1 /*- 2 * Copyright (c) 1991, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * This code is derived from software contributed to Berkeley by 6 * Kenneth Almquist. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 4. Neither the name of the University nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 */ 32 33 #ifndef lint 34 #if 0 35 static char sccsid[] = "@(#)parser.c 8.7 (Berkeley) 5/16/95"; 36 #endif 37 #endif /* not lint */ 38 #include <sys/cdefs.h> 39 __FBSDID("$FreeBSD$"); 40 41 #include <stdlib.h> 42 #include <unistd.h> 43 #include <stdio.h> 44 45 #include "shell.h" 46 #include "parser.h" 47 #include "nodes.h" 48 #include "expand.h" /* defines rmescapes() */ 49 #include "syntax.h" 50 #include "options.h" 51 #include "input.h" 52 #include "output.h" 53 #include "var.h" 54 #include "error.h" 55 #include "memalloc.h" 56 #include "mystring.h" 57 #include "alias.h" 58 #include "show.h" 59 #include "eval.h" 60 #include "exec.h" /* to check for special builtins */ 61 #ifndef NO_HISTORY 62 #include "myhistedit.h" 63 #endif 64 65 /* 66 * Shell command parser. 67 */ 68 69 #define EOFMARKLEN 79 70 #define PROMPTLEN 128 71 72 /* values of checkkwd variable */ 73 #define CHKALIAS 0x1 74 #define CHKKWD 0x2 75 #define CHKNL 0x4 76 77 /* values returned by readtoken */ 78 #include "token.h" 79 80 81 82 struct heredoc { 83 struct heredoc *next; /* next here document in list */ 84 union node *here; /* redirection node */ 85 char *eofmark; /* string indicating end of input */ 86 int striptabs; /* if set, strip leading tabs */ 87 }; 88 89 struct parser_temp { 90 struct parser_temp *next; 91 void *data; 92 }; 93 94 95 static struct heredoc *heredoclist; /* list of here documents to read */ 96 static int doprompt; /* if set, prompt the user */ 97 static int needprompt; /* true if interactive and at start of line */ 98 static int lasttoken; /* last token read */ 99 int tokpushback; /* last token pushed back */ 100 static char *wordtext; /* text of last word returned by readtoken */ 101 static int checkkwd; 102 static struct nodelist *backquotelist; 103 static union node *redirnode; 104 static struct heredoc *heredoc; 105 static int quoteflag; /* set if (part of) last token was quoted */ 106 static int startlinno; /* line # where last token started */ 107 static int funclinno; /* line # where the current function started */ 108 static struct parser_temp *parser_temp; 109 110 111 static union node *list(int, int); 112 static union node *andor(void); 113 static union node *pipeline(void); 114 static union node *command(void); 115 static union node *simplecmd(union node **, union node *); 116 static union node *makename(void); 117 static void parsefname(void); 118 static void parseheredoc(void); 119 static int peektoken(void); 120 static int readtoken(void); 121 static int xxreadtoken(void); 122 static int readtoken1(int, const char *, const char *, int); 123 static int noexpand(char *); 124 static void synexpect(int) __dead2; 125 static void synerror(const char *) __dead2; 126 static void setprompt(int); 127 128 129 static void * 130 parser_temp_alloc(size_t len) 131 { 132 struct parser_temp *t; 133 134 INTOFF; 135 t = ckmalloc(sizeof(*t)); 136 t->data = NULL; 137 t->next = parser_temp; 138 parser_temp = t; 139 t->data = ckmalloc(len); 140 INTON; 141 return t->data; 142 } 143 144 145 static void * 146 parser_temp_realloc(void *ptr, size_t len) 147 { 148 struct parser_temp *t; 149 150 INTOFF; 151 t = parser_temp; 152 if (ptr != t->data) 153 error("bug: parser_temp_realloc misused"); 154 t->data = ckrealloc(t->data, len); 155 INTON; 156 return t->data; 157 } 158 159 160 static void 161 parser_temp_free_upto(void *ptr) 162 { 163 struct parser_temp *t; 164 int done = 0; 165 166 INTOFF; 167 while (parser_temp != NULL && !done) { 168 t = parser_temp; 169 parser_temp = t->next; 170 done = t->data == ptr; 171 ckfree(t->data); 172 ckfree(t); 173 } 174 INTON; 175 if (!done) 176 error("bug: parser_temp_free_upto misused"); 177 } 178 179 180 static void 181 parser_temp_free_all(void) 182 { 183 struct parser_temp *t; 184 185 INTOFF; 186 while (parser_temp != NULL) { 187 t = parser_temp; 188 parser_temp = t->next; 189 ckfree(t->data); 190 ckfree(t); 191 } 192 INTON; 193 } 194 195 196 /* 197 * Read and parse a command. Returns NEOF on end of file. (NULL is a 198 * valid parse tree indicating a blank line.) 199 */ 200 201 union node * 202 parsecmd(int interact) 203 { 204 int t; 205 206 /* This assumes the parser is not re-entered, 207 * which could happen if we add command substitution on PS1/PS2. 208 */ 209 parser_temp_free_all(); 210 heredoclist = NULL; 211 212 tokpushback = 0; 213 doprompt = interact; 214 if (doprompt) 215 setprompt(1); 216 else 217 setprompt(0); 218 needprompt = 0; 219 t = readtoken(); 220 if (t == TEOF) 221 return NEOF; 222 if (t == TNL) 223 return NULL; 224 tokpushback++; 225 return list(1, 1); 226 } 227 228 229 static union node * 230 list(int nlflag, int erflag) 231 { 232 union node *ntop, *n1, *n2, *n3; 233 int tok; 234 235 checkkwd = CHKNL | CHKKWD | CHKALIAS; 236 if (!nlflag && !erflag && tokendlist[peektoken()]) 237 return NULL; 238 ntop = n1 = NULL; 239 for (;;) { 240 n2 = andor(); 241 tok = readtoken(); 242 if (tok == TBACKGND) { 243 if (n2 != NULL && n2->type == NPIPE) { 244 n2->npipe.backgnd = 1; 245 } else if (n2 != NULL && n2->type == NREDIR) { 246 n2->type = NBACKGND; 247 } else { 248 n3 = (union node *)stalloc(sizeof (struct nredir)); 249 n3->type = NBACKGND; 250 n3->nredir.n = n2; 251 n3->nredir.redirect = NULL; 252 n2 = n3; 253 } 254 } 255 if (ntop == NULL) 256 ntop = n2; 257 else if (n1 == NULL) { 258 n1 = (union node *)stalloc(sizeof (struct nbinary)); 259 n1->type = NSEMI; 260 n1->nbinary.ch1 = ntop; 261 n1->nbinary.ch2 = n2; 262 ntop = n1; 263 } 264 else { 265 n3 = (union node *)stalloc(sizeof (struct nbinary)); 266 n3->type = NSEMI; 267 n3->nbinary.ch1 = n1->nbinary.ch2; 268 n3->nbinary.ch2 = n2; 269 n1->nbinary.ch2 = n3; 270 n1 = n3; 271 } 272 switch (tok) { 273 case TBACKGND: 274 case TSEMI: 275 tok = readtoken(); 276 /* FALLTHROUGH */ 277 case TNL: 278 if (tok == TNL) { 279 parseheredoc(); 280 if (nlflag) 281 return ntop; 282 } else if (tok == TEOF && nlflag) { 283 parseheredoc(); 284 return ntop; 285 } else { 286 tokpushback++; 287 } 288 checkkwd = CHKNL | CHKKWD | CHKALIAS; 289 if (!nlflag && (erflag ? peektoken() == TEOF : 290 tokendlist[peektoken()])) 291 return ntop; 292 break; 293 case TEOF: 294 if (heredoclist) 295 parseheredoc(); 296 else 297 pungetc(); /* push back EOF on input */ 298 return ntop; 299 default: 300 if (nlflag || erflag) 301 synexpect(-1); 302 tokpushback++; 303 return ntop; 304 } 305 } 306 } 307 308 309 310 static union node * 311 andor(void) 312 { 313 union node *n1, *n2, *n3; 314 int t; 315 316 n1 = pipeline(); 317 for (;;) { 318 if ((t = readtoken()) == TAND) { 319 t = NAND; 320 } else if (t == TOR) { 321 t = NOR; 322 } else { 323 tokpushback++; 324 return n1; 325 } 326 n2 = pipeline(); 327 n3 = (union node *)stalloc(sizeof (struct nbinary)); 328 n3->type = t; 329 n3->nbinary.ch1 = n1; 330 n3->nbinary.ch2 = n2; 331 n1 = n3; 332 } 333 } 334 335 336 337 static union node * 338 pipeline(void) 339 { 340 union node *n1, *n2, *pipenode; 341 struct nodelist *lp, *prev; 342 int negate, t; 343 344 negate = 0; 345 checkkwd = CHKNL | CHKKWD | CHKALIAS; 346 TRACE(("pipeline: entered\n")); 347 while (readtoken() == TNOT) 348 negate = !negate; 349 tokpushback++; 350 n1 = command(); 351 if (readtoken() == TPIPE) { 352 pipenode = (union node *)stalloc(sizeof (struct npipe)); 353 pipenode->type = NPIPE; 354 pipenode->npipe.backgnd = 0; 355 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 356 pipenode->npipe.cmdlist = lp; 357 lp->n = n1; 358 do { 359 prev = lp; 360 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 361 checkkwd = CHKNL | CHKKWD | CHKALIAS; 362 t = readtoken(); 363 tokpushback++; 364 if (t == TNOT) 365 lp->n = pipeline(); 366 else 367 lp->n = command(); 368 prev->next = lp; 369 } while (readtoken() == TPIPE); 370 lp->next = NULL; 371 n1 = pipenode; 372 } 373 tokpushback++; 374 if (negate) { 375 n2 = (union node *)stalloc(sizeof (struct nnot)); 376 n2->type = NNOT; 377 n2->nnot.com = n1; 378 return n2; 379 } else 380 return n1; 381 } 382 383 384 385 static union node * 386 command(void) 387 { 388 union node *n1, *n2; 389 union node *ap, **app; 390 union node *cp, **cpp; 391 union node *redir, **rpp; 392 int t; 393 int is_subshell; 394 395 checkkwd = CHKNL | CHKKWD | CHKALIAS; 396 is_subshell = 0; 397 redir = NULL; 398 n1 = NULL; 399 rpp = &redir; 400 401 /* Check for redirection which may precede command */ 402 while (readtoken() == TREDIR) { 403 *rpp = n2 = redirnode; 404 rpp = &n2->nfile.next; 405 parsefname(); 406 } 407 tokpushback++; 408 409 switch (readtoken()) { 410 case TIF: 411 n1 = (union node *)stalloc(sizeof (struct nif)); 412 n1->type = NIF; 413 if ((n1->nif.test = list(0, 0)) == NULL) 414 synexpect(-1); 415 if (readtoken() != TTHEN) 416 synexpect(TTHEN); 417 n1->nif.ifpart = list(0, 0); 418 n2 = n1; 419 while (readtoken() == TELIF) { 420 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif)); 421 n2 = n2->nif.elsepart; 422 n2->type = NIF; 423 if ((n2->nif.test = list(0, 0)) == NULL) 424 synexpect(-1); 425 if (readtoken() != TTHEN) 426 synexpect(TTHEN); 427 n2->nif.ifpart = list(0, 0); 428 } 429 if (lasttoken == TELSE) 430 n2->nif.elsepart = list(0, 0); 431 else { 432 n2->nif.elsepart = NULL; 433 tokpushback++; 434 } 435 if (readtoken() != TFI) 436 synexpect(TFI); 437 checkkwd = CHKKWD | CHKALIAS; 438 break; 439 case TWHILE: 440 case TUNTIL: { 441 int got; 442 n1 = (union node *)stalloc(sizeof (struct nbinary)); 443 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL; 444 if ((n1->nbinary.ch1 = list(0, 0)) == NULL) 445 synexpect(-1); 446 if ((got=readtoken()) != TDO) { 447 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : "")); 448 synexpect(TDO); 449 } 450 n1->nbinary.ch2 = list(0, 0); 451 if (readtoken() != TDONE) 452 synexpect(TDONE); 453 checkkwd = CHKKWD | CHKALIAS; 454 break; 455 } 456 case TFOR: 457 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext)) 458 synerror("Bad for loop variable"); 459 n1 = (union node *)stalloc(sizeof (struct nfor)); 460 n1->type = NFOR; 461 n1->nfor.var = wordtext; 462 while (readtoken() == TNL) 463 ; 464 if (lasttoken == TWORD && ! quoteflag && equal(wordtext, "in")) { 465 app = ≈ 466 while (readtoken() == TWORD) { 467 n2 = (union node *)stalloc(sizeof (struct narg)); 468 n2->type = NARG; 469 n2->narg.text = wordtext; 470 n2->narg.backquote = backquotelist; 471 *app = n2; 472 app = &n2->narg.next; 473 } 474 *app = NULL; 475 n1->nfor.args = ap; 476 if (lasttoken != TNL && lasttoken != TSEMI) 477 synexpect(-1); 478 } else { 479 static char argvars[5] = { 480 CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0' 481 }; 482 n2 = (union node *)stalloc(sizeof (struct narg)); 483 n2->type = NARG; 484 n2->narg.text = argvars; 485 n2->narg.backquote = NULL; 486 n2->narg.next = NULL; 487 n1->nfor.args = n2; 488 /* 489 * Newline or semicolon here is optional (but note 490 * that the original Bourne shell only allowed NL). 491 */ 492 if (lasttoken != TNL && lasttoken != TSEMI) 493 tokpushback++; 494 } 495 checkkwd = CHKNL | CHKKWD | CHKALIAS; 496 if ((t = readtoken()) == TDO) 497 t = TDONE; 498 else if (t == TBEGIN) 499 t = TEND; 500 else 501 synexpect(-1); 502 n1->nfor.body = list(0, 0); 503 if (readtoken() != t) 504 synexpect(t); 505 checkkwd = CHKKWD | CHKALIAS; 506 break; 507 case TCASE: 508 n1 = (union node *)stalloc(sizeof (struct ncase)); 509 n1->type = NCASE; 510 if (readtoken() != TWORD) 511 synexpect(TWORD); 512 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg)); 513 n2->type = NARG; 514 n2->narg.text = wordtext; 515 n2->narg.backquote = backquotelist; 516 n2->narg.next = NULL; 517 while (readtoken() == TNL); 518 if (lasttoken != TWORD || ! equal(wordtext, "in")) 519 synerror("expecting \"in\""); 520 cpp = &n1->ncase.cases; 521 checkkwd = CHKNL | CHKKWD, readtoken(); 522 while (lasttoken != TESAC) { 523 *cpp = cp = (union node *)stalloc(sizeof (struct nclist)); 524 cp->type = NCLIST; 525 app = &cp->nclist.pattern; 526 if (lasttoken == TLP) 527 readtoken(); 528 for (;;) { 529 *app = ap = (union node *)stalloc(sizeof (struct narg)); 530 ap->type = NARG; 531 ap->narg.text = wordtext; 532 ap->narg.backquote = backquotelist; 533 checkkwd = CHKNL | CHKKWD; 534 if (readtoken() != TPIPE) 535 break; 536 app = &ap->narg.next; 537 readtoken(); 538 } 539 ap->narg.next = NULL; 540 if (lasttoken != TRP) 541 synexpect(TRP); 542 cp->nclist.body = list(0, 0); 543 544 checkkwd = CHKNL | CHKKWD | CHKALIAS; 545 if ((t = readtoken()) != TESAC) { 546 if (t == TENDCASE) 547 ; 548 else if (t == TFALLTHRU) 549 cp->type = NCLISTFALLTHRU; 550 else 551 synexpect(TENDCASE); 552 checkkwd = CHKNL | CHKKWD, readtoken(); 553 } 554 cpp = &cp->nclist.next; 555 } 556 *cpp = NULL; 557 checkkwd = CHKKWD | CHKALIAS; 558 break; 559 case TLP: 560 n1 = (union node *)stalloc(sizeof (struct nredir)); 561 n1->type = NSUBSHELL; 562 n1->nredir.n = list(0, 0); 563 n1->nredir.redirect = NULL; 564 if (readtoken() != TRP) 565 synexpect(TRP); 566 checkkwd = CHKKWD | CHKALIAS; 567 is_subshell = 1; 568 break; 569 case TBEGIN: 570 n1 = list(0, 0); 571 if (readtoken() != TEND) 572 synexpect(TEND); 573 checkkwd = CHKKWD | CHKALIAS; 574 break; 575 /* Handle an empty command like other simple commands. */ 576 case TBACKGND: 577 case TSEMI: 578 case TAND: 579 case TOR: 580 /* 581 * An empty command before a ; doesn't make much sense, and 582 * should certainly be disallowed in the case of `if ;'. 583 */ 584 if (!redir) 585 synexpect(-1); 586 case TNL: 587 case TEOF: 588 case TWORD: 589 case TRP: 590 tokpushback++; 591 n1 = simplecmd(rpp, redir); 592 return n1; 593 default: 594 synexpect(-1); 595 } 596 597 /* Now check for redirection which may follow command */ 598 while (readtoken() == TREDIR) { 599 *rpp = n2 = redirnode; 600 rpp = &n2->nfile.next; 601 parsefname(); 602 } 603 tokpushback++; 604 *rpp = NULL; 605 if (redir) { 606 if (!is_subshell) { 607 n2 = (union node *)stalloc(sizeof (struct nredir)); 608 n2->type = NREDIR; 609 n2->nredir.n = n1; 610 n1 = n2; 611 } 612 n1->nredir.redirect = redir; 613 } 614 615 return n1; 616 } 617 618 619 static union node * 620 simplecmd(union node **rpp, union node *redir) 621 { 622 union node *args, **app; 623 union node **orig_rpp = rpp; 624 union node *n = NULL; 625 int special; 626 int savecheckkwd; 627 628 /* If we don't have any redirections already, then we must reset */ 629 /* rpp to be the address of the local redir variable. */ 630 if (redir == 0) 631 rpp = &redir; 632 633 args = NULL; 634 app = &args; 635 /* 636 * We save the incoming value, because we need this for shell 637 * functions. There can not be a redirect or an argument between 638 * the function name and the open parenthesis. 639 */ 640 orig_rpp = rpp; 641 642 savecheckkwd = CHKALIAS; 643 644 for (;;) { 645 checkkwd = savecheckkwd; 646 if (readtoken() == TWORD) { 647 n = (union node *)stalloc(sizeof (struct narg)); 648 n->type = NARG; 649 n->narg.text = wordtext; 650 n->narg.backquote = backquotelist; 651 *app = n; 652 app = &n->narg.next; 653 if (savecheckkwd != 0 && !isassignment(wordtext)) 654 savecheckkwd = 0; 655 } else if (lasttoken == TREDIR) { 656 *rpp = n = redirnode; 657 rpp = &n->nfile.next; 658 parsefname(); /* read name of redirection file */ 659 } else if (lasttoken == TLP && app == &args->narg.next 660 && rpp == orig_rpp) { 661 /* We have a function */ 662 if (readtoken() != TRP) 663 synexpect(TRP); 664 funclinno = plinno; 665 /* 666 * - Require plain text. 667 * - Functions with '/' cannot be called. 668 * - Reject name=(). 669 * - Reject ksh extended glob patterns. 670 */ 671 if (!noexpand(n->narg.text) || quoteflag || 672 strchr(n->narg.text, '/') || 673 strchr("!%*+-=?@}~", 674 n->narg.text[strlen(n->narg.text) - 1])) 675 synerror("Bad function name"); 676 rmescapes(n->narg.text); 677 if (find_builtin(n->narg.text, &special) >= 0 && 678 special) 679 synerror("Cannot override a special builtin with a function"); 680 n->type = NDEFUN; 681 n->narg.next = command(); 682 funclinno = 0; 683 return n; 684 } else { 685 tokpushback++; 686 break; 687 } 688 } 689 *app = NULL; 690 *rpp = NULL; 691 n = (union node *)stalloc(sizeof (struct ncmd)); 692 n->type = NCMD; 693 n->ncmd.args = args; 694 n->ncmd.redirect = redir; 695 return n; 696 } 697 698 static union node * 699 makename(void) 700 { 701 union node *n; 702 703 n = (union node *)stalloc(sizeof (struct narg)); 704 n->type = NARG; 705 n->narg.next = NULL; 706 n->narg.text = wordtext; 707 n->narg.backquote = backquotelist; 708 return n; 709 } 710 711 void 712 fixredir(union node *n, const char *text, int err) 713 { 714 TRACE(("Fix redir %s %d\n", text, err)); 715 if (!err) 716 n->ndup.vname = NULL; 717 718 if (is_digit(text[0]) && text[1] == '\0') 719 n->ndup.dupfd = digit_val(text[0]); 720 else if (text[0] == '-' && text[1] == '\0') 721 n->ndup.dupfd = -1; 722 else { 723 724 if (err) 725 synerror("Bad fd number"); 726 else 727 n->ndup.vname = makename(); 728 } 729 } 730 731 732 static void 733 parsefname(void) 734 { 735 union node *n = redirnode; 736 737 if (readtoken() != TWORD) 738 synexpect(-1); 739 if (n->type == NHERE) { 740 struct heredoc *here = heredoc; 741 struct heredoc *p; 742 int i; 743 744 if (quoteflag == 0) 745 n->type = NXHERE; 746 TRACE(("Here document %d\n", n->type)); 747 if (here->striptabs) { 748 while (*wordtext == '\t') 749 wordtext++; 750 } 751 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN) 752 synerror("Illegal eof marker for << redirection"); 753 rmescapes(wordtext); 754 here->eofmark = wordtext; 755 here->next = NULL; 756 if (heredoclist == NULL) 757 heredoclist = here; 758 else { 759 for (p = heredoclist ; p->next ; p = p->next); 760 p->next = here; 761 } 762 } else if (n->type == NTOFD || n->type == NFROMFD) { 763 fixredir(n, wordtext, 0); 764 } else { 765 n->nfile.fname = makename(); 766 } 767 } 768 769 770 /* 771 * Input any here documents. 772 */ 773 774 static void 775 parseheredoc(void) 776 { 777 struct heredoc *here; 778 union node *n; 779 780 while (heredoclist) { 781 here = heredoclist; 782 heredoclist = here->next; 783 if (needprompt) { 784 setprompt(2); 785 needprompt = 0; 786 } 787 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX, 788 here->eofmark, here->striptabs); 789 n = (union node *)stalloc(sizeof (struct narg)); 790 n->narg.type = NARG; 791 n->narg.next = NULL; 792 n->narg.text = wordtext; 793 n->narg.backquote = backquotelist; 794 here->here->nhere.doc = n; 795 } 796 } 797 798 static int 799 peektoken(void) 800 { 801 int t; 802 803 t = readtoken(); 804 tokpushback++; 805 return (t); 806 } 807 808 static int 809 readtoken(void) 810 { 811 int t; 812 struct alias *ap; 813 #ifdef DEBUG 814 int alreadyseen = tokpushback; 815 #endif 816 817 top: 818 t = xxreadtoken(); 819 820 /* 821 * eat newlines 822 */ 823 if (checkkwd & CHKNL) { 824 while (t == TNL) { 825 parseheredoc(); 826 t = xxreadtoken(); 827 } 828 } 829 830 /* 831 * check for keywords and aliases 832 */ 833 if (t == TWORD && !quoteflag) 834 { 835 const char * const *pp; 836 837 if (checkkwd & CHKKWD) 838 for (pp = parsekwd; *pp; pp++) { 839 if (**pp == *wordtext && equal(*pp, wordtext)) 840 { 841 lasttoken = t = pp - parsekwd + KWDOFFSET; 842 TRACE(("keyword %s recognized\n", tokname[t])); 843 goto out; 844 } 845 } 846 if (checkkwd & CHKALIAS && 847 (ap = lookupalias(wordtext, 1)) != NULL) { 848 pushstring(ap->val, strlen(ap->val), ap); 849 goto top; 850 } 851 } 852 out: 853 if (t != TNOT) 854 checkkwd = 0; 855 856 #ifdef DEBUG 857 if (!alreadyseen) 858 TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 859 else 860 TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 861 #endif 862 return (t); 863 } 864 865 866 /* 867 * Read the next input token. 868 * If the token is a word, we set backquotelist to the list of cmds in 869 * backquotes. We set quoteflag to true if any part of the word was 870 * quoted. 871 * If the token is TREDIR, then we set redirnode to a structure containing 872 * the redirection. 873 * In all cases, the variable startlinno is set to the number of the line 874 * on which the token starts. 875 * 876 * [Change comment: here documents and internal procedures] 877 * [Readtoken shouldn't have any arguments. Perhaps we should make the 878 * word parsing code into a separate routine. In this case, readtoken 879 * doesn't need to have any internal procedures, but parseword does. 880 * We could also make parseoperator in essence the main routine, and 881 * have parseword (readtoken1?) handle both words and redirection.] 882 */ 883 884 #define RETURN(token) return lasttoken = token 885 886 static int 887 xxreadtoken(void) 888 { 889 int c; 890 891 if (tokpushback) { 892 tokpushback = 0; 893 return lasttoken; 894 } 895 if (needprompt) { 896 setprompt(2); 897 needprompt = 0; 898 } 899 startlinno = plinno; 900 for (;;) { /* until token or start of word found */ 901 c = pgetc_macro(); 902 switch (c) { 903 case ' ': case '\t': 904 continue; 905 case '#': 906 while ((c = pgetc()) != '\n' && c != PEOF); 907 pungetc(); 908 continue; 909 case '\\': 910 if (pgetc() == '\n') { 911 startlinno = ++plinno; 912 if (doprompt) 913 setprompt(2); 914 else 915 setprompt(0); 916 continue; 917 } 918 pungetc(); 919 goto breakloop; 920 case '\n': 921 plinno++; 922 needprompt = doprompt; 923 RETURN(TNL); 924 case PEOF: 925 RETURN(TEOF); 926 case '&': 927 if (pgetc() == '&') 928 RETURN(TAND); 929 pungetc(); 930 RETURN(TBACKGND); 931 case '|': 932 if (pgetc() == '|') 933 RETURN(TOR); 934 pungetc(); 935 RETURN(TPIPE); 936 case ';': 937 c = pgetc(); 938 if (c == ';') 939 RETURN(TENDCASE); 940 else if (c == '&') 941 RETURN(TFALLTHRU); 942 pungetc(); 943 RETURN(TSEMI); 944 case '(': 945 RETURN(TLP); 946 case ')': 947 RETURN(TRP); 948 default: 949 goto breakloop; 950 } 951 } 952 breakloop: 953 return readtoken1(c, BASESYNTAX, (char *)NULL, 0); 954 #undef RETURN 955 } 956 957 958 #define MAXNEST_static 8 959 struct tokenstate 960 { 961 const char *syntax; /* *SYNTAX */ 962 int parenlevel; /* levels of parentheses in arithmetic */ 963 enum tokenstate_category 964 { 965 TSTATE_TOP, 966 TSTATE_VAR_OLD, /* ${var+-=?}, inherits dquotes */ 967 TSTATE_VAR_NEW, /* other ${var...}, own dquote state */ 968 TSTATE_ARITH 969 } category; 970 }; 971 972 973 /* 974 * Called to parse command substitutions. 975 */ 976 977 static char * 978 parsebackq(char *out, struct nodelist **pbqlist, 979 int oldstyle, int dblquote, int quoted) 980 { 981 struct nodelist **nlpp; 982 union node *n; 983 char *volatile str; 984 struct jmploc jmploc; 985 struct jmploc *const savehandler = handler; 986 size_t savelen; 987 int saveprompt; 988 const int bq_startlinno = plinno; 989 char *volatile ostr = NULL; 990 struct parsefile *const savetopfile = getcurrentfile(); 991 struct heredoc *const saveheredoclist = heredoclist; 992 struct heredoc *here; 993 994 str = NULL; 995 if (setjmp(jmploc.loc)) { 996 popfilesupto(savetopfile); 997 if (str) 998 ckfree(str); 999 if (ostr) 1000 ckfree(ostr); 1001 heredoclist = saveheredoclist; 1002 handler = savehandler; 1003 if (exception == EXERROR) { 1004 startlinno = bq_startlinno; 1005 synerror("Error in command substitution"); 1006 } 1007 longjmp(handler->loc, 1); 1008 } 1009 INTOFF; 1010 savelen = out - stackblock(); 1011 if (savelen > 0) { 1012 str = ckmalloc(savelen); 1013 memcpy(str, stackblock(), savelen); 1014 } 1015 handler = &jmploc; 1016 heredoclist = NULL; 1017 INTON; 1018 if (oldstyle) { 1019 /* We must read until the closing backquote, giving special 1020 treatment to some slashes, and then push the string and 1021 reread it as input, interpreting it normally. */ 1022 char *oout; 1023 int c; 1024 int olen; 1025 1026 1027 STARTSTACKSTR(oout); 1028 for (;;) { 1029 if (needprompt) { 1030 setprompt(2); 1031 needprompt = 0; 1032 } 1033 CHECKSTRSPACE(2, oout); 1034 switch (c = pgetc()) { 1035 case '`': 1036 goto done; 1037 1038 case '\\': 1039 if ((c = pgetc()) == '\n') { 1040 plinno++; 1041 if (doprompt) 1042 setprompt(2); 1043 else 1044 setprompt(0); 1045 /* 1046 * If eating a newline, avoid putting 1047 * the newline into the new character 1048 * stream (via the USTPUTC after the 1049 * switch). 1050 */ 1051 continue; 1052 } 1053 if (c != '\\' && c != '`' && c != '$' 1054 && (!dblquote || c != '"')) 1055 USTPUTC('\\', oout); 1056 break; 1057 1058 case '\n': 1059 plinno++; 1060 needprompt = doprompt; 1061 break; 1062 1063 case PEOF: 1064 startlinno = plinno; 1065 synerror("EOF in backquote substitution"); 1066 break; 1067 1068 default: 1069 break; 1070 } 1071 USTPUTC(c, oout); 1072 } 1073 done: 1074 USTPUTC('\0', oout); 1075 olen = oout - stackblock(); 1076 INTOFF; 1077 ostr = ckmalloc(olen); 1078 memcpy(ostr, stackblock(), olen); 1079 setinputstring(ostr, 1); 1080 INTON; 1081 } 1082 nlpp = pbqlist; 1083 while (*nlpp) 1084 nlpp = &(*nlpp)->next; 1085 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 1086 (*nlpp)->next = NULL; 1087 1088 if (oldstyle) { 1089 saveprompt = doprompt; 1090 doprompt = 0; 1091 } 1092 1093 n = list(0, oldstyle); 1094 1095 if (oldstyle) 1096 doprompt = saveprompt; 1097 else { 1098 if (readtoken() != TRP) 1099 synexpect(TRP); 1100 } 1101 1102 (*nlpp)->n = n; 1103 if (oldstyle) { 1104 /* 1105 * Start reading from old file again, ignoring any pushed back 1106 * tokens left from the backquote parsing 1107 */ 1108 popfile(); 1109 tokpushback = 0; 1110 } 1111 STARTSTACKSTR(out); 1112 CHECKSTRSPACE(savelen + 1, out); 1113 INTOFF; 1114 if (str) { 1115 memcpy(out, str, savelen); 1116 STADJUST(savelen, out); 1117 ckfree(str); 1118 str = NULL; 1119 } 1120 if (ostr) { 1121 ckfree(ostr); 1122 ostr = NULL; 1123 } 1124 here = saveheredoclist; 1125 if (here != NULL) { 1126 while (here->next != NULL) 1127 here = here->next; 1128 here->next = heredoclist; 1129 heredoclist = saveheredoclist; 1130 } 1131 handler = savehandler; 1132 INTON; 1133 if (quoted) 1134 USTPUTC(CTLBACKQ | CTLQUOTE, out); 1135 else 1136 USTPUTC(CTLBACKQ, out); 1137 return out; 1138 } 1139 1140 1141 /* 1142 * Called to parse a backslash escape sequence inside $'...'. 1143 * The backslash has already been read. 1144 */ 1145 static char * 1146 readcstyleesc(char *out) 1147 { 1148 int c, v, i, n; 1149 1150 c = pgetc(); 1151 switch (c) { 1152 case '\0': 1153 synerror("Unterminated quoted string"); 1154 case '\n': 1155 plinno++; 1156 if (doprompt) 1157 setprompt(2); 1158 else 1159 setprompt(0); 1160 return out; 1161 case '\\': 1162 case '\'': 1163 case '"': 1164 v = c; 1165 break; 1166 case 'a': v = '\a'; break; 1167 case 'b': v = '\b'; break; 1168 case 'e': v = '\033'; break; 1169 case 'f': v = '\f'; break; 1170 case 'n': v = '\n'; break; 1171 case 'r': v = '\r'; break; 1172 case 't': v = '\t'; break; 1173 case 'v': v = '\v'; break; 1174 case 'x': 1175 v = 0; 1176 for (;;) { 1177 c = pgetc(); 1178 if (c >= '0' && c <= '9') 1179 v = (v << 4) + c - '0'; 1180 else if (c >= 'A' && c <= 'F') 1181 v = (v << 4) + c - 'A' + 10; 1182 else if (c >= 'a' && c <= 'f') 1183 v = (v << 4) + c - 'a' + 10; 1184 else 1185 break; 1186 } 1187 pungetc(); 1188 break; 1189 case '0': case '1': case '2': case '3': 1190 case '4': case '5': case '6': case '7': 1191 v = c - '0'; 1192 c = pgetc(); 1193 if (c >= '0' && c <= '7') { 1194 v <<= 3; 1195 v += c - '0'; 1196 c = pgetc(); 1197 if (c >= '0' && c <= '7') { 1198 v <<= 3; 1199 v += c - '0'; 1200 } else 1201 pungetc(); 1202 } else 1203 pungetc(); 1204 break; 1205 case 'c': 1206 c = pgetc(); 1207 if (c < 0x3f || c > 0x7a || c == 0x60) 1208 synerror("Bad escape sequence"); 1209 if (c == '\\' && pgetc() != '\\') 1210 synerror("Bad escape sequence"); 1211 if (c == '?') 1212 v = 127; 1213 else 1214 v = c & 0x1f; 1215 break; 1216 case 'u': 1217 case 'U': 1218 n = c == 'U' ? 8 : 4; 1219 v = 0; 1220 for (i = 0; i < n; i++) { 1221 c = pgetc(); 1222 if (c >= '0' && c <= '9') 1223 v = (v << 4) + c - '0'; 1224 else if (c >= 'A' && c <= 'F') 1225 v = (v << 4) + c - 'A' + 10; 1226 else if (c >= 'a' && c <= 'f') 1227 v = (v << 4) + c - 'a' + 10; 1228 else 1229 synerror("Bad escape sequence"); 1230 } 1231 if (v == 0 || (v >= 0xd800 && v <= 0xdfff)) 1232 synerror("Bad escape sequence"); 1233 /* We really need iconv here. */ 1234 if (initial_localeisutf8 && v > 127) { 1235 CHECKSTRSPACE(4, out); 1236 /* 1237 * We cannot use wctomb() as the locale may have 1238 * changed. 1239 */ 1240 if (v <= 0x7ff) { 1241 USTPUTC(0xc0 | v >> 6, out); 1242 USTPUTC(0x80 | (v & 0x3f), out); 1243 return out; 1244 } else if (v <= 0xffff) { 1245 USTPUTC(0xe0 | v >> 12, out); 1246 USTPUTC(0x80 | ((v >> 6) & 0x3f), out); 1247 USTPUTC(0x80 | (v & 0x3f), out); 1248 return out; 1249 } else if (v <= 0x10ffff) { 1250 USTPUTC(0xf0 | v >> 18, out); 1251 USTPUTC(0x80 | ((v >> 12) & 0x3f), out); 1252 USTPUTC(0x80 | ((v >> 6) & 0x3f), out); 1253 USTPUTC(0x80 | (v & 0x3f), out); 1254 return out; 1255 } 1256 } 1257 if (v > 127) 1258 v = '?'; 1259 break; 1260 default: 1261 synerror("Bad escape sequence"); 1262 } 1263 v = (char)v; 1264 /* 1265 * We can't handle NUL bytes. 1266 * POSIX says we should skip till the closing quote. 1267 */ 1268 if (v == '\0') { 1269 while ((c = pgetc()) != '\'') { 1270 if (c == '\\') 1271 c = pgetc(); 1272 if (c == PEOF) 1273 synerror("Unterminated quoted string"); 1274 } 1275 pungetc(); 1276 return out; 1277 } 1278 if (SQSYNTAX[v] == CCTL) 1279 USTPUTC(CTLESC, out); 1280 USTPUTC(v, out); 1281 return out; 1282 } 1283 1284 1285 /* 1286 * If eofmark is NULL, read a word or a redirection symbol. If eofmark 1287 * is not NULL, read a here document. In the latter case, eofmark is the 1288 * word which marks the end of the document and striptabs is true if 1289 * leading tabs should be stripped from the document. The argument firstc 1290 * is the first character of the input token or document. 1291 * 1292 * Because C does not have internal subroutines, I have simulated them 1293 * using goto's to implement the subroutine linkage. The following macros 1294 * will run code that appears at the end of readtoken1. 1295 */ 1296 1297 #define CHECKEND() {goto checkend; checkend_return:;} 1298 #define PARSEREDIR() {goto parseredir; parseredir_return:;} 1299 #define PARSESUB() {goto parsesub; parsesub_return:;} 1300 #define PARSEARITH() {goto parsearith; parsearith_return:;} 1301 1302 static int 1303 readtoken1(int firstc, char const *initialsyntax, const char *eofmark, 1304 int striptabs) 1305 { 1306 int c = firstc; 1307 char *out; 1308 int len; 1309 char line[EOFMARKLEN + 1]; 1310 struct nodelist *bqlist; 1311 int quotef; 1312 int newvarnest; 1313 int level; 1314 int synentry; 1315 struct tokenstate state_static[MAXNEST_static]; 1316 int maxnest = MAXNEST_static; 1317 struct tokenstate *state = state_static; 1318 int sqiscstyle = 0; 1319 1320 startlinno = plinno; 1321 quotef = 0; 1322 bqlist = NULL; 1323 newvarnest = 0; 1324 level = 0; 1325 state[level].syntax = initialsyntax; 1326 state[level].parenlevel = 0; 1327 state[level].category = TSTATE_TOP; 1328 1329 STARTSTACKSTR(out); 1330 loop: { /* for each line, until end of word */ 1331 CHECKEND(); /* set c to PEOF if at end of here document */ 1332 for (;;) { /* until end of line or end of word */ 1333 CHECKSTRSPACE(4, out); /* permit 4 calls to USTPUTC */ 1334 1335 synentry = state[level].syntax[c]; 1336 1337 switch(synentry) { 1338 case CNL: /* '\n' */ 1339 if (state[level].syntax == BASESYNTAX) 1340 goto endword; /* exit outer loop */ 1341 USTPUTC(c, out); 1342 plinno++; 1343 if (doprompt) 1344 setprompt(2); 1345 else 1346 setprompt(0); 1347 c = pgetc(); 1348 goto loop; /* continue outer loop */ 1349 case CSBACK: 1350 if (sqiscstyle) { 1351 out = readcstyleesc(out); 1352 break; 1353 } 1354 /* FALLTHROUGH */ 1355 case CWORD: 1356 USTPUTC(c, out); 1357 break; 1358 case CCTL: 1359 if (eofmark == NULL || initialsyntax != SQSYNTAX) 1360 USTPUTC(CTLESC, out); 1361 USTPUTC(c, out); 1362 break; 1363 case CBACK: /* backslash */ 1364 c = pgetc(); 1365 if (c == PEOF) { 1366 USTPUTC('\\', out); 1367 pungetc(); 1368 } else if (c == '\n') { 1369 plinno++; 1370 if (doprompt) 1371 setprompt(2); 1372 else 1373 setprompt(0); 1374 } else { 1375 if (state[level].syntax == DQSYNTAX && 1376 c != '\\' && c != '`' && c != '$' && 1377 (c != '"' || (eofmark != NULL && 1378 newvarnest == 0)) && 1379 (c != '}' || state[level].category != TSTATE_VAR_OLD)) 1380 USTPUTC('\\', out); 1381 if ((eofmark == NULL || 1382 newvarnest > 0) && 1383 state[level].syntax == BASESYNTAX) 1384 USTPUTC(CTLQUOTEMARK, out); 1385 if (SQSYNTAX[c] == CCTL) 1386 USTPUTC(CTLESC, out); 1387 USTPUTC(c, out); 1388 if ((eofmark == NULL || 1389 newvarnest > 0) && 1390 state[level].syntax == BASESYNTAX && 1391 state[level].category == TSTATE_VAR_OLD) 1392 USTPUTC(CTLQUOTEEND, out); 1393 quotef++; 1394 } 1395 break; 1396 case CSQUOTE: 1397 USTPUTC(CTLQUOTEMARK, out); 1398 state[level].syntax = SQSYNTAX; 1399 sqiscstyle = 0; 1400 break; 1401 case CDQUOTE: 1402 USTPUTC(CTLQUOTEMARK, out); 1403 state[level].syntax = DQSYNTAX; 1404 break; 1405 case CENDQUOTE: 1406 if (eofmark != NULL && newvarnest == 0) 1407 USTPUTC(c, out); 1408 else { 1409 if (state[level].category == TSTATE_VAR_OLD) 1410 USTPUTC(CTLQUOTEEND, out); 1411 state[level].syntax = BASESYNTAX; 1412 quotef++; 1413 } 1414 break; 1415 case CVAR: /* '$' */ 1416 PARSESUB(); /* parse substitution */ 1417 break; 1418 case CENDVAR: /* '}' */ 1419 if (level > 0 && 1420 ((state[level].category == TSTATE_VAR_OLD && 1421 state[level].syntax == 1422 state[level - 1].syntax) || 1423 (state[level].category == TSTATE_VAR_NEW && 1424 state[level].syntax == BASESYNTAX))) { 1425 if (state[level].category == TSTATE_VAR_NEW) 1426 newvarnest--; 1427 level--; 1428 USTPUTC(CTLENDVAR, out); 1429 } else { 1430 USTPUTC(c, out); 1431 } 1432 break; 1433 case CLP: /* '(' in arithmetic */ 1434 state[level].parenlevel++; 1435 USTPUTC(c, out); 1436 break; 1437 case CRP: /* ')' in arithmetic */ 1438 if (state[level].parenlevel > 0) { 1439 USTPUTC(c, out); 1440 --state[level].parenlevel; 1441 } else { 1442 if (pgetc() == ')') { 1443 if (level > 0 && 1444 state[level].category == TSTATE_ARITH) { 1445 level--; 1446 USTPUTC(CTLENDARI, out); 1447 } else 1448 USTPUTC(')', out); 1449 } else { 1450 /* 1451 * unbalanced parens 1452 * (don't 2nd guess - no error) 1453 */ 1454 pungetc(); 1455 USTPUTC(')', out); 1456 } 1457 } 1458 break; 1459 case CBQUOTE: /* '`' */ 1460 out = parsebackq(out, &bqlist, 1, 1461 state[level].syntax == DQSYNTAX && 1462 (eofmark == NULL || newvarnest > 0), 1463 state[level].syntax == DQSYNTAX || state[level].syntax == ARISYNTAX); 1464 break; 1465 case CEOF: 1466 goto endword; /* exit outer loop */ 1467 case CIGN: 1468 break; 1469 default: 1470 if (level == 0) 1471 goto endword; /* exit outer loop */ 1472 USTPUTC(c, out); 1473 } 1474 c = pgetc_macro(); 1475 } 1476 } 1477 endword: 1478 if (state[level].syntax == ARISYNTAX) 1479 synerror("Missing '))'"); 1480 if (state[level].syntax != BASESYNTAX && eofmark == NULL) 1481 synerror("Unterminated quoted string"); 1482 if (state[level].category == TSTATE_VAR_OLD || 1483 state[level].category == TSTATE_VAR_NEW) { 1484 startlinno = plinno; 1485 synerror("Missing '}'"); 1486 } 1487 if (state != state_static) 1488 parser_temp_free_upto(state); 1489 USTPUTC('\0', out); 1490 len = out - stackblock(); 1491 out = stackblock(); 1492 if (eofmark == NULL) { 1493 if ((c == '>' || c == '<') 1494 && quotef == 0 1495 && len <= 2 1496 && (*out == '\0' || is_digit(*out))) { 1497 PARSEREDIR(); 1498 return lasttoken = TREDIR; 1499 } else { 1500 pungetc(); 1501 } 1502 } 1503 quoteflag = quotef; 1504 backquotelist = bqlist; 1505 grabstackblock(len); 1506 wordtext = out; 1507 return lasttoken = TWORD; 1508 /* end of readtoken routine */ 1509 1510 1511 /* 1512 * Check to see whether we are at the end of the here document. When this 1513 * is called, c is set to the first character of the next input line. If 1514 * we are at the end of the here document, this routine sets the c to PEOF. 1515 */ 1516 1517 checkend: { 1518 if (eofmark) { 1519 if (striptabs) { 1520 while (c == '\t') 1521 c = pgetc(); 1522 } 1523 if (c == *eofmark) { 1524 if (pfgets(line, sizeof line) != NULL) { 1525 const char *p, *q; 1526 1527 p = line; 1528 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++); 1529 if ((*p == '\0' || *p == '\n') && *q == '\0') { 1530 c = PEOF; 1531 if (*p == '\n') { 1532 plinno++; 1533 needprompt = doprompt; 1534 } 1535 } else { 1536 pushstring(line, strlen(line), NULL); 1537 } 1538 } 1539 } 1540 } 1541 goto checkend_return; 1542 } 1543 1544 1545 /* 1546 * Parse a redirection operator. The variable "out" points to a string 1547 * specifying the fd to be redirected. The variable "c" contains the 1548 * first character of the redirection operator. 1549 */ 1550 1551 parseredir: { 1552 char fd = *out; 1553 union node *np; 1554 1555 np = (union node *)stalloc(sizeof (struct nfile)); 1556 if (c == '>') { 1557 np->nfile.fd = 1; 1558 c = pgetc(); 1559 if (c == '>') 1560 np->type = NAPPEND; 1561 else if (c == '&') 1562 np->type = NTOFD; 1563 else if (c == '|') 1564 np->type = NCLOBBER; 1565 else { 1566 np->type = NTO; 1567 pungetc(); 1568 } 1569 } else { /* c == '<' */ 1570 np->nfile.fd = 0; 1571 c = pgetc(); 1572 if (c == '<') { 1573 if (sizeof (struct nfile) != sizeof (struct nhere)) { 1574 np = (union node *)stalloc(sizeof (struct nhere)); 1575 np->nfile.fd = 0; 1576 } 1577 np->type = NHERE; 1578 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc)); 1579 heredoc->here = np; 1580 if ((c = pgetc()) == '-') { 1581 heredoc->striptabs = 1; 1582 } else { 1583 heredoc->striptabs = 0; 1584 pungetc(); 1585 } 1586 } else if (c == '&') 1587 np->type = NFROMFD; 1588 else if (c == '>') 1589 np->type = NFROMTO; 1590 else { 1591 np->type = NFROM; 1592 pungetc(); 1593 } 1594 } 1595 if (fd != '\0') 1596 np->nfile.fd = digit_val(fd); 1597 redirnode = np; 1598 goto parseredir_return; 1599 } 1600 1601 1602 /* 1603 * Parse a substitution. At this point, we have read the dollar sign 1604 * and nothing else. 1605 */ 1606 1607 parsesub: { 1608 char buf[10]; 1609 int subtype; 1610 int typeloc; 1611 int flags; 1612 char *p; 1613 static const char types[] = "}-+?="; 1614 int bracketed_name = 0; /* used to handle ${[0-9]*} variables */ 1615 int linno; 1616 int length; 1617 int c1; 1618 1619 c = pgetc(); 1620 if (c == '(') { /* $(command) or $((arith)) */ 1621 if (pgetc() == '(') { 1622 PARSEARITH(); 1623 } else { 1624 pungetc(); 1625 out = parsebackq(out, &bqlist, 0, 1626 state[level].syntax == DQSYNTAX && 1627 (eofmark == NULL || newvarnest > 0), 1628 state[level].syntax == DQSYNTAX || 1629 state[level].syntax == ARISYNTAX); 1630 } 1631 } else if (c == '{' || is_name(c) || is_special(c)) { 1632 USTPUTC(CTLVAR, out); 1633 typeloc = out - stackblock(); 1634 USTPUTC(VSNORMAL, out); 1635 subtype = VSNORMAL; 1636 flags = 0; 1637 if (c == '{') { 1638 bracketed_name = 1; 1639 c = pgetc(); 1640 subtype = 0; 1641 } 1642 varname: 1643 if (!is_eof(c) && is_name(c)) { 1644 length = 0; 1645 do { 1646 STPUTC(c, out); 1647 c = pgetc(); 1648 length++; 1649 } while (!is_eof(c) && is_in_name(c)); 1650 if (length == 6 && 1651 strncmp(out - length, "LINENO", length) == 0) { 1652 /* Replace the variable name with the 1653 * current line number. */ 1654 linno = plinno; 1655 if (funclinno != 0) 1656 linno -= funclinno - 1; 1657 snprintf(buf, sizeof(buf), "%d", linno); 1658 STADJUST(-6, out); 1659 STPUTS(buf, out); 1660 flags |= VSLINENO; 1661 } 1662 } else if (is_digit(c)) { 1663 if (bracketed_name) { 1664 do { 1665 STPUTC(c, out); 1666 c = pgetc(); 1667 } while (is_digit(c)); 1668 } else { 1669 STPUTC(c, out); 1670 c = pgetc(); 1671 } 1672 } else if (is_special(c)) { 1673 c1 = c; 1674 c = pgetc(); 1675 if (subtype == 0 && c1 == '#') { 1676 subtype = VSLENGTH; 1677 if (strchr(types, c) == NULL && c != ':' && 1678 c != '#' && c != '%') 1679 goto varname; 1680 c1 = c; 1681 c = pgetc(); 1682 if (c1 != '}' && c == '}') { 1683 pungetc(); 1684 c = c1; 1685 goto varname; 1686 } 1687 pungetc(); 1688 c = c1; 1689 c1 = '#'; 1690 subtype = 0; 1691 } 1692 USTPUTC(c1, out); 1693 } else { 1694 subtype = VSERROR; 1695 if (c == '}') 1696 pungetc(); 1697 else if (c == '\n' || c == PEOF) 1698 synerror("Unexpected end of line in substitution"); 1699 else 1700 USTPUTC(c, out); 1701 } 1702 if (subtype == 0) { 1703 switch (c) { 1704 case ':': 1705 flags |= VSNUL; 1706 c = pgetc(); 1707 /*FALLTHROUGH*/ 1708 default: 1709 p = strchr(types, c); 1710 if (p == NULL) { 1711 if (c == '\n' || c == PEOF) 1712 synerror("Unexpected end of line in substitution"); 1713 if (flags == VSNUL) 1714 STPUTC(':', out); 1715 STPUTC(c, out); 1716 subtype = VSERROR; 1717 } else 1718 subtype = p - types + VSNORMAL; 1719 break; 1720 case '%': 1721 case '#': 1722 { 1723 int cc = c; 1724 subtype = c == '#' ? VSTRIMLEFT : 1725 VSTRIMRIGHT; 1726 c = pgetc(); 1727 if (c == cc) 1728 subtype++; 1729 else 1730 pungetc(); 1731 break; 1732 } 1733 } 1734 } else if (subtype != VSERROR) { 1735 if (subtype == VSLENGTH && c != '}') 1736 subtype = VSERROR; 1737 pungetc(); 1738 } 1739 STPUTC('=', out); 1740 if (state[level].syntax == DQSYNTAX || 1741 state[level].syntax == ARISYNTAX) 1742 flags |= VSQUOTE; 1743 *(stackblock() + typeloc) = subtype | flags; 1744 if (subtype != VSNORMAL) { 1745 if (level + 1 >= maxnest) { 1746 maxnest *= 2; 1747 if (state == state_static) { 1748 state = parser_temp_alloc( 1749 maxnest * sizeof(*state)); 1750 memcpy(state, state_static, 1751 MAXNEST_static * sizeof(*state)); 1752 } else 1753 state = parser_temp_realloc(state, 1754 maxnest * sizeof(*state)); 1755 } 1756 level++; 1757 state[level].parenlevel = 0; 1758 if (subtype == VSMINUS || subtype == VSPLUS || 1759 subtype == VSQUESTION || subtype == VSASSIGN) { 1760 /* 1761 * For operators that were in the Bourne shell, 1762 * inherit the double-quote state. 1763 */ 1764 state[level].syntax = state[level - 1].syntax; 1765 state[level].category = TSTATE_VAR_OLD; 1766 } else { 1767 /* 1768 * The other operators take a pattern, 1769 * so go to BASESYNTAX. 1770 * Also, ' and " are now special, even 1771 * in here documents. 1772 */ 1773 state[level].syntax = BASESYNTAX; 1774 state[level].category = TSTATE_VAR_NEW; 1775 newvarnest++; 1776 } 1777 } 1778 } else if (c == '\'' && state[level].syntax == BASESYNTAX) { 1779 /* $'cstylequotes' */ 1780 USTPUTC(CTLQUOTEMARK, out); 1781 state[level].syntax = SQSYNTAX; 1782 sqiscstyle = 1; 1783 } else { 1784 USTPUTC('$', out); 1785 pungetc(); 1786 } 1787 goto parsesub_return; 1788 } 1789 1790 1791 /* 1792 * Parse an arithmetic expansion (indicate start of one and set state) 1793 */ 1794 parsearith: { 1795 1796 if (level + 1 >= maxnest) { 1797 maxnest *= 2; 1798 if (state == state_static) { 1799 state = parser_temp_alloc( 1800 maxnest * sizeof(*state)); 1801 memcpy(state, state_static, 1802 MAXNEST_static * sizeof(*state)); 1803 } else 1804 state = parser_temp_realloc(state, 1805 maxnest * sizeof(*state)); 1806 } 1807 level++; 1808 state[level].syntax = ARISYNTAX; 1809 state[level].parenlevel = 0; 1810 state[level].category = TSTATE_ARITH; 1811 USTPUTC(CTLARI, out); 1812 if (state[level - 1].syntax == DQSYNTAX) 1813 USTPUTC('"',out); 1814 else 1815 USTPUTC(' ',out); 1816 goto parsearith_return; 1817 } 1818 1819 } /* end of readtoken */ 1820 1821 1822 void 1823 resetparser(void) 1824 { 1825 tokpushback = 0; 1826 checkkwd = 0; 1827 } 1828 1829 1830 /* 1831 * Returns true if the text contains nothing to expand (no dollar signs 1832 * or backquotes). 1833 */ 1834 1835 static int 1836 noexpand(char *text) 1837 { 1838 char *p; 1839 char c; 1840 1841 p = text; 1842 while ((c = *p++) != '\0') { 1843 if ( c == CTLQUOTEMARK) 1844 continue; 1845 if (c == CTLESC) 1846 p++; 1847 else if (BASESYNTAX[(int)c] == CCTL) 1848 return 0; 1849 } 1850 return 1; 1851 } 1852 1853 1854 /* 1855 * Return true if the argument is a legal variable name (a letter or 1856 * underscore followed by zero or more letters, underscores, and digits). 1857 */ 1858 1859 int 1860 goodname(const char *name) 1861 { 1862 const char *p; 1863 1864 p = name; 1865 if (! is_name(*p)) 1866 return 0; 1867 while (*++p) { 1868 if (! is_in_name(*p)) 1869 return 0; 1870 } 1871 return 1; 1872 } 1873 1874 1875 int 1876 isassignment(const char *p) 1877 { 1878 if (!is_name(*p)) 1879 return 0; 1880 p++; 1881 for (;;) { 1882 if (*p == '=') 1883 return 1; 1884 else if (!is_in_name(*p)) 1885 return 0; 1886 p++; 1887 } 1888 } 1889 1890 1891 /* 1892 * Called when an unexpected token is read during the parse. The argument 1893 * is the token that is expected, or -1 if more than one type of token can 1894 * occur at this point. 1895 */ 1896 1897 static void 1898 synexpect(int token) 1899 { 1900 char msg[64]; 1901 1902 if (token >= 0) { 1903 fmtstr(msg, 64, "%s unexpected (expecting %s)", 1904 tokname[lasttoken], tokname[token]); 1905 } else { 1906 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]); 1907 } 1908 synerror(msg); 1909 } 1910 1911 1912 static void 1913 synerror(const char *msg) 1914 { 1915 if (commandname) 1916 outfmt(out2, "%s: %d: ", commandname, startlinno); 1917 outfmt(out2, "Syntax error: %s\n", msg); 1918 error((char *)NULL); 1919 } 1920 1921 static void 1922 setprompt(int which) 1923 { 1924 whichprompt = which; 1925 1926 #ifndef NO_HISTORY 1927 if (!el) 1928 #endif 1929 { 1930 out2str(getprompt(NULL)); 1931 flushout(out2); 1932 } 1933 } 1934 1935 /* 1936 * called by editline -- any expansions to the prompt 1937 * should be added here. 1938 */ 1939 char * 1940 getprompt(void *unused __unused) 1941 { 1942 static char ps[PROMPTLEN]; 1943 char *fmt; 1944 const char *pwd; 1945 int i, trim; 1946 static char internal_error[] = "??"; 1947 1948 /* 1949 * Select prompt format. 1950 */ 1951 switch (whichprompt) { 1952 case 0: 1953 fmt = nullstr; 1954 break; 1955 case 1: 1956 fmt = ps1val(); 1957 break; 1958 case 2: 1959 fmt = ps2val(); 1960 break; 1961 default: 1962 return internal_error; 1963 } 1964 1965 /* 1966 * Format prompt string. 1967 */ 1968 for (i = 0; (i < 127) && (*fmt != '\0'); i++, fmt++) 1969 if (*fmt == '\\') 1970 switch (*++fmt) { 1971 1972 /* 1973 * Hostname. 1974 * 1975 * \h specifies just the local hostname, 1976 * \H specifies fully-qualified hostname. 1977 */ 1978 case 'h': 1979 case 'H': 1980 ps[i] = '\0'; 1981 gethostname(&ps[i], PROMPTLEN - i); 1982 /* Skip to end of hostname. */ 1983 trim = (*fmt == 'h') ? '.' : '\0'; 1984 while ((ps[i+1] != '\0') && (ps[i+1] != trim)) 1985 i++; 1986 break; 1987 1988 /* 1989 * Working directory. 1990 * 1991 * \W specifies just the final component, 1992 * \w specifies the entire path. 1993 */ 1994 case 'W': 1995 case 'w': 1996 pwd = lookupvar("PWD"); 1997 if (pwd == NULL) 1998 pwd = "?"; 1999 if (*fmt == 'W' && 2000 *pwd == '/' && pwd[1] != '\0') 2001 strlcpy(&ps[i], strrchr(pwd, '/') + 1, 2002 PROMPTLEN - i); 2003 else 2004 strlcpy(&ps[i], pwd, PROMPTLEN - i); 2005 /* Skip to end of path. */ 2006 while (ps[i + 1] != '\0') 2007 i++; 2008 break; 2009 2010 /* 2011 * Superuser status. 2012 * 2013 * '$' for normal users, '#' for root. 2014 */ 2015 case '$': 2016 ps[i] = (geteuid() != 0) ? '$' : '#'; 2017 break; 2018 2019 /* 2020 * A literal \. 2021 */ 2022 case '\\': 2023 ps[i] = '\\'; 2024 break; 2025 2026 /* 2027 * Emit unrecognized formats verbatim. 2028 */ 2029 default: 2030 ps[i++] = '\\'; 2031 ps[i] = *fmt; 2032 break; 2033 } 2034 else 2035 ps[i] = *fmt; 2036 ps[i] = '\0'; 2037 return (ps); 2038 } 2039 2040 2041 const char * 2042 expandstr(const char *ps) 2043 { 2044 union node n; 2045 struct jmploc jmploc; 2046 struct jmploc *const savehandler = handler; 2047 const int saveprompt = doprompt; 2048 struct parsefile *const savetopfile = getcurrentfile(); 2049 struct parser_temp *const saveparser_temp = parser_temp; 2050 const char *result = NULL; 2051 2052 if (!setjmp(jmploc.loc)) { 2053 handler = &jmploc; 2054 parser_temp = NULL; 2055 setinputstring(ps, 1); 2056 doprompt = 0; 2057 readtoken1(pgetc(), DQSYNTAX, "\n\n", 0); 2058 if (backquotelist != NULL) 2059 error("Command substitution not allowed here"); 2060 2061 n.narg.type = NARG; 2062 n.narg.next = NULL; 2063 n.narg.text = wordtext; 2064 n.narg.backquote = backquotelist; 2065 2066 expandarg(&n, NULL, 0); 2067 result = stackblock(); 2068 INTOFF; 2069 } 2070 handler = savehandler; 2071 doprompt = saveprompt; 2072 popfilesupto(savetopfile); 2073 if (parser_temp != saveparser_temp) { 2074 parser_temp_free_all(); 2075 parser_temp = saveparser_temp; 2076 } 2077 if (result != NULL) { 2078 INTON; 2079 } else if (exception == EXINT) 2080 raise(SIGINT); 2081 return result; 2082 } 2083