.0.rb +4 -0
- data/lib/regexp_parser/syntax/versions/3.2.0.rb +4 -0
- data/lib/regexp_parser/syntax/versions.rb +1 -1
- data/lib/regexp_parser/syntax.rb +1 -1
- data/lib/regexp_parser/token.rb +9 -20
- data/lib/regexp_parser/version.rb +1 -1
- data/lib/regexp_parser.rb +0 -2
- data/regexp_parser.gemspec +20 -22
- metadata +37 -166
- data/lib/regexp_parser/scanner/properties/long.yml +0 -594
- data/lib/regexp_parser/scanner/properties/short.yml +0 -237
- data/lib/regexp_parser/syntax/tokens/anchor.rb +0 -15
- data/lib/regexp_parser/syntax/tokens/backref.rb +0 -24
- data/lib/regexp_parser/syntax/tokens/escape.rb +0 -30
- data/lib/regexp_parser/syntax/tokens/quantifier.rb +0 -35
- data/lib/regexp_parser/syntax/tokens/unicode_property.rb +0 -675
- data/lib/regexp_parser/syntax/tokens.rb +0 -45
- data/spec/expression/base_spec.rb +0 -104
- data/spec/expression/clone_spec.rb +0 -152
- data/spec/expression/conditional_spec.rb +0 -89
- data/spec/expression/free_space_spec.rb +0 -27
- data/spec/expression/methods/match_length_spec.rb +0 -161
- data/spec/expression/methods/match_spec.rb +0 -25
- data/spec/expression/methods/strfregexp_spec.rb +0 -224
- data/spec/expression/methods/tests_spec.rb +0 -99
- data/spec/expression/methods/traverse_spec.rb +0 -161
- data/spec/expression/options_spec.rb +0 -128
- data/spec/expression/subexpression_spec.rb +0 -50
- data/spec/expression/to_h_spec.rb +0 -26
- data/spec/expression/to_s_spec.rb +0 -108
- data/spec/lexer/all_spec.rb +0 -22
- data/spec/lexer/conditionals_spec.rb +0 -53
- data/spec/lexer/delimiters_spec.rb +0 -68
- data/spec/lexer/escapes_spec.rb +0 -14
- data/spec/lexer/keep_spec.rb +0 -10
- data/spec/lexer/literals_spec.rb +0 -64
- data/spec/lexer/nesting_spec.rb +0 -99
- data/spec/lexer/refcalls_spec.rb +0 -60
- data/spec/parser/all_spec.rb +0 -43
- data/spec/parser/alternation_spec.rb +0 -88
- data/spec/parser/anchors_spec.rb +0 -17
- data/spec/parser/conditionals_spec.rb +0 -179
- data/spec/parser/errors_spec.rb +0 -30
- data/spec/parser/escapes_spec.rb +0 -121
- data/spec/parser/free_space_spec.rb +0 -130
- data/spec/parser/groups_spec.rb +0 -108
- data/spec/parser/keep_spec.rb +0 -6
- data/spec/parser/options_spec.rb +0 -28
- data/spec/parser/posix_classes_spec.rb +0 -8
- data/spec/parser/properties_spec.rb +0 -115
- data/spec/parser/quantifiers_spec.rb +0 -68
- data/spec/parser/refcalls_spec.rb +0 -117
- data/spec/parser/set/intersections_spec.rb +0 -127
- data/spec/parser/set/ranges_spec.rb +0 -111
- data/spec/parser/sets_spec.rb +0 -178
- data/spec/parser/types_spec.rb +0 -18
- data/spec/scanner/all_spec.rb +0 -18
- data/spec/scanner/anchors_spec.rb +0 -21
- data/spec/scanner/conditionals_spec.rb +0 -128
- data/spec/scanner/delimiters_spec.rb +0 -52
- data/spec/scanner/errors_spec.rb +0 -67
- data/spec/scanner/escapes_spec.rb +0 -64
- data/spec/scanner/free_space_spec.rb +0 -165
- data/spec/scanner/groups_spec.rb +0 -61
- data/spec/scanner/keep_spec.rb +0 -10
- data/spec/scanner/literals_spec.rb +0 -39
- data/spec/scanner/meta_spec.rb +0 -18
- data/spec/scanner/options_spec.rb +0 -36
- data/spec/scanner/properties_spec.rb +0 -64
- data/spec/scanner/quantifiers_spec.rb +0 -25
- data/spec/scanner/refcalls_spec.rb +0 -55
- data/spec/scanner/sets_spec.rb +0 -151
- data/spec/scanner/types_spec.rb +0 -14
- data/spec/spec_helper.rb +0 -16
- data/spec/support/runner.rb +0 -42
- data/spec/support/shared_examples.rb +0 -77
- data/spec/support/warning_extractor.rb +0 -60
- data/spec/syntax/syntax_spec.rb +0 -48
- data/spec/syntax/syntax_token_map_spec.rb +0 -23
- data/spec/syntax/versions/1.8.6_spec.rb +0 -17
- data/spec/syntax/versions/1.9.1_spec.rb +0 -10
- data/spec/syntax/versions/1.9.3_spec.rb +0 -9
- data/spec/syntax/versions/2.0.0_spec.rb +0 -13
- data/spec/syntax/versions/2.2.0_spec.rb +0 -9
- data/spec/syntax/versions/aliases_spec.rb +0 -37
- data/spec/token/token_spec.rb +0 -85
@@ -0,0 +1,717 @@
|
|
1
|
+
module Regexp::Syntax
|
2
|
+
module Token
|
3
|
+
module UnicodeProperty
|
4
|
+
all = proc { |name| constants.grep(/#{name}/).flat_map(&method(:const_get)) }
|
5
|
+
|
6
|
+
CharType_V1_9_0 = %i[alnum alpha ascii blank cntrl digit graph
|
7
|
+
lower print punct space upper word xdigit]
|
8
|
+
|
9
|
+
CharType_V2_5_0 = %i[xposixpunct]
|
10
|
+
|
11
|
+
POSIX = %i[any assigned newline]
|
12
|
+
|
13
|
+
module Category
|
14
|
+
Letter = %i[letter uppercase_letter lowercase_letter
|
15
|
+
titlecase_letter modifier_letter other_letter]
|
16
|
+
|
17
|
+
Mark = %i[mark nonspacing_mark spacing_mark
|
18
|
+
enclosing_mark]
|
19
|
+
|
20
|
+
Number = %i[number decimal_number letter_number
|
21
|
+
other_number]
|
22
|
+
|
23
|
+
Punctuation = %i[punctuation connector_punctuation dash_punctuation
|
24
|
+
open_punctuation close_punctuation initial_punctuation
|
25
|
+
final_punctuation other_punctuation]
|
26
|
+
|
27
|
+
Symbol = %i[symbol math_symbol currency_symbol
|
28
|
+
modifier_symbol other_symbol]
|
29
|
+
|
30
|
+
Separator = %i[separator space_separator line_separator
|
31
|
+
paragraph_separator]
|
32
|
+
|
33
|
+
Codepoint = %i[other control format
|
34
|
+
surrogate private_use unassigned]
|
35
|
+
|
36
|
+
All = Letter + Mark + Number + Punctuation +
|
37
|
+
Symbol + Separator + Codepoint
|
38
|
+
end
|
39
|
+
|
40
|
+
Age_V1_9_3 = %i[age=1.1 age=2.0 age=2.1 age=3.0 age=3.1
|
41
|
+
age=3.2 age=4.0 age=4.1 age=5.0 age=5.1
|
42
|
+
age=5.2 age=6.0]
|
43
|
+
|
44
|
+
Age_V2_0_0 = %i[age=6.1]
|
45
|
+
|
46
|
+
Age_V2_2_0 = %i[age=6.2 age=6.3 age=7.0]
|
47
|
+
|
48
|
+
Age_V2_3_0 = %i[age=8.0]
|
49
|
+
|
50
|
+
Age_V2_4_0 = %i[age=9.0]
|
51
|
+
|
52
|
+
Age_V2_5_0 = %i[age=10.0]
|
53
|
+
|
54
|
+
Age_V2_6_0 = %i[age=11.0]
|
55
|
+
|
56
|
+
Age_V2_6_2 = %i[age=12.0]
|
57
|
+
|
58
|
+
Age_V2_6_3 = %i[age=12.1]
|
59
|
+
|
60
|
+
Age_V3_1_0 = %i[age=13.0]
|
61
|
+
|
62
|
+
Age_V3_2_0 = %i[age=14.0]
|
63
|
+
|
64
|
+
Age = all[:Age_V]
|
65
|
+
|
66
|
+
Derived_V1_9_0 = %i[
|
67
|
+
ascii_hex_digit
|
68
|
+
alphabetic
|
69
|
+
cased
|
70
|
+
changes_when_casefolded
|
71
|
+
changes_when_casemapped
|
72
|
+
changes_when_lowercased
|
73
|
+
changes_when_titlecased
|
74
|
+
changes_when_uppercased
|
75
|
+
case_ignorable
|
76
|
+
bidi_control
|
77
|
+
dash
|
78
|
+
deprecated
|
79
|
+
default_ignorable_code_point
|
80
|
+
diacritic
|
81
|
+
extender
|
82
|
+
grapheme_base
|
83
|
+
grapheme_extend
|
84
|
+
grapheme_link
|
85
|
+
hex_digit
|
86
|
+
hyphen
|
87
|
+
id_continue
|
88
|
+
ideographic
|
89
|
+
id_start
|
90
|
+
ids_binary_operator
|
91
|
+
ids_trinary_operator
|
92
|
+
join_control
|
93
|
+
logical_order_exception
|
94
|
+
lowercase
|
95
|
+
math
|
96
|
+
noncharacter_code_point
|
97
|
+
other_alphabetic
|
98
|
+
other_default_ignorable_code_point
|
99
|
+
other_grapheme_extend
|
100
|
+
other_id_continue
|
101
|
+
other_id_start
|
102
|
+
other_lowercase
|
103
|
+
other_math
|
104
|
+
other_uppercase
|
105
|
+
pattern_syntax
|
106
|
+
pattern_white_space
|
107
|
+
quotation_mark
|
108
|
+
radical
|
109
|
+
sentence_terminal
|
110
|
+
soft_dotted
|
111
|
+
terminal_punctuation
|
112
|
+
unified_ideograph
|
113
|
+
uppercase
|
114
|
+
variation_selector
|
115
|
+
white_space
|
116
|
+
xid_start
|
117
|
+
xid_continue
|
118
|
+
]
|
119
|
+
|
120
|
+
Derived_V2_0_0 = %i[
|
121
|
+
cased_letter
|
122
|
+
combining_mark
|
123
|
+
]
|
124
|
+
|
125
|
+
Derived_V2_4_0 = %i[
|
126
|
+
prepended_concatenation_mark
|
127
|
+
]
|
128
|
+
|
129
|
+
Derived_V2_5_0 = %i[
|
130
|
+
regional_indicator
|
131
|
+
]
|
132
|
+
|
133
|
+
Derived = all[:Derived_V]
|
134
|
+
|
135
|
+
Script_V1_9_0 = %i[
|
136
|
+
arabic
|
137
|
+
imperial_aramaic
|
138
|
+
armenian
|
139
|
+
avestan
|
140
|
+
balinese
|
141
|
+
bamum
|
142
|
+
bengali
|
143
|
+
bopomofo
|
144
|
+
braille
|
145
|
+
buginese
|
146
|
+
buhid
|
147
|
+
canadian_aboriginal
|
148
|
+
carian
|
149
|
+
cham
|
150
|
+
cherokee
|
151
|
+
coptic
|
152
|
+
cypriot
|
153
|
+
cyrillic
|
154
|
+
devanagari
|
155
|
+
deseret
|
156
|
+
egyptian_hieroglyphs
|
157
|
+
ethiopic
|
158
|
+
georgian
|
159
|
+
glagolitic
|
160
|
+
gothic
|
161
|
+
greek
|
162
|
+
gujarati
|
163
|
+
gurmukhi
|
164
|
+
hangul
|
165
|
+
han
|
166
|
+
hanunoo
|
167
|
+
hebrew
|
168
|
+
hiragana
|
169
|
+
old_italic
|
170
|
+
javanese
|
171
|
+
kayah_li
|
172
|
+
katakana
|
173
|
+
kharoshthi
|
174
|
+
khmer
|
175
|
+
kannada
|
176
|
+
kaithi
|
177
|
+
tai_tham
|
178
|
+
lao
|
179
|
+
latin
|
180
|
+
lepcha
|
181
|
+
limbu
|
182
|
+
linear_b
|
183
|
+
lisu
|
184
|
+
lycian
|
185
|
+
lydian
|
186
|
+
malayalam
|
187
|
+
mongolian
|
188
|
+
meetei_mayek
|
189
|
+
myanmar
|
190
|
+
nko
|
191
|
+
ogham
|
192
|
+
ol_chiki
|
193
|
+
old_turkic
|
194
|
+
oriya
|
195
|
+
osmanya
|
196
|
+
phags_pa
|
197
|
+
inscriptional_pahlavi
|
198
|
+
phoenician
|
199
|
+
inscriptional_parthian
|
200
|
+
rejang
|
201
|
+
runic
|
202
|
+
samaritan
|
203
|
+
old_south_arabian
|
204
|
+
saurashtra
|
205
|
+
shavian
|
206
|
+
sinhala
|
207
|
+
sundanese
|
208
|
+
syloti_nagri
|
209
|
+
syriac
|
210
|
+
tagbanwa
|
211
|
+
tai_le
|
212
|
+
new_tai_lue
|
213
|
+
tamil
|
214
|
+
tai_viet
|
215
|
+
telugu
|
216
|
+
tifinagh
|
217
|
+
tagalog
|
218
|
+
thaana
|
219
|
+
thai
|
220
|
+
tibetan
|
221
|
+
ugaritic
|
222
|
+
vai
|
223
|
+
old_persian
|
224
|
+
cuneiform
|
225
|
+
yi
|
226
|
+
inherited
|
227
|
+
common
|
228
|
+
unknown
|
229
|
+
]
|
230
|
+
|
231
|
+
Script_V1_9_3 = %i[
|
232
|
+
brahmi
|
233
|
+
batak
|
234
|
+
mandaic
|
235
|
+
]
|
236
|
+
|
237
|
+
Script_V2_0_0 = %i[
|
238
|
+
chakma
|
239
|
+
meroitic_cursive
|
240
|
+
meroitic_hieroglyphs
|
241
|
+
miao
|
242
|
+
sharada
|
243
|
+
sora_sompeng
|
244
|
+
takri
|
245
|
+
]
|
246
|
+
|
247
|
+
Script_V2_2_0 = %i[
|
248
|
+
caucasian_albanian
|
249
|
+
bassa_vah
|
250
|
+
duployan
|
251
|
+
elbasan
|
252
|
+
grantha
|
253
|
+
pahawh_hmong
|
254
|
+
khojki
|
255
|
+
linear_a
|
256
|
+
mahajani
|
257
|
+
manichaean
|
258
|
+
mende_kikakui
|
259
|
+
modi
|
260
|
+
mro
|
261
|
+
old_north_arabian
|
262
|
+
nabataean
|
263
|
+
palmyrene
|
264
|
+
pau_cin_hau
|
265
|
+
old_permic
|
266
|
+
psalter_pahlavi
|
267
|
+
siddham
|
268
|
+
khudawadi
|
269
|
+
tirhuta
|
270
|
+
warang_citi
|
271
|
+
]
|
272
|
+
|
273
|
+
Script_V2_3_0 = %i[
|
274
|
+
ahom
|
275
|
+
anatolian_hieroglyphs
|
276
|
+
hatran
|
277
|
+
multani
|
278
|
+
old_hungarian
|
279
|
+
signwriting
|
280
|
+
]
|
281
|
+
|
282
|
+
Script_V2_4_0 = %i[
|
283
|
+
adlam
|
284
|
+
bhaiksuki
|
285
|
+
marchen
|
286
|
+
newa
|
287
|
+
osage
|
288
|
+
tangut
|
289
|
+
]
|
290
|
+
|
291
|
+
Script_V2_5_0 = %i[
|
292
|
+
masaram_gondi
|
293
|
+
nushu
|
294
|
+
soyombo
|
295
|
+
zanabazar_square
|
296
|
+
]
|
297
|
+
|
298
|
+
Script_V2_6_0 = %i[
|
299
|
+
dogra
|
300
|
+
gunjala_gondi
|
301
|
+
hanifi_rohingya
|
302
|
+
makasar
|
303
|
+
medefaidrin
|
304
|
+
old_sogdian
|
305
|
+
sogdian
|
306
|
+
]
|
307
|
+
|
308
|
+
Script_V2_6_2 = %i[
|
309
|
+
elymaic
|
310
|
+
nandinagari
|
311
|
+
nyiakeng_puachue_hmong
|
312
|
+
wancho
|
313
|
+
]
|
314
|
+
|
315
|
+
Script_V3_1_0 = %i[
|
316
|
+
chorasmian
|
317
|
+
dives_akuru
|
318
|
+
khitan_small_script
|
319
|
+
yezidi
|
320
|
+
]
|
321
|
+
|
322
|
+
Script_V3_2_0 = %i[
|
323
|
+
cypro_minoan
|
324
|
+
old_uyghur
|
325
|
+
tangsa
|
326
|
+
toto
|
327
|
+
vithkuqi
|
328
|
+
]
|
329
|
+
|
330
|
+
Script = all[:Script_V]
|
331
|
+
|
332
|
+
UnicodeBlock_V1_9_0 = %i[
|
333
|
+
in_alphabetic_presentation_forms
|
334
|
+
in_arabic
|
335
|
+
in_armenian
|
336
|
+
in_arrows
|
337
|
+
in_basic_latin
|
338
|
+
in_bengali
|
339
|
+
in_block_elements
|
340
|
+
in_bopomofo_extended
|
341
|
+
in_bopomofo
|
342
|
+
in_box_drawing
|
343
|
+
in_braille_patterns
|
344
|
+
in_buhid
|
345
|
+
in_cjk_compatibility_forms
|
346
|
+
in_cjk_compatibility_ideographs
|
347
|
+
in_cjk_compatibility
|
348
|
+
in_cjk_radicals_supplement
|
349
|
+
in_cjk_symbols_and_punctuation
|
350
|
+
in_cjk_unified_ideographs_extension_a
|
351
|
+
in_cjk_unified_ideographs
|
352
|
+
in_cherokee
|
353
|
+
in_combining_diacritical_marks_for_symbols
|
354
|
+
in_combining_diacritical_marks
|
355
|
+
in_combining_half_marks
|
356
|
+
in_control_pictures
|
357
|
+
in_currency_symbols
|
358
|
+
in_cyrillic_supplement
|
359
|
+
in_cyrillic
|
360
|
+
in_devanagari
|
361
|
+
in_dingbats
|
362
|
+
in_enclosed_alphanumerics
|
363
|
+
in_enclosed_cjk_letters_and_months
|
364
|
+
in_ethiopic
|
365
|
+
in_general_punctuation
|
366
|
+
in_geometric_shapes
|
367
|
+
in_georgian
|
368
|
+
in_greek_extended
|
369
|
+
in_greek_and_coptic
|
370
|
+
in_gujarati
|
371
|
+
in_gurmukhi
|
372
|
+
in_halfwidth_and_fullwidth_forms
|
373
|
+
in_hangul_compatibility_jamo
|
374
|
+
in_hangul_jamo
|
375
|
+
in_hangul_syllables
|
376
|
+
in_hanunoo
|
377
|
+
in_hebrew
|
378
|
+
in_high_private_use_surrogates
|
379
|
+
in_high_surrogates
|
380
|
+
in_hiragana
|
381
|
+
in_ipa_extensions
|
382
|
+
in_ideographic_description_characters
|
383
|
+
in_kanbun
|
384
|
+
in_kangxi_radicals
|
385
|
+
in_kannada
|
386
|
+
in_katakana_phonetic_extensions
|
387
|
+
in_katakana
|
388
|
+
in_khmer_symbols
|
389
|
+
in_khmer
|
390
|
+
in_lao
|
391
|
+
in_latin_extended_additional
|
392
|
+
in_letterlike_symbols
|
393
|
+
in_limbu
|
394
|
+
in_low_surrogates
|
395
|
+
in_malayalam
|
396
|
+
in_mathematical_operators
|
397
|
+
in_miscellaneous_symbols_and_arrows
|
398
|
+
in_miscellaneous_symbols
|
399
|
+
in_miscellaneous_technical
|
400
|
+
in_mongolian
|
401
|
+
in_myanmar
|
402
|
+
in_number_forms
|
403
|
+
in_ogham
|
404
|
+
in_optical_character_recognition
|
405
|
+
in_oriya
|
406
|
+
in_phonetic_extensions
|
407
|
+
in_private_use_area
|
408
|
+
in_runic
|
409
|
+
in_sinhala
|
410
|
+
in_small_form_variants
|
411
|
+
in_spacing_modifier_letters
|
412
|
+
in_specials
|
413
|
+
in_superscripts_and_subscripts
|
414
|
+
in_supplemental_mathematical_operators
|
415
|
+
in_syriac
|
416
|
+
in_tagalog
|
417
|
+
in_tagbanwa
|
418
|
+
in_tai_le
|
419
|
+
in_tamil
|
420
|
+
in_telugu
|
421
|
+
in_thaana
|
422
|
+
in_thai
|
423
|
+
in_tibetan
|
424
|
+
in_unified_canadian_aboriginal_syllabics
|
425
|
+
in_variation_selectors
|
426
|
+
in_yi_radicals
|
427
|
+
in_yi_syllables
|
428
|
+
in_yijing_hexagram_symbols
|
429
|
+
]
|
430
|
+
|
431
|
+
UnicodeBlock_V2_0_0 = %i[
|
432
|
+
in_aegean_numbers
|
433
|
+
in_alchemical_symbols
|
434
|
+
in_ancient_greek_musical_notation
|
435
|
+
in_ancient_greek_numbers
|
436
|
+
in_ancient_symbols
|
437
|
+
in_arabic_extended_a
|
438
|
+
in_arabic_mathematical_alphabetic_symbols
|
439
|
+
in_arabic_presentation_forms_a
|
440
|
+
in_arabic_presentation_forms_b
|
441
|
+
in_arabic_supplement
|
442
|
+
in_avestan
|
443
|
+
in_balinese
|
444
|
+
in_bamum
|
445
|
+
in_bamum_supplement
|
446
|
+
in_batak
|
447
|
+
in_brahmi
|
448
|
+
in_buginese
|
449
|
+
in_byzantine_musical_symbols
|
450
|
+
in_cjk_compatibility_ideographs_supplement
|
451
|
+
in_cjk_strokes
|
452
|
+
in_cjk_unified_ideographs_extension_b
|
453
|
+
in_cjk_unified_ideographs_extension_c
|
454
|
+
in_cjk_unified_ideographs_extension_d
|
455
|
+
in_carian
|
456
|
+
in_chakma
|
457
|
+
in_cham
|
458
|
+
in_combining_diacritical_marks_supplement
|
459
|
+
in_common_indic_number_forms
|
460
|
+
in_coptic
|
461
|
+
in_counting_rod_numerals
|
462
|
+
in_cuneiform
|
463
|
+
in_cuneiform_numbers_and_punctuation
|
464
|
+
in_cypriot_syllabary
|
465
|
+
in_cyrillic_extended_a
|
466
|
+
in_cyrillic_extended_b
|
467
|
+
in_deseret
|
468
|
+
in_devanagari_extended
|
469
|
+
in_domino_tiles
|
470
|
+
in_egyptian_hieroglyphs
|
471
|
+
in_emoticons
|
472
|
+
in_enclosed_alphanumeric_supplement
|
473
|
+
in_enclosed_ideographic_supplement
|
474
|
+
in_ethiopic_extended
|
475
|
+
in_ethiopic_extended_a
|
476
|
+
in_ethiopic_supplement
|
477
|
+
in_georgian_supplement
|
478
|
+
in_glagolitic
|
479
|
+
in_gothic
|
480
|
+
in_hangul_jamo_extended_a
|
481
|
+
in_hangul_jamo_extended_b
|
482
|
+
in_imperial_aramaic
|
483
|
+
in_inscriptional_pahlavi
|
484
|
+
in_inscriptional_parthian
|
485
|
+
in_javanese
|
486
|
+
in_kaithi
|
487
|
+
in_kana_supplement
|
488
|
+
in_kayah_li
|
489
|
+
in_kharoshthi
|
490
|
+
in_latin_1_supplement
|
491
|
+
in_latin_extended_a
|
492
|
+
in_latin_extended_b
|
493
|
+
in_latin_extended_c
|
494
|
+
in_latin_extended_d
|
495
|
+
in_lepcha
|
496
|
+
in_linear_b_ideograms
|
497
|
+
in_linear_b_syllabary
|
498
|
+
in_lisu
|
499
|
+
in_lycian
|
500
|
+
in_lydian
|
501
|
+
in_mahjong_tiles
|
502
|
+
in_mandaic
|
503
|
+
in_mathematical_alphanumeric_symbols
|
504
|
+
in_meetei_mayek
|
505
|
+
in_meetei_mayek_extensions
|
506
|
+
in_meroitic_cursive
|
507
|
+
in_meroitic_hieroglyphs
|
508
|
+
in_miao
|
509
|
+
in_miscellaneous_mathematical_symbols_a
|
510
|
+
in_miscellaneous_mathematical_symbols_b
|
511
|
+
in_miscellaneous_symbols_and_pictographs
|
512
|
+
in_modifier_tone_letters
|
513
|
+
in_musical_symbols
|
514
|
+
in_myanmar_extended_a
|
515
|
+
in_nko
|
516
|
+
in_new_tai_lue
|
517
|
+
in_no_block
|
518
|
+
in_ol_chiki
|
519
|
+
in_old_italic
|
520
|
+
in_old_persian
|
521
|
+
in_old_south_arabian
|
522
|
+
in_old_turkic
|
523
|
+
in_osmanya
|
524
|
+
in_phags_pa
|
525
|
+
in_phaistos_disc
|
526
|
+
in_phoenician
|
527
|
+
in_phonetic_extensions_supplement
|
528
|
+
in_playing_cards
|
529
|
+
in_rejang
|
530
|
+
in_rumi_numeral_symbols
|
531
|
+
in_samaritan
|
532
|
+
in_saurashtra
|
533
|
+
in_sharada
|
534
|
+
in_shavian
|
535
|
+
in_sora_sompeng
|
536
|
+
in_sundanese
|
537
|
+
in_sundanese_supplement
|
538
|
+
in_supplemental_arrows_a
|
539
|
+
in_supplemental_arrows_b
|
540
|
+
in_supplemental_punctuation
|
541
|
+
in_supplementary_private_use_area_a
|
542
|
+
in_supplementary_private_use_area_b
|
543
|
+
in_syloti_nagri
|
544
|
+
in_tags
|
545
|
+
in_tai_tham
|
546
|
+
in_tai_viet
|
547
|
+
in_tai_xuan_jing_symbols
|
548
|
+
in_takri
|
549
|
+
in_tifinagh
|
550
|
+
in_transport_and_map_symbols
|
551
|
+
in_ugaritic
|
552
|
+
in_unified_canadian_aboriginal_syllabics_extended
|
553
|
+
in_vai
|
554
|
+
in_variation_selectors_supplement
|
555
|
+
in_vedic_extensions
|
556
|
+
in_vertical_forms
|
557
|
+
]
|
558
|
+
|
559
|
+
UnicodeBlock_V2_2_0 = %i[
|
560
|
+
in_bassa_vah
|
561
|
+
in_caucasian_albanian
|
562
|
+
in_combining_diacritical_marks_extended
|
563
|
+
in_coptic_epact_numbers
|
564
|
+
in_duployan
|
565
|
+
in_elbasan
|
566
|
+
in_geometric_shapes_extended
|
567
|
+
in_grantha
|
568
|
+
in_khojki
|
569
|
+
in_khudawadi
|
570
|
+
in_latin_extended_e
|
571
|
+
in_linear_a
|
572
|
+
in_mahajani
|
573
|
+
in_manichaean
|
574
|
+
in_mende_kikakui
|
575
|
+
in_modi
|
576
|
+
in_mro
|
577
|
+
in_myanmar_extended_b
|
578
|
+
in_nabataean
|
579
|
+
in_old_north_arabian
|
580
|
+
in_old_permic
|
581
|
+
in_ornamental_dingbats
|
582
|
+
in_pahawh_hmong
|
583
|
+
in_palmyrene
|
584
|
+
in_pau_cin_hau
|
585
|
+
in_psalter_pahlavi
|
586
|
+
in_shorthand_format_controls
|
587
|
+
in_siddham
|
588
|
+
in_sinhala_archaic_numbers
|
589
|
+
in_supplemental_arrows_c
|
590
|
+
in_tirhuta
|
591
|
+
in_warang_citi
|
592
|
+
]
|
593
|
+
|
594
|
+
UnicodeBlock_V2_3_0 = %i[
|
595
|
+
in_ahom
|
596
|
+
in_anatolian_hieroglyphs
|
597
|
+
in_cjk_unified_ideographs_extension_e
|
598
|
+
in_cherokee_supplement
|
599
|
+
in_early_dynastic_cuneiform
|
600
|
+
in_hatran
|
601
|
+
in_multani
|
602
|
+
in_old_hungarian
|
603
|
+
in_supplemental_symbols_and_pictographs
|
604
|
+
in_sutton_signwriting
|
605
|
+
]
|
606
|
+
|
607
|
+
UnicodeBlock_V2_4_0 = %i[
|
608
|
+
in_adlam
|
609
|
+
in_bhaiksuki
|
610
|
+
in_cyrillic_extended_c
|
611
|
+
in_glagolitic_supplement
|
612
|
+
in_ideographic_symbols_and_punctuation
|
613
|
+
in_marchen
|
614
|
+
in_mongolian_supplement
|
615
|
+
in_newa
|
616
|
+
in_osage
|
617
|
+
in_tangut
|
618
|
+
in_tangut_components
|
619
|
+
]
|
620
|
+
|
621
|
+
UnicodeBlock_V2_5_0 = %i[
|
622
|
+
in_cjk_unified_ideographs_extension_f
|
623
|
+
in_kana_extended_a
|
624
|
+
in_masaram_gondi
|
625
|
+
in_nushu
|
626
|
+
in_soyombo
|
627
|
+
in_syriac_supplement
|
628
|
+
in_zanabazar_square
|
629
|
+
]
|
630
|
+
|
631
|
+
UnicodeBlock_V2_6_0 = %i[
|
632
|
+
in_chess_symbols
|
633
|
+
in_dogra
|
634
|
+
in_georgian_extended
|
635
|
+
in_gunjala_gondi
|
636
|
+
in_hanifi_rohingya
|
637
|
+
in_indic_siyaq_numbers
|
638
|
+
in_makasar
|
639
|
+
in_mayan_numerals
|
640
|
+
in_medefaidrin
|
641
|
+
in_old_sogdian
|
642
|
+
in_sogdian
|
643
|
+
]
|
644
|
+
|
645
|
+
UnicodeBlock_V2_6_2 = %i[
|
646
|
+
in_egyptian_hieroglyph_format_controls
|
647
|
+
in_elymaic
|
648
|
+
in_nandinagari
|
649
|
+
in_nyiakeng_puachue_hmong
|
650
|
+
in_ottoman_siyaq_numbers
|
651
|
+
in_small_kana_extension
|
652
|
+
in_symbols_and_pictographs_extended_a
|
653
|
+
in_tamil_supplement
|
654
|
+
in_wancho
|
655
|
+
]
|
656
|
+
|
657
|
+
UnicodeBlock_V3_1_0 = %i[
|
658
|
+
in_chorasmian
|
659
|
+
in_cjk_unified_ideographs_extension_g
|
660
|
+
in_dives_akuru
|
661
|
+
in_khitan_small_script
|
662
|
+
in_lisu_supplement
|
663
|
+
in_symbols_for_legacy_computing
|
664
|
+
in_tangut_supplement
|
665
|
+
in_yezidi
|
666
|
+
]
|
667
|
+
|
668
|
+
UnicodeBlock_V3_2_0 = %i[
|
669
|
+
in_arabic_extended_b
|
670
|
+
in_cypro_minoan
|
671
|
+
in_ethiopic_extended_b
|
672
|
+
in_kana_extended_b
|
673
|
+
in_latin_extended_f
|
674
|
+
in_latin_extended_g
|
675
|
+
in_old_uyghur
|
676
|
+
in_tangsa
|
677
|
+
in_toto
|
678
|
+
in_unified_canadian_aboriginal_syllabics_extended_a
|
679
|
+
in_vithkuqi
|
680
|
+
in_znamenny_musical_notation
|
681
|
+
]
|
682
|
+
|
683
|
+
UnicodeBlock = all[:UnicodeBlock_V]
|
684
|
+
|
685
|
+
Emoji_V2_5_0 = %i[
|
686
|
+
emoji
|
687
|
+
emoji_component
|
688
|
+
emoji_modifier
|
689
|
+
emoji_modifier_base
|
690
|
+
emoji_presentation
|
691
|
+
]
|
692
|
+
|
693
|
+
Emoji = all[:Emoji_V]
|
694
|
+
|
695
|
+
V1_9_0 = Category::All + POSIX + all[:V1_9_0]
|
696
|
+
V1_9_3 = all[:V1_9_3]
|
697
|
+
V2_0_0 = all[:V2_0_0]
|
698
|
+
V2_2_0 = all[:V2_2_0]
|
699
|
+
V2_3_0 = all[:V2_3_0]
|
700
|
+
V2_4_0 = all[:V2_4_0]
|
701
|
+
V2_5_0 = all[:V2_5_0]
|
702
|
+
V2_6_0 = all[:V2_6_0]
|
703
|
+
V2_6_2 = all[:V2_6_2]
|
704
|
+
V2_6_3 = all[:V2_6_3]
|
705
|
+
V3_1_0 = all[:V3_1_0]
|
706
|
+
V3_2_0 = all[:V3_2_0]
|
707
|
+
|
708
|
+
All = all[/^V\d+_\d+_\d+$/]
|
709
|
+
|
710
|
+
Type = :property
|
711
|
+
NonType = :nonproperty
|
712
|
+
end
|
713
|
+
|
714
|
+
Map[UnicodeProperty::Type] = UnicodeProperty::All
|
715
|
+
Map[UnicodeProperty::NonType] = UnicodeProperty::All
|
716
|
+
end
|
717
|
+
end
|