regexp_parser 2.2.0 → 2.2.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +14 -2
- data/LICENSE +1 -1
- data/README.md +2 -2
- data/Rakefile +5 -8
- data/lib/regexp_parser/expression/classes/escape_sequence.rb +12 -7
- data/lib/regexp_parser/expression/methods/strfregexp.rb +1 -1
- data/lib/regexp_parser/scanner/properties/long.csv +604 -0
- data/lib/regexp_parser/scanner/properties/short.csv +242 -0
- data/lib/regexp_parser/scanner/scanner.rl +6 -4
- data/lib/regexp_parser/scanner.rb +126 -124
- data/lib/regexp_parser/syntax/base.rb +3 -5
- data/lib/regexp_parser/syntax/token/backreference.rb +7 -2
- data/lib/regexp_parser/syntax/versions/1.8.6.rb +2 -2
- data/lib/regexp_parser/syntax/versions/1.9.1.rb +1 -1
- data/lib/regexp_parser/version.rb +1 -1
- data/regexp_parser.gemspec +20 -22
- metadata +11 -143
- data/lib/regexp_parser/scanner/properties/long.yml +0 -607
- data/lib/regexp_parser/scanner/properties/short.yml +0 -245
- data/spec/expression/base_spec.rb +0 -104
- data/spec/expression/clone_spec.rb +0 -152
- data/spec/expression/conditional_spec.rb +0 -89
- data/spec/expression/free_space_spec.rb +0 -27
- data/spec/expression/methods/match_length_spec.rb +0 -161
- data/spec/expression/methods/match_spec.rb +0 -25
- data/spec/expression/methods/strfregexp_spec.rb +0 -224
- data/spec/expression/methods/tests_spec.rb +0 -99
- data/spec/expression/methods/traverse_spec.rb +0 -161
- data/spec/expression/options_spec.rb +0 -128
- data/spec/expression/subexpression_spec.rb +0 -50
- data/spec/expression/to_h_spec.rb +0 -26
- data/spec/expression/to_s_spec.rb +0 -108
- data/spec/lexer/all_spec.rb +0 -22
- data/spec/lexer/conditionals_spec.rb +0 -53
- data/spec/lexer/delimiters_spec.rb +0 -68
- data/spec/lexer/escapes_spec.rb +0 -14
- data/spec/lexer/keep_spec.rb +0 -10
- data/spec/lexer/literals_spec.rb +0 -64
- data/spec/lexer/nesting_spec.rb +0 -99
- data/spec/lexer/refcalls_spec.rb +0 -60
- data/spec/parser/all_spec.rb +0 -43
- data/spec/parser/alternation_spec.rb +0 -88
- data/spec/parser/anchors_spec.rb +0 -17
- data/spec/parser/conditionals_spec.rb +0 -179
- data/spec/parser/errors_spec.rb +0 -30
- data/spec/parser/escapes_spec.rb +0 -133
- data/spec/parser/free_space_spec.rb +0 -130
- data/spec/parser/groups_spec.rb +0 -108
- data/spec/parser/keep_spec.rb +0 -6
- data/spec/parser/options_spec.rb +0 -28
- data/spec/parser/posix_classes_spec.rb +0 -8
- data/spec/parser/properties_spec.rb +0 -117
- data/spec/parser/quantifiers_spec.rb +0 -68
- data/spec/parser/refcalls_spec.rb +0 -117
- data/spec/parser/set/intersections_spec.rb +0 -127
- data/spec/parser/set/ranges_spec.rb +0 -121
- data/spec/parser/sets_spec.rb +0 -178
- data/spec/parser/types_spec.rb +0 -18
- data/spec/scanner/all_spec.rb +0 -18
- data/spec/scanner/anchors_spec.rb +0 -21
- data/spec/scanner/conditionals_spec.rb +0 -128
- data/spec/scanner/delimiters_spec.rb +0 -52
- data/spec/scanner/errors_spec.rb +0 -67
- data/spec/scanner/escapes_spec.rb +0 -73
- data/spec/scanner/free_space_spec.rb +0 -165
- data/spec/scanner/groups_spec.rb +0 -61
- data/spec/scanner/keep_spec.rb +0 -10
- data/spec/scanner/literals_spec.rb +0 -39
- data/spec/scanner/meta_spec.rb +0 -18
- data/spec/scanner/options_spec.rb +0 -36
- data/spec/scanner/properties_spec.rb +0 -64
- data/spec/scanner/quantifiers_spec.rb +0 -25
- data/spec/scanner/refcalls_spec.rb +0 -55
- data/spec/scanner/sets_spec.rb +0 -151
- data/spec/scanner/types_spec.rb +0 -14
- data/spec/spec_helper.rb +0 -28
- data/spec/support/capturing_stderr.rb +0 -9
- data/spec/support/shared_examples.rb +0 -77
- 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 -38
- data/spec/token/token_spec.rb +0 -85
@@ -1,73 +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
|
-
# ineffectual literal escapes
|
15
|
-
# these cause "Unknown escape" warnings in Ruby for ascii chars,
|
16
|
-
# and simply drop the backslash for non-ascii chars (/\ü/.inspect == '/ü/').
|
17
|
-
# In terms of matching, Ruby treats them both like non-escaped literals.
|
18
|
-
include_examples 'scan', 'c\qt', 1 => [:escape, :literal, '\q', 1, 3]
|
19
|
-
include_examples 'scan', 'a\üc', 1 => [:escape, :literal, '\ü', 1, 3]
|
20
|
-
include_examples 'scan', 'a\😋c', 1 => [:escape, :literal, '\😋', 1, 3]
|
21
|
-
|
22
|
-
# these incomplete ref/call sequences are treated as literal escapes by Ruby
|
23
|
-
include_examples 'scan', 'c\gt', 1 => [:escape, :literal, '\g', 1, 3]
|
24
|
-
include_examples 'scan', 'c\kt', 1 => [:escape, :literal, '\k', 1, 3]
|
25
|
-
|
26
|
-
include_examples 'scan', 'a\012c', 1 => [:escape, :octal, '\012', 1, 5]
|
27
|
-
include_examples 'scan', 'a\0124', 1 => [:escape, :octal, '\012', 1, 5]
|
28
|
-
include_examples 'scan', '\712+7', 0 => [:escape, :octal, '\712', 0, 4]
|
29
|
-
|
30
|
-
include_examples 'scan', 'a\xA', 1 => [:escape, :hex, '\xA', 1, 4]
|
31
|
-
include_examples 'scan', 'a\x24c', 1 => [:escape, :hex, '\x24', 1, 5]
|
32
|
-
include_examples 'scan', 'a\x0640c', 1 => [:escape, :hex, '\x06', 1, 5]
|
33
|
-
|
34
|
-
include_examples 'scan', 'a\u0640c', 1 => [:escape, :codepoint, '\u0640', 1, 7]
|
35
|
-
include_examples 'scan', 'a\u{640 0641}c', 1 => [:escape, :codepoint_list, '\u{640 0641}', 1, 13]
|
36
|
-
include_examples 'scan', 'a\u{10FFFF}c', 1 => [:escape, :codepoint_list, '\u{10FFFF}', 1, 11]
|
37
|
-
|
38
|
-
include_examples 'scan', 'ab\\\xcd', 1 => [:escape, :backslash, '\\\\', 2, 4]
|
39
|
-
include_examples 'scan', 'ab\\\0cd', 1 => [:escape, :backslash, '\\\\', 2, 4]
|
40
|
-
include_examples 'scan', 'ab\\\Kcd', 1 => [:escape, :backslash, '\\\\', 2, 4]
|
41
|
-
|
42
|
-
include_examples 'scan', 'ab\^cd', 1 => [:escape, :bol, '\^', 2, 4]
|
43
|
-
include_examples 'scan', 'ab\$cd', 1 => [:escape, :eol, '\$', 2, 4]
|
44
|
-
include_examples 'scan', 'ab\[cd', 1 => [:escape, :set_open, '\[', 2, 4]
|
45
|
-
|
46
|
-
# Meta/control espaces
|
47
|
-
#
|
48
|
-
# After the following fix in Ruby 3.1, a Regexp#source containing meta/control
|
49
|
-
# escapes can only be set with the Regexp::new constructor.
|
50
|
-
# In Regexp literals, these escapes are now pre-processed to hex escapes.
|
51
|
-
#
|
52
|
-
# https://github.com/ruby/ruby/commit/11ae581a4a7f5d5f5ec6378872eab8f25381b1b9
|
53
|
-
n = ->(regexp_body){ Regexp.new(regexp_body.force_encoding('ascii-8bit'), 'n') }
|
54
|
-
|
55
|
-
include_examples 'scan', 'a\cBc', 1 => [:escape, :control, '\cB', 1, 4]
|
56
|
-
include_examples 'scan', 'a\c^c', 1 => [:escape, :control, '\c^', 1, 4]
|
57
|
-
include_examples 'scan', 'a\c\n', 1 => [:escape, :control, '\c\n', 1, 5]
|
58
|
-
include_examples 'scan', 'a\c\\\\b', 1 => [:escape, :control, '\c\\\\', 1, 5]
|
59
|
-
include_examples 'scan', 'a\C-bc', 1 => [:escape, :control, '\C-b', 1, 5]
|
60
|
-
include_examples 'scan', 'a\C-^b', 1 => [:escape, :control, '\C-^', 1, 5]
|
61
|
-
include_examples 'scan', 'a\C-\nb', 1 => [:escape, :control, '\C-\n', 1, 6]
|
62
|
-
include_examples 'scan', 'a\C-\\\\b', 1 => [:escape, :control, '\C-\\\\', 1, 6]
|
63
|
-
include_examples 'scan', n.('a\c\M-Bc'), 1 => [:escape, :control, '\c\M-B', 1, 7]
|
64
|
-
include_examples 'scan', n.('a\C-\M-Bc'), 1 => [:escape, :control, '\C-\M-B', 1, 8]
|
65
|
-
|
66
|
-
include_examples 'scan', n.('a\M-Bc'), 1 => [:escape, :meta_sequence, '\M-B', 1, 5]
|
67
|
-
include_examples 'scan', n.('a\M-\cBc'), 1 => [:escape, :meta_sequence, '\M-\cB', 1, 7]
|
68
|
-
include_examples 'scan', n.('a\M-\c^'), 1 => [:escape, :meta_sequence, '\M-\c^', 1, 7]
|
69
|
-
include_examples 'scan', n.('a\M-\c\n'), 1 => [:escape, :meta_sequence, '\M-\c\n', 1, 8]
|
70
|
-
include_examples 'scan', n.('a\M-\c\\\\'), 1 => [:escape, :meta_sequence, '\M-\c\\\\', 1, 8]
|
71
|
-
include_examples 'scan', n.('a\M-\C-Bc'), 1 => [:escape, :meta_sequence, '\M-\C-B', 1, 8]
|
72
|
-
include_examples 'scan', n.('a\M-\C-\\\\'), 1 => [:escape, :meta_sequence, '\M-\C-\\\\', 1, 9]
|
73
|
-
end
|
@@ -1,165 +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
|
-
|
43
|
-
# single line / no trailing newline (c.f. issue #66)
|
44
|
-
include_examples 'scan', /a # b/x,
|
45
|
-
0 => [:literal, :literal, 'a', 0, 1],
|
46
|
-
1 => [:free_space, :whitespace, ' ', 1, 2],
|
47
|
-
2 => [:free_space, :comment, "# b", 2, 5]
|
48
|
-
|
49
|
-
# without spaces (c.f. issue #66)
|
50
|
-
include_examples 'scan', /a#b/x,
|
51
|
-
0 => [:literal, :literal, 'a', 0, 1],
|
52
|
-
1 => [:free_space, :comment, "#b", 1, 3]
|
53
|
-
end
|
54
|
-
|
55
|
-
describe('scan free space inlined') do
|
56
|
-
include_examples 'scan', /a b(?x:c d e)f g/,
|
57
|
-
0 => [:literal, :literal, 'a b', 0, 3],
|
58
|
-
1 => [:group, :options, '(?x:', 3, 7],
|
59
|
-
2 => [:literal, :literal, 'c', 7, 8],
|
60
|
-
3 => [:free_space, :whitespace, ' ', 8, 9],
|
61
|
-
4 => [:literal, :literal, 'd', 9, 10],
|
62
|
-
5 => [:free_space, :whitespace, ' ', 10, 11],
|
63
|
-
6 => [:literal, :literal, 'e', 11, 12],
|
64
|
-
7 => [:group, :close, ')', 12, 13],
|
65
|
-
8 => [:literal, :literal, 'f g', 13, 16]
|
66
|
-
end
|
67
|
-
|
68
|
-
describe('scan free space nested') do
|
69
|
-
include_examples 'scan', /a b(?x:c d(?-x:e f)g h)i j/,
|
70
|
-
0 => [:literal, :literal, 'a b', 0, 3],
|
71
|
-
1 => [:group, :options, '(?x:', 3, 7],
|
72
|
-
2 => [:literal, :literal, 'c', 7, 8],
|
73
|
-
3 => [:free_space, :whitespace, ' ', 8, 9],
|
74
|
-
4 => [:literal, :literal, 'd', 9, 10],
|
75
|
-
5 => [:group, :options, '(?-x:', 10, 15],
|
76
|
-
6 => [:literal, :literal, 'e f', 15, 18],
|
77
|
-
7 => [:group, :close, ')', 18, 19],
|
78
|
-
8 => [:literal, :literal, 'g', 19, 20],
|
79
|
-
9 => [:free_space, :whitespace, ' ', 20, 21],
|
80
|
-
10 => [:literal, :literal, 'h', 21, 22],
|
81
|
-
11 => [:group, :close, ')', 22, 23],
|
82
|
-
12 => [:literal, :literal, 'i j', 23, 26]
|
83
|
-
end
|
84
|
-
|
85
|
-
describe('scan free space nested groups') do
|
86
|
-
include_examples 'scan', /(a (b(?x: (c d) (?-x:(e f) )g) h)i j)/,
|
87
|
-
0 => [:group, :capture, '(', 0, 1],
|
88
|
-
1 => [:literal, :literal, 'a ', 1, 3],
|
89
|
-
2 => [:group, :capture, '(', 3, 4],
|
90
|
-
3 => [:literal, :literal, 'b', 4, 5],
|
91
|
-
4 => [:group, :options, '(?x:', 5, 9],
|
92
|
-
5 => [:free_space, :whitespace, ' ', 9, 10],
|
93
|
-
6 => [:group, :capture, '(', 10, 11],
|
94
|
-
7 => [:literal, :literal, 'c', 11, 12],
|
95
|
-
8 => [:free_space, :whitespace, ' ', 12, 13],
|
96
|
-
9 => [:literal, :literal, 'd', 13, 14],
|
97
|
-
10 => [:group, :close, ')', 14, 15],
|
98
|
-
11 => [:free_space, :whitespace, ' ', 15, 16],
|
99
|
-
12 => [:group, :options, '(?-x:', 16, 21],
|
100
|
-
13 => [:group, :capture, '(', 21, 22],
|
101
|
-
14 => [:literal, :literal, 'e f', 22, 25],
|
102
|
-
15 => [:group, :close, ')', 25, 26],
|
103
|
-
16 => [:literal, :literal, ' ', 26, 27],
|
104
|
-
17 => [:group, :close, ')', 27, 28],
|
105
|
-
18 => [:literal, :literal, 'g', 28, 29],
|
106
|
-
19 => [:group, :close, ')', 29, 30],
|
107
|
-
20 => [:literal, :literal, ' h', 30, 32],
|
108
|
-
21 => [:group, :close, ')', 32, 33],
|
109
|
-
22 => [:literal, :literal, 'i j', 33, 36],
|
110
|
-
23 => [:group, :close, ')', 36, 37]
|
111
|
-
end
|
112
|
-
|
113
|
-
describe('scan free space switch groups') do
|
114
|
-
include_examples 'scan', /(a (b((?x) (c d) ((?-x)(e f) )g) h)i j)/,
|
115
|
-
0 => [:group, :capture, '(', 0, 1],
|
116
|
-
1 => [:literal, :literal, 'a ', 1, 3],
|
117
|
-
2 => [:group, :capture, '(', 3, 4],
|
118
|
-
3 => [:literal, :literal, 'b', 4, 5],
|
119
|
-
4 => [:group, :capture, '(', 5, 6],
|
120
|
-
5 => [:group, :options_switch, '(?x', 6, 9],
|
121
|
-
6 => [:group, :close, ')', 9, 10],
|
122
|
-
7 => [:free_space, :whitespace, ' ', 10, 11],
|
123
|
-
8 => [:group, :capture, '(', 11, 12],
|
124
|
-
9 => [:literal, :literal, 'c', 12, 13],
|
125
|
-
10 => [:free_space, :whitespace, ' ', 13, 14],
|
126
|
-
11 => [:literal, :literal, 'd', 14, 15],
|
127
|
-
12 => [:group, :close, ')', 15, 16],
|
128
|
-
13 => [:free_space, :whitespace, ' ', 16, 17],
|
129
|
-
14 => [:group, :capture, '(', 17, 18],
|
130
|
-
15 => [:group, :options_switch, '(?-x', 18, 22],
|
131
|
-
16 => [:group, :close, ')', 22, 23],
|
132
|
-
17 => [:group, :capture, '(', 23, 24],
|
133
|
-
18 => [:literal, :literal, 'e f', 24, 27],
|
134
|
-
19 => [:group, :close, ')', 27, 28],
|
135
|
-
20 => [:literal, :literal, ' ', 28, 29],
|
136
|
-
21 => [:group, :close, ')', 29, 30],
|
137
|
-
22 => [:literal, :literal, 'g', 30, 31],
|
138
|
-
23 => [:group, :close, ')', 31, 32],
|
139
|
-
24 => [:literal, :literal, ' h', 32, 34],
|
140
|
-
25 => [:group, :close, ')', 34, 35],
|
141
|
-
26 => [:literal, :literal, 'i j', 35, 38],
|
142
|
-
27 => [:group, :close, ')', 38, 39]
|
143
|
-
end
|
144
|
-
|
145
|
-
describe('scanning `#` in regular (non-x mode)') do # c.f. issue 70
|
146
|
-
include_examples 'scan', /a#bcd/,
|
147
|
-
0 => [:literal, :literal, 'a#bcd', 0, 5]
|
148
|
-
include_examples 'scan', /a # bcd/,
|
149
|
-
0 => [:literal, :literal, 'a # bcd', 0, 7]
|
150
|
-
|
151
|
-
include_examples 'scan', /a#\d/,
|
152
|
-
0 => [:literal, :literal, 'a#', 0, 2],
|
153
|
-
1 => [:type, :digit, '\d', 2, 4]
|
154
|
-
include_examples 'scan', /a # \d/,
|
155
|
-
0 => [:literal, :literal, 'a # ', 0, 4],
|
156
|
-
1 => [:type, :digit, '\d', 4, 6]
|
157
|
-
|
158
|
-
include_examples 'scan', /a#()/,
|
159
|
-
0 => [:literal, :literal, 'a#', 0, 2],
|
160
|
-
1 => [:group, :capture, '(', 2, 3]
|
161
|
-
include_examples 'scan', /a # ()/,
|
162
|
-
0 => [:literal, :literal, 'a # ', 0, 4],
|
163
|
-
1 => [:group, :capture, '(', 4, 5]
|
164
|
-
end
|
165
|
-
end
|
data/spec/scanner/groups_spec.rb
DELETED
@@ -1,61 +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
|
-
# Named groups
|
9
|
-
# only names that start with a hyphen or digit (ascii or other) are invalid
|
10
|
-
include_examples 'scan', '(?<name>abc)', 0 => [:group, :named_ab, '(?<name>', 0, 8]
|
11
|
-
include_examples 'scan', "(?'name'abc)", 0 => [:group, :named_sq, "(?'name'", 0, 8]
|
12
|
-
include_examples 'scan', '(?<name_1>abc)', 0 => [:group, :named_ab, '(?<name_1>', 0,10]
|
13
|
-
include_examples 'scan', "(?'name_1'abc)", 0 => [:group, :named_sq, "(?'name_1'", 0,10]
|
14
|
-
include_examples 'scan', '(?<name-1>abc)', 0 => [:group, :named_ab, '(?<name-1>', 0,10]
|
15
|
-
include_examples 'scan', "(?'name-1'abc)", 0 => [:group, :named_sq, "(?'name-1'", 0,10]
|
16
|
-
include_examples 'scan', "(?<name'1>abc)", 0 => [:group, :named_ab, "(?<name'1>", 0,10]
|
17
|
-
include_examples 'scan', "(?'name>1'abc)", 0 => [:group, :named_sq, "(?'name>1'", 0,10]
|
18
|
-
include_examples 'scan', '(?<üüuuüü>abc)', 0 => [:group, :named_ab, '(?<üüuuüü>', 0,10]
|
19
|
-
include_examples 'scan', "(?'üüuuüü'abc)", 0 => [:group, :named_sq, "(?'üüuuüü'", 0,10]
|
20
|
-
include_examples 'scan', "(?<😋1234😋>abc)", 0 => [:group, :named_ab, "(?<😋1234😋>", 0,10]
|
21
|
-
include_examples 'scan', "(?'😋1234😋'abc)", 0 => [:group, :named_sq, "(?'😋1234😋'", 0,10]
|
22
|
-
|
23
|
-
include_examples 'scan', '(?:abc)', 0 => [:group, :passive, '(?:', 0, 3]
|
24
|
-
include_examples 'scan', '(?:)', 0 => [:group, :passive, '(?:', 0, 3]
|
25
|
-
include_examples 'scan', '(?::)', 0 => [:group, :passive, '(?:', 0, 3]
|
26
|
-
|
27
|
-
# Comments
|
28
|
-
include_examples 'scan', '(?#abc)', 0 => [:group, :comment, '(?#abc)', 0, 7]
|
29
|
-
include_examples 'scan', '(?#)', 0 => [:group, :comment, '(?#)', 0, 4]
|
30
|
-
|
31
|
-
# Assertions
|
32
|
-
include_examples 'scan', '(?=abc)', 0 => [:assertion, :lookahead, '(?=', 0, 3]
|
33
|
-
include_examples 'scan', '(?!abc)', 0 => [:assertion, :nlookahead, '(?!', 0, 3]
|
34
|
-
include_examples 'scan', '(?<=abc)', 0 => [:assertion, :lookbehind, '(?<=', 0, 4]
|
35
|
-
include_examples 'scan', '(?<!abc)', 0 => [:assertion, :nlookbehind, '(?<!', 0, 4]
|
36
|
-
|
37
|
-
# Options
|
38
|
-
include_examples 'scan', '(?-mix:abc)', 0 => [:group, :options, '(?-mix:', 0, 7]
|
39
|
-
include_examples 'scan', '(?m-ix:abc)', 0 => [:group, :options, '(?m-ix:', 0, 7]
|
40
|
-
include_examples 'scan', '(?mi-x:abc)', 0 => [:group, :options, '(?mi-x:', 0, 7]
|
41
|
-
include_examples 'scan', '(?mix:abc)', 0 => [:group, :options, '(?mix:', 0, 6]
|
42
|
-
include_examples 'scan', '(?m:)', 0 => [:group, :options, '(?m:', 0, 4]
|
43
|
-
include_examples 'scan', '(?i:)', 0 => [:group, :options, '(?i:', 0, 4]
|
44
|
-
include_examples 'scan', '(?x:)', 0 => [:group, :options, '(?x:', 0, 4]
|
45
|
-
include_examples 'scan', '(?mix)', 0 => [:group, :options_switch, '(?mix', 0, 5]
|
46
|
-
include_examples 'scan', '(?d-mix:abc)', 0 => [:group, :options, '(?d-mix:', 0, 8]
|
47
|
-
include_examples 'scan', '(?a-mix:abc)', 0 => [:group, :options, '(?a-mix:', 0, 8]
|
48
|
-
include_examples 'scan', '(?u-mix:abc)', 0 => [:group, :options, '(?u-mix:', 0, 8]
|
49
|
-
include_examples 'scan', '(?da-m:abc)', 0 => [:group, :options, '(?da-m:', 0, 7]
|
50
|
-
include_examples 'scan', '(?du-x:abc)', 0 => [:group, :options, '(?du-x:', 0, 7]
|
51
|
-
include_examples 'scan', '(?dau-i:abc)', 0 => [:group, :options, '(?dau-i:', 0, 8]
|
52
|
-
include_examples 'scan', '(?dau:abc)', 0 => [:group, :options, '(?dau:', 0, 6]
|
53
|
-
include_examples 'scan', '(?d:)', 0 => [:group, :options, '(?d:', 0, 4]
|
54
|
-
include_examples 'scan', '(?a:)', 0 => [:group, :options, '(?a:', 0, 4]
|
55
|
-
include_examples 'scan', '(?u:)', 0 => [:group, :options, '(?u:', 0, 4]
|
56
|
-
include_examples 'scan', '(?dau)', 0 => [:group, :options_switch, '(?dau', 0, 5]
|
57
|
-
|
58
|
-
if ruby_version_at_least('2.4.1')
|
59
|
-
include_examples 'scan', '(?~abc)', 0 => [:group, :absence, '(?~', 0, 3]
|
60
|
-
end
|
61
|
-
end
|
data/spec/scanner/keep_spec.rb
DELETED
@@ -1,39 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe('UTF8 scanning') do
|
4
|
-
# ascii, single byte characters
|
5
|
-
include_examples 'scan', 'a',
|
6
|
-
0 => [:literal, :literal, 'a', 0, 1]
|
7
|
-
|
8
|
-
include_examples 'scan', 'ab+',
|
9
|
-
0 => [:literal, :literal, 'ab', 0, 2],
|
10
|
-
1 => [:quantifier, :one_or_more, '+', 2, 3]
|
11
|
-
|
12
|
-
# 2 byte wide characters
|
13
|
-
include_examples 'scan', 'äöü',
|
14
|
-
0 => [:literal, :literal, 'äöü', 0, 3]
|
15
|
-
|
16
|
-
# 3 byte wide characters, Japanese
|
17
|
-
include_examples 'scan', 'ab?れます+cd',
|
18
|
-
0 => [:literal, :literal, 'ab', 0, 2],
|
19
|
-
1 => [:quantifier, :zero_or_one, '?', 2, 3],
|
20
|
-
2 => [:literal, :literal, 'れます', 3, 6],
|
21
|
-
3 => [:quantifier, :one_or_more, '+', 6, 7],
|
22
|
-
4 => [:literal, :literal, 'cd', 7, 9]
|
23
|
-
|
24
|
-
# 4 byte wide characters, Osmanya
|
25
|
-
include_examples 'scan', '𐒀𐒁?𐒂ab+𐒃',
|
26
|
-
0 => [:literal, :literal, '𐒀𐒁', 0, 2],
|
27
|
-
1 => [:quantifier, :zero_or_one, '?', 2, 3],
|
28
|
-
2 => [:literal, :literal, '𐒂ab', 3, 6],
|
29
|
-
3 => [:quantifier, :one_or_more, '+', 6, 7],
|
30
|
-
4 => [:literal, :literal, '𐒃', 7, 8]
|
31
|
-
|
32
|
-
include_examples 'scan', 'mu𝄞?si*𝄫c+',
|
33
|
-
0 => [:literal, :literal, 'mu𝄞', 0, 3],
|
34
|
-
1 => [:quantifier, :zero_or_one, '?', 3, 4],
|
35
|
-
2 => [:literal, :literal, 'si', 4, 6],
|
36
|
-
3 => [:quantifier, :zero_or_more, '*', 6, 7],
|
37
|
-
4 => [:literal, :literal, '𝄫c', 7, 9],
|
38
|
-
5 => [:quantifier, :one_or_more, '+', 9, 10]
|
39
|
-
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
|
@@ -1,36 +0,0 @@
|
|
1
|
-
require 'spec_helper'
|
2
|
-
|
3
|
-
RSpec.describe('passing options to scan') do
|
4
|
-
def expect_type_tokens(tokens, type_tokens)
|
5
|
-
expect(tokens.map { |type, token, *| [type, token] }).to eq(type_tokens)
|
6
|
-
end
|
7
|
-
|
8
|
-
it 'raises if if scanning from a Regexp and options are passed' do
|
9
|
-
expect { RS.scan(/a+/, options: ::Regexp::EXTENDED) }.to raise_error(
|
10
|
-
ArgumentError,
|
11
|
-
'options cannot be supplied unless scanning a String'
|
12
|
-
)
|
13
|
-
end
|
14
|
-
|
15
|
-
it 'sets free_spacing based on options if scanning from a String' do
|
16
|
-
expect_type_tokens(
|
17
|
-
RS.scan('a+#c', options: ::Regexp::MULTILINE | ::Regexp::EXTENDED),
|
18
|
-
[
|
19
|
-
%i[literal literal],
|
20
|
-
%i[quantifier one_or_more],
|
21
|
-
%i[free_space comment]
|
22
|
-
]
|
23
|
-
)
|
24
|
-
end
|
25
|
-
|
26
|
-
it 'does not set free_spacing if scanning from a String and passing no options' do
|
27
|
-
expect_type_tokens(
|
28
|
-
RS.scan('a+#c'),
|
29
|
-
[
|
30
|
-
%i[literal literal],
|
31
|
-
%i[quantifier one_or_more],
|
32
|
-
%i[literal literal]
|
33
|
-
]
|
34
|
-
)
|
35
|
-
end
|
36
|
-
end
|
@@ -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
|
@@ -1,25 +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
|
-
|
21
|
-
# special case: chained quantifiers
|
22
|
-
include_examples 'scan', 'a+{2}{3}', 1 => [:quantifier, :one_or_more, '+', 1, 2]
|
23
|
-
include_examples 'scan', 'a+{2}{3}', 2 => [:quantifier, :interval, '{2}', 2, 5]
|
24
|
-
include_examples 'scan', 'a+{2}{3}', 3 => [:quantifier, :interval, '{3}', 5, 8]
|
25
|
-
end
|
@@ -1,55 +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
|
-
#
|
9
|
-
# NOTE: only \g supports forward-looking references using '+', e.g. \g<+1>
|
10
|
-
# refers to the next group, but \k<+1> refers to a group named '+1'.
|
11
|
-
# Inversely, only \k supports addition or substraction of a recursion level.
|
12
|
-
# E.g. \k<x+0> refers to a group named 'x' at the current recursion level,
|
13
|
-
# but \g<x+0> refers to a a group named 'x+0'.
|
14
|
-
#
|
15
|
-
include_examples 'scan', '(?<X>abc)\k<X>', 3 => [:backref, :name_ref_ab, '\k<X>', 9, 14]
|
16
|
-
include_examples 'scan', "(?<X>abc)\\k'X'", 3 => [:backref, :name_ref_sq, "\\k'X'", 9, 14]
|
17
|
-
|
18
|
-
include_examples 'scan', '(?<+1>abc)\k<+1>', 3 => [:backref, :name_ref_ab, '\k<+1>', 10, 16]
|
19
|
-
include_examples 'scan', "(?<+1>abc)\\k'+1'", 3 => [:backref, :name_ref_sq, "\\k'+1'", 10, 16]
|
20
|
-
|
21
|
-
include_examples 'scan', '(abc)\k<1>', 3 => [:backref, :number_ref_ab, '\k<1>', 5, 10]
|
22
|
-
include_examples 'scan', "(abc)\\k'1'", 3 => [:backref, :number_ref_sq, "\\k'1'", 5, 10]
|
23
|
-
|
24
|
-
include_examples 'scan', '(abc)\k<-1>', 3 => [:backref, :number_rel_ref_ab, '\k<-1>', 5, 11]
|
25
|
-
include_examples 'scan', "(abc)\\k'-1'", 3 => [:backref, :number_rel_ref_sq, "\\k'-1'", 5, 11]
|
26
|
-
|
27
|
-
# Sub-expression invocation, named, numbered, and relative
|
28
|
-
include_examples 'scan', '(?<X>abc)\g<X>', 3 => [:backref, :name_call_ab, '\g<X>', 9, 14]
|
29
|
-
include_examples 'scan', "(?<X>abc)\\g'X'", 3 => [:backref, :name_call_sq, "\\g'X'", 9, 14]
|
30
|
-
|
31
|
-
include_examples 'scan', '(?<X>abc)\g<X-1>', 3 => [:backref, :name_call_ab, '\g<X-1>', 9, 16]
|
32
|
-
include_examples 'scan', "(?<X>abc)\\g'X-1'", 3 => [:backref, :name_call_sq, "\\g'X-1'", 9, 16]
|
33
|
-
|
34
|
-
include_examples 'scan', '(abc)\g<1>', 3 => [:backref, :number_call_ab, '\g<1>', 5, 10]
|
35
|
-
include_examples 'scan', "(abc)\\g'1'", 3 => [:backref, :number_call_sq, "\\g'1'", 5, 10]
|
36
|
-
|
37
|
-
include_examples 'scan', 'a(b|\g<0>)', 4 => [:backref, :number_call_ab, '\g<0>', 4, 9]
|
38
|
-
include_examples 'scan', "a(b|\\g'0')", 4 => [:backref, :number_call_sq, "\\g'0'", 4, 9]
|
39
|
-
|
40
|
-
include_examples 'scan', '(abc)\g<-1>', 3 => [:backref, :number_rel_call_ab, '\g<-1>', 5, 11]
|
41
|
-
include_examples 'scan', "(abc)\\g'-1'", 3 => [:backref, :number_rel_call_sq, "\\g'-1'", 5, 11]
|
42
|
-
|
43
|
-
include_examples 'scan', '\g<+1>(abc)', 0 => [:backref, :number_rel_call_ab, '\g<+1>', 0, 6]
|
44
|
-
include_examples 'scan', "\\g'+1'(abc)", 0 => [:backref, :number_rel_call_sq, "\\g'+1'", 0, 6]
|
45
|
-
|
46
|
-
# Group back-references, with recursion level
|
47
|
-
include_examples 'scan', '(?<X>abc)\k<X-0>', 3 => [:backref, :name_recursion_ref_ab, '\k<X-0>', 9, 16]
|
48
|
-
include_examples 'scan', "(?<X>abc)\\k'X-0'", 3 => [:backref, :name_recursion_ref_sq, "\\k'X-0'", 9, 16]
|
49
|
-
|
50
|
-
include_examples 'scan', '(abc)\k<1-0>', 3 => [:backref, :number_recursion_ref_ab, '\k<1-0>', 5, 12]
|
51
|
-
include_examples 'scan', "(abc)\\k'1-0'", 3 => [:backref, :number_recursion_ref_sq, "\\k'1-0'", 5, 12]
|
52
|
-
|
53
|
-
include_examples 'scan', '(abc)\k<+1-0>', 3 => [:backref, :name_recursion_ref_ab, '\k<+1-0>', 5, 13]
|
54
|
-
include_examples 'scan', "(abc)\\k'+1-0'", 3 => [:backref, :name_recursion_ref_sq, "\\k'+1-0'", 5, 13]
|
55
|
-
end
|