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 * @(#)parser.c 8.7 (Berkeley) 5/16/95 37 * $FreeBSD: src/bin/sh/parser.c,v 1.29.2.9 2002/10/18 11:24:04 tjr Exp $ 38 * $DragonFly: src/bin/sh/parser.c,v 1.8 2006/09/28 22:29:44 pavalos Exp $ 39 */ 40 41 #include <stdlib.h> 42 43 #include "shell.h" 44 #include "parser.h" 45 #include "nodes.h" 46 #include "expand.h" /* defines rmescapes() */ 47 #include "redir.h" /* defines copyfd() */ 48 #include "syntax.h" 49 #include "options.h" 50 #include "input.h" 51 #include "output.h" 52 #include "var.h" 53 #include "error.h" 54 #include "memalloc.h" 55 #include "mystring.h" 56 #include "alias.h" 57 #include "show.h" 58 #include "eval.h" 59 #ifndef NO_HISTORY 60 #include "myhistedit.h" 61 #endif 62 63 /* 64 * Shell command parser. 65 */ 66 67 #define EOFMARKLEN 79 68 69 /* values returned by readtoken */ 70 #include "token.h" 71 72 73 74 struct heredoc { 75 struct heredoc *next; /* next here document in list */ 76 union node *here; /* redirection node */ 77 char *eofmark; /* string indicating end of input */ 78 int striptabs; /* if set, strip leading tabs */ 79 }; 80 81 82 83 STATIC struct heredoc *heredoclist; /* list of here documents to read */ 84 STATIC int parsebackquote; /* nonzero if we are inside backquotes */ 85 STATIC int doprompt; /* if set, prompt the user */ 86 STATIC int needprompt; /* true if interactive and at start of line */ 87 STATIC int lasttoken; /* last token read */ 88 MKINIT int tokpushback; /* last token pushed back */ 89 STATIC char *wordtext; /* text of last word returned by readtoken */ 90 MKINIT int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */ 91 STATIC struct nodelist *backquotelist; 92 STATIC union node *redirnode; 93 STATIC struct heredoc *heredoc; 94 STATIC int quoteflag; /* set if (part of) last token was quoted */ 95 STATIC int startlinno; /* line # where last token started */ 96 97 /* XXX When 'noaliases' is set to one, no alias expansion takes place. */ 98 static int noaliases = 0; 99 100 STATIC union node *list(int); 101 STATIC union node *andor(void); 102 STATIC union node *pipeline(void); 103 STATIC union node *command(void); 104 STATIC union node *simplecmd(union node **, union node *); 105 STATIC union node *makename(void); 106 STATIC void parsefname(void); 107 STATIC void parseheredoc(void); 108 STATIC int peektoken(void); 109 STATIC int readtoken(void); 110 STATIC int xxreadtoken(void); 111 STATIC int readtoken1(int, char const *, char *, int); 112 STATIC int noexpand(char *); 113 STATIC void synexpect(int); 114 STATIC void synerror(const char *); 115 STATIC void setprompt(int); 116 117 118 /* 119 * Read and parse a command. Returns NEOF on end of file. (NULL is a 120 * valid parse tree indicating a blank line.) 121 */ 122 123 union node * 124 parsecmd(int interact) 125 { 126 int t; 127 128 tokpushback = 0; 129 doprompt = interact; 130 if (doprompt) 131 setprompt(1); 132 else 133 setprompt(0); 134 needprompt = 0; 135 t = readtoken(); 136 if (t == TEOF) 137 return NEOF; 138 if (t == TNL) 139 return NULL; 140 tokpushback++; 141 return list(1); 142 } 143 144 145 STATIC union node * 146 list(int nlflag) 147 { 148 union node *n1, *n2, *n3; 149 int tok; 150 151 checkkwd = 2; 152 if (nlflag == 0 && tokendlist[peektoken()]) 153 return NULL; 154 n1 = NULL; 155 for (;;) { 156 n2 = andor(); 157 tok = readtoken(); 158 if (tok == TBACKGND) { 159 if (n2->type == NCMD || n2->type == NPIPE) { 160 n2->ncmd.backgnd = 1; 161 } else if (n2->type == NREDIR) { 162 n2->type = NBACKGND; 163 } else { 164 n3 = (union node *)stalloc(sizeof (struct nredir)); 165 n3->type = NBACKGND; 166 n3->nredir.n = n2; 167 n3->nredir.redirect = NULL; 168 n2 = n3; 169 } 170 } 171 if (n1 == NULL) { 172 n1 = n2; 173 } 174 else { 175 n3 = (union node *)stalloc(sizeof (struct nbinary)); 176 n3->type = NSEMI; 177 n3->nbinary.ch1 = n1; 178 n3->nbinary.ch2 = n2; 179 n1 = n3; 180 } 181 switch (tok) { 182 case TBACKGND: 183 case TSEMI: 184 tok = readtoken(); 185 /* fall through */ 186 case TNL: 187 if (tok == TNL) { 188 parseheredoc(); 189 if (nlflag) 190 return n1; 191 } else { 192 tokpushback++; 193 } 194 checkkwd = 2; 195 if (tokendlist[peektoken()]) 196 return n1; 197 break; 198 case TEOF: 199 if (heredoclist) 200 parseheredoc(); 201 else 202 pungetc(); /* push back EOF on input */ 203 return n1; 204 default: 205 if (nlflag) 206 synexpect(-1); 207 tokpushback++; 208 return n1; 209 } 210 } 211 } 212 213 214 215 STATIC union node * 216 andor(void) 217 { 218 union node *n1, *n2, *n3; 219 int t; 220 221 n1 = pipeline(); 222 for (;;) { 223 if ((t = readtoken()) == TAND) { 224 t = NAND; 225 } else if (t == TOR) { 226 t = NOR; 227 } else { 228 tokpushback++; 229 return n1; 230 } 231 n2 = pipeline(); 232 n3 = (union node *)stalloc(sizeof (struct nbinary)); 233 n3->type = t; 234 n3->nbinary.ch1 = n1; 235 n3->nbinary.ch2 = n2; 236 n1 = n3; 237 } 238 } 239 240 241 242 STATIC union node * 243 pipeline(void) 244 { 245 union node *n1, *n2, *pipenode; 246 struct nodelist *lp, *prev; 247 int negate; 248 249 negate = 0; 250 TRACE(("pipeline: entered\n")); 251 while (readtoken() == TNOT) 252 negate = !negate; 253 tokpushback++; 254 n1 = command(); 255 if (readtoken() == TPIPE) { 256 pipenode = (union node *)stalloc(sizeof (struct npipe)); 257 pipenode->type = NPIPE; 258 pipenode->npipe.backgnd = 0; 259 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 260 pipenode->npipe.cmdlist = lp; 261 lp->n = n1; 262 do { 263 prev = lp; 264 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 265 lp->n = command(); 266 prev->next = lp; 267 } while (readtoken() == TPIPE); 268 lp->next = NULL; 269 n1 = pipenode; 270 } 271 tokpushback++; 272 if (negate) { 273 n2 = (union node *)stalloc(sizeof (struct nnot)); 274 n2->type = NNOT; 275 n2->nnot.com = n1; 276 return n2; 277 } else 278 return n1; 279 } 280 281 282 283 STATIC union node * 284 command(void) 285 { 286 union node *n1, *n2; 287 union node *ap, **app; 288 union node *cp, **cpp; 289 union node *redir, **rpp; 290 int t, negate = 0; 291 292 checkkwd = 2; 293 redir = NULL; 294 n1 = NULL; 295 rpp = &redir; 296 297 /* Check for redirection which may precede command */ 298 while (readtoken() == TREDIR) { 299 *rpp = n2 = redirnode; 300 rpp = &n2->nfile.next; 301 parsefname(); 302 } 303 tokpushback++; 304 305 while (readtoken() == TNOT) { 306 TRACE(("command: TNOT recognized\n")); 307 negate = !negate; 308 } 309 tokpushback++; 310 311 switch (readtoken()) { 312 case TIF: 313 n1 = (union node *)stalloc(sizeof (struct nif)); 314 n1->type = NIF; 315 n1->nif.test = list(0); 316 if (readtoken() != TTHEN) 317 synexpect(TTHEN); 318 n1->nif.ifpart = list(0); 319 n2 = n1; 320 while (readtoken() == TELIF) { 321 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif)); 322 n2 = n2->nif.elsepart; 323 n2->type = NIF; 324 n2->nif.test = list(0); 325 if (readtoken() != TTHEN) 326 synexpect(TTHEN); 327 n2->nif.ifpart = list(0); 328 } 329 if (lasttoken == TELSE) 330 n2->nif.elsepart = list(0); 331 else { 332 n2->nif.elsepart = NULL; 333 tokpushback++; 334 } 335 if (readtoken() != TFI) 336 synexpect(TFI); 337 checkkwd = 1; 338 break; 339 case TWHILE: 340 case TUNTIL: { 341 int got; 342 n1 = (union node *)stalloc(sizeof (struct nbinary)); 343 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL; 344 n1->nbinary.ch1 = list(0); 345 if ((got=readtoken()) != TDO) { 346 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : "")); 347 synexpect(TDO); 348 } 349 n1->nbinary.ch2 = list(0); 350 if (readtoken() != TDONE) 351 synexpect(TDONE); 352 checkkwd = 1; 353 break; 354 } 355 case TFOR: 356 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext)) 357 synerror("Bad for loop variable"); 358 n1 = (union node *)stalloc(sizeof (struct nfor)); 359 n1->type = NFOR; 360 n1->nfor.var = wordtext; 361 if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) { 362 app = ≈ 363 while (readtoken() == TWORD) { 364 n2 = (union node *)stalloc(sizeof (struct narg)); 365 n2->type = NARG; 366 n2->narg.text = wordtext; 367 n2->narg.backquote = backquotelist; 368 *app = n2; 369 app = &n2->narg.next; 370 } 371 *app = NULL; 372 n1->nfor.args = ap; 373 if (lasttoken != TNL && lasttoken != TSEMI) 374 synexpect(-1); 375 } else { 376 static char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'}; 377 n2 = (union node *)stalloc(sizeof (struct narg)); 378 n2->type = NARG; 379 n2->narg.text = argvars; 380 n2->narg.backquote = NULL; 381 n2->narg.next = NULL; 382 n1->nfor.args = n2; 383 /* 384 * Newline or semicolon here is optional (but note 385 * that the original Bourne shell only allowed NL). 386 */ 387 if (lasttoken != TNL && lasttoken != TSEMI) 388 tokpushback++; 389 } 390 checkkwd = 2; 391 if ((t = readtoken()) == TDO) 392 t = TDONE; 393 else if (t == TBEGIN) 394 t = TEND; 395 else 396 synexpect(-1); 397 n1->nfor.body = list(0); 398 if (readtoken() != t) 399 synexpect(t); 400 checkkwd = 1; 401 break; 402 case TCASE: 403 n1 = (union node *)stalloc(sizeof (struct ncase)); 404 n1->type = NCASE; 405 if (readtoken() != TWORD) 406 synexpect(TWORD); 407 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg)); 408 n2->type = NARG; 409 n2->narg.text = wordtext; 410 n2->narg.backquote = backquotelist; 411 n2->narg.next = NULL; 412 while (readtoken() == TNL); 413 if (lasttoken != TWORD || ! equal(wordtext, "in")) 414 synerror("expecting \"in\""); 415 cpp = &n1->ncase.cases; 416 noaliases = 1; /* turn off alias expansion */ 417 checkkwd = 2, readtoken(); 418 while (lasttoken != TESAC) { 419 *cpp = cp = (union node *)stalloc(sizeof (struct nclist)); 420 cp->type = NCLIST; 421 app = &cp->nclist.pattern; 422 if (lasttoken == TLP) 423 readtoken(); 424 for (;;) { 425 *app = ap = (union node *)stalloc(sizeof (struct narg)); 426 ap->type = NARG; 427 ap->narg.text = wordtext; 428 ap->narg.backquote = backquotelist; 429 if (checkkwd = 2, readtoken() != TPIPE) 430 break; 431 app = &ap->narg.next; 432 readtoken(); 433 } 434 ap->narg.next = NULL; 435 if (lasttoken != TRP) 436 noaliases = 0, synexpect(TRP); 437 cp->nclist.body = list(0); 438 439 checkkwd = 2; 440 if ((t = readtoken()) != TESAC) { 441 if (t != TENDCASE) 442 noaliases = 0, synexpect(TENDCASE); 443 else 444 checkkwd = 2, readtoken(); 445 } 446 cpp = &cp->nclist.next; 447 } 448 noaliases = 0; /* reset alias expansion */ 449 *cpp = NULL; 450 checkkwd = 1; 451 break; 452 case TLP: 453 n1 = (union node *)stalloc(sizeof (struct nredir)); 454 n1->type = NSUBSHELL; 455 n1->nredir.n = list(0); 456 n1->nredir.redirect = NULL; 457 if (readtoken() != TRP) 458 synexpect(TRP); 459 checkkwd = 1; 460 break; 461 case TBEGIN: 462 n1 = list(0); 463 if (readtoken() != TEND) 464 synexpect(TEND); 465 checkkwd = 1; 466 break; 467 /* Handle an empty command like other simple commands. */ 468 case TSEMI: 469 case TAND: 470 case TOR: 471 /* 472 * An empty command before a ; doesn't make much sense, and 473 * should certainly be disallowed in the case of `if ;'. 474 */ 475 if (!redir) 476 synexpect(-1); 477 case TNL: 478 case TEOF: 479 case TWORD: 480 case TRP: 481 tokpushback++; 482 n1 = simplecmd(rpp, redir); 483 goto checkneg; 484 default: 485 synexpect(-1); 486 } 487 488 /* Now check for redirection which may follow command */ 489 while (readtoken() == TREDIR) { 490 *rpp = n2 = redirnode; 491 rpp = &n2->nfile.next; 492 parsefname(); 493 } 494 tokpushback++; 495 *rpp = NULL; 496 if (redir) { 497 if (n1->type != NSUBSHELL) { 498 n2 = (union node *)stalloc(sizeof (struct nredir)); 499 n2->type = NREDIR; 500 n2->nredir.n = n1; 501 n1 = n2; 502 } 503 n1->nredir.redirect = redir; 504 } 505 506 checkneg: 507 if (negate) { 508 n2 = (union node *)stalloc(sizeof (struct nnot)); 509 n2->type = NNOT; 510 n2->nnot.com = n1; 511 return n2; 512 } 513 else 514 return n1; 515 } 516 517 518 STATIC union node * 519 simplecmd(union node **rpp, union node *redir) 520 { 521 union node *args, **app; 522 union node **orig_rpp = rpp; 523 union node *n = NULL, *n2; 524 int negate = 0; 525 526 /* If we don't have any redirections already, then we must reset */ 527 /* rpp to be the address of the local redir variable. */ 528 if (redir == 0) 529 rpp = &redir; 530 531 args = NULL; 532 app = &args; 533 /* 534 * We save the incoming value, because we need this for shell 535 * functions. There can not be a redirect or an argument between 536 * the function name and the open parenthesis. 537 */ 538 orig_rpp = rpp; 539 540 while (readtoken() == TNOT) { 541 TRACE(("command: TNOT recognized\n")); 542 negate = !negate; 543 } 544 tokpushback++; 545 546 for (;;) { 547 if (readtoken() == TWORD) { 548 n = (union node *)stalloc(sizeof (struct narg)); 549 n->type = NARG; 550 n->narg.text = wordtext; 551 n->narg.backquote = backquotelist; 552 *app = n; 553 app = &n->narg.next; 554 } else if (lasttoken == TREDIR) { 555 *rpp = n = redirnode; 556 rpp = &n->nfile.next; 557 parsefname(); /* read name of redirection file */ 558 } else if (lasttoken == TLP && app == &args->narg.next 559 && rpp == orig_rpp) { 560 /* We have a function */ 561 if (readtoken() != TRP) 562 synexpect(TRP); 563 #ifdef notdef 564 if (! goodname(n->narg.text)) 565 synerror("Bad function name"); 566 #endif 567 n->type = NDEFUN; 568 n->narg.next = command(); 569 goto checkneg; 570 } else { 571 tokpushback++; 572 break; 573 } 574 } 575 *app = NULL; 576 *rpp = NULL; 577 n = (union node *)stalloc(sizeof (struct ncmd)); 578 n->type = NCMD; 579 n->ncmd.backgnd = 0; 580 n->ncmd.args = args; 581 n->ncmd.redirect = redir; 582 583 checkneg: 584 if (negate) { 585 n2 = (union node *)stalloc(sizeof (struct nnot)); 586 n2->type = NNOT; 587 n2->nnot.com = n; 588 return n2; 589 } 590 else 591 return n; 592 } 593 594 STATIC union node * 595 makename(void) 596 { 597 union node *n; 598 599 n = (union node *)stalloc(sizeof (struct narg)); 600 n->type = NARG; 601 n->narg.next = NULL; 602 n->narg.text = wordtext; 603 n->narg.backquote = backquotelist; 604 return n; 605 } 606 607 void 608 fixredir(union node *n, const char *text, int err) 609 { 610 TRACE(("Fix redir %s %d\n", text, err)); 611 if (!err) 612 n->ndup.vname = NULL; 613 614 if (is_digit(text[0]) && text[1] == '\0') 615 n->ndup.dupfd = digit_val(text[0]); 616 else if (text[0] == '-' && text[1] == '\0') 617 n->ndup.dupfd = -1; 618 else { 619 620 if (err) 621 synerror("Bad fd number"); 622 else 623 n->ndup.vname = makename(); 624 } 625 } 626 627 628 STATIC void 629 parsefname(void) 630 { 631 union node *n = redirnode; 632 633 if (readtoken() != TWORD) 634 synexpect(-1); 635 if (n->type == NHERE) { 636 struct heredoc *here = heredoc; 637 struct heredoc *p; 638 int i; 639 640 if (quoteflag == 0) 641 n->type = NXHERE; 642 TRACE(("Here document %d\n", n->type)); 643 if (here->striptabs) { 644 while (*wordtext == '\t') 645 wordtext++; 646 } 647 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN) 648 synerror("Illegal eof marker for << redirection"); 649 rmescapes(wordtext); 650 here->eofmark = wordtext; 651 here->next = NULL; 652 if (heredoclist == NULL) 653 heredoclist = here; 654 else { 655 for (p = heredoclist ; p->next ; p = p->next); 656 p->next = here; 657 } 658 } else if (n->type == NTOFD || n->type == NFROMFD) { 659 fixredir(n, wordtext, 0); 660 } else { 661 n->nfile.fname = makename(); 662 } 663 } 664 665 666 /* 667 * Input any here documents. 668 */ 669 670 STATIC void 671 parseheredoc(void) 672 { 673 struct heredoc *here; 674 union node *n; 675 676 while (heredoclist) { 677 here = heredoclist; 678 heredoclist = here->next; 679 if (needprompt) { 680 setprompt(2); 681 needprompt = 0; 682 } 683 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX, 684 here->eofmark, here->striptabs); 685 n = (union node *)stalloc(sizeof (struct narg)); 686 n->narg.type = NARG; 687 n->narg.next = NULL; 688 n->narg.text = wordtext; 689 n->narg.backquote = backquotelist; 690 here->here->nhere.doc = n; 691 } 692 } 693 694 STATIC int 695 peektoken(void) 696 { 697 int t; 698 699 t = readtoken(); 700 tokpushback++; 701 return (t); 702 } 703 704 STATIC int 705 readtoken(void) 706 { 707 int t; 708 int savecheckkwd = checkkwd; 709 struct alias *ap; 710 #ifdef DEBUG 711 int alreadyseen = tokpushback; 712 #endif 713 714 top: 715 t = xxreadtoken(); 716 717 if (checkkwd) { 718 /* 719 * eat newlines 720 */ 721 if (checkkwd == 2) { 722 checkkwd = 0; 723 while (t == TNL) { 724 parseheredoc(); 725 t = xxreadtoken(); 726 } 727 } else 728 checkkwd = 0; 729 /* 730 * check for keywords and aliases 731 */ 732 if (t == TWORD && !quoteflag) 733 { 734 const char * const *pp; 735 736 for (pp = parsekwd; *pp; pp++) { 737 if (**pp == *wordtext && equal(*pp, wordtext)) 738 { 739 lasttoken = t = pp - parsekwd + KWDOFFSET; 740 TRACE(("keyword %s recognized\n", tokname[t])); 741 goto out; 742 } 743 } 744 if (noaliases == 0 && 745 (ap = lookupalias(wordtext, 1)) != NULL) { 746 pushstring(ap->val, strlen(ap->val), ap); 747 checkkwd = savecheckkwd; 748 goto top; 749 } 750 } 751 out: 752 checkkwd = (t == TNOT) ? savecheckkwd : 0; 753 } 754 #ifdef DEBUG 755 if (!alreadyseen) 756 TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 757 else 758 TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 759 #endif 760 return (t); 761 } 762 763 764 /* 765 * Read the next input token. 766 * If the token is a word, we set backquotelist to the list of cmds in 767 * backquotes. We set quoteflag to true if any part of the word was 768 * quoted. 769 * If the token is TREDIR, then we set redirnode to a structure containing 770 * the redirection. 771 * In all cases, the variable startlinno is set to the number of the line 772 * on which the token starts. 773 * 774 * [Change comment: here documents and internal procedures] 775 * [Readtoken shouldn't have any arguments. Perhaps we should make the 776 * word parsing code into a separate routine. In this case, readtoken 777 * doesn't need to have any internal procedures, but parseword does. 778 * We could also make parseoperator in essence the main routine, and 779 * have parseword (readtoken1?) handle both words and redirection.] 780 */ 781 782 #define RETURN(token) return lasttoken = token 783 784 STATIC int 785 xxreadtoken(void) 786 { 787 int c; 788 789 if (tokpushback) { 790 tokpushback = 0; 791 return lasttoken; 792 } 793 if (needprompt) { 794 setprompt(2); 795 needprompt = 0; 796 } 797 startlinno = plinno; 798 for (;;) { /* until token or start of word found */ 799 c = pgetc_macro(); 800 if (c == ' ' || c == '\t') 801 continue; /* quick check for white space first */ 802 switch (c) { 803 case ' ': case '\t': 804 continue; 805 case '#': 806 while ((c = pgetc()) != '\n' && c != PEOF); 807 pungetc(); 808 continue; 809 case '\\': 810 if (pgetc() == '\n') { 811 startlinno = ++plinno; 812 if (doprompt) 813 setprompt(2); 814 else 815 setprompt(0); 816 continue; 817 } 818 pungetc(); 819 goto breakloop; 820 case '\n': 821 plinno++; 822 needprompt = doprompt; 823 RETURN(TNL); 824 case PEOF: 825 RETURN(TEOF); 826 case '&': 827 if (pgetc() == '&') 828 RETURN(TAND); 829 pungetc(); 830 RETURN(TBACKGND); 831 case '|': 832 if (pgetc() == '|') 833 RETURN(TOR); 834 pungetc(); 835 RETURN(TPIPE); 836 case ';': 837 if (pgetc() == ';') 838 RETURN(TENDCASE); 839 pungetc(); 840 RETURN(TSEMI); 841 case '(': 842 RETURN(TLP); 843 case ')': 844 RETURN(TRP); 845 default: 846 goto breakloop; 847 } 848 } 849 breakloop: 850 return readtoken1(c, BASESYNTAX, (char *)NULL, 0); 851 #undef RETURN 852 } 853 854 855 856 /* 857 * If eofmark is NULL, read a word or a redirection symbol. If eofmark 858 * is not NULL, read a here document. In the latter case, eofmark is the 859 * word which marks the end of the document and striptabs is true if 860 * leading tabs should be stripped from the document. The argument firstc 861 * is the first character of the input token or document. 862 * 863 * Because C does not have internal subroutines, I have simulated them 864 * using goto's to implement the subroutine linkage. The following macros 865 * will run code that appears at the end of readtoken1. 866 */ 867 868 #define CHECKEND() {goto checkend; checkend_return:;} 869 #define PARSEREDIR() {goto parseredir; parseredir_return:;} 870 #define PARSESUB() {goto parsesub; parsesub_return:;} 871 #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;} 872 #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;} 873 #define PARSEARITH() {goto parsearith; parsearith_return:;} 874 875 STATIC int 876 readtoken1(int firstc, char const *syntax, char *eofmark, int striptabs) 877 { 878 int c = firstc; 879 char *out; 880 int len; 881 char line[EOFMARKLEN + 1]; 882 struct nodelist *bqlist; 883 int quotef; 884 int dblquote; 885 int varnest; /* levels of variables expansion */ 886 int arinest; /* levels of arithmetic expansion */ 887 int parenlevel; /* levels of parens in arithmetic */ 888 int oldstyle; 889 char const *prevsyntax; /* syntax before arithmetic */ 890 int synentry; 891 #if __GNUC__ 892 /* Avoid longjmp clobbering */ 893 (void) &out; 894 (void) "ef; 895 (void) &dblquote; 896 (void) &varnest; 897 (void) &arinest; 898 (void) &parenlevel; 899 (void) &oldstyle; 900 (void) &prevsyntax; 901 (void) &syntax; 902 (void) &synentry; 903 #endif 904 905 startlinno = plinno; 906 dblquote = 0; 907 if (syntax == DQSYNTAX) 908 dblquote = 1; 909 quotef = 0; 910 bqlist = NULL; 911 varnest = 0; 912 arinest = 0; 913 parenlevel = 0; 914 915 STARTSTACKSTR(out); 916 loop: { /* for each line, until end of word */ 917 CHECKEND(); /* set c to PEOF if at end of here document */ 918 for (;;) { /* until end of line or end of word */ 919 CHECKSTRSPACE(3, out); /* permit 3 calls to USTPUTC */ 920 921 synentry = syntax[c]; 922 923 switch(synentry) { 924 case CNL: /* '\n' */ 925 if (syntax == BASESYNTAX) 926 goto endword; /* exit outer loop */ 927 USTPUTC(c, out); 928 plinno++; 929 if (doprompt) 930 setprompt(2); 931 else 932 setprompt(0); 933 c = pgetc(); 934 goto loop; /* continue outer loop */ 935 case CWORD: 936 USTPUTC(c, out); 937 break; 938 case CCTL: 939 if (eofmark == NULL || dblquote) 940 USTPUTC(CTLESC, out); 941 USTPUTC(c, out); 942 break; 943 case CBACK: /* backslash */ 944 c = pgetc(); 945 if (c == PEOF) { 946 USTPUTC('\\', out); 947 pungetc(); 948 } else if (c == '\n') { 949 if (doprompt) 950 setprompt(2); 951 else 952 setprompt(0); 953 } else { 954 if (dblquote && c != '\\' && 955 c != '`' && c != '$' && 956 (c != '"' || eofmark != NULL)) 957 USTPUTC('\\', out); 958 if (SQSYNTAX[c] == CCTL) 959 USTPUTC(CTLESC, out); 960 else if (eofmark == NULL) 961 USTPUTC(CTLQUOTEMARK, out); 962 USTPUTC(c, out); 963 quotef++; 964 } 965 break; 966 case CSQUOTE: 967 if (eofmark == NULL) 968 USTPUTC(CTLQUOTEMARK, out); 969 syntax = SQSYNTAX; 970 break; 971 case CDQUOTE: 972 if (eofmark == NULL) 973 USTPUTC(CTLQUOTEMARK, out); 974 syntax = DQSYNTAX; 975 dblquote = 1; 976 break; 977 case CENDQUOTE: 978 if (eofmark != NULL && arinest == 0 && 979 varnest == 0) { 980 USTPUTC(c, out); 981 } else { 982 if (arinest) { 983 syntax = ARISYNTAX; 984 dblquote = 0; 985 } else if (eofmark == NULL) { 986 syntax = BASESYNTAX; 987 dblquote = 0; 988 } 989 quotef++; 990 } 991 break; 992 case CVAR: /* '$' */ 993 PARSESUB(); /* parse substitution */ 994 break; 995 case CENDVAR: /* '}' */ 996 if (varnest > 0) { 997 varnest--; 998 USTPUTC(CTLENDVAR, out); 999 } else { 1000 USTPUTC(c, out); 1001 } 1002 break; 1003 case CLP: /* '(' in arithmetic */ 1004 parenlevel++; 1005 USTPUTC(c, out); 1006 break; 1007 case CRP: /* ')' in arithmetic */ 1008 if (parenlevel > 0) { 1009 USTPUTC(c, out); 1010 --parenlevel; 1011 } else { 1012 if (pgetc() == ')') { 1013 if (--arinest == 0) { 1014 USTPUTC(CTLENDARI, out); 1015 syntax = prevsyntax; 1016 if (syntax == DQSYNTAX) 1017 dblquote = 1; 1018 else 1019 dblquote = 0; 1020 } else 1021 USTPUTC(')', out); 1022 } else { 1023 /* 1024 * unbalanced parens 1025 * (don't 2nd guess - no error) 1026 */ 1027 pungetc(); 1028 USTPUTC(')', out); 1029 } 1030 } 1031 break; 1032 case CBQUOTE: /* '`' */ 1033 PARSEBACKQOLD(); 1034 break; 1035 case CEOF: 1036 goto endword; /* exit outer loop */ 1037 default: 1038 if (varnest == 0) 1039 goto endword; /* exit outer loop */ 1040 USTPUTC(c, out); 1041 } 1042 c = pgetc_macro(); 1043 } 1044 } 1045 endword: 1046 if (syntax == ARISYNTAX) 1047 synerror("Missing '))'"); 1048 if (syntax != BASESYNTAX && ! parsebackquote && eofmark == NULL) 1049 synerror("Unterminated quoted string"); 1050 if (varnest != 0) { 1051 startlinno = plinno; 1052 synerror("Missing '}'"); 1053 } 1054 USTPUTC('\0', out); 1055 len = out - stackblock(); 1056 out = stackblock(); 1057 if (eofmark == NULL) { 1058 if ((c == '>' || c == '<') 1059 && quotef == 0 1060 && len <= 2 1061 && (*out == '\0' || is_digit(*out))) { 1062 PARSEREDIR(); 1063 return lasttoken = TREDIR; 1064 } else { 1065 pungetc(); 1066 } 1067 } 1068 quoteflag = quotef; 1069 backquotelist = bqlist; 1070 grabstackblock(len); 1071 wordtext = out; 1072 return lasttoken = TWORD; 1073 /* end of readtoken routine */ 1074 1075 1076 1077 /* 1078 * Check to see whether we are at the end of the here document. When this 1079 * is called, c is set to the first character of the next input line. If 1080 * we are at the end of the here document, this routine sets the c to PEOF. 1081 */ 1082 1083 checkend: { 1084 if (eofmark) { 1085 if (striptabs) { 1086 while (c == '\t') 1087 c = pgetc(); 1088 } 1089 if (c == *eofmark) { 1090 if (pfgets(line, sizeof line) != NULL) { 1091 char *p, *q; 1092 1093 p = line; 1094 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++); 1095 if (*p == '\n' && *q == '\0') { 1096 c = PEOF; 1097 plinno++; 1098 needprompt = doprompt; 1099 } else { 1100 pushstring(line, strlen(line), NULL); 1101 } 1102 } 1103 } 1104 } 1105 goto checkend_return; 1106 } 1107 1108 1109 /* 1110 * Parse a redirection operator. The variable "out" points to a string 1111 * specifying the fd to be redirected. The variable "c" contains the 1112 * first character of the redirection operator. 1113 */ 1114 1115 parseredir: { 1116 char fd = *out; 1117 union node *np; 1118 1119 np = (union node *)stalloc(sizeof (struct nfile)); 1120 if (c == '>') { 1121 np->nfile.fd = 1; 1122 c = pgetc(); 1123 if (c == '>') 1124 np->type = NAPPEND; 1125 else if (c == '&') 1126 np->type = NTOFD; 1127 else if (c == '|') 1128 np->type = NCLOBBER; 1129 else { 1130 np->type = NTO; 1131 pungetc(); 1132 } 1133 } else { /* c == '<' */ 1134 np->nfile.fd = 0; 1135 c = pgetc(); 1136 if (c == '<') { 1137 if (sizeof (struct nfile) != sizeof (struct nhere)) { 1138 np = (union node *)stalloc(sizeof (struct nhere)); 1139 np->nfile.fd = 0; 1140 } 1141 np->type = NHERE; 1142 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc)); 1143 heredoc->here = np; 1144 if ((c = pgetc()) == '-') { 1145 heredoc->striptabs = 1; 1146 } else { 1147 heredoc->striptabs = 0; 1148 pungetc(); 1149 } 1150 } else if (c == '&') 1151 np->type = NFROMFD; 1152 else if (c == '>') 1153 np->type = NFROMTO; 1154 else { 1155 np->type = NFROM; 1156 pungetc(); 1157 } 1158 } 1159 if (fd != '\0') 1160 np->nfile.fd = digit_val(fd); 1161 redirnode = np; 1162 goto parseredir_return; 1163 } 1164 1165 1166 /* 1167 * Parse a substitution. At this point, we have read the dollar sign 1168 * and nothing else. 1169 */ 1170 1171 parsesub: { 1172 int subtype; 1173 int typeloc; 1174 int flags; 1175 char *p; 1176 static const char types[] = "}-+?="; 1177 int bracketed_name = 0; /* used to handle ${[0-9]*} variables */ 1178 1179 c = pgetc(); 1180 if (c != '(' && c != '{' && (is_eof(c) || !is_name(c)) && 1181 !is_special(c)) { 1182 USTPUTC('$', out); 1183 pungetc(); 1184 } else if (c == '(') { /* $(command) or $((arith)) */ 1185 if (pgetc() == '(') { 1186 PARSEARITH(); 1187 } else { 1188 pungetc(); 1189 PARSEBACKQNEW(); 1190 } 1191 } else { 1192 USTPUTC(CTLVAR, out); 1193 typeloc = out - stackblock(); 1194 USTPUTC(VSNORMAL, out); 1195 subtype = VSNORMAL; 1196 if (c == '{') { 1197 bracketed_name = 1; 1198 c = pgetc(); 1199 if (c == '#') { 1200 if ((c = pgetc()) == '}') 1201 c = '#'; 1202 else 1203 subtype = VSLENGTH; 1204 } 1205 else 1206 subtype = 0; 1207 } 1208 if (!is_eof(c) && is_name(c)) { 1209 do { 1210 STPUTC(c, out); 1211 c = pgetc(); 1212 } while (!is_eof(c) && is_in_name(c)); 1213 } else if (is_digit(c)) { 1214 if (bracketed_name) { 1215 do { 1216 STPUTC(c, out); 1217 c = pgetc(); 1218 } while (is_digit(c)); 1219 } else { 1220 STPUTC(c, out); 1221 c = pgetc(); 1222 } 1223 } else { 1224 if (! is_special(c)) 1225 badsub: synerror("Bad substitution"); 1226 USTPUTC(c, out); 1227 c = pgetc(); 1228 } 1229 STPUTC('=', out); 1230 flags = 0; 1231 if (subtype == 0) { 1232 switch (c) { 1233 case ':': 1234 flags = VSNUL; 1235 c = pgetc(); 1236 /*FALLTHROUGH*/ 1237 default: 1238 p = strchr(types, c); 1239 if (p == NULL) 1240 goto badsub; 1241 subtype = p - types + VSNORMAL; 1242 break; 1243 case '%': 1244 case '#': 1245 { 1246 int cc = c; 1247 subtype = c == '#' ? VSTRIMLEFT : 1248 VSTRIMRIGHT; 1249 c = pgetc(); 1250 if (c == cc) 1251 subtype++; 1252 else 1253 pungetc(); 1254 break; 1255 } 1256 } 1257 } else { 1258 pungetc(); 1259 } 1260 if (subtype != VSLENGTH && (dblquote || arinest)) 1261 flags |= VSQUOTE; 1262 *(stackblock() + typeloc) = subtype | flags; 1263 if (subtype != VSNORMAL) 1264 varnest++; 1265 } 1266 goto parsesub_return; 1267 } 1268 1269 1270 /* 1271 * Called to parse command substitutions. Newstyle is set if the command 1272 * is enclosed inside $(...); nlpp is a pointer to the head of the linked 1273 * list of commands (passed by reference), and savelen is the number of 1274 * characters on the top of the stack which must be preserved. 1275 */ 1276 1277 parsebackq: { 1278 struct nodelist **nlpp; 1279 int savepbq; 1280 union node *n; 1281 char *volatile str; 1282 struct jmploc jmploc; 1283 struct jmploc *volatile savehandler; 1284 int savelen; 1285 int saveprompt; 1286 #if __GNUC__ 1287 /* Avoid longjmp clobbering */ 1288 (void) &saveprompt; 1289 #endif 1290 1291 savepbq = parsebackquote; 1292 if (setjmp(jmploc.loc)) { 1293 if (str) 1294 ckfree(str); 1295 parsebackquote = 0; 1296 handler = savehandler; 1297 longjmp(handler->loc, 1); 1298 } 1299 INTOFF; 1300 str = NULL; 1301 savelen = out - stackblock(); 1302 if (savelen > 0) { 1303 str = ckmalloc(savelen); 1304 memcpy(str, stackblock(), savelen); 1305 } 1306 savehandler = handler; 1307 handler = &jmploc; 1308 INTON; 1309 if (oldstyle) { 1310 /* We must read until the closing backquote, giving special 1311 treatment to some slashes, and then push the string and 1312 reread it as input, interpreting it normally. */ 1313 char *pout; 1314 int pc; 1315 int psavelen; 1316 char *pstr; 1317 1318 1319 STARTSTACKSTR(pout); 1320 for (;;) { 1321 if (needprompt) { 1322 setprompt(2); 1323 needprompt = 0; 1324 } 1325 switch (pc = pgetc()) { 1326 case '`': 1327 goto done; 1328 1329 case '\\': 1330 if ((pc = pgetc()) == '\n') { 1331 plinno++; 1332 if (doprompt) 1333 setprompt(2); 1334 else 1335 setprompt(0); 1336 /* 1337 * If eating a newline, avoid putting 1338 * the newline into the new character 1339 * stream (via the STPUTC after the 1340 * switch). 1341 */ 1342 continue; 1343 } 1344 if (pc != '\\' && pc != '`' && pc != '$' 1345 && (!dblquote || pc != '"')) 1346 STPUTC('\\', pout); 1347 break; 1348 1349 case '\n': 1350 plinno++; 1351 needprompt = doprompt; 1352 break; 1353 1354 case PEOF: 1355 startlinno = plinno; 1356 synerror("EOF in backquote substitution"); 1357 break; 1358 1359 default: 1360 break; 1361 } 1362 STPUTC(pc, pout); 1363 } 1364 done: 1365 STPUTC('\0', pout); 1366 psavelen = pout - stackblock(); 1367 if (psavelen > 0) { 1368 pstr = ckmalloc(psavelen); 1369 memcpy(pstr, stackblock(), psavelen); 1370 setinputstring(pstr, 1); 1371 } 1372 } 1373 nlpp = &bqlist; 1374 while (*nlpp) 1375 nlpp = &(*nlpp)->next; 1376 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 1377 (*nlpp)->next = NULL; 1378 parsebackquote = oldstyle; 1379 1380 if (oldstyle) { 1381 saveprompt = doprompt; 1382 doprompt = 0; 1383 } 1384 1385 n = list(0); 1386 1387 if (oldstyle) 1388 doprompt = saveprompt; 1389 else { 1390 if (readtoken() != TRP) 1391 synexpect(TRP); 1392 } 1393 1394 (*nlpp)->n = n; 1395 if (oldstyle) { 1396 /* 1397 * Start reading from old file again, ignoring any pushed back 1398 * tokens left from the backquote parsing 1399 */ 1400 popfile(); 1401 tokpushback = 0; 1402 } 1403 while (stackblocksize() <= savelen) 1404 growstackblock(); 1405 STARTSTACKSTR(out); 1406 if (str) { 1407 memcpy(out, str, savelen); 1408 STADJUST(savelen, out); 1409 INTOFF; 1410 ckfree(str); 1411 str = NULL; 1412 INTON; 1413 } 1414 parsebackquote = savepbq; 1415 handler = savehandler; 1416 if (arinest || dblquote) 1417 USTPUTC(CTLBACKQ | CTLQUOTE, out); 1418 else 1419 USTPUTC(CTLBACKQ, out); 1420 if (oldstyle) 1421 goto parsebackq_oldreturn; 1422 else 1423 goto parsebackq_newreturn; 1424 } 1425 1426 /* 1427 * Parse an arithmetic expansion (indicate start of one and set state) 1428 */ 1429 parsearith: { 1430 1431 if (++arinest == 1) { 1432 prevsyntax = syntax; 1433 syntax = ARISYNTAX; 1434 USTPUTC(CTLARI, out); 1435 if (dblquote) 1436 USTPUTC('"',out); 1437 else 1438 USTPUTC(' ',out); 1439 } else { 1440 /* 1441 * we collapse embedded arithmetic expansion to 1442 * parenthesis, which should be equivalent 1443 */ 1444 USTPUTC('(', out); 1445 } 1446 goto parsearith_return; 1447 } 1448 1449 } /* end of readtoken */ 1450 1451 1452 1453 #ifdef mkinit 1454 RESET { 1455 tokpushback = 0; 1456 checkkwd = 0; 1457 } 1458 #endif 1459 1460 /* 1461 * Returns true if the text contains nothing to expand (no dollar signs 1462 * or backquotes). 1463 */ 1464 1465 STATIC int 1466 noexpand(char *text) 1467 { 1468 char *p; 1469 char c; 1470 1471 p = text; 1472 while ((c = *p++) != '\0') { 1473 if ( c == CTLQUOTEMARK) 1474 continue; 1475 if (c == CTLESC) 1476 p++; 1477 else if (BASESYNTAX[(int)c] == CCTL) 1478 return 0; 1479 } 1480 return 1; 1481 } 1482 1483 1484 /* 1485 * Return true if the argument is a legal variable name (a letter or 1486 * underscore followed by zero or more letters, underscores, and digits). 1487 */ 1488 1489 int 1490 goodname(char *name) 1491 { 1492 char *p; 1493 1494 p = name; 1495 if (! is_name(*p)) 1496 return 0; 1497 while (*++p) { 1498 if (! is_in_name(*p)) 1499 return 0; 1500 } 1501 return 1; 1502 } 1503 1504 1505 /* 1506 * Called when an unexpected token is read during the parse. The argument 1507 * is the token that is expected, or -1 if more than one type of token can 1508 * occur at this point. 1509 */ 1510 1511 STATIC void 1512 synexpect(int token) 1513 { 1514 char msg[64]; 1515 1516 if (token >= 0) { 1517 fmtstr(msg, 64, "%s unexpected (expecting %s)", 1518 tokname[lasttoken], tokname[token]); 1519 } else { 1520 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]); 1521 } 1522 synerror(msg); 1523 } 1524 1525 1526 STATIC void 1527 synerror(const char *msg) 1528 { 1529 if (commandname) 1530 outfmt(&errout, "%s: %d: ", commandname, startlinno); 1531 outfmt(&errout, "Syntax error: %s\n", msg); 1532 error((char *)NULL); 1533 } 1534 1535 STATIC void 1536 setprompt(int which) 1537 { 1538 whichprompt = which; 1539 1540 #ifndef NO_HISTORY 1541 if (!el) 1542 #endif 1543 out2str(getprompt(NULL)); 1544 } 1545 1546 /* 1547 * called by editline -- any expansions to the prompt 1548 * should be added here. 1549 */ 1550 const char * 1551 getprompt(void *unused __unused) 1552 { 1553 switch (whichprompt) { 1554 case 0: 1555 return ""; 1556 case 1: 1557 return ps1val(); 1558 case 2: 1559 return ps2val(); 1560 default: 1561 return "<internal prompt error>"; 1562 } 1563 } 1564