regexp_parser 1.3.0 → 1.7.0

Sign up to get free protection for your applications and to get access to all the features.
Files changed (170) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +65 -1
  3. data/Gemfile +3 -3
  4. data/README.md +10 -14
  5. data/Rakefile +3 -4
  6. data/lib/regexp_parser/expression.rb +28 -53
  7. data/lib/regexp_parser/expression/classes/backref.rb +18 -10
  8. data/lib/regexp_parser/expression/classes/conditional.rb +7 -2
  9. data/lib/regexp_parser/expression/classes/escape.rb +0 -4
  10. data/lib/regexp_parser/expression/classes/group.rb +4 -2
  11. data/lib/regexp_parser/expression/classes/keep.rb +1 -3
  12. data/lib/regexp_parser/expression/methods/match.rb +13 -0
  13. data/lib/regexp_parser/expression/methods/match_length.rb +172 -0
  14. data/lib/regexp_parser/expression/methods/options.rb +35 -0
  15. data/lib/regexp_parser/expression/methods/strfregexp.rb +0 -1
  16. data/lib/regexp_parser/expression/methods/tests.rb +6 -15
  17. data/lib/regexp_parser/expression/methods/traverse.rb +3 -1
  18. data/lib/regexp_parser/expression/quantifier.rb +2 -2
  19. data/lib/regexp_parser/expression/sequence.rb +3 -6
  20. data/lib/regexp_parser/expression/sequence_operation.rb +2 -6
  21. data/lib/regexp_parser/expression/subexpression.rb +3 -5
  22. data/lib/regexp_parser/lexer.rb +30 -44
  23. data/lib/regexp_parser/parser.rb +47 -24
  24. data/lib/regexp_parser/scanner.rb +1159 -1329
  25. data/lib/regexp_parser/scanner/char_type.rl +0 -3
  26. data/lib/regexp_parser/scanner/properties/long.yml +34 -1
  27. data/lib/regexp_parser/scanner/properties/short.yml +12 -0
  28. data/lib/regexp_parser/scanner/scanner.rl +82 -190
  29. data/lib/regexp_parser/syntax/tokens.rb +2 -10
  30. data/lib/regexp_parser/syntax/tokens/unicode_property.rb +72 -21
  31. data/lib/regexp_parser/syntax/versions/2.6.0.rb +10 -0
  32. data/lib/regexp_parser/syntax/versions/2.6.2.rb +10 -0
  33. data/lib/regexp_parser/syntax/versions/2.6.3.rb +10 -0
  34. data/lib/regexp_parser/version.rb +1 -1
  35. data/regexp_parser.gemspec +3 -3
  36. data/spec/expression/base_spec.rb +94 -0
  37. data/spec/expression/clone_spec.rb +120 -0
  38. data/spec/expression/conditional_spec.rb +89 -0
  39. data/spec/expression/free_space_spec.rb +27 -0
  40. data/spec/expression/methods/match_length_spec.rb +161 -0
  41. data/spec/expression/methods/match_spec.rb +25 -0
  42. data/spec/expression/methods/strfregexp_spec.rb +224 -0
  43. data/spec/expression/methods/tests_spec.rb +99 -0
  44. data/spec/expression/methods/traverse_spec.rb +161 -0
  45. data/spec/expression/options_spec.rb +128 -0
  46. data/spec/expression/root_spec.rb +9 -0
  47. data/spec/expression/sequence_spec.rb +9 -0
  48. data/spec/expression/subexpression_spec.rb +50 -0
  49. data/spec/expression/to_h_spec.rb +26 -0
  50. data/spec/expression/to_s_spec.rb +100 -0
  51. data/spec/lexer/all_spec.rb +22 -0
  52. data/spec/lexer/conditionals_spec.rb +53 -0
  53. data/spec/lexer/escapes_spec.rb +14 -0
  54. data/spec/lexer/keep_spec.rb +10 -0
  55. data/spec/lexer/literals_spec.rb +89 -0
  56. data/spec/lexer/nesting_spec.rb +99 -0
  57. data/spec/lexer/refcalls_spec.rb +55 -0
  58. data/spec/parser/all_spec.rb +43 -0
  59. data/spec/parser/alternation_spec.rb +88 -0
  60. data/spec/parser/anchors_spec.rb +17 -0
  61. data/spec/parser/conditionals_spec.rb +179 -0
  62. data/spec/parser/errors_spec.rb +30 -0
  63. data/spec/parser/escapes_spec.rb +121 -0
  64. data/spec/parser/free_space_spec.rb +130 -0
  65. data/spec/parser/groups_spec.rb +108 -0
  66. data/spec/parser/keep_spec.rb +6 -0
  67. data/spec/parser/posix_classes_spec.rb +8 -0
  68. data/spec/parser/properties_spec.rb +115 -0
  69. data/spec/parser/quantifiers_spec.rb +51 -0
  70. data/spec/parser/refcalls_spec.rb +112 -0
  71. data/spec/parser/set/intersections_spec.rb +127 -0
  72. data/spec/parser/set/ranges_spec.rb +111 -0
  73. data/spec/parser/sets_spec.rb +178 -0
  74. data/spec/parser/types_spec.rb +18 -0
  75. data/spec/scanner/all_spec.rb +18 -0
  76. data/spec/scanner/anchors_spec.rb +21 -0
  77. data/spec/scanner/conditionals_spec.rb +128 -0
  78. data/spec/scanner/errors_spec.rb +68 -0
  79. data/spec/scanner/escapes_spec.rb +53 -0
  80. data/spec/scanner/free_space_spec.rb +133 -0
  81. data/spec/scanner/groups_spec.rb +52 -0
  82. data/spec/scanner/keep_spec.rb +10 -0
  83. data/spec/scanner/literals_spec.rb +49 -0
  84. data/spec/scanner/meta_spec.rb +18 -0
  85. data/spec/scanner/properties_spec.rb +64 -0
  86. data/spec/scanner/quantifiers_spec.rb +20 -0
  87. data/spec/scanner/refcalls_spec.rb +36 -0
  88. data/spec/scanner/sets_spec.rb +102 -0
  89. data/spec/scanner/types_spec.rb +14 -0
  90. data/spec/spec_helper.rb +15 -0
  91. data/{test → spec}/support/runner.rb +9 -8
  92. data/spec/support/shared_examples.rb +77 -0
  93. data/{test → spec}/support/warning_extractor.rb +5 -7
  94. data/spec/syntax/syntax_spec.rb +48 -0
  95. data/spec/syntax/syntax_token_map_spec.rb +23 -0
  96. data/spec/syntax/versions/1.8.6_spec.rb +17 -0
  97. data/spec/syntax/versions/1.9.1_spec.rb +10 -0
  98. data/spec/syntax/versions/1.9.3_spec.rb +9 -0
  99. data/spec/syntax/versions/2.0.0_spec.rb +13 -0
  100. data/spec/syntax/versions/2.2.0_spec.rb +9 -0
  101. data/spec/syntax/versions/aliases_spec.rb +37 -0
  102. data/spec/token/token_spec.rb +85 -0
  103. metadata +144 -143
  104. data/test/expression/test_all.rb +0 -12
  105. data/test/expression/test_base.rb +0 -90
  106. data/test/expression/test_clone.rb +0 -89
  107. data/test/expression/test_conditionals.rb +0 -113
  108. data/test/expression/test_free_space.rb +0 -35
  109. data/test/expression/test_set.rb +0 -84
  110. data/test/expression/test_strfregexp.rb +0 -230
  111. data/test/expression/test_subexpression.rb +0 -58
  112. data/test/expression/test_tests.rb +0 -99
  113. data/test/expression/test_to_h.rb +0 -59
  114. data/test/expression/test_to_s.rb +0 -104
  115. data/test/expression/test_traverse.rb +0 -161
  116. data/test/helpers.rb +0 -10
  117. data/test/lexer/test_all.rb +0 -41
  118. data/test/lexer/test_conditionals.rb +0 -127
  119. data/test/lexer/test_keep.rb +0 -24
  120. data/test/lexer/test_literals.rb +0 -130
  121. data/test/lexer/test_nesting.rb +0 -132
  122. data/test/lexer/test_refcalls.rb +0 -56
  123. data/test/parser/set/test_intersections.rb +0 -127
  124. data/test/parser/set/test_ranges.rb +0 -111
  125. data/test/parser/test_all.rb +0 -64
  126. data/test/parser/test_alternation.rb +0 -92
  127. data/test/parser/test_anchors.rb +0 -34
  128. data/test/parser/test_conditionals.rb +0 -187
  129. data/test/parser/test_errors.rb +0 -63
  130. data/test/parser/test_escapes.rb +0 -134
  131. data/test/parser/test_free_space.rb +0 -139
  132. data/test/parser/test_groups.rb +0 -289
  133. data/test/parser/test_keep.rb +0 -21
  134. data/test/parser/test_posix_classes.rb +0 -27
  135. data/test/parser/test_properties.rb +0 -133
  136. data/test/parser/test_quantifiers.rb +0 -301
  137. data/test/parser/test_refcalls.rb +0 -186
  138. data/test/parser/test_sets.rb +0 -179
  139. data/test/parser/test_types.rb +0 -50
  140. data/test/scanner/test_all.rb +0 -38
  141. data/test/scanner/test_anchors.rb +0 -38
  142. data/test/scanner/test_conditionals.rb +0 -184
  143. data/test/scanner/test_errors.rb +0 -91
  144. data/test/scanner/test_escapes.rb +0 -56
  145. data/test/scanner/test_free_space.rb +0 -200
  146. data/test/scanner/test_groups.rb +0 -79
  147. data/test/scanner/test_keep.rb +0 -35
  148. data/test/scanner/test_literals.rb +0 -89
  149. data/test/scanner/test_meta.rb +0 -40
  150. data/test/scanner/test_properties.rb +0 -312
  151. data/test/scanner/test_quantifiers.rb +0 -37
  152. data/test/scanner/test_refcalls.rb +0 -52
  153. data/test/scanner/test_scripts.rb +0 -53
  154. data/test/scanner/test_sets.rb +0 -119
  155. data/test/scanner/test_types.rb +0 -35
  156. data/test/scanner/test_unicode_blocks.rb +0 -30
  157. data/test/support/disable_autotest.rb +0 -8
  158. data/test/syntax/test_all.rb +0 -6
  159. data/test/syntax/test_syntax.rb +0 -61
  160. data/test/syntax/test_syntax_token_map.rb +0 -25
  161. data/test/syntax/versions/test_1.8.rb +0 -55
  162. data/test/syntax/versions/test_1.9.1.rb +0 -36
  163. data/test/syntax/versions/test_1.9.3.rb +0 -32
  164. data/test/syntax/versions/test_2.0.0.rb +0 -37
  165. data/test/syntax/versions/test_2.2.0.rb +0 -32
  166. data/test/syntax/versions/test_aliases.rb +0 -129
  167. data/test/syntax/versions/test_all.rb +0 -5
  168. data/test/test_all.rb +0 -5
  169. data/test/token/test_all.rb +0 -2
  170. data/test/token/test_token.rb +0 -107
@@ -0,0 +1,161 @@
1
+ require 'spec_helper'
2
+
3
+ RSpec.describe('Subexpression#traverse') do
4
+ specify('Subexpression#traverse') do
5
+ root = RP.parse(/a(b(c(d)))|g[h-i]j|klmn/)
6
+
7
+ enters = 0
8
+ visits = 0
9
+ exits = 0
10
+
11
+ root.traverse do |event, _exp, _index|
12
+ enters = (enters + 1) if event == :enter
13
+ visits = (visits + 1) if event == :visit
14
+ exits = (exits + 1) if event == :exit
15
+ end
16
+
17
+ expect(enters).to eq 9
18
+ expect(enters).to eq exits
19
+
20
+ expect(visits).to eq 9
21
+ end
22
+
23
+ specify('Subexpression#traverse including self') do
24
+ root = RP.parse(/a(b(c(d)))|g[h-i]j|klmn/)
25
+
26
+ enters = 0
27
+ visits = 0
28
+ exits = 0
29
+
30
+ root.traverse(true) do |event, _exp, _index|
31
+ enters = (enters + 1) if event == :enter
32
+ visits = (visits + 1) if event == :visit
33
+ exits = (exits + 1) if event == :exit
34
+ end
35
+
36
+ expect(enters).to eq 10
37
+ expect(enters).to eq exits
38
+
39
+ expect(visits).to eq 9
40
+ end
41
+
42
+ specify('Subexpression#traverse without a block') do
43
+ root = RP.parse(/abc/)
44
+ enum = root.traverse
45
+
46
+ expect(enum).to be_a(Enumerator)
47
+ event, expr, idx = enum.next
48
+ expect(event).to eq(:visit)
49
+ expect(expr).to be_a(Regexp::Expression::Literal)
50
+ expect(idx).to eq(0)
51
+ end
52
+
53
+ specify('Subexpression#walk alias') do
54
+ root = RP.parse(/abc/)
55
+
56
+ expect(root).to respond_to(:walk)
57
+ end
58
+
59
+ specify('Subexpression#each_expression') do
60
+ root = RP.parse(/a(?x:b(c))|g[h-k]/)
61
+
62
+ count = 0
63
+ root.each_expression { count += 1 }
64
+
65
+ expect(count).to eq 13
66
+ end
67
+
68
+ specify('Subexpression#each_expression including self') do
69
+ root = RP.parse(/a(?x:b(c))|g[h-k]/)
70
+
71
+ count = 0
72
+ root.each_expression(true) { count += 1 }
73
+
74
+ expect(count).to eq 14
75
+ end
76
+
77
+ specify('Subexpression#each_expression indices') do
78
+ root = RP.parse(/a(b)c/)
79
+
80
+ indices = []
81
+ root.each_expression { |_exp, index| (indices << index) }
82
+
83
+ expect(indices).to eq [0, 1, 0, 2]
84
+ end
85
+
86
+ specify('Subexpression#each_expression indices including self') do
87
+ root = RP.parse(/a(b)c/)
88
+
89
+ indices = []
90
+ root.each_expression(true) { |_exp, index| (indices << index) }
91
+
92
+ expect(indices).to eq [0, 0, 1, 0, 2]
93
+ end
94
+
95
+ specify('Subexpression#each_expression without a block') do
96
+ root = RP.parse(/abc/)
97
+ enum = root.each_expression
98
+
99
+ expect(enum).to be_a(Enumerator)
100
+ expr, idx = enum.next
101
+ expect(expr).to be_a(Regexp::Expression::Literal)
102
+ expect(idx).to eq(0)
103
+ end
104
+
105
+ specify('Subexpression#flat_map without block') do
106
+ root = RP.parse(/a(b([c-e]+))?/)
107
+
108
+ array = root.flat_map
109
+
110
+ expect(array).to be_instance_of(Array)
111
+ expect(array.length).to eq 8
112
+
113
+ array.each do |item|
114
+ expect(item).to be_instance_of(Array)
115
+ expect(item.length).to eq 2
116
+ expect(item.first).to be_a(Regexp::Expression::Base)
117
+ expect(item.last).to be_a(Integer)
118
+ end
119
+ end
120
+
121
+ specify('Subexpression#flat_map without block including self') do
122
+ root = RP.parse(/a(b([c-e]+))?/)
123
+
124
+ array = root.flat_map(true)
125
+
126
+ expect(array).to be_instance_of(Array)
127
+ expect(array.length).to eq 9
128
+ end
129
+
130
+ specify('Subexpression#flat_map indices') do
131
+ root = RP.parse(/a(b([c-e]+))?f*g/)
132
+
133
+ indices = root.flat_map { |_exp, index| index }
134
+
135
+ expect(indices).to eq [0, 1, 0, 1, 0, 0, 0, 1, 2, 3]
136
+ end
137
+
138
+ specify('Subexpression#flat_map indices including self') do
139
+ root = RP.parse(/a(b([c-e]+))?f*g/)
140
+
141
+ indices = root.flat_map(true) { |_exp, index| index }
142
+
143
+ expect(indices).to eq [0, 0, 1, 0, 1, 0, 0, 0, 1, 2, 3]
144
+ end
145
+
146
+ specify('Subexpression#flat_map expressions') do
147
+ root = RP.parse(/a(b(c(d)))/)
148
+
149
+ levels = root.flat_map { |exp, _index| [exp.level, exp.text] if exp.terminal? }.compact
150
+
151
+ expect(levels).to eq [[0, 'a'], [1, 'b'], [2, 'c'], [3, 'd']]
152
+ end
153
+
154
+ specify('Subexpression#flat_map expressions including self') do
155
+ root = RP.parse(/a(b(c(d)))/)
156
+
157
+ levels = root.flat_map(true) { |exp, _index| [exp.level, exp.to_s] }.compact
158
+
159
+ expect(levels).to eq [[nil, 'a(b(c(d)))'], [0, 'a'], [0, '(b(c(d)))'], [1, 'b'], [1, '(c(d))'], [2, 'c'], [2, '(d)'], [3, 'd']]
160
+ end
161
+ end
@@ -0,0 +1,128 @@
1
+ require 'spec_helper'
2
+
3
+ RSpec.describe('Expression#options') do
4
+ it 'returns a hash of options/flags that affect the expression' do
5
+ exp = RP.parse(/a/ix)[0]
6
+ expect(exp).to be_a Literal
7
+ expect(exp.options).to eq(i: true, x: true)
8
+ end
9
+
10
+ it 'includes options that are locally enabled via special groups' do
11
+ exp = RP.parse(/(?x)(?m:a)/i)[1][0]
12
+ expect(exp).to be_a Literal
13
+ expect(exp.options).to eq(i: true, m: true, x: true)
14
+ end
15
+
16
+ it 'excludes locally disabled options' do
17
+ exp = RP.parse(/(?x)(?-im:a)/i)[1][0]
18
+ expect(exp).to be_a Literal
19
+ expect(exp.options).to eq(x: true)
20
+ end
21
+
22
+ it 'gives correct precedence to negative options' do
23
+ # Negative options have precedence. E.g. /(?i-i)a/ is case-sensitive.
24
+ regexp = /(?i-i:a)/
25
+ expect(regexp).to match 'a'
26
+ expect(regexp).not_to match 'A'
27
+
28
+ exp = RP.parse(regexp)[0][0]
29
+ expect(exp).to be_a Literal
30
+ expect(exp.options).to eq({})
31
+ end
32
+
33
+ it 'correctly handles multiple negative option parts' do
34
+ regexp = /(?--m--mx--) . /mx
35
+ expect(regexp).to match ' . '
36
+ expect(regexp).not_to match '.'
37
+ expect(regexp).not_to match "\n"
38
+
39
+ exp = RP.parse(regexp)[2]
40
+ expect(exp.options).to eq({})
41
+ end
42
+
43
+ it 'gives correct precedence when encountering multiple encoding flags' do
44
+ # Any encoding flag overrides all previous encoding flags. If there are
45
+ # multiple encoding flags in an options string, the last one wins.
46
+ # E.g. /(?dau)\w/ matches UTF8 chars but /(?dua)\w/ only ASCII chars.
47
+ regexp1 = /(?dau)\w/
48
+ regexp2 = /(?dua)\w/
49
+ expect(regexp1).to match 'ü'
50
+ expect(regexp2).not_to match 'ü'
51
+
52
+ exp1 = RP.parse(regexp1)[1]
53
+ exp2 = RP.parse(regexp2)[1]
54
+ expect(exp1.options).to eq(u: true)
55
+ expect(exp2.options).to eq(a: true)
56
+ end
57
+
58
+ it 'is accessible via shortcuts' do
59
+ exp = Root.build
60
+
61
+ expect { exp.options[:i] = true }
62
+ .to change { exp.i? }.from(false).to(true)
63
+ .and change { exp.ignore_case? }.from(false).to(true)
64
+ .and change { exp.case_insensitive? }.from(false).to(true)
65
+
66
+ expect { exp.options[:m] = true }
67
+ .to change { exp.m? }.from(false).to(true)
68
+ .and change { exp.multiline? }.from(false).to(true)
69
+
70
+ expect { exp.options[:x] = true }
71
+ .to change { exp.x? }.from(false).to(true)
72
+ .and change { exp.extended? }.from(false).to(true)
73
+ .and change { exp.free_spacing? }.from(false).to(true)
74
+
75
+ expect { exp.options[:a] = true }
76
+ .to change { exp.a? }.from(false).to(true)
77
+ .and change { exp.ascii_classes? }.from(false).to(true)
78
+
79
+ expect { exp.options[:d] = true }
80
+ .to change { exp.d? }.from(false).to(true)
81
+ .and change { exp.default_classes? }.from(false).to(true)
82
+
83
+ expect { exp.options[:u] = true }
84
+ .to change { exp.u? }.from(false).to(true)
85
+ .and change { exp.unicode_classes? }.from(false).to(true)
86
+ end
87
+
88
+ RSpec.shared_examples '#options' do |regexp, path, klass|
89
+ it "works for expression class #{klass}" do
90
+ exp = RP.parse(/#{regexp.source}/i).dig(*path)
91
+ expect(exp).to be_a(klass)
92
+ expect(exp).to be_i
93
+ expect(exp).not_to be_x
94
+ end
95
+ end
96
+
97
+ include_examples '#options', //, [], Root
98
+ include_examples '#options', /a/, [0], Literal
99
+ include_examples '#options', /\A/, [0], Anchor::Base
100
+ include_examples '#options', /\d/, [0], CharacterType::Base
101
+ include_examples '#options', /\n/, [0], EscapeSequence::Base
102
+ include_examples '#options', /\K/, [0], Keep::Mark
103
+ include_examples '#options', /./, [0], CharacterType::Any
104
+ include_examples '#options', /(a)/, [0], Group::Base
105
+ include_examples '#options', /(a)/, [0, 0], Literal
106
+ include_examples '#options', /(?=a)/, [0], Assertion::Base
107
+ include_examples '#options', /(?=a)/, [0, 0], Literal
108
+ include_examples '#options', /(a|b)/, [0], Group::Base
109
+ include_examples '#options', /(a|b)/, [0, 0], Alternation
110
+ include_examples '#options', /(a|b)/, [0, 0, 0], Alternative
111
+ include_examples '#options', /(a|b)/, [0, 0, 0, 0], Literal
112
+ include_examples '#options', /(a)\1/, [1], Backreference::Base
113
+ include_examples '#options', /(a)\k<1>/, [1], Backreference::Number
114
+ include_examples '#options', /(a)\g<1>/, [1], Backreference::NumberCall
115
+ include_examples '#options', /[a]/, [0], CharacterSet
116
+ include_examples '#options', /[a]/, [0, 0], Literal
117
+ include_examples '#options', /[a-z]/, [0, 0], CharacterSet::Range
118
+ include_examples '#options', /[a-z]/, [0, 0, 0], Literal
119
+ include_examples '#options', /[a&&z]/, [0, 0], CharacterSet::Intersection
120
+ include_examples '#options', /[a&&z]/, [0, 0, 0], CharacterSet::IntersectedSequence
121
+ include_examples '#options', /[a&&z]/, [0, 0, 0, 0], Literal
122
+ include_examples '#options', /[[:ascii:]]/, [0, 0], PosixClass
123
+ include_examples '#options', /\p{word}/, [0], UnicodeProperty::Base
124
+ include_examples '#options', /(a)(?(1)b|c)/, [1], Conditional::Expression
125
+ include_examples '#options', /(a)(?(1)b|c)/, [1, 0], Conditional::Condition
126
+ include_examples '#options', /(a)(?(1)b|c)/, [1, 1], Conditional::Branch
127
+ include_examples '#options', /(a)(?(1)b|c)/, [1, 1, 0], Literal
128
+ end
@@ -0,0 +1,9 @@
1
+ require 'spec_helper'
2
+
3
+ RSpec.describe(Regexp::Expression::Root) do
4
+ describe('#initialize') do
5
+ it 'supports the old, nonstandard arity for backwards compatibility' do
6
+ expect { Root.new }.to output.to_stderr
7
+ end
8
+ end
9
+ end
@@ -0,0 +1,9 @@
1
+ require 'spec_helper'
2
+
3
+ RSpec.describe(Regexp::Expression::Sequence) do
4
+ describe('#initialize') do
5
+ it 'supports the old, nonstandard arity for backwards compatibility' do
6
+ expect { Sequence.new(0, 0, 0) }.to output.to_stderr
7
+ end
8
+ end
9
+ end
@@ -0,0 +1,50 @@
1
+ require 'spec_helper'
2
+
3
+ RSpec.describe(Regexp::Expression::Subexpression) do
4
+ specify('#ts, #te') do
5
+ regx = /abcd|ghij|klmn|pqur/
6
+ root = RP.parse(regx)
7
+
8
+ alt = root.first
9
+
10
+ { 0 => [0, 4], 1 => [5, 9], 2 => [10, 14], 3 => [15, 19] }.each do |index, span|
11
+ sequence = alt[index]
12
+
13
+ expect(sequence.ts).to eq span[0]
14
+ expect(sequence.te).to eq span[1]
15
+ end
16
+ end
17
+
18
+ specify('#nesting_level') do
19
+ root = RP.parse(/a(b(\d|[ef-g[h]]))/)
20
+
21
+ tests = {
22
+ 'a' => 1,
23
+ 'b' => 2,
24
+ '\d|[ef-g[h]]' => 3, # alternation
25
+ '\d' => 4, # first alternative
26
+ '[ef-g[h]]' => 4, # second alternative
27
+ 'e' => 5,
28
+ 'f-g' => 5,
29
+ 'f' => 6,
30
+ 'g' => 6,
31
+ 'h' => 6,
32
+ }
33
+
34
+ root.each_expression do |exp|
35
+ next unless expected_nesting_level = tests.delete(exp.to_s)
36
+ expect(expected_nesting_level).to eq exp.nesting_level
37
+ end
38
+
39
+ expect(tests).to be_empty
40
+ end
41
+
42
+ specify('#dig') do
43
+ root = RP.parse(/(((a)))/)
44
+
45
+ expect(root.dig(0).to_s).to eq '(((a)))'
46
+ expect(root.dig(0, 0, 0, 0).to_s).to eq 'a'
47
+ expect(root.dig(0, 0, 0, 0, 0)).to be_nil
48
+ expect(root.dig(3, 7)).to be_nil
49
+ end
50
+ end
@@ -0,0 +1,26 @@
1
+ require 'spec_helper'
2
+
3
+ RSpec.describe('Expression#to_h') do
4
+ specify('Root#to_h') do
5
+ root = RP.parse('abc')
6
+
7
+ hash = root.to_h
8
+
9
+ expect(token: :root, type: :expression, text: 'abc', starts_at: 0, length: 3, quantifier: nil, options: {}, level: nil, set_level: nil, conditional_level: nil, expressions: [{ token: :literal, type: :literal, text: 'abc', starts_at: 0, length: 3, quantifier: nil, options: {}, level: 0, set_level: 0, conditional_level: 0 }]).to eq hash
10
+ end
11
+
12
+ specify('Quantifier#to_h') do
13
+ root = RP.parse('a{2,4}')
14
+ exp = root.expressions.at(0)
15
+
16
+ hash = exp.quantifier.to_h
17
+
18
+ expect(max: 4, min: 2, mode: :greedy, text: '{2,4}', token: :interval).to eq hash
19
+ end
20
+
21
+ specify('Conditional#to_h') do
22
+ root = RP.parse('(?<A>a)(?(<A>)b|c)', 'ruby/2.0')
23
+
24
+ expect { root.to_h }.not_to(raise_error)
25
+ end
26
+ end
@@ -0,0 +1,100 @@
1
+ require 'spec_helper'
2
+
3
+ RSpec.describe('Expression#to_s') do
4
+ specify('literal alternation') do
5
+ pattern = 'abcd|ghij|klmn|pqur'
6
+
7
+ expect(RP.parse(pattern).to_s).to eq pattern
8
+ end
9
+
10
+ specify('quantified alternations') do
11
+ pattern = '(?:a?[b]+(c){2}|d+[e]*(f)?)|(?:g+[h]?(i){2,3}|j*[k]{3,5}(l)?)'
12
+
13
+ expect(RP.parse(pattern).to_s).to eq pattern
14
+ end
15
+
16
+ specify('quantified sets') do
17
+ pattern = '[abc]+|[^def]{3,6}'
18
+
19
+ expect(RP.parse(pattern).to_s).to eq pattern
20
+ end
21
+
22
+ specify('property sets') do
23
+ pattern = '[\\a\\b\\p{Lu}\\P{Z}\\c\\d]+'
24
+
25
+ expect(RP.parse(pattern, 'ruby/1.9').to_s).to eq pattern
26
+ end
27
+
28
+ specify('groups') do
29
+ pattern = "(a(?>b(?:c(?<n>d(?'N'e)??f)+g)*+h)*i)++"
30
+
31
+ expect(RP.parse(pattern, 'ruby/1.9').to_s).to eq pattern
32
+ end
33
+
34
+ specify('assertions') do
35
+ pattern = '(a+(?=b+(?!c+(?<=d+(?<!e+)?f+)?g+)?h+)?i+)?'
36
+
37
+ expect(RP.parse(pattern, 'ruby/1.9').to_s).to eq pattern
38
+ end
39
+
40
+ specify('comments') do
41
+ pattern = '(?#start)a(?#middle)b(?#end)'
42
+
43
+ expect(RP.parse(pattern).to_s).to eq pattern
44
+ end
45
+
46
+ specify('options') do
47
+ pattern = '(?mix:start)a(?-mix:middle)b(?i-mx:end)'
48
+
49
+ expect(RP.parse(pattern).to_s).to eq pattern
50
+ end
51
+
52
+ specify('url') do
53
+ pattern = ('(^$)|(^(http|https):\\/\\/[a-z0-9]+([\\-\\.]{1}[a-z0-9]+)*' + '\\.[a-z]{2,5}(([0-9]{1,5})?\\/.*)?$)')
54
+
55
+ expect(RP.parse(pattern).to_s).to eq pattern
56
+ end
57
+
58
+ specify('multiline source') do
59
+ multiline = /
60
+ \A
61
+ a? # One letter
62
+ b{2,5} # Another one
63
+ [c-g]+ # A set
64
+ \z
65
+ /x
66
+
67
+ expect(RP.parse(multiline).to_s).to eq multiline.source
68
+ end
69
+
70
+ specify('multiline #to_s') do
71
+ multiline = /
72
+ \A
73
+ a? # One letter
74
+ b{2,5} # Another one
75
+ [c-g]+ # A set
76
+ \z
77
+ /x
78
+
79
+ expect(RP.parse(multiline.to_s).to_s).to eq multiline.to_s
80
+ end
81
+
82
+ # Free spacing expressions that use spaces between quantifiers and their
83
+ # targets do not produce identical results due to the way quantifiers are
84
+ # applied to expressions (members, not nodes) and the merging of consecutive
85
+ # space nodes. This tests that they produce equivalent results.
86
+ specify('multiline equivalence') do
87
+ multiline = /
88
+ \A
89
+ a ? # One letter
90
+ b {2,5} # Another one
91
+ [c-g] + # A set
92
+ \z
93
+ /x
94
+
95
+ str = 'bbbcged'
96
+ root = RP.parse(multiline)
97
+
98
+ expect(Regexp.new(root.to_s, Regexp::EXTENDED).match(str)[0]).to eq multiline.match(str)[0]
99
+ end
100
+ end