bin/sh/parser.c

   1 /*-
   2  * Copyright (c) 1991, 1993
   3  *      The Regents of the University of California.  All rights reserved.
   4  *
   5  * This code is derived from software contributed to Berkeley by
   6  * Kenneth Almquist.
   7  *
   8  * Redistribution and use in source and binary forms, with or without
   9  * modification, are permitted provided that the following conditions
  10  * are met:
  11  * 1. Redistributions of source code must retain the above copyright
  12  *    notice, this list of conditions and the following disclaimer.
  13  * 2. Redistributions in binary form must reproduce the above copyright
  14  *    notice, this list of conditions and the following disclaimer in the
  15  *    documentation and/or other materials provided with the distribution.
  16  * 4. Neither the name of the University nor the names of its contributors
  17  *    may be used to endorse or promote products derived from this software
  18  *    without specific prior written permission.
  19  *
  20  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  30  * SUCH DAMAGE.
  31  */
  32
  33 #ifndef lint
  34 #if 0
  35 static char sccsid[] = "@(#)parser.c    8.7 (Berkeley) 5/16/95";
  36 #endif
  37 #endif /* not lint */
  38 #include <sys/cdefs.h>
  39 __FBSDID("$FreeBSD$");
  40
  41 #include <stdlib.h>
  42 #include <unistd.h>
  43 #include <stdio.h>
  44
  45 #include "shell.h"
  46 #include "parser.h"
  47 #include "nodes.h"
  48 #include "expand.h"     /* defines rmescapes() */
  49 #include "syntax.h"
  50 #include "options.h"
  51 #include "input.h"
  52 #include "output.h"
  53 #include "var.h"
  54 #include "error.h"
  55 #include "memalloc.h"
  56 #include "mystring.h"
  57 #include "alias.h"
  58 #include "show.h"
  59 #include "eval.h"
  60 #include "exec.h"       /* to check for special builtins */
  61 #ifndef NO_HISTORY
  62 #include "myhistedit.h"
  63 #endif
  64
  65 /*
  66  * Shell command parser.
  67  */
  68
  69 #define EOFMARKLEN      79
  70 #define PROMPTLEN       128
  71
  72 /* values of checkkwd variable */
  73 #define CHKALIAS        0x1
  74 #define CHKKWD          0x2
  75 #define CHKNL           0x4
  76
  77 /* values returned by readtoken */
  78 #include "token.h"
  79
  80
  81
  82 struct heredoc {
  83         struct heredoc *next;   /* next here document in list */
  84         union node *here;               /* redirection node */
  85         char *eofmark;          /* string indicating end of input */
  86         int striptabs;          /* if set, strip leading tabs */
  87 };
  88
  89 struct parser_temp {
  90         struct parser_temp *next;
  91         void *data;
  92 };
  93
  94
  95 static struct heredoc *heredoclist;     /* list of here documents to read */
  96 static int doprompt;            /* if set, prompt the user */
  97 static int needprompt;          /* true if interactive and at start of line */
  98 static int lasttoken;           /* last token read */
  99 static int tokpushback;         /* last token pushed back */
 100 static char *wordtext;          /* text of last word returned by readtoken */
 101 static int checkkwd;
 102 static struct nodelist *backquotelist;
 103 static union node *redirnode;
 104 static struct heredoc *heredoc;
 105 static int quoteflag;           /* set if (part of) last token was quoted */
 106 static int startlinno;          /* line # where last token started */
 107 static int funclinno;           /* line # where the current function started */
 108 static struct parser_temp *parser_temp;
 109
 110
 111 static union node *list(int);
 112 static union node *andor(void);
 113 static union node *pipeline(void);
 114 static union node *command(void);
 115 static union node *simplecmd(union node **, union node *);
 116 static union node *makename(void);
 117 static union node *makebinary(int type, union node *n1, union node *n2);
 118 static void parsefname(void);
 119 static void parseheredoc(void);
 120 static int peektoken(void);
 121 static int readtoken(void);
 122 static int xxreadtoken(void);
 123 static int readtoken1(int, const char *, const char *, int);
 124 static int noexpand(char *);
 125 static void consumetoken(int);
 126 static void synexpect(int) __dead2;
 127 static void synerror(const char *) __dead2;
 128 static void setprompt(int);
 129
 130
 131 static void *
 132 parser_temp_alloc(size_t len)
 133 {
 134         struct parser_temp *t;
 135
 136         INTOFF;
 137         t = ckmalloc(sizeof(*t));
 138         t->data = NULL;
 139         t->next = parser_temp;
 140         parser_temp = t;
 141         t->data = ckmalloc(len);
 142         INTON;
 143         return t->data;
 144 }
 145
 146
 147 static void *
 148 parser_temp_realloc(void *ptr, size_t len)
 149 {
 150         struct parser_temp *t;
 151
 152         INTOFF;
 153         t = parser_temp;
 154         if (ptr != t->data)
 155                 error("bug: parser_temp_realloc misused");
 156         t->data = ckrealloc(t->data, len);
 157         INTON;
 158         return t->data;
 159 }
 160
 161
 162 static void
 163 parser_temp_free_upto(void *ptr)
 164 {
 165         struct parser_temp *t;
 166         int done = 0;
 167
 168         INTOFF;
 169         while (parser_temp != NULL && !done) {
 170                 t = parser_temp;
 171                 parser_temp = t->next;
 172                 done = t->data == ptr;
 173                 ckfree(t->data);
 174                 ckfree(t);
 175         }
 176         INTON;
 177         if (!done)
 178                 error("bug: parser_temp_free_upto misused");
 179 }
 180
 181
 182 static void
 183 parser_temp_free_all(void)
 184 {
 185         struct parser_temp *t;
 186
 187         INTOFF;
 188         while (parser_temp != NULL) {
 189                 t = parser_temp;
 190                 parser_temp = t->next;
 191                 ckfree(t->data);
 192                 ckfree(t);
 193         }
 194         INTON;
 195 }
 196
 197
 198 /*
 199  * Read and parse a command.  Returns NEOF on end of file.  (NULL is a
 200  * valid parse tree indicating a blank line.)
 201  */
 202
 203 union node *
 204 parsecmd(int interact)
 205 {
 206         int t;
 207
 208         /* This assumes the parser is not re-entered,
 209          * which could happen if we add command substitution on PS1/PS2.
 210          */
 211         parser_temp_free_all();
 212         heredoclist = NULL;
 213
 214         tokpushback = 0;
 215         checkkwd = 0;
 216         doprompt = interact;
 217         if (doprompt)
 218                 setprompt(1);
 219         else
 220                 setprompt(0);
 221         needprompt = 0;
 222         t = readtoken();
 223         if (t == TEOF)
 224                 return NEOF;
 225         if (t == TNL)
 226                 return NULL;
 227         tokpushback++;
 228         return list(1);
 229 }
 230
 231
 232 /*
 233  * Read and parse words for wordexp.
 234  * Returns a list of NARG nodes; NULL if there are no words.
 235  */
 236 union node *
 237 parsewordexp(void)
 238 {
 239         union node *n, *first = NULL, **pnext;
 240         int t;
 241
 242         /* This assumes the parser is not re-entered,
 243          * which could happen if we add command substitution on PS1/PS2.
 244          */
 245         parser_temp_free_all();
 246         heredoclist = NULL;
 247
 248         tokpushback = 0;
 249         checkkwd = 0;
 250         doprompt = 0;
 251         setprompt(0);
 252         needprompt = 0;
 253         pnext = &first;
 254         while ((t = readtoken()) != TEOF) {
 255                 if (t != TWORD)
 256                         synexpect(TWORD);
 257                 n = makename();
 258                 *pnext = n;
 259                 pnext = &n->narg.next;
 260         }
 261         return first;
 262 }
 263
 264
 265 static union node *
 266 list(int nlflag)
 267 {
 268         union node *ntop, *n1, *n2, *n3;
 269         int tok;
 270
 271         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 272         if (!nlflag && tokendlist[peektoken()])
 273                 return NULL;
 274         ntop = n1 = NULL;
 275         for (;;) {
 276                 n2 = andor();
 277                 tok = readtoken();
 278                 if (tok == TBACKGND) {
 279                         if (n2 != NULL && n2->type == NPIPE) {
 280                                 n2->npipe.backgnd = 1;
 281                         } else if (n2 != NULL && n2->type == NREDIR) {
 282                                 n2->type = NBACKGND;
 283                         } else {
 284                                 n3 = (union node *)stalloc(sizeof (struct nredir));
 285                                 n3->type = NBACKGND;
 286                                 n3->nredir.n = n2;
 287                                 n3->nredir.redirect = NULL;
 288                                 n2 = n3;
 289                         }
 290                 }
 291                 if (ntop == NULL)
 292                         ntop = n2;
 293                 else if (n1 == NULL) {
 294                         n1 = makebinary(NSEMI, ntop, n2);
 295                         ntop = n1;
 296                 }
 297                 else {
 298                         n3 = makebinary(NSEMI, n1->nbinary.ch2, n2);
 299                         n1->nbinary.ch2 = n3;
 300                         n1 = n3;
 301                 }
 302                 switch (tok) {
 303                 case TBACKGND:
 304                 case TSEMI:
 305                         tok = readtoken();
 306                         /* FALLTHROUGH */
 307                 case TNL:
 308                         if (tok == TNL) {
 309                                 parseheredoc();
 310                                 if (nlflag)
 311                                         return ntop;
 312                         } else if (tok == TEOF && nlflag) {
 313                                 parseheredoc();
 314                                 return ntop;
 315                         } else {
 316                                 tokpushback++;
 317                         }
 318                         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 319                         if (!nlflag && tokendlist[peektoken()])
 320                                 return ntop;
 321                         break;
 322                 case TEOF:
 323                         if (heredoclist)
 324                                 parseheredoc();
 325                         else
 326                                 pungetc();              /* push back EOF on input */
 327                         return ntop;
 328                 default:
 329                         if (nlflag)
 330                                 synexpect(-1);
 331                         tokpushback++;
 332                         return ntop;
 333                 }
 334         }
 335 }
 336
 337
 338
 339 static union node *
 340 andor(void)
 341 {
 342         union node *n;
 343         int t;
 344
 345         n = pipeline();
 346         for (;;) {
 347                 if ((t = readtoken()) == TAND) {
 348                         t = NAND;
 349                 } else if (t == TOR) {
 350                         t = NOR;
 351                 } else {
 352                         tokpushback++;
 353                         return n;
 354                 }
 355                 n = makebinary(t, n, pipeline());
 356         }
 357 }
 358
 359
 360
 361 static union node *
 362 pipeline(void)
 363 {
 364         union node *n1, *n2, *pipenode;
 365         struct nodelist *lp, *prev;
 366         int negate, t;
 367
 368         negate = 0;
 369         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 370         TRACE(("pipeline: entered\n"));
 371         while (readtoken() == TNOT)
 372                 negate = !negate;
 373         tokpushback++;
 374         n1 = command();
 375         if (readtoken() == TPIPE) {
 376                 pipenode = (union node *)stalloc(sizeof (struct npipe));
 377                 pipenode->type = NPIPE;
 378                 pipenode->npipe.backgnd = 0;
 379                 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
 380                 pipenode->npipe.cmdlist = lp;
 381                 lp->n = n1;
 382                 do {
 383                         prev = lp;
 384                         lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
 385                         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 386                         t = readtoken();
 387                         tokpushback++;
 388                         if (t == TNOT)
 389                                 lp->n = pipeline();
 390                         else
 391                                 lp->n = command();
 392                         prev->next = lp;
 393                 } while (readtoken() == TPIPE);
 394                 lp->next = NULL;
 395                 n1 = pipenode;
 396         }
 397         tokpushback++;
 398         if (negate) {
 399                 n2 = (union node *)stalloc(sizeof (struct nnot));
 400                 n2->type = NNOT;
 401                 n2->nnot.com = n1;
 402                 return n2;
 403         } else
 404                 return n1;
 405 }
 406
 407
 408
 409 static union node *
 410 command(void)
 411 {
 412         union node *n1, *n2;
 413         union node *ap, **app;
 414         union node *cp, **cpp;
 415         union node *redir, **rpp;
 416         int t;
 417         int is_subshell;
 418
 419         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 420         is_subshell = 0;
 421         redir = NULL;
 422         n1 = NULL;
 423         rpp = &redir;
 424
 425         /* Check for redirection which may precede command */
 426         while (readtoken() == TREDIR) {
 427                 *rpp = n2 = redirnode;
 428                 rpp = &n2->nfile.next;
 429                 parsefname();
 430         }
 431         tokpushback++;
 432
 433         switch (readtoken()) {
 434         case TIF:
 435                 n1 = (union node *)stalloc(sizeof (struct nif));
 436                 n1->type = NIF;
 437                 if ((n1->nif.test = list(0)) == NULL)
 438                         synexpect(-1);
 439                 consumetoken(TTHEN);
 440                 n1->nif.ifpart = list(0);
 441                 n2 = n1;
 442                 while (readtoken() == TELIF) {
 443                         n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif));
 444                         n2 = n2->nif.elsepart;
 445                         n2->type = NIF;
 446                         if ((n2->nif.test = list(0)) == NULL)
 447                                 synexpect(-1);
 448                         consumetoken(TTHEN);
 449                         n2->nif.ifpart = list(0);
 450                 }
 451                 if (lasttoken == TELSE)
 452                         n2->nif.elsepart = list(0);
 453                 else {
 454                         n2->nif.elsepart = NULL;
 455                         tokpushback++;
 456                 }
 457                 consumetoken(TFI);
 458                 checkkwd = CHKKWD | CHKALIAS;
 459                 break;
 460         case TWHILE:
 461         case TUNTIL:
 462                 t = lasttoken;
 463                 if ((n1 = list(0)) == NULL)
 464                         synexpect(-1);
 465                 consumetoken(TDO);
 466                 n1 = makebinary((t == TWHILE)? NWHILE : NUNTIL, n1, list(0));
 467                 consumetoken(TDONE);
 468                 checkkwd = CHKKWD | CHKALIAS;
 469                 break;
 470         case TFOR:
 471                 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext))
 472                         synerror("Bad for loop variable");
 473                 n1 = (union node *)stalloc(sizeof (struct nfor));
 474                 n1->type = NFOR;
 475                 n1->nfor.var = wordtext;
 476                 while (readtoken() == TNL)
 477                         ;
 478                 if (lasttoken == TWORD && ! quoteflag && equal(wordtext, "in")) {
 479                         app = &ap;
 480                         while (readtoken() == TWORD) {
 481                                 n2 = makename();
 482                                 *app = n2;
 483                                 app = &n2->narg.next;
 484                         }
 485                         *app = NULL;
 486                         n1->nfor.args = ap;
 487                         if (lasttoken != TNL && lasttoken != TSEMI)
 488                                 synexpect(-1);
 489                 } else {
 490                         static char argvars[5] = {
 491                                 CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'
 492                         };
 493                         n2 = (union node *)stalloc(sizeof (struct narg));
 494                         n2->type = NARG;
 495                         n2->narg.text = argvars;
 496                         n2->narg.backquote = NULL;
 497                         n2->narg.next = NULL;
 498                         n1->nfor.args = n2;
 499                         /*
 500                          * Newline or semicolon here is optional (but note
 501                          * that the original Bourne shell only allowed NL).
 502                          */
 503                         if (lasttoken != TNL && lasttoken != TSEMI)
 504                                 tokpushback++;
 505                 }
 506                 checkkwd = CHKNL | CHKKWD | CHKALIAS;
 507                 if ((t = readtoken()) == TDO)
 508                         t = TDONE;
 509                 else if (t == TBEGIN)
 510                         t = TEND;
 511                 else
 512                         synexpect(-1);
 513                 n1->nfor.body = list(0);
 514                 consumetoken(t);
 515                 checkkwd = CHKKWD | CHKALIAS;
 516                 break;
 517         case TCASE:
 518                 n1 = (union node *)stalloc(sizeof (struct ncase));
 519                 n1->type = NCASE;
 520                 consumetoken(TWORD);
 521                 n1->ncase.expr = makename();
 522                 while (readtoken() == TNL);
 523                 if (lasttoken != TWORD || ! equal(wordtext, "in"))
 524                         synerror("expecting \"in\"");
 525                 cpp = &n1->ncase.cases;
 526                 checkkwd = CHKNL | CHKKWD, readtoken();
 527                 while (lasttoken != TESAC) {
 528                         *cpp = cp = (union node *)stalloc(sizeof (struct nclist));
 529                         cp->type = NCLIST;
 530                         app = &cp->nclist.pattern;
 531                         if (lasttoken == TLP)
 532                                 readtoken();
 533                         for (;;) {
 534                                 *app = ap = makename();
 535                                 checkkwd = CHKNL | CHKKWD;
 536                                 if (readtoken() != TPIPE)
 537                                         break;
 538                                 app = &ap->narg.next;
 539                                 readtoken();
 540                         }
 541                         ap->narg.next = NULL;
 542                         if (lasttoken != TRP)
 543                                 synexpect(TRP);
 544                         cp->nclist.body = list(0);
 545
 546                         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 547                         if ((t = readtoken()) != TESAC) {
 548                                 if (t == TENDCASE)
 549                                         ;
 550                                 else if (t == TFALLTHRU)
 551                                         cp->type = NCLISTFALLTHRU;
 552                                 else
 553                                         synexpect(TENDCASE);
 554                                 checkkwd = CHKNL | CHKKWD, readtoken();
 555                         }
 556                         cpp = &cp->nclist.next;
 557                 }
 558                 *cpp = NULL;
 559                 checkkwd = CHKKWD | CHKALIAS;
 560                 break;
 561         case TLP:
 562                 n1 = (union node *)stalloc(sizeof (struct nredir));
 563                 n1->type = NSUBSHELL;
 564                 n1->nredir.n = list(0);
 565                 n1->nredir.redirect = NULL;
 566                 consumetoken(TRP);
 567                 checkkwd = CHKKWD | CHKALIAS;
 568                 is_subshell = 1;
 569                 break;
 570         case TBEGIN:
 571                 n1 = list(0);
 572                 consumetoken(TEND);
 573                 checkkwd = CHKKWD | CHKALIAS;
 574                 break;
 575         /* A simple command must have at least one redirection or word. */
 576         case TBACKGND:
 577         case TSEMI:
 578         case TAND:
 579         case TOR:
 580         case TPIPE:
 581         case TENDCASE:
 582         case TFALLTHRU:
 583         case TEOF:
 584         case TNL:
 585         case TRP:
 586                 if (!redir)
 587                         synexpect(-1);
 588         case TWORD:
 589                 tokpushback++;
 590                 n1 = simplecmd(rpp, redir);
 591                 return n1;
 592         default:
 593                 synexpect(-1);
 594         }
 595
 596         /* Now check for redirection which may follow command */
 597         while (readtoken() == TREDIR) {
 598                 *rpp = n2 = redirnode;
 599                 rpp = &n2->nfile.next;
 600                 parsefname();
 601         }
 602         tokpushback++;
 603         *rpp = NULL;
 604         if (redir) {
 605                 if (!is_subshell) {
 606                         n2 = (union node *)stalloc(sizeof (struct nredir));
 607                         n2->type = NREDIR;
 608                         n2->nredir.n = n1;
 609                         n1 = n2;
 610                 }
 611                 n1->nredir.redirect = redir;
 612         }
 613
 614         return n1;
 615 }
 616
 617
 618 static union node *
 619 simplecmd(union node **rpp, union node *redir)
 620 {
 621         union node *args, **app;
 622         union node **orig_rpp = rpp;
 623         union node *n = NULL;
 624         int special;
 625         int savecheckkwd;
 626
 627         /* If we don't have any redirections already, then we must reset */
 628         /* rpp to be the address of the local redir variable.  */
 629         if (redir == 0)
 630                 rpp = &redir;
 631
 632         args = NULL;
 633         app = &args;
 634         /*
 635          * We save the incoming value, because we need this for shell
 636          * functions.  There can not be a redirect or an argument between
 637          * the function name and the open parenthesis.
 638          */
 639         orig_rpp = rpp;
 640
 641         savecheckkwd = CHKALIAS;
 642
 643         for (;;) {
 644                 checkkwd = savecheckkwd;
 645                 if (readtoken() == TWORD) {
 646                         n = makename();
 647                         *app = n;
 648                         app = &n->narg.next;
 649                         if (savecheckkwd != 0 && !isassignment(wordtext))
 650                                 savecheckkwd = 0;
 651                 } else if (lasttoken == TREDIR) {
 652                         *rpp = n = redirnode;
 653                         rpp = &n->nfile.next;
 654                         parsefname();   /* read name of redirection file */
 655                 } else if (lasttoken == TLP && app == &args->narg.next
 656                                             && rpp == orig_rpp) {
 657                         /* We have a function */
 658                         consumetoken(TRP);
 659                         funclinno = plinno;
 660                         /*
 661                          * - Require plain text.
 662                          * - Functions with '/' cannot be called.
 663                          * - Reject name=().
 664                          * - Reject ksh extended glob patterns.
 665                          */
 666                         if (!noexpand(n->narg.text) || quoteflag ||
 667                             strchr(n->narg.text, '/') ||
 668                             strchr("!%*+-=?@}~",
 669                                 n->narg.text[strlen(n->narg.text) - 1]))
 670                                 synerror("Bad function name");
 671                         rmescapes(n->narg.text);
 672                         if (find_builtin(n->narg.text, &special) >= 0 &&
 673                             special)
 674                                 synerror("Cannot override a special builtin with a function");
 675                         n->type = NDEFUN;
 676                         n->narg.next = command();
 677                         funclinno = 0;
 678                         return n;
 679                 } else {
 680                         tokpushback++;
 681                         break;
 682                 }
 683         }
 684         *app = NULL;
 685         *rpp = NULL;
 686         n = (union node *)stalloc(sizeof (struct ncmd));
 687         n->type = NCMD;
 688         n->ncmd.args = args;
 689         n->ncmd.redirect = redir;
 690         return n;
 691 }
 692
 693 static union node *
 694 makename(void)
 695 {
 696         union node *n;
 697
 698         n = (union node *)stalloc(sizeof (struct narg));
 699         n->type = NARG;
 700         n->narg.next = NULL;
 701         n->narg.text = wordtext;
 702         n->narg.backquote = backquotelist;
 703         return n;
 704 }
 705
 706 static union node *
 707 makebinary(int type, union node *n1, union node *n2)
 708 {
 709         union node *n;
 710
 711         n = (union node *)stalloc(sizeof (struct nbinary));
 712         n->type = type;
 713         n->nbinary.ch1 = n1;
 714         n->nbinary.ch2 = n2;
 715         return (n);
 716 }
 717
 718 void
 719 forcealias(void)
 720 {
 721         checkkwd |= CHKALIAS;
 722 }
 723
 724 void
 725 fixredir(union node *n, const char *text, int err)
 726 {
 727         TRACE(("Fix redir %s %d\n", text, err));
 728         if (!err)
 729                 n->ndup.vname = NULL;
 730
 731         if (is_digit(text[0]) && text[1] == '\0')
 732                 n->ndup.dupfd = digit_val(text[0]);
 733         else if (text[0] == '-' && text[1] == '\0')
 734                 n->ndup.dupfd = -1;
 735         else {
 736
 737                 if (err)
 738                         synerror("Bad fd number");
 739                 else
 740                         n->ndup.vname = makename();
 741         }
 742 }
 743
 744
 745 static void
 746 parsefname(void)
 747 {
 748         union node *n = redirnode;
 749
 750         consumetoken(TWORD);
 751         if (n->type == NHERE) {
 752                 struct heredoc *here = heredoc;
 753                 struct heredoc *p;
 754                 int i;
 755
 756                 if (quoteflag == 0)
 757                         n->type = NXHERE;
 758                 TRACE(("Here document %d\n", n->type));
 759                 if (here->striptabs) {
 760                         while (*wordtext == '\t')
 761                                 wordtext++;
 762                 }
 763                 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN)
 764                         synerror("Illegal eof marker for << redirection");
 765                 rmescapes(wordtext);
 766                 here->eofmark = wordtext;
 767                 here->next = NULL;
 768                 if (heredoclist == NULL)
 769                         heredoclist = here;
 770                 else {
 771                         for (p = heredoclist ; p->next ; p = p->next);
 772                         p->next = here;
 773                 }
 774         } else if (n->type == NTOFD || n->type == NFROMFD) {
 775                 fixredir(n, wordtext, 0);
 776         } else {
 777                 n->nfile.fname = makename();
 778         }
 779 }
 780
 781
 782 /*
 783  * Input any here documents.
 784  */
 785
 786 static void
 787 parseheredoc(void)
 788 {
 789         struct heredoc *here;
 790         union node *n;
 791
 792         while (heredoclist) {
 793                 here = heredoclist;
 794                 heredoclist = here->next;
 795                 if (needprompt) {
 796                         setprompt(2);
 797                         needprompt = 0;
 798                 }
 799                 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX,
 800                                 here->eofmark, here->striptabs);
 801                 n = makename();
 802                 here->here->nhere.doc = n;
 803         }
 804 }
 805
 806 static int
 807 peektoken(void)
 808 {
 809         int t;
 810
 811         t = readtoken();
 812         tokpushback++;
 813         return (t);
 814 }
 815
 816 static int
 817 readtoken(void)
 818 {
 819         int t;
 820         struct alias *ap;
 821 #ifdef DEBUG
 822         int alreadyseen = tokpushback;
 823 #endif
 824
 825         top:
 826         t = xxreadtoken();
 827
 828         /*
 829          * eat newlines
 830          */
 831         if (checkkwd & CHKNL) {
 832                 while (t == TNL) {
 833                         parseheredoc();
 834                         t = xxreadtoken();
 835                 }
 836         }
 837
 838         /*
 839          * check for keywords and aliases
 840          */
 841         if (t == TWORD && !quoteflag)
 842         {
 843                 const char * const *pp;
 844
 845                 if (checkkwd & CHKKWD)
 846                         for (pp = parsekwd; *pp; pp++) {
 847                                 if (**pp == *wordtext && equal(*pp, wordtext))
 848                                 {
 849                                         lasttoken = t = pp - parsekwd + KWDOFFSET;
 850                                         TRACE(("keyword %s recognized\n", tokname[t]));
 851                                         goto out;
 852                                 }
 853                         }
 854                 if (checkkwd & CHKALIAS &&
 855                     (ap = lookupalias(wordtext, 1)) != NULL) {
 856                         pushstring(ap->val, strlen(ap->val), ap);
 857                         goto top;
 858                 }
 859         }
 860 out:
 861         if (t != TNOT)
 862                 checkkwd = 0;
 863
 864 #ifdef DEBUG
 865         if (!alreadyseen)
 866             TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
 867         else
 868             TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
 869 #endif
 870         return (t);
 871 }
 872
 873
 874 /*
 875  * Read the next input token.
 876  * If the token is a word, we set backquotelist to the list of cmds in
 877  *      backquotes.  We set quoteflag to true if any part of the word was
 878  *      quoted.
 879  * If the token is TREDIR, then we set redirnode to a structure containing
 880  *      the redirection.
 881  * In all cases, the variable startlinno is set to the number of the line
 882  *      on which the token starts.
 883  *
 884  * [Change comment:  here documents and internal procedures]
 885  * [Readtoken shouldn't have any arguments.  Perhaps we should make the
 886  *  word parsing code into a separate routine.  In this case, readtoken
 887  *  doesn't need to have any internal procedures, but parseword does.
 888  *  We could also make parseoperator in essence the main routine, and
 889  *  have parseword (readtoken1?) handle both words and redirection.]
 890  */
 891
 892 #define RETURN(token)   return lasttoken = token
 893
 894 static int
 895 xxreadtoken(void)
 896 {
 897         int c;
 898
 899         if (tokpushback) {
 900                 tokpushback = 0;
 901                 return lasttoken;
 902         }
 903         if (needprompt) {
 904                 setprompt(2);
 905                 needprompt = 0;
 906         }
 907         startlinno = plinno;
 908         for (;;) {      /* until token or start of word found */
 909                 c = pgetc_macro();
 910                 switch (c) {
 911                 case ' ': case '\t':
 912                         continue;
 913                 case '#':
 914                         while ((c = pgetc()) != '\n' && c != PEOF);
 915                         pungetc();
 916                         continue;
 917                 case '\\':
 918                         if (pgetc() == '\n') {
 919                                 startlinno = ++plinno;
 920                                 if (doprompt)
 921                                         setprompt(2);
 922                                 else
 923                                         setprompt(0);
 924                                 continue;
 925                         }
 926                         pungetc();
 927                         goto breakloop;
 928                 case '\n':
 929                         plinno++;
 930                         needprompt = doprompt;
 931                         RETURN(TNL);
 932                 case PEOF:
 933                         RETURN(TEOF);
 934                 case '&':
 935                         if (pgetc() == '&')
 936                                 RETURN(TAND);
 937                         pungetc();
 938                         RETURN(TBACKGND);
 939                 case '|':
 940                         if (pgetc() == '|')
 941                                 RETURN(TOR);
 942                         pungetc();
 943                         RETURN(TPIPE);
 944                 case ';':
 945                         c = pgetc();
 946                         if (c == ';')
 947                                 RETURN(TENDCASE);
 948                         else if (c == '&')
 949                                 RETURN(TFALLTHRU);
 950                         pungetc();
 951                         RETURN(TSEMI);
 952                 case '(':
 953                         RETURN(TLP);
 954                 case ')':
 955                         RETURN(TRP);
 956                 default:
 957                         goto breakloop;
 958                 }
 959         }
 960 breakloop:
 961         return readtoken1(c, BASESYNTAX, (char *)NULL, 0);
 962 #undef RETURN
 963 }
 964
 965
 966 #define MAXNEST_static 8
 967 struct tokenstate
 968 {
 969         const char *syntax; /* *SYNTAX */
 970         int parenlevel; /* levels of parentheses in arithmetic */
 971         enum tokenstate_category
 972         {
 973                 TSTATE_TOP,
 974                 TSTATE_VAR_OLD, /* ${var+-=?}, inherits dquotes */
 975                 TSTATE_VAR_NEW, /* other ${var...}, own dquote state */
 976                 TSTATE_ARITH
 977         } category;
 978 };
 979
 980
 981 /*
 982  * Called to parse command substitutions.
 983  */
 984
 985 static char *
 986 parsebackq(char *out, struct nodelist **pbqlist,
 987                 int oldstyle, int dblquote, int quoted)
 988 {
 989         struct nodelist **nlpp;
 990         union node *n;
 991         char *volatile str;
 992         struct jmploc jmploc;
 993         struct jmploc *const savehandler = handler;
 994         size_t savelen;
 995         int saveprompt;
 996         const int bq_startlinno = plinno;
 997         char *volatile ostr = NULL;
 998         struct parsefile *const savetopfile = getcurrentfile();
 999         struct heredoc *const saveheredoclist = heredoclist;
1000         struct heredoc *here;
1001
1002         str = NULL;
1003         if (setjmp(jmploc.loc)) {
1004                 popfilesupto(savetopfile);
1005                 if (str)
1006                         ckfree(str);
1007                 if (ostr)
1008                         ckfree(ostr);
1009                 heredoclist = saveheredoclist;
1010                 handler = savehandler;
1011                 if (exception == EXERROR) {
1012                         startlinno = bq_startlinno;
1013                         synerror("Error in command substitution");
1014                 }
1015                 longjmp(handler->loc, 1);
1016         }
1017         INTOFF;
1018         savelen = out - stackblock();
1019         if (savelen > 0) {
1020                 str = ckmalloc(savelen);
1021                 memcpy(str, stackblock(), savelen);
1022         }
1023         handler = &jmploc;
1024         heredoclist = NULL;
1025         INTON;
1026         if (oldstyle) {
1027                 /* We must read until the closing backquote, giving special
1028                    treatment to some slashes, and then push the string and
1029                    reread it as input, interpreting it normally.  */
1030                 char *oout;
1031                 int c;
1032                 int olen;
1033
1034
1035                 STARTSTACKSTR(oout);
1036                 for (;;) {
1037                         if (needprompt) {
1038                                 setprompt(2);
1039                                 needprompt = 0;
1040                         }
1041                         CHECKSTRSPACE(2, oout);
1042                         switch (c = pgetc()) {
1043                         case '`':
1044                                 goto done;
1045
1046                         case '\\':
1047                                 if ((c = pgetc()) == '\n') {
1048                                         plinno++;
1049                                         if (doprompt)
1050                                                 setprompt(2);
1051                                         else
1052                                                 setprompt(0);
1053                                         /*
1054                                          * If eating a newline, avoid putting
1055                                          * the newline into the new character
1056                                          * stream (via the USTPUTC after the
1057                                          * switch).
1058                                          */
1059                                         continue;
1060                                 }
1061                                 if (c != '\\' && c != '`' && c != '$'
1062                                     && (!dblquote || c != '"'))
1063                                         USTPUTC('\\', oout);
1064                                 break;
1065
1066                         case '\n':
1067                                 plinno++;
1068                                 needprompt = doprompt;
1069                                 break;
1070
1071                         case PEOF:
1072                                 startlinno = plinno;
1073                                 synerror("EOF in backquote substitution");
1074                                 break;
1075
1076                         default:
1077                                 break;
1078                         }
1079                         USTPUTC(c, oout);
1080                 }
1081 done:
1082                 USTPUTC('\0', oout);
1083                 olen = oout - stackblock();
1084                 INTOFF;
1085                 ostr = ckmalloc(olen);
1086                 memcpy(ostr, stackblock(), olen);
1087                 setinputstring(ostr, 1);
1088                 INTON;
1089         }
1090         nlpp = pbqlist;
1091         while (*nlpp)
1092                 nlpp = &(*nlpp)->next;
1093         *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist));
1094         (*nlpp)->next = NULL;
1095
1096         if (oldstyle) {
1097                 saveprompt = doprompt;
1098                 doprompt = 0;
1099         }
1100
1101         n = list(0);
1102
1103         if (oldstyle) {
1104                 if (peektoken() != TEOF)
1105                         synexpect(-1);
1106                 doprompt = saveprompt;
1107         } else
1108                 consumetoken(TRP);
1109
1110         (*nlpp)->n = n;
1111         if (oldstyle) {
1112                 /*
1113                  * Start reading from old file again, ignoring any pushed back
1114                  * tokens left from the backquote parsing
1115                  */
1116                 popfile();
1117                 tokpushback = 0;
1118         }
1119         STARTSTACKSTR(out);
1120         CHECKSTRSPACE(savelen + 1, out);
1121         INTOFF;
1122         if (str) {
1123                 memcpy(out, str, savelen);
1124                 STADJUST(savelen, out);
1125                 ckfree(str);
1126                 str = NULL;
1127         }
1128         if (ostr) {
1129                 ckfree(ostr);
1130                 ostr = NULL;
1131         }
1132         here = saveheredoclist;
1133         if (here != NULL) {
1134                 while (here->next != NULL)
1135                         here = here->next;
1136                 here->next = heredoclist;
1137                 heredoclist = saveheredoclist;
1138         }
1139         handler = savehandler;
1140         INTON;
1141         if (quoted)
1142                 USTPUTC(CTLBACKQ | CTLQUOTE, out);
1143         else
1144                 USTPUTC(CTLBACKQ, out);
1145         return out;
1146 }
1147
1148
1149 /*
1150  * Called to parse a backslash escape sequence inside $'...'.
1151  * The backslash has already been read.
1152  */
1153 static char *
1154 readcstyleesc(char *out)
1155 {
1156         int c, v, i, n;
1157
1158         c = pgetc();
1159         switch (c) {
1160         case '\0':
1161                 synerror("Unterminated quoted string");
1162         case '\n':
1163                 plinno++;
1164                 if (doprompt)
1165                         setprompt(2);
1166                 else
1167                         setprompt(0);
1168                 return out;
1169         case '\\':
1170         case '\'':
1171         case '"':
1172                 v = c;
1173                 break;
1174         case 'a': v = '\a'; break;
1175         case 'b': v = '\b'; break;
1176         case 'e': v = '\033'; break;
1177         case 'f': v = '\f'; break;
1178         case 'n': v = '\n'; break;
1179         case 'r': v = '\r'; break;
1180         case 't': v = '\t'; break;
1181         case 'v': v = '\v'; break;
1182         case 'x':
1183                   v = 0;
1184                   for (;;) {
1185                           c = pgetc();
1186                           if (c >= '0' && c <= '9')
1187                                   v = (v << 4) + c - '0';
1188                           else if (c >= 'A' && c <= 'F')
1189                                   v = (v << 4) + c - 'A' + 10;
1190                           else if (c >= 'a' && c <= 'f')
1191                                   v = (v << 4) + c - 'a' + 10;
1192                           else
1193                                   break;
1194                   }
1195                   pungetc();
1196                   break;
1197         case '0': case '1': case '2': case '3':
1198         case '4': case '5': case '6': case '7':
1199                   v = c - '0';
1200                   c = pgetc();
1201                   if (c >= '0' && c <= '7') {
1202                           v <<= 3;
1203                           v += c - '0';
1204                           c = pgetc();
1205                           if (c >= '0' && c <= '7') {
1206                                   v <<= 3;
1207                                   v += c - '0';
1208                           } else
1209                                   pungetc();
1210                   } else
1211                           pungetc();
1212                   break;
1213         case 'c':
1214                   c = pgetc();
1215                   if (c < 0x3f || c > 0x7a || c == 0x60)
1216                           synerror("Bad escape sequence");
1217                   if (c == '\\' && pgetc() != '\\')
1218                           synerror("Bad escape sequence");
1219                   if (c == '?')
1220                           v = 127;
1221                   else
1222                           v = c & 0x1f;
1223                   break;
1224         case 'u':
1225         case 'U':
1226                   n = c == 'U' ? 8 : 4;
1227                   v = 0;
1228                   for (i = 0; i < n; i++) {
1229                           c = pgetc();
1230                           if (c >= '0' && c <= '9')
1231                                   v = (v << 4) + c - '0';
1232                           else if (c >= 'A' && c <= 'F')
1233                                   v = (v << 4) + c - 'A' + 10;
1234                           else if (c >= 'a' && c <= 'f')
1235                                   v = (v << 4) + c - 'a' + 10;
1236                           else
1237                                   synerror("Bad escape sequence");
1238                   }
1239                   if (v == 0 || (v >= 0xd800 && v <= 0xdfff))
1240                           synerror("Bad escape sequence");
1241                   /* We really need iconv here. */
1242                   if (initial_localeisutf8 && v > 127) {
1243                           CHECKSTRSPACE(4, out);
1244                           /*
1245                            * We cannot use wctomb() as the locale may have
1246                            * changed.
1247                            */
1248                           if (v <= 0x7ff) {
1249                                   USTPUTC(0xc0 | v >> 6, out);
1250                                   USTPUTC(0x80 | (v & 0x3f), out);
1251                                   return out;
1252                           } else if (v <= 0xffff) {
1253                                   USTPUTC(0xe0 | v >> 12, out);
1254                                   USTPUTC(0x80 | ((v >> 6) & 0x3f), out);
1255                                   USTPUTC(0x80 | (v & 0x3f), out);
1256                                   return out;
1257                           } else if (v <= 0x10ffff) {
1258                                   USTPUTC(0xf0 | v >> 18, out);
1259                                   USTPUTC(0x80 | ((v >> 12) & 0x3f), out);
1260                                   USTPUTC(0x80 | ((v >> 6) & 0x3f), out);
1261                                   USTPUTC(0x80 | (v & 0x3f), out);
1262                                   return out;
1263                           }
1264                   }
1265                   if (v > 127)
1266                           v = '?';
1267                   break;
1268         default:
1269                   synerror("Bad escape sequence");
1270         }
1271         v = (char)v;
1272         /*
1273          * We can't handle NUL bytes.
1274          * POSIX says we should skip till the closing quote.
1275          */
1276         if (v == '\0') {
1277                 while ((c = pgetc()) != '\'') {
1278                         if (c == '\\')
1279                                 c = pgetc();
1280                         if (c == PEOF)
1281                                 synerror("Unterminated quoted string");
1282                 }
1283                 pungetc();
1284                 return out;
1285         }
1286         if (SQSYNTAX[v] == CCTL)
1287                 USTPUTC(CTLESC, out);
1288         USTPUTC(v, out);
1289         return out;
1290 }
1291
1292
1293 /*
1294  * If eofmark is NULL, read a word or a redirection symbol.  If eofmark
1295  * is not NULL, read a here document.  In the latter case, eofmark is the
1296  * word which marks the end of the document and striptabs is true if
1297  * leading tabs should be stripped from the document.  The argument firstc
1298  * is the first character of the input token or document.
1299  *
1300  * Because C does not have internal subroutines, I have simulated them
1301  * using goto's to implement the subroutine linkage.  The following macros
1302  * will run code that appears at the end of readtoken1.
1303  */
1304
1305 #define CHECKEND()      {goto checkend; checkend_return:;}
1306 #define PARSEREDIR()    {goto parseredir; parseredir_return:;}
1307 #define PARSESUB()      {goto parsesub; parsesub_return:;}
1308 #define PARSEARITH()    {goto parsearith; parsearith_return:;}
1309
1310 static int
1311 readtoken1(int firstc, char const *initialsyntax, const char *eofmark,
1312     int striptabs)
1313 {
1314         int c = firstc;
1315         char *out;
1316         int len;
1317         char line[EOFMARKLEN + 1];
1318         struct nodelist *bqlist;
1319         int quotef;
1320         int newvarnest;
1321         int level;
1322         int synentry;
1323         struct tokenstate state_static[MAXNEST_static];
1324         int maxnest = MAXNEST_static;
1325         struct tokenstate *state = state_static;
1326         int sqiscstyle = 0;
1327
1328         startlinno = plinno;
1329         quotef = 0;
1330         bqlist = NULL;
1331         newvarnest = 0;
1332         level = 0;
1333         state[level].syntax = initialsyntax;
1334         state[level].parenlevel = 0;
1335         state[level].category = TSTATE_TOP;
1336
1337         STARTSTACKSTR(out);
1338         loop: { /* for each line, until end of word */
1339                 CHECKEND();     /* set c to PEOF if at end of here document */
1340                 for (;;) {      /* until end of line or end of word */
1341                         CHECKSTRSPACE(4, out);  /* permit 4 calls to USTPUTC */
1342
1343                         synentry = state[level].syntax[c];
1344
1345                         switch(synentry) {
1346                         case CNL:       /* '\n' */
1347                                 if (state[level].syntax == BASESYNTAX)
1348                                         goto endword;   /* exit outer loop */
1349                                 USTPUTC(c, out);
1350                                 plinno++;
1351                                 if (doprompt)
1352                                         setprompt(2);
1353                                 else
1354                                         setprompt(0);
1355                                 c = pgetc();
1356                                 goto loop;              /* continue outer loop */
1357                         case CSBACK:
1358                                 if (sqiscstyle) {
1359                                         out = readcstyleesc(out);
1360                                         break;
1361                                 }
1362                                 /* FALLTHROUGH */
1363                         case CWORD:
1364                                 USTPUTC(c, out);
1365                                 break;
1366                         case CCTL:
1367                                 if (eofmark == NULL || initialsyntax != SQSYNTAX)
1368                                         USTPUTC(CTLESC, out);
1369                                 USTPUTC(c, out);
1370                                 break;
1371                         case CBACK:     /* backslash */
1372                                 c = pgetc();
1373                                 if (c == PEOF) {
1374                                         USTPUTC('\\', out);
1375                                         pungetc();
1376                                 } else if (c == '\n') {
1377                                         plinno++;
1378                                         if (doprompt)
1379                                                 setprompt(2);
1380                                         else
1381                                                 setprompt(0);
1382                                 } else {
1383                                         if (state[level].syntax == DQSYNTAX &&
1384                                             c != '\\' && c != '`' && c != '$' &&
1385                                             (c != '"' || (eofmark != NULL &&
1386                                                 newvarnest == 0)) &&
1387                                             (c != '}' || state[level].category != TSTATE_VAR_OLD))
1388                                                 USTPUTC('\\', out);
1389                                         if ((eofmark == NULL ||
1390                                             newvarnest > 0) &&
1391                                             state[level].syntax == BASESYNTAX)
1392                                                 USTPUTC(CTLQUOTEMARK, out);
1393                                         if (SQSYNTAX[c] == CCTL)
1394                                                 USTPUTC(CTLESC, out);
1395                                         USTPUTC(c, out);
1396                                         if ((eofmark == NULL ||
1397                                             newvarnest > 0) &&
1398                                             state[level].syntax == BASESYNTAX &&
1399                                             state[level].category == TSTATE_VAR_OLD)
1400                                                 USTPUTC(CTLQUOTEEND, out);
1401                                         quotef++;
1402                                 }
1403                                 break;
1404                         case CSQUOTE:
1405                                 USTPUTC(CTLQUOTEMARK, out);
1406                                 state[level].syntax = SQSYNTAX;
1407                                 sqiscstyle = 0;
1408                                 break;
1409                         case CDQUOTE:
1410                                 USTPUTC(CTLQUOTEMARK, out);
1411                                 state[level].syntax = DQSYNTAX;
1412                                 break;
1413                         case CENDQUOTE:
1414                                 if (eofmark != NULL && newvarnest == 0)
1415                                         USTPUTC(c, out);
1416                                 else {
1417                                         if (state[level].category == TSTATE_VAR_OLD)
1418                                                 USTPUTC(CTLQUOTEEND, out);
1419                                         state[level].syntax = BASESYNTAX;
1420                                         quotef++;
1421                                 }
1422                                 break;
1423                         case CVAR:      /* '$' */
1424                                 PARSESUB();             /* parse substitution */
1425                                 break;
1426                         case CENDVAR:   /* '}' */
1427                                 if (level > 0 &&
1428                                     ((state[level].category == TSTATE_VAR_OLD &&
1429                                       state[level].syntax ==
1430                                       state[level - 1].syntax) ||
1431                                     (state[level].category == TSTATE_VAR_NEW &&
1432                                      state[level].syntax == BASESYNTAX))) {
1433                                         if (state[level].category == TSTATE_VAR_NEW)
1434                                                 newvarnest--;
1435                                         level--;
1436                                         USTPUTC(CTLENDVAR, out);
1437                                 } else {
1438                                         USTPUTC(c, out);
1439                                 }
1440                                 break;
1441                         case CLP:       /* '(' in arithmetic */
1442                                 state[level].parenlevel++;
1443                                 USTPUTC(c, out);
1444                                 break;
1445                         case CRP:       /* ')' in arithmetic */
1446                                 if (state[level].parenlevel > 0) {
1447                                         USTPUTC(c, out);
1448                                         --state[level].parenlevel;
1449                                 } else {
1450                                         if (pgetc() == ')') {
1451                                                 if (level > 0 &&
1452                                                     state[level].category == TSTATE_ARITH) {
1453                                                         level--;
1454                                                         USTPUTC(CTLENDARI, out);
1455                                                 } else
1456                                                         USTPUTC(')', out);
1457                                         } else {
1458                                                 /*
1459                                                  * unbalanced parens
1460                                                  *  (don't 2nd guess - no error)
1461                                                  */
1462                                                 pungetc();
1463                                                 USTPUTC(')', out);
1464                                         }
1465                                 }
1466                                 break;
1467                         case CBQUOTE:   /* '`' */
1468                                 out = parsebackq(out, &bqlist, 1,
1469                                     state[level].syntax == DQSYNTAX &&
1470                                     (eofmark == NULL || newvarnest > 0),
1471                                     state[level].syntax == DQSYNTAX || state[level].syntax == ARISYNTAX);
1472                                 break;
1473                         case CEOF:
1474                                 goto endword;           /* exit outer loop */
1475                         case CIGN:
1476                                 break;
1477                         default:
1478                                 if (level == 0)
1479                                         goto endword;   /* exit outer loop */
1480                                 USTPUTC(c, out);
1481                         }
1482                         c = pgetc_macro();
1483                 }
1484         }
1485 endword:
1486         if (state[level].syntax == ARISYNTAX)
1487                 synerror("Missing '))'");
1488         if (state[level].syntax != BASESYNTAX && eofmark == NULL)
1489                 synerror("Unterminated quoted string");
1490         if (state[level].category == TSTATE_VAR_OLD ||
1491             state[level].category == TSTATE_VAR_NEW) {
1492                 startlinno = plinno;
1493                 synerror("Missing '}'");
1494         }
1495         if (state != state_static)
1496                 parser_temp_free_upto(state);
1497         USTPUTC('\0', out);
1498         len = out - stackblock();
1499         out = stackblock();
1500         if (eofmark == NULL) {
1501                 if ((c == '>' || c == '<')
1502                  && quotef == 0
1503                  && len <= 2
1504                  && (*out == '\0' || is_digit(*out))) {
1505                         PARSEREDIR();
1506                         return lasttoken = TREDIR;
1507                 } else {
1508                         pungetc();
1509                 }
1510         }
1511         quoteflag = quotef;
1512         backquotelist = bqlist;
1513         grabstackblock(len);
1514         wordtext = out;
1515         return lasttoken = TWORD;
1516 /* end of readtoken routine */
1517
1518
1519 /*
1520  * Check to see whether we are at the end of the here document.  When this
1521  * is called, c is set to the first character of the next input line.  If
1522  * we are at the end of the here document, this routine sets the c to PEOF.
1523  */
1524
1525 checkend: {
1526         if (eofmark) {
1527                 if (striptabs) {
1528                         while (c == '\t')
1529                                 c = pgetc();
1530                 }
1531                 if (c == *eofmark) {
1532                         if (pfgets(line, sizeof line) != NULL) {
1533                                 const char *p, *q;
1534
1535                                 p = line;
1536                                 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++);
1537                                 if ((*p == '\0' || *p == '\n') && *q == '\0') {
1538                                         c = PEOF;
1539                                         if (*p == '\n') {
1540                                                 plinno++;
1541                                                 needprompt = doprompt;
1542                                         }
1543                                 } else {
1544                                         pushstring(line, strlen(line), NULL);
1545                                 }
1546                         }
1547                 }
1548         }
1549         goto checkend_return;
1550 }
1551
1552
1553 /*
1554  * Parse a redirection operator.  The variable "out" points to a string
1555  * specifying the fd to be redirected.  The variable "c" contains the
1556  * first character of the redirection operator.
1557  */
1558
1559 parseredir: {
1560         char fd = *out;
1561         union node *np;
1562
1563         np = (union node *)stalloc(sizeof (struct nfile));
1564         if (c == '>') {
1565                 np->nfile.fd = 1;
1566                 c = pgetc();
1567                 if (c == '>')
1568                         np->type = NAPPEND;
1569                 else if (c == '&')
1570                         np->type = NTOFD;
1571                 else if (c == '|')
1572                         np->type = NCLOBBER;
1573                 else {
1574                         np->type = NTO;
1575                         pungetc();
1576                 }
1577         } else {        /* c == '<' */
1578                 np->nfile.fd = 0;
1579                 c = pgetc();
1580                 if (c == '<') {
1581                         if (sizeof (struct nfile) != sizeof (struct nhere)) {
1582                                 np = (union node *)stalloc(sizeof (struct nhere));
1583                                 np->nfile.fd = 0;
1584                         }
1585                         np->type = NHERE;
1586                         heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc));
1587                         heredoc->here = np;
1588                         if ((c = pgetc()) == '-') {
1589                                 heredoc->striptabs = 1;
1590                         } else {
1591                                 heredoc->striptabs = 0;
1592                                 pungetc();
1593                         }
1594                 } else if (c == '&')
1595                         np->type = NFROMFD;
1596                 else if (c == '>')
1597                         np->type = NFROMTO;
1598                 else {
1599                         np->type = NFROM;
1600                         pungetc();
1601                 }
1602         }
1603         if (fd != '\0')
1604                 np->nfile.fd = digit_val(fd);
1605         redirnode = np;
1606         goto parseredir_return;
1607 }
1608
1609
1610 /*
1611  * Parse a substitution.  At this point, we have read the dollar sign
1612  * and nothing else.
1613  */
1614
1615 parsesub: {
1616         char buf[10];
1617         int subtype;
1618         int typeloc;
1619         int flags;
1620         char *p;
1621         static const char types[] = "}-+?=";
1622         int bracketed_name = 0; /* used to handle ${[0-9]*} variables */
1623         int linno;
1624         int length;
1625         int c1;
1626
1627         c = pgetc();
1628         if (c == '(') { /* $(command) or $((arith)) */
1629                 if (pgetc() == '(') {
1630                         PARSEARITH();
1631                 } else {
1632                         pungetc();
1633                         out = parsebackq(out, &bqlist, 0,
1634                             state[level].syntax == DQSYNTAX &&
1635                             (eofmark == NULL || newvarnest > 0),
1636                             state[level].syntax == DQSYNTAX ||
1637                             state[level].syntax == ARISYNTAX);
1638                 }
1639         } else if (c == '{' || is_name(c) || is_special(c)) {
1640                 USTPUTC(CTLVAR, out);
1641                 typeloc = out - stackblock();
1642                 USTPUTC(VSNORMAL, out);
1643                 subtype = VSNORMAL;
1644                 flags = 0;
1645                 if (c == '{') {
1646                         bracketed_name = 1;
1647                         c = pgetc();
1648                         subtype = 0;
1649                 }
1650 varname:
1651                 if (!is_eof(c) && is_name(c)) {
1652                         length = 0;
1653                         do {
1654                                 STPUTC(c, out);
1655                                 c = pgetc();
1656                                 length++;
1657                         } while (!is_eof(c) && is_in_name(c));
1658                         if (length == 6 &&
1659                             strncmp(out - length, "LINENO", length) == 0) {
1660                                 /* Replace the variable name with the
1661                                  * current line number. */
1662                                 linno = plinno;
1663                                 if (funclinno != 0)
1664                                         linno -= funclinno - 1;
1665                                 snprintf(buf, sizeof(buf), "%d", linno);
1666                                 STADJUST(-6, out);
1667                                 STPUTS(buf, out);
1668                                 flags |= VSLINENO;
1669                         }
1670                 } else if (is_digit(c)) {
1671                         if (bracketed_name) {
1672                                 do {
1673                                         STPUTC(c, out);
1674                                         c = pgetc();
1675                                 } while (is_digit(c));
1676                         } else {
1677                                 STPUTC(c, out);
1678                                 c = pgetc();
1679                         }
1680                 } else if (is_special(c)) {
1681                         c1 = c;
1682                         c = pgetc();
1683                         if (subtype == 0 && c1 == '#') {
1684                                 subtype = VSLENGTH;
1685                                 if (strchr(types, c) == NULL && c != ':' &&
1686                                     c != '#' && c != '%')
1687                                         goto varname;
1688                                 c1 = c;
1689                                 c = pgetc();
1690                                 if (c1 != '}' && c == '}') {
1691                                         pungetc();
1692                                         c = c1;
1693                                         goto varname;
1694                                 }
1695                                 pungetc();
1696                                 c = c1;
1697                                 c1 = '#';
1698                                 subtype = 0;
1699                         }
1700                         USTPUTC(c1, out);
1701                 } else {
1702                         subtype = VSERROR;
1703                         if (c == '}')
1704                                 pungetc();
1705                         else if (c == '\n' || c == PEOF)
1706                                 synerror("Unexpected end of line in substitution");
1707                         else if (BASESYNTAX[c] != CCTL)
1708                                 USTPUTC(c, out);
1709                 }
1710                 if (subtype == 0) {
1711                         switch (c) {
1712                         case ':':
1713                                 flags |= VSNUL;
1714                                 c = pgetc();
1715                                 /*FALLTHROUGH*/
1716                         default:
1717                                 p = strchr(types, c);
1718                                 if (p == NULL) {
1719                                         if (c == '\n' || c == PEOF)
1720                                                 synerror("Unexpected end of line in substitution");
1721                                         if (flags == VSNUL)
1722                                                 STPUTC(':', out);
1723                                         if (BASESYNTAX[c] != CCTL)
1724                                                 STPUTC(c, out);
1725                                         subtype = VSERROR;
1726                                 } else
1727                                         subtype = p - types + VSNORMAL;
1728                                 break;
1729                         case '%':
1730                         case '#':
1731                                 {
1732                                         int cc = c;
1733                                         subtype = c == '#' ? VSTRIMLEFT :
1734                                                              VSTRIMRIGHT;
1735                                         c = pgetc();
1736                                         if (c == cc)
1737                                                 subtype++;
1738                                         else
1739                                                 pungetc();
1740                                         break;
1741                                 }
1742                         }
1743                 } else if (subtype != VSERROR) {
1744                         if (subtype == VSLENGTH && c != '}')
1745                                 subtype = VSERROR;
1746                         pungetc();
1747                 }
1748                 STPUTC('=', out);
1749                 if (state[level].syntax == DQSYNTAX ||
1750                     state[level].syntax == ARISYNTAX)
1751                         flags |= VSQUOTE;
1752                 *(stackblock() + typeloc) = subtype | flags;
1753                 if (subtype != VSNORMAL) {
1754                         if (level + 1 >= maxnest) {
1755                                 maxnest *= 2;
1756                                 if (state == state_static) {
1757                                         state = parser_temp_alloc(
1758                                             maxnest * sizeof(*state));
1759                                         memcpy(state, state_static,
1760                                             MAXNEST_static * sizeof(*state));
1761                                 } else
1762                                         state = parser_temp_realloc(state,
1763                                             maxnest * sizeof(*state));
1764                         }
1765                         level++;
1766                         state[level].parenlevel = 0;
1767                         if (subtype == VSMINUS || subtype == VSPLUS ||
1768                             subtype == VSQUESTION || subtype == VSASSIGN) {
1769                                 /*
1770                                  * For operators that were in the Bourne shell,
1771                                  * inherit the double-quote state.
1772                                  */
1773                                 state[level].syntax = state[level - 1].syntax;
1774                                 state[level].category = TSTATE_VAR_OLD;
1775                         } else {
1776                                 /*
1777                                  * The other operators take a pattern,
1778                                  * so go to BASESYNTAX.
1779                                  * Also, ' and " are now special, even
1780                                  * in here documents.
1781                                  */
1782                                 state[level].syntax = BASESYNTAX;
1783                                 state[level].category = TSTATE_VAR_NEW;
1784                                 newvarnest++;
1785                         }
1786                 }
1787         } else if (c == '\'' && state[level].syntax == BASESYNTAX) {
1788                 /* $'cstylequotes' */
1789                 USTPUTC(CTLQUOTEMARK, out);
1790                 state[level].syntax = SQSYNTAX;
1791                 sqiscstyle = 1;
1792         } else {
1793                 USTPUTC('$', out);
1794                 pungetc();
1795         }
1796         goto parsesub_return;
1797 }
1798
1799
1800 /*
1801  * Parse an arithmetic expansion (indicate start of one and set state)
1802  */
1803 parsearith: {
1804
1805         if (level + 1 >= maxnest) {
1806                 maxnest *= 2;
1807                 if (state == state_static) {
1808                         state = parser_temp_alloc(
1809                             maxnest * sizeof(*state));
1810                         memcpy(state, state_static,
1811                             MAXNEST_static * sizeof(*state));
1812                 } else
1813                         state = parser_temp_realloc(state,
1814                             maxnest * sizeof(*state));
1815         }
1816         level++;
1817         state[level].syntax = ARISYNTAX;
1818         state[level].parenlevel = 0;
1819         state[level].category = TSTATE_ARITH;
1820         USTPUTC(CTLARI, out);
1821         if (state[level - 1].syntax == DQSYNTAX)
1822                 USTPUTC('"',out);
1823         else
1824                 USTPUTC(' ',out);
1825         goto parsearith_return;
1826 }
1827
1828 } /* end of readtoken */
1829
1830
1831 /*
1832  * Returns true if the text contains nothing to expand (no dollar signs
1833  * or backquotes).
1834  */
1835
1836 static int
1837 noexpand(char *text)
1838 {
1839         char *p;
1840         char c;
1841
1842         p = text;
1843         while ((c = *p++) != '\0') {
1844                 if ( c == CTLQUOTEMARK)
1845                         continue;
1846                 if (c == CTLESC)
1847                         p++;
1848                 else if (BASESYNTAX[(int)c] == CCTL)
1849                         return 0;
1850         }
1851         return 1;
1852 }
1853
1854
1855 /*
1856  * Return true if the argument is a legal variable name (a letter or
1857  * underscore followed by zero or more letters, underscores, and digits).
1858  */
1859
1860 int
1861 goodname(const char *name)
1862 {
1863         const char *p;
1864
1865         p = name;
1866         if (! is_name(*p))
1867                 return 0;
1868         while (*++p) {
1869                 if (! is_in_name(*p))
1870                         return 0;
1871         }
1872         return 1;
1873 }
1874
1875
1876 int
1877 isassignment(const char *p)
1878 {
1879         if (!is_name(*p))
1880                 return 0;
1881         p++;
1882         for (;;) {
1883                 if (*p == '=')
1884                         return 1;
1885                 else if (!is_in_name(*p))
1886                         return 0;
1887                 p++;
1888         }
1889 }
1890
1891
1892 static void
1893 consumetoken(int token)
1894 {
1895         if (readtoken() != token)
1896                 synexpect(token);
1897 }
1898
1899
1900 /*
1901  * Called when an unexpected token is read during the parse.  The argument
1902  * is the token that is expected, or -1 if more than one type of token can
1903  * occur at this point.
1904  */
1905
1906 static void
1907 synexpect(int token)
1908 {
1909         char msg[64];
1910
1911         if (token >= 0) {
1912                 fmtstr(msg, 64, "%s unexpected (expecting %s)",
1913                         tokname[lasttoken], tokname[token]);
1914         } else {
1915                 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]);
1916         }
1917         synerror(msg);
1918 }
1919
1920
1921 static void
1922 synerror(const char *msg)
1923 {
1924         if (commandname)
1925                 outfmt(out2, "%s: %d: ", commandname, startlinno);
1926         outfmt(out2, "Syntax error: %s\n", msg);
1927         error((char *)NULL);
1928 }
1929
1930 static void
1931 setprompt(int which)
1932 {
1933         whichprompt = which;
1934
1935 #ifndef NO_HISTORY
1936         if (!el)
1937 #endif
1938         {
1939                 out2str(getprompt(NULL));
1940                 flushout(out2);
1941         }
1942 }
1943
1944 /*
1945  * called by editline -- any expansions to the prompt
1946  *    should be added here.
1947  */
1948 char *
1949 getprompt(void *unused __unused)
1950 {
1951         static char ps[PROMPTLEN];
1952         char *fmt;
1953         const char *pwd;
1954         int i, trim;
1955         static char internal_error[] = "??";
1956
1957         /*
1958          * Select prompt format.
1959          */
1960         switch (whichprompt) {
1961         case 0:
1962                 fmt = nullstr;
1963                 break;
1964         case 1:
1965                 fmt = ps1val();
1966                 break;
1967         case 2:
1968                 fmt = ps2val();
1969                 break;
1970         default:
1971                 return internal_error;
1972         }
1973
1974         /*
1975          * Format prompt string.
1976          */
1977         for (i = 0; (i < PROMPTLEN - 1) && (*fmt != '\0'); i++, fmt++)
1978                 if (*fmt == '\\')
1979                         switch (*++fmt) {
1980
1981                                 /*
1982                                  * Hostname.
1983                                  *
1984                                  * \h specifies just the local hostname,
1985                                  * \H specifies fully-qualified hostname.
1986                                  */
1987                         case 'h':
1988                         case 'H':
1989                                 ps[i] = '\0';
1990                                 gethostname(&ps[i], PROMPTLEN - i - 1);
1991                                 ps[PROMPTLEN - 1] = '\0';
1992                                 /* Skip to end of hostname. */
1993                                 trim = (*fmt == 'h') ? '.' : '\0';
1994                                 while ((ps[i] != '\0') && (ps[i] != trim))
1995                                         i++;
1996                                 --i;
1997                                 break;
1998
1999                                 /*
2000                                  * Working directory.
2001                                  *
2002                                  * \W specifies just the final component,
2003                                  * \w specifies the entire path.
2004                                  */
2005                         case 'W':
2006                         case 'w':
2007                                 pwd = lookupvar("PWD");
2008                                 if (pwd == NULL || *pwd == '\0')
2009                                         pwd = "?";
2010                                 if (*fmt == 'W' &&
2011                                     *pwd == '/' && pwd[1] != '\0')
2012                                         strlcpy(&ps[i], strrchr(pwd, '/') + 1,
2013                                             PROMPTLEN - i);
2014                                 else
2015                                         strlcpy(&ps[i], pwd, PROMPTLEN - i);
2016                                 /* Skip to end of path. */
2017                                 while (ps[i + 1] != '\0')
2018                                         i++;
2019                                 break;
2020
2021                                 /*
2022                                  * Superuser status.
2023                                  *
2024                                  * '$' for normal users, '#' for root.
2025                                  */
2026                         case '$':
2027                                 ps[i] = (geteuid() != 0) ? '$' : '#';
2028                                 break;
2029
2030                                 /*
2031                                  * A literal \.
2032                                  */
2033                         case '\\':
2034                                 ps[i] = '\\';
2035                                 break;
2036
2037                                 /*
2038                                  * Emit unrecognized formats verbatim.
2039                                  */
2040                         default:
2041                                 ps[i] = '\\';
2042                                 if (i < PROMPTLEN - 2)
2043                                         ps[++i] = *fmt;
2044                                 break;
2045                         }
2046                 else
2047                         ps[i] = *fmt;
2048         ps[i] = '\0';
2049         return (ps);
2050 }
2051
2052
2053 const char *
2054 expandstr(const char *ps)
2055 {
2056         union node n;
2057         struct jmploc jmploc;
2058         struct jmploc *const savehandler = handler;
2059         const int saveprompt = doprompt;
2060         struct parsefile *const savetopfile = getcurrentfile();
2061         struct parser_temp *const saveparser_temp = parser_temp;
2062         const char *result = NULL;
2063
2064         if (!setjmp(jmploc.loc)) {
2065                 handler = &jmploc;
2066                 parser_temp = NULL;
2067                 setinputstring(ps, 1);
2068                 doprompt = 0;
2069                 readtoken1(pgetc(), DQSYNTAX, "\n\n", 0);
2070                 if (backquotelist != NULL)
2071                         error("Command substitution not allowed here");
2072
2073                 n.narg.type = NARG;
2074                 n.narg.next = NULL;
2075                 n.narg.text = wordtext;
2076                 n.narg.backquote = backquotelist;
2077
2078                 expandarg(&n, NULL, 0);
2079                 result = stackblock();
2080                 INTOFF;
2081         }
2082         handler = savehandler;
2083         doprompt = saveprompt;
2084         popfilesupto(savetopfile);
2085         if (parser_temp != saveparser_temp) {
2086                 parser_temp_free_all();
2087                 parser_temp = saveparser_temp;
2088         }
2089         if (result != NULL) {
2090                 INTON;
2091         } else if (exception == EXINT)
2092                 raise(SIGINT);
2093         return result;
2094 }