re2 2.15.0.rc1-x86-linux-gnu
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.rspec +3 -0
- data/Gemfile +11 -0
- data/LICENSE-DEPENDENCIES.txt +237 -0
- data/LICENSE.txt +28 -0
- data/README.md +396 -0
- data/Rakefile +94 -0
- data/dependencies.yml +7 -0
- data/ext/re2/extconf.rb +332 -0
- data/ext/re2/re2.cc +2254 -0
- data/ext/re2/recipes.rb +54 -0
- data/lib/3.1/re2.so +0 -0
- data/lib/3.2/re2.so +0 -0
- data/lib/3.3/re2.so +0 -0
- data/lib/3.4/re2.so +0 -0
- data/lib/re2/regexp.rb +72 -0
- data/lib/re2/scanner.rb +26 -0
- data/lib/re2/string.rb +38 -0
- data/lib/re2/version.rb +14 -0
- data/lib/re2.rb +20 -0
- data/re2.gemspec +47 -0
- data/spec/kernel_spec.rb +37 -0
- data/spec/re2/match_data_spec.rb +411 -0
- data/spec/re2/regexp_spec.rb +911 -0
- data/spec/re2/scanner_spec.rb +275 -0
- data/spec/re2/set_spec.rb +231 -0
- data/spec/re2/string_spec.rb +62 -0
- data/spec/re2_spec.rb +201 -0
- data/spec/spec_helper.rb +31 -0
- metadata +129 -0
@@ -0,0 +1,911 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
RSpec.describe RE2::Regexp do
|
4
|
+
describe "#initialize" do
|
5
|
+
it "returns an instance given only a pattern" do
|
6
|
+
re = RE2::Regexp.new('woo')
|
7
|
+
|
8
|
+
expect(re).to be_a(RE2::Regexp)
|
9
|
+
end
|
10
|
+
|
11
|
+
it "returns an instance given a pattern and options" do
|
12
|
+
re = RE2::Regexp.new('woo', case_sensitive: false)
|
13
|
+
|
14
|
+
expect(re).to be_a(RE2::Regexp)
|
15
|
+
end
|
16
|
+
|
17
|
+
it "accepts patterns containing null bytes" do
|
18
|
+
re = RE2::Regexp.new("a\0b")
|
19
|
+
|
20
|
+
expect(re.pattern).to eq("a\0b")
|
21
|
+
end
|
22
|
+
|
23
|
+
it "raises an error if given an inappropriate type" do
|
24
|
+
expect { RE2::Regexp.new(nil) }.to raise_error(TypeError)
|
25
|
+
end
|
26
|
+
|
27
|
+
it "allows invalid patterns to be created" do
|
28
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
29
|
+
|
30
|
+
expect(re).to be_a(RE2::Regexp)
|
31
|
+
end
|
32
|
+
|
33
|
+
it "supports passing something that can be coerced to a String as input" do
|
34
|
+
re = RE2::Regexp.new(StringLike.new('w(o)(o)'))
|
35
|
+
|
36
|
+
expect(re).to be_a(RE2::Regexp)
|
37
|
+
end
|
38
|
+
end
|
39
|
+
|
40
|
+
describe ".compile" do
|
41
|
+
it "returns an instance given only a pattern" do
|
42
|
+
re = RE2::Regexp.compile('woo')
|
43
|
+
|
44
|
+
expect(re).to be_a(RE2::Regexp)
|
45
|
+
end
|
46
|
+
|
47
|
+
it "returns an instance given a pattern and options" do
|
48
|
+
re = RE2::Regexp.compile('woo', case_sensitive: false)
|
49
|
+
expect(re).to be_a(RE2::Regexp)
|
50
|
+
end
|
51
|
+
|
52
|
+
it "accepts patterns containing null bytes" do
|
53
|
+
re = RE2::Regexp.compile("a\0b")
|
54
|
+
|
55
|
+
expect(re.pattern).to eq("a\0b")
|
56
|
+
end
|
57
|
+
|
58
|
+
it "raises an error if given an inappropriate type" do
|
59
|
+
expect { RE2::Regexp.compile(nil) }.to raise_error(TypeError)
|
60
|
+
end
|
61
|
+
|
62
|
+
it "allows invalid patterns to be created" do
|
63
|
+
re = RE2::Regexp.compile('???', log_errors: false)
|
64
|
+
|
65
|
+
expect(re).to be_a(RE2::Regexp)
|
66
|
+
end
|
67
|
+
|
68
|
+
it "supports passing something that can be coerced to a String as input" do
|
69
|
+
re = RE2::Regexp.compile(StringLike.new('w(o)(o)'))
|
70
|
+
|
71
|
+
expect(re).to be_a(RE2::Regexp)
|
72
|
+
end
|
73
|
+
end
|
74
|
+
|
75
|
+
describe "#options" do
|
76
|
+
it "returns a hash of options" do
|
77
|
+
options = RE2::Regexp.new('woo').options
|
78
|
+
expect(options).to be_a(Hash)
|
79
|
+
end
|
80
|
+
|
81
|
+
it "is populated with default options when nothing has been set" do
|
82
|
+
expect(RE2::Regexp.new('woo').options).to include(
|
83
|
+
utf8: true,
|
84
|
+
posix_syntax: false,
|
85
|
+
longest_match: false,
|
86
|
+
log_errors: true,
|
87
|
+
literal: false,
|
88
|
+
never_nl: false,
|
89
|
+
case_sensitive: true,
|
90
|
+
perl_classes: false,
|
91
|
+
word_boundary: false,
|
92
|
+
one_line: false
|
93
|
+
)
|
94
|
+
end
|
95
|
+
|
96
|
+
it "is populated with overridden options when specified" do
|
97
|
+
options = RE2::Regexp.new('woo', case_sensitive: false).options
|
98
|
+
|
99
|
+
expect(options).to include(case_sensitive: false)
|
100
|
+
end
|
101
|
+
end
|
102
|
+
|
103
|
+
describe "#error" do
|
104
|
+
it "returns nil if there is no error" do
|
105
|
+
error = RE2::Regexp.new('woo').error
|
106
|
+
|
107
|
+
expect(error).to be_nil
|
108
|
+
end
|
109
|
+
|
110
|
+
# Use log_errors: false to suppress RE2's logging to STDERR.
|
111
|
+
it "contains the error string if there is an error" do
|
112
|
+
error = RE2::Regexp.new('wo(o', log_errors: false).error
|
113
|
+
|
114
|
+
expect(error).to eq("missing ): wo(o")
|
115
|
+
end
|
116
|
+
end
|
117
|
+
|
118
|
+
describe "#error_arg" do
|
119
|
+
it "returns nil if there is no error" do
|
120
|
+
error_arg = RE2::Regexp.new('woo').error_arg
|
121
|
+
|
122
|
+
expect(error_arg).to be_nil
|
123
|
+
end
|
124
|
+
|
125
|
+
it "returns the offending portion of the pattern if there is an error" do
|
126
|
+
error_arg = RE2::Regexp.new('wo(o', log_errors: false).error_arg
|
127
|
+
|
128
|
+
expect(error_arg).to eq("wo(o")
|
129
|
+
end
|
130
|
+
end
|
131
|
+
|
132
|
+
describe "#program_size" do
|
133
|
+
it "returns a numeric value" do
|
134
|
+
program_size = RE2::Regexp.new('w(o)(o)').program_size
|
135
|
+
|
136
|
+
expect(program_size).to be_an(Integer)
|
137
|
+
end
|
138
|
+
|
139
|
+
it "returns -1 for an invalid pattern" do
|
140
|
+
program_size = RE2::Regexp.new('???', log_errors: false).program_size
|
141
|
+
|
142
|
+
expect(program_size).to eq(-1)
|
143
|
+
end
|
144
|
+
end
|
145
|
+
|
146
|
+
describe "#to_str" do
|
147
|
+
it "returns the original pattern" do
|
148
|
+
string = RE2::Regexp.new('w(o)(o)').to_str
|
149
|
+
|
150
|
+
expect(string).to eq("w(o)(o)")
|
151
|
+
end
|
152
|
+
|
153
|
+
it "returns the pattern even if invalid" do
|
154
|
+
string = RE2::Regexp.new('???', log_errors: false).to_str
|
155
|
+
|
156
|
+
expect(string).to eq("???")
|
157
|
+
end
|
158
|
+
end
|
159
|
+
|
160
|
+
describe "#pattern" do
|
161
|
+
it "returns the original pattern" do
|
162
|
+
pattern = RE2::Regexp.new('w(o)(o)').pattern
|
163
|
+
|
164
|
+
expect(pattern).to eq("w(o)(o)")
|
165
|
+
end
|
166
|
+
|
167
|
+
it "returns the pattern even if invalid" do
|
168
|
+
pattern = RE2::Regexp.new('???', log_errors: false).pattern
|
169
|
+
|
170
|
+
expect(pattern).to eq("???")
|
171
|
+
end
|
172
|
+
end
|
173
|
+
|
174
|
+
describe "#inspect" do
|
175
|
+
it "shows the class name and original pattern" do
|
176
|
+
string = RE2::Regexp.new('w(o)(o)').inspect
|
177
|
+
|
178
|
+
expect(string).to eq("#<RE2::Regexp /w(o)(o)/>")
|
179
|
+
end
|
180
|
+
|
181
|
+
it "respects the pattern's original encoding" do
|
182
|
+
string = RE2::Regexp.new('w(o)(o)', utf8: false).inspect
|
183
|
+
|
184
|
+
expect(string.encoding).to eq(Encoding::ISO_8859_1)
|
185
|
+
end
|
186
|
+
end
|
187
|
+
|
188
|
+
describe "#utf8?" do
|
189
|
+
it "returns true by default" do
|
190
|
+
expect(RE2::Regexp.new('woo')).to be_utf8
|
191
|
+
end
|
192
|
+
|
193
|
+
it "can be overridden on initialization" do
|
194
|
+
re = RE2::Regexp.new('woo', utf8: false)
|
195
|
+
|
196
|
+
expect(re).to_not be_utf8
|
197
|
+
end
|
198
|
+
end
|
199
|
+
|
200
|
+
describe "#posix_syntax?" do
|
201
|
+
it "returns false by default" do
|
202
|
+
expect(RE2::Regexp.new('woo')).to_not be_posix_syntax
|
203
|
+
end
|
204
|
+
|
205
|
+
it "can be overridden on initialization" do
|
206
|
+
re = RE2::Regexp.new('woo', posix_syntax: true)
|
207
|
+
|
208
|
+
expect(re).to be_posix_syntax
|
209
|
+
end
|
210
|
+
end
|
211
|
+
|
212
|
+
describe "#literal?" do
|
213
|
+
it "returns false by default" do
|
214
|
+
expect(RE2::Regexp.new('woo')).to_not be_literal
|
215
|
+
end
|
216
|
+
|
217
|
+
it "can be overridden on initialization" do
|
218
|
+
re = RE2::Regexp.new('woo', literal: true)
|
219
|
+
|
220
|
+
expect(re).to be_literal
|
221
|
+
end
|
222
|
+
end
|
223
|
+
|
224
|
+
describe "#never_nl?" do
|
225
|
+
it "returns false by default" do
|
226
|
+
expect(RE2::Regexp.new('woo')).to_not be_never_nl
|
227
|
+
end
|
228
|
+
|
229
|
+
it "can be overridden on initialization" do
|
230
|
+
re = RE2::Regexp.new('woo', never_nl: true)
|
231
|
+
|
232
|
+
expect(re).to be_never_nl
|
233
|
+
end
|
234
|
+
end
|
235
|
+
|
236
|
+
describe "#case_sensitive?" do
|
237
|
+
it "returns true by default" do
|
238
|
+
expect(RE2::Regexp.new('woo')).to be_case_sensitive
|
239
|
+
end
|
240
|
+
|
241
|
+
it "can be overridden on initialization" do
|
242
|
+
re = RE2::Regexp.new('woo', case_sensitive: false)
|
243
|
+
expect(re).to_not be_case_sensitive
|
244
|
+
end
|
245
|
+
end
|
246
|
+
|
247
|
+
describe "#case_insensitive?" do
|
248
|
+
it "returns false by default" do
|
249
|
+
expect(RE2::Regexp.new('woo')).to_not be_case_insensitive
|
250
|
+
end
|
251
|
+
|
252
|
+
it "can be overridden on initialization" do
|
253
|
+
re = RE2::Regexp.new('woo', case_sensitive: false)
|
254
|
+
|
255
|
+
expect(re).to be_case_insensitive
|
256
|
+
end
|
257
|
+
end
|
258
|
+
|
259
|
+
describe "#casefold?" do
|
260
|
+
it "returns true by default" do
|
261
|
+
expect(RE2::Regexp.new('woo')).to_not be_casefold
|
262
|
+
end
|
263
|
+
|
264
|
+
it "can be overridden on initialization" do
|
265
|
+
re = RE2::Regexp.new('woo', case_sensitive: false)
|
266
|
+
|
267
|
+
expect(re).to be_casefold
|
268
|
+
end
|
269
|
+
end
|
270
|
+
|
271
|
+
describe "#longest_match?" do
|
272
|
+
it "returns false by default" do
|
273
|
+
expect(RE2::Regexp.new('woo')).to_not be_casefold
|
274
|
+
end
|
275
|
+
|
276
|
+
it "can be overridden on initialization" do
|
277
|
+
re = RE2::Regexp.new('woo', longest_match: true)
|
278
|
+
|
279
|
+
expect(re).to be_longest_match
|
280
|
+
end
|
281
|
+
end
|
282
|
+
|
283
|
+
describe "#log_errors?" do
|
284
|
+
it "returns true by default" do
|
285
|
+
expect(RE2::Regexp.new('woo')).to be_log_errors
|
286
|
+
end
|
287
|
+
|
288
|
+
it "can be overridden on initialization" do
|
289
|
+
re = RE2::Regexp.new('woo', log_errors: false)
|
290
|
+
|
291
|
+
expect(re).to_not be_log_errors
|
292
|
+
end
|
293
|
+
end
|
294
|
+
|
295
|
+
describe "#perl_classes?" do
|
296
|
+
it "returns false by default" do
|
297
|
+
expect(RE2::Regexp.new('woo')).to_not be_perl_classes
|
298
|
+
end
|
299
|
+
|
300
|
+
it "can be overridden on initialization" do
|
301
|
+
re = RE2::Regexp.new('woo', perl_classes: true)
|
302
|
+
|
303
|
+
expect(re).to be_perl_classes
|
304
|
+
end
|
305
|
+
end
|
306
|
+
|
307
|
+
describe "#word_boundary?" do
|
308
|
+
it "returns false by default" do
|
309
|
+
expect(RE2::Regexp.new('woo')).to_not be_word_boundary
|
310
|
+
end
|
311
|
+
|
312
|
+
it "can be overridden on initialization" do
|
313
|
+
re = RE2::Regexp.new('woo', word_boundary: true)
|
314
|
+
|
315
|
+
expect(re).to be_word_boundary
|
316
|
+
end
|
317
|
+
end
|
318
|
+
|
319
|
+
describe "#one_line?" do
|
320
|
+
it "returns false by default" do
|
321
|
+
expect(RE2::Regexp.new('woo')).to_not be_one_line
|
322
|
+
end
|
323
|
+
|
324
|
+
it "can be overridden on initialization" do
|
325
|
+
re = RE2::Regexp.new('woo', one_line: true)
|
326
|
+
|
327
|
+
expect(re).to be_one_line
|
328
|
+
end
|
329
|
+
end
|
330
|
+
|
331
|
+
describe "#max_mem" do
|
332
|
+
it "returns the default max memory" do
|
333
|
+
expect(RE2::Regexp.new('woo').max_mem).to eq(8388608)
|
334
|
+
end
|
335
|
+
|
336
|
+
it "can be overridden on initialization" do
|
337
|
+
re = RE2::Regexp.new('woo', max_mem: 1024)
|
338
|
+
|
339
|
+
expect(re.max_mem).to eq(1024)
|
340
|
+
end
|
341
|
+
end
|
342
|
+
|
343
|
+
describe "#match" do
|
344
|
+
it "returns match data given only text if the pattern has capturing groups" do
|
345
|
+
re = RE2::Regexp.new('My name is (\w+) (\w+)')
|
346
|
+
|
347
|
+
expect(re.match("My name is Alice Bloggs")).to be_a(RE2::MatchData)
|
348
|
+
end
|
349
|
+
|
350
|
+
it "returns only true or false given only text if the pattern has no capturing groups" do
|
351
|
+
re = RE2::Regexp.new('My name is \w+ \w+')
|
352
|
+
|
353
|
+
expect(re.match("My name is Alice Bloggs")).to eq(true)
|
354
|
+
end
|
355
|
+
|
356
|
+
it "supports matching against text containing null bytes" do
|
357
|
+
re = RE2::Regexp.new("a\0b")
|
358
|
+
|
359
|
+
expect(re.match("a\0b")).to eq(true)
|
360
|
+
end
|
361
|
+
|
362
|
+
it "returns nil if the text does not match the pattern" do
|
363
|
+
re = RE2::Regexp.new('My name is (\w+) (\w+)')
|
364
|
+
|
365
|
+
expect(re.match("My age is 99")).to be_nil
|
366
|
+
end
|
367
|
+
|
368
|
+
it "accepts text that can be coerced to a string" do
|
369
|
+
re = RE2::Regexp.new('My name is (\w+) (\w+)')
|
370
|
+
|
371
|
+
expect(re.match(StringLike.new("My name is Alice Bloggs"))).to be_a(RE2::MatchData)
|
372
|
+
end
|
373
|
+
|
374
|
+
it "raises an exception when given text that cannot be coerced to a string" do
|
375
|
+
re = RE2::Regexp.new('My name is (\w+) (\w+)')
|
376
|
+
|
377
|
+
expect { re.match(nil) }.to raise_error(TypeError)
|
378
|
+
end
|
379
|
+
|
380
|
+
it "returns nil with an invalid pattern" do
|
381
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
382
|
+
|
383
|
+
expect(re.match("My name is Alice Bloggs")).to be_nil
|
384
|
+
end
|
385
|
+
|
386
|
+
it "returns nil with an invalid pattern and options" do
|
387
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
388
|
+
|
389
|
+
expect(re.match('foo bar', startpos: 1)).to be_nil
|
390
|
+
end
|
391
|
+
|
392
|
+
it "accepts an offset at which to start matching", :aggregate_failures do
|
393
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
394
|
+
md = re.match("one two three", startpos: 4)
|
395
|
+
|
396
|
+
expect(md[1]).to eq("two")
|
397
|
+
expect(md[2]).to eq("three")
|
398
|
+
end
|
399
|
+
|
400
|
+
it "returns nil if using a starting offset past the end of the text" do
|
401
|
+
skip "Underlying RE2::Match does not have endpos argument" unless RE2::Regexp.match_has_endpos_argument?
|
402
|
+
|
403
|
+
re = RE2::Regexp.new('(\w+) (\w+)', log_errors: false)
|
404
|
+
|
405
|
+
expect(re.match("one two three", startpos: 20, endpos: 21)).to be_nil
|
406
|
+
end
|
407
|
+
|
408
|
+
it "raises an exception when given a negative starting offset" do
|
409
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
410
|
+
|
411
|
+
expect { re.match("one two three", startpos: -1) }.to raise_error(ArgumentError, "startpos should be >= 0")
|
412
|
+
end
|
413
|
+
|
414
|
+
it "raises an exception when given a starting offset past the default ending offset" do
|
415
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
416
|
+
|
417
|
+
expect { re.match("one two three", startpos: 30) }.to raise_error(ArgumentError, "startpos should be <= endpos")
|
418
|
+
end
|
419
|
+
|
420
|
+
it "accepts an offset at which to end matching", :aggregate_failures do
|
421
|
+
skip "Underlying RE2::Match does not have endpos argument" unless RE2::Regexp.match_has_endpos_argument?
|
422
|
+
|
423
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
424
|
+
md = re.match("one two three", endpos: 6)
|
425
|
+
|
426
|
+
expect(md[1]).to eq("one")
|
427
|
+
expect(md[2]).to eq("tw")
|
428
|
+
end
|
429
|
+
|
430
|
+
it "returns nil if using a ending offset at the start of the text" do
|
431
|
+
skip "Underlying RE2::Match does not have endpos argument" unless RE2::Regexp.match_has_endpos_argument?
|
432
|
+
|
433
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
434
|
+
|
435
|
+
expect(re.match("one two three", endpos: 0)).to be_nil
|
436
|
+
end
|
437
|
+
|
438
|
+
it "raises an exception when given a negative ending offset" do
|
439
|
+
skip "Underlying RE2::Match does not have endpos argument" unless RE2::Regexp.match_has_endpos_argument?
|
440
|
+
|
441
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
442
|
+
|
443
|
+
expect { re.match("one two three", endpos: -1) }.to raise_error(ArgumentError, "endpos should be >= 0")
|
444
|
+
end
|
445
|
+
|
446
|
+
it "raises an exception when given an ending offset before the starting offset" do
|
447
|
+
skip "Underlying RE2::Match does not have endpos argument" unless RE2::Regexp.match_has_endpos_argument?
|
448
|
+
|
449
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
450
|
+
|
451
|
+
expect { re.match("one two three", startpos: 3, endpos: 0) }.to raise_error(ArgumentError, "startpos should be <= endpos")
|
452
|
+
end
|
453
|
+
|
454
|
+
it "raises an error if given an ending offset and RE2 does not support it" do
|
455
|
+
skip "Underlying RE2::Match has endpos argument" if RE2::Regexp.match_has_endpos_argument?
|
456
|
+
|
457
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
458
|
+
|
459
|
+
expect { re.match("one two three", endpos: 3) }.to raise_error(RE2::Regexp::UnsupportedError)
|
460
|
+
end
|
461
|
+
|
462
|
+
it "does not anchor matches by default when extracting submatches" do
|
463
|
+
re = RE2::Regexp.new('(two)')
|
464
|
+
|
465
|
+
expect(re.match("one two three")).to be_a(RE2::MatchData)
|
466
|
+
end
|
467
|
+
|
468
|
+
it "does not anchor matches by default without extracting submatches" do
|
469
|
+
re = RE2::Regexp.new('(two)')
|
470
|
+
|
471
|
+
expect(re.match("one two three", submatches: 0)).to eq(true)
|
472
|
+
end
|
473
|
+
|
474
|
+
it "can explicitly match without anchoring when extracting submatches" do
|
475
|
+
re = RE2::Regexp.new('(two)')
|
476
|
+
|
477
|
+
expect(re.match("one two three", anchor: :unanchored)).to be_a(RE2::MatchData)
|
478
|
+
end
|
479
|
+
|
480
|
+
it "can explicitly match with neither anchoring nor extracting submatches" do
|
481
|
+
re = RE2::Regexp.new('(two)')
|
482
|
+
|
483
|
+
expect(re.match("one two three", anchor: :unanchored, submatches: 0)).to eq(true)
|
484
|
+
end
|
485
|
+
|
486
|
+
it "can anchor matches at the start when extracting submatches", :aggregate_failures do
|
487
|
+
re = RE2::Regexp.new('(two)')
|
488
|
+
|
489
|
+
expect(re.match("two three", anchor: :anchor_start)).to be_a(RE2::MatchData)
|
490
|
+
expect(re.match("one two three", anchor: :anchor_start)).to be_nil
|
491
|
+
end
|
492
|
+
|
493
|
+
it "can anchor matches at the start without extracting submatches", :aggregate_failures do
|
494
|
+
re = RE2::Regexp.new('(two)')
|
495
|
+
|
496
|
+
expect(re.match("two three", anchor: :anchor_start, submatches: 0)).to eq(true)
|
497
|
+
expect(re.match("one two three", anchor: :anchor_start, submatches: 0)).to eq(false)
|
498
|
+
end
|
499
|
+
|
500
|
+
it "can anchor matches at both ends when extracting submatches", :aggregate_failures do
|
501
|
+
re = RE2::Regexp.new('(two)')
|
502
|
+
|
503
|
+
expect(re.match("two three", anchor: :anchor_both)).to be_nil
|
504
|
+
expect(re.match("two", anchor: :anchor_both)).to be_a(RE2::MatchData)
|
505
|
+
end
|
506
|
+
|
507
|
+
it "does not anchor matches when given a nil anchor" do
|
508
|
+
re = RE2::Regexp.new('(two)')
|
509
|
+
|
510
|
+
expect(re.match("one two three", anchor: nil)).to be_a(RE2::MatchData)
|
511
|
+
end
|
512
|
+
|
513
|
+
it "raises an exception when given an invalid anchor" do
|
514
|
+
re = RE2::Regexp.new('(two)')
|
515
|
+
|
516
|
+
expect { re.match("one two three", anchor: :invalid) }.to raise_error(ArgumentError, "anchor should be one of: :unanchored, :anchor_start, :anchor_both")
|
517
|
+
end
|
518
|
+
|
519
|
+
it "raises an exception when given a non-symbol anchor" do
|
520
|
+
re = RE2::Regexp.new('(two)')
|
521
|
+
|
522
|
+
expect { re.match("one two three", anchor: 0) }.to raise_error(TypeError)
|
523
|
+
end
|
524
|
+
|
525
|
+
it "extracts all submatches by default", :aggregate_failures do
|
526
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
527
|
+
md = re.match("one two three")
|
528
|
+
|
529
|
+
expect(md[1]).to eq("one")
|
530
|
+
expect(md[2]).to eq("two")
|
531
|
+
expect(md[3]).to eq("three")
|
532
|
+
end
|
533
|
+
|
534
|
+
it "supports extracting submatches containing null bytes" do
|
535
|
+
re = RE2::Regexp.new("(a\0b)")
|
536
|
+
md = re.match("a\0bc")
|
537
|
+
|
538
|
+
expect(md[1]).to eq("a\0b")
|
539
|
+
end
|
540
|
+
|
541
|
+
it "extracts a specific number of submatches", :aggregate_failures do
|
542
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
543
|
+
md = re.match("one two three", submatches: 2)
|
544
|
+
|
545
|
+
expect(md[1]).to eq("one")
|
546
|
+
expect(md[2]).to eq("two")
|
547
|
+
expect(md[3]).to be_nil
|
548
|
+
end
|
549
|
+
|
550
|
+
it "pads submatches with nil when requesting more than the number of capturing groups" do
|
551
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
552
|
+
md = re.match("one two three", submatches: 5)
|
553
|
+
|
554
|
+
expect(md.to_a).to eq(["one two three", "one", "two", "three", nil, nil])
|
555
|
+
end
|
556
|
+
|
557
|
+
it "raises an exception when given a negative number of submatches" do
|
558
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
559
|
+
|
560
|
+
expect { re.match("one two three", submatches: -1) }.to raise_error(ArgumentError, "number of matches should be >= 0")
|
561
|
+
end
|
562
|
+
|
563
|
+
it "raises an exception when given a non-numeric number of submatches" do
|
564
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
565
|
+
|
566
|
+
expect { re.match("one two three", submatches: :invalid) }.to raise_error(TypeError)
|
567
|
+
end
|
568
|
+
|
569
|
+
it "defaults to extracting all submatches when given nil", :aggregate_failures do
|
570
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
571
|
+
md = re.match("one two three", submatches: nil)
|
572
|
+
|
573
|
+
expect(md[1]).to eq("one")
|
574
|
+
expect(md[2]).to eq("two")
|
575
|
+
expect(md[3]).to eq("three")
|
576
|
+
end
|
577
|
+
|
578
|
+
it "accepts passing the number of submatches instead of options for backward compatibility", :aggregate_failures do
|
579
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
580
|
+
md = re.match("one two three", 2)
|
581
|
+
|
582
|
+
expect(md[1]).to eq("one")
|
583
|
+
expect(md[2]).to eq("two")
|
584
|
+
expect(md[3]).to be_nil
|
585
|
+
end
|
586
|
+
|
587
|
+
it "raises an exception when given invalid options" do
|
588
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
589
|
+
|
590
|
+
expect { re.match("one two three", :invalid) }.to raise_error(TypeError)
|
591
|
+
end
|
592
|
+
|
593
|
+
it "accepts anything that can be coerced to a hash as options", :aggregate_failures do
|
594
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
595
|
+
|
596
|
+
expect(re.match("one two three", nil)).to be_a(RE2::MatchData)
|
597
|
+
end
|
598
|
+
end
|
599
|
+
|
600
|
+
describe "#match?" do
|
601
|
+
it "returns only true or false even if there are capturing groups", :aggregate_failures do
|
602
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
603
|
+
|
604
|
+
expect(re.match?("My name is Alice Bloggs")).to eq(true)
|
605
|
+
expect(re.match?("My age is 99")).to eq(false)
|
606
|
+
end
|
607
|
+
|
608
|
+
it "returns false if the pattern is invalid" do
|
609
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
610
|
+
|
611
|
+
expect(re.match?("My name is Alice Bloggs")).to eq(false)
|
612
|
+
end
|
613
|
+
|
614
|
+
it "raises an exception if text cannot be coerced to a string" do
|
615
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
616
|
+
|
617
|
+
expect { re.match?(0) }.to raise_error(TypeError)
|
618
|
+
end
|
619
|
+
end
|
620
|
+
|
621
|
+
describe "#partial_match?" do
|
622
|
+
it "returns only true or false even if there are capturing groups", :aggregate_failures do
|
623
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
624
|
+
|
625
|
+
expect(re.partial_match?("My name is Alice Bloggs")).to eq(true)
|
626
|
+
expect(re.partial_match?("My age is 99")).to eq(false)
|
627
|
+
end
|
628
|
+
|
629
|
+
it "supports matching against text containing null bytes", :aggregate_failures do
|
630
|
+
re = RE2::Regexp.new("a\0b")
|
631
|
+
|
632
|
+
expect(re.partial_match?("a\0b")).to eq(true)
|
633
|
+
expect(re.partial_match?("ab")).to eq(false)
|
634
|
+
end
|
635
|
+
|
636
|
+
it "returns false if the pattern is invalid" do
|
637
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
638
|
+
|
639
|
+
expect(re.partial_match?("My name is Alice Bloggs")).to eq(false)
|
640
|
+
end
|
641
|
+
|
642
|
+
it "raises an exception if text cannot be coerced to a string" do
|
643
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
644
|
+
|
645
|
+
expect { re.partial_match?(0) }.to raise_error(TypeError)
|
646
|
+
end
|
647
|
+
end
|
648
|
+
|
649
|
+
describe "#=~" do
|
650
|
+
it "returns only true or false even if there are capturing groups", :aggregate_failures do
|
651
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
652
|
+
|
653
|
+
expect(re =~ "My name is Alice Bloggs").to eq(true)
|
654
|
+
expect(re =~ "My age is 99").to eq(false)
|
655
|
+
end
|
656
|
+
|
657
|
+
it "supports matching against text containing null bytes", :aggregate_failures do
|
658
|
+
re = RE2::Regexp.new("a\0b")
|
659
|
+
|
660
|
+
expect(re =~ "a\0b").to eq(true)
|
661
|
+
expect(re =~ "ab").to eq(false)
|
662
|
+
end
|
663
|
+
|
664
|
+
it "returns false if the pattern is invalid" do
|
665
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
666
|
+
|
667
|
+
expect(re =~ "My name is Alice Bloggs").to eq(false)
|
668
|
+
end
|
669
|
+
|
670
|
+
it "raises an exception if text cannot be coerced to a string" do
|
671
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
672
|
+
|
673
|
+
expect { re =~ 0 }.to raise_error(TypeError)
|
674
|
+
end
|
675
|
+
end
|
676
|
+
|
677
|
+
describe "#===" do
|
678
|
+
it "returns only true or false even if there are capturing groups", :aggregate_failures do
|
679
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
680
|
+
|
681
|
+
expect(re === "My name is Alice Bloggs").to eq(true)
|
682
|
+
expect(re === "My age is 99").to eq(false)
|
683
|
+
end
|
684
|
+
|
685
|
+
it "returns false if the pattern is invalid" do
|
686
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
687
|
+
|
688
|
+
expect(re === "My name is Alice Bloggs").to eq(false)
|
689
|
+
end
|
690
|
+
|
691
|
+
it "raises an exception if text cannot be coerced to a string" do
|
692
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
693
|
+
|
694
|
+
expect { re === 0 }.to raise_error(TypeError)
|
695
|
+
end
|
696
|
+
end
|
697
|
+
|
698
|
+
describe "#full_match?" do
|
699
|
+
it "returns only true or false even if there are capturing groups", :aggregate_failures do
|
700
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
701
|
+
|
702
|
+
expect(re.full_match?("My name is Alice Bloggs")).to eq(true)
|
703
|
+
expect(re.full_match?("My name is Alice Bloggs and I am 99")).to eq(false)
|
704
|
+
end
|
705
|
+
|
706
|
+
it "supports matching against text containing null bytes", :aggregate_failures do
|
707
|
+
re = RE2::Regexp.new("a\0b")
|
708
|
+
|
709
|
+
expect(re.full_match?("a\0b")).to eq(true)
|
710
|
+
expect(re.full_match?("a\0bc")).to eq(false)
|
711
|
+
end
|
712
|
+
|
713
|
+
it "returns false if the pattern is invalid" do
|
714
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
715
|
+
|
716
|
+
expect(re.full_match?("My name is Alice Bloggs")).to eq(false)
|
717
|
+
end
|
718
|
+
|
719
|
+
it "raises an exception if text cannot be coerced to a string" do
|
720
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
721
|
+
|
722
|
+
expect { re.full_match?(0) }.to raise_error(TypeError)
|
723
|
+
end
|
724
|
+
end
|
725
|
+
|
726
|
+
describe "#ok?" do
|
727
|
+
it "returns true for valid patterns", :aggregate_failures do
|
728
|
+
expect(RE2::Regexp.new('woo')).to be_ok
|
729
|
+
expect(RE2::Regexp.new('wo(o)')).to be_ok
|
730
|
+
expect(RE2::Regexp.new('((\d)\w+){3,}')).to be_ok
|
731
|
+
end
|
732
|
+
|
733
|
+
it "returns false for invalid patterns", :aggregate_failures do
|
734
|
+
expect(RE2::Regexp.new('wo(o', log_errors: false)).to_not be_ok
|
735
|
+
expect(RE2::Regexp.new('wo[o', log_errors: false)).to_not be_ok
|
736
|
+
expect(RE2::Regexp.new('*', log_errors: false)).to_not be_ok
|
737
|
+
end
|
738
|
+
end
|
739
|
+
|
740
|
+
describe ".escape" do
|
741
|
+
it "transforms a string into a regexp" do
|
742
|
+
expect(RE2::Regexp.escape("1.5-2.0?")).to eq('1\.5\-2\.0\?')
|
743
|
+
end
|
744
|
+
end
|
745
|
+
|
746
|
+
describe ".quote" do
|
747
|
+
it "transforms a string into a regexp" do
|
748
|
+
expect(RE2::Regexp.quote("1.5-2.0?")).to eq('1\.5\-2\.0\?')
|
749
|
+
end
|
750
|
+
end
|
751
|
+
|
752
|
+
describe "#number_of_capturing_groups" do
|
753
|
+
it "returns the number of groups in a pattern", :aggregate_failures do
|
754
|
+
expect(RE2::Regexp.new('(a)(b)(c)').number_of_capturing_groups).to eq(3)
|
755
|
+
expect(RE2::Regexp.new('abc').number_of_capturing_groups).to eq(0)
|
756
|
+
expect(RE2::Regexp.new('a((b)c)').number_of_capturing_groups).to eq(2)
|
757
|
+
end
|
758
|
+
|
759
|
+
it "returns -1 for an invalid pattern" do
|
760
|
+
expect(RE2::Regexp.new('???', log_errors: false).number_of_capturing_groups).to eq(-1)
|
761
|
+
end
|
762
|
+
end
|
763
|
+
|
764
|
+
describe "#named_capturing_groups" do
|
765
|
+
it "returns a hash of names to indices" do
|
766
|
+
expect(RE2::Regexp.new('(?P<bob>a)').named_capturing_groups).to be_a(Hash)
|
767
|
+
end
|
768
|
+
|
769
|
+
it "maps names to indices with only one group" do
|
770
|
+
groups = RE2::Regexp.new('(?P<bob>a)').named_capturing_groups
|
771
|
+
|
772
|
+
expect(groups).to eq("bob" => 1)
|
773
|
+
end
|
774
|
+
|
775
|
+
it "maps names to indices with several groups" do
|
776
|
+
groups = RE2::Regexp.new('(?P<bob>a)(o)(?P<rob>e)').named_capturing_groups
|
777
|
+
|
778
|
+
expect(groups).to eq("bob" => 1, "rob" => 3)
|
779
|
+
end
|
780
|
+
|
781
|
+
it "returns an empty hash for an invalid regexp" do
|
782
|
+
expect(RE2::Regexp.new('???', log_errors: false).named_capturing_groups).to be_empty
|
783
|
+
end
|
784
|
+
end
|
785
|
+
|
786
|
+
describe "#scan" do
|
787
|
+
it "returns a scanner" do
|
788
|
+
r = RE2::Regexp.new('(\w+)')
|
789
|
+
scanner = r.scan("It is a truth universally acknowledged")
|
790
|
+
|
791
|
+
expect(scanner).to be_a(RE2::Scanner)
|
792
|
+
end
|
793
|
+
|
794
|
+
it "raises a type error if given invalid input" do
|
795
|
+
r = RE2::Regexp.new('(\w+)')
|
796
|
+
|
797
|
+
expect { r.scan(nil) }.to raise_error(TypeError)
|
798
|
+
end
|
799
|
+
end
|
800
|
+
|
801
|
+
describe "#partial_match" do
|
802
|
+
it "matches the pattern anywhere within the given text" do
|
803
|
+
r = RE2::Regexp.new('f(o+)')
|
804
|
+
|
805
|
+
expect(r.partial_match("foo bar")).to be_a(RE2::MatchData)
|
806
|
+
end
|
807
|
+
|
808
|
+
it "returns true or false if there are no capturing groups" do
|
809
|
+
r = RE2::Regexp.new('fo+')
|
810
|
+
|
811
|
+
expect(r.partial_match("foo bar")).to eq(true)
|
812
|
+
end
|
813
|
+
|
814
|
+
it "can set the number of submatches to extract", :aggregate_failures do
|
815
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
816
|
+
m = r.partial_match("fooaa bar", submatches: 1)
|
817
|
+
|
818
|
+
expect(m[1]).to eq("oo")
|
819
|
+
expect(m[2]).to be_nil
|
820
|
+
|
821
|
+
m = r.partial_match("fooaa bar", submatches: 2)
|
822
|
+
|
823
|
+
expect(m[1]).to eq("oo")
|
824
|
+
expect(m[2]).to eq("aa")
|
825
|
+
end
|
826
|
+
|
827
|
+
it "raises an error if given non-hash options" do
|
828
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
829
|
+
|
830
|
+
expect { r.partial_match("fooaa bar", "not a hash") }.to raise_error(TypeError)
|
831
|
+
end
|
832
|
+
|
833
|
+
it "accepts options that can be coerced to a hash", :aggregate_failures do
|
834
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
835
|
+
|
836
|
+
m = r.partial_match("fooaa bar", nil)
|
837
|
+
expect(m[1]).to eq('oo')
|
838
|
+
|
839
|
+
m = r.partial_match("fooaa bar", [])
|
840
|
+
expect(m[1]).to eq('oo')
|
841
|
+
end
|
842
|
+
|
843
|
+
it "accepts anything that can be coerced to a string" do
|
844
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
845
|
+
|
846
|
+
expect(r.partial_match(StringLike.new("fooaa bar"))).to be_a(RE2::MatchData)
|
847
|
+
end
|
848
|
+
|
849
|
+
it "does not allow the anchor to be overridden" do
|
850
|
+
r = RE2::Regexp.new('(\d+)')
|
851
|
+
|
852
|
+
expect(r.partial_match('ruby:1234', anchor: :anchor_both)).to be_a(RE2::MatchData)
|
853
|
+
end
|
854
|
+
end
|
855
|
+
|
856
|
+
describe "#full_match" do
|
857
|
+
it "only matches the pattern if all of the given text matches", :aggregate_failures do
|
858
|
+
r = RE2::Regexp.new('f(o+)')
|
859
|
+
|
860
|
+
expect(r.full_match("foo")).to be_a(RE2::MatchData)
|
861
|
+
expect(r.full_match("foo bar")).to be_nil
|
862
|
+
end
|
863
|
+
|
864
|
+
it "returns true or false if there are no capturing groups" do
|
865
|
+
r = RE2::Regexp.new('fo+')
|
866
|
+
|
867
|
+
expect(r.full_match("foo")).to eq(true)
|
868
|
+
end
|
869
|
+
|
870
|
+
it "can set the number of submatches to extract", :aggregate_failures do
|
871
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
872
|
+
m = r.full_match("fooaa", submatches: 1)
|
873
|
+
|
874
|
+
expect(m[1]).to eq("oo")
|
875
|
+
expect(m[2]).to be_nil
|
876
|
+
|
877
|
+
m = r.full_match("fooaa", submatches: 2)
|
878
|
+
|
879
|
+
expect(m[1]).to eq("oo")
|
880
|
+
expect(m[2]).to eq("aa")
|
881
|
+
end
|
882
|
+
|
883
|
+
it "raises an error if given non-hash options" do
|
884
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
885
|
+
|
886
|
+
expect { r.full_match("fooaa", "not a hash") }.to raise_error(TypeError)
|
887
|
+
end
|
888
|
+
|
889
|
+
it "accepts options that can be coerced to a hash", :aggregate_failures do
|
890
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
891
|
+
|
892
|
+
m = r.full_match("fooaa", nil)
|
893
|
+
expect(m[1]).to eq("oo")
|
894
|
+
|
895
|
+
m = r.full_match("fooaa", [])
|
896
|
+
expect(m[1]).to eq("oo")
|
897
|
+
end
|
898
|
+
|
899
|
+
it "accepts anything that can be coerced to a string" do
|
900
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
901
|
+
|
902
|
+
expect(r.full_match(StringLike.new("fooaa"), submatches: 0)).to eq(true)
|
903
|
+
end
|
904
|
+
|
905
|
+
it "does not allow the anchor to be overridden" do
|
906
|
+
r = RE2::Regexp.new('(\d+)')
|
907
|
+
|
908
|
+
expect(r.full_match('ruby:1234', anchor: :unanchored)).to be_nil
|
909
|
+
end
|
910
|
+
end
|
911
|
+
end
|