re2 2.4.3-x86-mingw32 → 2.6.0.rc1-x86-mingw32
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +249 -192
- data/Rakefile +1 -1
- data/ext/re2/extconf.rb +6 -70
- data/ext/re2/re2.cc +450 -263
- data/ext/re2/recipes.rb +8 -0
- data/lib/2.6/re2.so +0 -0
- data/lib/2.7/re2.so +0 -0
- data/lib/3.0/re2.so +0 -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/re2/regexp.rb +70 -0
- data/lib/re2/scanner.rb +9 -0
- data/lib/re2/string.rb +10 -59
- data/lib/re2/version.rb +10 -1
- data/lib/re2.rb +7 -3
- data/re2.gemspec +3 -2
- data/spec/kernel_spec.rb +2 -2
- data/spec/re2/match_data_spec.rb +64 -25
- data/spec/re2/regexp_spec.rb +492 -113
- data/spec/re2/scanner_spec.rb +3 -8
- data/spec/re2/set_spec.rb +18 -18
- data/spec/re2_spec.rb +4 -4
- metadata +11 -9
data/spec/re2/regexp_spec.rb
CHANGED
@@ -2,11 +2,13 @@ RSpec.describe RE2::Regexp do
|
|
2
2
|
describe "#initialize" do
|
3
3
|
it "returns an instance given only a pattern" do
|
4
4
|
re = RE2::Regexp.new('woo')
|
5
|
+
|
5
6
|
expect(re).to be_a(RE2::Regexp)
|
6
7
|
end
|
7
8
|
|
8
9
|
it "returns an instance given a pattern and options" do
|
9
|
-
re = RE2::Regexp.new('woo', :
|
10
|
+
re = RE2::Regexp.new('woo', case_sensitive: false)
|
11
|
+
|
10
12
|
expect(re).to be_a(RE2::Regexp)
|
11
13
|
end
|
12
14
|
|
@@ -15,7 +17,8 @@ RSpec.describe RE2::Regexp do
|
|
15
17
|
end
|
16
18
|
|
17
19
|
it "allows invalid patterns to be created" do
|
18
|
-
re = RE2::Regexp.new('???', :
|
20
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
21
|
+
|
19
22
|
expect(re).to be_a(RE2::Regexp)
|
20
23
|
end
|
21
24
|
|
@@ -29,11 +32,12 @@ RSpec.describe RE2::Regexp do
|
|
29
32
|
describe ".compile" do
|
30
33
|
it "returns an instance given only a pattern" do
|
31
34
|
re = RE2::Regexp.compile('woo')
|
35
|
+
|
32
36
|
expect(re).to be_a(RE2::Regexp)
|
33
37
|
end
|
34
38
|
|
35
39
|
it "returns an instance given a pattern and options" do
|
36
|
-
re = RE2::Regexp.compile('woo', :
|
40
|
+
re = RE2::Regexp.compile('woo', case_sensitive: false)
|
37
41
|
expect(re).to be_a(RE2::Regexp)
|
38
42
|
end
|
39
43
|
|
@@ -42,7 +46,8 @@ RSpec.describe RE2::Regexp do
|
|
42
46
|
end
|
43
47
|
|
44
48
|
it "allows invalid patterns to be created" do
|
45
|
-
re = RE2::Regexp.compile('???', :
|
49
|
+
re = RE2::Regexp.compile('???', log_errors: false)
|
50
|
+
|
46
51
|
expect(re).to be_a(RE2::Regexp)
|
47
52
|
end
|
48
53
|
|
@@ -60,34 +65,38 @@ RSpec.describe RE2::Regexp do
|
|
60
65
|
end
|
61
66
|
|
62
67
|
it "is populated with default options when nothing has been set" do
|
63
|
-
|
64
|
-
|
65
|
-
|
66
|
-
|
67
|
-
|
68
|
-
|
69
|
-
|
70
|
-
|
71
|
-
|
72
|
-
|
73
|
-
|
68
|
+
expect(RE2::Regexp.new('woo').options).to include(
|
69
|
+
utf8: true,
|
70
|
+
posix_syntax: false,
|
71
|
+
longest_match: false,
|
72
|
+
log_errors: true,
|
73
|
+
literal: false,
|
74
|
+
never_nl: false,
|
75
|
+
case_sensitive: true,
|
76
|
+
perl_classes: false,
|
77
|
+
word_boundary: false,
|
78
|
+
one_line: false
|
79
|
+
)
|
74
80
|
end
|
75
81
|
|
76
82
|
it "is populated with overridden options when specified" do
|
77
|
-
options = RE2::Regexp.new('woo', :
|
78
|
-
|
83
|
+
options = RE2::Regexp.new('woo', case_sensitive: false).options
|
84
|
+
|
85
|
+
expect(options).to include(case_sensitive: false)
|
79
86
|
end
|
80
87
|
end
|
81
88
|
|
82
89
|
describe "#error" do
|
83
90
|
it "returns nil if there is no error" do
|
84
91
|
error = RE2::Regexp.new('woo').error
|
92
|
+
|
85
93
|
expect(error).to be_nil
|
86
94
|
end
|
87
95
|
|
88
|
-
# Use log_errors
|
96
|
+
# Use log_errors: false to suppress RE2's logging to STDERR.
|
89
97
|
it "contains the error string if there is an error" do
|
90
|
-
error = RE2::Regexp.new('wo(o', :
|
98
|
+
error = RE2::Regexp.new('wo(o', log_errors: false).error
|
99
|
+
|
91
100
|
expect(error).to eq("missing ): wo(o")
|
92
101
|
end
|
93
102
|
end
|
@@ -95,11 +104,13 @@ RSpec.describe RE2::Regexp do
|
|
95
104
|
describe "#error_arg" do
|
96
105
|
it "returns nil if there is no error" do
|
97
106
|
error_arg = RE2::Regexp.new('woo').error_arg
|
107
|
+
|
98
108
|
expect(error_arg).to be_nil
|
99
109
|
end
|
100
110
|
|
101
|
-
it "returns the offending portion of the
|
102
|
-
error_arg = RE2::Regexp.new('wo(o', :
|
111
|
+
it "returns the offending portion of the pattern if there is an error" do
|
112
|
+
error_arg = RE2::Regexp.new('wo(o', log_errors: false).error_arg
|
113
|
+
|
103
114
|
expect(error_arg).to eq("wo(o")
|
104
115
|
end
|
105
116
|
end
|
@@ -112,7 +123,8 @@ RSpec.describe RE2::Regexp do
|
|
112
123
|
end
|
113
124
|
|
114
125
|
it "returns -1 for an invalid pattern" do
|
115
|
-
program_size = RE2::Regexp.new('???', :
|
126
|
+
program_size = RE2::Regexp.new('???', log_errors: false).program_size
|
127
|
+
|
116
128
|
expect(program_size).to eq(-1)
|
117
129
|
end
|
118
130
|
end
|
@@ -120,18 +132,27 @@ RSpec.describe RE2::Regexp do
|
|
120
132
|
describe "#to_str" do
|
121
133
|
it "returns the original pattern" do
|
122
134
|
string = RE2::Regexp.new('w(o)(o)').to_str
|
135
|
+
|
123
136
|
expect(string).to eq("w(o)(o)")
|
124
137
|
end
|
138
|
+
|
139
|
+
it "returns the pattern even if invalid" do
|
140
|
+
string = RE2::Regexp.new('???', log_errors: false).to_str
|
141
|
+
|
142
|
+
expect(string).to eq("???")
|
143
|
+
end
|
125
144
|
end
|
126
145
|
|
127
146
|
describe "#pattern" do
|
128
147
|
it "returns the original pattern" do
|
129
148
|
pattern = RE2::Regexp.new('w(o)(o)').pattern
|
149
|
+
|
130
150
|
expect(pattern).to eq("w(o)(o)")
|
131
151
|
end
|
132
152
|
|
133
153
|
it "returns the pattern even if invalid" do
|
134
|
-
pattern = RE2::Regexp.new('???', :
|
154
|
+
pattern = RE2::Regexp.new('???', log_errors: false).pattern
|
155
|
+
|
135
156
|
expect(pattern).to eq("???")
|
136
157
|
end
|
137
158
|
end
|
@@ -139,8 +160,15 @@ RSpec.describe RE2::Regexp do
|
|
139
160
|
describe "#inspect" do
|
140
161
|
it "shows the class name and original pattern" do
|
141
162
|
string = RE2::Regexp.new('w(o)(o)').inspect
|
163
|
+
|
142
164
|
expect(string).to eq("#<RE2::Regexp /w(o)(o)/>")
|
143
165
|
end
|
166
|
+
|
167
|
+
it "respects the pattern's original encoding" do
|
168
|
+
string = RE2::Regexp.new('w(o)(o)', utf8: false).inspect
|
169
|
+
|
170
|
+
expect(string.encoding).to eq(Encoding::ISO_8859_1)
|
171
|
+
end
|
144
172
|
end
|
145
173
|
|
146
174
|
describe "#utf8?" do
|
@@ -149,7 +177,8 @@ RSpec.describe RE2::Regexp do
|
|
149
177
|
end
|
150
178
|
|
151
179
|
it "can be overridden on initialization" do
|
152
|
-
re = RE2::Regexp.new('woo', :
|
180
|
+
re = RE2::Regexp.new('woo', utf8: false)
|
181
|
+
|
153
182
|
expect(re).to_not be_utf8
|
154
183
|
end
|
155
184
|
end
|
@@ -160,7 +189,8 @@ RSpec.describe RE2::Regexp do
|
|
160
189
|
end
|
161
190
|
|
162
191
|
it "can be overridden on initialization" do
|
163
|
-
re = RE2::Regexp.new('woo', :
|
192
|
+
re = RE2::Regexp.new('woo', posix_syntax: true)
|
193
|
+
|
164
194
|
expect(re).to be_posix_syntax
|
165
195
|
end
|
166
196
|
end
|
@@ -171,7 +201,8 @@ RSpec.describe RE2::Regexp do
|
|
171
201
|
end
|
172
202
|
|
173
203
|
it "can be overridden on initialization" do
|
174
|
-
re = RE2::Regexp.new('woo', :
|
204
|
+
re = RE2::Regexp.new('woo', literal: true)
|
205
|
+
|
175
206
|
expect(re).to be_literal
|
176
207
|
end
|
177
208
|
end
|
@@ -182,7 +213,8 @@ RSpec.describe RE2::Regexp do
|
|
182
213
|
end
|
183
214
|
|
184
215
|
it "can be overridden on initialization" do
|
185
|
-
re = RE2::Regexp.new('woo', :
|
216
|
+
re = RE2::Regexp.new('woo', never_nl: true)
|
217
|
+
|
186
218
|
expect(re).to be_never_nl
|
187
219
|
end
|
188
220
|
end
|
@@ -193,7 +225,7 @@ RSpec.describe RE2::Regexp do
|
|
193
225
|
end
|
194
226
|
|
195
227
|
it "can be overridden on initialization" do
|
196
|
-
re = RE2::Regexp.new('woo', :
|
228
|
+
re = RE2::Regexp.new('woo', case_sensitive: false)
|
197
229
|
expect(re).to_not be_case_sensitive
|
198
230
|
end
|
199
231
|
end
|
@@ -204,7 +236,8 @@ RSpec.describe RE2::Regexp do
|
|
204
236
|
end
|
205
237
|
|
206
238
|
it "can be overridden on initialization" do
|
207
|
-
re = RE2::Regexp.new('woo', :
|
239
|
+
re = RE2::Regexp.new('woo', case_sensitive: false)
|
240
|
+
|
208
241
|
expect(re).to be_case_insensitive
|
209
242
|
end
|
210
243
|
end
|
@@ -215,7 +248,8 @@ RSpec.describe RE2::Regexp do
|
|
215
248
|
end
|
216
249
|
|
217
250
|
it "can be overridden on initialization" do
|
218
|
-
re = RE2::Regexp.new('woo', :
|
251
|
+
re = RE2::Regexp.new('woo', case_sensitive: false)
|
252
|
+
|
219
253
|
expect(re).to be_casefold
|
220
254
|
end
|
221
255
|
end
|
@@ -226,7 +260,8 @@ RSpec.describe RE2::Regexp do
|
|
226
260
|
end
|
227
261
|
|
228
262
|
it "can be overridden on initialization" do
|
229
|
-
re = RE2::Regexp.new('woo', :
|
263
|
+
re = RE2::Regexp.new('woo', longest_match: true)
|
264
|
+
|
230
265
|
expect(re).to be_longest_match
|
231
266
|
end
|
232
267
|
end
|
@@ -237,7 +272,8 @@ RSpec.describe RE2::Regexp do
|
|
237
272
|
end
|
238
273
|
|
239
274
|
it "can be overridden on initialization" do
|
240
|
-
re = RE2::Regexp.new('woo', :
|
275
|
+
re = RE2::Regexp.new('woo', log_errors: false)
|
276
|
+
|
241
277
|
expect(re).to_not be_log_errors
|
242
278
|
end
|
243
279
|
end
|
@@ -248,7 +284,8 @@ RSpec.describe RE2::Regexp do
|
|
248
284
|
end
|
249
285
|
|
250
286
|
it "can be overridden on initialization" do
|
251
|
-
re = RE2::Regexp.new('woo', :
|
287
|
+
re = RE2::Regexp.new('woo', perl_classes: true)
|
288
|
+
|
252
289
|
expect(re).to be_perl_classes
|
253
290
|
end
|
254
291
|
end
|
@@ -259,7 +296,8 @@ RSpec.describe RE2::Regexp do
|
|
259
296
|
end
|
260
297
|
|
261
298
|
it "can be overridden on initialization" do
|
262
|
-
re = RE2::Regexp.new('woo', :
|
299
|
+
re = RE2::Regexp.new('woo', word_boundary: true)
|
300
|
+
|
263
301
|
expect(re).to be_word_boundary
|
264
302
|
end
|
265
303
|
end
|
@@ -270,7 +308,8 @@ RSpec.describe RE2::Regexp do
|
|
270
308
|
end
|
271
309
|
|
272
310
|
it "can be overridden on initialization" do
|
273
|
-
re = RE2::Regexp.new('woo', :
|
311
|
+
re = RE2::Regexp.new('woo', one_line: true)
|
312
|
+
|
274
313
|
expect(re).to be_one_line
|
275
314
|
end
|
276
315
|
end
|
@@ -281,144 +320,372 @@ RSpec.describe RE2::Regexp do
|
|
281
320
|
end
|
282
321
|
|
283
322
|
it "can be overridden on initialization" do
|
284
|
-
re = RE2::Regexp.new('woo', :
|
323
|
+
re = RE2::Regexp.new('woo', max_mem: 1024)
|
324
|
+
|
285
325
|
expect(re.max_mem).to eq(1024)
|
286
326
|
end
|
287
327
|
end
|
288
328
|
|
289
329
|
describe "#match" do
|
290
|
-
|
330
|
+
it "returns match data given only text if the pattern has capturing groups" do
|
331
|
+
re = RE2::Regexp.new('My name is (\w+) (\w+)')
|
291
332
|
|
292
|
-
|
293
|
-
md = re.match("My name is Robert Paulson")
|
294
|
-
expect(md).to be_a(RE2::MatchData)
|
333
|
+
expect(re.match("My name is Alice Bloggs")).to be_a(RE2::MatchData)
|
295
334
|
end
|
296
335
|
|
297
|
-
it "returns
|
298
|
-
|
336
|
+
it "returns only true or false given only text if the pattern has no capturing groups" do
|
337
|
+
re = RE2::Regexp.new('My name is \w+ \w+')
|
338
|
+
|
339
|
+
expect(re.match("My name is Alice Bloggs")).to eq(true)
|
299
340
|
end
|
300
341
|
|
301
|
-
it "returns
|
302
|
-
|
303
|
-
|
342
|
+
it "returns nil if the text does not match the pattern" do
|
343
|
+
re = RE2::Regexp.new('My name is (\w+) (\w+)')
|
344
|
+
|
345
|
+
expect(re.match("My age is 99")).to be_nil
|
304
346
|
end
|
305
347
|
|
306
|
-
it "
|
307
|
-
re = RE2::Regexp.new('My name is')
|
348
|
+
it "accepts text that can be coerced to a string" do
|
349
|
+
re = RE2::Regexp.new('My name is (\w+) (\w+)')
|
308
350
|
|
309
|
-
expect(re.match(
|
351
|
+
expect(re.match(StringLike.new("My name is Alice Bloggs"))).to be_a(RE2::MatchData)
|
310
352
|
end
|
311
353
|
|
312
|
-
it "raises an exception when given
|
354
|
+
it "raises an exception when given text that cannot be coerced to a string" do
|
355
|
+
re = RE2::Regexp.new('My name is (\w+) (\w+)')
|
356
|
+
|
313
357
|
expect { re.match(nil) }.to raise_error(TypeError)
|
314
358
|
end
|
315
359
|
|
316
|
-
it "
|
317
|
-
|
360
|
+
it "returns nil with an invalid pattern" do
|
361
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
362
|
+
|
363
|
+
expect(re.match("My name is Alice Bloggs")).to be_nil
|
364
|
+
end
|
365
|
+
|
366
|
+
it "returns nil with an invalid pattern and options" do
|
367
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
368
|
+
|
369
|
+
expect(re.match('foo bar', startpos: 1)).to be_nil
|
318
370
|
end
|
319
371
|
|
320
|
-
it "
|
321
|
-
|
372
|
+
it "accepts an offset at which to start matching", :aggregate_failures do
|
373
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
374
|
+
md = re.match("one two three", startpos: 4)
|
375
|
+
|
376
|
+
expect(md[1]).to eq("two")
|
377
|
+
expect(md[2]).to eq("three")
|
322
378
|
end
|
323
379
|
|
324
|
-
it "returns nil
|
325
|
-
|
326
|
-
|
380
|
+
it "returns nil if using a starting offset past the end of the text" do
|
381
|
+
skip "Underlying RE2::Match does not have endpos argument" unless RE2::Regexp.match_has_endpos_argument?
|
382
|
+
|
383
|
+
re = RE2::Regexp.new('(\w+) (\w+)', log_errors: false)
|
384
|
+
|
385
|
+
expect(re.match("one two three", startpos: 20, endpos: 21)).to be_nil
|
327
386
|
end
|
328
387
|
|
329
|
-
|
330
|
-
|
388
|
+
it "raises an exception when given a negative starting offset" do
|
389
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
331
390
|
|
332
|
-
|
333
|
-
|
334
|
-
|
391
|
+
expect { re.match("one two three", startpos: -1) }.to raise_error(ArgumentError, "startpos should be >= 0")
|
392
|
+
end
|
393
|
+
|
394
|
+
it "raises an exception when given a starting offset past the default ending offset" do
|
395
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
396
|
+
|
397
|
+
expect { re.match("one two three", startpos: 30) }.to raise_error(ArgumentError, "startpos should be <= endpos")
|
398
|
+
end
|
335
399
|
|
336
|
-
|
337
|
-
|
338
|
-
end
|
400
|
+
it "accepts an offset at which to end matching", :aggregate_failures do
|
401
|
+
skip "Underlying RE2::Match does not have endpos argument" unless RE2::Regexp.match_has_endpos_argument?
|
339
402
|
|
340
|
-
|
341
|
-
|
342
|
-
end
|
403
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
404
|
+
md = re.match("one two three", endpos: 6)
|
343
405
|
|
344
|
-
|
345
|
-
|
346
|
-
end
|
406
|
+
expect(md[1]).to eq("one")
|
407
|
+
expect(md[2]).to eq("tw")
|
347
408
|
end
|
348
409
|
|
349
|
-
|
350
|
-
|
410
|
+
it "returns nil if using a ending offset at the start of the text" do
|
411
|
+
skip "Underlying RE2::Match does not have endpos argument" unless RE2::Regexp.match_has_endpos_argument?
|
412
|
+
|
413
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
414
|
+
|
415
|
+
expect(re.match("one two three", endpos: 0)).to be_nil
|
416
|
+
end
|
351
417
|
|
352
|
-
|
353
|
-
|
354
|
-
end
|
418
|
+
it "raises an exception when given a negative ending offset" do
|
419
|
+
skip "Underlying RE2::Match does not have endpos argument" unless RE2::Regexp.match_has_endpos_argument?
|
355
420
|
|
356
|
-
|
357
|
-
|
358
|
-
|
421
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
422
|
+
|
423
|
+
expect { re.match("one two three", endpos: -1) }.to raise_error(ArgumentError, "endpos should be >= 0")
|
424
|
+
end
|
425
|
+
|
426
|
+
it "raises an exception when given an ending offset before the starting offset" do
|
427
|
+
skip "Underlying RE2::Match does not have endpos argument" unless RE2::Regexp.match_has_endpos_argument?
|
428
|
+
|
429
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
430
|
+
|
431
|
+
expect { re.match("one two three", startpos: 3, endpos: 0) }.to raise_error(ArgumentError, "startpos should be <= endpos")
|
432
|
+
end
|
433
|
+
|
434
|
+
it "raises an error if given an ending offset and RE2 does not support it" do
|
435
|
+
skip "Underlying RE2::Match has endpos argument" if RE2::Regexp.match_has_endpos_argument?
|
436
|
+
|
437
|
+
re = RE2::Regexp.new('(\w+) (\w+)')
|
438
|
+
|
439
|
+
expect { re.match("one two three", endpos: 3) }.to raise_error(RE2::Regexp::UnsupportedError)
|
440
|
+
end
|
441
|
+
|
442
|
+
it "does not anchor matches by default when extracting submatches" do
|
443
|
+
re = RE2::Regexp.new('(two)')
|
444
|
+
|
445
|
+
expect(re.match("one two three")).to be_a(RE2::MatchData)
|
446
|
+
end
|
447
|
+
|
448
|
+
it "does not anchor matches by default without extracting submatches" do
|
449
|
+
re = RE2::Regexp.new('(two)')
|
450
|
+
|
451
|
+
expect(re.match("one two three", submatches: 0)).to eq(true)
|
452
|
+
end
|
453
|
+
|
454
|
+
it "can explicitly match without anchoring when extracting submatches" do
|
455
|
+
re = RE2::Regexp.new('(two)')
|
456
|
+
|
457
|
+
expect(re.match("one two three", anchor: :unanchored)).to be_a(RE2::MatchData)
|
458
|
+
end
|
459
|
+
|
460
|
+
it "can explicitly match with neither anchoring nor extracting submatches" do
|
461
|
+
re = RE2::Regexp.new('(two)')
|
462
|
+
|
463
|
+
expect(re.match("one two three", anchor: :unanchored, submatches: 0)).to eq(true)
|
464
|
+
end
|
465
|
+
|
466
|
+
it "can anchor matches at the start when extracting submatches", :aggregate_failures do
|
467
|
+
re = RE2::Regexp.new('(two)')
|
468
|
+
|
469
|
+
expect(re.match("two three", anchor: :anchor_start)).to be_a(RE2::MatchData)
|
470
|
+
expect(re.match("one two three", anchor: :anchor_start)).to be_nil
|
471
|
+
end
|
472
|
+
|
473
|
+
it "can anchor matches at the start without extracting submatches", :aggregate_failures do
|
474
|
+
re = RE2::Regexp.new('(two)')
|
475
|
+
|
476
|
+
expect(re.match("two three", anchor: :anchor_start, submatches: 0)).to eq(true)
|
477
|
+
expect(re.match("one two three", anchor: :anchor_start, submatches: 0)).to eq(false)
|
478
|
+
end
|
479
|
+
|
480
|
+
it "can anchor matches at both ends when extracting submatches", :aggregate_failures do
|
481
|
+
re = RE2::Regexp.new('(two)')
|
482
|
+
|
483
|
+
expect(re.match("two three", anchor: :anchor_both)).to be_nil
|
484
|
+
expect(re.match("two", anchor: :anchor_both)).to be_a(RE2::MatchData)
|
485
|
+
end
|
486
|
+
|
487
|
+
it "does not anchor matches when given a nil anchor" do
|
488
|
+
re = RE2::Regexp.new('(two)')
|
489
|
+
|
490
|
+
expect(re.match("one two three", anchor: nil)).to be_a(RE2::MatchData)
|
491
|
+
end
|
492
|
+
|
493
|
+
it "raises an exception when given an invalid anchor" do
|
494
|
+
re = RE2::Regexp.new('(two)')
|
495
|
+
|
496
|
+
expect { re.match("one two three", anchor: :invalid) }.to raise_error(ArgumentError, "anchor should be one of: :unanchored, :anchor_start, :anchor_both")
|
497
|
+
end
|
498
|
+
|
499
|
+
it "raises an exception when given a non-symbol anchor" do
|
500
|
+
re = RE2::Regexp.new('(two)')
|
501
|
+
|
502
|
+
expect { re.match("one two three", anchor: 0) }.to raise_error(TypeError)
|
503
|
+
end
|
504
|
+
|
505
|
+
it "extracts all submatches by default", :aggregate_failures do
|
506
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
507
|
+
md = re.match("one two three")
|
508
|
+
|
509
|
+
expect(md[1]).to eq("one")
|
510
|
+
expect(md[2]).to eq("two")
|
511
|
+
expect(md[3]).to eq("three")
|
512
|
+
end
|
513
|
+
|
514
|
+
it "extracts a specific number of submatches", :aggregate_failures do
|
515
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
516
|
+
md = re.match("one two three", submatches: 2)
|
517
|
+
|
518
|
+
expect(md[1]).to eq("one")
|
519
|
+
expect(md[2]).to eq("two")
|
520
|
+
expect(md[3]).to be_nil
|
521
|
+
end
|
522
|
+
|
523
|
+
it "pads submatches with nil when requesting more than the number of capturing groups" do
|
524
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
525
|
+
md = re.match("one two three", submatches: 5)
|
526
|
+
|
527
|
+
expect(md.to_a).to eq(["one two three", "one", "two", "three", nil, nil])
|
528
|
+
end
|
529
|
+
|
530
|
+
it "raises an exception when given a negative number of submatches" do
|
531
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
532
|
+
|
533
|
+
expect { re.match("one two three", submatches: -1) }.to raise_error(ArgumentError, "number of matches should be >= 0")
|
534
|
+
end
|
535
|
+
|
536
|
+
it "raises an exception when given a non-numeric number of submatches" do
|
537
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
538
|
+
|
539
|
+
expect { re.match("one two three", submatches: :invalid) }.to raise_error(TypeError)
|
540
|
+
end
|
359
541
|
|
360
|
-
|
361
|
-
|
362
|
-
|
363
|
-
end
|
542
|
+
it "defaults to extracting all submatches when given nil", :aggregate_failures do
|
543
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
544
|
+
md = re.match("one two three", submatches: nil)
|
364
545
|
|
365
|
-
|
366
|
-
|
367
|
-
|
368
|
-
|
369
|
-
|
370
|
-
|
546
|
+
expect(md[1]).to eq("one")
|
547
|
+
expect(md[2]).to eq("two")
|
548
|
+
expect(md[3]).to eq("three")
|
549
|
+
end
|
550
|
+
|
551
|
+
it "accepts passing the number of submatches instead of options for backward compatibility", :aggregate_failures do
|
552
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
553
|
+
md = re.match("one two three", 2)
|
554
|
+
|
555
|
+
expect(md[1]).to eq("one")
|
556
|
+
expect(md[2]).to eq("two")
|
557
|
+
expect(md[3]).to be_nil
|
558
|
+
end
|
559
|
+
|
560
|
+
it "raises an exception when given invalid options" do
|
561
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
562
|
+
|
563
|
+
expect { re.match("one two three", :invalid) }.to raise_error(TypeError)
|
564
|
+
end
|
565
|
+
|
566
|
+
it "accepts anything that can be coerced to a hash as options", :aggregate_failures do
|
567
|
+
re = RE2::Regexp.new('(\w+) (\w+) (\w+)')
|
568
|
+
|
569
|
+
expect(re.match("one two three", nil)).to be_a(RE2::MatchData)
|
371
570
|
end
|
372
571
|
end
|
373
572
|
|
374
573
|
describe "#match?" do
|
375
|
-
it "returns only true or false if
|
574
|
+
it "returns only true or false even if there are capturing groups", :aggregate_failures do
|
376
575
|
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
377
|
-
|
576
|
+
|
577
|
+
expect(re.match?("My name is Alice Bloggs")).to eq(true)
|
378
578
|
expect(re.match?("My age is 99")).to eq(false)
|
379
579
|
end
|
380
580
|
|
381
581
|
it "returns false if the pattern is invalid" do
|
382
|
-
re = RE2::Regexp.new('???', :
|
383
|
-
|
582
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
583
|
+
|
584
|
+
expect(re.match?("My name is Alice Bloggs")).to eq(false)
|
585
|
+
end
|
586
|
+
|
587
|
+
it "raises an exception if text cannot be coerced to a string" do
|
588
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
589
|
+
|
590
|
+
expect { re.match?(0) }.to raise_error(TypeError)
|
591
|
+
end
|
592
|
+
end
|
593
|
+
|
594
|
+
describe "#partial_match?" do
|
595
|
+
it "returns only true or false even if there are capturing groups", :aggregate_failures do
|
596
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
597
|
+
|
598
|
+
expect(re.partial_match?("My name is Alice Bloggs")).to eq(true)
|
599
|
+
expect(re.partial_match?("My age is 99")).to eq(false)
|
600
|
+
end
|
601
|
+
|
602
|
+
it "returns false if the pattern is invalid" do
|
603
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
604
|
+
|
605
|
+
expect(re.partial_match?("My name is Alice Bloggs")).to eq(false)
|
606
|
+
end
|
607
|
+
|
608
|
+
it "raises an exception if text cannot be coerced to a string" do
|
609
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
610
|
+
|
611
|
+
expect { re.partial_match?(0) }.to raise_error(TypeError)
|
384
612
|
end
|
385
613
|
end
|
386
614
|
|
387
615
|
describe "#=~" do
|
388
|
-
it "returns only true or false if
|
616
|
+
it "returns only true or false even if there are capturing groups", :aggregate_failures do
|
389
617
|
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
390
|
-
|
618
|
+
|
619
|
+
expect(re =~ "My name is Alice Bloggs").to eq(true)
|
391
620
|
expect(re =~ "My age is 99").to eq(false)
|
392
621
|
end
|
393
|
-
end
|
394
622
|
|
395
|
-
|
396
|
-
|
623
|
+
it "returns false if the pattern is invalid" do
|
624
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
625
|
+
|
626
|
+
expect(re =~ "My name is Alice Bloggs").to eq(false)
|
627
|
+
end
|
628
|
+
|
629
|
+
it "raises an exception if text cannot be coerced to a string" do
|
397
630
|
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
398
|
-
|
399
|
-
expect
|
631
|
+
|
632
|
+
expect { re =~ 0 }.to raise_error(TypeError)
|
400
633
|
end
|
401
634
|
end
|
402
635
|
|
403
636
|
describe "#===" do
|
404
|
-
it "returns only true or false if
|
637
|
+
it "returns only true or false even if there are capturing groups", :aggregate_failures do
|
405
638
|
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
406
|
-
|
639
|
+
|
640
|
+
expect(re === "My name is Alice Bloggs").to eq(true)
|
407
641
|
expect(re === "My age is 99").to eq(false)
|
408
642
|
end
|
643
|
+
|
644
|
+
it "returns false if the pattern is invalid" do
|
645
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
646
|
+
|
647
|
+
expect(re === "My name is Alice Bloggs").to eq(false)
|
648
|
+
end
|
649
|
+
|
650
|
+
it "raises an exception if text cannot be coerced to a string" do
|
651
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
652
|
+
|
653
|
+
expect { re === 0 }.to raise_error(TypeError)
|
654
|
+
end
|
655
|
+
end
|
656
|
+
|
657
|
+
describe "#full_match?" do
|
658
|
+
it "returns only true or false even if there are capturing groups", :aggregate_failures do
|
659
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
660
|
+
|
661
|
+
expect(re.full_match?("My name is Alice Bloggs")).to eq(true)
|
662
|
+
expect(re.full_match?("My name is Alice Bloggs and I am 99")).to eq(false)
|
663
|
+
end
|
664
|
+
|
665
|
+
it "returns false if the pattern is invalid" do
|
666
|
+
re = RE2::Regexp.new('???', log_errors: false)
|
667
|
+
|
668
|
+
expect(re.full_match?("My name is Alice Bloggs")).to eq(false)
|
669
|
+
end
|
670
|
+
|
671
|
+
it "raises an exception if text cannot be coerced to a string" do
|
672
|
+
re = RE2::Regexp.new('My name is (\S+) (\S+)')
|
673
|
+
|
674
|
+
expect { re.full_match?(0) }.to raise_error(TypeError)
|
675
|
+
end
|
409
676
|
end
|
410
677
|
|
411
678
|
describe "#ok?" do
|
412
|
-
it "returns true for valid
|
679
|
+
it "returns true for valid patterns", :aggregate_failures do
|
413
680
|
expect(RE2::Regexp.new('woo')).to be_ok
|
414
681
|
expect(RE2::Regexp.new('wo(o)')).to be_ok
|
415
682
|
expect(RE2::Regexp.new('((\d)\w+){3,}')).to be_ok
|
416
683
|
end
|
417
684
|
|
418
|
-
it "returns false for invalid
|
419
|
-
expect(RE2::Regexp.new('wo(o', :
|
420
|
-
expect(RE2::Regexp.new('wo[o', :
|
421
|
-
expect(RE2::Regexp.new('*', :
|
685
|
+
it "returns false for invalid patterns", :aggregate_failures do
|
686
|
+
expect(RE2::Regexp.new('wo(o', log_errors: false)).to_not be_ok
|
687
|
+
expect(RE2::Regexp.new('wo[o', log_errors: false)).to_not be_ok
|
688
|
+
expect(RE2::Regexp.new('*', log_errors: false)).to_not be_ok
|
422
689
|
end
|
423
690
|
end
|
424
691
|
|
@@ -435,14 +702,14 @@ RSpec.describe RE2::Regexp do
|
|
435
702
|
end
|
436
703
|
|
437
704
|
describe "#number_of_capturing_groups" do
|
438
|
-
it "returns the number of groups in a
|
705
|
+
it "returns the number of groups in a pattern", :aggregate_failures do
|
439
706
|
expect(RE2::Regexp.new('(a)(b)(c)').number_of_capturing_groups).to eq(3)
|
440
707
|
expect(RE2::Regexp.new('abc').number_of_capturing_groups).to eq(0)
|
441
708
|
expect(RE2::Regexp.new('a((b)c)').number_of_capturing_groups).to eq(2)
|
442
709
|
end
|
443
710
|
|
444
|
-
it "returns -1 for an invalid
|
445
|
-
expect(RE2::Regexp.new('???', :
|
711
|
+
it "returns -1 for an invalid pattern" do
|
712
|
+
expect(RE2::Regexp.new('???', log_errors: false).number_of_capturing_groups).to eq(-1)
|
446
713
|
end
|
447
714
|
end
|
448
715
|
|
@@ -453,17 +720,18 @@ RSpec.describe RE2::Regexp do
|
|
453
720
|
|
454
721
|
it "maps names to indices with only one group" do
|
455
722
|
groups = RE2::Regexp.new('(?P<bob>a)').named_capturing_groups
|
456
|
-
|
723
|
+
|
724
|
+
expect(groups).to eq("bob" => 1)
|
457
725
|
end
|
458
726
|
|
459
727
|
it "maps names to indices with several groups" do
|
460
728
|
groups = RE2::Regexp.new('(?P<bob>a)(o)(?P<rob>e)').named_capturing_groups
|
461
|
-
|
462
|
-
expect(groups
|
729
|
+
|
730
|
+
expect(groups).to eq("bob" => 1, "rob" => 3)
|
463
731
|
end
|
464
732
|
|
465
733
|
it "returns an empty hash for an invalid regexp" do
|
466
|
-
expect(RE2::Regexp.new('???', :
|
734
|
+
expect(RE2::Regexp.new('???', log_errors: false).named_capturing_groups).to be_empty
|
467
735
|
end
|
468
736
|
end
|
469
737
|
|
@@ -475,4 +743,115 @@ RSpec.describe RE2::Regexp do
|
|
475
743
|
expect(scanner).to be_a(RE2::Scanner)
|
476
744
|
end
|
477
745
|
end
|
746
|
+
|
747
|
+
describe "#partial_match" do
|
748
|
+
it "matches the pattern anywhere within the given text" do
|
749
|
+
r = RE2::Regexp.new('f(o+)')
|
750
|
+
|
751
|
+
expect(r.partial_match("foo bar")).to be_a(RE2::MatchData)
|
752
|
+
end
|
753
|
+
|
754
|
+
it "returns true or false if there are no capturing groups" do
|
755
|
+
r = RE2::Regexp.new('fo+')
|
756
|
+
|
757
|
+
expect(r.partial_match("foo bar")).to eq(true)
|
758
|
+
end
|
759
|
+
|
760
|
+
it "can set the number of submatches to extract", :aggregate_failures do
|
761
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
762
|
+
m = r.partial_match("fooaa bar", submatches: 1)
|
763
|
+
|
764
|
+
expect(m[1]).to eq("oo")
|
765
|
+
expect(m[2]).to be_nil
|
766
|
+
|
767
|
+
m = r.partial_match("fooaa bar", submatches: 2)
|
768
|
+
|
769
|
+
expect(m[1]).to eq("oo")
|
770
|
+
expect(m[2]).to eq("aa")
|
771
|
+
end
|
772
|
+
|
773
|
+
it "raises an error if given non-hash options" do
|
774
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
775
|
+
|
776
|
+
expect { r.partial_match("fooaa bar", "not a hash") }.to raise_error(TypeError)
|
777
|
+
end
|
778
|
+
|
779
|
+
it "accepts options that can be coerced to a hash", :aggregate_failures do
|
780
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
781
|
+
|
782
|
+
m = r.partial_match("fooaa bar", nil)
|
783
|
+
expect(m[1]).to eq('oo')
|
784
|
+
|
785
|
+
m = r.partial_match("fooaa bar", [])
|
786
|
+
expect(m[1]).to eq('oo')
|
787
|
+
end
|
788
|
+
|
789
|
+
it "accepts anything that can be coerced to a string" do
|
790
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
791
|
+
|
792
|
+
expect(r.partial_match(StringLike.new("fooaa bar"))).to be_a(RE2::MatchData)
|
793
|
+
end
|
794
|
+
|
795
|
+
it "does not allow the anchor to be overridden" do
|
796
|
+
r = RE2::Regexp.new('(\d+)')
|
797
|
+
|
798
|
+
expect(r.partial_match('ruby:1234', anchor: :anchor_both)).to be_a(RE2::MatchData)
|
799
|
+
end
|
800
|
+
end
|
801
|
+
|
802
|
+
describe "#full_match" do
|
803
|
+
it "only matches the pattern if all of the given text matches", :aggregate_failures do
|
804
|
+
r = RE2::Regexp.new('f(o+)')
|
805
|
+
|
806
|
+
expect(r.full_match("foo")).to be_a(RE2::MatchData)
|
807
|
+
expect(r.full_match("foo bar")).to be_nil
|
808
|
+
end
|
809
|
+
|
810
|
+
it "returns true or false if there are no capturing groups" do
|
811
|
+
r = RE2::Regexp.new('fo+')
|
812
|
+
|
813
|
+
expect(r.full_match("foo")).to eq(true)
|
814
|
+
end
|
815
|
+
|
816
|
+
it "can set the number of submatches to extract", :aggregate_failures do
|
817
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
818
|
+
m = r.full_match("fooaa", submatches: 1)
|
819
|
+
|
820
|
+
expect(m[1]).to eq("oo")
|
821
|
+
expect(m[2]).to be_nil
|
822
|
+
|
823
|
+
m = r.full_match("fooaa", submatches: 2)
|
824
|
+
|
825
|
+
expect(m[1]).to eq("oo")
|
826
|
+
expect(m[2]).to eq("aa")
|
827
|
+
end
|
828
|
+
|
829
|
+
it "raises an error if given non-hash options" do
|
830
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
831
|
+
|
832
|
+
expect { r.full_match("fooaa", "not a hash") }.to raise_error(TypeError)
|
833
|
+
end
|
834
|
+
|
835
|
+
it "accepts options that can be coerced to a hash", :aggregate_failures do
|
836
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
837
|
+
|
838
|
+
m = r.full_match("fooaa", nil)
|
839
|
+
expect(m[1]).to eq("oo")
|
840
|
+
|
841
|
+
m = r.full_match("fooaa", [])
|
842
|
+
expect(m[1]).to eq("oo")
|
843
|
+
end
|
844
|
+
|
845
|
+
it "accepts anything that can be coerced to a string" do
|
846
|
+
r = RE2::Regexp.new('f(o+)(a+)')
|
847
|
+
|
848
|
+
expect(r.full_match(StringLike.new("fooaa"), submatches: 0)).to eq(true)
|
849
|
+
end
|
850
|
+
|
851
|
+
it "does not allow the anchor to be overridden" do
|
852
|
+
r = RE2::Regexp.new('(\d+)')
|
853
|
+
|
854
|
+
expect(r.full_match('ruby:1234', anchor: :unanchored)).to be_nil
|
855
|
+
end
|
856
|
+
end
|
478
857
|
end
|