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 * $Id$ 37 */ 38 39 #ifndef lint 40 static char const sccsid[] = "@(#)parser.c 8.7 (Berkeley) 5/16/95"; 41 #endif /* not lint */ 42 43 #include <stdlib.h> 44 45 #include "shell.h" 46 #include "parser.h" 47 #include "nodes.h" 48 #include "expand.h" /* defines rmescapes() */ 49 #include "redir.h" /* defines copyfd() */ 50 #include "syntax.h" 51 #include "options.h" 52 #include "input.h" 53 #include "output.h" 54 #include "var.h" 55 #include "error.h" 56 #include "memalloc.h" 57 #include "mystring.h" 58 #include "alias.h" 59 #include "show.h" 60 #ifndef NO_HISTORY 61 #include "myhistedit.h" 62 #endif 63 64 /* 65 * Shell command parser. 66 */ 67 68 #define EOFMARKLEN 79 69 70 /* values returned by readtoken */ 71 #include "token.h" 72 73 74 75 struct heredoc { 76 struct heredoc *next; /* next here document in list */ 77 union node *here; /* redirection node */ 78 char *eofmark; /* string indicating end of input */ 79 int striptabs; /* if set, strip leading tabs */ 80 }; 81 82 83 84 struct heredoc *heredoclist; /* list of here documents to read */ 85 int parsebackquote; /* nonzero if we are inside backquotes */ 86 int doprompt; /* if set, prompt the user */ 87 int needprompt; /* true if interactive and at start of line */ 88 int lasttoken; /* last token read */ 89 MKINIT int tokpushback; /* last token pushed back */ 90 char *wordtext; /* text of last word returned by readtoken */ 91 MKINIT int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */ 92 struct nodelist *backquotelist; 93 union node *redirnode; 94 struct heredoc *heredoc; 95 int quoteflag; /* set if (part of) last token was quoted */ 96 int startlinno; /* line # where last token started */ 97 98 /* XXX When 'noaliases' is set to one, no alias expansion takes place. */ 99 static int noaliases = 0; 100 101 #define GDB_HACK 1 /* avoid local declarations which gdb can't handle */ 102 #ifdef GDB_HACK 103 static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'}; 104 static const char types[] = "}-+?="; 105 #endif 106 107 108 STATIC union node *list __P((int)); 109 STATIC union node *andor __P((void)); 110 STATIC union node *pipeline __P((void)); 111 STATIC union node *command __P((void)); 112 STATIC union node *simplecmd __P((union node **, union node *)); 113 STATIC union node *makename __P((void)); 114 STATIC void parsefname __P((void)); 115 STATIC void parseheredoc __P((void)); 116 STATIC int peektoken __P((void)); 117 STATIC int readtoken __P((void)); 118 STATIC int xxreadtoken __P((void)); 119 STATIC int readtoken1 __P((int, char const *, char *, int)); 120 STATIC int noexpand __P((char *)); 121 STATIC void synexpect __P((int)); 122 STATIC void synerror __P((char *)); 123 STATIC void setprompt __P((int)); 124 125 126 /* 127 * Read and parse a command. Returns NEOF on end of file. (NULL is a 128 * valid parse tree indicating a blank line.) 129 */ 130 131 union node * 132 parsecmd(interact) 133 int interact; 134 { 135 int t; 136 137 doprompt = interact; 138 if (doprompt) 139 setprompt(1); 140 else 141 setprompt(0); 142 needprompt = 0; 143 t = readtoken(); 144 if (t == TEOF) 145 return NEOF; 146 if (t == TNL) 147 return NULL; 148 tokpushback++; 149 return list(1); 150 } 151 152 153 STATIC union node * 154 list(nlflag) 155 int nlflag; 156 { 157 union node *n1, *n2, *n3; 158 int tok; 159 160 checkkwd = 2; 161 if (nlflag == 0 && tokendlist[peektoken()]) 162 return NULL; 163 n1 = NULL; 164 for (;;) { 165 n2 = andor(); 166 tok = readtoken(); 167 if (tok == TBACKGND) { 168 if (n2->type == NCMD || n2->type == NPIPE) { 169 n2->ncmd.backgnd = 1; 170 } else if (n2->type == NREDIR) { 171 n2->type = NBACKGND; 172 } else { 173 n3 = (union node *)stalloc(sizeof (struct nredir)); 174 n3->type = NBACKGND; 175 n3->nredir.n = n2; 176 n3->nredir.redirect = NULL; 177 n2 = n3; 178 } 179 } 180 if (n1 == NULL) { 181 n1 = n2; 182 } 183 else { 184 n3 = (union node *)stalloc(sizeof (struct nbinary)); 185 n3->type = NSEMI; 186 n3->nbinary.ch1 = n1; 187 n3->nbinary.ch2 = n2; 188 n1 = n3; 189 } 190 switch (tok) { 191 case TBACKGND: 192 case TSEMI: 193 tok = readtoken(); 194 /* fall through */ 195 case TNL: 196 if (tok == TNL) { 197 parseheredoc(); 198 if (nlflag) 199 return n1; 200 } else { 201 tokpushback++; 202 } 203 checkkwd = 2; 204 if (tokendlist[peektoken()]) 205 return n1; 206 break; 207 case TEOF: 208 if (heredoclist) 209 parseheredoc(); 210 else 211 pungetc(); /* push back EOF on input */ 212 return n1; 213 default: 214 if (nlflag) 215 synexpect(-1); 216 tokpushback++; 217 return n1; 218 } 219 } 220 } 221 222 223 224 STATIC union node * 225 andor() { 226 union node *n1, *n2, *n3; 227 int t; 228 229 n1 = pipeline(); 230 for (;;) { 231 if ((t = readtoken()) == TAND) { 232 t = NAND; 233 } else if (t == TOR) { 234 t = NOR; 235 } else { 236 tokpushback++; 237 return n1; 238 } 239 n2 = pipeline(); 240 n3 = (union node *)stalloc(sizeof (struct nbinary)); 241 n3->type = t; 242 n3->nbinary.ch1 = n1; 243 n3->nbinary.ch2 = n2; 244 n1 = n3; 245 } 246 } 247 248 249 250 STATIC union node * 251 pipeline() { 252 union node *n1, *pipenode; 253 struct nodelist *lp, *prev; 254 255 TRACE(("pipeline: entered\n")); 256 n1 = command(); 257 if (readtoken() == TPIPE) { 258 pipenode = (union node *)stalloc(sizeof (struct npipe)); 259 pipenode->type = NPIPE; 260 pipenode->npipe.backgnd = 0; 261 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 262 pipenode->npipe.cmdlist = lp; 263 lp->n = n1; 264 do { 265 prev = lp; 266 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 267 lp->n = command(); 268 prev->next = lp; 269 } while (readtoken() == TPIPE); 270 lp->next = NULL; 271 n1 = pipenode; 272 } 273 tokpushback++; 274 return n1; 275 } 276 277 278 279 STATIC union node * 280 command() { 281 union node *n1, *n2; 282 union node *ap, **app; 283 union node *cp, **cpp; 284 union node *redir, **rpp; 285 int t, negate = 0; 286 287 checkkwd = 2; 288 redir = NULL; 289 n1 = NULL; 290 rpp = &redir; 291 292 /* Check for redirection which may precede command */ 293 while (readtoken() == TREDIR) { 294 *rpp = n2 = redirnode; 295 rpp = &n2->nfile.next; 296 parsefname(); 297 } 298 tokpushback++; 299 300 while (readtoken() == TNOT) { 301 TRACE(("command: TNOT recognized\n")); 302 negate = !negate; 303 } 304 tokpushback++; 305 306 switch (readtoken()) { 307 case TIF: 308 n1 = (union node *)stalloc(sizeof (struct nif)); 309 n1->type = NIF; 310 n1->nif.test = list(0); 311 if (readtoken() != TTHEN) 312 synexpect(TTHEN); 313 n1->nif.ifpart = list(0); 314 n2 = n1; 315 while (readtoken() == TELIF) { 316 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif)); 317 n2 = n2->nif.elsepart; 318 n2->type = NIF; 319 n2->nif.test = list(0); 320 if (readtoken() != TTHEN) 321 synexpect(TTHEN); 322 n2->nif.ifpart = list(0); 323 } 324 if (lasttoken == TELSE) 325 n2->nif.elsepart = list(0); 326 else { 327 n2->nif.elsepart = NULL; 328 tokpushback++; 329 } 330 if (readtoken() != TFI) 331 synexpect(TFI); 332 checkkwd = 1; 333 break; 334 case TWHILE: 335 case TUNTIL: { 336 int got; 337 n1 = (union node *)stalloc(sizeof (struct nbinary)); 338 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL; 339 n1->nbinary.ch1 = list(0); 340 if ((got=readtoken()) != TDO) { 341 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : "")); 342 synexpect(TDO); 343 } 344 n1->nbinary.ch2 = list(0); 345 if (readtoken() != TDONE) 346 synexpect(TDONE); 347 checkkwd = 1; 348 break; 349 } 350 case TFOR: 351 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext)) 352 synerror("Bad for loop variable"); 353 n1 = (union node *)stalloc(sizeof (struct nfor)); 354 n1->type = NFOR; 355 n1->nfor.var = wordtext; 356 if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) { 357 app = ≈ 358 while (readtoken() == TWORD) { 359 n2 = (union node *)stalloc(sizeof (struct narg)); 360 n2->type = NARG; 361 n2->narg.text = wordtext; 362 n2->narg.backquote = backquotelist; 363 *app = n2; 364 app = &n2->narg.next; 365 } 366 *app = NULL; 367 n1->nfor.args = ap; 368 if (lasttoken != TNL && lasttoken != TSEMI) 369 synexpect(-1); 370 } else { 371 #ifndef GDB_HACK 372 static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, 373 '@', '=', '\0'}; 374 #endif 375 n2 = (union node *)stalloc(sizeof (struct narg)); 376 n2->type = NARG; 377 n2->narg.text = (char *)argvars; 378 n2->narg.backquote = NULL; 379 n2->narg.next = NULL; 380 n1->nfor.args = n2; 381 /* 382 * Newline or semicolon here is optional (but note 383 * that the original Bourne shell only allowed NL). 384 */ 385 if (lasttoken != TNL && lasttoken != TSEMI) 386 tokpushback++; 387 } 388 checkkwd = 2; 389 if ((t = readtoken()) == TDO) 390 t = TDONE; 391 else if (t == TBEGIN) 392 t = TEND; 393 else 394 synexpect(-1); 395 n1->nfor.body = list(0); 396 if (readtoken() != t) 397 synexpect(t); 398 checkkwd = 1; 399 break; 400 case TCASE: 401 n1 = (union node *)stalloc(sizeof (struct ncase)); 402 n1->type = NCASE; 403 if (readtoken() != TWORD) 404 synexpect(TWORD); 405 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg)); 406 n2->type = NARG; 407 n2->narg.text = wordtext; 408 n2->narg.backquote = backquotelist; 409 n2->narg.next = NULL; 410 while (readtoken() == TNL); 411 if (lasttoken != TWORD || ! equal(wordtext, "in")) 412 synerror("expecting \"in\""); 413 cpp = &n1->ncase.cases; 414 noaliases = 1; /* turn off alias expansion */ 415 checkkwd = 2, readtoken(); 416 do { 417 *cpp = cp = (union node *)stalloc(sizeof (struct nclist)); 418 cp->type = NCLIST; 419 app = &cp->nclist.pattern; 420 for (;;) { 421 *app = ap = (union node *)stalloc(sizeof (struct narg)); 422 ap->type = NARG; 423 ap->narg.text = wordtext; 424 ap->narg.backquote = backquotelist; 425 if (checkkwd = 2, readtoken() != TPIPE) 426 break; 427 app = &ap->narg.next; 428 readtoken(); 429 } 430 ap->narg.next = NULL; 431 if (lasttoken != TRP) 432 noaliases = 0, synexpect(TRP); 433 cp->nclist.body = list(0); 434 435 checkkwd = 2; 436 if ((t = readtoken()) != TESAC) { 437 if (t != TENDCASE) 438 noaliases = 0, synexpect(TENDCASE); 439 else 440 checkkwd = 2, readtoken(); 441 } 442 cpp = &cp->nclist.next; 443 } while(lasttoken != TESAC); 444 noaliases = 0; /* reset alias expansion */ 445 *cpp = NULL; 446 checkkwd = 1; 447 break; 448 case TLP: 449 n1 = (union node *)stalloc(sizeof (struct nredir)); 450 n1->type = NSUBSHELL; 451 n1->nredir.n = list(0); 452 n1->nredir.redirect = NULL; 453 if (readtoken() != TRP) 454 synexpect(TRP); 455 checkkwd = 1; 456 break; 457 case TBEGIN: 458 n1 = list(0); 459 if (readtoken() != TEND) 460 synexpect(TEND); 461 checkkwd = 1; 462 break; 463 /* Handle an empty command like other simple commands. */ 464 case TSEMI: 465 /* 466 * An empty command before a ; doesn't make much sense, and 467 * should certainly be disallowed in the case of `if ;'. 468 */ 469 if (!redir) 470 synexpect(-1); 471 case TAND: 472 case TOR: 473 case TNL: 474 case TEOF: 475 case TWORD: 476 case TRP: 477 tokpushback++; 478 n1 = simplecmd(rpp, redir); 479 goto checkneg; 480 default: 481 synexpect(-1); 482 } 483 484 /* Now check for redirection which may follow command */ 485 while (readtoken() == TREDIR) { 486 *rpp = n2 = redirnode; 487 rpp = &n2->nfile.next; 488 parsefname(); 489 } 490 tokpushback++; 491 *rpp = NULL; 492 if (redir) { 493 if (n1->type != NSUBSHELL) { 494 n2 = (union node *)stalloc(sizeof (struct nredir)); 495 n2->type = NREDIR; 496 n2->nredir.n = n1; 497 n1 = n2; 498 } 499 n1->nredir.redirect = redir; 500 } 501 502 checkneg: 503 if (negate) { 504 n2 = (union node *)stalloc(sizeof (struct nnot)); 505 n2->type = NNOT; 506 n2->nnot.com = n1; 507 return n2; 508 } 509 else 510 return n1; 511 } 512 513 514 STATIC union node * 515 simplecmd(rpp, redir) 516 union node **rpp, *redir; 517 { 518 union node *args, **app; 519 union node **orig_rpp = rpp; 520 union node *n = NULL, *n2; 521 int negate = 0; 522 523 /* If we don't have any redirections already, then we must reset */ 524 /* rpp to be the address of the local redir variable. */ 525 if (redir == 0) 526 rpp = &redir; 527 528 args = NULL; 529 app = &args; 530 /* 531 * We save the incoming value, because we need this for shell 532 * functions. There can not be a redirect or an argument between 533 * the function name and the open parenthesis. 534 */ 535 orig_rpp = rpp; 536 537 while (readtoken() == TNOT) { 538 TRACE(("command: TNOT recognized\n")); 539 negate = !negate; 540 } 541 tokpushback++; 542 543 for (;;) { 544 if (readtoken() == TWORD) { 545 n = (union node *)stalloc(sizeof (struct narg)); 546 n->type = NARG; 547 n->narg.text = wordtext; 548 n->narg.backquote = backquotelist; 549 *app = n; 550 app = &n->narg.next; 551 } else if (lasttoken == TREDIR) { 552 *rpp = n = redirnode; 553 rpp = &n->nfile.next; 554 parsefname(); /* read name of redirection file */ 555 } else if (lasttoken == TLP && app == &args->narg.next 556 && rpp == orig_rpp) { 557 /* We have a function */ 558 if (readtoken() != TRP) 559 synexpect(TRP); 560 #ifdef notdef 561 if (! goodname(n->narg.text)) 562 synerror("Bad function name"); 563 #endif 564 n->type = NDEFUN; 565 n->narg.next = command(); 566 goto checkneg; 567 } else { 568 tokpushback++; 569 break; 570 } 571 } 572 *app = NULL; 573 *rpp = NULL; 574 n = (union node *)stalloc(sizeof (struct ncmd)); 575 n->type = NCMD; 576 n->ncmd.backgnd = 0; 577 n->ncmd.args = args; 578 n->ncmd.redirect = redir; 579 580 checkneg: 581 if (negate) { 582 n2 = (union node *)stalloc(sizeof (struct nnot)); 583 n2->type = NNOT; 584 n2->nnot.com = n; 585 return n2; 586 } 587 else 588 return n; 589 } 590 591 STATIC union node * 592 makename() { 593 union node *n; 594 595 n = (union node *)stalloc(sizeof (struct narg)); 596 n->type = NARG; 597 n->narg.next = NULL; 598 n->narg.text = wordtext; 599 n->narg.backquote = backquotelist; 600 return n; 601 } 602 603 void fixredir(n, text, err) 604 union node *n; 605 const char *text; 606 int err; 607 { 608 TRACE(("Fix redir %s %d\n", text, err)); 609 if (!err) 610 n->ndup.vname = NULL; 611 612 if (is_digit(text[0]) && text[1] == '\0') 613 n->ndup.dupfd = digit_val(text[0]); 614 else if (text[0] == '-' && text[1] == '\0') 615 n->ndup.dupfd = -1; 616 else { 617 618 if (err) 619 synerror("Bad fd number"); 620 else 621 n->ndup.vname = makename(); 622 } 623 } 624 625 626 STATIC void 627 parsefname() { 628 union node *n = redirnode; 629 630 if (readtoken() != TWORD) 631 synexpect(-1); 632 if (n->type == NHERE) { 633 struct heredoc *here = heredoc; 634 struct heredoc *p; 635 int i; 636 637 if (quoteflag == 0) 638 n->type = NXHERE; 639 TRACE(("Here document %d\n", n->type)); 640 if (here->striptabs) { 641 while (*wordtext == '\t') 642 wordtext++; 643 } 644 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN) 645 synerror("Illegal eof marker for << redirection"); 646 rmescapes(wordtext); 647 here->eofmark = wordtext; 648 here->next = NULL; 649 if (heredoclist == NULL) 650 heredoclist = here; 651 else { 652 for (p = heredoclist ; p->next ; p = p->next); 653 p->next = here; 654 } 655 } else if (n->type == NTOFD || n->type == NFROMFD) { 656 fixredir(n, wordtext, 0); 657 } else { 658 n->nfile.fname = makename(); 659 } 660 } 661 662 663 /* 664 * Input any here documents. 665 */ 666 667 STATIC void 668 parseheredoc() { 669 struct heredoc *here; 670 union node *n; 671 672 while (heredoclist) { 673 here = heredoclist; 674 heredoclist = here->next; 675 if (needprompt) { 676 setprompt(2); 677 needprompt = 0; 678 } 679 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX, 680 here->eofmark, here->striptabs); 681 n = (union node *)stalloc(sizeof (struct narg)); 682 n->narg.type = NARG; 683 n->narg.next = NULL; 684 n->narg.text = wordtext; 685 n->narg.backquote = backquotelist; 686 here->here->nhere.doc = n; 687 } 688 } 689 690 STATIC int 691 peektoken() { 692 int t; 693 694 t = readtoken(); 695 tokpushback++; 696 return (t); 697 } 698 699 STATIC int xxreadtoken(); 700 701 STATIC int 702 readtoken() { 703 int t; 704 int savecheckkwd = checkkwd; 705 struct alias *ap; 706 #ifdef DEBUG 707 int alreadyseen = tokpushback; 708 #endif 709 710 top: 711 t = xxreadtoken(); 712 713 if (checkkwd) { 714 /* 715 * eat newlines 716 */ 717 if (checkkwd == 2) { 718 checkkwd = 0; 719 while (t == TNL) { 720 parseheredoc(); 721 t = xxreadtoken(); 722 } 723 } else 724 checkkwd = 0; 725 /* 726 * check for keywords and aliases 727 */ 728 if (t == TWORD && !quoteflag) 729 { 730 register char * const *pp; 731 732 for (pp = (char **)parsekwd; *pp; pp++) { 733 if (**pp == *wordtext && equal(*pp, wordtext)) 734 { 735 lasttoken = t = pp - parsekwd + KWDOFFSET; 736 TRACE(("keyword %s recognized\n", tokname[t])); 737 goto out; 738 } 739 } 740 if (noaliases == 0 && 741 (ap = lookupalias(wordtext, 1)) != NULL) { 742 pushstring(ap->val, strlen(ap->val), ap); 743 checkkwd = savecheckkwd; 744 goto top; 745 } 746 } 747 out: 748 checkkwd = (t == TNOT) ? savecheckkwd : 0; 749 } 750 #ifdef DEBUG 751 if (!alreadyseen) 752 TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 753 else 754 TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 755 #endif 756 return (t); 757 } 758 759 760 /* 761 * Read the next input token. 762 * If the token is a word, we set backquotelist to the list of cmds in 763 * backquotes. We set quoteflag to true if any part of the word was 764 * quoted. 765 * If the token is TREDIR, then we set redirnode to a structure containing 766 * the redirection. 767 * In all cases, the variable startlinno is set to the number of the line 768 * on which the token starts. 769 * 770 * [Change comment: here documents and internal procedures] 771 * [Readtoken shouldn't have any arguments. Perhaps we should make the 772 * word parsing code into a separate routine. In this case, readtoken 773 * doesn't need to have any internal procedures, but parseword does. 774 * We could also make parseoperator in essence the main routine, and 775 * have parseword (readtoken1?) handle both words and redirection.] 776 */ 777 778 #define RETURN(token) return lasttoken = token 779 780 STATIC int 781 xxreadtoken() { 782 register c; 783 784 if (tokpushback) { 785 tokpushback = 0; 786 return lasttoken; 787 } 788 if (needprompt) { 789 setprompt(2); 790 needprompt = 0; 791 } 792 startlinno = plinno; 793 for (;;) { /* until token or start of word found */ 794 c = pgetc_macro(); 795 if (c == ' ' || c == '\t') 796 continue; /* quick check for white space first */ 797 switch (c) { 798 case ' ': case '\t': 799 continue; 800 case '#': 801 while ((c = pgetc()) != '\n' && c != PEOF); 802 pungetc(); 803 continue; 804 case '\\': 805 if (pgetc() == '\n') { 806 startlinno = ++plinno; 807 if (doprompt) 808 setprompt(2); 809 else 810 setprompt(0); 811 continue; 812 } 813 pungetc(); 814 goto breakloop; 815 case '\n': 816 plinno++; 817 needprompt = doprompt; 818 RETURN(TNL); 819 case PEOF: 820 RETURN(TEOF); 821 case '&': 822 if (pgetc() == '&') 823 RETURN(TAND); 824 pungetc(); 825 RETURN(TBACKGND); 826 case '|': 827 if (pgetc() == '|') 828 RETURN(TOR); 829 pungetc(); 830 RETURN(TPIPE); 831 case ';': 832 if (pgetc() == ';') 833 RETURN(TENDCASE); 834 pungetc(); 835 RETURN(TSEMI); 836 case '(': 837 RETURN(TLP); 838 case ')': 839 RETURN(TRP); 840 default: 841 goto breakloop; 842 } 843 } 844 breakloop: 845 return readtoken1(c, BASESYNTAX, (char *)NULL, 0); 846 #undef RETURN 847 } 848 849 850 851 /* 852 * If eofmark is NULL, read a word or a redirection symbol. If eofmark 853 * is not NULL, read a here document. In the latter case, eofmark is the 854 * word which marks the end of the document and striptabs is true if 855 * leading tabs should be stripped from the document. The argument firstc 856 * is the first character of the input token or document. 857 * 858 * Because C does not have internal subroutines, I have simulated them 859 * using goto's to implement the subroutine linkage. The following macros 860 * will run code that appears at the end of readtoken1. 861 */ 862 863 #define CHECKEND() {goto checkend; checkend_return:;} 864 #define PARSEREDIR() {goto parseredir; parseredir_return:;} 865 #define PARSESUB() {goto parsesub; parsesub_return:;} 866 #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;} 867 #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;} 868 #define PARSEARITH() {goto parsearith; parsearith_return:;} 869 870 STATIC int 871 readtoken1(firstc, syntax, eofmark, striptabs) 872 int firstc; 873 char const *syntax; 874 char *eofmark; 875 int striptabs; 876 { 877 int c = firstc; 878 char *out; 879 int len; 880 char line[EOFMARKLEN + 1]; 881 struct nodelist *bqlist; 882 int quotef; 883 int dblquote; 884 int varnest; /* levels of variables expansion */ 885 int arinest; /* levels of arithmetic expansion */ 886 int parenlevel; /* levels of parens in arithmetic */ 887 int oldstyle; 888 char const *prevsyntax; /* syntax before arithmetic */ 889 #if __GNUC__ 890 /* Avoid longjmp clobbering */ 891 (void) &out; 892 (void) "ef; 893 (void) &dblquote; 894 (void) &varnest; 895 (void) &arinest; 896 (void) &parenlevel; 897 (void) &oldstyle; 898 (void) &prevsyntax; 899 (void) &syntax; 900 #endif 901 902 startlinno = plinno; 903 dblquote = 0; 904 if (syntax == DQSYNTAX) 905 dblquote = 1; 906 quotef = 0; 907 bqlist = NULL; 908 varnest = 0; 909 arinest = 0; 910 parenlevel = 0; 911 912 STARTSTACKSTR(out); 913 loop: { /* for each line, until end of word */ 914 #if ATTY 915 if (c == '\034' && doprompt 916 && attyset() && ! equal(termval(), "emacs")) { 917 attyline(); 918 if (syntax == BASESYNTAX) 919 return readtoken(); 920 c = pgetc(); 921 goto loop; 922 } 923 #endif 924 CHECKEND(); /* set c to PEOF if at end of here document */ 925 for (;;) { /* until end of line or end of word */ 926 CHECKSTRSPACE(3, out); /* permit 3 calls to USTPUTC */ 927 switch(syntax[c]) { 928 case CNL: /* '\n' */ 929 if (syntax == BASESYNTAX) 930 goto endword; /* exit outer loop */ 931 USTPUTC(c, out); 932 plinno++; 933 if (doprompt) 934 setprompt(2); 935 else 936 setprompt(0); 937 c = pgetc(); 938 goto loop; /* continue outer loop */ 939 case CWORD: 940 USTPUTC(c, out); 941 break; 942 case CCTL: 943 if (eofmark == NULL || dblquote) 944 USTPUTC(CTLESC, out); 945 USTPUTC(c, out); 946 break; 947 case CBACK: /* backslash */ 948 c = pgetc(); 949 if (c == PEOF) { 950 USTPUTC('\\', out); 951 pungetc(); 952 } else if (c == '\n') { 953 if (doprompt) 954 setprompt(2); 955 else 956 setprompt(0); 957 } else { 958 if (dblquote && c != '\\' && c != '`' && c != '$' 959 && (c != '"' || eofmark != NULL)) 960 USTPUTC('\\', out); 961 if (SQSYNTAX[c] == CCTL) 962 USTPUTC(CTLESC, out); 963 USTPUTC(c, out); 964 quotef++; 965 } 966 break; 967 case CSQUOTE: 968 syntax = SQSYNTAX; 969 break; 970 case CDQUOTE: 971 syntax = DQSYNTAX; 972 dblquote = 1; 973 break; 974 case CENDQUOTE: 975 if (eofmark) { 976 USTPUTC(c, out); 977 } else { 978 if (arinest) 979 syntax = ARISYNTAX; 980 else 981 syntax = BASESYNTAX; 982 quotef++; 983 dblquote = 0; 984 } 985 break; 986 case CVAR: /* '$' */ 987 PARSESUB(); /* parse substitution */ 988 break; 989 case CENDVAR: /* '}' */ 990 if (varnest > 0) { 991 varnest--; 992 USTPUTC(CTLENDVAR, out); 993 } else { 994 USTPUTC(c, out); 995 } 996 break; 997 case CLP: /* '(' in arithmetic */ 998 parenlevel++; 999 USTPUTC(c, out); 1000 break; 1001 case CRP: /* ')' in arithmetic */ 1002 if (parenlevel > 0) { 1003 USTPUTC(c, out); 1004 --parenlevel; 1005 } else { 1006 if (pgetc() == ')') { 1007 if (--arinest == 0) { 1008 USTPUTC(CTLENDARI, out); 1009 syntax = prevsyntax; 1010 } else 1011 USTPUTC(')', out); 1012 } else { 1013 /* 1014 * unbalanced parens 1015 * (don't 2nd guess - no error) 1016 */ 1017 pungetc(); 1018 USTPUTC(')', out); 1019 } 1020 } 1021 break; 1022 case CBQUOTE: /* '`' */ 1023 PARSEBACKQOLD(); 1024 break; 1025 case CEOF: 1026 goto endword; /* exit outer loop */ 1027 default: 1028 if (varnest == 0) 1029 goto endword; /* exit outer loop */ 1030 USTPUTC(c, out); 1031 } 1032 c = pgetc_macro(); 1033 } 1034 } 1035 endword: 1036 if (syntax == ARISYNTAX) 1037 synerror("Missing '))'"); 1038 if (syntax != BASESYNTAX && ! parsebackquote && eofmark == NULL) 1039 synerror("Unterminated quoted string"); 1040 if (varnest != 0) { 1041 startlinno = plinno; 1042 synerror("Missing '}'"); 1043 } 1044 USTPUTC('\0', out); 1045 len = out - stackblock(); 1046 out = stackblock(); 1047 if (eofmark == NULL) { 1048 if ((c == '>' || c == '<') 1049 && quotef == 0 1050 && len <= 2 1051 && (*out == '\0' || is_digit(*out))) { 1052 PARSEREDIR(); 1053 return lasttoken = TREDIR; 1054 } else { 1055 pungetc(); 1056 } 1057 } 1058 quoteflag = quotef; 1059 backquotelist = bqlist; 1060 grabstackblock(len); 1061 wordtext = out; 1062 return lasttoken = TWORD; 1063 /* end of readtoken routine */ 1064 1065 1066 1067 /* 1068 * Check to see whether we are at the end of the here document. When this 1069 * is called, c is set to the first character of the next input line. If 1070 * we are at the end of the here document, this routine sets the c to PEOF. 1071 */ 1072 1073 checkend: { 1074 if (eofmark) { 1075 if (striptabs) { 1076 while (c == '\t') 1077 c = pgetc(); 1078 } 1079 if (c == *eofmark) { 1080 if (pfgets(line, sizeof line) != NULL) { 1081 register char *p, *q; 1082 1083 p = line; 1084 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++); 1085 if (*p == '\n' && *q == '\0') { 1086 c = PEOF; 1087 plinno++; 1088 needprompt = doprompt; 1089 } else { 1090 pushstring(line, strlen(line), NULL); 1091 } 1092 } 1093 } 1094 } 1095 goto checkend_return; 1096 } 1097 1098 1099 /* 1100 * Parse a redirection operator. The variable "out" points to a string 1101 * specifying the fd to be redirected. The variable "c" contains the 1102 * first character of the redirection operator. 1103 */ 1104 1105 parseredir: { 1106 char fd = *out; 1107 union node *np; 1108 1109 np = (union node *)stalloc(sizeof (struct nfile)); 1110 if (c == '>') { 1111 np->nfile.fd = 1; 1112 c = pgetc(); 1113 if (c == '>') 1114 np->type = NAPPEND; 1115 else if (c == '&') 1116 np->type = NTOFD; 1117 else { 1118 np->type = NTO; 1119 pungetc(); 1120 } 1121 } else { /* c == '<' */ 1122 np->nfile.fd = 0; 1123 c = pgetc(); 1124 if (c == '<') { 1125 if (sizeof (struct nfile) != sizeof (struct nhere)) { 1126 np = (union node *)stalloc(sizeof (struct nhere)); 1127 np->nfile.fd = 0; 1128 } 1129 np->type = NHERE; 1130 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc)); 1131 heredoc->here = np; 1132 if ((c = pgetc()) == '-') { 1133 heredoc->striptabs = 1; 1134 } else { 1135 heredoc->striptabs = 0; 1136 pungetc(); 1137 } 1138 } else if (c == '&') 1139 np->type = NFROMFD; 1140 else { 1141 np->type = NFROM; 1142 pungetc(); 1143 } 1144 } 1145 if (fd != '\0') 1146 np->nfile.fd = digit_val(fd); 1147 redirnode = np; 1148 goto parseredir_return; 1149 } 1150 1151 1152 /* 1153 * Parse a substitution. At this point, we have read the dollar sign 1154 * and nothing else. 1155 */ 1156 1157 parsesub: { 1158 int subtype; 1159 int typeloc; 1160 int flags; 1161 char *p; 1162 #ifndef GDB_HACK 1163 static const char types[] = "}-+?="; 1164 #endif 1165 int bracketed_name = 0; /* used to handle ${[0-9]*} variables */ 1166 1167 c = pgetc(); 1168 if (c != '(' && c != '{' && !is_name(c) && !is_special(c)) { 1169 USTPUTC('$', out); 1170 pungetc(); 1171 } else if (c == '(') { /* $(command) or $((arith)) */ 1172 if (pgetc() == '(') { 1173 PARSEARITH(); 1174 } else { 1175 pungetc(); 1176 PARSEBACKQNEW(); 1177 } 1178 } else { 1179 USTPUTC(CTLVAR, out); 1180 typeloc = out - stackblock(); 1181 USTPUTC(VSNORMAL, out); 1182 subtype = VSNORMAL; 1183 if (c == '{') { 1184 bracketed_name = 1; 1185 c = pgetc(); 1186 if (c == '#') { 1187 if ((c = pgetc()) == '}') 1188 c = '#'; 1189 else 1190 subtype = VSLENGTH; 1191 } 1192 else 1193 subtype = 0; 1194 } 1195 if (is_name(c)) { 1196 do { 1197 STPUTC(c, out); 1198 c = pgetc(); 1199 } while (is_in_name(c)); 1200 } else if (is_digit(c)) { 1201 if (bracketed_name) { 1202 do { 1203 STPUTC(c, out); 1204 c = pgetc(); 1205 } while (is_digit(c)); 1206 } else { 1207 STPUTC(c, out); 1208 c = pgetc(); 1209 } 1210 } else { 1211 if (! is_special(c)) 1212 badsub: synerror("Bad substitution"); 1213 USTPUTC(c, out); 1214 c = pgetc(); 1215 } 1216 STPUTC('=', out); 1217 flags = 0; 1218 if (subtype == 0) { 1219 switch (c) { 1220 case ':': 1221 flags = VSNUL; 1222 c = pgetc(); 1223 /*FALLTHROUGH*/ 1224 default: 1225 p = strchr(types, c); 1226 if (p == NULL) 1227 goto badsub; 1228 subtype = p - types + VSNORMAL; 1229 break; 1230 case '%': 1231 case '#': 1232 { 1233 int cc = c; 1234 subtype = c == '#' ? VSTRIMLEFT : 1235 VSTRIMRIGHT; 1236 c = pgetc(); 1237 if (c == cc) 1238 subtype++; 1239 else 1240 pungetc(); 1241 break; 1242 } 1243 } 1244 } else { 1245 pungetc(); 1246 } 1247 if (dblquote || arinest) 1248 flags |= VSQUOTE; 1249 *(stackblock() + typeloc) = subtype | flags; 1250 if (subtype != VSNORMAL) 1251 varnest++; 1252 } 1253 goto parsesub_return; 1254 } 1255 1256 1257 /* 1258 * Called to parse command substitutions. Newstyle is set if the command 1259 * is enclosed inside $(...); nlpp is a pointer to the head of the linked 1260 * list of commands (passed by reference), and savelen is the number of 1261 * characters on the top of the stack which must be preserved. 1262 */ 1263 1264 parsebackq: { 1265 struct nodelist **nlpp; 1266 int savepbq; 1267 union node *n; 1268 char *volatile str; 1269 struct jmploc jmploc; 1270 struct jmploc *volatile savehandler; 1271 int savelen; 1272 int saveprompt; 1273 #if __GNUC__ 1274 /* Avoid longjmp clobbering */ 1275 (void) &saveprompt; 1276 #endif 1277 1278 savepbq = parsebackquote; 1279 if (setjmp(jmploc.loc)) { 1280 if (str) 1281 ckfree(str); 1282 parsebackquote = 0; 1283 handler = savehandler; 1284 longjmp(handler->loc, 1); 1285 } 1286 INTOFF; 1287 str = NULL; 1288 savelen = out - stackblock(); 1289 if (savelen > 0) { 1290 str = ckmalloc(savelen); 1291 memcpy(str, stackblock(), savelen); 1292 } 1293 savehandler = handler; 1294 handler = &jmploc; 1295 INTON; 1296 if (oldstyle) { 1297 /* We must read until the closing backquote, giving special 1298 treatment to some slashes, and then push the string and 1299 reread it as input, interpreting it normally. */ 1300 register char *out; 1301 register c; 1302 int savelen; 1303 char *str; 1304 1305 1306 STARTSTACKSTR(out); 1307 for (;;) { 1308 if (needprompt) { 1309 setprompt(2); 1310 needprompt = 0; 1311 } 1312 switch (c = pgetc()) { 1313 case '`': 1314 goto done; 1315 1316 case '\\': 1317 if ((c = pgetc()) == '\n') { 1318 plinno++; 1319 if (doprompt) 1320 setprompt(2); 1321 else 1322 setprompt(0); 1323 /* 1324 * If eating a newline, avoid putting 1325 * the newline into the new character 1326 * stream (via the STPUTC after the 1327 * switch). 1328 */ 1329 continue; 1330 } 1331 if (c != '\\' && c != '`' && c != '$' 1332 && (!dblquote || c != '"')) 1333 STPUTC('\\', out); 1334 break; 1335 1336 case '\n': 1337 plinno++; 1338 needprompt = doprompt; 1339 break; 1340 1341 case PEOF: 1342 startlinno = plinno; 1343 synerror("EOF in backquote substitution"); 1344 break; 1345 1346 default: 1347 break; 1348 } 1349 STPUTC(c, out); 1350 } 1351 done: 1352 STPUTC('\0', out); 1353 savelen = out - stackblock(); 1354 if (savelen > 0) { 1355 str = ckmalloc(savelen); 1356 memcpy(str, stackblock(), savelen); 1357 setinputstring(str, 1); 1358 } 1359 } 1360 nlpp = &bqlist; 1361 while (*nlpp) 1362 nlpp = &(*nlpp)->next; 1363 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 1364 (*nlpp)->next = NULL; 1365 parsebackquote = oldstyle; 1366 1367 if (oldstyle) { 1368 saveprompt = doprompt; 1369 doprompt = 0; 1370 } 1371 1372 n = list(0); 1373 1374 if (oldstyle) 1375 doprompt = saveprompt; 1376 else { 1377 if (readtoken() != TRP) 1378 synexpect(TRP); 1379 } 1380 1381 (*nlpp)->n = n; 1382 if (oldstyle) { 1383 /* 1384 * Start reading from old file again, ignoring any pushed back 1385 * tokens left from the backquote parsing 1386 */ 1387 popfile(); 1388 tokpushback = 0; 1389 } 1390 while (stackblocksize() <= savelen) 1391 growstackblock(); 1392 STARTSTACKSTR(out); 1393 if (str) { 1394 memcpy(out, str, savelen); 1395 STADJUST(savelen, out); 1396 INTOFF; 1397 ckfree(str); 1398 str = NULL; 1399 INTON; 1400 } 1401 parsebackquote = savepbq; 1402 handler = savehandler; 1403 if (arinest || dblquote) 1404 USTPUTC(CTLBACKQ | CTLQUOTE, out); 1405 else 1406 USTPUTC(CTLBACKQ, out); 1407 if (oldstyle) 1408 goto parsebackq_oldreturn; 1409 else 1410 goto parsebackq_newreturn; 1411 } 1412 1413 /* 1414 * Parse an arithmetic expansion (indicate start of one and set state) 1415 */ 1416 parsearith: { 1417 1418 if (++arinest == 1) { 1419 prevsyntax = syntax; 1420 syntax = ARISYNTAX; 1421 USTPUTC(CTLARI, out); 1422 } else { 1423 /* 1424 * we collapse embedded arithmetic expansion to 1425 * parenthesis, which should be equivalent 1426 */ 1427 USTPUTC('(', out); 1428 } 1429 goto parsearith_return; 1430 } 1431 1432 } /* end of readtoken */ 1433 1434 1435 1436 #ifdef mkinit 1437 RESET { 1438 tokpushback = 0; 1439 checkkwd = 0; 1440 } 1441 #endif 1442 1443 /* 1444 * Returns true if the text contains nothing to expand (no dollar signs 1445 * or backquotes). 1446 */ 1447 1448 STATIC int 1449 noexpand(text) 1450 char *text; 1451 { 1452 register char *p; 1453 register char c; 1454 1455 p = text; 1456 while ((c = *p++) != '\0') { 1457 if (c == CTLESC) 1458 p++; 1459 else if (BASESYNTAX[c] == CCTL) 1460 return 0; 1461 } 1462 return 1; 1463 } 1464 1465 1466 /* 1467 * Return true if the argument is a legal variable name (a letter or 1468 * underscore followed by zero or more letters, underscores, and digits). 1469 */ 1470 1471 int 1472 goodname(name) 1473 char *name; 1474 { 1475 register char *p; 1476 1477 p = name; 1478 if (! is_name(*p)) 1479 return 0; 1480 while (*++p) { 1481 if (! is_in_name(*p)) 1482 return 0; 1483 } 1484 return 1; 1485 } 1486 1487 1488 /* 1489 * Called when an unexpected token is read during the parse. The argument 1490 * is the token that is expected, or -1 if more than one type of token can 1491 * occur at this point. 1492 */ 1493 1494 STATIC void 1495 synexpect(token) 1496 int token; 1497 { 1498 char msg[64]; 1499 1500 if (token >= 0) { 1501 fmtstr(msg, 64, "%s unexpected (expecting %s)", 1502 tokname[lasttoken], tokname[token]); 1503 } else { 1504 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]); 1505 } 1506 synerror(msg); 1507 } 1508 1509 1510 STATIC void 1511 synerror(msg) 1512 char *msg; 1513 { 1514 if (commandname) 1515 outfmt(&errout, "%s: %d: ", commandname, startlinno); 1516 outfmt(&errout, "Syntax error: %s\n", msg); 1517 error((char *)NULL); 1518 } 1519 1520 STATIC void 1521 setprompt(which) 1522 int which; 1523 { 1524 whichprompt = which; 1525 1526 #ifndef NO_HISTORY 1527 if (!el) 1528 #endif 1529 out2str(getprompt(NULL)); 1530 } 1531 1532 /* 1533 * called by editline -- any expansions to the prompt 1534 * should be added here. 1535 */ 1536 char * 1537 getprompt(unused) 1538 void *unused; 1539 { 1540 switch (whichprompt) { 1541 case 0: 1542 return ""; 1543 case 1: 1544 return ps1val(); 1545 case 2: 1546 return ps2val(); 1547 default: 1548 return "<internal prompt error>"; 1549 } 1550 } 1551