| 1 | # Test out Oil's regular expression syntax. |
| 2 | |
| 3 | #### /^.$/ |
| 4 | shopt -s oil:all |
| 5 | var pat = '' |
| 6 | |
| 7 | setvar pat = /^.$/ |
| 8 | echo pat=$pat |
| 9 | |
| 10 | setvar pat = /%start dot %end/ |
| 11 | echo pat=$pat |
| 12 | |
| 13 | if ('' ~ pat) { # ERE syntax |
| 14 | echo yes |
| 15 | } else { |
| 16 | echo no |
| 17 | } |
| 18 | # $pat is same as pat |
| 19 | if ('f' ~ pat) { # ERE syntax |
| 20 | echo yes |
| 21 | } else { |
| 22 | echo no |
| 23 | } |
| 24 | |
| 25 | ## STDOUT: |
| 26 | pat=^.$ |
| 27 | pat=^.$ |
| 28 | no |
| 29 | yes |
| 30 | ## END |
| 31 | |
| 32 | |
| 33 | #### /.+/ |
| 34 | shopt -s oil:all |
| 35 | |
| 36 | var pat = /.+/ |
| 37 | echo $pat |
| 38 | |
| 39 | var s = 'foo' |
| 40 | if (s ~ pat) { # ERE syntax |
| 41 | echo yes |
| 42 | } |
| 43 | var empty = '' |
| 44 | if (empty ~ pat) { echo yes } else { echo no } |
| 45 | ## STDOUT: |
| 46 | .+ |
| 47 | yes |
| 48 | no |
| 49 | ## END |
| 50 | |
| 51 | #### Repeat {1,3} etc. |
| 52 | var pat = null |
| 53 | |
| 54 | setvar pat = /d{2}/ |
| 55 | echo $pat |
| 56 | setvar pat = /d{1,3}/ |
| 57 | echo $pat |
| 58 | setvar pat = /d{1,}/ |
| 59 | echo $pat |
| 60 | setvar pat = /d{,3}/ |
| 61 | echo $pat |
| 62 | |
| 63 | |
| 64 | ## STDOUT: |
| 65 | [[:digit:]]{2} |
| 66 | [[:digit:]]{1,3} |
| 67 | [[:digit:]]{1,} |
| 68 | [[:digit:]]{,3} |
| 69 | ## END |
| 70 | |
| 71 | |
| 72 | #### d+ digit+ ~d+ ~digit+ |
| 73 | shopt -s oil:all |
| 74 | |
| 75 | var pat = '' |
| 76 | |
| 77 | setvar pat = /d+/ |
| 78 | echo $pat |
| 79 | if ('42' ~ pat) { echo yes } |
| 80 | |
| 81 | var empty = '' |
| 82 | if (empty ~ pat) { echo yes } else { echo no } |
| 83 | |
| 84 | setvar pat = /digit+/ |
| 85 | echo $pat |
| 86 | setvar pat = /~d+/ |
| 87 | echo $pat |
| 88 | setvar pat = /~digit+/ |
| 89 | echo $pat |
| 90 | |
| 91 | |
| 92 | ## STDOUT: |
| 93 | [[:digit:]]+ |
| 94 | yes |
| 95 | no |
| 96 | [[:digit:]]+ |
| 97 | [^[:digit:]]+ |
| 98 | [^[:digit:]]+ |
| 99 | ## END |
| 100 | |
| 101 | #### Alternation and sequence |
| 102 | var pat = '' |
| 103 | setvar pat = /s d+ | w*/ |
| 104 | echo $pat |
| 105 | setvar pat = /s d+ or w*/ |
| 106 | echo $pat |
| 107 | ## STDOUT: |
| 108 | [[:space:]][[:digit:]]+|[[:alpha:][:digit:]_]* |
| 109 | [[:space:]][[:digit:]]+|[[:alpha:][:digit:]_]* |
| 110 | ## END |
| 111 | |
| 112 | #### Char Class Ranges |
| 113 | shopt -s oil:all |
| 114 | |
| 115 | var pat = '' |
| 116 | setvar pat = /[0-9 a-f]+/ |
| 117 | echo $pat |
| 118 | # This is equivalent |
| 119 | setvar pat = /['0' - '9' 'a' - 'f']+/ |
| 120 | echo $pat |
| 121 | |
| 122 | if ('0123' ~ pat) { echo yes } else { echo no } |
| 123 | if ('zzz' ~ pat) { echo yes } else { echo no } |
| 124 | if ('' ~ pat) { echo yes } else { echo no } |
| 125 | ## STDOUT: |
| 126 | [0-9a-f]+ |
| 127 | [0-9a-f]+ |
| 128 | yes |
| 129 | no |
| 130 | no |
| 131 | ## END |
| 132 | |
| 133 | #### Char Class Set |
| 134 | shopt -s oil:all |
| 135 | var pat = '' |
| 136 | |
| 137 | # This is NOT allowed |
| 138 | # setvar pat = /[a b c]+/ |
| 139 | |
| 140 | setvar pat = /['abc']+/ |
| 141 | echo $pat |
| 142 | |
| 143 | if ('cbcb' ~ pat) { echo yes } else { echo no } |
| 144 | if ('0123' ~ pat) { echo yes } else { echo no } |
| 145 | if ('' ~ pat) { echo yes } else { echo no } |
| 146 | ## STDOUT: |
| 147 | [abc]+ |
| 148 | yes |
| 149 | no |
| 150 | no |
| 151 | ## END |
| 152 | |
| 153 | #### Range with escaped characters |
| 154 | shopt -s oil:all |
| 155 | |
| 156 | var pat = null |
| 157 | |
| 158 | setvar pat = / [ \x00 - \x0f ] / |
| 159 | echo $pat | od -A n -t x1 |
| 160 | |
| 161 | ## STDOUT: |
| 162 | 5b 00 2d 0f 5d 0a |
| 163 | ## END |
| 164 | |
| 165 | |
| 166 | #### Group () |
| 167 | shopt -s oil:all |
| 168 | var pat = '' |
| 169 | |
| 170 | setvar pat = /(%start s or d d)/ |
| 171 | echo $pat |
| 172 | |
| 173 | if (' foo' ~ pat) { echo yes } else { echo no } |
| 174 | if ('-00-' ~ pat) { echo yes } else { echo no } |
| 175 | if ('foo' ~ pat) { echo yes } else { echo no } |
| 176 | |
| 177 | ## STDOUT: |
| 178 | (^[[:space:]]|[[:digit:]][[:digit:]]) |
| 179 | yes |
| 180 | yes |
| 181 | no |
| 182 | ## END |
| 183 | |
| 184 | #### Capture is acceptable as a group |
| 185 | shopt -s oil:all |
| 186 | var pat = /<%start s | d d>/ |
| 187 | echo $pat |
| 188 | ## STDOUT: |
| 189 | (^[[:space:]]|[[:digit:]][[:digit:]]) |
| 190 | ## END |
| 191 | |
| 192 | #### Named Capture Decays Without Name |
| 193 | shopt -s oil:all |
| 194 | var pat = /<d+ : month>/ |
| 195 | echo $pat |
| 196 | |
| 197 | if ('123' ~ pat) { |
| 198 | echo yes |
| 199 | } |
| 200 | |
| 201 | ## STDOUT: |
| 202 | ([[:digit:]]+) |
| 203 | yes |
| 204 | ## END |
| 205 | |
| 206 | #### Named Capture With ~ Assigns Variable |
| 207 | shopt -s oil:all |
| 208 | var pat = /<d+ : month>/ |
| 209 | echo $pat |
| 210 | |
| 211 | if ('123' ~ pat) { |
| 212 | echo yes |
| 213 | = month |
| 214 | } |
| 215 | ## STDOUT: |
| 216 | ([[:digit:]]+) |
| 217 | yes |
| 218 | TODO MONTH |
| 219 | ## END |
| 220 | |
| 221 | #### literal '' |
| 222 | shopt -s oil:all |
| 223 | var pat = '' |
| 224 | |
| 225 | setvar pat = /'abc' 'def'/ |
| 226 | echo $pat |
| 227 | |
| 228 | #setvar pat = /'abc' '^ + * ?'/ |
| 229 | #echo $pat |
| 230 | |
| 231 | if ('abcde' ~ pat) { echo yes } else { echo no } |
| 232 | if ('abcdef' ~ pat) { echo yes } else { echo no } |
| 233 | |
| 234 | ## STDOUT: |
| 235 | abcdef |
| 236 | no |
| 237 | yes |
| 238 | ## END |
| 239 | |
| 240 | #### double quoted, $x, and ${x} |
| 241 | shopt -s oil:all |
| 242 | var pat = '' |
| 243 | |
| 244 | var x = 'x' |
| 245 | var y = 'y' |
| 246 | setvar pat = / $x ${x} "abc" "$x${y}"/ |
| 247 | echo $pat |
| 248 | |
| 249 | if ('xxabcx' ~ pat) { echo yes } else { echo no } |
| 250 | if ('xxabcxyf' ~ pat) { echo yes } else { echo no } |
| 251 | |
| 252 | ## STDOUT: |
| 253 | xxabcxy |
| 254 | no |
| 255 | yes |
| 256 | ## END |
| 257 | |
| 258 | #### @splice |
| 259 | shopt -s oil:all |
| 260 | var d = /d+/; |
| 261 | var ip = / @d '.' @d '.' @d '.' @d / |
| 262 | echo $ip |
| 263 | if ('0.0.0.0' ~ ip) { echo yes } else { echo no } |
| 264 | if ('0.0.0' ~ ip) { echo yes } else { echo no } |
| 265 | ## STDOUT: |
| 266 | [[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+ |
| 267 | yes |
| 268 | no |
| 269 | ## END |
| 270 | |
| 271 | #### splice with capital letters |
| 272 | shopt -s oil:all |
| 273 | var D = /d+/; |
| 274 | var ip = / D '.' D '.' D '.' D / |
| 275 | echo $ip |
| 276 | if ('0.0.0.0' ~ ip) { echo yes } else { echo no } |
| 277 | if ('0.0.0' ~ ip) { echo yes } else { echo no } |
| 278 | ## STDOUT: |
| 279 | [[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+\.[[:digit:]]+ |
| 280 | yes |
| 281 | no |
| 282 | ## END |
| 283 | |
| 284 | #### Matching escaped tab character |
| 285 | shopt -s oil:all |
| 286 | |
| 287 | # BUG: need C strings in array literal |
| 288 | var lines=@($'aa\tbb' $'cc\tdd') |
| 289 | |
| 290 | var pat = / ('a' [\t] 'b') / |
| 291 | write pat=$pat |
| 292 | write @lines | egrep $pat |
| 293 | |
| 294 | ## stdout-json: "pat=(a[\t]b)\naa\tbb\n" |
| 295 | |
| 296 | #### Match non-ASCII byte denoted using c'\xff' |
| 297 | shopt -s oil:all |
| 298 | var pat = /[ c'\xff' ]/; |
| 299 | |
| 300 | echo $pat | od -A n -t x1 |
| 301 | if (c'\xff' ~ pat) { echo yes } else { echo no } |
| 302 | if (c'\xfe' ~ pat) { echo yes } else { echo no } |
| 303 | |
| 304 | ## STDOUT: |
| 305 | 5b ff 5d 0a |
| 306 | yes |
| 307 | no |
| 308 | ## END |
| 309 | |
| 310 | #### Match non-ASCII byte denoted using \xff |
| 311 | shopt -s oil:all |
| 312 | var pat = /[ \xff ]/; |
| 313 | |
| 314 | echo $pat | od -A n -t x1 |
| 315 | if (c'\xff' ~ pat) { echo yes } else { echo no } |
| 316 | if (c'\xfe' ~ pat) { echo yes } else { echo no } |
| 317 | |
| 318 | ## STDOUT: |
| 319 | 5b ff 5d 0a |
| 320 | yes |
| 321 | no |
| 322 | ## END |
| 323 | |
| 324 | #### ERE can express Unicode escapes that are in the ASCII range |
| 325 | shopt -s oil:all |
| 326 | var pat = /[ \u007f ]/; |
| 327 | |
| 328 | echo $pat | od -A n -t x1 |
| 329 | if (c'\x7f' ~ pat) { echo yes } else { echo no } |
| 330 | if (c'\x7e' ~ pat) { echo yes } else { echo no } |
| 331 | |
| 332 | ## STDOUT: |
| 333 | 5b 7f 5d 0a |
| 334 | yes |
| 335 | no |
| 336 | ## END |
| 337 | |
| 338 | #### ERE can't express higher Unicode escapes |
| 339 | shopt -s oil:all |
| 340 | var pat = /[ \u00ff ]/; |
| 341 | |
| 342 | echo $pat | od -A n -t x1 |
| 343 | if (c'\x7f' ~ pat) { echo yes } else { echo no } |
| 344 | if (c'\x7e' ~ pat) { echo yes } else { echo no } |
| 345 | |
| 346 | ## status: 1 |
| 347 | ## stdout-json: "" |
| 348 | |
| 349 | #### non-ASCII bytes must be singleton terms, e.g. '\x7f\xff' is disallowed |
| 350 | var bytes = c'\x7f\xff' |
| 351 | var pat = / [ $bytes ] / |
| 352 | echo $pat |
| 353 | ## status: 1 |
| 354 | ## stdout-json: "" |
| 355 | |
| 356 | #### Matching escaped tab character |
| 357 | shopt -s oil:all |
| 358 | |
| 359 | # BUG: need C strings in array literal |
| 360 | var lines=@($'aa\tbb' $'cc\tdd') |
| 361 | |
| 362 | var pat = / ('a' [\t] 'b') / |
| 363 | write pat=$pat |
| 364 | write @lines | egrep $pat |
| 365 | |
| 366 | ## stdout-json: "pat=(a[\t]b)\naa\tbb\n" |
| 367 | |
| 368 | #### Matching ] and \ and ' and " in character classes |
| 369 | shopt -s oil:all |
| 370 | |
| 371 | # BUG: need C strings in array literal |
| 372 | var lines=@( |
| 373 | 'backslash \' |
| 374 | 'rbracket ]' |
| 375 | 'lbracket [' |
| 376 | "sq '" |
| 377 | 'dq "' |
| 378 | ) |
| 379 | |
| 380 | # Weird GNU quirk: ] has to come first! |
| 381 | # []abc] works. But [abc\]] does NOT work. Stupid rule! |
| 382 | |
| 383 | var pat = / [ ']' \\ \' \" ] / |
| 384 | write pat=$pat |
| 385 | write @lines | egrep $pat |
| 386 | |
| 387 | ## STDOUT: |
| 388 | pat=[]\\'"] |
| 389 | backslash \ |
| 390 | rbracket ] |
| 391 | sq ' |
| 392 | dq " |
| 393 | ## END |
| 394 | |
| 395 | #### Matching literal hyphen in character classes |
| 396 | shopt -s oil:all |
| 397 | |
| 398 | var literal = '-' |
| 399 | var pat = / [ 'a' $literal 'b' ${literal} "-" ] / |
| 400 | write pat=$pat |
| 401 | write 'c-d' 'ab' 'cd' | grep $pat |
| 402 | ## STDOUT: |
| 403 | pat=[a\-b\-\-] |
| 404 | c-d |
| 405 | ab |
| 406 | ## END |
| 407 | |
| 408 | #### Repeated String Literal With Single Char |
| 409 | shopt -s oil:all |
| 410 | |
| 411 | var literal = 'f' |
| 412 | var pat = null |
| 413 | |
| 414 | setvar pat = / %start $literal+ %end / |
| 415 | echo $pat |
| 416 | setvar pat = / %start ($literal)+ %end / |
| 417 | echo $pat |
| 418 | |
| 419 | if ('fff' ~ pat) { echo yes } |
| 420 | if ('foo' !~ pat) { echo no } |
| 421 | |
| 422 | ## STDOUT: |
| 423 | ^f+$ |
| 424 | ^(f)+$ |
| 425 | yes |
| 426 | no |
| 427 | ## END |
| 428 | |
| 429 | #### Error when unparenthesized string of more than one character is repeated |
| 430 | shopt -s oil:all |
| 431 | |
| 432 | var literal = 'foo' |
| 433 | var pat = null |
| 434 | |
| 435 | setvar pat = / %start $literal+ %end / |
| 436 | echo $pat |
| 437 | setvar pat = / %start ($literal)+ %end / |
| 438 | echo $pat |
| 439 | |
| 440 | if ('foofoo' ~ pat) { echo yes } |
| 441 | if ('foof' !~ pat) { echo no } |
| 442 | |
| 443 | ## status: 1 |
| 444 | ## stdout-json: "" |
| 445 | |
| 446 | #### Instead of c'foo\\bar' use 'foo' \\ 'bar' |
| 447 | shopt -s oil:all |
| 448 | var pat = /'foo' \\ 'bar'/ |
| 449 | echo $pat |
| 450 | |
| 451 | if (r'foo\bar' ~ pat) { echo yes } |
| 452 | if (r'foo.bar' !~ pat) { echo no } |
| 453 | ## STDOUT: |
| 454 | foo\\bar |
| 455 | yes |
| 456 | no |
| 457 | ## END |
| 458 | |
| 459 | #### Negation of Character Class |
| 460 | shopt -s oil:all |
| 461 | |
| 462 | var pat = / ~[ a-z ] / |
| 463 | echo $pat |
| 464 | |
| 465 | if ('0' ~ pat) { echo yes } |
| 466 | if ('a' !~ pat) { echo no } |
| 467 | |
| 468 | ## STDOUT: |
| 469 | [^a-z] |
| 470 | yes |
| 471 | no |
| 472 | ## END |
| 473 | |
| 474 | #### Posix and Perl class in class literals |
| 475 | shopt -s oil:all |
| 476 | |
| 477 | var pat = null |
| 478 | |
| 479 | setvar pat = / [ space 'z' ] / |
| 480 | echo $pat |
| 481 | #setvar pat = / [ ~space 'z' ] / |
| 482 | #echo $pat |
| 483 | |
| 484 | # PROBLEM: can't negate individual POSIX classes. They would have to be a Perl |
| 485 | # class to be \D or \S. |
| 486 | # [[:space:]z] negates the whole thing! |
| 487 | # [^[:space:]] |
| 488 | |
| 489 | setvar pat = / [ digit 'z' ] / |
| 490 | echo $pat |
| 491 | #setvar pat = / [ ~digit 'z' ] / |
| 492 | #echo $pat |
| 493 | |
| 494 | ## STDOUT: |
| 495 | [[:space:]z] |
| 496 | [[:digit:]z] |
| 497 | ## END |
| 498 | |
| 499 | #### [~d] can't be negated because it's a literal character |
| 500 | setvar pat = / [ ~d 'z' ] / |
| 501 | echo $pat |
| 502 | ## status: 2 |
| 503 | ## stdout-json: "" |
| 504 | |
| 505 | #### [~digit] can't be negated in POSIX ERE (but yes in Perl) |
| 506 | var pat = null |
| 507 | setvar pat = / [ ~digit 'z' ] / |
| 508 | echo $pat |
| 509 | ## status: 1 |
| 510 | ## stdout-json: "" |
| 511 | |
| 512 | #### Long Python Example |
| 513 | |
| 514 | # https://docs.python.org/3/reference/lexical_analysis.html#integer-literals |
| 515 | |
| 516 | # integer ::= decinteger | bininteger | octinteger | hexinteger |
| 517 | # decinteger ::= nonzerodigit (["_"] digit)* | "0"+ (["_"] "0")* |
| 518 | # bininteger ::= "0" ("b" | "B") (["_"] bindigit)+ |
| 519 | # octinteger ::= "0" ("o" | "O") (["_"] octdigit)+ |
| 520 | # hexinteger ::= "0" ("x" | "X") (["_"] hexdigit)+ |
| 521 | # nonzerodigit ::= "1"..."9" |
| 522 | # digit ::= "0"..."9" |
| 523 | # bindigit ::= "0" | "1" |
| 524 | # octdigit ::= "0"..."7" |
| 525 | # hexdigit ::= digit | "a"..."f" | "A"..."F" |
| 526 | |
| 527 | shopt -s oil:all |
| 528 | |
| 529 | DecDigit = / [0-9] / |
| 530 | BinDigit = / [0-1] / |
| 531 | OctDigit = / [0-7] / |
| 532 | HexDigit = / [0-9 a-f A-F] / # note: not splicing Digit into character class |
| 533 | |
| 534 | DecInt = / [1-9] ('_'? DecDigit)* | '0'+ ('_'? '0')* / |
| 535 | BinInt = / '0' [b B] ('_'? BinDigit)+ / |
| 536 | OctInt = / '0' [o O] ('_'? OctDigit)+ / |
| 537 | HexInt = / '0' [x X] ('_'? HexDigit)+ / |
| 538 | |
| 539 | Integer = / %start (DecInt | BinInt | OctInt | HexInt) %end / |
| 540 | |
| 541 | #echo $Integer |
| 542 | |
| 543 | if ( '123' ~ Integer) { echo 'Y' } |
| 544 | if ( 'zzz' !~ Integer) { echo 'N' } |
| 545 | |
| 546 | if ('123_000' ~ Integer) { echo 'Y decimal' } |
| 547 | if ('000_123' !~ Integer) { echo 'N decimal' } |
| 548 | |
| 549 | if ( '0b100' ~ Integer) { echo 'Y binary' } |
| 550 | if ( '0b102' !~ Integer) { echo 'N binary' } |
| 551 | |
| 552 | if ( '0o755' ~ Integer) { echo 'Y octal' } |
| 553 | if ( '0o778' !~ Integer) { echo 'N octal' } |
| 554 | |
| 555 | if ( '0xFF' ~ Integer) { echo 'Y hex' } |
| 556 | if ( '0xFG' !~ Integer) { echo 'N hex' } |
| 557 | |
| 558 | ## STDOUT: |
| 559 | Y |
| 560 | N |
| 561 | Y decimal |
| 562 | N decimal |
| 563 | Y binary |
| 564 | N binary |
| 565 | Y octal |
| 566 | N octal |
| 567 | Y hex |
| 568 | N hex |
| 569 | ## END |
| 570 | |
| 571 | #### Invalid sh operation on eggex |
| 572 | var pat = / d+ / |
| 573 | #pat[invalid]=1 |
| 574 | pat[invalid]+=1 |
| 575 | ## status: 1 |
| 576 | ## stdout-json: "" |