| 1 |
# See: https://github.com/rust-lang/regex/issues/48 |
| 2 |
[[test]] |
| 3 |
name = "invalid-regex-no-crash-100" |
| 4 |
regex = '(*)' |
| 5 |
haystack = "" |
| 6 |
matches = [] |
| 7 |
compiles = false |
| 8 |
|
| 9 |
# See: https://github.com/rust-lang/regex/issues/48 |
| 10 |
[[test]] |
| 11 |
name = "invalid-regex-no-crash-200" |
| 12 |
regex = '(?:?)' |
| 13 |
haystack = "" |
| 14 |
matches = [] |
| 15 |
compiles = false |
| 16 |
|
| 17 |
# See: https://github.com/rust-lang/regex/issues/48 |
| 18 |
[[test]] |
| 19 |
name = "invalid-regex-no-crash-300" |
| 20 |
regex = '(?)' |
| 21 |
haystack = "" |
| 22 |
matches = [] |
| 23 |
compiles = false |
| 24 |
|
| 25 |
# See: https://github.com/rust-lang/regex/issues/48 |
| 26 |
[[test]] |
| 27 |
name = "invalid-regex-no-crash-400" |
| 28 |
regex = '*' |
| 29 |
haystack = "" |
| 30 |
matches = [] |
| 31 |
compiles = false |
| 32 |
|
| 33 |
# See: https://github.com/rust-lang/regex/issues/75 |
| 34 |
[[test]] |
| 35 |
name = "unsorted-binary-search-100" |
| 36 |
regex = '(?i-u)[a_]+' |
| 37 |
haystack = "A_" |
| 38 |
matches = [[0, 2]] |
| 39 |
|
| 40 |
# See: https://github.com/rust-lang/regex/issues/75 |
| 41 |
[[test]] |
| 42 |
name = "unsorted-binary-search-200" |
| 43 |
regex = '(?i-u)[A_]+' |
| 44 |
haystack = "a_" |
| 45 |
matches = [[0, 2]] |
| 46 |
|
| 47 |
# See: https://github.com/rust-lang/regex/issues/76 |
| 48 |
[[test]] |
| 49 |
name = "unicode-case-lower-nocase-flag" |
| 50 |
regex = '(?i)\p{Ll}+' |
| 51 |
haystack = "ΛΘΓΔα" |
| 52 |
matches = [[0, 10]] |
| 53 |
|
| 54 |
# See: https://github.com/rust-lang/regex/issues/99 |
| 55 |
[[test]] |
| 56 |
name = "negated-char-class-100" |
| 57 |
regex = '(?i)[^x]' |
| 58 |
haystack = "x" |
| 59 |
matches = [] |
| 60 |
|
| 61 |
# See: https://github.com/rust-lang/regex/issues/99 |
| 62 |
[[test]] |
| 63 |
name = "negated-char-class-200" |
| 64 |
regex = '(?i)[^x]' |
| 65 |
haystack = "X" |
| 66 |
matches = [] |
| 67 |
|
| 68 |
# See: https://github.com/rust-lang/regex/issues/101 |
| 69 |
[[test]] |
| 70 |
name = "ascii-word-underscore" |
| 71 |
regex = '[[:word:]]' |
| 72 |
haystack = "_" |
| 73 |
matches = [[0, 1]] |
| 74 |
|
| 75 |
# See: https://github.com/rust-lang/regex/issues/129 |
| 76 |
[[test]] |
| 77 |
name = "captures-repeat" |
| 78 |
regex = '([a-f]){2}(?P<foo>[x-z])' |
| 79 |
haystack = "abx" |
| 80 |
matches = [ |
| 81 |
[[0, 3], [1, 2], [2, 3]], |
| 82 |
] |
| 83 |
|
| 84 |
# See: https://github.com/rust-lang/regex/issues/153 |
| 85 |
[[test]] |
| 86 |
name = "alt-in-alt-100" |
| 87 |
regex = 'ab?|$' |
| 88 |
haystack = "az" |
| 89 |
matches = [[0, 1], [2, 2]] |
| 90 |
|
| 91 |
# See: https://github.com/rust-lang/regex/issues/153 |
| 92 |
[[test]] |
| 93 |
name = "alt-in-alt-200" |
| 94 |
regex = '^(?:.*?)(?:\n|\r\n?|$)' |
| 95 |
haystack = "ab\rcd" |
| 96 |
matches = [[0, 3]] |
| 97 |
|
| 98 |
# See: https://github.com/rust-lang/regex/issues/169 |
| 99 |
[[test]] |
| 100 |
name = "leftmost-first-prefix" |
| 101 |
regex = 'z*azb' |
| 102 |
haystack = "azb" |
| 103 |
matches = [[0, 3]] |
| 104 |
|
| 105 |
# See: https://github.com/rust-lang/regex/issues/191 |
| 106 |
[[test]] |
| 107 |
name = "many-alternates" |
| 108 |
regex = '1|2|3|4|5|6|7|8|9|10|int' |
| 109 |
haystack = "int" |
| 110 |
matches = [[0, 3]] |
| 111 |
|
| 112 |
# See: https://github.com/rust-lang/regex/issues/204 |
| 113 |
[[test]] |
| 114 |
name = "word-boundary-alone-100" |
| 115 |
regex = '\b' |
| 116 |
haystack = "Should this (work?)" |
| 117 |
matches = [[0, 0], [6, 6], [7, 7], [11, 11], [13, 13], [17, 17]] |
| 118 |
|
| 119 |
# See: https://github.com/rust-lang/regex/issues/204 |
| 120 |
[[test]] |
| 121 |
name = "word-boundary-alone-200" |
| 122 |
regex = '\b' |
| 123 |
haystack = "a b c" |
| 124 |
matches = [[0, 0], [1, 1], [2, 2], [3, 3], [4, 4], [5, 5]] |
| 125 |
|
| 126 |
# See: https://github.com/rust-lang/regex/issues/264 |
| 127 |
[[test]] |
| 128 |
name = "word-boundary-ascii-no-capture" |
| 129 |
regex = '\B' |
| 130 |
haystack = "\U00028F3E" |
| 131 |
matches = [[0, 0], [1, 1], [2, 2], [3, 3], [4, 4]] |
| 132 |
unicode = false |
| 133 |
utf8 = false |
| 134 |
|
| 135 |
# See: https://github.com/rust-lang/regex/issues/264 |
| 136 |
[[test]] |
| 137 |
name = "word-boundary-ascii-capture" |
| 138 |
regex = '(?:\B)' |
| 139 |
haystack = "\U00028F3E" |
| 140 |
matches = [[0, 0], [1, 1], [2, 2], [3, 3], [4, 4]] |
| 141 |
unicode = false |
| 142 |
utf8 = false |
| 143 |
|
| 144 |
# See: https://github.com/rust-lang/regex/issues/268 |
| 145 |
[[test]] |
| 146 |
name = "partial-anchor" |
| 147 |
regex = '^a|b' |
| 148 |
haystack = "ba" |
| 149 |
matches = [[0, 1]] |
| 150 |
|
| 151 |
# See: https://github.com/rust-lang/regex/issues/271 |
| 152 |
[[test]] |
| 153 |
name = "endl-or-word-boundary" |
| 154 |
regex = '(?m:$)|(?-u:\b)' |
| 155 |
haystack = "\U0006084E" |
| 156 |
matches = [[4, 4]] |
| 157 |
|
| 158 |
# See: https://github.com/rust-lang/regex/issues/271 |
| 159 |
[[test]] |
| 160 |
name = "zero-or-end" |
| 161 |
regex = '(?i-u:\x00)|$' |
| 162 |
haystack = "\U000E682F" |
| 163 |
matches = [[4, 4]] |
| 164 |
|
| 165 |
# See: https://github.com/rust-lang/regex/issues/271 |
| 166 |
[[test]] |
| 167 |
name = "y-or-endl" |
| 168 |
regex = '(?i-u:y)|(?m:$)' |
| 169 |
haystack = "\U000B4331" |
| 170 |
matches = [[4, 4]] |
| 171 |
|
| 172 |
# See: https://github.com/rust-lang/regex/issues/271 |
| 173 |
[[test]] |
| 174 |
name = "word-boundary-start-x" |
| 175 |
regex = '(?u:\b)^(?-u:X)' |
| 176 |
haystack = "X" |
| 177 |
matches = [[0, 1]] |
| 178 |
|
| 179 |
# See: https://github.com/rust-lang/regex/issues/271 |
| 180 |
[[test]] |
| 181 |
name = "word-boundary-ascii-start-x" |
| 182 |
regex = '(?-u:\b)^(?-u:X)' |
| 183 |
haystack = "X" |
| 184 |
matches = [[0, 1]] |
| 185 |
|
| 186 |
# See: https://github.com/rust-lang/regex/issues/271 |
| 187 |
[[test]] |
| 188 |
name = "end-not-word-boundary" |
| 189 |
regex = '$\B' |
| 190 |
haystack = "\U0005C124\U000B576C" |
| 191 |
matches = [[8, 8]] |
| 192 |
unicode = false |
| 193 |
utf8 = false |
| 194 |
|
| 195 |
# See: https://github.com/rust-lang/regex/issues/280 |
| 196 |
[[test]] |
| 197 |
name = "partial-anchor-alternate-begin" |
| 198 |
regex = '^a|z' |
| 199 |
haystack = "yyyyya" |
| 200 |
matches = [] |
| 201 |
|
| 202 |
# See: https://github.com/rust-lang/regex/issues/280 |
| 203 |
[[test]] |
| 204 |
name = "partial-anchor-alternate-end" |
| 205 |
regex = 'a$|z' |
| 206 |
haystack = "ayyyyy" |
| 207 |
matches = [] |
| 208 |
|
| 209 |
# See: https://github.com/rust-lang/regex/issues/289 |
| 210 |
[[test]] |
| 211 |
name = "lits-unambiguous-100" |
| 212 |
regex = '(?:ABC|CDA|BC)X' |
| 213 |
haystack = "CDAX" |
| 214 |
matches = [[0, 4]] |
| 215 |
|
| 216 |
# See: https://github.com/rust-lang/regex/issues/291 |
| 217 |
[[test]] |
| 218 |
name = "lits-unambiguous-200" |
| 219 |
regex = '((IMG|CAM|MG|MB2)_|(DSCN|CIMG))(?P<n>[0-9]+)$' |
| 220 |
haystack = "CIMG2341" |
| 221 |
matches = [ |
| 222 |
[[0, 8], [0, 4], [], [0, 4], [4, 8]], |
| 223 |
] |
| 224 |
|
| 225 |
# See: https://github.com/rust-lang/regex/issues/303 |
| 226 |
# |
| 227 |
# 2022-09-19: This has now been "properly" fixed in that empty character |
| 228 |
# classes are fully supported as something that can never match. This test |
| 229 |
# used to be marked as 'compiles = false', but now it works. |
| 230 |
[[test]] |
| 231 |
name = "negated-full-byte-range" |
| 232 |
regex = '[^\x00-\xFF]' |
| 233 |
haystack = "" |
| 234 |
matches = [] |
| 235 |
compiles = true |
| 236 |
unicode = false |
| 237 |
utf8 = false |
| 238 |
|
| 239 |
# See: https://github.com/rust-lang/regex/issues/321 |
| 240 |
[[test]] |
| 241 |
name = "strange-anchor-non-complete-prefix" |
| 242 |
regex = 'a^{2}' |
| 243 |
haystack = "" |
| 244 |
matches = [] |
| 245 |
|
| 246 |
# See: https://github.com/rust-lang/regex/issues/321 |
| 247 |
[[test]] |
| 248 |
name = "strange-anchor-non-complete-suffix" |
| 249 |
regex = '${2}a' |
| 250 |
haystack = "" |
| 251 |
matches = [] |
| 252 |
|
| 253 |
# See: https://github.com/rust-lang/regex/issues/334 |
| 254 |
# See: https://github.com/rust-lang/regex/issues/557 |
| 255 |
[[test]] |
| 256 |
name = "captures-after-dfa-premature-end-100" |
| 257 |
regex = 'a(b*(X|$))?' |
| 258 |
haystack = "abcbX" |
| 259 |
matches = [ |
| 260 |
[[0, 1], [], []], |
| 261 |
] |
| 262 |
|
| 263 |
# See: https://github.com/rust-lang/regex/issues/334 |
| 264 |
# See: https://github.com/rust-lang/regex/issues/557 |
| 265 |
[[test]] |
| 266 |
name = "captures-after-dfa-premature-end-200" |
| 267 |
regex = 'a(bc*(X|$))?' |
| 268 |
haystack = "abcbX" |
| 269 |
matches = [ |
| 270 |
[[0, 1], [], []], |
| 271 |
] |
| 272 |
|
| 273 |
# See: https://github.com/rust-lang/regex/issues/334 |
| 274 |
# See: https://github.com/rust-lang/regex/issues/557 |
| 275 |
[[test]] |
| 276 |
name = "captures-after-dfa-premature-end-300" |
| 277 |
regex = '(aa$)?' |
| 278 |
haystack = "aaz" |
| 279 |
matches = [ |
| 280 |
[[0, 0], []], |
| 281 |
[[1, 1], []], |
| 282 |
[[2, 2], []], |
| 283 |
[[3, 3], []], |
| 284 |
] |
| 285 |
|
| 286 |
# Plucked from "Why aren’t regular expressions a lingua franca? an empirical |
| 287 |
# study on the re-use and portability of regular expressions", The ACM Joint |
| 288 |
# European Software Engineering Conference and Symposium on the Foundations of |
| 289 |
# Software Engineering (ESEC/FSE), 2019. |
| 290 |
# |
| 291 |
# Link: https://dl.acm.org/doi/pdf/10.1145/3338906.3338909 |
| 292 |
[[test]] |
| 293 |
name = "captures-after-dfa-premature-end-400" |
| 294 |
regex = '(a)\d*\.?\d+\b' |
| 295 |
haystack = "a0.0c" |
| 296 |
matches = [ |
| 297 |
[[0, 2], [0, 1]], |
| 298 |
] |
| 299 |
|
| 300 |
# See: https://github.com/rust-lang/regex/issues/437 |
| 301 |
[[test]] |
| 302 |
name = "literal-panic" |
| 303 |
regex = 'typename type\-parameter\-[0-9]+\-[0-9]+::.+' |
| 304 |
haystack = "test" |
| 305 |
matches = [] |
| 306 |
|
| 307 |
# See: https://github.com/rust-lang/regex/issues/527 |
| 308 |
[[test]] |
| 309 |
name = "empty-flag-expr" |
| 310 |
regex = '(?:(?:(?x)))' |
| 311 |
haystack = "" |
| 312 |
matches = [[0, 0]] |
| 313 |
|
| 314 |
# See: https://github.com/rust-lang/regex/issues/533 |
| 315 |
#[[tests]] |
| 316 |
#name = "blank-matches-nothing-between-space-and-tab" |
| 317 |
#regex = '[[:blank:]]' |
| 318 |
#input = '\x0A\x0B\x0C\x0D\x0E\x0F\x10\x11\x12\x13\x14\x15\x16\x17\x18\x19\x1A\x1B\x1C\x1D\x1E\x1F' |
| 319 |
#match = false |
| 320 |
#unescape = true |
| 321 |
|
| 322 |
# See: https://github.com/rust-lang/regex/issues/533 |
| 323 |
#[[tests]] |
| 324 |
#name = "blank-matches-nothing-between-space-and-tab-inverted" |
| 325 |
#regex = '^[[:^blank:]]+$' |
| 326 |
#input = '\x0A\x0B\x0C\x0D\x0E\x0F\x10\x11\x12\x13\x14\x15\x16\x17\x18\x19\x1A\x1B\x1C\x1D\x1E\x1F' |
| 327 |
#match = true |
| 328 |
#unescape = true |
| 329 |
|
| 330 |
# See: https://github.com/rust-lang/regex/issues/555 |
| 331 |
[[test]] |
| 332 |
name = "invalid-repetition" |
| 333 |
regex = '(?m){1,1}' |
| 334 |
haystack = "" |
| 335 |
matches = [] |
| 336 |
compiles = false |
| 337 |
|
| 338 |
# See: https://github.com/rust-lang/regex/issues/640 |
| 339 |
[[test]] |
| 340 |
name = "flags-are-unset" |
| 341 |
regex = '(?:(?i)foo)|Bar' |
| 342 |
haystack = "foo Foo bar Bar" |
| 343 |
matches = [[0, 3], [4, 7], [12, 15]] |
| 344 |
|
| 345 |
# Note that 'Ј' is not 'j', but cyrillic Je |
| 346 |
# https://en.wikipedia.org/wiki/Je_(Cyrillic) |
| 347 |
# |
| 348 |
# See: https://github.com/rust-lang/regex/issues/659 |
| 349 |
[[test]] |
| 350 |
name = "empty-group-with-unicode" |
| 351 |
regex = '(?:)Ј01' |
| 352 |
haystack = 'zЈ01' |
| 353 |
matches = [[1, 5]] |
| 354 |
|
| 355 |
# See: https://github.com/rust-lang/regex/issues/579 |
| 356 |
[[test]] |
| 357 |
name = "word-boundary-weird" |
| 358 |
regex = '\b..\b' |
| 359 |
haystack = "I have 12, he has 2!" |
| 360 |
matches = [[0, 2], [7, 9], [9, 11], [11, 13], [17, 19]] |
| 361 |
|
| 362 |
# See: https://github.com/rust-lang/regex/issues/579 |
| 363 |
[[test]] |
| 364 |
name = "word-boundary-weird-ascii" |
| 365 |
regex = '\b..\b' |
| 366 |
haystack = "I have 12, he has 2!" |
| 367 |
matches = [[0, 2], [7, 9], [9, 11], [11, 13], [17, 19]] |
| 368 |
unicode = false |
| 369 |
utf8 = false |
| 370 |
|
| 371 |
# See: https://github.com/rust-lang/regex/issues/579 |
| 372 |
[[test]] |
| 373 |
name = "word-boundary-weird-minimal-ascii" |
| 374 |
regex = '\b..\b' |
| 375 |
haystack = "az,,b" |
| 376 |
matches = [[0, 2], [2, 4]] |
| 377 |
unicode = false |
| 378 |
utf8 = false |
| 379 |
|
| 380 |
# See: https://github.com/BurntSushi/ripgrep/issues/1203 |
| 381 |
[[test]] |
| 382 |
name = "reverse-suffix-100" |
| 383 |
regex = '[0-4][0-4][0-4]000' |
| 384 |
haystack = "153.230000" |
| 385 |
matches = [[4, 10]] |
| 386 |
|
| 387 |
# See: https://github.com/BurntSushi/ripgrep/issues/1203 |
| 388 |
[[test]] |
| 389 |
name = "reverse-suffix-200" |
| 390 |
regex = '[0-9][0-9][0-9]000' |
| 391 |
haystack = "153.230000\n" |
| 392 |
matches = [[4, 10]] |
| 393 |
|
| 394 |
# This is a tricky case for the reverse suffix optimization, because it |
| 395 |
# finds the 'foobar' match but the reverse scan must fail to find a match by |
| 396 |
# correctly dealing with the word boundary following the 'foobar' literal when |
| 397 |
# computing the start state. |
| 398 |
# |
| 399 |
# This test exists because I tried to break the following assumption that |
| 400 |
# is currently in the code: that if a suffix is found and the reverse scan |
| 401 |
# succeeds, then it's guaranteed that there is an overall match. Namely, the |
| 402 |
# 'is_match' routine does *not* do another forward scan in this case because of |
| 403 |
# this assumption. |
| 404 |
[[test]] |
| 405 |
name = "reverse-suffix-300" |
| 406 |
regex = '\w+foobar\b' |
| 407 |
haystack = "xyzfoobarZ" |
| 408 |
matches = [] |
| 409 |
unicode = false |
| 410 |
utf8 = false |
| 411 |
|
| 412 |
# See: https://github.com/BurntSushi/ripgrep/issues/1247 |
| 413 |
[[test]] |
| 414 |
name = "stops" |
| 415 |
regex = '\bs(?:[ab])' |
| 416 |
haystack = 's\xE4' |
| 417 |
matches = [] |
| 418 |
unescape = true |
| 419 |
utf8 = false |
| 420 |
|
| 421 |
# See: https://github.com/BurntSushi/ripgrep/issues/1247 |
| 422 |
[[test]] |
| 423 |
name = "stops-ascii" |
| 424 |
regex = '(?-u:\b)s(?:[ab])' |
| 425 |
haystack = 's\xE4' |
| 426 |
matches = [] |
| 427 |
unescape = true |
| 428 |
utf8 = false |
| 429 |
|
| 430 |
# See: https://github.com/rust-lang/regex/issues/850 |
| 431 |
[[test]] |
| 432 |
name = "adjacent-line-boundary-100" |
| 433 |
regex = '(?m)^(?:[^ ]+?)$' |
| 434 |
haystack = "line1\nline2" |
| 435 |
matches = [[0, 5], [6, 11]] |
| 436 |
|
| 437 |
# Continued. |
| 438 |
[[test]] |
| 439 |
name = "adjacent-line-boundary-200" |
| 440 |
regex = '(?m)^(?:[^ ]+?)$' |
| 441 |
haystack = "A\nB" |
| 442 |
matches = [[0, 1], [2, 3]] |
| 443 |
|
| 444 |
# There is no issue for this bug. |
| 445 |
[[test]] |
| 446 |
name = "anchored-prefix-100" |
| 447 |
regex = '^a[[:^space:]]' |
| 448 |
haystack = "a " |
| 449 |
matches = [] |
| 450 |
|
| 451 |
# There is no issue for this bug. |
| 452 |
[[test]] |
| 453 |
name = "anchored-prefix-200" |
| 454 |
regex = '^a[[:^space:]]' |
| 455 |
haystack = "foo boo a" |
| 456 |
matches = [] |
| 457 |
|
| 458 |
# There is no issue for this bug. |
| 459 |
[[test]] |
| 460 |
name = "anchored-prefix-300" |
| 461 |
regex = '^-[a-z]' |
| 462 |
haystack = "r-f" |
| 463 |
matches = [] |
| 464 |
|
| 465 |
# Tests that a possible Aho-Corasick optimization works correctly. It only |
| 466 |
# kicks in when we have a lot of literals. By "works correctly," we mean that |
| 467 |
# leftmost-first match semantics are properly respected. That is, samwise |
| 468 |
# should match, not sam. |
| 469 |
# |
| 470 |
# There is no issue for this bug. |
| 471 |
[[test]] |
| 472 |
name = "aho-corasick-100" |
| 473 |
regex = 'samwise|sam|a|b|c|d|e|f|g|h|i|j|k|l|m|n|o|p|q|r|s|t|u|v|w|x|y|z|A|B|C|D|E|F|G|H|I|J|K|L|M|N|O|P|Q|R|S|T|U|V|W|X|Y|Z' |
| 474 |
haystack = "samwise" |
| 475 |
matches = [[0, 7]] |
| 476 |
|
| 477 |
# See: https://github.com/rust-lang/regex/issues/921 |
| 478 |
[[test]] |
| 479 |
name = "interior-anchor-capture" |
| 480 |
regex = '(a$)b$' |
| 481 |
haystack = 'ab' |
| 482 |
matches = [] |
| 483 |
|
| 484 |
# I found this bug in the course of adding some of the regexes that Ruff uses |
| 485 |
# to rebar. It turns out that the lazy DFA was finding a match that was being |
| 486 |
# rejected by the one-pass DFA. Yikes. I then minimized the regex and haystack. |
| 487 |
# |
| 488 |
# Source: https://github.com/charliermarsh/ruff/blob/a919041ddaa64cdf6f216f90dd0480dab69fd3ba/crates/ruff/src/rules/pycodestyle/rules/whitespace_around_keywords.rs#L52 |
| 489 |
[[test]] |
| 490 |
name = "ruff-whitespace-around-keywords" |
| 491 |
regex = '^(a|ab)$' |
| 492 |
haystack = "ab" |
| 493 |
anchored = true |
| 494 |
unicode = false |
| 495 |
utf8 = true |
| 496 |
matches = [[[0, 2], [0, 2]]] |
| 497 |
|
| 498 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 499 |
[[test]] |
| 500 |
name = "i429-0" |
| 501 |
regex = '(?:(?-u:\b)|(?u:h))+' |
| 502 |
haystack = "h" |
| 503 |
unicode = true |
| 504 |
utf8 = false |
| 505 |
matches = [[0, 0], [1, 1]] |
| 506 |
|
| 507 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 508 |
[[test]] |
| 509 |
name = "i429-1" |
| 510 |
regex = '(?u:\B)' |
| 511 |
haystack = "鋸" |
| 512 |
unicode = true |
| 513 |
utf8 = false |
| 514 |
matches = [] |
| 515 |
|
| 516 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 517 |
[[test]] |
| 518 |
name = "i429-2" |
| 519 |
regex = '(?:(?u:\b)|(?s-u:.))+' |
| 520 |
haystack = "oB" |
| 521 |
unicode = true |
| 522 |
utf8 = false |
| 523 |
matches = [[0, 0], [1, 2]] |
| 524 |
|
| 525 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 526 |
[[test]] |
| 527 |
name = "i429-3" |
| 528 |
regex = '(?:(?-u:\B)|(?su:.))+' |
| 529 |
haystack = "\U000FEF80" |
| 530 |
unicode = true |
| 531 |
utf8 = false |
| 532 |
matches = [[0, 0], [1, 1], [2, 2], [3, 3], [4, 4]] |
| 533 |
|
| 534 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 535 |
[[test]] |
| 536 |
name = "i429-3-utf8" |
| 537 |
regex = '(?:(?-u:\B)|(?su:.))+' |
| 538 |
haystack = "\U000FEF80" |
| 539 |
unicode = true |
| 540 |
utf8 = true |
| 541 |
matches = [[0, 0], [4, 4]] |
| 542 |
|
| 543 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 544 |
[[test]] |
| 545 |
name = "i429-4" |
| 546 |
regex = '(?m:$)(?m:^)(?su:.)' |
| 547 |
haystack = "\n‣" |
| 548 |
unicode = true |
| 549 |
utf8 = false |
| 550 |
matches = [[0, 1]] |
| 551 |
|
| 552 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 553 |
[[test]] |
| 554 |
name = "i429-5" |
| 555 |
regex = '(?m:$)^(?m:^)' |
| 556 |
haystack = "\n" |
| 557 |
unicode = true |
| 558 |
utf8 = false |
| 559 |
matches = [[0, 0]] |
| 560 |
|
| 561 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 562 |
[[test]] |
| 563 |
name = "i429-6" |
| 564 |
regex = '(?P<kp>(?iu:do)(?m:$))*' |
| 565 |
haystack = "dodo" |
| 566 |
unicode = true |
| 567 |
utf8 = false |
| 568 |
matches = [ |
| 569 |
[[0, 0], []], |
| 570 |
[[1, 1], []], |
| 571 |
[[2, 4], [2, 4]], |
| 572 |
] |
| 573 |
|
| 574 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 575 |
[[test]] |
| 576 |
name = "i429-7" |
| 577 |
regex = '(?u:\B)' |
| 578 |
haystack = "䡁" |
| 579 |
unicode = true |
| 580 |
utf8 = false |
| 581 |
matches = [] |
| 582 |
|
| 583 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 584 |
[[test]] |
| 585 |
name = "i429-8" |
| 586 |
regex = '(?:(?-u:\b)|(?u:[\u{0}-W]))+' |
| 587 |
haystack = "0" |
| 588 |
unicode = true |
| 589 |
utf8 = false |
| 590 |
matches = [[0, 0], [1, 1]] |
| 591 |
|
| 592 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 593 |
[[test]] |
| 594 |
name = "i429-9" |
| 595 |
regex = '((?m:$)(?-u:\B)(?s-u:.)(?-u:\B)$)' |
| 596 |
haystack = "\n\n" |
| 597 |
unicode = true |
| 598 |
utf8 = false |
| 599 |
matches = [ |
| 600 |
[[1, 2], [1, 2]], |
| 601 |
] |
| 602 |
|
| 603 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 604 |
[[test]] |
| 605 |
name = "i429-10" |
| 606 |
regex = '(?m:$)(?m:$)^(?su:.)' |
| 607 |
haystack = "\n\u0081¨\u200a" |
| 608 |
unicode = true |
| 609 |
utf8 = false |
| 610 |
matches = [[0, 1]] |
| 611 |
|
| 612 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 613 |
[[test]] |
| 614 |
name = "i429-11" |
| 615 |
regex = '(?-u:\B)(?m:^)' |
| 616 |
haystack = "0\n" |
| 617 |
unicode = true |
| 618 |
utf8 = false |
| 619 |
matches = [[2, 2]] |
| 620 |
|
| 621 |
# From: https://github.com/rust-lang/regex/issues/429 |
| 622 |
[[test]] |
| 623 |
name = "i429-12" |
| 624 |
regex = '(?:(?u:\b)|(?-u:.))+' |
| 625 |
haystack = "0" |
| 626 |
unicode = true |
| 627 |
utf8 = false |
| 628 |
matches = [[0, 0], [1, 1]] |
| 629 |
|
| 630 |
# From: https://github.com/rust-lang/regex/issues/969 |
| 631 |
[[test]] |
| 632 |
name = "i969" |
| 633 |
regex = 'c.*d\z' |
| 634 |
haystack = "ababcd" |
| 635 |
bounds = [4, 6] |
| 636 |
search-kind = "earliest" |
| 637 |
matches = [[4, 6]] |
| 638 |
|
| 639 |
# I found this during the regex-automata migration. This is the fowler basic |
| 640 |
# 154 test, but without anchored = true and without a match limit. |
| 641 |
# |
| 642 |
# This test caught a subtle bug in the hybrid reverse DFA search, where it |
| 643 |
# would skip over the termination condition if it entered a start state. This |
| 644 |
# was a double bug. Firstly, the reverse DFA shouldn't have had start states |
| 645 |
# specialized in the first place, and thus it shouldn't have possible to detect |
| 646 |
# that the DFA had entered a start state. The second bug was that the start |
| 647 |
# state handling was incorrect by jumping over the termination condition. |
| 648 |
[[test]] |
| 649 |
name = "fowler-basic154-unanchored" |
| 650 |
regex = '''a([bc]*)c*''' |
| 651 |
haystack = '''abc''' |
| 652 |
matches = [[[0, 3], [1, 3]]] |
| 653 |
|
| 654 |
# From: https://github.com/rust-lang/regex/issues/981 |
| 655 |
# |
| 656 |
# This was never really a problem in the new architecture because the |
| 657 |
# regex-automata engines are far more principled about how they deal with |
| 658 |
# look-around. (This was one of the many reasons I wanted to re-work the |
| 659 |
# original regex crate engines.) |
| 660 |
[[test]] |
| 661 |
name = "word-boundary-interact-poorly-with-literal-optimizations" |
| 662 |
regex = '(?i:(?:\b|_)win(?:32|64|dows)?(?:\b|_))' |
| 663 |
haystack = 'ubi-Darwin-x86_64.tar.gz' |
| 664 |
matches = [] |
| 665 |
|
| 666 |
# This was found during fuzz testing of regex. It provoked a panic in the meta |
| 667 |
# engine as a result of the reverse suffix optimization. Namely, it hit a case |
| 668 |
# where a suffix match was found, a corresponding reverse match was found, but |
| 669 |
# the forward search turned up no match. The forward search should always match |
| 670 |
# if the suffix and reverse search match. |
| 671 |
# |
| 672 |
# This in turn uncovered an inconsistency between the PikeVM and the DFA (lazy |
| 673 |
# and fully compiled) engines. It was caused by a mishandling of the collection |
| 674 |
# of NFA state IDs in the generic determinization code (which is why both types |
| 675 |
# of DFA were impacted). Namely, when a fail state was encountered (that's the |
| 676 |
# `[^\s\S]` in the pattern below), then it would just stop collecting states. |
| 677 |
# But that's not correct since a later state could lead to a match. |
| 678 |
[[test]] |
| 679 |
name = "impossible-branch" |
| 680 |
regex = '.*[^\s\S]A|B' |
| 681 |
haystack = "B" |
| 682 |
matches = [[0, 1]] |
| 683 |
|
| 684 |
# This was found during fuzz testing in regex-lite. The regex crate never |
| 685 |
# suffered from this bug, but it causes regex-lite to incorrectly compile |
| 686 |
# captures. |
| 687 |
[[test]] |
| 688 |
name = "captures-wrong-order" |
| 689 |
regex = '(a){0}(a)' |
| 690 |
haystack = 'a' |
| 691 |
matches = [[[0, 1], [], [0, 1]]] |
| 692 |
|
| 693 |
# This tests a bug in how quit states are handled in the DFA. At some point |
| 694 |
# during development, the DFAs were tweaked slightly such that if they hit |
| 695 |
# a quit state (which means, they hit a byte that the caller configured should |
| 696 |
# stop the search), then it might not return an error necessarily. Namely, if a |
| 697 |
# match had already been found, then it would be returned instead of an error. |
| 698 |
# |
| 699 |
# But this is actually wrong! Why? Because even though a match had been found, |
| 700 |
# it wouldn't be fully correct to return it once a quit state has been seen |
| 701 |
# because you can't determine whether the match offset returned is the correct |
| 702 |
# greedy/leftmost-first match. Since you can't complete the search as requested |
| 703 |
# by the caller, the DFA should just stop and return an error. |
| 704 |
# |
| 705 |
# Interestingly, this does seem to produce an unavoidable difference between |
| 706 |
# 'try_is_match().unwrap()' and 'try_find().unwrap().is_some()' for the DFAs. |
| 707 |
# The former will stop immediately once a match is known to occur and return |
| 708 |
# 'Ok(true)', where as the latter could find the match but quit with an |
| 709 |
# 'Err(..)' first. |
| 710 |
# |
| 711 |
# Thankfully, I believe this inconsistency between 'is_match()' and 'find()' |
| 712 |
# cannot be observed in the higher level meta regex API because it specifically |
| 713 |
# will try another engine that won't fail in the case of a DFA failing. |
| 714 |
# |
| 715 |
# This regression happened in the regex crate rewrite, but before anything got |
| 716 |
# released. |
| 717 |
[[test]] |
| 718 |
name = "negated-unicode-word-boundary-dfa-fail" |
| 719 |
regex = '\B.*' |
| 720 |
haystack = "!\u02D7" |
| 721 |
matches = [[0, 3]] |
| 722 |
|
| 723 |
# This failure was found in the *old* regex crate (prior to regex 1.9), but |
| 724 |
# I didn't investigate why. My best guess is that it's a literal optimization |
| 725 |
# bug. It didn't occur in the rewrite. |
| 726 |
[[test]] |
| 727 |
name = "missed-match" |
| 728 |
regex = 'e..+e.ee>' |
| 729 |
haystack = 'Zeee.eZZZZZZZZeee>eeeeeee>' |
| 730 |
matches = [[1, 26]] |
| 731 |
|
| 732 |
# This test came from the 'ignore' crate and tripped a bug in how accelerated |
| 733 |
# DFA states were handled in an overlapping search. |
| 734 |
[[test]] |
| 735 |
name = "regex-to-glob" |
| 736 |
regex = ['(?-u)^path1/[^/]*$'] |
| 737 |
haystack = "path1/foo" |
| 738 |
matches = [[0, 9]] |
| 739 |
utf8 = false |
| 740 |
match-kind = "all" |
| 741 |
search-kind = "overlapping" |
| 742 |
|
| 743 |
# See: https://github.com/rust-lang/regex/issues/1060 |
| 744 |
[[test]] |
| 745 |
name = "reverse-inner-plus-shorter-than-expected" |
| 746 |
regex = '(?:(\d+)[:.])?(\d{1,2})[:.](\d{2})' |
| 747 |
haystack = '102:12:39' |
| 748 |
matches = [[[0, 9], [0, 3], [4, 6], [7, 9]]] |
| 749 |
|
| 750 |
# Like reverse-inner-plus-shorter-than-expected, but using a far simpler regex |
| 751 |
# to demonstrate the extent of the rot. Sigh. |
| 752 |
# |
| 753 |
# See: https://github.com/rust-lang/regex/issues/1060 |
| 754 |
[[test]] |
| 755 |
name = "reverse-inner-short" |
| 756 |
regex = '(?:([0-9][0-9][0-9]):)?([0-9][0-9]):([0-9][0-9])' |
| 757 |
haystack = '102:12:39' |
| 758 |
matches = [[[0, 9], [0, 3], [4, 6], [7, 9]]] |
| 759 |
|
| 760 |
# This regression test was found via the RegexSet APIs. It triggered a |
| 761 |
# particular code path where a regex was compiled with 'All' match semantics |
| 762 |
# (to support overlapping search), but got funneled down into a standard |
| 763 |
# leftmost search when calling 'is_match'. This is fine on its own, but the |
| 764 |
# leftmost search will use a prefilter and that's where this went awry. |
| 765 |
# |
| 766 |
# Namely, since 'All' semantics were used, the aho-corasick prefilter was |
| 767 |
# incorrectly compiled with 'Standard' semantics. This was wrong because |
| 768 |
# 'Standard' immediately attempts to report a match at every position, even if |
| 769 |
# that would mean reporting a match past the leftmost match before reporting |
| 770 |
# the leftmost match. This breaks the prefilter contract of never having false |
| 771 |
# negatives and leads overall to the engine not finding a match. |
| 772 |
# |
| 773 |
# See: https://github.com/rust-lang/regex/issues/1070 |
| 774 |
[[test]] |
| 775 |
name = "prefilter-with-aho-corasick-standard-semantics" |
| 776 |
regex = '(?m)^ *v [0-9]' |
| 777 |
haystack = 'v 0' |
| 778 |
matches = [ |
| 779 |
{ id = 0, spans = [[0, 3]] }, |
| 780 |
] |
| 781 |
match-kind = "all" |
| 782 |
search-kind = "overlapping" |
| 783 |
unicode = true |
| 784 |
utf8 = true |
| 785 |
|
| 786 |
# This tests that the PikeVM and the meta regex agree on a particular regex. |
| 787 |
# This test previously failed when the ad hoc engines inside the meta engine |
| 788 |
# did not handle quit states correctly. Namely, the Unicode word boundary here |
| 789 |
# combined with a non-ASCII codepoint provokes the quit state. The ad hoc |
| 790 |
# engines were previously returning a match even after entering the quit state |
| 791 |
# if a match had been previously detected, but this is incorrect. The reason |
| 792 |
# is that if a quit state is found, then the search must give up *immediately* |
| 793 |
# because it prevents the search from finding the "proper" leftmost-first |
| 794 |
# match. If it instead returns a match that has been found, it risks reporting |
| 795 |
# an improper match, as it did in this case. |
| 796 |
# |
| 797 |
# See: https://github.com/rust-lang/regex/issues/1046 |
| 798 |
[[test]] |
| 799 |
name = "non-prefix-literal-quit-state" |
| 800 |
regex = '.+\b\n' |
| 801 |
haystack = "β77\n" |
| 802 |
matches = [[0, 5]] |
| 803 |
|
| 804 |
# This is a regression test for some errant HIR interval set operations that |
| 805 |
# were made in the regex-syntax 0.8.0 release and then reverted in 0.8.1. The |
| 806 |
# issue here is that the HIR produced from the regex had out-of-order ranges. |
| 807 |
# |
| 808 |
# See: https://github.com/rust-lang/regex/issues/1103 |
| 809 |
# Ref: https://github.com/rust-lang/regex/pull/1051 |
| 810 |
# Ref: https://github.com/rust-lang/regex/pull/1102 |
| 811 |
[[test]] |
| 812 |
name = "hir-optimization-out-of-order-class" |
| 813 |
regex = '^[[:alnum:]./-]+$' |
| 814 |
haystack = "a-b" |
| 815 |
matches = [[0, 3]] |
| 816 |
|
| 817 |
# This is a regression test for an improper reverse suffix optimization. This |
| 818 |
# occurred when I "broadened" the applicability of the optimization to include |
| 819 |
# multiple possible literal suffixes instead of only sticking to a non-empty |
| 820 |
# longest common suffix. It turns out that, at least given how the reverse |
| 821 |
# suffix optimization works, we need to stick to the longest common suffix for |
| 822 |
# now. |
| 823 |
# |
| 824 |
# See: https://github.com/rust-lang/regex/issues/1110 |
| 825 |
# See also: https://github.com/astral-sh/ruff/pull/7980 |
| 826 |
[[test]] |
| 827 |
name = 'improper-reverse-suffix-optimization' |
| 828 |
regex = '(\\N\{[^}]+})|([{}])' |
| 829 |
haystack = 'hiya \N{snowman} bye' |
| 830 |
matches = [[[5, 16], [5, 16], []]] |