| 1 | ## oils_failures_allowed: 0
|
| 2 |
|
| 3 | #### s ~ regex and s !~ regex
|
| 4 | shopt -s ysh:upgrade
|
| 5 |
|
| 6 | var s = 'foo'
|
| 7 | if (s ~ '.([[:alpha:]]+)') { # ERE syntax
|
| 8 | echo matches
|
| 9 | argv.py $[_group(0)] $[_group(1)]
|
| 10 | }
|
| 11 | if (s !~ '[[:digit:]]+') {
|
| 12 | echo "does not match"
|
| 13 | argv.py $[_group(0)] $[_group(1)]
|
| 14 | }
|
| 15 |
|
| 16 | if (s ~ '[[:digit:]]+') {
|
| 17 | echo "matches"
|
| 18 | }
|
| 19 | # Should be cleared now
|
| 20 | # should this be Undef rather than ''?
|
| 21 | try {
|
| 22 | var x = _group(0)
|
| 23 | }
|
| 24 | if (_status === 3) {
|
| 25 | echo 'got expected status 3'
|
| 26 | }
|
| 27 |
|
| 28 | try {
|
| 29 | var y = _group(1)
|
| 30 | }
|
| 31 | if (_status === 3) {
|
| 32 | echo 'got expected status 3'
|
| 33 | }
|
| 34 |
|
| 35 | ## STDOUT:
|
| 36 | matches
|
| 37 | ['foo', 'oo']
|
| 38 | does not match
|
| 39 | ['foo', 'oo']
|
| 40 | got expected status 3
|
| 41 | got expected status 3
|
| 42 | ## END
|
| 43 |
|
| 44 | #### Invalid regex has libc error message
|
| 45 |
|
| 46 | shopt -s ysh:upgrade
|
| 47 |
|
| 48 | # Hm it's hard to test this, we can't get stderr of YSH from within YSH?
|
| 49 | #redir 2>err.txt {
|
| 50 | # if ('abc' ~ '+') {
|
| 51 | # echo 'bad'
|
| 52 | # }
|
| 53 | #}
|
| 54 |
|
| 55 | if ('abc' ~ '+') {
|
| 56 | echo 'bad'
|
| 57 | }
|
| 58 |
|
| 59 | ## status: 2
|
| 60 | ## STDOUT:
|
| 61 | ## END
|
| 62 |
|
| 63 | #### Eggex flags to ignore case are respected
|
| 64 | shopt -s ysh:upgrade
|
| 65 |
|
| 66 | # based on Python's spelling
|
| 67 | var pat = / 'abc' ; i /
|
| 68 | var pat2 = / @pat 'def' ; reg_icase / # this is allowed
|
| 69 |
|
| 70 | if ('-abcdef-' ~ pat2) {
|
| 71 | echo 'yes'
|
| 72 | }
|
| 73 |
|
| 74 | if ('-ABCDEF-' ~ pat2) {
|
| 75 | echo 'yes'
|
| 76 | }
|
| 77 |
|
| 78 | if ('ABCDE' ~ pat2) {
|
| 79 | echo 'BUG'
|
| 80 | }
|
| 81 |
|
| 82 | ## STDOUT:
|
| 83 | yes
|
| 84 | yes
|
| 85 | ## END
|
| 86 |
|
| 87 | #### Eggex flags to treat newlines as special are respected
|
| 88 | shopt -s ysh:upgrade
|
| 89 |
|
| 90 | if (u'abc123\n' ~ / digit %end /) {
|
| 91 | echo 'BUG'
|
| 92 | }
|
| 93 | if (u'abc\n123' ~ / %start digit /) {
|
| 94 | echo 'BUG'
|
| 95 | }
|
| 96 |
|
| 97 | if (u'abc123\n' ~ / digit %end ; reg_newline /) {
|
| 98 | echo 'yes'
|
| 99 | }
|
| 100 | if (u'abc\n123' ~ / %start digit ; reg_newline /) {
|
| 101 | echo 'yes'
|
| 102 | }
|
| 103 |
|
| 104 | if (u'\n' ~ / . /) {
|
| 105 | echo 'yes'
|
| 106 | }
|
| 107 | if (u'\n' ~ / !digit /) {
|
| 108 | echo 'yes'
|
| 109 | }
|
| 110 |
|
| 111 | if (u'\n' ~ / . ; reg_newline /) {
|
| 112 | echo 'BUG'
|
| 113 | }
|
| 114 | if (u'\n' ~ / !digit ; reg_newline /) {
|
| 115 | echo 'BUG'
|
| 116 | }
|
| 117 |
|
| 118 | ## STDOUT:
|
| 119 | yes
|
| 120 | yes
|
| 121 | yes
|
| 122 | yes
|
| 123 | ## END
|
| 124 |
|
| 125 | #### leftMatch() combined with REG_NEWLINE - interaction with ^
|
| 126 | shopt --set ysh:upgrade
|
| 127 |
|
| 128 | # REG_NEWLINE
|
| 129 | # . does NOT match newline
|
| 130 | # [^a] does NOT match newline
|
| 131 | # ^ matches the empty string after a newline (regardless of REG_NOTBOL)
|
| 132 | # $ matches the empty string before a newline (regardless of REG_NOTBOL)
|
| 133 | #
|
| 134 | # note: REG_NOTBOL is not used in leftMatch()
|
| 135 |
|
| 136 | var lines = '''
|
| 137 | one
|
| 138 | 2
|
| 139 | three
|
| 140 | '''
|
| 141 |
|
| 142 | if (0) {
|
| 143 | = lines
|
| 144 | = ${p}
|
| 145 | echo ---
|
| 146 | }
|
| 147 |
|
| 148 | proc show-matches(; pattern) {
|
| 149 | var pos = 0
|
| 150 | while (true) {
|
| 151 | var m = lines.leftMatch(pattern, pos=pos)
|
| 152 | #var m = lines.search(pattern, pos=pos)
|
| 153 |
|
| 154 | if (m is null) {
|
| 155 | return
|
| 156 | }
|
| 157 |
|
| 158 | #pp test_ ([m.group(0), m.group(1), m.group(2)])
|
| 159 | pp test_ ([pos, m.group(0)])
|
| 160 |
|
| 161 | setvar pos = m.end(0)
|
| 162 | }
|
| 163 | }
|
| 164 |
|
| 165 | var p1 = / <capture [a-z] as letter> /
|
| 166 | var p2 = / <capture [a-z] as letter> ; ; reg_newline /
|
| 167 |
|
| 168 | show-matches (p1)
|
| 169 | echo
|
| 170 | show-matches (p2)
|
| 171 |
|
| 172 | ## STDOUT:
|
| 173 | (List) [0,"o"]
|
| 174 | (List) [1,"n"]
|
| 175 | (List) [2,"e"]
|
| 176 |
|
| 177 | (List) [0,"o"]
|
| 178 | (List) [1,"n"]
|
| 179 | (List) [2,"e"]
|
| 180 | ## END
|
| 181 |
|
| 182 | #### Bug fix: leftMatch() can be used with a|b (ERE precedence is respected)
|
| 183 | shopt --set ysh:upgrade
|
| 184 |
|
| 185 | var lines = '''
|
| 186 | one
|
| 187 | 2
|
| 188 | three
|
| 189 | '''
|
| 190 |
|
| 191 | proc show-matches(; pattern) {
|
| 192 | var pos = 0
|
| 193 | while (true) {
|
| 194 | var m = lines.leftMatch(pattern, pos=pos)
|
| 195 | #var m = lines.search(pattern, pos=pos)
|
| 196 |
|
| 197 | if (m is null) {
|
| 198 | return
|
| 199 | }
|
| 200 |
|
| 201 | #pp test_ ([m.group(0), m.group(1), m.group(2)])
|
| 202 | pp test_ ([pos, m.group(0)])
|
| 203 |
|
| 204 | setvar pos = m.end(0)
|
| 205 | }
|
| 206 | }
|
| 207 |
|
| 208 | # Alternation requires special logic
|
| 209 | var p3 = / <capture [0-9] as number> | <capture [a-z] as letter> /
|
| 210 | var p4 = / <capture [0-9] as number> | <capture [a-z] as letter> ; ; reg_newline /
|
| 211 |
|
| 212 | var m = lines.leftMatch(p3, pos=3)
|
| 213 | assert [m is null]
|
| 214 |
|
| 215 | var m = lines.leftMatch(p3, pos=4)
|
| 216 | pp test_ ([m.group(0), m.group(1), m.group(2)])
|
| 217 |
|
| 218 | var m = lines.leftMatch(p3, pos=5)
|
| 219 | assert [m is null]
|
| 220 |
|
| 221 | var m = lines.leftMatch(p3, pos=6)
|
| 222 | pp test_ ([m.group(0), m.group(1), m.group(2)])
|
| 223 |
|
| 224 | echo
|
| 225 | show-matches (p3)
|
| 226 |
|
| 227 | echo
|
| 228 | show-matches (p4)
|
| 229 |
|
| 230 | ## STDOUT:
|
| 231 | (List) ["2","2",null]
|
| 232 | (List) ["t",null,"t"]
|
| 233 |
|
| 234 | (List) [0,"o"]
|
| 235 | (List) [1,"n"]
|
| 236 | (List) [2,"e"]
|
| 237 |
|
| 238 | (List) [0,"o"]
|
| 239 | (List) [1,"n"]
|
| 240 | (List) [2,"e"]
|
| 241 | ## END
|
| 242 |
|
| 243 | #### Positional captures with _group
|
| 244 | shopt -s ysh:upgrade
|
| 245 |
|
| 246 | var x = 'zz 2020-08-20'
|
| 247 |
|
| 248 | if [[ $x =~ ([[:digit:]]+)-([[:digit:]]+) ]] {
|
| 249 | argv.py "${BASH_REMATCH[@]}"
|
| 250 | }
|
| 251 |
|
| 252 | # THIS IS A NO-OP. The variable is SHADOWED by the special name.
|
| 253 | # I think that's OK.
|
| 254 | setvar BASH_REMATCH = :| reset |
|
| 255 |
|
| 256 | if (x ~ /<capture d+> '-' <capture d+>/) {
|
| 257 | argv.py "${BASH_REMATCH[@]}"
|
| 258 | argv.py $[_group(0)] $[_group(1)] $[_group(2)]
|
| 259 |
|
| 260 | # TODO: Also test _start() and _end()
|
| 261 | }
|
| 262 | ## STDOUT:
|
| 263 | ['2020-08', '2020', '08']
|
| 264 | ['2020-08', '2020', '08']
|
| 265 | ['2020-08', '2020', '08']
|
| 266 | ## END
|
| 267 |
|
| 268 | #### _group() returns null when group doesn't match
|
| 269 | shopt -s ysh:upgrade
|
| 270 |
|
| 271 | var pat = / <capture 'a'> | <capture 'b'> /
|
| 272 | if ('b' ~ pat) {
|
| 273 | echo "$[_group(1)] $[_group(2)]"
|
| 274 | }
|
| 275 | ## STDOUT:
|
| 276 | null b
|
| 277 | ## END
|
| 278 |
|
| 279 | #### _start() and _end()
|
| 280 | shopt -s ysh:upgrade
|
| 281 |
|
| 282 | var s = 'foo123bar'
|
| 283 | if (s ~ /digit+/) {
|
| 284 | echo start=$[_start(0)] end=$[_end(0)]
|
| 285 | }
|
| 286 | echo ---
|
| 287 |
|
| 288 | if (s ~ / <capture [a-z]+> <capture digit+> /) {
|
| 289 | echo start=$[_start(1)] end=$[_end(1)]
|
| 290 | echo start=$[_start(2)] end=$[_end(2)]
|
| 291 | }
|
| 292 | echo ---
|
| 293 |
|
| 294 | if (s ~ / <capture [a-z]+> | <capture digit+> /) {
|
| 295 | echo start=$[_start(1)] end=$[_end(1)]
|
| 296 | echo start=$[_start(2)] end=$[_end(2)]
|
| 297 | }
|
| 298 |
|
| 299 | ## STDOUT:
|
| 300 | start=3 end=6
|
| 301 | ---
|
| 302 | start=0 end=3
|
| 303 | start=3 end=6
|
| 304 | ---
|
| 305 | start=0 end=3
|
| 306 | start=-1 end=-1
|
| 307 | ## END
|
| 308 |
|
| 309 | #### Str->search() method returns value.Match object
|
| 310 |
|
| 311 | var s = '= Hi5- Bye6-'
|
| 312 |
|
| 313 | var m = s.search(/ <capture [a-z]+ > <capture d+> '-' ; i /)
|
| 314 | echo "g0 $[m.start(0)] $[m.end(0)] $[m.group(0)]"
|
| 315 | echo "g1 $[m.start(1)] $[m.end(1)] $[m.group(1)]"
|
| 316 | echo "g2 $[m.start(2)] $[m.end(2)] $[m.group(2)]"
|
| 317 |
|
| 318 | echo ---
|
| 319 |
|
| 320 | var pos = m.end(0) # search from end position
|
| 321 | var m = s.search(/ <capture [a-z]+ > <capture d+> '-' ; i /, pos=pos)
|
| 322 | echo "g0 $[m.start(0)] $[m.end(0)] $[m.group(0)]"
|
| 323 | echo "g1 $[m.start(1)] $[m.end(1)] $[m.group(1)]"
|
| 324 | echo "g2 $[m.start(2)] $[m.end(2)] $[m.group(2)]"
|
| 325 |
|
| 326 | ## STDOUT:
|
| 327 | g0 2 6 Hi5-
|
| 328 | g1 2 4 Hi
|
| 329 | g2 4 5 5
|
| 330 | ---
|
| 331 | g0 7 12 Bye6-
|
| 332 | g1 7 10 Bye
|
| 333 | g2 10 11 6
|
| 334 | ## END
|
| 335 |
|
| 336 | #### Str->search() only matches %start ^ when pos == 0
|
| 337 |
|
| 338 | shopt -s ysh:upgrade
|
| 339 |
|
| 340 | var anchored = / %start <capture d+> '-' /
|
| 341 | var free = / <capture d+> '-' /
|
| 342 |
|
| 343 | var s = '12-34-'
|
| 344 |
|
| 345 | for pat in ([anchored, free]) {
|
| 346 | echo "pat=$pat"
|
| 347 |
|
| 348 | var pos = 0
|
| 349 | while (true) {
|
| 350 | var m = s.search(pat, pos=pos)
|
| 351 | if (not m) {
|
| 352 | break
|
| 353 | }
|
| 354 | echo $[m.group(0)]
|
| 355 | setvar pos = m.end(0)
|
| 356 | }
|
| 357 |
|
| 358 | }
|
| 359 |
|
| 360 | ## STDOUT:
|
| 361 | pat=^([[:digit:]]+)-
|
| 362 | 12-
|
| 363 | pat=([[:digit:]]+)-
|
| 364 | 12-
|
| 365 | 34-
|
| 366 | ## END
|
| 367 |
|
| 368 |
|
| 369 | #### search() and leftMatch() accept ERE string
|
| 370 |
|
| 371 | var s = '= hi5- bye6-'
|
| 372 |
|
| 373 | var m = s.search('([[:alpha:]]+)([[:digit:]]+)-')
|
| 374 | echo "g0 $[m.start(0)] $[m.end(0)] $[m.group(0)]"
|
| 375 | echo "g1 $[m.start(1)] $[m.end(1)] $[m.group(1)]"
|
| 376 | echo "g2 $[m.start(2)] $[m.end(2)] $[m.group(2)]"
|
| 377 | echo ---
|
| 378 |
|
| 379 | var m = s[2:].leftMatch('([[:alpha:]]+)([[:digit:]]+)-')
|
| 380 | echo "g0 $[m.start(0)] $[m.end(0)] $[m.group(0)]"
|
| 381 | echo "g1 $[m.start(1)] $[m.end(1)] $[m.group(1)]"
|
| 382 | echo "g2 $[m.start(2)] $[m.end(2)] $[m.group(2)]"
|
| 383 |
|
| 384 | ## STDOUT:
|
| 385 | g0 2 6 hi5-
|
| 386 | g1 2 4 hi
|
| 387 | g2 4 5 5
|
| 388 | ---
|
| 389 | g0 0 4 hi5-
|
| 390 | g1 0 2 hi
|
| 391 | g2 2 3 5
|
| 392 | ## END
|
| 393 |
|
| 394 | #### Str.leftMatch() can implement lexer pattern
|
| 395 |
|
| 396 | shopt -s ysh:upgrade
|
| 397 |
|
| 398 | var lexer = / <capture d+> | <capture [a-z]+> | <capture s+> /
|
| 399 | #echo $lexer
|
| 400 |
|
| 401 | proc show-tokens (s) {
|
| 402 | var pos = 0
|
| 403 |
|
| 404 | while (true) {
|
| 405 | echo "pos=$pos"
|
| 406 |
|
| 407 | var m = s.leftMatch(lexer, pos=pos)
|
| 408 | if (not m) {
|
| 409 | break
|
| 410 | }
|
| 411 | # TODO: add groups()
|
| 412 | #var groups = [m.group(1), m.group(2), m.group(3)]
|
| 413 | echo "$[m.group(1)]/$[m.group(2)]/$[m.group(3)]/"
|
| 414 |
|
| 415 | echo
|
| 416 |
|
| 417 | setvar pos = m.end(0)
|
| 418 | }
|
| 419 | }
|
| 420 |
|
| 421 | show-tokens 'ab 12'
|
| 422 |
|
| 423 | echo '==='
|
| 424 |
|
| 425 | # There's a token here that doesn't leftMatch()
|
| 426 | show-tokens 'ab+12'
|
| 427 |
|
| 428 | ## STDOUT:
|
| 429 | pos=0
|
| 430 | null/ab/null/
|
| 431 |
|
| 432 | pos=2
|
| 433 | null/null/ /
|
| 434 |
|
| 435 | pos=3
|
| 436 | 12/null/null/
|
| 437 |
|
| 438 | pos=5
|
| 439 | ===
|
| 440 | pos=0
|
| 441 | null/ab/null/
|
| 442 |
|
| 443 | pos=2
|
| 444 | ## END
|
| 445 |
|
| 446 | #### Named captures with m.group()
|
| 447 | shopt -s ysh:all
|
| 448 |
|
| 449 | var s = 'zz 2020-08-20'
|
| 450 | var pat = /<capture d+ as year> '-' <capture d+ as month>/
|
| 451 |
|
| 452 | var m = s.search(pat)
|
| 453 | argv.py $[m.group('year')] $[m.group('month')]
|
| 454 | echo $[m.start('year')] $[m.end('year')]
|
| 455 | echo $[m.start('month')] $[m.end('month')]
|
| 456 |
|
| 457 | argv.py $[m.group('oops')]
|
| 458 | echo 'error'
|
| 459 |
|
| 460 | ## status: 3
|
| 461 | ## STDOUT:
|
| 462 | ['2020', '08']
|
| 463 | 3 7
|
| 464 | 8 10
|
| 465 | ## END
|
| 466 |
|
| 467 | #### Named captures with _group() _start() _end()
|
| 468 | shopt -s ysh:all
|
| 469 |
|
| 470 | var x = 'zz 2020-08-20'
|
| 471 |
|
| 472 | if (x ~ /<capture d+ as year> '-' <capture d+ as month>/) {
|
| 473 | argv.py $[_group('year')] $[_group('month')]
|
| 474 | echo $[_start('year')] $[_end('year')]
|
| 475 | echo $[_start('month')] $[_end('month')]
|
| 476 | }
|
| 477 |
|
| 478 | argv.py $[_group('oops')]
|
| 479 |
|
| 480 | ## status: 3
|
| 481 | ## STDOUT:
|
| 482 | ['2020', '08']
|
| 483 | 3 7
|
| 484 | 8 10
|
| 485 | ## END
|
| 486 |
|
| 487 | #### Named Capture Decays Without Name
|
| 488 | shopt -s ysh:all
|
| 489 | var pat = /<capture d+ as month>/
|
| 490 | echo $pat
|
| 491 |
|
| 492 | if ('123' ~ pat) {
|
| 493 | echo yes
|
| 494 | }
|
| 495 |
|
| 496 | ## STDOUT:
|
| 497 | ([[:digit:]]+)
|
| 498 | yes
|
| 499 | ## END
|
| 500 |
|
| 501 | #### Nested Named Capture Uses ( ordering
|
| 502 |
|
| 503 | shopt -s ysh:upgrade
|
| 504 |
|
| 505 | var Date = /<capture d+ as year> '-' <capture d+ as month>/
|
| 506 | var Time = /<capture d+ as hour> ':' <capture d+ as minute> (':' <capture d+ as secs>)? /
|
| 507 |
|
| 508 | var pat = / 'when: ' (<capture Date> | <capture Time as two>) /
|
| 509 | #echo $pat
|
| 510 |
|
| 511 | proc show-groups (; m) {
|
| 512 | echo 0 $[m.group(0)]
|
| 513 | echo 1 $[m.group(1)] # this is everything except when
|
| 514 | echo 2 $[m.group(2)]
|
| 515 | echo
|
| 516 | echo $[m.group('two')]
|
| 517 | echo $[m.group('year')] $[m.group('month')]
|
| 518 | echo $[m.group('hour')] $[m.group('minute')] $[m.group('secs')]
|
| 519 | }
|
| 520 |
|
| 521 | var m = 'when: 2023-10'.leftMatch(pat)
|
| 522 |
|
| 523 | show-groups (m)
|
| 524 |
|
| 525 | var m = 'when: 23:30'.leftMatch(pat)
|
| 526 |
|
| 527 | echo ---
|
| 528 | show-groups (m)
|
| 529 |
|
| 530 | var m = 'when: 23:30:59'.leftMatch(pat)
|
| 531 |
|
| 532 | echo ---
|
| 533 | show-groups (m)
|
| 534 |
|
| 535 | ## STDOUT:
|
| 536 | 0 when: 2023-10
|
| 537 | 1 2023-10
|
| 538 | 2 2023-10
|
| 539 |
|
| 540 | null
|
| 541 | 2023 10
|
| 542 | null null null
|
| 543 | ---
|
| 544 | 0 when: 23:30
|
| 545 | 1 23:30
|
| 546 | 2 null
|
| 547 |
|
| 548 | 23:30
|
| 549 | null null
|
| 550 | 23 30 null
|
| 551 | ---
|
| 552 | 0 when: 23:30:59
|
| 553 | 1 23:30:59
|
| 554 | 2 null
|
| 555 |
|
| 556 | 23:30:59
|
| 557 | null null
|
| 558 | 23 30 59
|
| 559 | ## END
|
| 560 |
|
| 561 | #### Capture with Type Conversion Func
|
| 562 | shopt -s ysh:upgrade
|
| 563 |
|
| 564 | var s = 'hi 42-3.14'
|
| 565 | var pat = / <capture d+: int> '-' <capture d+ '.' d+ : float> /
|
| 566 |
|
| 567 | if (s ~ pat) {
|
| 568 | var g1 = _group(1) # Int
|
| 569 | var g2 = _group(2) # Float
|
| 570 | echo $[type(g1)] $[type(g2)]
|
| 571 | }
|
| 572 |
|
| 573 | var m = s.search(pat)
|
| 574 | if (m) {
|
| 575 | echo $[m.group(1) => type()] $[m.group(2) => type()]
|
| 576 | }
|
| 577 |
|
| 578 | ## STDOUT:
|
| 579 | Int Float
|
| 580 | Int Float
|
| 581 | ## END
|
| 582 |
|
| 583 |
|
| 584 | #### Named Capture with Type Conversion Func
|
| 585 | shopt -s ysh:upgrade
|
| 586 |
|
| 587 | func floatNegate(x) {
|
| 588 | return (-float(x))
|
| 589 | }
|
| 590 |
|
| 591 | var s = 'hi 42-3.14'
|
| 592 | var pat = / <capture d+ as left: int> '-' <capture d+ '.' d+ as right: floatNegate> /
|
| 593 |
|
| 594 | if (s ~ pat) {
|
| 595 | var g1 = _group('left') # Int
|
| 596 | var g2 = _group('right') # Float
|
| 597 | echo $g2
|
| 598 | echo $[type(g1)] $[type(g2)]
|
| 599 | }
|
| 600 |
|
| 601 | var m = s.search(pat)
|
| 602 | if (m) {
|
| 603 | echo $[m.group('right')]
|
| 604 | echo $[m.group('left') => type()] $[m.group('right') => type()]
|
| 605 | }
|
| 606 |
|
| 607 | ## STDOUT:
|
| 608 | -3.14
|
| 609 | Int Float
|
| 610 | -3.14
|
| 611 | Int Float
|
| 612 | ## END
|
| 613 |
|
| 614 | #### Can't splice eggex with different flags
|
| 615 | shopt -s ysh:upgrade
|
| 616 |
|
| 617 | var pat = / 'abc' ; i /
|
| 618 | var pat2 = / @pat 'def' ; reg_icase / # this is allowed
|
| 619 |
|
| 620 | var pat3 = / @pat 'def' /
|
| 621 | = pat3
|
| 622 |
|
| 623 | ## status: 1
|
| 624 | ## STDOUT:
|
| 625 | ## END
|
| 626 |
|
| 627 | #### Eggex with translation preference has arbitrary flags
|
| 628 | shopt -s ysh:upgrade
|
| 629 |
|
| 630 | # TODO: can provide introspection so users can translate it?
|
| 631 | # This is kind of a speculative corner of the language.
|
| 632 |
|
| 633 | var pat = / d+ ; ignorecase ; PCRE /
|
| 634 |
|
| 635 | # This uses ERE, as a test
|
| 636 | if ('ab 12' ~ pat) {
|
| 637 | echo yes
|
| 638 | }
|
| 639 |
|
| 640 | ## STDOUT:
|
| 641 | yes
|
| 642 | ## END
|
| 643 |
|
| 644 |
|
| 645 | #### Invalid sh operation on eggex
|
| 646 | var pat = / d+ /
|
| 647 | #pat[invalid]=1
|
| 648 | pat[invalid]+=1
|
| 649 | ## status: 1
|
| 650 | ## stdout-json: ""
|
| 651 |
|
| 652 | #### Long Python Example
|
| 653 |
|
| 654 | # https://docs.python.org/3/reference/lexical_analysis.html#integer-literals
|
| 655 |
|
| 656 | # integer ::= decinteger | bininteger | octinteger | hexinteger
|
| 657 | # decinteger ::= nonzerodigit (["_"] digit)* | "0"+ (["_"] "0")*
|
| 658 | # bininteger ::= "0" ("b" | "B") (["_"] bindigit)+
|
| 659 | # octinteger ::= "0" ("o" | "O") (["_"] octdigit)+
|
| 660 | # hexinteger ::= "0" ("x" | "X") (["_"] hexdigit)+
|
| 661 | # nonzerodigit ::= "1"..."9"
|
| 662 | # digit ::= "0"..."9"
|
| 663 | # bindigit ::= "0" | "1"
|
| 664 | # octdigit ::= "0"..."7"
|
| 665 | # hexdigit ::= digit | "a"..."f" | "A"..."F"
|
| 666 |
|
| 667 | shopt -s ysh:all
|
| 668 |
|
| 669 | const DecDigit = / [0-9] /
|
| 670 | const BinDigit = / [0-1] /
|
| 671 | const OctDigit = / [0-7] /
|
| 672 | const HexDigit = / [0-9 a-f A-F] / # note: not splicing Digit into character class
|
| 673 |
|
| 674 | const DecInt = / [1-9] ('_'? DecDigit)* | '0'+ ('_'? '0')* /
|
| 675 | const BinInt = / '0' [b B] ('_'? BinDigit)+ /
|
| 676 | const OctInt = / '0' [o O] ('_'? OctDigit)+ /
|
| 677 | const HexInt = / '0' [x X] ('_'? HexDigit)+ /
|
| 678 |
|
| 679 | const Integer = / %start (DecInt | BinInt | OctInt | HexInt) %end /
|
| 680 |
|
| 681 | #echo $Integer
|
| 682 |
|
| 683 | if ( '123' ~ Integer) { echo 'Y' }
|
| 684 | if ( 'zzz' !~ Integer) { echo 'N' }
|
| 685 |
|
| 686 | if ('123_000' ~ Integer) { echo 'Y decimal' }
|
| 687 | if ('000_123' !~ Integer) { echo 'N decimal' }
|
| 688 |
|
| 689 | if ( '0b100' ~ Integer) { echo 'Y binary' }
|
| 690 | if ( '0b102' !~ Integer) { echo 'N binary' }
|
| 691 |
|
| 692 | if ( '0o755' ~ Integer) { echo 'Y octal' }
|
| 693 | if ( '0o778' !~ Integer) { echo 'N octal' }
|
| 694 |
|
| 695 | if ( '0xFF' ~ Integer) { echo 'Y hex' }
|
| 696 | if ( '0xFG' !~ Integer) { echo 'N hex' }
|
| 697 |
|
| 698 | ## STDOUT:
|
| 699 | Y
|
| 700 | N
|
| 701 | Y decimal
|
| 702 | N decimal
|
| 703 | Y binary
|
| 704 | N binary
|
| 705 | Y octal
|
| 706 | N octal
|
| 707 | Y hex
|
| 708 | N hex
|
| 709 | ## END
|
| 710 |
|
| 711 | #### Regex in a loop (bug regression)
|
| 712 |
|
| 713 | shopt --set ysh:all
|
| 714 |
|
| 715 | var content = [ 1, 2 ]
|
| 716 | var i = 0
|
| 717 | while (i < len(content)) {
|
| 718 | var line = content[i]
|
| 719 | write $[content[i]]
|
| 720 | if (str(line) ~ / s* 'imports' s* '=' s* .* /) {
|
| 721 | exit
|
| 722 | }
|
| 723 | setvar i += 1
|
| 724 | }
|
| 725 |
|
| 726 | ## STDOUT:
|
| 727 | 1
|
| 728 | 2
|
| 729 | ## END
|
| 730 |
|
| 731 |
|
| 732 | #### Regex in a loop depending on var
|
| 733 |
|
| 734 | shopt --set ysh:all
|
| 735 |
|
| 736 | var lines = ['foo', 'bar']
|
| 737 | for line in (lines) {
|
| 738 | write "line $line"
|
| 739 |
|
| 740 | # = / $line /
|
| 741 |
|
| 742 | if ("x$line" ~ / dot @line /) {
|
| 743 | #if (line ~ / $line /) {
|
| 744 | write "matched $line"
|
| 745 | }
|
| 746 | }
|
| 747 |
|
| 748 | ## STDOUT:
|
| 749 | line foo
|
| 750 | matched foo
|
| 751 | line bar
|
| 752 | matched bar
|
| 753 | ## END
|
| 754 |
|
| 755 |
|
| 756 | #### Regex with [ (bug regression)
|
| 757 | shopt --set ysh:all
|
| 758 |
|
| 759 | if ('[' ~ / '[' /) {
|
| 760 | echo 'sq'
|
| 761 | }
|
| 762 |
|
| 763 | if ('[' ~ / [ '[' ] /) {
|
| 764 | echo 'char class'
|
| 765 | }
|
| 766 |
|
| 767 | # User-reported string
|
| 768 | if ("a" ~ / s* 'imports' s* '=' s* '[' /) {
|
| 769 | echo "yes"
|
| 770 | }
|
| 771 |
|
| 772 | ## STDOUT:
|
| 773 | sq
|
| 774 | char class
|
| 775 | ## END
|
| 776 |
|
| 777 | #### Str.replace(Str, Str)
|
| 778 | shopt --set ysh:all
|
| 779 |
|
| 780 | var mystr = 'abca'
|
| 781 | write $[mystr.replace('a', 'A')] # Two matches
|
| 782 | write $[mystr.replace('b', 'B')] # One match
|
| 783 | write $[mystr.replace('x', 'y')] # No matches
|
| 784 |
|
| 785 | write $[mystr.replace('abc', '')] # Empty substitution
|
| 786 | write $[mystr.replace('', 'new')] # Empty substring
|
| 787 | ## STDOUT:
|
| 788 | AbcA
|
| 789 | aBca
|
| 790 | abca
|
| 791 | a
|
| 792 | newanewbnewcnewanew
|
| 793 | ## END
|
| 794 |
|
| 795 | #### Str.replace(Eggex, Str)
|
| 796 | shopt --set ysh:all
|
| 797 |
|
| 798 | var mystr = 'mangled----kebab--case'
|
| 799 | write $[mystr.replace(/ '-'+ /, '-')]
|
| 800 |
|
| 801 | setvar mystr = 'smaller-to-bigger'
|
| 802 | write $[mystr.replace(/ '-'+ /, '---')]
|
| 803 | ## STDOUT:
|
| 804 | mangled-kebab-case
|
| 805 | smaller---to---bigger
|
| 806 | ## END
|
| 807 |
|
| 808 | #### Str.replace(Eggex, Expr)
|
| 809 | shopt --set ysh:all
|
| 810 |
|
| 811 | var mystr = 'name: Bob'
|
| 812 | write $[mystr.replace(/ 'name: ' <capture dot+> /, ^"Hello $1")]
|
| 813 | write $[mystr.replace(/ 'name: ' <capture dot+> /, ^"Hello $1 (extracted from '$0')")]
|
| 814 | ## STDOUT:
|
| 815 | Hello Bob
|
| 816 | Hello Bob (extracted from 'name: Bob')
|
| 817 | ## END
|
| 818 |
|
| 819 | #### Str.replace(*, Expr), $0
|
| 820 | shopt --set ysh:all
|
| 821 |
|
| 822 | # Functionality
|
| 823 | var mystr = 'class Foo: # this class is called Foo'
|
| 824 | write $[mystr.replace("Foo", ^"$0Bar")]
|
| 825 | write $[mystr.replace(/ 'Foo' /, ^"$0Bar")]
|
| 826 |
|
| 827 | # Edge-cases
|
| 828 | var dollar0 = "$0"
|
| 829 | #echo dollar0=$dollar0
|
| 830 | #echo "0 = $0"
|
| 831 |
|
| 832 | var expected = "f($dollar0)($dollar0)"
|
| 833 | #echo "expected = $expected"
|
| 834 |
|
| 835 | # Eager replacement
|
| 836 | assert [expected === "foo".replace("o", "($0)")]
|
| 837 |
|
| 838 | assert ['f(o)(o)' === "foo".replace("o", ^"($0)")]
|
| 839 |
|
| 840 | func f() { return ( "<$0>" ) }
|
| 841 | assert ["<$dollar0>" === f()]
|
| 842 |
|
| 843 | assert ['f<o><o>' === "foo".replace("o", ^[f()])]
|
| 844 |
|
| 845 | ## STDOUT:
|
| 846 | class FooBar: # this class is called FooBar
|
| 847 | class FooBar: # this class is called FooBar
|
| 848 | ## END
|
| 849 |
|
| 850 | #### Str.replace(Eggex, Expr), scopes
|
| 851 | shopt --set ysh:all
|
| 852 |
|
| 853 | var mystr = '123'
|
| 854 |
|
| 855 | var anotherVar = 'surprise!'
|
| 856 | write $[mystr.replace(/ <capture d+> /, ^"Hello $1 ($anotherVar)")]
|
| 857 |
|
| 858 | var globalName = '456'
|
| 859 | write $[mystr.replace(/ <capture d+ as globalName> /, ^"Hello $globalName")]
|
| 860 |
|
| 861 | write $[mystr.replace(/ <capture d+ as localName> /, ^"Hello $localName, $globalName")]
|
| 862 | ## STDOUT:
|
| 863 | Hello 123 (surprise!)
|
| 864 | Hello 123
|
| 865 | Hello 123, 456
|
| 866 | ## END
|
| 867 |
|
| 868 | #### Str.replace(Eggex, *, count)
|
| 869 | shopt --set ysh:all
|
| 870 |
|
| 871 | var mystr = '1abc2abc3abc'
|
| 872 |
|
| 873 | for count in (-2..<4) {
|
| 874 | write $[mystr.replace('abc', "-", count=count)]
|
| 875 | write $[mystr.replace('abc', ^"-", count=count)]
|
| 876 | write $[mystr.replace(/ [a-z]+ /, "-", count=count)]
|
| 877 | write $[mystr.replace(/ [a-z]+ /, "-", count=count)]
|
| 878 | }
|
| 879 | ## STDOUT:
|
| 880 | 1-2-3-
|
| 881 | 1-2-3-
|
| 882 | 1-2-3-
|
| 883 | 1-2-3-
|
| 884 | 1-2-3-
|
| 885 | 1-2-3-
|
| 886 | 1-2-3-
|
| 887 | 1-2-3-
|
| 888 | 1abc2abc3abc
|
| 889 | 1abc2abc3abc
|
| 890 | 1abc2abc3abc
|
| 891 | 1abc2abc3abc
|
| 892 | 1-2abc3abc
|
| 893 | 1-2abc3abc
|
| 894 | 1-2abc3abc
|
| 895 | 1-2abc3abc
|
| 896 | 1-2-3abc
|
| 897 | 1-2-3abc
|
| 898 | 1-2-3abc
|
| 899 | 1-2-3abc
|
| 900 | 1-2-3-
|
| 901 | 1-2-3-
|
| 902 | 1-2-3-
|
| 903 | 1-2-3-
|
| 904 | ## END
|
| 905 |
|
| 906 | #### Str.replace(Str, Str), empty new/old strings
|
| 907 | var mystr = 'abca'
|
| 908 | write $[mystr.replace('abc', '')] # Empty substitution
|
| 909 | write $[mystr.replace('', 'new')] # Empty substring
|
| 910 | write $[mystr.replace('', 'new', count=1)] # Empty substring, count != -1
|
| 911 | write $[mystr.replace('', 'new', count=10)] # Empty substring, count too large
|
| 912 | ## STDOUT:
|
| 913 | a
|
| 914 | newanewbnewcnewanew
|
| 915 | newabca
|
| 916 | newanewbnewcnewanew
|
| 917 | ## END
|
| 918 |
|
| 919 | #### Str.replace(Eggex, Lazy), convert_func
|
| 920 | shopt --set ysh:all
|
| 921 |
|
| 922 | var mystr = '123'
|
| 923 |
|
| 924 | write $[mystr.replace(/ <capture d+ as n : int> /, ^"$[n + 1]")]
|
| 925 |
|
| 926 | # values automatically get stringified
|
| 927 | write $[mystr.replace(/ <capture d+ as n : int> /, ^"$1")]
|
| 928 |
|
| 929 | func not_str(inp) {
|
| 930 | return ({ "value": inp })
|
| 931 | }
|
| 932 |
|
| 933 | # should fail to stringify $1
|
| 934 | try { call mystr.replace(/ <capture d+ : not_str> /, ^"$1") }
|
| 935 | write status=$_status
|
| 936 | ## STDOUT:
|
| 937 | 124
|
| 938 | 123
|
| 939 | status=3
|
| 940 | ## END
|
| 941 |
|
| 942 | #### Str.replace(Eggex, *), eflags
|
| 943 | shopt --set ysh:all
|
| 944 |
|
| 945 | var mystr = b'1-2-3\n4-5'
|
| 946 | write $[mystr.replace(/ d+ /, ^"[$0]")]
|
| 947 | write $[mystr.replace(/ ^ d+ /, ^"[$0]")]
|
| 948 | write $[mystr.replace(/ ^ d+ ; reg_newline /, ^"[$0]")]
|
| 949 | ## STDOUT:
|
| 950 | [1]-[2]-[3]
|
| 951 | [4]-[5]
|
| 952 | [1]-2-3
|
| 953 | 4-5
|
| 954 | [1]-2-3
|
| 955 | [4]-5
|
| 956 | ## END
|
| 957 |
|
| 958 | #### Str.replace(Eggex, *), guard against infinite loop
|
| 959 | shopt --set ysh:all
|
| 960 |
|
| 961 | var mystr = 'foo bar baz'
|
| 962 | write $[mystr.replace(/ space* /, ' ')]
|
| 963 | ## status: 3
|
| 964 | ## STDOUT:
|
| 965 | ## END
|
| 966 |
|
| 967 | #### Str.replace(Eggex, *), str cannot contain NUL bytes
|
| 968 | shopt --set ysh:all
|
| 969 |
|
| 970 | var mystr = b'foo bar baz\y00'
|
| 971 | write $[mystr.replace(/ space+ /, ' ')]
|
| 972 | ## status: 3
|
| 973 | ## STDOUT:
|
| 974 | ## END
|
| 975 |
|
| 976 | #### Str.replace() at top level
|
| 977 | shopt --set ysh:upgrade
|
| 978 |
|
| 979 | var s = 'mystr'
|
| 980 | var pat = / 's' <capture dot> /
|
| 981 | var template = ^"[$x $0 $1 $x]"
|
| 982 | pp test_ (template)
|
| 983 |
|
| 984 | var x = 'x'
|
| 985 |
|
| 986 | var new = s.replace(pat, template)
|
| 987 | echo 'replace ' $new
|
| 988 |
|
| 989 | func myreplace(s, template) {
|
| 990 | return (s.replace(pat, template))
|
| 991 | }
|
| 992 |
|
| 993 | echo myreplace $[myreplace(s, template)]
|
| 994 |
|
| 995 | ## STDOUT:
|
| 996 | <Expr>
|
| 997 | replace my[x st t x]r
|
| 998 | myreplace my[x st t x]r
|
| 999 | ## END
|
| 1000 |
|
| 1001 | #### Str.replace() lexical scope with ^""
|
| 1002 | shopt --set ysh:upgrade
|
| 1003 |
|
| 1004 | var s = 'mystr'
|
| 1005 | var pat = / 's' <capture dot> /
|
| 1006 |
|
| 1007 | proc p {
|
| 1008 | var x = 'x'
|
| 1009 | var template = ^"[$x $0 $1 $x]"
|
| 1010 | pp test_ (template)
|
| 1011 |
|
| 1012 | var new = s.replace(pat, template)
|
| 1013 | echo 'replace ' $new
|
| 1014 |
|
| 1015 | func myreplace(s, template) {
|
| 1016 | return (s.replace(pat, template))
|
| 1017 | }
|
| 1018 |
|
| 1019 | echo myreplace $[myreplace(s, template)]
|
| 1020 | }
|
| 1021 |
|
| 1022 | p
|
| 1023 |
|
| 1024 | ## STDOUT:
|
| 1025 | <Expr>
|
| 1026 | replace my[x st t x]r
|
| 1027 | myreplace my[x st t x]r
|
| 1028 | ## END
|
| 1029 |
|
| 1030 | #### Str.replace() lexical scope with ^[]
|
| 1031 | shopt --set ysh:upgrade
|
| 1032 |
|
| 1033 | var s = 'mystr'
|
| 1034 | var pat = / 's' <capture dot> /
|
| 1035 |
|
| 1036 | proc p {
|
| 1037 | var x = 'x'
|
| 1038 | var template = ^['[' ++ x ++ ' ' ++ $0 ++ ' ' ++ $1 ++ ' ' ++ x ++ ']']
|
| 1039 | pp test_ (template)
|
| 1040 |
|
| 1041 | var new = s.replace(pat, template)
|
| 1042 | echo 'replace ' $new
|
| 1043 |
|
| 1044 | func myreplace(s, template) {
|
| 1045 | return (s.replace(pat, template))
|
| 1046 | }
|
| 1047 |
|
| 1048 | echo myreplace $[myreplace(s, template)]
|
| 1049 | }
|
| 1050 |
|
| 1051 | p
|
| 1052 |
|
| 1053 | ## STDOUT:
|
| 1054 | <Expr>
|
| 1055 | replace my[x st t x]r
|
| 1056 | myreplace my[x st t x]r
|
| 1057 | ## END
|