regexp_parser 2.11.1 → 2.11.2
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.
- checksums.yaml +4 -4
- data/Gemfile +2 -0
- data/Rakefile +2 -0
- data/lib/regexp_parser/error.rb +2 -0
- data/lib/regexp_parser/expression/base.rb +2 -0
- data/lib/regexp_parser/expression/classes/alternation.rb +2 -0
- data/lib/regexp_parser/expression/classes/anchor.rb +2 -0
- data/lib/regexp_parser/expression/classes/backreference.rb +2 -0
- data/lib/regexp_parser/expression/classes/character_set/intersection.rb +2 -0
- data/lib/regexp_parser/expression/classes/character_set/range.rb +2 -0
- data/lib/regexp_parser/expression/classes/character_set.rb +2 -0
- data/lib/regexp_parser/expression/classes/character_type.rb +2 -0
- data/lib/regexp_parser/expression/classes/conditional.rb +2 -0
- data/lib/regexp_parser/expression/classes/escape_sequence.rb +2 -0
- data/lib/regexp_parser/expression/classes/free_space.rb +2 -0
- data/lib/regexp_parser/expression/classes/group.rb +2 -0
- data/lib/regexp_parser/expression/classes/keep.rb +2 -0
- data/lib/regexp_parser/expression/classes/literal.rb +2 -0
- data/lib/regexp_parser/expression/classes/posix_class.rb +2 -0
- data/lib/regexp_parser/expression/classes/root.rb +2 -0
- data/lib/regexp_parser/expression/classes/unicode_property.rb +2 -0
- data/lib/regexp_parser/expression/methods/construct.rb +2 -0
- data/lib/regexp_parser/expression/methods/escape_sequence_char.rb +2 -0
- data/lib/regexp_parser/expression/methods/escape_sequence_codepoint.rb +2 -0
- data/lib/regexp_parser/expression/methods/human_name.rb +2 -0
- data/lib/regexp_parser/expression/methods/match.rb +2 -0
- data/lib/regexp_parser/expression/methods/match_length.rb +2 -0
- data/lib/regexp_parser/expression/methods/negative.rb +2 -0
- data/lib/regexp_parser/expression/methods/options.rb +2 -0
- data/lib/regexp_parser/expression/methods/parts.rb +2 -0
- data/lib/regexp_parser/expression/methods/printing.rb +2 -0
- data/lib/regexp_parser/expression/methods/referenced_expressions.rb +2 -0
- data/lib/regexp_parser/expression/methods/strfregexp.rb +2 -0
- data/lib/regexp_parser/expression/methods/tests.rb +2 -0
- data/lib/regexp_parser/expression/methods/traverse.rb +2 -0
- data/lib/regexp_parser/expression/quantifier.rb +3 -1
- data/lib/regexp_parser/expression/sequence.rb +2 -0
- data/lib/regexp_parser/expression/sequence_operation.rb +2 -0
- data/lib/regexp_parser/expression/shared.rb +2 -0
- data/lib/regexp_parser/expression/subexpression.rb +2 -0
- data/lib/regexp_parser/expression.rb +2 -0
- data/lib/regexp_parser/lexer.rb +2 -0
- data/lib/regexp_parser/parser.rb +2 -0
- data/lib/regexp_parser/scanner/errors/premature_end_error.rb +2 -0
- data/lib/regexp_parser/scanner/errors/scanner_error.rb +2 -0
- data/lib/regexp_parser/scanner/errors/validation_error.rb +2 -0
- data/lib/regexp_parser/scanner/scanner.rl +9 -1
- data/lib/regexp_parser/scanner.rb +10 -2
- data/lib/regexp_parser/syntax/any.rb +2 -0
- data/lib/regexp_parser/syntax/base.rb +2 -0
- data/lib/regexp_parser/syntax/token/anchor.rb +5 -3
- data/lib/regexp_parser/syntax/token/assertion.rb +4 -2
- data/lib/regexp_parser/syntax/token/backreference.rb +8 -6
- data/lib/regexp_parser/syntax/token/character_set.rb +3 -1
- data/lib/regexp_parser/syntax/token/character_type.rb +6 -4
- data/lib/regexp_parser/syntax/token/conditional.rb +5 -3
- data/lib/regexp_parser/syntax/token/escape.rb +9 -7
- data/lib/regexp_parser/syntax/token/group.rb +8 -6
- data/lib/regexp_parser/syntax/token/keep.rb +3 -1
- data/lib/regexp_parser/syntax/token/meta.rb +4 -2
- data/lib/regexp_parser/syntax/token/posix_class.rb +4 -2
- data/lib/regexp_parser/syntax/token/quantifier.rb +8 -6
- data/lib/regexp_parser/syntax/token/unicode_property.rb +51 -49
- data/lib/regexp_parser/syntax/token/virtual.rb +5 -3
- data/lib/regexp_parser/syntax/token.rb +5 -3
- data/lib/regexp_parser/syntax/version_lookup.rb +4 -2
- data/lib/regexp_parser/syntax/versions/1.8.6.rb +2 -0
- data/lib/regexp_parser/syntax/versions/1.9.1.rb +2 -0
- data/lib/regexp_parser/syntax/versions/1.9.3.rb +2 -0
- data/lib/regexp_parser/syntax/versions/2.0.0.rb +2 -0
- data/lib/regexp_parser/syntax/versions/2.2.0.rb +2 -0
- data/lib/regexp_parser/syntax/versions/2.3.0.rb +2 -0
- data/lib/regexp_parser/syntax/versions/2.4.0.rb +2 -0
- data/lib/regexp_parser/syntax/versions/2.4.1.rb +2 -0
- data/lib/regexp_parser/syntax/versions/2.5.0.rb +2 -0
- data/lib/regexp_parser/syntax/versions/2.6.0.rb +2 -0
- data/lib/regexp_parser/syntax/versions/2.6.2.rb +2 -0
- data/lib/regexp_parser/syntax/versions/2.6.3.rb +2 -0
- data/lib/regexp_parser/syntax/versions/3.1.0.rb +2 -0
- data/lib/regexp_parser/syntax/versions/3.2.0.rb +2 -0
- data/lib/regexp_parser/syntax/versions.rb +2 -0
- data/lib/regexp_parser/syntax.rb +2 -0
- data/lib/regexp_parser/token.rb +2 -0
- data/lib/regexp_parser/version.rb +3 -1
- data/lib/regexp_parser.rb +2 -0
- data/regexp_parser.gemspec +2 -0
- metadata +1 -1
@@ -1,3 +1,5 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
1
3
|
module Regexp::Syntax
|
2
4
|
module Token
|
3
5
|
module Quantifier
|
@@ -5,23 +7,23 @@ module Regexp::Syntax
|
|
5
7
|
zero_or_one
|
6
8
|
zero_or_more
|
7
9
|
one_or_more
|
8
|
-
]
|
10
|
+
].freeze
|
9
11
|
|
10
12
|
Reluctant = %i[
|
11
13
|
zero_or_one_reluctant
|
12
14
|
zero_or_more_reluctant
|
13
15
|
one_or_more_reluctant
|
14
|
-
]
|
16
|
+
].freeze
|
15
17
|
|
16
18
|
Possessive = %i[
|
17
19
|
zero_or_one_possessive
|
18
20
|
zero_or_more_possessive
|
19
21
|
one_or_more_possessive
|
20
|
-
]
|
22
|
+
].freeze
|
21
23
|
|
22
|
-
Interval = %i[interval]
|
23
|
-
IntervalReluctant = %i[interval_reluctant]
|
24
|
-
IntervalPossessive = %i[interval_possessive]
|
24
|
+
Interval = %i[interval].freeze
|
25
|
+
IntervalReluctant = %i[interval_reluctant].freeze
|
26
|
+
IntervalPossessive = %i[interval_possessive].freeze
|
25
27
|
|
26
28
|
IntervalAll = Interval + IntervalReluctant + IntervalPossessive
|
27
29
|
|
@@ -1,37 +1,39 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
1
3
|
module Regexp::Syntax
|
2
4
|
module Token
|
3
5
|
module UnicodeProperty
|
4
6
|
all = proc { |name| constants.grep(/#{name}/).flat_map(&method(:const_get)) }
|
5
7
|
|
6
8
|
CharType_V1_9_0 = %i[alnum alpha ascii blank cntrl digit graph
|
7
|
-
lower print punct space upper word xdigit]
|
9
|
+
lower print punct space upper word xdigit].freeze
|
8
10
|
|
9
|
-
CharType_V2_5_0 = %i[xposixpunct]
|
11
|
+
CharType_V2_5_0 = %i[xposixpunct].freeze
|
10
12
|
|
11
|
-
POSIX = %i[any assigned newline]
|
13
|
+
POSIX = %i[any assigned newline].freeze
|
12
14
|
|
13
15
|
module Category
|
14
16
|
Letter = %i[letter uppercase_letter lowercase_letter
|
15
|
-
titlecase_letter modifier_letter other_letter]
|
17
|
+
titlecase_letter modifier_letter other_letter].freeze
|
16
18
|
|
17
19
|
Mark = %i[mark nonspacing_mark spacing_mark
|
18
|
-
enclosing_mark]
|
20
|
+
enclosing_mark].freeze
|
19
21
|
|
20
22
|
Number = %i[number decimal_number letter_number
|
21
|
-
other_number]
|
23
|
+
other_number].freeze
|
22
24
|
|
23
25
|
Punctuation = %i[punctuation connector_punctuation dash_punctuation
|
24
26
|
open_punctuation close_punctuation initial_punctuation
|
25
|
-
final_punctuation other_punctuation]
|
27
|
+
final_punctuation other_punctuation].freeze
|
26
28
|
|
27
29
|
Symbol = %i[symbol math_symbol currency_symbol
|
28
|
-
modifier_symbol other_symbol]
|
30
|
+
modifier_symbol other_symbol].freeze
|
29
31
|
|
30
32
|
Separator = %i[separator space_separator line_separator
|
31
|
-
paragraph_separator]
|
33
|
+
paragraph_separator].freeze
|
32
34
|
|
33
35
|
Codepoint = %i[other control format
|
34
|
-
surrogate private_use unassigned]
|
36
|
+
surrogate private_use unassigned].freeze
|
35
37
|
|
36
38
|
All = Letter + Mark + Number + Punctuation +
|
37
39
|
Symbol + Separator + Codepoint
|
@@ -39,27 +41,27 @@ module Regexp::Syntax
|
|
39
41
|
|
40
42
|
Age_V1_9_3 = %i[age=1.1 age=2.0 age=2.1 age=3.0 age=3.1
|
41
43
|
age=3.2 age=4.0 age=4.1 age=5.0 age=5.1
|
42
|
-
age=5.2 age=6.0]
|
44
|
+
age=5.2 age=6.0].freeze
|
43
45
|
|
44
|
-
Age_V2_0_0 = %i[age=6.1]
|
46
|
+
Age_V2_0_0 = %i[age=6.1].freeze
|
45
47
|
|
46
|
-
Age_V2_2_0 = %i[age=6.2 age=6.3 age=7.0]
|
48
|
+
Age_V2_2_0 = %i[age=6.2 age=6.3 age=7.0].freeze
|
47
49
|
|
48
|
-
Age_V2_3_0 = %i[age=8.0]
|
50
|
+
Age_V2_3_0 = %i[age=8.0].freeze
|
49
51
|
|
50
|
-
Age_V2_4_0 = %i[age=9.0]
|
52
|
+
Age_V2_4_0 = %i[age=9.0].freeze
|
51
53
|
|
52
|
-
Age_V2_5_0 = %i[age=10.0]
|
54
|
+
Age_V2_5_0 = %i[age=10.0].freeze
|
53
55
|
|
54
|
-
Age_V2_6_0 = %i[age=11.0]
|
56
|
+
Age_V2_6_0 = %i[age=11.0].freeze
|
55
57
|
|
56
|
-
Age_V2_6_2 = %i[age=12.0]
|
58
|
+
Age_V2_6_2 = %i[age=12.0].freeze
|
57
59
|
|
58
|
-
Age_V2_6_3 = %i[age=12.1]
|
60
|
+
Age_V2_6_3 = %i[age=12.1].freeze
|
59
61
|
|
60
|
-
Age_V3_1_0 = %i[age=13.0]
|
62
|
+
Age_V3_1_0 = %i[age=13.0].freeze
|
61
63
|
|
62
|
-
Age_V3_2_0 = %i[age=14.0 age=15.0]
|
64
|
+
Age_V3_2_0 = %i[age=14.0 age=15.0].freeze
|
63
65
|
|
64
66
|
Age_V3_5_0 = %i[age=15.1]
|
65
67
|
|
@@ -117,20 +119,20 @@ module Regexp::Syntax
|
|
117
119
|
white_space
|
118
120
|
xid_start
|
119
121
|
xid_continue
|
120
|
-
]
|
122
|
+
].freeze
|
121
123
|
|
122
124
|
Derived_V2_0_0 = %i[
|
123
125
|
cased_letter
|
124
126
|
combining_mark
|
125
|
-
]
|
127
|
+
].freeze
|
126
128
|
|
127
129
|
Derived_V2_4_0 = %i[
|
128
130
|
prepended_concatenation_mark
|
129
|
-
]
|
131
|
+
].freeze
|
130
132
|
|
131
133
|
Derived_V2_5_0 = %i[
|
132
134
|
regional_indicator
|
133
|
-
]
|
135
|
+
].freeze
|
134
136
|
|
135
137
|
Derived_V3_5_0 = %i[
|
136
138
|
id_compat_math_continue
|
@@ -234,13 +236,13 @@ module Regexp::Syntax
|
|
234
236
|
inherited
|
235
237
|
common
|
236
238
|
unknown
|
237
|
-
]
|
239
|
+
].freeze
|
238
240
|
|
239
241
|
Script_V1_9_3 = %i[
|
240
242
|
brahmi
|
241
243
|
batak
|
242
244
|
mandaic
|
243
|
-
]
|
245
|
+
].freeze
|
244
246
|
|
245
247
|
Script_V2_0_0 = %i[
|
246
248
|
chakma
|
@@ -250,7 +252,7 @@ module Regexp::Syntax
|
|
250
252
|
sharada
|
251
253
|
sora_sompeng
|
252
254
|
takri
|
253
|
-
]
|
255
|
+
].freeze
|
254
256
|
|
255
257
|
Script_V2_2_0 = %i[
|
256
258
|
caucasian_albanian
|
@@ -276,7 +278,7 @@ module Regexp::Syntax
|
|
276
278
|
khudawadi
|
277
279
|
tirhuta
|
278
280
|
warang_citi
|
279
|
-
]
|
281
|
+
].freeze
|
280
282
|
|
281
283
|
Script_V2_3_0 = %i[
|
282
284
|
ahom
|
@@ -285,7 +287,7 @@ module Regexp::Syntax
|
|
285
287
|
multani
|
286
288
|
old_hungarian
|
287
289
|
signwriting
|
288
|
-
]
|
290
|
+
].freeze
|
289
291
|
|
290
292
|
Script_V2_4_0 = %i[
|
291
293
|
adlam
|
@@ -294,14 +296,14 @@ module Regexp::Syntax
|
|
294
296
|
newa
|
295
297
|
osage
|
296
298
|
tangut
|
297
|
-
]
|
299
|
+
].freeze
|
298
300
|
|
299
301
|
Script_V2_5_0 = %i[
|
300
302
|
masaram_gondi
|
301
303
|
nushu
|
302
304
|
soyombo
|
303
305
|
zanabazar_square
|
304
|
-
]
|
306
|
+
].freeze
|
305
307
|
|
306
308
|
Script_V2_6_0 = %i[
|
307
309
|
dogra
|
@@ -311,21 +313,21 @@ module Regexp::Syntax
|
|
311
313
|
medefaidrin
|
312
314
|
old_sogdian
|
313
315
|
sogdian
|
314
|
-
]
|
316
|
+
].freeze
|
315
317
|
|
316
318
|
Script_V2_6_2 = %i[
|
317
319
|
elymaic
|
318
320
|
nandinagari
|
319
321
|
nyiakeng_puachue_hmong
|
320
322
|
wancho
|
321
|
-
]
|
323
|
+
].freeze
|
322
324
|
|
323
325
|
Script_V3_1_0 = %i[
|
324
326
|
chorasmian
|
325
327
|
dives_akuru
|
326
328
|
khitan_small_script
|
327
329
|
yezidi
|
328
|
-
]
|
330
|
+
].freeze
|
329
331
|
|
330
332
|
Script_V3_2_0 = %i[
|
331
333
|
cypro_minoan
|
@@ -335,7 +337,7 @@ module Regexp::Syntax
|
|
335
337
|
tangsa
|
336
338
|
toto
|
337
339
|
vithkuqi
|
338
|
-
]
|
340
|
+
].freeze
|
339
341
|
|
340
342
|
Script = all[:Script_V]
|
341
343
|
|
@@ -436,7 +438,7 @@ module Regexp::Syntax
|
|
436
438
|
in_yi_radicals
|
437
439
|
in_yi_syllables
|
438
440
|
in_yijing_hexagram_symbols
|
439
|
-
]
|
441
|
+
].freeze
|
440
442
|
|
441
443
|
UnicodeBlock_V2_0_0 = %i[
|
442
444
|
in_aegean_numbers
|
@@ -564,7 +566,7 @@ module Regexp::Syntax
|
|
564
566
|
in_variation_selectors_supplement
|
565
567
|
in_vedic_extensions
|
566
568
|
in_vertical_forms
|
567
|
-
]
|
569
|
+
].freeze
|
568
570
|
|
569
571
|
UnicodeBlock_V2_2_0 = %i[
|
570
572
|
in_bassa_vah
|
@@ -599,7 +601,7 @@ module Regexp::Syntax
|
|
599
601
|
in_supplemental_arrows_c
|
600
602
|
in_tirhuta
|
601
603
|
in_warang_citi
|
602
|
-
]
|
604
|
+
].freeze
|
603
605
|
|
604
606
|
UnicodeBlock_V2_3_0 = %i[
|
605
607
|
in_ahom
|
@@ -612,7 +614,7 @@ module Regexp::Syntax
|
|
612
614
|
in_old_hungarian
|
613
615
|
in_supplemental_symbols_and_pictographs
|
614
616
|
in_sutton_signwriting
|
615
|
-
]
|
617
|
+
].freeze
|
616
618
|
|
617
619
|
UnicodeBlock_V2_4_0 = %i[
|
618
620
|
in_adlam
|
@@ -626,7 +628,7 @@ module Regexp::Syntax
|
|
626
628
|
in_osage
|
627
629
|
in_tangut
|
628
630
|
in_tangut_components
|
629
|
-
]
|
631
|
+
].freeze
|
630
632
|
|
631
633
|
UnicodeBlock_V2_5_0 = %i[
|
632
634
|
in_cjk_unified_ideographs_extension_f
|
@@ -636,7 +638,7 @@ module Regexp::Syntax
|
|
636
638
|
in_soyombo
|
637
639
|
in_syriac_supplement
|
638
640
|
in_zanabazar_square
|
639
|
-
]
|
641
|
+
].freeze
|
640
642
|
|
641
643
|
UnicodeBlock_V2_6_0 = %i[
|
642
644
|
in_chess_symbols
|
@@ -650,7 +652,7 @@ module Regexp::Syntax
|
|
650
652
|
in_medefaidrin
|
651
653
|
in_old_sogdian
|
652
654
|
in_sogdian
|
653
|
-
]
|
655
|
+
].freeze
|
654
656
|
|
655
657
|
UnicodeBlock_V2_6_2 = %i[
|
656
658
|
in_egyptian_hieroglyph_format_controls
|
@@ -662,7 +664,7 @@ module Regexp::Syntax
|
|
662
664
|
in_symbols_and_pictographs_extended_a
|
663
665
|
in_tamil_supplement
|
664
666
|
in_wancho
|
665
|
-
]
|
667
|
+
].freeze
|
666
668
|
|
667
669
|
UnicodeBlock_V3_1_0 = %i[
|
668
670
|
in_chorasmian
|
@@ -673,7 +675,7 @@ module Regexp::Syntax
|
|
673
675
|
in_symbols_for_legacy_computing
|
674
676
|
in_tangut_supplement
|
675
677
|
in_yezidi
|
676
|
-
]
|
678
|
+
].freeze
|
677
679
|
|
678
680
|
UnicodeBlock_V3_2_0 = %i[
|
679
681
|
in_arabic_extended_b
|
@@ -695,7 +697,7 @@ module Regexp::Syntax
|
|
695
697
|
in_unified_canadian_aboriginal_syllabics_extended_a
|
696
698
|
in_vithkuqi
|
697
699
|
in_znamenny_musical_notation
|
698
|
-
]
|
700
|
+
].freeze
|
699
701
|
|
700
702
|
UnicodeBlock_V3_5_0 = %i[
|
701
703
|
in_cjk_unified_ideographs_extension_i
|
@@ -709,11 +711,11 @@ module Regexp::Syntax
|
|
709
711
|
emoji_modifier
|
710
712
|
emoji_modifier_base
|
711
713
|
emoji_presentation
|
712
|
-
]
|
714
|
+
].freeze
|
713
715
|
|
714
716
|
Emoji_V2_6_0 = %i[
|
715
717
|
extended_pictographic
|
716
|
-
]
|
718
|
+
].freeze
|
717
719
|
|
718
720
|
Enumerated_V2_4_0 = %i[
|
719
721
|
grapheme_cluster_break=control
|
@@ -729,7 +731,7 @@ module Regexp::Syntax
|
|
729
731
|
grapheme_cluster_break=t
|
730
732
|
grapheme_cluster_break=v
|
731
733
|
grapheme_cluster_break=zwj
|
732
|
-
]
|
734
|
+
].freeze
|
733
735
|
|
734
736
|
Enumerated = all[:Enumerated_V]
|
735
737
|
|
@@ -1,10 +1,12 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
1
3
|
module Regexp::Syntax
|
2
4
|
module Token
|
3
5
|
module Virtual
|
4
|
-
Root = %i[root]
|
5
|
-
Sequence = %i[sequence]
|
6
|
+
Root = %i[root].freeze
|
7
|
+
Sequence = %i[sequence].freeze
|
6
8
|
|
7
|
-
All = %i[root sequence]
|
9
|
+
All = %i[root sequence].freeze
|
8
10
|
Type = :expression
|
9
11
|
end
|
10
12
|
end
|
@@ -1,15 +1,17 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
1
3
|
# Define the base module and the simplest of tokens.
|
2
4
|
module Regexp::Syntax
|
3
5
|
module Token
|
4
|
-
Map =
|
6
|
+
Map = Hash.new
|
5
7
|
|
6
8
|
module Literal
|
7
|
-
All = %i[literal]
|
9
|
+
All = %i[literal].freeze
|
8
10
|
Type = :literal
|
9
11
|
end
|
10
12
|
|
11
13
|
module FreeSpace
|
12
|
-
All = %i[comment whitespace]
|
14
|
+
All = %i[comment whitespace].freeze
|
13
15
|
Type = :free_space
|
14
16
|
end
|
15
17
|
|
@@ -1,7 +1,9 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
1
3
|
module Regexp::Syntax
|
2
4
|
VERSION_FORMAT = '\Aruby/\d+\.\d+(\.\d+)?\z'
|
3
|
-
VERSION_REGEXP = /#{VERSION_FORMAT}
|
4
|
-
VERSION_CONST_REGEXP = /\AV\d+_\d+(?:_\d+)?\z
|
5
|
+
VERSION_REGEXP = /#{VERSION_FORMAT}/.freeze
|
6
|
+
VERSION_CONST_REGEXP = /\AV\d+_\d+(?:_\d+)?\z/.freeze
|
5
7
|
|
6
8
|
class InvalidVersionNameError < Regexp::Syntax::SyntaxError
|
7
9
|
def initialize(name)
|
data/lib/regexp_parser/syntax.rb
CHANGED
data/lib/regexp_parser/token.rb
CHANGED
data/lib/regexp_parser.rb
CHANGED
data/regexp_parser.gemspec
CHANGED