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 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by the University of 19 * California, Berkeley and its contributors. 20 * 4. Neither the name of the University nor the names of its contributors 21 * may be used to endorse or promote products derived from this software 22 * without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 34 * SUCH DAMAGE. 35 */ 36 37 #ifndef lint 38 #if 0 39 static char sccsid[] = "@(#)parser.c 8.7 (Berkeley) 5/16/95"; 40 #endif 41 #endif /* not lint */ 42 #include <sys/cdefs.h> 43 __FBSDID("$FreeBSD$"); 44 45 #include <stdlib.h> 46 47 #include "shell.h" 48 #include "parser.h" 49 #include "nodes.h" 50 #include "expand.h" /* defines rmescapes() */ 51 #include "syntax.h" 52 #include "options.h" 53 #include "input.h" 54 #include "output.h" 55 #include "var.h" 56 #include "error.h" 57 #include "memalloc.h" 58 #include "mystring.h" 59 #include "alias.h" 60 #include "show.h" 61 #include "eval.h" 62 #ifndef NO_HISTORY 63 #include "myhistedit.h" 64 #endif 65 66 /* 67 * Shell command parser. 68 */ 69 70 #define EOFMARKLEN 79 71 72 /* values returned by readtoken */ 73 #include "token.h" 74 75 76 77 struct heredoc { 78 struct heredoc *next; /* next here document in list */ 79 union node *here; /* redirection node */ 80 char *eofmark; /* string indicating end of input */ 81 int striptabs; /* if set, strip leading tabs */ 82 }; 83 84 85 86 STATIC struct heredoc *heredoclist; /* list of here documents to read */ 87 STATIC int parsebackquote; /* nonzero if we are inside backquotes */ 88 STATIC int doprompt; /* if set, prompt the user */ 89 STATIC int needprompt; /* true if interactive and at start of line */ 90 STATIC int lasttoken; /* last token read */ 91 MKINIT int tokpushback; /* last token pushed back */ 92 STATIC char *wordtext; /* text of last word returned by readtoken */ 93 MKINIT int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */ 94 STATIC struct nodelist *backquotelist; 95 STATIC union node *redirnode; 96 STATIC struct heredoc *heredoc; 97 STATIC int quoteflag; /* set if (part of) last token was quoted */ 98 STATIC int startlinno; /* line # where last token started */ 99 100 /* XXX When 'noaliases' is set to one, no alias expansion takes place. */ 101 static int noaliases = 0; 102 103 #define GDB_HACK 1 /* avoid local declarations which gdb can't handle */ 104 #ifdef GDB_HACK 105 static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'}; 106 static const char types[] = "}-+?="; 107 #endif 108 109 110 STATIC union node *list(int); 111 STATIC union node *andor(void); 112 STATIC union node *pipeline(void); 113 STATIC union node *command(void); 114 STATIC union node *simplecmd(union node **, union node *); 115 STATIC union node *makename(void); 116 STATIC void parsefname(void); 117 STATIC void parseheredoc(void); 118 STATIC int peektoken(void); 119 STATIC int readtoken(void); 120 STATIC int xxreadtoken(void); 121 STATIC int readtoken1(int, char const *, char *, int); 122 STATIC int noexpand(char *); 123 STATIC void synexpect(int); 124 STATIC void synerror(char *); 125 STATIC void setprompt(int); 126 127 128 /* 129 * Read and parse a command. Returns NEOF on end of file. (NULL is a 130 * valid parse tree indicating a blank line.) 131 */ 132 133 union node * 134 parsecmd(int interact) 135 { 136 int t; 137 138 tokpushback = 0; 139 doprompt = interact; 140 if (doprompt) 141 setprompt(1); 142 else 143 setprompt(0); 144 needprompt = 0; 145 t = readtoken(); 146 if (t == TEOF) 147 return NEOF; 148 if (t == TNL) 149 return NULL; 150 tokpushback++; 151 return list(1); 152 } 153 154 155 STATIC union node * 156 list(int nlflag) 157 { 158 union node *n1, *n2, *n3; 159 int tok; 160 161 checkkwd = 2; 162 if (nlflag == 0 && tokendlist[peektoken()]) 163 return NULL; 164 n1 = NULL; 165 for (;;) { 166 n2 = andor(); 167 tok = readtoken(); 168 if (tok == TBACKGND) { 169 if (n2->type == NCMD || n2->type == NPIPE) { 170 n2->ncmd.backgnd = 1; 171 } else if (n2->type == NREDIR) { 172 n2->type = NBACKGND; 173 } else { 174 n3 = (union node *)stalloc(sizeof (struct nredir)); 175 n3->type = NBACKGND; 176 n3->nredir.n = n2; 177 n3->nredir.redirect = NULL; 178 n2 = n3; 179 } 180 } 181 if (n1 == NULL) { 182 n1 = n2; 183 } 184 else { 185 n3 = (union node *)stalloc(sizeof (struct nbinary)); 186 n3->type = NSEMI; 187 n3->nbinary.ch1 = n1; 188 n3->nbinary.ch2 = n2; 189 n1 = n3; 190 } 191 switch (tok) { 192 case TBACKGND: 193 case TSEMI: 194 tok = readtoken(); 195 /* FALLTHROUGH */ 196 case TNL: 197 if (tok == TNL) { 198 parseheredoc(); 199 if (nlflag) 200 return n1; 201 } else { 202 tokpushback++; 203 } 204 checkkwd = 2; 205 if (tokendlist[peektoken()]) 206 return n1; 207 break; 208 case TEOF: 209 if (heredoclist) 210 parseheredoc(); 211 else 212 pungetc(); /* push back EOF on input */ 213 return n1; 214 default: 215 if (nlflag) 216 synexpect(-1); 217 tokpushback++; 218 return n1; 219 } 220 } 221 } 222 223 224 225 STATIC union node * 226 andor(void) 227 { 228 union node *n1, *n2, *n3; 229 int t; 230 231 n1 = pipeline(); 232 for (;;) { 233 if ((t = readtoken()) == TAND) { 234 t = NAND; 235 } else if (t == TOR) { 236 t = NOR; 237 } else { 238 tokpushback++; 239 return n1; 240 } 241 n2 = pipeline(); 242 n3 = (union node *)stalloc(sizeof (struct nbinary)); 243 n3->type = t; 244 n3->nbinary.ch1 = n1; 245 n3->nbinary.ch2 = n2; 246 n1 = n3; 247 } 248 } 249 250 251 252 STATIC union node * 253 pipeline(void) 254 { 255 union node *n1, *n2, *pipenode; 256 struct nodelist *lp, *prev; 257 int negate; 258 259 negate = 0; 260 TRACE(("pipeline: entered\n")); 261 while (readtoken() == TNOT) 262 negate = !negate; 263 tokpushback++; 264 n1 = command(); 265 if (readtoken() == TPIPE) { 266 pipenode = (union node *)stalloc(sizeof (struct npipe)); 267 pipenode->type = NPIPE; 268 pipenode->npipe.backgnd = 0; 269 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 270 pipenode->npipe.cmdlist = lp; 271 lp->n = n1; 272 do { 273 prev = lp; 274 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 275 lp->n = command(); 276 prev->next = lp; 277 } while (readtoken() == TPIPE); 278 lp->next = NULL; 279 n1 = pipenode; 280 } 281 tokpushback++; 282 if (negate) { 283 n2 = (union node *)stalloc(sizeof (struct nnot)); 284 n2->type = NNOT; 285 n2->nnot.com = n1; 286 return n2; 287 } else 288 return n1; 289 } 290 291 292 293 STATIC union node * 294 command(void) 295 { 296 union node *n1, *n2; 297 union node *ap, **app; 298 union node *cp, **cpp; 299 union node *redir, **rpp; 300 int t, negate = 0; 301 302 checkkwd = 2; 303 redir = NULL; 304 n1 = NULL; 305 rpp = &redir; 306 307 /* Check for redirection which may precede command */ 308 while (readtoken() == TREDIR) { 309 *rpp = n2 = redirnode; 310 rpp = &n2->nfile.next; 311 parsefname(); 312 } 313 tokpushback++; 314 315 while (readtoken() == TNOT) { 316 TRACE(("command: TNOT recognized\n")); 317 negate = !negate; 318 } 319 tokpushback++; 320 321 switch (readtoken()) { 322 case TIF: 323 n1 = (union node *)stalloc(sizeof (struct nif)); 324 n1->type = NIF; 325 if ((n1->nif.test = list(0)) == NULL) 326 synexpect(-1); 327 if (readtoken() != TTHEN) 328 synexpect(TTHEN); 329 n1->nif.ifpart = list(0); 330 n2 = n1; 331 while (readtoken() == TELIF) { 332 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif)); 333 n2 = n2->nif.elsepart; 334 n2->type = NIF; 335 if ((n2->nif.test = list(0)) == NULL) 336 synexpect(-1); 337 if (readtoken() != TTHEN) 338 synexpect(TTHEN); 339 n2->nif.ifpart = list(0); 340 } 341 if (lasttoken == TELSE) 342 n2->nif.elsepart = list(0); 343 else { 344 n2->nif.elsepart = NULL; 345 tokpushback++; 346 } 347 if (readtoken() != TFI) 348 synexpect(TFI); 349 checkkwd = 1; 350 break; 351 case TWHILE: 352 case TUNTIL: { 353 int got; 354 n1 = (union node *)stalloc(sizeof (struct nbinary)); 355 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL; 356 if ((n1->nbinary.ch1 = list(0)) == NULL) 357 synexpect(-1); 358 if ((got=readtoken()) != TDO) { 359 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : "")); 360 synexpect(TDO); 361 } 362 n1->nbinary.ch2 = list(0); 363 if (readtoken() != TDONE) 364 synexpect(TDONE); 365 checkkwd = 1; 366 break; 367 } 368 case TFOR: 369 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext)) 370 synerror("Bad for loop variable"); 371 n1 = (union node *)stalloc(sizeof (struct nfor)); 372 n1->type = NFOR; 373 n1->nfor.var = wordtext; 374 if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) { 375 app = ≈ 376 while (readtoken() == TWORD) { 377 n2 = (union node *)stalloc(sizeof (struct narg)); 378 n2->type = NARG; 379 n2->narg.text = wordtext; 380 n2->narg.backquote = backquotelist; 381 *app = n2; 382 app = &n2->narg.next; 383 } 384 *app = NULL; 385 n1->nfor.args = ap; 386 if (lasttoken != TNL && lasttoken != TSEMI) 387 synexpect(-1); 388 } else { 389 #ifndef GDB_HACK 390 static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, 391 '@', '=', '\0'}; 392 #endif 393 n2 = (union node *)stalloc(sizeof (struct narg)); 394 n2->type = NARG; 395 n2->narg.text = (char *)argvars; 396 n2->narg.backquote = NULL; 397 n2->narg.next = NULL; 398 n1->nfor.args = n2; 399 /* 400 * Newline or semicolon here is optional (but note 401 * that the original Bourne shell only allowed NL). 402 */ 403 if (lasttoken != TNL && lasttoken != TSEMI) 404 tokpushback++; 405 } 406 checkkwd = 2; 407 if ((t = readtoken()) == TDO) 408 t = TDONE; 409 else if (t == TBEGIN) 410 t = TEND; 411 else 412 synexpect(-1); 413 n1->nfor.body = list(0); 414 if (readtoken() != t) 415 synexpect(t); 416 checkkwd = 1; 417 break; 418 case TCASE: 419 n1 = (union node *)stalloc(sizeof (struct ncase)); 420 n1->type = NCASE; 421 if (readtoken() != TWORD) 422 synexpect(TWORD); 423 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg)); 424 n2->type = NARG; 425 n2->narg.text = wordtext; 426 n2->narg.backquote = backquotelist; 427 n2->narg.next = NULL; 428 while (readtoken() == TNL); 429 if (lasttoken != TWORD || ! equal(wordtext, "in")) 430 synerror("expecting \"in\""); 431 cpp = &n1->ncase.cases; 432 noaliases = 1; /* turn off alias expansion */ 433 checkkwd = 2, readtoken(); 434 while (lasttoken != TESAC) { 435 *cpp = cp = (union node *)stalloc(sizeof (struct nclist)); 436 cp->type = NCLIST; 437 app = &cp->nclist.pattern; 438 if (lasttoken == TLP) 439 readtoken(); 440 for (;;) { 441 *app = ap = (union node *)stalloc(sizeof (struct narg)); 442 ap->type = NARG; 443 ap->narg.text = wordtext; 444 ap->narg.backquote = backquotelist; 445 if (checkkwd = 2, readtoken() != TPIPE) 446 break; 447 app = &ap->narg.next; 448 readtoken(); 449 } 450 ap->narg.next = NULL; 451 if (lasttoken != TRP) 452 noaliases = 0, synexpect(TRP); 453 cp->nclist.body = list(0); 454 455 checkkwd = 2; 456 if ((t = readtoken()) != TESAC) { 457 if (t != TENDCASE) 458 noaliases = 0, synexpect(TENDCASE); 459 else 460 checkkwd = 2, readtoken(); 461 } 462 cpp = &cp->nclist.next; 463 } 464 noaliases = 0; /* reset alias expansion */ 465 *cpp = NULL; 466 checkkwd = 1; 467 break; 468 case TLP: 469 n1 = (union node *)stalloc(sizeof (struct nredir)); 470 n1->type = NSUBSHELL; 471 n1->nredir.n = list(0); 472 n1->nredir.redirect = NULL; 473 if (readtoken() != TRP) 474 synexpect(TRP); 475 checkkwd = 1; 476 break; 477 case TBEGIN: 478 n1 = list(0); 479 if (readtoken() != TEND) 480 synexpect(TEND); 481 checkkwd = 1; 482 break; 483 /* Handle an empty command like other simple commands. */ 484 case TSEMI: 485 case TAND: 486 case TOR: 487 /* 488 * An empty command before a ; doesn't make much sense, and 489 * should certainly be disallowed in the case of `if ;'. 490 */ 491 if (!redir) 492 synexpect(-1); 493 case TNL: 494 case TEOF: 495 case TWORD: 496 case TRP: 497 tokpushback++; 498 n1 = simplecmd(rpp, redir); 499 goto checkneg; 500 default: 501 synexpect(-1); 502 } 503 504 /* Now check for redirection which may follow command */ 505 while (readtoken() == TREDIR) { 506 *rpp = n2 = redirnode; 507 rpp = &n2->nfile.next; 508 parsefname(); 509 } 510 tokpushback++; 511 *rpp = NULL; 512 if (redir) { 513 if (n1->type != NSUBSHELL) { 514 n2 = (union node *)stalloc(sizeof (struct nredir)); 515 n2->type = NREDIR; 516 n2->nredir.n = n1; 517 n1 = n2; 518 } 519 n1->nredir.redirect = redir; 520 } 521 522 checkneg: 523 if (negate) { 524 n2 = (union node *)stalloc(sizeof (struct nnot)); 525 n2->type = NNOT; 526 n2->nnot.com = n1; 527 return n2; 528 } 529 else 530 return n1; 531 } 532 533 534 STATIC union node * 535 simplecmd(union node **rpp, union node *redir) 536 { 537 union node *args, **app; 538 union node **orig_rpp = rpp; 539 union node *n = NULL, *n2; 540 int negate = 0; 541 542 /* If we don't have any redirections already, then we must reset */ 543 /* rpp to be the address of the local redir variable. */ 544 if (redir == 0) 545 rpp = &redir; 546 547 args = NULL; 548 app = &args; 549 /* 550 * We save the incoming value, because we need this for shell 551 * functions. There can not be a redirect or an argument between 552 * the function name and the open parenthesis. 553 */ 554 orig_rpp = rpp; 555 556 while (readtoken() == TNOT) { 557 TRACE(("command: TNOT recognized\n")); 558 negate = !negate; 559 } 560 tokpushback++; 561 562 for (;;) { 563 if (readtoken() == TWORD) { 564 n = (union node *)stalloc(sizeof (struct narg)); 565 n->type = NARG; 566 n->narg.text = wordtext; 567 n->narg.backquote = backquotelist; 568 *app = n; 569 app = &n->narg.next; 570 } else if (lasttoken == TREDIR) { 571 *rpp = n = redirnode; 572 rpp = &n->nfile.next; 573 parsefname(); /* read name of redirection file */ 574 } else if (lasttoken == TLP && app == &args->narg.next 575 && rpp == orig_rpp) { 576 /* We have a function */ 577 if (readtoken() != TRP) 578 synexpect(TRP); 579 #ifdef notdef 580 if (! goodname(n->narg.text)) 581 synerror("Bad function name"); 582 #endif 583 n->type = NDEFUN; 584 n->narg.next = command(); 585 goto checkneg; 586 } else { 587 tokpushback++; 588 break; 589 } 590 } 591 *app = NULL; 592 *rpp = NULL; 593 n = (union node *)stalloc(sizeof (struct ncmd)); 594 n->type = NCMD; 595 n->ncmd.backgnd = 0; 596 n->ncmd.args = args; 597 n->ncmd.redirect = redir; 598 599 checkneg: 600 if (negate) { 601 n2 = (union node *)stalloc(sizeof (struct nnot)); 602 n2->type = NNOT; 603 n2->nnot.com = n; 604 return n2; 605 } 606 else 607 return n; 608 } 609 610 STATIC union node * 611 makename(void) 612 { 613 union node *n; 614 615 n = (union node *)stalloc(sizeof (struct narg)); 616 n->type = NARG; 617 n->narg.next = NULL; 618 n->narg.text = wordtext; 619 n->narg.backquote = backquotelist; 620 return n; 621 } 622 623 void fixredir(union node *n, const char *text, int err) 624 { 625 TRACE(("Fix redir %s %d\n", text, err)); 626 if (!err) 627 n->ndup.vname = NULL; 628 629 if (is_digit(text[0]) && text[1] == '\0') 630 n->ndup.dupfd = digit_val(text[0]); 631 else if (text[0] == '-' && text[1] == '\0') 632 n->ndup.dupfd = -1; 633 else { 634 635 if (err) 636 synerror("Bad fd number"); 637 else 638 n->ndup.vname = makename(); 639 } 640 } 641 642 643 STATIC void 644 parsefname(void) 645 { 646 union node *n = redirnode; 647 648 if (readtoken() != TWORD) 649 synexpect(-1); 650 if (n->type == NHERE) { 651 struct heredoc *here = heredoc; 652 struct heredoc *p; 653 int i; 654 655 if (quoteflag == 0) 656 n->type = NXHERE; 657 TRACE(("Here document %d\n", n->type)); 658 if (here->striptabs) { 659 while (*wordtext == '\t') 660 wordtext++; 661 } 662 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN) 663 synerror("Illegal eof marker for << redirection"); 664 rmescapes(wordtext); 665 here->eofmark = wordtext; 666 here->next = NULL; 667 if (heredoclist == NULL) 668 heredoclist = here; 669 else { 670 for (p = heredoclist ; p->next ; p = p->next); 671 p->next = here; 672 } 673 } else if (n->type == NTOFD || n->type == NFROMFD) { 674 fixredir(n, wordtext, 0); 675 } else { 676 n->nfile.fname = makename(); 677 } 678 } 679 680 681 /* 682 * Input any here documents. 683 */ 684 685 STATIC void 686 parseheredoc(void) 687 { 688 struct heredoc *here; 689 union node *n; 690 691 while (heredoclist) { 692 here = heredoclist; 693 heredoclist = here->next; 694 if (needprompt) { 695 setprompt(2); 696 needprompt = 0; 697 } 698 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX, 699 here->eofmark, here->striptabs); 700 n = (union node *)stalloc(sizeof (struct narg)); 701 n->narg.type = NARG; 702 n->narg.next = NULL; 703 n->narg.text = wordtext; 704 n->narg.backquote = backquotelist; 705 here->here->nhere.doc = n; 706 } 707 } 708 709 STATIC int 710 peektoken(void) 711 { 712 int t; 713 714 t = readtoken(); 715 tokpushback++; 716 return (t); 717 } 718 719 STATIC int 720 readtoken(void) 721 { 722 int t; 723 int savecheckkwd = checkkwd; 724 struct alias *ap; 725 #ifdef DEBUG 726 int alreadyseen = tokpushback; 727 #endif 728 729 top: 730 t = xxreadtoken(); 731 732 if (checkkwd) { 733 /* 734 * eat newlines 735 */ 736 if (checkkwd == 2) { 737 checkkwd = 0; 738 while (t == TNL) { 739 parseheredoc(); 740 t = xxreadtoken(); 741 } 742 } else 743 checkkwd = 0; 744 /* 745 * check for keywords and aliases 746 */ 747 if (t == TWORD && !quoteflag) 748 { 749 const char * const *pp; 750 751 for (pp = parsekwd; *pp; pp++) { 752 if (**pp == *wordtext && equal(*pp, wordtext)) 753 { 754 lasttoken = t = pp - parsekwd + KWDOFFSET; 755 TRACE(("keyword %s recognized\n", tokname[t])); 756 goto out; 757 } 758 } 759 if (noaliases == 0 && 760 (ap = lookupalias(wordtext, 1)) != NULL) { 761 pushstring(ap->val, strlen(ap->val), ap); 762 checkkwd = savecheckkwd; 763 goto top; 764 } 765 } 766 out: 767 checkkwd = (t == TNOT) ? savecheckkwd : 0; 768 } 769 #ifdef DEBUG 770 if (!alreadyseen) 771 TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 772 else 773 TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 774 #endif 775 return (t); 776 } 777 778 779 /* 780 * Read the next input token. 781 * If the token is a word, we set backquotelist to the list of cmds in 782 * backquotes. We set quoteflag to true if any part of the word was 783 * quoted. 784 * If the token is TREDIR, then we set redirnode to a structure containing 785 * the redirection. 786 * In all cases, the variable startlinno is set to the number of the line 787 * on which the token starts. 788 * 789 * [Change comment: here documents and internal procedures] 790 * [Readtoken shouldn't have any arguments. Perhaps we should make the 791 * word parsing code into a separate routine. In this case, readtoken 792 * doesn't need to have any internal procedures, but parseword does. 793 * We could also make parseoperator in essence the main routine, and 794 * have parseword (readtoken1?) handle both words and redirection.] 795 */ 796 797 #define RETURN(token) return lasttoken = token 798 799 STATIC int 800 xxreadtoken(void) 801 { 802 int c; 803 804 if (tokpushback) { 805 tokpushback = 0; 806 return lasttoken; 807 } 808 if (needprompt) { 809 setprompt(2); 810 needprompt = 0; 811 } 812 startlinno = plinno; 813 for (;;) { /* until token or start of word found */ 814 c = pgetc_macro(); 815 if (c == ' ' || c == '\t') 816 continue; /* quick check for white space first */ 817 switch (c) { 818 case ' ': case '\t': 819 continue; 820 case '#': 821 while ((c = pgetc()) != '\n' && c != PEOF); 822 pungetc(); 823 continue; 824 case '\\': 825 if (pgetc() == '\n') { 826 startlinno = ++plinno; 827 if (doprompt) 828 setprompt(2); 829 else 830 setprompt(0); 831 continue; 832 } 833 pungetc(); 834 goto breakloop; 835 case '\n': 836 plinno++; 837 needprompt = doprompt; 838 RETURN(TNL); 839 case PEOF: 840 RETURN(TEOF); 841 case '&': 842 if (pgetc() == '&') 843 RETURN(TAND); 844 pungetc(); 845 RETURN(TBACKGND); 846 case '|': 847 if (pgetc() == '|') 848 RETURN(TOR); 849 pungetc(); 850 RETURN(TPIPE); 851 case ';': 852 if (pgetc() == ';') 853 RETURN(TENDCASE); 854 pungetc(); 855 RETURN(TSEMI); 856 case '(': 857 RETURN(TLP); 858 case ')': 859 RETURN(TRP); 860 default: 861 goto breakloop; 862 } 863 } 864 breakloop: 865 return readtoken1(c, BASESYNTAX, (char *)NULL, 0); 866 #undef RETURN 867 } 868 869 870 871 /* 872 * If eofmark is NULL, read a word or a redirection symbol. If eofmark 873 * is not NULL, read a here document. In the latter case, eofmark is the 874 * word which marks the end of the document and striptabs is true if 875 * leading tabs should be stripped from the document. The argument firstc 876 * is the first character of the input token or document. 877 * 878 * Because C does not have internal subroutines, I have simulated them 879 * using goto's to implement the subroutine linkage. The following macros 880 * will run code that appears at the end of readtoken1. 881 */ 882 883 #define CHECKEND() {goto checkend; checkend_return:;} 884 #define PARSEREDIR() {goto parseredir; parseredir_return:;} 885 #define PARSESUB() {goto parsesub; parsesub_return:;} 886 #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;} 887 #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;} 888 #define PARSEARITH() {goto parsearith; parsearith_return:;} 889 890 STATIC int 891 readtoken1(int firstc, char const *syntax, char *eofmark, int striptabs) 892 { 893 int c = firstc; 894 char *out; 895 int len; 896 char line[EOFMARKLEN + 1]; 897 struct nodelist *bqlist; 898 int quotef; 899 int dblquote; 900 int varnest; /* levels of variables expansion */ 901 int arinest; /* levels of arithmetic expansion */ 902 int parenlevel; /* levels of parens in arithmetic */ 903 int oldstyle; 904 char const *prevsyntax; /* syntax before arithmetic */ 905 int synentry; 906 #if __GNUC__ 907 /* Avoid longjmp clobbering */ 908 (void) &out; 909 (void) "ef; 910 (void) &dblquote; 911 (void) &varnest; 912 (void) &arinest; 913 (void) &parenlevel; 914 (void) &oldstyle; 915 (void) &prevsyntax; 916 (void) &syntax; 917 (void) &synentry; 918 #endif 919 920 startlinno = plinno; 921 dblquote = 0; 922 if (syntax == DQSYNTAX) 923 dblquote = 1; 924 quotef = 0; 925 bqlist = NULL; 926 varnest = 0; 927 arinest = 0; 928 parenlevel = 0; 929 930 STARTSTACKSTR(out); 931 loop: { /* for each line, until end of word */ 932 CHECKEND(); /* set c to PEOF if at end of here document */ 933 for (;;) { /* until end of line or end of word */ 934 CHECKSTRSPACE(3, out); /* permit 3 calls to USTPUTC */ 935 936 synentry = syntax[c]; 937 938 switch(synentry) { 939 case CNL: /* '\n' */ 940 if (syntax == BASESYNTAX) 941 goto endword; /* exit outer loop */ 942 USTPUTC(c, out); 943 plinno++; 944 if (doprompt) 945 setprompt(2); 946 else 947 setprompt(0); 948 c = pgetc(); 949 goto loop; /* continue outer loop */ 950 case CWORD: 951 USTPUTC(c, out); 952 break; 953 case CCTL: 954 if (eofmark == NULL || dblquote) 955 USTPUTC(CTLESC, out); 956 USTPUTC(c, out); 957 break; 958 case CBACK: /* backslash */ 959 c = pgetc(); 960 if (c == PEOF) { 961 USTPUTC('\\', out); 962 pungetc(); 963 } else if (c == '\n') { 964 if (doprompt) 965 setprompt(2); 966 else 967 setprompt(0); 968 } else { 969 if (dblquote && c != '\\' && 970 c != '`' && c != '$' && 971 (c != '"' || eofmark != NULL)) 972 USTPUTC('\\', out); 973 if (SQSYNTAX[c] == CCTL) 974 USTPUTC(CTLESC, out); 975 else if (eofmark == NULL) 976 USTPUTC(CTLQUOTEMARK, out); 977 USTPUTC(c, out); 978 quotef++; 979 } 980 break; 981 case CSQUOTE: 982 if (eofmark == NULL) 983 USTPUTC(CTLQUOTEMARK, out); 984 syntax = SQSYNTAX; 985 break; 986 case CDQUOTE: 987 if (eofmark == NULL) 988 USTPUTC(CTLQUOTEMARK, out); 989 syntax = DQSYNTAX; 990 dblquote = 1; 991 break; 992 case CENDQUOTE: 993 if (eofmark != NULL && arinest == 0 && 994 varnest == 0) { 995 USTPUTC(c, out); 996 } else { 997 if (arinest) { 998 syntax = ARISYNTAX; 999 dblquote = 0; 1000 } else if (eofmark == NULL) { 1001 syntax = BASESYNTAX; 1002 dblquote = 0; 1003 } 1004 quotef++; 1005 } 1006 break; 1007 case CVAR: /* '$' */ 1008 PARSESUB(); /* parse substitution */ 1009 break; 1010 case CENDVAR: /* '}' */ 1011 if (varnest > 0) { 1012 varnest--; 1013 USTPUTC(CTLENDVAR, out); 1014 } else { 1015 USTPUTC(c, out); 1016 } 1017 break; 1018 case CLP: /* '(' in arithmetic */ 1019 parenlevel++; 1020 USTPUTC(c, out); 1021 break; 1022 case CRP: /* ')' in arithmetic */ 1023 if (parenlevel > 0) { 1024 USTPUTC(c, out); 1025 --parenlevel; 1026 } else { 1027 if (pgetc() == ')') { 1028 if (--arinest == 0) { 1029 USTPUTC(CTLENDARI, out); 1030 syntax = prevsyntax; 1031 if (syntax == DQSYNTAX) 1032 dblquote = 1; 1033 else 1034 dblquote = 0; 1035 } else 1036 USTPUTC(')', out); 1037 } else { 1038 /* 1039 * unbalanced parens 1040 * (don't 2nd guess - no error) 1041 */ 1042 pungetc(); 1043 USTPUTC(')', out); 1044 } 1045 } 1046 break; 1047 case CBQUOTE: /* '`' */ 1048 PARSEBACKQOLD(); 1049 break; 1050 case CEOF: 1051 goto endword; /* exit outer loop */ 1052 default: 1053 if (varnest == 0) 1054 goto endword; /* exit outer loop */ 1055 USTPUTC(c, out); 1056 } 1057 c = pgetc_macro(); 1058 } 1059 } 1060 endword: 1061 if (syntax == ARISYNTAX) 1062 synerror("Missing '))'"); 1063 if (syntax != BASESYNTAX && ! parsebackquote && eofmark == NULL) 1064 synerror("Unterminated quoted string"); 1065 if (varnest != 0) { 1066 startlinno = plinno; 1067 synerror("Missing '}'"); 1068 } 1069 USTPUTC('\0', out); 1070 len = out - stackblock(); 1071 out = stackblock(); 1072 if (eofmark == NULL) { 1073 if ((c == '>' || c == '<') 1074 && quotef == 0 1075 && len <= 2 1076 && (*out == '\0' || is_digit(*out))) { 1077 PARSEREDIR(); 1078 return lasttoken = TREDIR; 1079 } else { 1080 pungetc(); 1081 } 1082 } 1083 quoteflag = quotef; 1084 backquotelist = bqlist; 1085 grabstackblock(len); 1086 wordtext = out; 1087 return lasttoken = TWORD; 1088 /* end of readtoken routine */ 1089 1090 1091 1092 /* 1093 * Check to see whether we are at the end of the here document. When this 1094 * is called, c is set to the first character of the next input line. If 1095 * we are at the end of the here document, this routine sets the c to PEOF. 1096 */ 1097 1098 checkend: { 1099 if (eofmark) { 1100 if (striptabs) { 1101 while (c == '\t') 1102 c = pgetc(); 1103 } 1104 if (c == *eofmark) { 1105 if (pfgets(line, sizeof line) != NULL) { 1106 char *p, *q; 1107 1108 p = line; 1109 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++); 1110 if (*p == '\n' && *q == '\0') { 1111 c = PEOF; 1112 plinno++; 1113 needprompt = doprompt; 1114 } else { 1115 pushstring(line, strlen(line), NULL); 1116 } 1117 } 1118 } 1119 } 1120 goto checkend_return; 1121 } 1122 1123 1124 /* 1125 * Parse a redirection operator. The variable "out" points to a string 1126 * specifying the fd to be redirected. The variable "c" contains the 1127 * first character of the redirection operator. 1128 */ 1129 1130 parseredir: { 1131 char fd = *out; 1132 union node *np; 1133 1134 np = (union node *)stalloc(sizeof (struct nfile)); 1135 if (c == '>') { 1136 np->nfile.fd = 1; 1137 c = pgetc(); 1138 if (c == '>') 1139 np->type = NAPPEND; 1140 else if (c == '&') 1141 np->type = NTOFD; 1142 else if (c == '|') 1143 np->type = NCLOBBER; 1144 else { 1145 np->type = NTO; 1146 pungetc(); 1147 } 1148 } else { /* c == '<' */ 1149 np->nfile.fd = 0; 1150 c = pgetc(); 1151 if (c == '<') { 1152 if (sizeof (struct nfile) != sizeof (struct nhere)) { 1153 np = (union node *)stalloc(sizeof (struct nhere)); 1154 np->nfile.fd = 0; 1155 } 1156 np->type = NHERE; 1157 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc)); 1158 heredoc->here = np; 1159 if ((c = pgetc()) == '-') { 1160 heredoc->striptabs = 1; 1161 } else { 1162 heredoc->striptabs = 0; 1163 pungetc(); 1164 } 1165 } else if (c == '&') 1166 np->type = NFROMFD; 1167 else if (c == '>') 1168 np->type = NFROMTO; 1169 else { 1170 np->type = NFROM; 1171 pungetc(); 1172 } 1173 } 1174 if (fd != '\0') 1175 np->nfile.fd = digit_val(fd); 1176 redirnode = np; 1177 goto parseredir_return; 1178 } 1179 1180 1181 /* 1182 * Parse a substitution. At this point, we have read the dollar sign 1183 * and nothing else. 1184 */ 1185 1186 parsesub: { 1187 int subtype; 1188 int typeloc; 1189 int flags; 1190 char *p; 1191 #ifndef GDB_HACK 1192 static const char types[] = "}-+?="; 1193 #endif 1194 int bracketed_name = 0; /* used to handle ${[0-9]*} variables */ 1195 1196 c = pgetc(); 1197 if (c != '(' && c != '{' && !is_name(c) && !is_special(c)) { 1198 USTPUTC('$', out); 1199 pungetc(); 1200 } else if (c == '(') { /* $(command) or $((arith)) */ 1201 if (pgetc() == '(') { 1202 PARSEARITH(); 1203 } else { 1204 pungetc(); 1205 PARSEBACKQNEW(); 1206 } 1207 } else { 1208 USTPUTC(CTLVAR, out); 1209 typeloc = out - stackblock(); 1210 USTPUTC(VSNORMAL, out); 1211 subtype = VSNORMAL; 1212 if (c == '{') { 1213 bracketed_name = 1; 1214 c = pgetc(); 1215 if (c == '#') { 1216 if ((c = pgetc()) == '}') 1217 c = '#'; 1218 else 1219 subtype = VSLENGTH; 1220 } 1221 else 1222 subtype = 0; 1223 } 1224 if (is_name(c)) { 1225 do { 1226 STPUTC(c, out); 1227 c = pgetc(); 1228 } while (is_in_name(c)); 1229 } else if (is_digit(c)) { 1230 if (bracketed_name) { 1231 do { 1232 STPUTC(c, out); 1233 c = pgetc(); 1234 } while (is_digit(c)); 1235 } else { 1236 STPUTC(c, out); 1237 c = pgetc(); 1238 } 1239 } else { 1240 if (! is_special(c)) 1241 badsub: synerror("Bad substitution"); 1242 USTPUTC(c, out); 1243 c = pgetc(); 1244 } 1245 STPUTC('=', out); 1246 flags = 0; 1247 if (subtype == 0) { 1248 switch (c) { 1249 case ':': 1250 flags = VSNUL; 1251 c = pgetc(); 1252 /*FALLTHROUGH*/ 1253 default: 1254 p = strchr(types, c); 1255 if (p == NULL) 1256 goto badsub; 1257 subtype = p - types + VSNORMAL; 1258 break; 1259 case '%': 1260 case '#': 1261 { 1262 int cc = c; 1263 subtype = c == '#' ? VSTRIMLEFT : 1264 VSTRIMRIGHT; 1265 c = pgetc(); 1266 if (c == cc) 1267 subtype++; 1268 else 1269 pungetc(); 1270 break; 1271 } 1272 } 1273 } else { 1274 pungetc(); 1275 } 1276 if (subtype != VSLENGTH && (dblquote || arinest)) 1277 flags |= VSQUOTE; 1278 *(stackblock() + typeloc) = subtype | flags; 1279 if (subtype != VSNORMAL) 1280 varnest++; 1281 } 1282 goto parsesub_return; 1283 } 1284 1285 1286 /* 1287 * Called to parse command substitutions. Newstyle is set if the command 1288 * is enclosed inside $(...); nlpp is a pointer to the head of the linked 1289 * list of commands (passed by reference), and savelen is the number of 1290 * characters on the top of the stack which must be preserved. 1291 */ 1292 1293 parsebackq: { 1294 struct nodelist **nlpp; 1295 int savepbq; 1296 union node *n; 1297 char *volatile str; 1298 struct jmploc jmploc; 1299 struct jmploc *volatile savehandler; 1300 int savelen; 1301 int saveprompt; 1302 #if __GNUC__ 1303 /* Avoid longjmp clobbering */ 1304 (void) &saveprompt; 1305 #endif 1306 1307 savepbq = parsebackquote; 1308 if (setjmp(jmploc.loc)) { 1309 if (str) 1310 ckfree(str); 1311 parsebackquote = 0; 1312 handler = savehandler; 1313 longjmp(handler->loc, 1); 1314 } 1315 INTOFF; 1316 str = NULL; 1317 savelen = out - stackblock(); 1318 if (savelen > 0) { 1319 str = ckmalloc(savelen); 1320 memcpy(str, stackblock(), savelen); 1321 } 1322 savehandler = handler; 1323 handler = &jmploc; 1324 INTON; 1325 if (oldstyle) { 1326 /* We must read until the closing backquote, giving special 1327 treatment to some slashes, and then push the string and 1328 reread it as input, interpreting it normally. */ 1329 char *out; 1330 int c; 1331 int savelen; 1332 char *str; 1333 1334 1335 STARTSTACKSTR(out); 1336 for (;;) { 1337 if (needprompt) { 1338 setprompt(2); 1339 needprompt = 0; 1340 } 1341 switch (c = pgetc()) { 1342 case '`': 1343 goto done; 1344 1345 case '\\': 1346 if ((c = pgetc()) == '\n') { 1347 plinno++; 1348 if (doprompt) 1349 setprompt(2); 1350 else 1351 setprompt(0); 1352 /* 1353 * If eating a newline, avoid putting 1354 * the newline into the new character 1355 * stream (via the STPUTC after the 1356 * switch). 1357 */ 1358 continue; 1359 } 1360 if (c != '\\' && c != '`' && c != '$' 1361 && (!dblquote || c != '"')) 1362 STPUTC('\\', out); 1363 break; 1364 1365 case '\n': 1366 plinno++; 1367 needprompt = doprompt; 1368 break; 1369 1370 case PEOF: 1371 startlinno = plinno; 1372 synerror("EOF in backquote substitution"); 1373 break; 1374 1375 default: 1376 break; 1377 } 1378 STPUTC(c, out); 1379 } 1380 done: 1381 STPUTC('\0', out); 1382 savelen = out - stackblock(); 1383 if (savelen > 0) { 1384 str = ckmalloc(savelen); 1385 memcpy(str, stackblock(), savelen); 1386 setinputstring(str, 1); 1387 } 1388 } 1389 nlpp = &bqlist; 1390 while (*nlpp) 1391 nlpp = &(*nlpp)->next; 1392 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 1393 (*nlpp)->next = NULL; 1394 parsebackquote = oldstyle; 1395 1396 if (oldstyle) { 1397 saveprompt = doprompt; 1398 doprompt = 0; 1399 } 1400 1401 n = list(0); 1402 1403 if (oldstyle) 1404 doprompt = saveprompt; 1405 else { 1406 if (readtoken() != TRP) 1407 synexpect(TRP); 1408 } 1409 1410 (*nlpp)->n = n; 1411 if (oldstyle) { 1412 /* 1413 * Start reading from old file again, ignoring any pushed back 1414 * tokens left from the backquote parsing 1415 */ 1416 popfile(); 1417 tokpushback = 0; 1418 } 1419 while (stackblocksize() <= savelen) 1420 growstackblock(); 1421 STARTSTACKSTR(out); 1422 if (str) { 1423 memcpy(out, str, savelen); 1424 STADJUST(savelen, out); 1425 INTOFF; 1426 ckfree(str); 1427 str = NULL; 1428 INTON; 1429 } 1430 parsebackquote = savepbq; 1431 handler = savehandler; 1432 if (arinest || dblquote) 1433 USTPUTC(CTLBACKQ | CTLQUOTE, out); 1434 else 1435 USTPUTC(CTLBACKQ, out); 1436 if (oldstyle) 1437 goto parsebackq_oldreturn; 1438 else 1439 goto parsebackq_newreturn; 1440 } 1441 1442 /* 1443 * Parse an arithmetic expansion (indicate start of one and set state) 1444 */ 1445 parsearith: { 1446 1447 if (++arinest == 1) { 1448 prevsyntax = syntax; 1449 syntax = ARISYNTAX; 1450 USTPUTC(CTLARI, out); 1451 if (dblquote) 1452 USTPUTC('"',out); 1453 else 1454 USTPUTC(' ',out); 1455 } else { 1456 /* 1457 * we collapse embedded arithmetic expansion to 1458 * parenthesis, which should be equivalent 1459 */ 1460 USTPUTC('(', out); 1461 } 1462 goto parsearith_return; 1463 } 1464 1465 } /* end of readtoken */ 1466 1467 1468 1469 #ifdef mkinit 1470 RESET { 1471 tokpushback = 0; 1472 checkkwd = 0; 1473 } 1474 #endif 1475 1476 /* 1477 * Returns true if the text contains nothing to expand (no dollar signs 1478 * or backquotes). 1479 */ 1480 1481 STATIC int 1482 noexpand(char *text) 1483 { 1484 char *p; 1485 char c; 1486 1487 p = text; 1488 while ((c = *p++) != '\0') { 1489 if ( c == CTLQUOTEMARK) 1490 continue; 1491 if (c == CTLESC) 1492 p++; 1493 else if (BASESYNTAX[(int)c] == CCTL) 1494 return 0; 1495 } 1496 return 1; 1497 } 1498 1499 1500 /* 1501 * Return true if the argument is a legal variable name (a letter or 1502 * underscore followed by zero or more letters, underscores, and digits). 1503 */ 1504 1505 int 1506 goodname(char *name) 1507 { 1508 char *p; 1509 1510 p = name; 1511 if (! is_name(*p)) 1512 return 0; 1513 while (*++p) { 1514 if (! is_in_name(*p)) 1515 return 0; 1516 } 1517 return 1; 1518 } 1519 1520 1521 /* 1522 * Called when an unexpected token is read during the parse. The argument 1523 * is the token that is expected, or -1 if more than one type of token can 1524 * occur at this point. 1525 */ 1526 1527 STATIC void 1528 synexpect(int token) 1529 { 1530 char msg[64]; 1531 1532 if (token >= 0) { 1533 fmtstr(msg, 64, "%s unexpected (expecting %s)", 1534 tokname[lasttoken], tokname[token]); 1535 } else { 1536 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]); 1537 } 1538 synerror(msg); 1539 } 1540 1541 1542 STATIC void 1543 synerror(char *msg) 1544 { 1545 if (commandname) 1546 outfmt(&errout, "%s: %d: ", commandname, startlinno); 1547 outfmt(&errout, "Syntax error: %s\n", msg); 1548 error((char *)NULL); 1549 } 1550 1551 STATIC void 1552 setprompt(int which) 1553 { 1554 whichprompt = which; 1555 1556 #ifndef NO_HISTORY 1557 if (!el) 1558 #endif 1559 out2str(getprompt(NULL)); 1560 } 1561 1562 /* 1563 * called by editline -- any expansions to the prompt 1564 * should be added here. 1565 */ 1566 char * 1567 getprompt(void *unused __unused) 1568 { 1569 switch (whichprompt) { 1570 case 0: 1571 return ""; 1572 case 1: 1573 return ps1val(); 1574 case 2: 1575 return ps2val(); 1576 default: 1577 return "<internal prompt error>"; 1578 } 1579 } 1580