summaryrefslogtreecommitdiffstats
path: root/third_party/rust/regex/testdata/utf8.toml
blob: 39e284b38280392f511a045d58ce2e03f2a228d7 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
# These test the UTF-8 modes expose by regex-automata. Namely, when utf8 is
# true, then we promise that the haystack is valid UTF-8. (Otherwise behavior
# is unspecified.) This also corresponds to building the regex engine with the
# following two guarantees:
#
# 1) For any non-empty match reported, its span is guaranteed to correspond to
# valid UTF-8.
# 2) All empty or zero-width matches reported must never split a UTF-8
# encoded codepoint. If the haystack has invalid UTF-8, then this results in
# unspecified behavior.
#
# The (2) is in particular what we focus our testing on since (1) is generally
# guaranteed by regex-syntax's AST-to-HIR translator and is well tested there.
# The thing with (2) is that it can't be described in the HIR, so the regex
# engines have to handle that case. Thus, we test it here.
#
# Note that it is possible to build a regex that has property (1) but not
# (2), and vice versa. This is done by building the HIR with 'utf8=true' but
# building the Thompson NFA with 'utf8=false'. We don't test that here because
# the harness doesn't expose a way to enable or disable UTF-8 mode with that
# granularity. Instead, those combinations are lightly tested via doc examples.
# That's not to say that (1) without (2) is uncommon. Indeed, ripgrep uses it
# because it cannot guarantee that its haystack is valid UTF-8.

# This tests that an empty regex doesn't split a codepoint.
[[test]]
name = "empty-utf8yes"
regex = ''
haystack = '☃'
matches = [[0, 0], [3, 3]]
unicode = true
utf8 = true

# Tests the overlapping case of the above.
[[test]]
name = "empty-utf8yes-overlapping"
regex = ''
haystack = '☃'
matches = [[0, 0], [3, 3]]
unicode = true
utf8 = true
match-kind = "all"
search-kind = "overlapping"

# This tests that an empty regex DOES split a codepoint when utf=false.
[[test]]
name = "empty-utf8no"
regex = ''
haystack = '☃'
matches = [[0, 0], [1, 1], [2, 2], [3, 3]]
unicode = true
utf8 = false

# Tests the overlapping case of the above.
[[test]]
name = "empty-utf8no-overlapping"
regex = ''
haystack = '☃'
matches = [[0, 0], [1, 1], [2, 2], [3, 3]]
unicode = true
utf8 = false
match-kind = "all"
search-kind = "overlapping"

# This tests that an empty regex doesn't split a codepoint, even if we give
# it bounds entirely within the codepoint.
#
# This is one of the trickier cases and is what motivated the current UTF-8
# mode design. In particular, at one point, this test failed the 'is_match'
# variant of the test but not 'find'. This is because the 'is_match' code path
# is specifically optimized for "was a match found" rather than "where is the
# match." In the former case, you don't really care about the empty-vs-non-empty
# matches, and thus, the codepoint splitting filtering logic wasn't getting
# applied. (In multiple ways across multiple regex engines.) In this way, you
# can wind up with a situation where 'is_match' says "yes," but 'find' says,
# "I didn't find anything." Which is... not great.
#
# I could have decided to say that providing boundaries that themselves split
# a codepoint would have unspecified behavior. But I couldn't quite convince
# myself that such boundaries were the only way to get an inconsistency between
# 'is_match' and 'find'.
#
# Note that I also tried to come up with a test like this that fails without
# using `bounds`. Specifically, a test where 'is_match' and 'find' disagree.
# But I couldn't do it, and I'm tempted to conclude it is impossible. The
# fundamental problem is that you need to simultaneously produce an empty match
# that splits a codepoint while *not* matching before or after the codepoint.
[[test]]
name = "empty-utf8yes-bounds"
regex = ''
haystack = '𝛃'
bounds = [1, 3]
matches = []
unicode = true
utf8 = true

# Tests the overlapping case of the above.
[[test]]
name = "empty-utf8yes-bounds-overlapping"
regex = ''
haystack = '𝛃'
bounds = [1, 3]
matches = []
unicode = true
utf8 = true
match-kind = "all"
search-kind = "overlapping"

# This tests that an empty regex splits a codepoint when the bounds are
# entirely within the codepoint.
[[test]]
name = "empty-utf8no-bounds"
regex = ''
haystack = '𝛃'
bounds = [1, 3]
matches = [[1, 1], [2, 2], [3, 3]]
unicode = true
utf8 = false

# Tests the overlapping case of the above.
[[test]]
name = "empty-utf8no-bounds-overlapping"
regex = ''
haystack = '𝛃'
bounds = [1, 3]
matches = [[1, 1], [2, 2], [3, 3]]
unicode = true
utf8 = false
match-kind = "all"
search-kind = "overlapping"

# In this test, we anchor the search. Since the start position is also a UTF-8
# boundary, we get a match.
[[test]]
name = "empty-utf8yes-anchored"
regex = ''
haystack = '𝛃'
matches = [[0, 0]]
anchored = true
unicode = true
utf8 = true

# Tests the overlapping case of the above.
[[test]]
name = "empty-utf8yes-anchored-overlapping"
regex = ''
haystack = '𝛃'
matches = [[0, 0]]
anchored = true
unicode = true
utf8 = true
match-kind = "all"
search-kind = "overlapping"

# Same as above, except with UTF-8 mode disabled. It almost doesn't change the
# result, except for the fact that since this is an anchored search and we
# always find all matches, the test harness will keep reporting matches until
# none are found. Because it's anchored, matches will be reported so long as
# they are directly adjacent. Since with UTF-8 mode the next anchored search
# after the match at [0, 0] fails, iteration stops (and doesn't find the last
# match at [4, 4]).
[[test]]
name = "empty-utf8no-anchored"
regex = ''
haystack = '𝛃'
matches = [[0, 0], [1, 1], [2, 2], [3, 3], [4, 4]]
anchored = true
unicode = true
utf8 = false

# Tests the overlapping case of the above.
#
# Note that overlapping anchored searches are a little weird, and it's not
# totally clear what their semantics ought to be. For now, we just test the
# current behavior of our test shim that implements overlapping search. (This
# is one of the reasons why we don't really expose regex-level overlapping
# searches.)
[[test]]
name = "empty-utf8no-anchored-overlapping"
regex = ''
haystack = '𝛃'
matches = [[0, 0]]
anchored = true
unicode = true
utf8 = false
match-kind = "all"
search-kind = "overlapping"

# In this test, we anchor the search, but also set bounds. The bounds start the
# search in the middle of a codepoint, so there should never be a match.
[[test]]
name = "empty-utf8yes-anchored-bounds"
regex = ''
haystack = '𝛃'
matches = []
bounds = [1, 3]
anchored = true
unicode = true
utf8 = true

# Tests the overlapping case of the above.
[[test]]
name = "empty-utf8yes-anchored-bounds-overlapping"
regex = ''
haystack = '𝛃'
matches = []
bounds = [1, 3]
anchored = true
unicode = true
utf8 = true
match-kind = "all"
search-kind = "overlapping"

# Same as above, except with UTF-8 mode disabled. Without UTF-8 mode enabled,
# matching within a codepoint is allowed. And remember, as in the anchored test
# above with UTF-8 mode disabled, iteration will report all adjacent matches.
# The matches at [0, 0] and [4, 4] are not included because of the bounds of
# the search.
[[test]]
name = "empty-utf8no-anchored-bounds"
regex = ''
haystack = '𝛃'
bounds = [1, 3]
matches = [[1, 1], [2, 2], [3, 3]]
anchored = true
unicode = true
utf8 = false

# Tests the overlapping case of the above.
#
# Note that overlapping anchored searches are a little weird, and it's not
# totally clear what their semantics ought to be. For now, we just test the
# current behavior of our test shim that implements overlapping search. (This
# is one of the reasons why we don't really expose regex-level overlapping
# searches.)
[[test]]
name = "empty-utf8no-anchored-bounds-overlapping"
regex = ''
haystack = '𝛃'
bounds = [1, 3]
matches = [[1, 1]]
anchored = true
unicode = true
utf8 = false
match-kind = "all"
search-kind = "overlapping"

# This tests that we find the match at the end of the string when the bounds
# exclude the first match.
[[test]]
name = "empty-utf8yes-startbound"
regex = ''
haystack = '𝛃'
bounds = [1, 4]
matches = [[4, 4]]
unicode = true
utf8 = true

# Tests the overlapping case of the above.
[[test]]
name = "empty-utf8yes-startbound-overlapping"
regex = ''
haystack = '𝛃'
bounds = [1, 4]
matches = [[4, 4]]
unicode = true
utf8 = true
match-kind = "all"
search-kind = "overlapping"

# Same as above, except since UTF-8 mode is disabled, we also find the matches
# inbetween that split the codepoint.
[[test]]
name = "empty-utf8no-startbound"
regex = ''
haystack = '𝛃'
bounds = [1, 4]
matches = [[1, 1], [2, 2], [3, 3], [4, 4]]
unicode = true
utf8 = false

# Tests the overlapping case of the above.
[[test]]
name = "empty-utf8no-startbound-overlapping"
regex = ''
haystack = '𝛃'
bounds = [1, 4]
matches = [[1, 1], [2, 2], [3, 3], [4, 4]]
unicode = true
utf8 = false
match-kind = "all"
search-kind = "overlapping"

# This tests that we don't find any matches in an anchored search, even when
# the bounds include a match (at the end).
[[test]]
name = "empty-utf8yes-anchored-startbound"
regex = ''
haystack = '𝛃'
bounds = [1, 4]
matches = []
anchored = true
unicode = true
utf8 = true

# Tests the overlapping case of the above.
[[test]]
name = "empty-utf8yes-anchored-startbound-overlapping"
regex = ''
haystack = '𝛃'
bounds = [1, 4]
matches = []
anchored = true
unicode = true
utf8 = true
match-kind = "all"
search-kind = "overlapping"

# Same as above, except since UTF-8 mode is disabled, we also find the matches
# inbetween that split the codepoint. Even though this is an anchored search,
# since the matches are adjacent, we find all of them.
[[test]]
name = "empty-utf8no-anchored-startbound"
regex = ''
haystack = '𝛃'
bounds = [1, 4]
matches = [[1, 1], [2, 2], [3, 3], [4, 4]]
anchored = true
unicode = true
utf8 = false

# Tests the overlapping case of the above.
#
# Note that overlapping anchored searches are a little weird, and it's not
# totally clear what their semantics ought to be. For now, we just test the
# current behavior of our test shim that implements overlapping search. (This
# is one of the reasons why we don't really expose regex-level overlapping
# searches.)
[[test]]
name = "empty-utf8no-anchored-startbound-overlapping"
regex = ''
haystack = '𝛃'
bounds = [1, 4]
matches = [[1, 1]]
anchored = true
unicode = true
utf8 = false
match-kind = "all"
search-kind = "overlapping"

# This tests that we find the match at the end of the haystack in UTF-8 mode
# when our bounds only include the empty string at the end of the haystack.
[[test]]
name = "empty-utf8yes-anchored-endbound"
regex = ''
haystack = '𝛃'
bounds = [4, 4]
matches = [[4, 4]]
anchored = true
unicode = true
utf8 = true

# Tests the overlapping case of the above.
[[test]]
name = "empty-utf8yes-anchored-endbound-overlapping"
regex = ''
haystack = '𝛃'
bounds = [4, 4]
matches = [[4, 4]]
anchored = true
unicode = true
utf8 = true
match-kind = "all"
search-kind = "overlapping"

# Same as above, but with UTF-8 mode disabled. Results remain the same since
# the only possible match does not split a codepoint.
[[test]]
name = "empty-utf8no-anchored-endbound"
regex = ''
haystack = '𝛃'
bounds = [4, 4]
matches = [[4, 4]]
anchored = true
unicode = true
utf8 = false

# Tests the overlapping case of the above.
[[test]]
name = "empty-utf8no-anchored-endbound-overlapping"
regex = ''
haystack = '𝛃'
bounds = [4, 4]
matches = [[4, 4]]
anchored = true
unicode = true
utf8 = false
match-kind = "all"
search-kind = "overlapping"