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.
Files changed (165) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +364 -22
  3. data/Gemfile +8 -2
  4. data/LICENSE +1 -1
  5. data/README.md +124 -88
  6. data/Rakefile +6 -70
  7. data/lib/regexp_parser/error.rb +4 -0
  8. data/lib/regexp_parser/expression/base.rb +76 -0
  9. data/lib/regexp_parser/expression/classes/alternation.rb +1 -1
  10. data/lib/regexp_parser/expression/classes/anchor.rb +0 -2
  11. data/lib/regexp_parser/expression/classes/{backref.rb → backreference.rb} +22 -2
  12. data/lib/regexp_parser/expression/classes/{set → character_set}/range.rb +4 -8
  13. data/lib/regexp_parser/expression/classes/{set.rb → character_set.rb} +3 -4
  14. data/lib/regexp_parser/expression/classes/{type.rb → character_type.rb} +0 -2
  15. data/lib/regexp_parser/expression/classes/conditional.rb +11 -5
  16. data/lib/regexp_parser/expression/classes/{escape.rb → escape_sequence.rb} +15 -7
  17. data/lib/regexp_parser/expression/classes/free_space.rb +5 -5
  18. data/lib/regexp_parser/expression/classes/group.rb +28 -15
  19. data/lib/regexp_parser/expression/classes/keep.rb +2 -0
  20. data/lib/regexp_parser/expression/classes/literal.rb +1 -5
  21. data/lib/regexp_parser/expression/classes/posix_class.rb +5 -1
  22. data/lib/regexp_parser/expression/classes/root.rb +4 -19
  23. data/lib/regexp_parser/expression/classes/{property.rb → unicode_property.rb} +5 -3
  24. data/lib/regexp_parser/expression/methods/construct.rb +41 -0
  25. data/lib/regexp_parser/expression/methods/human_name.rb +43 -0
  26. data/lib/regexp_parser/expression/methods/match_length.rb +11 -7
  27. data/lib/regexp_parser/expression/methods/parts.rb +23 -0
  28. data/lib/regexp_parser/expression/methods/printing.rb +26 -0
  29. data/lib/regexp_parser/expression/methods/strfregexp.rb +1 -1
  30. data/lib/regexp_parser/expression/methods/tests.rb +47 -1
  31. data/lib/regexp_parser/expression/methods/traverse.rb +34 -18
  32. data/lib/regexp_parser/expression/quantifier.rb +57 -17
  33. data/lib/regexp_parser/expression/sequence.rb +11 -47
  34. data/lib/regexp_parser/expression/sequence_operation.rb +4 -9
  35. data/lib/regexp_parser/expression/shared.rb +111 -0
  36. data/lib/regexp_parser/expression/subexpression.rb +27 -19
  37. data/lib/regexp_parser/expression.rb +14 -141
  38. data/lib/regexp_parser/lexer.rb +83 -41
  39. data/lib/regexp_parser/parser.rb +371 -429
  40. data/lib/regexp_parser/scanner/char_type.rl +11 -11
  41. data/lib/regexp_parser/scanner/errors/premature_end_error.rb +8 -0
  42. data/lib/regexp_parser/scanner/errors/scanner_error.rb +6 -0
  43. data/lib/regexp_parser/scanner/errors/validation_error.rb +63 -0
  44. data/lib/regexp_parser/scanner/properties/long.csv +633 -0
  45. data/lib/regexp_parser/scanner/properties/short.csv +248 -0
  46. data/lib/regexp_parser/scanner/property.rl +4 -4
  47. data/lib/regexp_parser/scanner/scanner.rl +295 -368
  48. data/lib/regexp_parser/scanner.rb +1405 -1674
  49. data/lib/regexp_parser/syntax/any.rb +2 -7
  50. data/lib/regexp_parser/syntax/base.rb +92 -67
  51. data/lib/regexp_parser/syntax/token/anchor.rb +15 -0
  52. data/lib/regexp_parser/syntax/{tokens → token}/assertion.rb +2 -2
  53. data/lib/regexp_parser/syntax/token/backreference.rb +33 -0
  54. data/lib/regexp_parser/syntax/token/character_set.rb +16 -0
  55. data/lib/regexp_parser/syntax/{tokens → token}/character_type.rb +3 -3
  56. data/lib/regexp_parser/syntax/{tokens → token}/conditional.rb +3 -3
  57. data/lib/regexp_parser/syntax/token/escape.rb +33 -0
  58. data/lib/regexp_parser/syntax/{tokens → token}/group.rb +7 -7
  59. data/lib/regexp_parser/syntax/{tokens → token}/keep.rb +1 -1
  60. data/lib/regexp_parser/syntax/token/meta.rb +20 -0
  61. data/lib/regexp_parser/syntax/{tokens → token}/posix_class.rb +3 -3
  62. data/lib/regexp_parser/syntax/token/quantifier.rb +35 -0
  63. data/lib/regexp_parser/syntax/token/unicode_property.rb +733 -0
  64. data/lib/regexp_parser/syntax/token/virtual.rb +11 -0
  65. data/lib/regexp_parser/syntax/token.rb +45 -0
  66. data/lib/regexp_parser/syntax/version_lookup.rb +19 -36
  67. data/lib/regexp_parser/syntax/versions/1.8.6.rb +13 -20
  68. data/lib/regexp_parser/syntax/versions/1.9.1.rb +10 -17
  69. data/lib/regexp_parser/syntax/versions/1.9.3.rb +3 -10
  70. data/lib/regexp_parser/syntax/versions/2.0.0.rb +8 -15
  71. data/lib/regexp_parser/syntax/versions/2.2.0.rb +3 -9
  72. data/lib/regexp_parser/syntax/versions/2.3.0.rb +3 -9
  73. data/lib/regexp_parser/syntax/versions/2.4.0.rb +3 -9
  74. data/lib/regexp_parser/syntax/versions/2.4.1.rb +2 -8
  75. data/lib/regexp_parser/syntax/versions/2.5.0.rb +3 -9
  76. data/lib/regexp_parser/syntax/versions/2.6.0.rb +3 -9
  77. data/lib/regexp_parser/syntax/versions/2.6.2.rb +3 -9
  78. data/lib/regexp_parser/syntax/versions/2.6.3.rb +3 -9
  79. data/lib/regexp_parser/syntax/versions/3.1.0.rb +4 -0
  80. data/lib/regexp_parser/syntax/versions/3.2.0.rb +4 -0
  81. data/lib/regexp_parser/syntax/versions.rb +3 -1
  82. data/lib/regexp_parser/syntax.rb +8 -6
  83. data/lib/regexp_parser/token.rb +9 -20
  84. data/lib/regexp_parser/version.rb +1 -1
  85. data/lib/regexp_parser.rb +0 -2
  86. data/regexp_parser.gemspec +20 -22
  87. metadata +49 -166
  88. data/lib/regexp_parser/scanner/properties/long.yml +0 -594
  89. data/lib/regexp_parser/scanner/properties/short.yml +0 -237
  90. data/lib/regexp_parser/syntax/tokens/anchor.rb +0 -15
  91. data/lib/regexp_parser/syntax/tokens/backref.rb +0 -24
  92. data/lib/regexp_parser/syntax/tokens/character_set.rb +0 -13
  93. data/lib/regexp_parser/syntax/tokens/escape.rb +0 -30
  94. data/lib/regexp_parser/syntax/tokens/meta.rb +0 -13
  95. data/lib/regexp_parser/syntax/tokens/quantifier.rb +0 -35
  96. data/lib/regexp_parser/syntax/tokens/unicode_property.rb +0 -675
  97. data/lib/regexp_parser/syntax/tokens.rb +0 -45
  98. data/spec/expression/base_spec.rb +0 -94
  99. data/spec/expression/clone_spec.rb +0 -120
  100. data/spec/expression/conditional_spec.rb +0 -89
  101. data/spec/expression/free_space_spec.rb +0 -27
  102. data/spec/expression/methods/match_length_spec.rb +0 -161
  103. data/spec/expression/methods/match_spec.rb +0 -25
  104. data/spec/expression/methods/strfregexp_spec.rb +0 -224
  105. data/spec/expression/methods/tests_spec.rb +0 -99
  106. data/spec/expression/methods/traverse_spec.rb +0 -161
  107. data/spec/expression/options_spec.rb +0 -128
  108. data/spec/expression/root_spec.rb +0 -9
  109. data/spec/expression/sequence_spec.rb +0 -9
  110. data/spec/expression/subexpression_spec.rb +0 -50
  111. data/spec/expression/to_h_spec.rb +0 -26
  112. data/spec/expression/to_s_spec.rb +0 -100
  113. data/spec/lexer/all_spec.rb +0 -22
  114. data/spec/lexer/conditionals_spec.rb +0 -53
  115. data/spec/lexer/escapes_spec.rb +0 -14
  116. data/spec/lexer/keep_spec.rb +0 -10
  117. data/spec/lexer/literals_spec.rb +0 -89
  118. data/spec/lexer/nesting_spec.rb +0 -99
  119. data/spec/lexer/refcalls_spec.rb +0 -55
  120. data/spec/parser/all_spec.rb +0 -43
  121. data/spec/parser/alternation_spec.rb +0 -88
  122. data/spec/parser/anchors_spec.rb +0 -17
  123. data/spec/parser/conditionals_spec.rb +0 -179
  124. data/spec/parser/errors_spec.rb +0 -30
  125. data/spec/parser/escapes_spec.rb +0 -121
  126. data/spec/parser/free_space_spec.rb +0 -130
  127. data/spec/parser/groups_spec.rb +0 -108
  128. data/spec/parser/keep_spec.rb +0 -6
  129. data/spec/parser/posix_classes_spec.rb +0 -8
  130. data/spec/parser/properties_spec.rb +0 -115
  131. data/spec/parser/quantifiers_spec.rb +0 -51
  132. data/spec/parser/refcalls_spec.rb +0 -112
  133. data/spec/parser/set/intersections_spec.rb +0 -127
  134. data/spec/parser/set/ranges_spec.rb +0 -111
  135. data/spec/parser/sets_spec.rb +0 -178
  136. data/spec/parser/types_spec.rb +0 -18
  137. data/spec/scanner/all_spec.rb +0 -18
  138. data/spec/scanner/anchors_spec.rb +0 -21
  139. data/spec/scanner/conditionals_spec.rb +0 -128
  140. data/spec/scanner/errors_spec.rb +0 -68
  141. data/spec/scanner/escapes_spec.rb +0 -53
  142. data/spec/scanner/free_space_spec.rb +0 -133
  143. data/spec/scanner/groups_spec.rb +0 -52
  144. data/spec/scanner/keep_spec.rb +0 -10
  145. data/spec/scanner/literals_spec.rb +0 -49
  146. data/spec/scanner/meta_spec.rb +0 -18
  147. data/spec/scanner/properties_spec.rb +0 -64
  148. data/spec/scanner/quantifiers_spec.rb +0 -20
  149. data/spec/scanner/refcalls_spec.rb +0 -36
  150. data/spec/scanner/sets_spec.rb +0 -102
  151. data/spec/scanner/types_spec.rb +0 -14
  152. data/spec/spec_helper.rb +0 -15
  153. data/spec/support/runner.rb +0 -42
  154. data/spec/support/shared_examples.rb +0 -77
  155. data/spec/support/warning_extractor.rb +0 -60
  156. data/spec/syntax/syntax_spec.rb +0 -48
  157. data/spec/syntax/syntax_token_map_spec.rb +0 -23
  158. data/spec/syntax/versions/1.8.6_spec.rb +0 -17
  159. data/spec/syntax/versions/1.9.1_spec.rb +0 -10
  160. data/spec/syntax/versions/1.9.3_spec.rb +0 -9
  161. data/spec/syntax/versions/2.0.0_spec.rb +0 -13
  162. data/spec/syntax/versions/2.2.0_spec.rb +0 -9
  163. data/spec/syntax/versions/aliases_spec.rb +0 -37
  164. data/spec/token/token_spec.rb +0 -85
  165. /data/lib/regexp_parser/expression/classes/{set → character_set}/intersection.rb +0 -0
@@ -1,64 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe('Property scanning') do
4
- RSpec.shared_examples 'scan property' do |text, token|
5
- it("scans \\p{#{text}} as property #{token}") do
6
- result = RS.scan("\\p{#{text}}")[0]
7
- expect(result[0..1]).to eq [:property, token]
8
- end
9
-
10
- it("scans \\P{#{text}} as nonproperty #{token}") do
11
- result = RS.scan("\\P{#{text}}")[0]
12
- expect(result[0..1]).to eq [:nonproperty, token]
13
- end
14
-
15
- it("scans \\p{^#{text}} as nonproperty #{token}") do
16
- result = RS.scan("\\p{^#{text}}")[0]
17
- expect(result[0..1]).to eq [:nonproperty, token]
18
- end
19
-
20
- it("scans double-negated \\P{^#{text}} as property #{token}") do
21
- result = RS.scan("\\P{^#{text}}")[0]
22
- expect(result[0..1]).to eq [:property, token]
23
- end
24
- end
25
-
26
- include_examples 'scan property', 'Alnum', :alnum
27
-
28
- include_examples 'scan property', 'XPosixPunct', :xposixpunct
29
-
30
- include_examples 'scan property', 'Newline', :newline
31
-
32
- include_examples 'scan property', 'Any', :any
33
-
34
- include_examples 'scan property', 'Assigned', :assigned
35
-
36
- include_examples 'scan property', 'Age=1.1', :'age=1.1'
37
- include_examples 'scan property', 'Age=10.0', :'age=10.0'
38
-
39
- include_examples 'scan property', 'ahex', :ascii_hex_digit
40
- include_examples 'scan property', 'ASCII_Hex_Digit', :ascii_hex_digit # test underscore
41
-
42
- include_examples 'scan property', 'sd', :soft_dotted
43
- include_examples 'scan property', 'Soft-Dotted', :soft_dotted # test dash
44
-
45
- include_examples 'scan property', 'Egyp', :egyptian_hieroglyphs
46
- include_examples 'scan property', 'Egyptian Hieroglyphs', :egyptian_hieroglyphs # test whitespace
47
-
48
- include_examples 'scan property', 'Linb', :linear_b
49
- include_examples 'scan property', 'Linear-B', :linear_b # test dash
50
-
51
- include_examples 'scan property', 'InArabic', :in_arabic # test block
52
- include_examples 'scan property', 'in Arabic', :in_arabic # test block w. whitespace
53
- include_examples 'scan property', 'In_Arabic', :in_arabic # test block w. underscore
54
-
55
- include_examples 'scan property', 'Yiii', :yi
56
- include_examples 'scan property', 'Yi', :yi
57
-
58
- include_examples 'scan property', 'Zinh', :inherited
59
- include_examples 'scan property', 'Inherited', :inherited
60
- include_examples 'scan property', 'Qaai', :inherited
61
-
62
- include_examples 'scan property', 'Zzzz', :unknown
63
- include_examples 'scan property', 'Unknown', :unknown
64
- end
@@ -1,20 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe('Quantifier scanning') do
4
- include_examples 'scan', 'a?', 1 => [:quantifier, :zero_or_one, '?', 1, 2]
5
- include_examples 'scan', 'a??', 1 => [:quantifier, :zero_or_one_reluctant, '??', 1, 3]
6
- include_examples 'scan', 'a?+', 1 => [:quantifier, :zero_or_one_possessive, '?+', 1, 3]
7
-
8
- include_examples 'scan', 'a*', 1 => [:quantifier, :zero_or_more, '*', 1, 2]
9
- include_examples 'scan', 'a*?', 1 => [:quantifier, :zero_or_more_reluctant, '*?', 1, 3]
10
- include_examples 'scan', 'a*+', 1 => [:quantifier, :zero_or_more_possessive, '*+', 1, 3]
11
-
12
- include_examples 'scan', 'a+', 1 => [:quantifier, :one_or_more, '+', 1, 2]
13
- include_examples 'scan', 'a+?', 1 => [:quantifier, :one_or_more_reluctant, '+?', 1, 3]
14
- include_examples 'scan', 'a++', 1 => [:quantifier, :one_or_more_possessive, '++', 1, 3]
15
-
16
- include_examples 'scan', 'a{2}', 1 => [:quantifier, :interval, '{2}', 1, 4]
17
- include_examples 'scan', 'a{2,}', 1 => [:quantifier, :interval, '{2,}', 1, 5]
18
- include_examples 'scan', 'a{,2}', 1 => [:quantifier, :interval, '{,2}', 1, 5]
19
- include_examples 'scan', 'a{2,4}', 1 => [:quantifier, :interval, '{2,4}', 1, 6]
20
- end
@@ -1,36 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe('RefCall scanning') do
4
- # Traditional numerical group back-reference
5
- include_examples 'scan', '(abc)\1' , 3 => [:backref, :number, '\1', 5, 7]
6
-
7
- # Group back-references, named, numbered, and relative
8
- include_examples 'scan', '(?<X>abc)\k<X>', 3 => [:backref, :name_ref_ab, '\k<X>', 9, 14]
9
- include_examples 'scan', "(?<X>abc)\\k'X'", 3 => [:backref, :name_ref_sq, "\\k'X'", 9, 14]
10
-
11
- include_examples 'scan', '(abc)\k<1>', 3 => [:backref, :number_ref_ab, '\k<1>', 5, 10]
12
- include_examples 'scan', "(abc)\\k'1'", 3 => [:backref, :number_ref_sq, "\\k'1'", 5, 10]
13
-
14
- include_examples 'scan', '(abc)\k<-1>', 3 => [:backref, :number_rel_ref_ab, '\k<-1>', 5, 11]
15
- include_examples 'scan', "(abc)\\k'-1'", 3 => [:backref, :number_rel_ref_sq, "\\k'-1'", 5, 11]
16
-
17
- # Sub-expression invocation, named, numbered, and relative
18
- include_examples 'scan', '(?<X>abc)\g<X>', 3 => [:backref, :name_call_ab, '\g<X>', 9, 14]
19
- include_examples 'scan', "(?<X>abc)\\g'X'", 3 => [:backref, :name_call_sq, "\\g'X'", 9, 14]
20
-
21
- include_examples 'scan', '(abc)\g<1>', 3 => [:backref, :number_call_ab, '\g<1>', 5, 10]
22
- include_examples 'scan', "(abc)\\g'1'", 3 => [:backref, :number_call_sq, "\\g'1'", 5, 10]
23
-
24
- include_examples 'scan', '(abc)\g<-1>', 3 => [:backref, :number_rel_call_ab, '\g<-1>', 5, 11]
25
- include_examples 'scan', "(abc)\\g'-1'", 3 => [:backref, :number_rel_call_sq, "\\g'-1'", 5, 11]
26
-
27
- include_examples 'scan', '\g<+1>(abc)', 0 => [:backref, :number_rel_call_ab, '\g<+1>', 0, 6]
28
- include_examples 'scan', "\\g'+1'(abc)", 0 => [:backref, :number_rel_call_sq, "\\g'+1'", 0, 6]
29
-
30
- # Group back-references, with recursion level
31
- include_examples 'scan', '(?<X>abc)\k<X-0>', 3 => [:backref, :name_recursion_ref_ab, '\k<X-0>', 9, 16]
32
- include_examples 'scan', "(?<X>abc)\\k'X-0'", 3 => [:backref, :name_recursion_ref_sq, "\\k'X-0'", 9, 16]
33
-
34
- include_examples 'scan', '(abc)\k<1-0>', 3 => [:backref, :number_recursion_ref_ab, '\k<1-0>', 5, 12]
35
- include_examples 'scan', "(abc)\\k'1-0'", 3 => [:backref, :number_recursion_ref_sq, "\\k'1-0'", 5, 12]
36
- end
@@ -1,102 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe('Set scanning') do
4
- include_examples 'scan', /[a]/, 0 => [:set, :open, '[', 0, 1]
5
- include_examples 'scan', /[b]/, 2 => [:set, :close, ']', 2, 3]
6
- include_examples 'scan', /[^n]/, 1 => [:set, :negate, '^', 1, 2]
7
-
8
- include_examples 'scan', /[c]/, 1 => [:literal, :literal, 'c', 1, 2]
9
- include_examples 'scan', /[\b]/, 1 => [:escape, :backspace, '\b', 1, 3]
10
- include_examples 'scan', /[A\bX]/, 2 => [:escape, :backspace, '\b', 2, 4]
11
-
12
- include_examples 'scan', /[.]/, 1 => [:literal, :literal, '.', 1, 2]
13
- include_examples 'scan', /[?]/, 1 => [:literal, :literal, '?', 1, 2]
14
- include_examples 'scan', /[*]/, 1 => [:literal, :literal, '*', 1, 2]
15
- include_examples 'scan', /[+]/, 1 => [:literal, :literal, '+', 1, 2]
16
- include_examples 'scan', /[{]/, 1 => [:literal, :literal, '{', 1, 2]
17
- include_examples 'scan', /[}]/, 1 => [:literal, :literal, '}', 1, 2]
18
- include_examples 'scan', /[<]/, 1 => [:literal, :literal, '<', 1, 2]
19
- include_examples 'scan', /[>]/, 1 => [:literal, :literal, '>', 1, 2]
20
-
21
- include_examples 'scan', /[äöü]/, 2 => [:literal, :literal, 'ö', 3, 5]
22
-
23
- include_examples 'scan', /[\x20]/, 1 => [:escape, :hex, '\x20', 1, 5]
24
-
25
- include_examples 'scan', '[\.]', 1 => [:escape, :dot, '\.', 1, 3]
26
- include_examples 'scan', '[\!]', 1 => [:escape, :literal, '\!', 1, 3]
27
- include_examples 'scan', '[\#]', 1 => [:escape, :literal, '\#', 1, 3]
28
- include_examples 'scan', '[\\]]', 1 => [:escape, :set_close, '\]', 1, 3]
29
- include_examples 'scan', '[\\\\]', 1 => [:escape, :backslash, '\\\\', 1, 3]
30
- include_examples 'scan', '[\A]', 1 => [:escape, :literal, '\A', 1, 3]
31
- include_examples 'scan', '[\z]', 1 => [:escape, :literal, '\z', 1, 3]
32
- include_examples 'scan', '[\g]', 1 => [:escape, :literal, '\g', 1, 3]
33
- include_examples 'scan', '[\K]', 1 => [:escape, :literal, '\K', 1, 3]
34
- include_examples 'scan', '[\R]', 1 => [:escape, :literal, '\R', 1, 3]
35
- include_examples 'scan', '[\X]', 1 => [:escape, :literal, '\X', 1, 3]
36
- include_examples 'scan', '[\c2]', 1 => [:escape, :literal, '\c', 1, 3]
37
- include_examples 'scan', '[\B]', 1 => [:escape, :literal, '\B', 1, 3]
38
- include_examples 'scan', '[a\-c]', 2 => [:escape, :literal, '\-', 2, 4]
39
-
40
- include_examples 'scan', /[\d]/, 1 => [:type, :digit, '\d', 1, 3]
41
- include_examples 'scan', /[\da-z]/, 1 => [:type, :digit, '\d', 1, 3]
42
- include_examples 'scan', /[\D]/, 1 => [:type, :nondigit, '\D', 1, 3]
43
-
44
- include_examples 'scan', /[\h]/, 1 => [:type, :hex, '\h', 1, 3]
45
- include_examples 'scan', /[\H]/, 1 => [:type, :nonhex, '\H', 1, 3]
46
-
47
- include_examples 'scan', /[\s]/, 1 => [:type, :space, '\s', 1, 3]
48
- include_examples 'scan', /[\S]/, 1 => [:type, :nonspace, '\S', 1, 3]
49
-
50
- include_examples 'scan', /[\w]/, 1 => [:type, :word, '\w', 1, 3]
51
- include_examples 'scan', /[\W]/, 1 => [:type, :nonword, '\W', 1, 3]
52
-
53
- include_examples 'scan', /[a-b]/, 1 => [:literal, :literal, 'a', 1, 2]
54
- include_examples 'scan', /[a-c]/, 2 => [:set, :range, '-', 2, 3]
55
- include_examples 'scan', /[a-d]/, 3 => [:literal, :literal, 'd', 3, 4]
56
- include_examples 'scan', /[a-b-]/, 4 => [:literal, :literal, '-', 4, 5]
57
- include_examples 'scan', /[-a]/, 1 => [:literal, :literal, '-', 1, 2]
58
- include_examples 'scan', /[a-c^]/, 4 => [:literal, :literal, '^', 4, 5]
59
- include_examples 'scan', /[a-bd-f]/, 2 => [:set, :range, '-', 2, 3]
60
- include_examples 'scan', /[a-cd-f]/, 5 => [:set, :range, '-', 5, 6]
61
-
62
- include_examples 'scan', /[a[:digit:]c]/, 2 => [:posixclass, :digit, '[:digit:]', 2, 11]
63
- include_examples 'scan', /[[:digit:][:space:]]/, 2 => [:posixclass, :space, '[:space:]', 10, 19]
64
- include_examples 'scan', /[[:^digit:]]/, 1 => [:nonposixclass, :digit, '[:^digit:]', 1, 11]
65
-
66
- include_examples 'scan', /[a[.a-b.]c]/, 2 => [:set, :collation, '[.a-b.]', 2, 9]
67
- include_examples 'scan', /[a[=e=]c]/, 2 => [:set, :equivalent, '[=e=]', 2, 7]
68
-
69
- include_examples 'scan', /[a-d&&g-h]/, 4 => [:set, :intersection, '&&', 4, 6]
70
- include_examples 'scan', /[a&&]/, 2 => [:set, :intersection, '&&', 2, 4]
71
- include_examples 'scan', /[&&z]/, 1 => [:set, :intersection, '&&', 1, 3]
72
-
73
- include_examples 'scan', /[a\p{digit}c]/, 2 => [:property, :digit, '\p{digit}', 2, 11]
74
- include_examples 'scan', /[a\P{digit}c]/, 2 => [:nonproperty, :digit, '\P{digit}', 2, 11]
75
- include_examples 'scan', /[a\p{^digit}c]/, 2 => [:nonproperty, :digit, '\p{^digit}', 2, 12]
76
- include_examples 'scan', /[a\P{^digit}c]/, 2 => [:property, :digit, '\P{^digit}', 2, 12]
77
-
78
- include_examples 'scan', /[a\p{ALPHA}c]/, 2 => [:property, :alpha, '\p{ALPHA}', 2, 11]
79
- include_examples 'scan', /[a\p{P}c]/, 2 => [:property, :punctuation,'\p{P}', 2, 7]
80
- include_examples 'scan', /[a\p{P}\P{P}c]/, 3 => [:nonproperty, :punctuation,'\P{P}', 7, 12]
81
-
82
- include_examples 'scan', /[\x20-\x27]/,
83
- 1 => [:escape, :hex, '\x20', 1, 5],
84
- 2 => [:set, :range, '-', 5, 6],
85
- 3 => [:escape, :hex, '\x27', 6, 10]
86
-
87
- include_examples 'scan', /[a-w&&[^c-g]z]/,
88
- 5 => [:set, :open, '[', 6, 7],
89
- 6 => [:set, :negate, '^', 7, 8],
90
- 8 => [:set, :range, '-', 9, 10],
91
- 10=> [:set, :close, ']', 11, 12]
92
-
93
- specify('set literal encoding') do
94
- text = RS.scan('[a]')[1][2].to_s
95
- expect(text).to eq 'a'
96
- expect(text.encoding.to_s).to eq 'UTF-8'
97
-
98
- text = RS.scan("[\u{1F632}]")[1][2].to_s
99
- expect(text).to eq "\u{1F632}"
100
- expect(text.encoding.to_s).to eq 'UTF-8'
101
- end
102
- end
@@ -1,14 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe('Type scanning') do
4
- include_examples 'scan', 'a\\dc', 1 => [:type, :digit, '\\d', 1, 3]
5
- include_examples 'scan', 'a\\Dc', 1 => [:type, :nondigit, '\\D', 1, 3]
6
- include_examples 'scan', 'a\\hc', 1 => [:type, :hex, '\\h', 1, 3]
7
- include_examples 'scan', 'a\\Hc', 1 => [:type, :nonhex, '\\H', 1, 3]
8
- include_examples 'scan', 'a\\sc', 1 => [:type, :space, '\\s', 1, 3]
9
- include_examples 'scan', 'a\\Sc', 1 => [:type, :nonspace, '\\S', 1, 3]
10
- include_examples 'scan', 'a\\wc', 1 => [:type, :word, '\\w', 1, 3]
11
- include_examples 'scan', 'a\\Wc', 1 => [:type, :nonword, '\\W', 1, 3]
12
- include_examples 'scan', 'a\\Rc', 1 => [:type, :linebreak, '\\R', 1, 3]
13
- include_examples 'scan', 'a\\Xc', 1 => [:type, :xgrapheme, '\\X', 1, 3]
14
- end
data/spec/spec_helper.rb DELETED
@@ -1,15 +0,0 @@
1
- require 'regexp_parser'
2
- require 'regexp_property_values'
3
- require_relative 'support/shared_examples'
4
-
5
- RS = Regexp::Scanner
6
- RL = Regexp::Lexer
7
- RP = Regexp::Parser
8
- RE = Regexp::Expression
9
- T = Regexp::Syntax::Token
10
-
11
- include Regexp::Expression
12
-
13
- def ruby_version_at_least(version)
14
- Gem::Version.new(RUBY_VERSION.dup) >= Gem::Version.new(version)
15
- end
@@ -1,42 +0,0 @@
1
- require 'pathname'
2
- require 'rspec'
3
-
4
- module RegexpParserSpec
5
- class Runner
6
- def initialize(arguments, warning_whitelist)
7
- @arguments = arguments
8
- @warning_whitelist = warning_whitelist
9
- end
10
-
11
- def run
12
- spec_status = nil
13
-
14
- Warning::Filter.new(warning_whitelist).assert_expected_warnings_only do
15
- setup
16
- spec_status = run_rspec
17
- end
18
-
19
- spec_status
20
- end
21
-
22
- private
23
-
24
- def setup
25
- $VERBOSE = true
26
-
27
- spec_files.each(&method(:require))
28
- end
29
-
30
- def run_rspec
31
- RSpec::Core::Runner.run([])
32
- end
33
-
34
- def spec_files
35
- arguments
36
- .map { |path| Pathname.new(path).expand_path.freeze }
37
- .select(&:file?)
38
- end
39
-
40
- attr_reader :arguments, :warning_whitelist
41
- end
42
- end
@@ -1,77 +0,0 @@
1
- RSpec.shared_examples 'syntax' do |klass, opts|
2
- opts[:implements].each do |type, tokens|
3
- tokens.each do |token|
4
- it("implements #{token} #{type}") do
5
- expect(klass.implements?(type, token)).to be true
6
- end
7
- end
8
- end
9
-
10
- opts[:excludes] && opts[:excludes].each do |type, tokens|
11
- tokens.each do |token|
12
- it("does not implement #{token} #{type}") do
13
- expect(klass.implements?(type, token)).to be false
14
- end
15
- end
16
- end
17
- end
18
-
19
- RSpec.shared_examples 'scan' do |pattern, checks|
20
- context "given the pattern #{pattern}" do
21
- before(:all) { @tokens = Regexp::Scanner.scan(pattern) }
22
-
23
- checks.each do |index, (type, token, text, ts, te)|
24
- it "scans token #{index} as #{token} #{type} at #{ts}..#{te}" do
25
- result = @tokens.at(index)
26
-
27
- expect(result[0]).to eq type
28
- expect(result[1]).to eq token
29
- expect(result[2]).to eq text
30
- expect(result[3]).to eq ts
31
- expect(result[4]).to eq te
32
- end
33
- end
34
- end
35
- end
36
-
37
- RSpec.shared_examples 'lex' do |pattern, checks|
38
- context "given the pattern #{pattern}" do
39
- before(:all) { @tokens = Regexp::Lexer.lex(pattern) }
40
-
41
- checks.each do |index, (type, token, text, ts, te, lvl, set_lvl, cond_lvl)|
42
- it "lexes token #{index} as #{token} #{type} at #{lvl}, #{set_lvl}, #{cond_lvl}" do
43
- struct = @tokens.at(index)
44
-
45
- expect(struct.type).to eq type
46
- expect(struct.token).to eq token
47
- expect(struct.text).to eq text
48
- expect(struct.ts).to eq ts
49
- expect(struct.te).to eq te
50
- expect(struct.level).to eq lvl
51
- expect(struct.set_level).to eq set_lvl
52
- expect(struct.conditional_level).to eq cond_lvl
53
- end
54
- end
55
- end
56
- end
57
-
58
- RSpec.shared_examples 'parse' do |pattern, checks|
59
- context "given the pattern #{pattern}" do
60
- before(:all) { @root = Regexp::Parser.parse(pattern, '*') }
61
-
62
- checks.each do |path, (type, token, klass, attributes)|
63
- it "parses expression at #{path} as #{klass}" do
64
- exp = @root.dig(*path)
65
-
66
- expect(exp).to be_instance_of(klass)
67
- expect(exp.type).to eq type
68
- expect(exp.token).to eq token
69
-
70
- attributes && attributes.each do |method, value|
71
- expect(exp.send(method)).to eq(value),
72
- "expected expression at #{path} to have #{method} #{value}"
73
- end
74
- end
75
- end
76
- end
77
- end
@@ -1,60 +0,0 @@
1
- require 'set'
2
- require 'delegate'
3
-
4
- module RegexpParserSpec
5
- class Warning
6
- class UnexpectedWarnings < StandardError
7
- MSG = 'Unexpected warnings: %s'.freeze
8
-
9
- def initialize(warnings)
10
- super(MSG % warnings.join("\n"))
11
- end
12
- end
13
-
14
- class Filter
15
- def initialize(whitelist)
16
- @whitelist = whitelist
17
- end
18
-
19
- def assert_expected_warnings_only
20
- original = $stderr
21
- $stderr = Extractor.new(original, @whitelist)
22
-
23
- yield
24
-
25
- assert_no_warnings($stderr.warnings)
26
- ensure
27
- $stderr = original
28
- end
29
-
30
- private
31
-
32
- def assert_no_warnings(warnings)
33
- raise UnexpectedWarnings, warnings.to_a if warnings.any?
34
- end
35
- end
36
-
37
- class Extractor < DelegateClass(IO)
38
- PATTERN = /\A(?:.+):(?:\d+): warning: (?:.+)\n\z/
39
-
40
- def initialize(io, whitelist)
41
- @whitelist = whitelist
42
- @warnings = Set.new
43
- super(io)
44
- end
45
-
46
- def write(message)
47
- return super if PATTERN !~ message
48
-
49
- warning = message.chomp
50
- @warnings << warning if @whitelist.none?(&warning.method(:include?))
51
-
52
- self
53
- end
54
-
55
- def warnings
56
- @warnings.dup.freeze
57
- end
58
- end
59
- end
60
- end
@@ -1,48 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe(Regexp::Syntax) do
4
- specify('unknown name') do
5
- expect { Regexp::Syntax.new('ruby/1.0') }.to raise_error(Regexp::Syntax::UnknownSyntaxNameError)
6
- end
7
-
8
- specify('new') do
9
- expect(Regexp::Syntax.new('ruby/1.9.3')).to be_instance_of(Regexp::Syntax::V1_9_3)
10
- end
11
-
12
- specify('new any') do
13
- expect(Regexp::Syntax.new('any')).to be_instance_of(Regexp::Syntax::Any)
14
- expect(Regexp::Syntax.new('*')).to be_instance_of(Regexp::Syntax::Any)
15
- end
16
-
17
- specify('not implemented') do
18
- expect { RP.parse('\\p{alpha}', 'ruby/1.8') }.to raise_error(Regexp::Syntax::NotImplementedError)
19
- end
20
-
21
- specify('supported?') do
22
- expect(Regexp::Syntax.supported?('ruby/1.1.1')).to be false
23
- expect(Regexp::Syntax.supported?('ruby/2.4.3')).to be true
24
- expect(Regexp::Syntax.supported?('ruby/2.5')).to be true
25
- end
26
-
27
- specify('invalid version') do
28
- expect { Regexp::Syntax.version_class('2.0.0') }.to raise_error(Regexp::Syntax::InvalidVersionNameError)
29
-
30
- expect { Regexp::Syntax.version_class('ruby/20') }.to raise_error(Regexp::Syntax::InvalidVersionNameError)
31
- end
32
-
33
- specify('version class tiny version') do
34
- expect(Regexp::Syntax.version_class('ruby/1.9.3')).to eq Regexp::Syntax::V1_9_3
35
-
36
- expect(Regexp::Syntax.version_class('ruby/2.3.1')).to eq Regexp::Syntax::V2_3_1
37
- end
38
-
39
- specify('version class minor version') do
40
- expect(Regexp::Syntax.version_class('ruby/1.9')).to eq Regexp::Syntax::V1_9
41
-
42
- expect(Regexp::Syntax.version_class('ruby/2.3')).to eq Regexp::Syntax::V2_3
43
- end
44
-
45
- specify('raises for unknown constant lookups') do
46
- expect { Regexp::Syntax::V1 }.to raise_error(/V1/)
47
- end
48
- end
@@ -1,23 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe(Regexp::Syntax::Token::Map) do
4
- let(:map) { Regexp::Syntax::Token::Map }
5
-
6
- specify('is complete') do
7
- latest_syntax = Regexp::Syntax.new('ruby/2.9')
8
-
9
- latest_syntax.features.each do |type, tokens|
10
- tokens.each { |token| expect(map[type]).to include(token) }
11
- end
12
- end
13
-
14
- specify('contains no duplicate type/token combinations') do
15
- combinations = map.flat_map do |type, tokens|
16
- tokens.map { |token| "#{type} #{token}" }
17
- end
18
-
19
- non_uniq = combinations.group_by { |str| str }.select { |_, v| v.count > 1 }
20
-
21
- expect(non_uniq.keys).to be_empty
22
- end
23
- end
@@ -1,17 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe(Regexp::Syntax::V1_8_6) do
4
- include_examples 'syntax', Regexp::Syntax.new('ruby/1.8.6'),
5
- implements: {
6
- assertion: T::Assertion::Lookahead,
7
- backref: [:number],
8
- escape: T::Escape::Basic + T::Escape::ASCII + T::Escape::Meta + T::Escape::Control,
9
- group: T::Group::V1_8_6,
10
- quantifier: T::Quantifier::Greedy + T::Quantifier::Reluctant + T::Quantifier::Interval + T::Quantifier::IntervalReluctant
11
- },
12
- excludes: {
13
- assertion: T::Assertion::Lookbehind,
14
- backref: T::Backreference::All - [:number] + T::SubexpressionCall::All,
15
- quantifier: T::Quantifier::Possessive
16
- }
17
- end
@@ -1,10 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe(Regexp::Syntax::V1_9_1) do
4
- include_examples 'syntax', Regexp::Syntax.new('ruby/1.9.1'),
5
- implements: {
6
- escape: T::Escape::Hex + T::Escape::Octal + T::Escape::Unicode,
7
- type: T::CharacterType::Hex,
8
- quantifier: T::Quantifier::Greedy + T::Quantifier::Reluctant + T::Quantifier::Possessive
9
- }
10
- end
@@ -1,9 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe(Regexp::Syntax::V1_9_3) do
4
- include_examples 'syntax', Regexp::Syntax.new('ruby/1.9.3'),
5
- implements: {
6
- property: T::UnicodeProperty::Script_V1_9_3 + T::UnicodeProperty::Age_V1_9_3,
7
- nonproperty: T::UnicodeProperty::Script_V1_9_3 + T::UnicodeProperty::Age_V1_9_3
8
- }
9
- end
@@ -1,13 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe(Regexp::Syntax::V2_0_0) do
4
- include_examples 'syntax', Regexp::Syntax.new('ruby/2.0.0'),
5
- implements: {
6
- property: T::UnicodeProperty::Age_V2_0_0,
7
- nonproperty: T::UnicodeProperty::Age_V2_0_0
8
- },
9
- excludes: {
10
- property: [:newline],
11
- nonproperty: [:newline]
12
- }
13
- end
@@ -1,9 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe(Regexp::Syntax::V2_2_0) do
4
- include_examples 'syntax', Regexp::Syntax.new('ruby/2.2.0'),
5
- implements: {
6
- property: T::UnicodeProperty::Script_V2_2_0 + T::UnicodeProperty::Age_V2_2_0,
7
- nonproperty: T::UnicodeProperty::Script_V2_2_0 + T::UnicodeProperty::Age_V2_2_0
8
- }
9
- end
@@ -1,37 +0,0 @@
1
- require 'spec_helper'
2
-
3
- RSpec.describe(Regexp::Syntax) do
4
- RSpec.shared_examples 'syntax alias' do |string, klass|
5
- it "aliases #{string} to #{klass}" do
6
- syntax = Regexp::Syntax.new(string)
7
- expect(syntax).to be_a klass
8
- end
9
- end
10
-
11
- include_examples 'syntax alias', 'ruby/1.8.6', Regexp::Syntax::V1_8_6
12
- include_examples 'syntax alias', 'ruby/1.8', Regexp::Syntax::V1_8_6
13
- include_examples 'syntax alias', 'ruby/1.9.1', Regexp::Syntax::V1_9_1
14
- include_examples 'syntax alias', 'ruby/1.9', Regexp::Syntax::V1_9_3
15
- include_examples 'syntax alias', 'ruby/2.0.0', Regexp::Syntax::V1_9
16
- include_examples 'syntax alias', 'ruby/2.0', Regexp::Syntax::V2_0_0
17
- include_examples 'syntax alias', 'ruby/2.1', Regexp::Syntax::V2_0_0
18
- include_examples 'syntax alias', 'ruby/2.2.0', Regexp::Syntax::V2_0_0
19
- include_examples 'syntax alias', 'ruby/2.2.10', Regexp::Syntax::V2_0_0
20
- include_examples 'syntax alias', 'ruby/2.2', Regexp::Syntax::V2_0_0
21
- include_examples 'syntax alias', 'ruby/2.3.0', Regexp::Syntax::V2_3_0
22
- include_examples 'syntax alias', 'ruby/2.3', Regexp::Syntax::V2_3_0
23
- include_examples 'syntax alias', 'ruby/2.4.0', Regexp::Syntax::V2_4_0
24
- include_examples 'syntax alias', 'ruby/2.4.1', Regexp::Syntax::V2_4_1
25
- include_examples 'syntax alias', 'ruby/2.5.0', Regexp::Syntax::V2_4_1
26
- include_examples 'syntax alias', 'ruby/2.5', Regexp::Syntax::V2_5_0
27
- include_examples 'syntax alias', 'ruby/2.6.0', Regexp::Syntax::V2_5_0
28
- include_examples 'syntax alias', 'ruby/2.6.2', Regexp::Syntax::V2_6_2
29
- include_examples 'syntax alias', 'ruby/2.6.3', Regexp::Syntax::V2_6_3
30
- include_examples 'syntax alias', 'ruby/2.6', Regexp::Syntax::V2_6_3
31
-
32
- specify('future alias warning') do
33
- expect { Regexp::Syntax.new('ruby/5.0') }
34
- .to output(/This library .* but you are running .* \(feature set of .*\)/)
35
- .to_stderr
36
- end
37
- end