regexp_parser 1.7.0 → 2.8.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +364 -22
- data/Gemfile +8 -2
- data/LICENSE +1 -1
- data/README.md +124 -88
- data/Rakefile +6 -70
- data/lib/regexp_parser/error.rb +4 -0
- data/lib/regexp_parser/expression/base.rb +76 -0
- data/lib/regexp_parser/expression/classes/alternation.rb +1 -1
- data/lib/regexp_parser/expression/classes/anchor.rb +0 -2
- data/lib/regexp_parser/expression/classes/{backref.rb → backreference.rb} +22 -2
- data/lib/regexp_parser/expression/classes/{set → character_set}/range.rb +4 -8
- data/lib/regexp_parser/expression/classes/{set.rb → character_set.rb} +3 -4
- data/lib/regexp_parser/expression/classes/{type.rb → character_type.rb} +0 -2
- data/lib/regexp_parser/expression/classes/conditional.rb +11 -5
- data/lib/regexp_parser/expression/classes/{escape.rb → escape_sequence.rb} +15 -7
- data/lib/regexp_parser/expression/classes/free_space.rb +5 -5
- data/lib/regexp_parser/expression/classes/group.rb +28 -15
- 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/posix_class.rb +5 -1
- data/lib/regexp_parser/expression/classes/root.rb +4 -19
- data/lib/regexp_parser/expression/classes/{property.rb → unicode_property.rb} +5 -3
- data/lib/regexp_parser/expression/methods/construct.rb +41 -0
- data/lib/regexp_parser/expression/methods/human_name.rb +43 -0
- data/lib/regexp_parser/expression/methods/match_length.rb +11 -7
- data/lib/regexp_parser/expression/methods/parts.rb +23 -0
- data/lib/regexp_parser/expression/methods/printing.rb +26 -0
- data/lib/regexp_parser/expression/methods/strfregexp.rb +1 -1
- data/lib/regexp_parser/expression/methods/tests.rb +47 -1
- data/lib/regexp_parser/expression/methods/traverse.rb +34 -18
- data/lib/regexp_parser/expression/quantifier.rb +57 -17
- data/lib/regexp_parser/expression/sequence.rb +11 -47
- data/lib/regexp_parser/expression/sequence_operation.rb +4 -9
- data/lib/regexp_parser/expression/shared.rb +111 -0
- data/lib/regexp_parser/expression/subexpression.rb +27 -19
- data/lib/regexp_parser/expression.rb +14 -141
- data/lib/regexp_parser/lexer.rb +83 -41
- data/lib/regexp_parser/parser.rb +371 -429
- data/lib/regexp_parser/scanner/char_type.rl +11 -11
- data/lib/regexp_parser/scanner/errors/premature_end_error.rb +8 -0
- data/lib/regexp_parser/scanner/errors/scanner_error.rb +6 -0
- data/lib/regexp_parser/scanner/errors/validation_error.rb +63 -0
- data/lib/regexp_parser/scanner/properties/long.csv +633 -0
- data/lib/regexp_parser/scanner/properties/short.csv +248 -0
- data/lib/regexp_parser/scanner/property.rl +4 -4
- data/lib/regexp_parser/scanner/scanner.rl +295 -368
- data/lib/regexp_parser/scanner.rb +1405 -1674
- data/lib/regexp_parser/syntax/any.rb +2 -7
- data/lib/regexp_parser/syntax/base.rb +92 -67
- 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 +33 -0
- data/lib/regexp_parser/syntax/token/character_set.rb +16 -0
- 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 +33 -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/token/meta.rb +20 -0
- 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 +733 -0
- data/lib/regexp_parser/syntax/token/virtual.rb +11 -0
- data/lib/regexp_parser/syntax/token.rb +45 -0
- data/lib/regexp_parser/syntax/version_lookup.rb +19 -36
- 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 +3 -1
- data/lib/regexp_parser/syntax.rb +8 -6
- 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 +49 -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/character_set.rb +0 -13
- data/lib/regexp_parser/syntax/tokens/escape.rb +0 -30
- data/lib/regexp_parser/syntax/tokens/meta.rb +0 -13
- 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 -94
- data/spec/expression/clone_spec.rb +0 -120
- 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/root_spec.rb +0 -9
- data/spec/expression/sequence_spec.rb +0 -9
- 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 -100
- data/spec/lexer/all_spec.rb +0 -22
- data/spec/lexer/conditionals_spec.rb +0 -53
- data/spec/lexer/escapes_spec.rb +0 -14
- data/spec/lexer/keep_spec.rb +0 -10
- data/spec/lexer/literals_spec.rb +0 -89
- data/spec/lexer/nesting_spec.rb +0 -99
- data/spec/lexer/refcalls_spec.rb +0 -55
- 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/posix_classes_spec.rb +0 -8
- data/spec/parser/properties_spec.rb +0 -115
- data/spec/parser/quantifiers_spec.rb +0 -51
- data/spec/parser/refcalls_spec.rb +0 -112
- 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/errors_spec.rb +0 -68
- data/spec/scanner/escapes_spec.rb +0 -53
- data/spec/scanner/free_space_spec.rb +0 -133
- data/spec/scanner/groups_spec.rb +0 -52
- data/spec/scanner/keep_spec.rb +0 -10
- data/spec/scanner/literals_spec.rb +0 -49
- data/spec/scanner/meta_spec.rb +0 -18
- data/spec/scanner/properties_spec.rb +0 -64
- data/spec/scanner/quantifiers_spec.rb +0 -20
- data/spec/scanner/refcalls_spec.rb +0 -36
- data/spec/scanner/sets_spec.rb +0 -102
- data/spec/scanner/types_spec.rb +0 -14
- data/spec/spec_helper.rb +0 -15
- 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
- /data/lib/regexp_parser/expression/classes/{set → character_set}/intersection.rb +0 -0
data/spec/parser/types_spec.rb
DELETED
@@ -1,18 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe('CharacterType parsing') do
|
4
|
-
include_examples 'parse', /a\dc/, 1 => [:type, :digit, CharacterType::Digit]
|
5
|
-
include_examples 'parse', /a\Dc/, 1 => [:type, :nondigit, CharacterType::NonDigit]
|
6
|
-
|
7
|
-
include_examples 'parse', /a\sc/, 1 => [:type, :space, CharacterType::Space]
|
8
|
-
include_examples 'parse', /a\Sc/, 1 => [:type, :nonspace, CharacterType::NonSpace]
|
9
|
-
|
10
|
-
include_examples 'parse', /a\hc/, 1 => [:type, :hex, CharacterType::Hex]
|
11
|
-
include_examples 'parse', /a\Hc/, 1 => [:type, :nonhex, CharacterType::NonHex]
|
12
|
-
|
13
|
-
include_examples 'parse', /a\wc/, 1 => [:type, :word, CharacterType::Word]
|
14
|
-
include_examples 'parse', /a\Wc/, 1 => [:type, :nonword, CharacterType::NonWord]
|
15
|
-
|
16
|
-
include_examples 'parse', 'a\\Rc', 1 => [:type, :linebreak, CharacterType::Linebreak]
|
17
|
-
include_examples 'parse', 'a\\Xc', 1 => [:type, :xgrapheme, CharacterType::ExtendedGrapheme]
|
18
|
-
end
|
data/spec/scanner/all_spec.rb
DELETED
@@ -1,18 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe(Regexp::Scanner) do
|
4
|
-
specify('scanner returns an array') do
|
5
|
-
expect(RS.scan('abc')).to be_instance_of(Array)
|
6
|
-
end
|
7
|
-
|
8
|
-
specify('scanner returns tokens as arrays') do
|
9
|
-
tokens = RS.scan('^abc+[^one]{2,3}\\b\\d\\\\C-C$')
|
10
|
-
expect(tokens).to all(be_a Array)
|
11
|
-
expect(tokens.map(&:length)).to all(eq 5)
|
12
|
-
end
|
13
|
-
|
14
|
-
specify('scanner token count') do
|
15
|
-
re = /^(one|two){2,3}([^d\]efm-qz\,\-]*)(ghi)+$/i
|
16
|
-
expect(RS.scan(re).length).to eq 28
|
17
|
-
end
|
18
|
-
end
|
@@ -1,21 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe('Anchor scanning') do
|
4
|
-
include_examples 'scan', '^abc', 0 => [:anchor, :bol, '^', 0, 1]
|
5
|
-
include_examples 'scan', 'abc$', 1 => [:anchor, :eol, '$', 3, 4]
|
6
|
-
|
7
|
-
include_examples 'scan', '\Aabc', 0 => [:anchor, :bos, '\A', 0, 2]
|
8
|
-
include_examples 'scan', 'abc\z', 1 => [:anchor, :eos, '\z', 3, 5]
|
9
|
-
include_examples 'scan', 'abc\Z', 1 => [:anchor, :eos_ob_eol, '\Z', 3, 5]
|
10
|
-
|
11
|
-
include_examples 'scan', 'a\bc', 1 => [:anchor, :word_boundary, '\b', 1, 3]
|
12
|
-
include_examples 'scan', 'a\Bc', 1 => [:anchor, :nonword_boundary, '\B', 1, 3]
|
13
|
-
|
14
|
-
include_examples 'scan', 'a\Gc', 1 => [:anchor, :match_start, '\G', 1, 3]
|
15
|
-
|
16
|
-
include_examples 'scan', "\\\\Ac", 0 => [:escape, :backslash, '\\\\', 0, 2]
|
17
|
-
include_examples 'scan', "a\\\\z", 1 => [:escape, :backslash, '\\\\', 1, 3]
|
18
|
-
include_examples 'scan', "a\\\\Z", 1 => [:escape, :backslash, '\\\\', 1, 3]
|
19
|
-
include_examples 'scan', "a\\\\bc", 1 => [:escape, :backslash, '\\\\', 1, 3]
|
20
|
-
include_examples 'scan', "a\\\\Bc", 1 => [:escape, :backslash, '\\\\', 1, 3]
|
21
|
-
end
|
@@ -1,128 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe('Conditional scanning') do
|
4
|
-
include_examples 'scan', /(a)(?(1)T|F)1/, 3 => [:conditional, :open, '(?', 3, 5]
|
5
|
-
include_examples 'scan', /(a)(?(1)T|F)2/, 4 => [:conditional, :condition_open, '(', 5, 6]
|
6
|
-
include_examples 'scan', /(a)(?(1)T|F)3/, 5 => [:conditional, :condition, '1', 6, 7]
|
7
|
-
include_examples 'scan', /(a)(?(1)T|F)4/, 6 => [:conditional, :condition_close, ')', 7, 8]
|
8
|
-
include_examples 'scan', /(a)(?(1)T|F)5/, 7 => [:literal, :literal, 'T', 8, 9]
|
9
|
-
include_examples 'scan', /(a)(?(1)T|F)6/, 8 => [:conditional, :separator, '|', 9, 10]
|
10
|
-
include_examples 'scan', /(a)(?(1)T|F)7/, 9 => [:literal, :literal, 'F', 10, 11]
|
11
|
-
include_examples 'scan', /(a)(?(1)T|F)8/, 10 => [:conditional, :close, ')', 11, 12]
|
12
|
-
include_examples 'scan', /(a)(?(1)TRUE)9/, 8 => [:conditional, :close, ')', 12, 13]
|
13
|
-
include_examples 'scan', /(a)(?(1)TRUE|)10/, 8 => [:conditional, :separator, '|', 12, 13]
|
14
|
-
include_examples 'scan', /(a)(?(1)TRUE|)11/, 9 => [:conditional, :close, ')', 13, 14]
|
15
|
-
include_examples 'scan', /(?<N>A)(?(<N>)T|F)1/, 5 => [:conditional, :condition, '<N>', 10, 13]
|
16
|
-
include_examples 'scan', /(?'N'A)(?('N')T|F)2/, 5 => [:conditional, :condition, "'N'", 10, 13]
|
17
|
-
|
18
|
-
include_examples 'scan', /(a(b(c)))(?(1)(?(2)d|(?(3)e|f))|(?(2)(?(1)g|h)))/,
|
19
|
-
0 => [:group, :capture, '(', 0, 1],
|
20
|
-
1 => [:literal, :literal, 'a', 1, 2],
|
21
|
-
2 => [:group, :capture, '(', 2, 3],
|
22
|
-
3 => [:literal, :literal, 'b', 3, 4],
|
23
|
-
4 => [:group, :capture, '(', 4, 5],
|
24
|
-
5 => [:literal, :literal, 'c', 5, 6],
|
25
|
-
6 => [:group, :close, ')', 6, 7],
|
26
|
-
7 => [:group, :close, ')', 7, 8],
|
27
|
-
8 => [:group, :close, ')', 8, 9],
|
28
|
-
9 => [:conditional, :open, '(?', 9, 11],
|
29
|
-
10 => [:conditional, :condition_open, '(', 11, 12],
|
30
|
-
11 => [:conditional, :condition, '1', 12, 13],
|
31
|
-
12 => [:conditional, :condition_close, ')', 13, 14],
|
32
|
-
13 => [:conditional, :open, '(?', 14, 16],
|
33
|
-
14 => [:conditional, :condition_open, '(', 16, 17],
|
34
|
-
15 => [:conditional, :condition, '2', 17, 18],
|
35
|
-
16 => [:conditional, :condition_close, ')', 18, 19],
|
36
|
-
17 => [:literal, :literal, 'd', 19, 20],
|
37
|
-
18 => [:conditional, :separator, '|', 20, 21],
|
38
|
-
19 => [:conditional, :open, '(?', 21, 23],
|
39
|
-
20 => [:conditional, :condition_open, '(', 23, 24],
|
40
|
-
21 => [:conditional, :condition, '3', 24, 25],
|
41
|
-
22 => [:conditional, :condition_close, ')', 25, 26],
|
42
|
-
23 => [:literal, :literal, 'e', 26, 27],
|
43
|
-
24 => [:conditional, :separator, '|', 27, 28],
|
44
|
-
25 => [:literal, :literal, 'f', 28, 29],
|
45
|
-
26 => [:conditional, :close, ')', 29, 30],
|
46
|
-
27 => [:conditional, :close, ')', 30, 31],
|
47
|
-
28 => [:conditional, :separator, '|', 31, 32],
|
48
|
-
29 => [:conditional, :open, '(?', 32, 34],
|
49
|
-
30 => [:conditional, :condition_open, '(', 34, 35],
|
50
|
-
31 => [:conditional, :condition, '2', 35, 36],
|
51
|
-
32 => [:conditional, :condition_close, ')', 36, 37],
|
52
|
-
33 => [:conditional, :open, '(?', 37, 39],
|
53
|
-
34 => [:conditional, :condition_open, '(', 39, 40],
|
54
|
-
35 => [:conditional, :condition, '1', 40, 41],
|
55
|
-
36 => [:conditional, :condition_close, ')', 41, 42],
|
56
|
-
37 => [:literal, :literal, 'g', 42, 43],
|
57
|
-
38 => [:conditional, :separator, '|', 43, 44],
|
58
|
-
39 => [:literal, :literal, 'h', 44, 45],
|
59
|
-
40 => [:conditional, :close, ')', 45, 46],
|
60
|
-
41 => [:conditional, :close, ')', 46, 47],
|
61
|
-
42 => [:conditional, :close, ')', 47, 48]
|
62
|
-
|
63
|
-
include_examples 'scan', /((a)|(b)|((?(2)(c(d|e)+)?|(?(3)f|(?(4)(g|(h)(i)))))))/,
|
64
|
-
0 => [:group, :capture, '(', 0, 1],
|
65
|
-
1 => [:group, :capture, '(', 1, 2],
|
66
|
-
2 => [:literal, :literal, 'a', 2, 3],
|
67
|
-
3 => [:group, :close, ')', 3, 4],
|
68
|
-
4 => [:meta, :alternation, '|', 4, 5],
|
69
|
-
5 => [:group, :capture, '(', 5, 6],
|
70
|
-
6 => [:literal, :literal, 'b', 6, 7],
|
71
|
-
7 => [:group, :close, ')', 7, 8],
|
72
|
-
8 => [:meta, :alternation, '|', 8, 9],
|
73
|
-
9 => [:group, :capture, '(', 9, 10],
|
74
|
-
10 => [:conditional, :open, '(?', 10, 12],
|
75
|
-
11 => [:conditional, :condition_open, '(', 12, 13],
|
76
|
-
12 => [:conditional, :condition, '2', 13, 14],
|
77
|
-
13 => [:conditional, :condition_close, ')', 14, 15],
|
78
|
-
14 => [:group, :capture, '(', 15, 16],
|
79
|
-
15 => [:literal, :literal, 'c', 16, 17],
|
80
|
-
16 => [:group, :capture, '(', 17, 18],
|
81
|
-
17 => [:literal, :literal, 'd', 18, 19],
|
82
|
-
18 => [:meta, :alternation, '|', 19, 20],
|
83
|
-
19 => [:literal, :literal, 'e', 20, 21],
|
84
|
-
20 => [:group, :close, ')', 21, 22],
|
85
|
-
21 => [:quantifier, :one_or_more, '+', 22, 23],
|
86
|
-
22 => [:group, :close, ')', 23, 24],
|
87
|
-
23 => [:quantifier, :zero_or_one, '?', 24, 25],
|
88
|
-
24 => [:conditional, :separator, '|', 25, 26],
|
89
|
-
25 => [:conditional, :open, '(?', 26, 28],
|
90
|
-
26 => [:conditional, :condition_open, '(', 28, 29],
|
91
|
-
27 => [:conditional, :condition, '3', 29, 30],
|
92
|
-
28 => [:conditional, :condition_close, ')', 30, 31],
|
93
|
-
29 => [:literal, :literal, 'f', 31, 32],
|
94
|
-
30 => [:conditional, :separator, '|', 32, 33],
|
95
|
-
31 => [:conditional, :open, '(?', 33, 35],
|
96
|
-
32 => [:conditional, :condition_open, '(', 35, 36],
|
97
|
-
33 => [:conditional, :condition, '4', 36, 37],
|
98
|
-
34 => [:conditional, :condition_close, ')', 37, 38],
|
99
|
-
35 => [:group, :capture, '(', 38, 39],
|
100
|
-
36 => [:literal, :literal, 'g', 39, 40],
|
101
|
-
37 => [:meta, :alternation, '|', 40, 41],
|
102
|
-
38 => [:group, :capture, '(', 41, 42],
|
103
|
-
39 => [:literal, :literal, 'h', 42, 43],
|
104
|
-
40 => [:group, :close, ')', 43, 44],
|
105
|
-
41 => [:group, :capture, '(', 44, 45],
|
106
|
-
42 => [:literal, :literal, 'i', 45, 46],
|
107
|
-
43 => [:group, :close, ')', 46, 47],
|
108
|
-
44 => [:group, :close, ')', 47, 48],
|
109
|
-
45 => [:conditional, :close, ')', 48, 49],
|
110
|
-
46 => [:conditional, :close, ')', 49, 50],
|
111
|
-
47 => [:conditional, :close, ')', 50, 51],
|
112
|
-
48 => [:group, :close, ')', 51, 52],
|
113
|
-
49 => [:group, :close, ')', 52, 53]
|
114
|
-
|
115
|
-
include_examples 'scan', /(a)(?(1)(b|c|d)|(e|f|g))(h)(?(2)(i|j|k)|(l|m|n))|o|p/,
|
116
|
-
9 => [:meta, :alternation, '|', 10, 11],
|
117
|
-
11 => [:meta, :alternation, '|', 12, 13],
|
118
|
-
14 => [:conditional, :separator, '|', 15, 16],
|
119
|
-
17 => [:meta, :alternation, '|', 18, 19],
|
120
|
-
19 => [:meta, :alternation, '|', 20, 21],
|
121
|
-
32 => [:meta, :alternation, '|', 34, 35],
|
122
|
-
34 => [:meta, :alternation, '|', 36, 37],
|
123
|
-
37 => [:conditional, :separator, '|', 39, 40],
|
124
|
-
40 => [:meta, :alternation, '|', 42, 43],
|
125
|
-
42 => [:meta, :alternation, '|', 44, 45],
|
126
|
-
46 => [:meta, :alternation, '|', 48, 49],
|
127
|
-
48 => [:meta, :alternation, '|', 50, 51]
|
128
|
-
end
|
data/spec/scanner/errors_spec.rb
DELETED
@@ -1,68 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe(Regexp::Scanner) do
|
4
|
-
RSpec.shared_examples 'scan error' do |error, issue, source|
|
5
|
-
it "raises #{error} for #{issue} `#{source}`" do
|
6
|
-
expect { RS.scan(source) }.to raise_error(error)
|
7
|
-
end
|
8
|
-
end
|
9
|
-
|
10
|
-
include_examples 'scan error', RS::PrematureEndError, 'unbalanced set', '[a'
|
11
|
-
include_examples 'scan error', RS::PrematureEndError, 'unbalanced set', '[[:alpha:]'
|
12
|
-
include_examples 'scan error', RS::PrematureEndError, 'unbalanced group', '(abc'
|
13
|
-
include_examples 'scan error', RS::PrematureEndError, 'unbalanced interval', 'a{1,2'
|
14
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in property', '\p{asci'
|
15
|
-
include_examples 'scan error', RS::PrematureEndError, 'incomplete property', '\p{ascii abc'
|
16
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof options', '(?mix'
|
17
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof escape', '\\'
|
18
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in hex escape', '\x'
|
19
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in cp escape', '\u'
|
20
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in cp escape', '\u0'
|
21
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in cp escape', '\u00'
|
22
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in cp escape', '\u000'
|
23
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in cp escape', '\u{'
|
24
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in cp escape', '\u{00'
|
25
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in cp escape', '\u{0000'
|
26
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in cp escape', '\u{0000 '
|
27
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in cp escape', '\u{0000 0000'
|
28
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in c-seq', '\c'
|
29
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in c-seq', '\c\M'
|
30
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in c-seq', '\c\M-'
|
31
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in c-seq', '\C'
|
32
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in c-seq', '\C-'
|
33
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in c-seq', '\C-\M'
|
34
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in c-seq', '\C-\M-'
|
35
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in m-seq', '\M'
|
36
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in m-seq', '\M-'
|
37
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in m-seq', '\M-\\'
|
38
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in m-seq', '\M-\c'
|
39
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in m-seq', '\M-\C'
|
40
|
-
include_examples 'scan error', RS::PrematureEndError, 'eof in m-seq', '\M-\C-'
|
41
|
-
include_examples 'scan error', RS::InvalidSequenceError, 'invalid hex', '\xZ'
|
42
|
-
include_examples 'scan error', RS::InvalidSequenceError, 'invalid hex', '\xZ0'
|
43
|
-
include_examples 'scan error', RS::InvalidSequenceError, 'invalid c-seq', '\cü'
|
44
|
-
include_examples 'scan error', RS::InvalidSequenceError, 'invalid c-seq', '\c\M-ü'
|
45
|
-
include_examples 'scan error', RS::InvalidSequenceError, 'invalid c-seq', '\C-ü'
|
46
|
-
include_examples 'scan error', RS::InvalidSequenceError, 'invalid c-seq', '\C-\M-ü'
|
47
|
-
include_examples 'scan error', RS::InvalidSequenceError, 'invalid m-seq', '\M-ü'
|
48
|
-
include_examples 'scan error', RS::InvalidSequenceError, 'invalid m-seq', '\M-\cü'
|
49
|
-
include_examples 'scan error', RS::InvalidSequenceError, 'invalid m-seq', '\M-\C-ü'
|
50
|
-
include_examples 'scan error', RS::ScannerError, 'invalid c-seq', '\Ca'
|
51
|
-
include_examples 'scan error', RS::ScannerError, 'invalid m-seq', '\Ma'
|
52
|
-
include_examples 'scan error', RS::InvalidGroupError, 'invalid group', "(?'')"
|
53
|
-
include_examples 'scan error', RS::InvalidGroupError, 'invalid group', "(?''empty-name)"
|
54
|
-
include_examples 'scan error', RS::InvalidGroupError, 'invalid group', '(?<>)'
|
55
|
-
include_examples 'scan error', RS::InvalidGroupError, 'invalid group', '(?<>empty-name)'
|
56
|
-
include_examples 'scan error', RS::InvalidGroupOption, 'invalid option', '(?foo)'
|
57
|
-
include_examples 'scan error', RS::InvalidGroupOption, 'invalid option', '(?mix abc)'
|
58
|
-
include_examples 'scan error', RS::InvalidGroupOption, 'invalid option', '(?mix^bc'
|
59
|
-
include_examples 'scan error', RS::InvalidGroupOption, 'invalid option', '(?)'
|
60
|
-
include_examples 'scan error', RS::InvalidGroupOption, 'invalid neg option', '(?-foo)'
|
61
|
-
include_examples 'scan error', RS::InvalidGroupOption, 'invalid neg option', '(?-u)'
|
62
|
-
include_examples 'scan error', RS::InvalidGroupOption, 'invalid neg option', '(?-mixu)'
|
63
|
-
include_examples 'scan error', RS::InvalidBackrefError, 'empty backref', '\k<>'
|
64
|
-
include_examples 'scan error', RS::InvalidBackrefError, 'empty backref', '\k\'\''
|
65
|
-
include_examples 'scan error', RS::InvalidBackrefError, 'empty refcall', '\g<>'
|
66
|
-
include_examples 'scan error', RS::InvalidBackrefError, 'empty refcall', '\g\'\''
|
67
|
-
include_examples 'scan error', RS::UnknownUnicodePropertyError, 'unknown property', '\p{foobar}'
|
68
|
-
end
|
@@ -1,53 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe('Escape scanning') do
|
4
|
-
include_examples 'scan', /c\at/, 1 => [:escape, :bell, '\a', 1, 3]
|
5
|
-
|
6
|
-
# not an escape outside a character set
|
7
|
-
include_examples 'scan', /c\bt/, 1 => [:anchor, :word_boundary, '\b', 1, 3]
|
8
|
-
|
9
|
-
include_examples 'scan', /c\ft/, 1 => [:escape, :form_feed, '\f', 1, 3]
|
10
|
-
include_examples 'scan', /c\nt/, 1 => [:escape, :newline, '\n', 1, 3]
|
11
|
-
include_examples 'scan', /c\tt/, 1 => [:escape, :tab, '\t', 1, 3]
|
12
|
-
include_examples 'scan', /c\vt/, 1 => [:escape, :vertical_tab, '\v', 1, 3]
|
13
|
-
|
14
|
-
include_examples 'scan', 'c\qt', 1 => [:escape, :literal, '\q', 1, 3]
|
15
|
-
|
16
|
-
include_examples 'scan', 'a\012c', 1 => [:escape, :octal, '\012', 1, 5]
|
17
|
-
include_examples 'scan', 'a\0124', 1 => [:escape, :octal, '\012', 1, 5]
|
18
|
-
include_examples 'scan', '\712+7', 0 => [:escape, :octal, '\712', 0, 4]
|
19
|
-
|
20
|
-
include_examples 'scan', 'a\x24c', 1 => [:escape, :hex, '\x24', 1, 5]
|
21
|
-
include_examples 'scan', 'a\x0640c', 1 => [:escape, :hex, '\x06', 1, 5]
|
22
|
-
|
23
|
-
include_examples 'scan', 'a\u0640c', 1 => [:escape, :codepoint, '\u0640', 1, 7]
|
24
|
-
include_examples 'scan', 'a\u{640 0641}c', 1 => [:escape, :codepoint_list, '\u{640 0641}', 1, 13]
|
25
|
-
include_examples 'scan', 'a\u{10FFFF}c', 1 => [:escape, :codepoint_list, '\u{10FFFF}', 1, 11]
|
26
|
-
|
27
|
-
include_examples 'scan', /a\cBc/, 1 => [:escape, :control, '\cB', 1, 4]
|
28
|
-
include_examples 'scan', /a\c^c/, 1 => [:escape, :control, '\c^', 1, 4]
|
29
|
-
include_examples 'scan', /a\c\n/, 1 => [:escape, :control, '\c\n', 1, 5]
|
30
|
-
include_examples 'scan', /a\c\\b/, 1 => [:escape, :control, '\c\\\\', 1, 5]
|
31
|
-
include_examples 'scan', /a\C-bc/, 1 => [:escape, :control, '\C-b', 1, 5]
|
32
|
-
include_examples 'scan', /a\C-^b/, 1 => [:escape, :control, '\C-^', 1, 5]
|
33
|
-
include_examples 'scan', /a\C-\nb/, 1 => [:escape, :control, '\C-\n', 1, 6]
|
34
|
-
include_examples 'scan', /a\C-\\b/, 1 => [:escape, :control, '\C-\\\\', 1, 6]
|
35
|
-
include_examples 'scan', /a\c\M-Bc/n, 1 => [:escape, :control, '\c\M-B', 1, 7]
|
36
|
-
include_examples 'scan', /a\C-\M-Bc/n, 1 => [:escape, :control, '\C-\M-B', 1, 8]
|
37
|
-
|
38
|
-
include_examples 'scan', /a\M-Bc/n, 1 => [:escape, :meta_sequence, '\M-B', 1, 5]
|
39
|
-
include_examples 'scan', /a\M-\cBc/n, 1 => [:escape, :meta_sequence, '\M-\cB', 1, 7]
|
40
|
-
include_examples 'scan', /a\M-\c^/n, 1 => [:escape, :meta_sequence, '\M-\c^', 1, 7]
|
41
|
-
include_examples 'scan', /a\M-\c\n/n, 1 => [:escape, :meta_sequence, '\M-\c\n', 1, 8]
|
42
|
-
include_examples 'scan', /a\M-\c\\/n, 1 => [:escape, :meta_sequence, '\M-\c\\\\', 1, 8]
|
43
|
-
include_examples 'scan', /a\M-\C-Bc/n, 1 => [:escape, :meta_sequence, '\M-\C-B', 1, 8]
|
44
|
-
include_examples 'scan', /a\M-\C-\\/n, 1 => [:escape, :meta_sequence, '\M-\C-\\\\', 1, 9]
|
45
|
-
|
46
|
-
include_examples 'scan', 'ab\\\xcd', 1 => [:escape, :backslash, '\\\\', 2, 4]
|
47
|
-
include_examples 'scan', 'ab\\\0cd', 1 => [:escape, :backslash, '\\\\', 2, 4]
|
48
|
-
include_examples 'scan', 'ab\\\Kcd', 1 => [:escape, :backslash, '\\\\', 2, 4]
|
49
|
-
|
50
|
-
include_examples 'scan', 'ab\^cd', 1 => [:escape, :bol, '\^', 2, 4]
|
51
|
-
include_examples 'scan', 'ab\$cd', 1 => [:escape, :eol, '\$', 2, 4]
|
52
|
-
include_examples 'scan', 'ab\[cd', 1 => [:escape, :set_open, '\[', 2, 4]
|
53
|
-
end
|
@@ -1,133 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe('FreeSpace scanning') do
|
4
|
-
describe('scan free space tokens') do
|
5
|
-
let(:tokens) { RS.scan(/
|
6
|
-
a
|
7
|
-
b ? c *
|
8
|
-
d {2,3}
|
9
|
-
e + | f +
|
10
|
-
/x) }
|
11
|
-
|
12
|
-
0.upto(24).select(&:even?).each do |i|
|
13
|
-
it "scans #{i} as free space" do
|
14
|
-
expect(tokens[i][0]).to eq :free_space
|
15
|
-
expect(tokens[i][1]).to eq :whitespace
|
16
|
-
end
|
17
|
-
end
|
18
|
-
0.upto(24).reject(&:even?).each do |i|
|
19
|
-
it "does not scan #{i} as free space" do
|
20
|
-
expect(tokens[i][0]).not_to eq :free_space
|
21
|
-
expect(tokens[i][1]).not_to eq :whitespace
|
22
|
-
end
|
23
|
-
end
|
24
|
-
|
25
|
-
it 'sets the correct text' do
|
26
|
-
[0, 2, 10, 14].each { |i| expect(tokens[i][2]).to eq "\n " }
|
27
|
-
[4, 6, 8, 12].each { |i| expect(tokens[i][2]).to eq ' ' }
|
28
|
-
end
|
29
|
-
end
|
30
|
-
|
31
|
-
describe('scan free space comments') do
|
32
|
-
include_examples 'scan', /
|
33
|
-
a + # A + comment
|
34
|
-
b ? # B ? comment
|
35
|
-
c {2,3} # C {2,3} comment
|
36
|
-
d + | e + # D|E comment
|
37
|
-
/x,
|
38
|
-
5 => [:free_space, :comment, "# A + comment\n", 11, 25],
|
39
|
-
11 => [:free_space, :comment, "# B ? comment\n", 37, 51],
|
40
|
-
17 => [:free_space, :comment, "# C {2,3} comment\n", 66, 84],
|
41
|
-
29 => [:free_space, :comment, "# D|E comment\n", 100, 114]
|
42
|
-
end
|
43
|
-
|
44
|
-
describe('scan free space inlined') do
|
45
|
-
include_examples 'scan', /a b(?x:c d e)f g/,
|
46
|
-
0 => [:literal, :literal, 'a b', 0, 3],
|
47
|
-
1 => [:group, :options, '(?x:', 3, 7],
|
48
|
-
2 => [:literal, :literal, 'c', 7, 8],
|
49
|
-
3 => [:free_space, :whitespace, ' ', 8, 9],
|
50
|
-
4 => [:literal, :literal, 'd', 9, 10],
|
51
|
-
5 => [:free_space, :whitespace, ' ', 10, 11],
|
52
|
-
6 => [:literal, :literal, 'e', 11, 12],
|
53
|
-
7 => [:group, :close, ')', 12, 13],
|
54
|
-
8 => [:literal, :literal, 'f g', 13, 16]
|
55
|
-
end
|
56
|
-
|
57
|
-
describe('scan free space nested') do
|
58
|
-
include_examples 'scan', /a b(?x:c d(?-x:e f)g h)i j/,
|
59
|
-
0 => [:literal, :literal, 'a b', 0, 3],
|
60
|
-
1 => [:group, :options, '(?x:', 3, 7],
|
61
|
-
2 => [:literal, :literal, 'c', 7, 8],
|
62
|
-
3 => [:free_space, :whitespace, ' ', 8, 9],
|
63
|
-
4 => [:literal, :literal, 'd', 9, 10],
|
64
|
-
5 => [:group, :options, '(?-x:', 10, 15],
|
65
|
-
6 => [:literal, :literal, 'e f', 15, 18],
|
66
|
-
7 => [:group, :close, ')', 18, 19],
|
67
|
-
8 => [:literal, :literal, 'g', 19, 20],
|
68
|
-
9 => [:free_space, :whitespace, ' ', 20, 21],
|
69
|
-
10 => [:literal, :literal, 'h', 21, 22],
|
70
|
-
11 => [:group, :close, ')', 22, 23],
|
71
|
-
12 => [:literal, :literal, 'i j', 23, 26]
|
72
|
-
end
|
73
|
-
|
74
|
-
describe('scan free space nested groups') do
|
75
|
-
include_examples 'scan', /(a (b(?x: (c d) (?-x:(e f) )g) h)i j)/,
|
76
|
-
0 => [:group, :capture, '(', 0, 1],
|
77
|
-
1 => [:literal, :literal, 'a ', 1, 3],
|
78
|
-
2 => [:group, :capture, '(', 3, 4],
|
79
|
-
3 => [:literal, :literal, 'b', 4, 5],
|
80
|
-
4 => [:group, :options, '(?x:', 5, 9],
|
81
|
-
5 => [:free_space, :whitespace, ' ', 9, 10],
|
82
|
-
6 => [:group, :capture, '(', 10, 11],
|
83
|
-
7 => [:literal, :literal, 'c', 11, 12],
|
84
|
-
8 => [:free_space, :whitespace, ' ', 12, 13],
|
85
|
-
9 => [:literal, :literal, 'd', 13, 14],
|
86
|
-
10 => [:group, :close, ')', 14, 15],
|
87
|
-
11 => [:free_space, :whitespace, ' ', 15, 16],
|
88
|
-
12 => [:group, :options, '(?-x:', 16, 21],
|
89
|
-
13 => [:group, :capture, '(', 21, 22],
|
90
|
-
14 => [:literal, :literal, 'e f', 22, 25],
|
91
|
-
15 => [:group, :close, ')', 25, 26],
|
92
|
-
16 => [:literal, :literal, ' ', 26, 27],
|
93
|
-
17 => [:group, :close, ')', 27, 28],
|
94
|
-
18 => [:literal, :literal, 'g', 28, 29],
|
95
|
-
19 => [:group, :close, ')', 29, 30],
|
96
|
-
20 => [:literal, :literal, ' h', 30, 32],
|
97
|
-
21 => [:group, :close, ')', 32, 33],
|
98
|
-
22 => [:literal, :literal, 'i j', 33, 36],
|
99
|
-
23 => [:group, :close, ')', 36, 37]
|
100
|
-
end
|
101
|
-
|
102
|
-
describe('scan free space switch groups') do
|
103
|
-
include_examples 'scan', /(a (b((?x) (c d) ((?-x)(e f) )g) h)i j)/,
|
104
|
-
0 => [:group, :capture, '(', 0, 1],
|
105
|
-
1 => [:literal, :literal, 'a ', 1, 3],
|
106
|
-
2 => [:group, :capture, '(', 3, 4],
|
107
|
-
3 => [:literal, :literal, 'b', 4, 5],
|
108
|
-
4 => [:group, :capture, '(', 5, 6],
|
109
|
-
5 => [:group, :options_switch, '(?x', 6, 9],
|
110
|
-
6 => [:group, :close, ')', 9, 10],
|
111
|
-
7 => [:free_space, :whitespace, ' ', 10, 11],
|
112
|
-
8 => [:group, :capture, '(', 11, 12],
|
113
|
-
9 => [:literal, :literal, 'c', 12, 13],
|
114
|
-
10 => [:free_space, :whitespace, ' ', 13, 14],
|
115
|
-
11 => [:literal, :literal, 'd', 14, 15],
|
116
|
-
12 => [:group, :close, ')', 15, 16],
|
117
|
-
13 => [:free_space, :whitespace, ' ', 16, 17],
|
118
|
-
14 => [:group, :capture, '(', 17, 18],
|
119
|
-
15 => [:group, :options_switch, '(?-x', 18, 22],
|
120
|
-
16 => [:group, :close, ')', 22, 23],
|
121
|
-
17 => [:group, :capture, '(', 23, 24],
|
122
|
-
18 => [:literal, :literal, 'e f', 24, 27],
|
123
|
-
19 => [:group, :close, ')', 27, 28],
|
124
|
-
20 => [:literal, :literal, ' ', 28, 29],
|
125
|
-
21 => [:group, :close, ')', 29, 30],
|
126
|
-
22 => [:literal, :literal, 'g', 30, 31],
|
127
|
-
23 => [:group, :close, ')', 31, 32],
|
128
|
-
24 => [:literal, :literal, ' h', 32, 34],
|
129
|
-
25 => [:group, :close, ')', 34, 35],
|
130
|
-
26 => [:literal, :literal, 'i j', 35, 38],
|
131
|
-
27 => [:group, :close, ')', 38, 39]
|
132
|
-
end
|
133
|
-
end
|
data/spec/scanner/groups_spec.rb
DELETED
@@ -1,52 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe('Group scanning') do
|
4
|
-
# Group types
|
5
|
-
include_examples 'scan', '(?>abc)', 0 => [:group, :atomic, '(?>', 0, 3]
|
6
|
-
include_examples 'scan', '(abc)', 0 => [:group, :capture, '(', 0, 1]
|
7
|
-
|
8
|
-
include_examples 'scan', '(?<name>abc)', 0 => [:group, :named_ab, '(?<name>', 0, 8]
|
9
|
-
include_examples 'scan', "(?'name'abc)", 0 => [:group, :named_sq, "(?'name'", 0, 8]
|
10
|
-
|
11
|
-
include_examples 'scan', '(?<name_1>abc)', 0 => [:group, :named_ab, '(?<name_1>', 0,10]
|
12
|
-
include_examples 'scan', "(?'name_1'abc)", 0 => [:group, :named_sq, "(?'name_1'", 0,10]
|
13
|
-
|
14
|
-
include_examples 'scan', '(?:abc)', 0 => [:group, :passive, '(?:', 0, 3]
|
15
|
-
include_examples 'scan', '(?:)', 0 => [:group, :passive, '(?:', 0, 3]
|
16
|
-
include_examples 'scan', '(?::)', 0 => [:group, :passive, '(?:', 0, 3]
|
17
|
-
|
18
|
-
# Comments
|
19
|
-
include_examples 'scan', '(?#abc)', 0 => [:group, :comment, '(?#abc)', 0, 7]
|
20
|
-
include_examples 'scan', '(?#)', 0 => [:group, :comment, '(?#)', 0, 4]
|
21
|
-
|
22
|
-
# Assertions
|
23
|
-
include_examples 'scan', '(?=abc)', 0 => [:assertion, :lookahead, '(?=', 0, 3]
|
24
|
-
include_examples 'scan', '(?!abc)', 0 => [:assertion, :nlookahead, '(?!', 0, 3]
|
25
|
-
include_examples 'scan', '(?<=abc)', 0 => [:assertion, :lookbehind, '(?<=', 0, 4]
|
26
|
-
include_examples 'scan', '(?<!abc)', 0 => [:assertion, :nlookbehind, '(?<!', 0, 4]
|
27
|
-
|
28
|
-
# Options
|
29
|
-
include_examples 'scan', '(?-mix:abc)', 0 => [:group, :options, '(?-mix:', 0, 7]
|
30
|
-
include_examples 'scan', '(?m-ix:abc)', 0 => [:group, :options, '(?m-ix:', 0, 7]
|
31
|
-
include_examples 'scan', '(?mi-x:abc)', 0 => [:group, :options, '(?mi-x:', 0, 7]
|
32
|
-
include_examples 'scan', '(?mix:abc)', 0 => [:group, :options, '(?mix:', 0, 6]
|
33
|
-
include_examples 'scan', '(?m:)', 0 => [:group, :options, '(?m:', 0, 4]
|
34
|
-
include_examples 'scan', '(?i:)', 0 => [:group, :options, '(?i:', 0, 4]
|
35
|
-
include_examples 'scan', '(?x:)', 0 => [:group, :options, '(?x:', 0, 4]
|
36
|
-
include_examples 'scan', '(?mix)', 0 => [:group, :options_switch, '(?mix', 0, 5]
|
37
|
-
include_examples 'scan', '(?d-mix:abc)', 0 => [:group, :options, '(?d-mix:', 0, 8]
|
38
|
-
include_examples 'scan', '(?a-mix:abc)', 0 => [:group, :options, '(?a-mix:', 0, 8]
|
39
|
-
include_examples 'scan', '(?u-mix:abc)', 0 => [:group, :options, '(?u-mix:', 0, 8]
|
40
|
-
include_examples 'scan', '(?da-m:abc)', 0 => [:group, :options, '(?da-m:', 0, 7]
|
41
|
-
include_examples 'scan', '(?du-x:abc)', 0 => [:group, :options, '(?du-x:', 0, 7]
|
42
|
-
include_examples 'scan', '(?dau-i:abc)', 0 => [:group, :options, '(?dau-i:', 0, 8]
|
43
|
-
include_examples 'scan', '(?dau:abc)', 0 => [:group, :options, '(?dau:', 0, 6]
|
44
|
-
include_examples 'scan', '(?d:)', 0 => [:group, :options, '(?d:', 0, 4]
|
45
|
-
include_examples 'scan', '(?a:)', 0 => [:group, :options, '(?a:', 0, 4]
|
46
|
-
include_examples 'scan', '(?u:)', 0 => [:group, :options, '(?u:', 0, 4]
|
47
|
-
include_examples 'scan', '(?dau)', 0 => [:group, :options_switch, '(?dau', 0, 5]
|
48
|
-
|
49
|
-
if ruby_version_at_least('2.4.1')
|
50
|
-
include_examples 'scan', '(?~abc)', 0 => [:group, :absence, '(?~', 0, 3]
|
51
|
-
end
|
52
|
-
end
|
data/spec/scanner/keep_spec.rb
DELETED
@@ -1,49 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe('UTF8 scanning') do
|
4
|
-
# ascii, single byte characters
|
5
|
-
include_examples 'scan', 'a', 0 => [:literal, :literal, 'a', 0, 1]
|
6
|
-
|
7
|
-
include_examples 'scan', 'ab+', 0 => [:literal, :literal, 'ab', 0, 2]
|
8
|
-
include_examples 'scan', 'ab+', 1 => [:quantifier, :one_or_more, '+', 2, 3]
|
9
|
-
|
10
|
-
# 2 byte wide characters, Arabic
|
11
|
-
include_examples 'scan', 'aاbبcت', 0 => [:literal, :literal, 'aاbبcت', 0, 9]
|
12
|
-
|
13
|
-
include_examples 'scan', 'aاbبت?', 0 => [:literal, :literal, 'aاbبت', 0, 8]
|
14
|
-
include_examples 'scan', 'aاbبت?', 1 => [:quantifier, :zero_or_one, '?', 8, 9]
|
15
|
-
|
16
|
-
include_examples 'scan', 'aا?bبcت+', 0 => [:literal, :literal, 'aا', 0, 3]
|
17
|
-
include_examples 'scan', 'aا?bبcت+', 1 => [:quantifier, :zero_or_one, '?', 3, 4]
|
18
|
-
include_examples 'scan', 'aا?bبcت+', 2 => [:literal, :literal, 'bبcت', 4, 10]
|
19
|
-
include_examples 'scan', 'aا?bبcت+', 3 => [:quantifier, :one_or_more, '+', 10, 11]
|
20
|
-
|
21
|
-
include_examples 'scan', 'a(اbب+)cت?', 0 => [:literal, :literal, 'a', 0, 1]
|
22
|
-
include_examples 'scan', 'a(اbب+)cت?', 1 => [:group, :capture, '(', 1, 2]
|
23
|
-
include_examples 'scan', 'a(اbب+)cت?', 2 => [:literal, :literal, 'اbب', 2, 7]
|
24
|
-
include_examples 'scan', 'a(اbب+)cت?', 3 => [:quantifier, :one_or_more, '+', 7, 8]
|
25
|
-
include_examples 'scan', 'a(اbب+)cت?', 4 => [:group, :close, ')', 8, 9]
|
26
|
-
include_examples 'scan', 'a(اbب+)cت?', 5 => [:literal, :literal, 'cت', 9, 12]
|
27
|
-
include_examples 'scan', 'a(اbب+)cت?', 6 => [:quantifier, :zero_or_one, '?', 12, 13]
|
28
|
-
|
29
|
-
# 3 byte wide characters, Japanese
|
30
|
-
include_examples 'scan', 'ab?れます+cd', 0 => [:literal, :literal, 'ab', 0, 2]
|
31
|
-
include_examples 'scan', 'ab?れます+cd', 1 => [:quantifier, :zero_or_one, '?', 2, 3]
|
32
|
-
include_examples 'scan', 'ab?れます+cd', 2 => [:literal, :literal, 'れます', 3, 12]
|
33
|
-
include_examples 'scan', 'ab?れます+cd', 3 => [:quantifier, :one_or_more, '+', 12, 13]
|
34
|
-
include_examples 'scan', 'ab?れます+cd', 4 => [:literal, :literal, 'cd', 13, 15]
|
35
|
-
|
36
|
-
# 4 byte wide characters, Osmanya
|
37
|
-
include_examples 'scan', '𐒀𐒁?𐒂ab+𐒃', 0 => [:literal, :literal, '𐒀𐒁', 0, 8]
|
38
|
-
include_examples 'scan', '𐒀𐒁?𐒂ab+𐒃', 1 => [:quantifier, :zero_or_one, '?', 8, 9]
|
39
|
-
include_examples 'scan', '𐒀𐒁?𐒂ab+𐒃', 2 => [:literal, :literal, '𐒂ab', 9, 15]
|
40
|
-
include_examples 'scan', '𐒀𐒁?𐒂ab+𐒃', 3 => [:quantifier, :one_or_more, '+', 15, 16]
|
41
|
-
include_examples 'scan', '𐒀𐒁?𐒂ab+𐒃', 4 => [:literal, :literal, '𐒃', 16, 20]
|
42
|
-
|
43
|
-
include_examples 'scan', 'mu𝄞?si*𝄫c+', 0 => [:literal, :literal, 'mu𝄞', 0, 6]
|
44
|
-
include_examples 'scan', 'mu𝄞?si*𝄫c+', 1 => [:quantifier, :zero_or_one, '?', 6, 7]
|
45
|
-
include_examples 'scan', 'mu𝄞?si*𝄫c+', 2 => [:literal, :literal, 'si', 7, 9]
|
46
|
-
include_examples 'scan', 'mu𝄞?si*𝄫c+', 3 => [:quantifier, :zero_or_more, '*', 9, 10]
|
47
|
-
include_examples 'scan', 'mu𝄞?si*𝄫c+', 4 => [:literal, :literal, '𝄫c', 10, 15]
|
48
|
-
include_examples 'scan', 'mu𝄞?si*𝄫c+', 5 => [:quantifier, :one_or_more, '+', 15, 16]
|
49
|
-
end
|
data/spec/scanner/meta_spec.rb
DELETED
@@ -1,18 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe('Meta scanning') do
|
4
|
-
include_examples 'scan', /abc??|def*+|ghi+/,
|
5
|
-
0 => [:literal, :literal, 'abc', 0, 3],
|
6
|
-
1 => [:quantifier, :zero_or_one_reluctant, '??', 3, 5],
|
7
|
-
2 => [:meta, :alternation, '|', 5, 6],
|
8
|
-
3 => [:literal, :literal, 'def', 6, 9],
|
9
|
-
4 => [:quantifier, :zero_or_more_possessive, '*+', 9, 11],
|
10
|
-
5 => [:meta, :alternation, '|', 11, 12]
|
11
|
-
|
12
|
-
include_examples 'scan', /(a\|b)|(c|d)\|(e[|]f)/,
|
13
|
-
2 => [:escape, :alternation, '\|', 2, 4],
|
14
|
-
5 => [:meta, :alternation, '|', 6, 7],
|
15
|
-
8 => [:meta, :alternation, '|', 9, 10],
|
16
|
-
11 => [:escape, :alternation, '\|', 12, 14],
|
17
|
-
15 => [:literal, :literal, '|', 17, 18]
|
18
|
-
end
|