bin/sh/parser.c

   1 /*-
   2  * SPDX-License-Identifier: BSD-3-Clause
   3  *
   4  * Copyright (c) 1991, 1993
   5  *      The Regents of the University of California.  All rights reserved.
   6  *
   7  * This code is derived from software contributed to Berkeley by
   8  * Kenneth Almquist.
   9  *
  10  * Redistribution and use in source and binary forms, with or without
  11  * modification, are permitted provided that the following conditions
  12  * are met:
  13  * 1. Redistributions of source code must retain the above copyright
  14  *    notice, this list of conditions and the following disclaimer.
  15  * 2. Redistributions in binary form must reproduce the above copyright
  16  *    notice, this list of conditions and the following disclaimer in the
  17  *    documentation and/or other materials provided with the distribution.
  18  * 3. Neither the name of the University nor the names of its contributors
  19  *    may be used to endorse or promote products derived from this software
  20  *    without specific prior written permission.
  21  *
  22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  32  * SUCH DAMAGE.
  33  */
  34
  35 #ifndef lint
  36 #if 0
  37 static char sccsid[] = "@(#)parser.c    8.7 (Berkeley) 5/16/95";
  38 #endif
  39 #endif /* not lint */
  40 #include <sys/cdefs.h>
  41 __FBSDID("$FreeBSD$");
  42
  43 #include <stdlib.h>
  44 #include <unistd.h>
  45 #include <stdio.h>
  46
  47 #include "shell.h"
  48 #include "parser.h"
  49 #include "nodes.h"
  50 #include "expand.h"     /* defines rmescapes() */
  51 #include "syntax.h"
  52 #include "options.h"
  53 #include "input.h"
  54 #include "output.h"
  55 #include "var.h"
  56 #include "error.h"
  57 #include "memalloc.h"
  58 #include "mystring.h"
  59 #include "alias.h"
  60 #include "show.h"
  61 #include "eval.h"
  62 #include "exec.h"       /* to check for special builtins */
  63 #ifndef NO_HISTORY
  64 #include "myhistedit.h"
  65 #endif
  66
  67 /*
  68  * Shell command parser.
  69  */
  70
  71 #define PROMPTLEN       128
  72
  73 /* values of checkkwd variable */
  74 #define CHKALIAS        0x1
  75 #define CHKKWD          0x2
  76 #define CHKNL           0x4
  77
  78 /* values returned by readtoken */
  79 #include "token.h"
  80
  81
  82
  83 struct heredoc {
  84         struct heredoc *next;   /* next here document in list */
  85         union node *here;               /* redirection node */
  86         char *eofmark;          /* string indicating end of input */
  87         int striptabs;          /* if set, strip leading tabs */
  88 };
  89
  90 struct parser_temp {
  91         struct parser_temp *next;
  92         void *data;
  93 };
  94
  95
  96 static struct heredoc *heredoclist;     /* list of here documents to read */
  97 static int doprompt;            /* if set, prompt the user */
  98 static int needprompt;          /* true if interactive and at start of line */
  99 static int lasttoken;           /* last token read */
 100 static int tokpushback;         /* last token pushed back */
 101 static char *wordtext;          /* text of last word returned by readtoken */
 102 static int checkkwd;
 103 static struct nodelist *backquotelist;
 104 static union node *redirnode;
 105 static struct heredoc *heredoc;
 106 static int quoteflag;           /* set if (part of) last token was quoted */
 107 static int startlinno;          /* line # where last token started */
 108 static int funclinno;           /* line # where the current function started */
 109 static struct parser_temp *parser_temp;
 110
 111 #define NOEOFMARK ((const char *)&heredoclist)
 112
 113
 114 static union node *list(int);
 115 static union node *andor(void);
 116 static union node *pipeline(void);
 117 static union node *command(void);
 118 static union node *simplecmd(union node **, union node *);
 119 static union node *makename(void);
 120 static union node *makebinary(int type, union node *n1, union node *n2);
 121 static void parsefname(void);
 122 static void parseheredoc(void);
 123 static int peektoken(void);
 124 static int readtoken(void);
 125 static int xxreadtoken(void);
 126 static int readtoken1(int, const char *, const char *, int);
 127 static int noexpand(char *);
 128 static void consumetoken(int);
 129 static void synexpect(int) __dead2;
 130 static void synerror(const char *) __dead2;
 131 static void setprompt(int);
 132 static int pgetc_linecont(void);
 133
 134
 135 static void *
 136 parser_temp_alloc(size_t len)
 137 {
 138         struct parser_temp *t;
 139
 140         INTOFF;
 141         t = ckmalloc(sizeof(*t));
 142         t->data = NULL;
 143         t->next = parser_temp;
 144         parser_temp = t;
 145         t->data = ckmalloc(len);
 146         INTON;
 147         return t->data;
 148 }
 149
 150
 151 static void *
 152 parser_temp_realloc(void *ptr, size_t len)
 153 {
 154         struct parser_temp *t;
 155
 156         INTOFF;
 157         t = parser_temp;
 158         if (ptr != t->data)
 159                 error("bug: parser_temp_realloc misused");
 160         t->data = ckrealloc(t->data, len);
 161         INTON;
 162         return t->data;
 163 }
 164
 165
 166 static void
 167 parser_temp_free_upto(void *ptr)
 168 {
 169         struct parser_temp *t;
 170         int done = 0;
 171
 172         INTOFF;
 173         while (parser_temp != NULL && !done) {
 174                 t = parser_temp;
 175                 parser_temp = t->next;
 176                 done = t->data == ptr;
 177                 ckfree(t->data);
 178                 ckfree(t);
 179         }
 180         INTON;
 181         if (!done)
 182                 error("bug: parser_temp_free_upto misused");
 183 }
 184
 185
 186 static void
 187 parser_temp_free_all(void)
 188 {
 189         struct parser_temp *t;
 190
 191         INTOFF;
 192         while (parser_temp != NULL) {
 193                 t = parser_temp;
 194                 parser_temp = t->next;
 195                 ckfree(t->data);
 196                 ckfree(t);
 197         }
 198         INTON;
 199 }
 200
 201
 202 /*
 203  * Read and parse a command.  Returns NEOF on end of file.  (NULL is a
 204  * valid parse tree indicating a blank line.)
 205  */
 206
 207 union node *
 208 parsecmd(int interact)
 209 {
 210         int t;
 211
 212         /* This assumes the parser is not re-entered,
 213          * which could happen if we add command substitution on PS1/PS2.
 214          */
 215         parser_temp_free_all();
 216         heredoclist = NULL;
 217
 218         tokpushback = 0;
 219         checkkwd = 0;
 220         doprompt = interact;
 221         if (doprompt)
 222                 setprompt(1);
 223         else
 224                 setprompt(0);
 225         needprompt = 0;
 226         t = readtoken();
 227         if (t == TEOF)
 228                 return NEOF;
 229         if (t == TNL)
 230                 return NULL;
 231         tokpushback++;
 232         return list(1);
 233 }
 234
 235
 236 /*
 237  * Read and parse words for wordexp.
 238  * Returns a list of NARG nodes; NULL if there are no words.
 239  */
 240 union node *
 241 parsewordexp(void)
 242 {
 243         union node *n, *first = NULL, **pnext;
 244         int t;
 245
 246         /* This assumes the parser is not re-entered,
 247          * which could happen if we add command substitution on PS1/PS2.
 248          */
 249         parser_temp_free_all();
 250         heredoclist = NULL;
 251
 252         tokpushback = 0;
 253         checkkwd = 0;
 254         doprompt = 0;
 255         setprompt(0);
 256         needprompt = 0;
 257         pnext = &first;
 258         while ((t = readtoken()) != TEOF) {
 259                 if (t != TWORD)
 260                         synexpect(TWORD);
 261                 n = makename();
 262                 *pnext = n;
 263                 pnext = &n->narg.next;
 264         }
 265         return first;
 266 }
 267
 268
 269 static union node *
 270 list(int nlflag)
 271 {
 272         union node *ntop, *n1, *n2, *n3;
 273         int tok;
 274
 275         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 276         if (!nlflag && tokendlist[peektoken()])
 277                 return NULL;
 278         ntop = n1 = NULL;
 279         for (;;) {
 280                 n2 = andor();
 281                 tok = readtoken();
 282                 if (tok == TBACKGND) {
 283                         if (n2 != NULL && n2->type == NPIPE) {
 284                                 n2->npipe.backgnd = 1;
 285                         } else if (n2 != NULL && n2->type == NREDIR) {
 286                                 n2->type = NBACKGND;
 287                         } else {
 288                                 n3 = (union node *)stalloc(sizeof (struct nredir));
 289                                 n3->type = NBACKGND;
 290                                 n3->nredir.n = n2;
 291                                 n3->nredir.redirect = NULL;
 292                                 n2 = n3;
 293                         }
 294                 }
 295                 if (ntop == NULL)
 296                         ntop = n2;
 297                 else if (n1 == NULL) {
 298                         n1 = makebinary(NSEMI, ntop, n2);
 299                         ntop = n1;
 300                 }
 301                 else {
 302                         n3 = makebinary(NSEMI, n1->nbinary.ch2, n2);
 303                         n1->nbinary.ch2 = n3;
 304                         n1 = n3;
 305                 }
 306                 switch (tok) {
 307                 case TBACKGND:
 308                 case TSEMI:
 309                         tok = readtoken();
 310                         /* FALLTHROUGH */
 311                 case TNL:
 312                         if (tok == TNL) {
 313                                 parseheredoc();
 314                                 if (nlflag)
 315                                         return ntop;
 316                         } else if (tok == TEOF && nlflag) {
 317                                 parseheredoc();
 318                                 return ntop;
 319                         } else {
 320                                 tokpushback++;
 321                         }
 322                         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 323                         if (!nlflag && tokendlist[peektoken()])
 324                                 return ntop;
 325                         break;
 326                 case TEOF:
 327                         if (heredoclist)
 328                                 parseheredoc();
 329                         else
 330                                 pungetc();              /* push back EOF on input */
 331                         return ntop;
 332                 default:
 333                         if (nlflag)
 334                                 synexpect(-1);
 335                         tokpushback++;
 336                         return ntop;
 337                 }
 338         }
 339 }
 340
 341
 342
 343 static union node *
 344 andor(void)
 345 {
 346         union node *n;
 347         int t;
 348
 349         n = pipeline();
 350         for (;;) {
 351                 if ((t = readtoken()) == TAND) {
 352                         t = NAND;
 353                 } else if (t == TOR) {
 354                         t = NOR;
 355                 } else {
 356                         tokpushback++;
 357                         return n;
 358                 }
 359                 n = makebinary(t, n, pipeline());
 360         }
 361 }
 362
 363
 364
 365 static union node *
 366 pipeline(void)
 367 {
 368         union node *n1, *n2, *pipenode;
 369         struct nodelist *lp, *prev;
 370         int negate, t;
 371
 372         negate = 0;
 373         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 374         TRACE(("pipeline: entered\n"));
 375         while (readtoken() == TNOT)
 376                 negate = !negate;
 377         tokpushback++;
 378         n1 = command();
 379         if (readtoken() == TPIPE) {
 380                 pipenode = (union node *)stalloc(sizeof (struct npipe));
 381                 pipenode->type = NPIPE;
 382                 pipenode->npipe.backgnd = 0;
 383                 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
 384                 pipenode->npipe.cmdlist = lp;
 385                 lp->n = n1;
 386                 do {
 387                         prev = lp;
 388                         lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
 389                         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 390                         t = readtoken();
 391                         tokpushback++;
 392                         if (t == TNOT)
 393                                 lp->n = pipeline();
 394                         else
 395                                 lp->n = command();
 396                         prev->next = lp;
 397                 } while (readtoken() == TPIPE);
 398                 lp->next = NULL;
 399                 n1 = pipenode;
 400         }
 401         tokpushback++;
 402         if (negate) {
 403                 n2 = (union node *)stalloc(sizeof (struct nnot));
 404                 n2->type = NNOT;
 405                 n2->nnot.com = n1;
 406                 return n2;
 407         } else
 408                 return n1;
 409 }
 410
 411
 412
 413 static union node *
 414 command(void)
 415 {
 416         union node *n1, *n2;
 417         union node *ap, **app;
 418         union node *cp, **cpp;
 419         union node *redir, **rpp;
 420         int t;
 421         int is_subshell;
 422
 423         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 424         is_subshell = 0;
 425         redir = NULL;
 426         n1 = NULL;
 427         rpp = &redir;
 428
 429         /* Check for redirection which may precede command */
 430         while (readtoken() == TREDIR) {
 431                 *rpp = n2 = redirnode;
 432                 rpp = &n2->nfile.next;
 433                 parsefname();
 434         }
 435         tokpushback++;
 436
 437         switch (readtoken()) {
 438         case TIF:
 439                 n1 = (union node *)stalloc(sizeof (struct nif));
 440                 n1->type = NIF;
 441                 if ((n1->nif.test = list(0)) == NULL)
 442                         synexpect(-1);
 443                 consumetoken(TTHEN);
 444                 n1->nif.ifpart = list(0);
 445                 n2 = n1;
 446                 while (readtoken() == TELIF) {
 447                         n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif));
 448                         n2 = n2->nif.elsepart;
 449                         n2->type = NIF;
 450                         if ((n2->nif.test = list(0)) == NULL)
 451                                 synexpect(-1);
 452                         consumetoken(TTHEN);
 453                         n2->nif.ifpart = list(0);
 454                 }
 455                 if (lasttoken == TELSE)
 456                         n2->nif.elsepart = list(0);
 457                 else {
 458                         n2->nif.elsepart = NULL;
 459                         tokpushback++;
 460                 }
 461                 consumetoken(TFI);
 462                 checkkwd = CHKKWD | CHKALIAS;
 463                 break;
 464         case TWHILE:
 465         case TUNTIL:
 466                 t = lasttoken;
 467                 if ((n1 = list(0)) == NULL)
 468                         synexpect(-1);
 469                 consumetoken(TDO);
 470                 n1 = makebinary((t == TWHILE)? NWHILE : NUNTIL, n1, list(0));
 471                 consumetoken(TDONE);
 472                 checkkwd = CHKKWD | CHKALIAS;
 473                 break;
 474         case TFOR:
 475                 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext))
 476                         synerror("Bad for loop variable");
 477                 n1 = (union node *)stalloc(sizeof (struct nfor));
 478                 n1->type = NFOR;
 479                 n1->nfor.var = wordtext;
 480                 while (readtoken() == TNL)
 481                         ;
 482                 if (lasttoken == TWORD && ! quoteflag && equal(wordtext, "in")) {
 483                         app = &ap;
 484                         while (readtoken() == TWORD) {
 485                                 n2 = makename();
 486                                 *app = n2;
 487                                 app = &n2->narg.next;
 488                         }
 489                         *app = NULL;
 490                         n1->nfor.args = ap;
 491                         if (lasttoken != TNL && lasttoken != TSEMI)
 492                                 synexpect(-1);
 493                 } else {
 494                         static char argvars[5] = {
 495                                 CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'
 496                         };
 497                         n2 = (union node *)stalloc(sizeof (struct narg));
 498                         n2->type = NARG;
 499                         n2->narg.text = argvars;
 500                         n2->narg.backquote = NULL;
 501                         n2->narg.next = NULL;
 502                         n1->nfor.args = n2;
 503                         /*
 504                          * Newline or semicolon here is optional (but note
 505                          * that the original Bourne shell only allowed NL).
 506                          */
 507                         if (lasttoken != TNL && lasttoken != TSEMI)
 508                                 tokpushback++;
 509                 }
 510                 checkkwd = CHKNL | CHKKWD | CHKALIAS;
 511                 if ((t = readtoken()) == TDO)
 512                         t = TDONE;
 513                 else if (t == TBEGIN)
 514                         t = TEND;
 515                 else
 516                         synexpect(-1);
 517                 n1->nfor.body = list(0);
 518                 consumetoken(t);
 519                 checkkwd = CHKKWD | CHKALIAS;
 520                 break;
 521         case TCASE:
 522                 n1 = (union node *)stalloc(sizeof (struct ncase));
 523                 n1->type = NCASE;
 524                 consumetoken(TWORD);
 525                 n1->ncase.expr = makename();
 526                 while (readtoken() == TNL);
 527                 if (lasttoken != TWORD || ! equal(wordtext, "in"))
 528                         synerror("expecting \"in\"");
 529                 cpp = &n1->ncase.cases;
 530                 checkkwd = CHKNL | CHKKWD, readtoken();
 531                 while (lasttoken != TESAC) {
 532                         *cpp = cp = (union node *)stalloc(sizeof (struct nclist));
 533                         cp->type = NCLIST;
 534                         app = &cp->nclist.pattern;
 535                         if (lasttoken == TLP)
 536                                 readtoken();
 537                         for (;;) {
 538                                 *app = ap = makename();
 539                                 checkkwd = CHKNL | CHKKWD;
 540                                 if (readtoken() != TPIPE)
 541                                         break;
 542                                 app = &ap->narg.next;
 543                                 readtoken();
 544                         }
 545                         ap->narg.next = NULL;
 546                         if (lasttoken != TRP)
 547                                 synexpect(TRP);
 548                         cp->nclist.body = list(0);
 549
 550                         checkkwd = CHKNL | CHKKWD | CHKALIAS;
 551                         if ((t = readtoken()) != TESAC) {
 552                                 if (t == TENDCASE)
 553                                         ;
 554                                 else if (t == TFALLTHRU)
 555                                         cp->type = NCLISTFALLTHRU;
 556                                 else
 557                                         synexpect(TENDCASE);
 558                                 checkkwd = CHKNL | CHKKWD, readtoken();
 559                         }
 560                         cpp = &cp->nclist.next;
 561                 }
 562                 *cpp = NULL;
 563                 checkkwd = CHKKWD | CHKALIAS;
 564                 break;
 565         case TLP:
 566                 n1 = (union node *)stalloc(sizeof (struct nredir));
 567                 n1->type = NSUBSHELL;
 568                 n1->nredir.n = list(0);
 569                 n1->nredir.redirect = NULL;
 570                 consumetoken(TRP);
 571                 checkkwd = CHKKWD | CHKALIAS;
 572                 is_subshell = 1;
 573                 break;
 574         case TBEGIN:
 575                 n1 = list(0);
 576                 consumetoken(TEND);
 577                 checkkwd = CHKKWD | CHKALIAS;
 578                 break;
 579         /* A simple command must have at least one redirection or word. */
 580         case TBACKGND:
 581         case TSEMI:
 582         case TAND:
 583         case TOR:
 584         case TPIPE:
 585         case TENDCASE:
 586         case TFALLTHRU:
 587         case TEOF:
 588         case TNL:
 589         case TRP:
 590                 if (!redir)
 591                         synexpect(-1);
 592         case TWORD:
 593                 tokpushback++;
 594                 n1 = simplecmd(rpp, redir);
 595                 return n1;
 596         default:
 597                 synexpect(-1);
 598         }
 599
 600         /* Now check for redirection which may follow command */
 601         while (readtoken() == TREDIR) {
 602                 *rpp = n2 = redirnode;
 603                 rpp = &n2->nfile.next;
 604                 parsefname();
 605         }
 606         tokpushback++;
 607         *rpp = NULL;
 608         if (redir) {
 609                 if (!is_subshell) {
 610                         n2 = (union node *)stalloc(sizeof (struct nredir));
 611                         n2->type = NREDIR;
 612                         n2->nredir.n = n1;
 613                         n1 = n2;
 614                 }
 615                 n1->nredir.redirect = redir;
 616         }
 617
 618         return n1;
 619 }
 620
 621
 622 static union node *
 623 simplecmd(union node **rpp, union node *redir)
 624 {
 625         union node *args, **app;
 626         union node **orig_rpp = rpp;
 627         union node *n = NULL;
 628         int special;
 629         int savecheckkwd;
 630
 631         /* If we don't have any redirections already, then we must reset */
 632         /* rpp to be the address of the local redir variable.  */
 633         if (redir == NULL)
 634                 rpp = &redir;
 635
 636         args = NULL;
 637         app = &args;
 638         /*
 639          * We save the incoming value, because we need this for shell
 640          * functions.  There can not be a redirect or an argument between
 641          * the function name and the open parenthesis.
 642          */
 643         orig_rpp = rpp;
 644
 645         savecheckkwd = CHKALIAS;
 646
 647         for (;;) {
 648                 checkkwd = savecheckkwd;
 649                 if (readtoken() == TWORD) {
 650                         n = makename();
 651                         *app = n;
 652                         app = &n->narg.next;
 653                         if (savecheckkwd != 0 && !isassignment(wordtext))
 654                                 savecheckkwd = 0;
 655                 } else if (lasttoken == TREDIR) {
 656                         *rpp = n = redirnode;
 657                         rpp = &n->nfile.next;
 658                         parsefname();   /* read name of redirection file */
 659                 } else if (lasttoken == TLP && app == &args->narg.next
 660                                             && rpp == orig_rpp) {
 661                         /* We have a function */
 662                         consumetoken(TRP);
 663                         funclinno = plinno;
 664                         /*
 665                          * - Require plain text.
 666                          * - Functions with '/' cannot be called.
 667                          * - Reject name=().
 668                          * - Reject ksh extended glob patterns.
 669                          */
 670                         if (!noexpand(n->narg.text) || quoteflag ||
 671                             strchr(n->narg.text, '/') ||
 672                             strchr("!%*+-=?@}~",
 673                                 n->narg.text[strlen(n->narg.text) - 1]))
 674                                 synerror("Bad function name");
 675                         rmescapes(n->narg.text);
 676                         if (find_builtin(n->narg.text, &special) >= 0 &&
 677                             special)
 678                                 synerror("Cannot override a special builtin with a function");
 679                         n->type = NDEFUN;
 680                         n->narg.next = command();
 681                         funclinno = 0;
 682                         return n;
 683                 } else {
 684                         tokpushback++;
 685                         break;
 686                 }
 687         }
 688         *app = NULL;
 689         *rpp = NULL;
 690         n = (union node *)stalloc(sizeof (struct ncmd));
 691         n->type = NCMD;
 692         n->ncmd.args = args;
 693         n->ncmd.redirect = redir;
 694         return n;
 695 }
 696
 697 static union node *
 698 makename(void)
 699 {
 700         union node *n;
 701
 702         n = (union node *)stalloc(sizeof (struct narg));
 703         n->type = NARG;
 704         n->narg.next = NULL;
 705         n->narg.text = wordtext;
 706         n->narg.backquote = backquotelist;
 707         return n;
 708 }
 709
 710 static union node *
 711 makebinary(int type, union node *n1, union node *n2)
 712 {
 713         union node *n;
 714
 715         n = (union node *)stalloc(sizeof (struct nbinary));
 716         n->type = type;
 717         n->nbinary.ch1 = n1;
 718         n->nbinary.ch2 = n2;
 719         return (n);
 720 }
 721
 722 void
 723 forcealias(void)
 724 {
 725         checkkwd |= CHKALIAS;
 726 }
 727
 728 void
 729 fixredir(union node *n, const char *text, int err)
 730 {
 731         TRACE(("Fix redir %s %d\n", text, err));
 732         if (!err)
 733                 n->ndup.vname = NULL;
 734
 735         if (is_digit(text[0]) && text[1] == '\0')
 736                 n->ndup.dupfd = digit_val(text[0]);
 737         else if (text[0] == '-' && text[1] == '\0')
 738                 n->ndup.dupfd = -1;
 739         else {
 740
 741                 if (err)
 742                         synerror("Bad fd number");
 743                 else
 744                         n->ndup.vname = makename();
 745         }
 746 }
 747
 748
 749 static void
 750 parsefname(void)
 751 {
 752         union node *n = redirnode;
 753
 754         consumetoken(TWORD);
 755         if (n->type == NHERE) {
 756                 struct heredoc *here = heredoc;
 757                 struct heredoc *p;
 758
 759                 if (quoteflag == 0)
 760                         n->type = NXHERE;
 761                 TRACE(("Here document %d\n", n->type));
 762                 if (here->striptabs) {
 763                         while (*wordtext == '\t')
 764                                 wordtext++;
 765                 }
 766                 if (! noexpand(wordtext))
 767                         synerror("Illegal eof marker for << redirection");
 768                 rmescapes(wordtext);
 769                 here->eofmark = wordtext;
 770                 here->next = NULL;
 771                 if (heredoclist == NULL)
 772                         heredoclist = here;
 773                 else {
 774                         for (p = heredoclist ; p->next ; p = p->next);
 775                         p->next = here;
 776                 }
 777         } else if (n->type == NTOFD || n->type == NFROMFD) {
 778                 fixredir(n, wordtext, 0);
 779         } else {
 780                 n->nfile.fname = makename();
 781         }
 782 }
 783
 784
 785 /*
 786  * Input any here documents.
 787  */
 788
 789 static void
 790 parseheredoc(void)
 791 {
 792         struct heredoc *here;
 793         union node *n;
 794
 795         while (heredoclist) {
 796                 here = heredoclist;
 797                 heredoclist = here->next;
 798                 if (needprompt) {
 799                         setprompt(2);
 800                         needprompt = 0;
 801                 }
 802                 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX,
 803                                 here->eofmark, here->striptabs);
 804                 n = makename();
 805                 here->here->nhere.doc = n;
 806         }
 807 }
 808
 809 static int
 810 peektoken(void)
 811 {
 812         int t;
 813
 814         t = readtoken();
 815         tokpushback++;
 816         return (t);
 817 }
 818
 819 static int
 820 readtoken(void)
 821 {
 822         int t;
 823         struct alias *ap;
 824 #ifdef DEBUG
 825         int alreadyseen = tokpushback;
 826 #endif
 827
 828         top:
 829         t = xxreadtoken();
 830
 831         /*
 832          * eat newlines
 833          */
 834         if (checkkwd & CHKNL) {
 835                 while (t == TNL) {
 836                         parseheredoc();
 837                         t = xxreadtoken();
 838                 }
 839         }
 840
 841         /*
 842          * check for keywords and aliases
 843          */
 844         if (t == TWORD && !quoteflag)
 845         {
 846                 const char * const *pp;
 847
 848                 if (checkkwd & CHKKWD)
 849                         for (pp = parsekwd; *pp; pp++) {
 850                                 if (**pp == *wordtext && equal(*pp, wordtext))
 851                                 {
 852                                         lasttoken = t = pp - parsekwd + KWDOFFSET;
 853                                         TRACE(("keyword %s recognized\n", tokname[t]));
 854                                         goto out;
 855                                 }
 856                         }
 857                 if (checkkwd & CHKALIAS &&
 858                     (ap = lookupalias(wordtext, 1)) != NULL) {
 859                         pushstring(ap->val, strlen(ap->val), ap);
 860                         goto top;
 861                 }
 862         }
 863 out:
 864         if (t != TNOT)
 865                 checkkwd = 0;
 866
 867 #ifdef DEBUG
 868         if (!alreadyseen)
 869             TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
 870         else
 871             TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
 872 #endif
 873         return (t);
 874 }
 875
 876
 877 /*
 878  * Read the next input token.
 879  * If the token is a word, we set backquotelist to the list of cmds in
 880  *      backquotes.  We set quoteflag to true if any part of the word was
 881  *      quoted.
 882  * If the token is TREDIR, then we set redirnode to a structure containing
 883  *      the redirection.
 884  * In all cases, the variable startlinno is set to the number of the line
 885  *      on which the token starts.
 886  *
 887  * [Change comment:  here documents and internal procedures]
 888  * [Readtoken shouldn't have any arguments.  Perhaps we should make the
 889  *  word parsing code into a separate routine.  In this case, readtoken
 890  *  doesn't need to have any internal procedures, but parseword does.
 891  *  We could also make parseoperator in essence the main routine, and
 892  *  have parseword (readtoken1?) handle both words and redirection.]
 893  */
 894
 895 #define RETURN(token)   return lasttoken = token
 896
 897 static int
 898 xxreadtoken(void)
 899 {
 900         int c;
 901
 902         if (tokpushback) {
 903                 tokpushback = 0;
 904                 return lasttoken;
 905         }
 906         if (needprompt) {
 907                 setprompt(2);
 908                 needprompt = 0;
 909         }
 910         startlinno = plinno;
 911         for (;;) {      /* until token or start of word found */
 912                 c = pgetc_macro();
 913                 switch (c) {
 914                 case ' ': case '\t':
 915                         continue;
 916                 case '#':
 917                         while ((c = pgetc()) != '\n' && c != PEOF);
 918                         pungetc();
 919                         continue;
 920                 case '\\':
 921                         if (pgetc() == '\n') {
 922                                 startlinno = ++plinno;
 923                                 if (doprompt)
 924                                         setprompt(2);
 925                                 else
 926                                         setprompt(0);
 927                                 continue;
 928                         }
 929                         pungetc();
 930                         /* FALLTHROUGH */
 931                 default:
 932                         return readtoken1(c, BASESYNTAX, (char *)NULL, 0);
 933                 case '\n':
 934                         plinno++;
 935                         needprompt = doprompt;
 936                         RETURN(TNL);
 937                 case PEOF:
 938                         RETURN(TEOF);
 939                 case '&':
 940                         if (pgetc_linecont() == '&')
 941                                 RETURN(TAND);
 942                         pungetc();
 943                         RETURN(TBACKGND);
 944                 case '|':
 945                         if (pgetc_linecont() == '|')
 946                                 RETURN(TOR);
 947                         pungetc();
 948                         RETURN(TPIPE);
 949                 case ';':
 950                         c = pgetc_linecont();
 951                         if (c == ';')
 952                                 RETURN(TENDCASE);
 953                         else if (c == '&')
 954                                 RETURN(TFALLTHRU);
 955                         pungetc();
 956                         RETURN(TSEMI);
 957                 case '(':
 958                         RETURN(TLP);
 959                 case ')':
 960                         RETURN(TRP);
 961                 }
 962         }
 963 #undef RETURN
 964 }
 965
 966
 967 #define MAXNEST_static 8
 968 struct tokenstate
 969 {
 970         const char *syntax; /* *SYNTAX */
 971         int parenlevel; /* levels of parentheses in arithmetic */
 972         enum tokenstate_category
 973         {
 974                 TSTATE_TOP,
 975                 TSTATE_VAR_OLD, /* ${var+-=?}, inherits dquotes */
 976                 TSTATE_VAR_NEW, /* other ${var...}, own dquote state */
 977                 TSTATE_ARITH
 978         } category;
 979 };
 980
 981
 982 /*
 983  * Check to see whether we are at the end of the here document.  When this
 984  * is called, c is set to the first character of the next input line.  If
 985  * we are at the end of the here document, this routine sets the c to PEOF.
 986  * The new value of c is returned.
 987  */
 988
 989 static int
 990 checkend(int c, const char *eofmark, int striptabs)
 991 {
 992         if (striptabs) {
 993                 while (c == '\t')
 994                         c = pgetc();
 995         }
 996         if (c == *eofmark) {
 997                 int c2;
 998                 const char *q;
 999
1000                 for (q = eofmark + 1; c2 = pgetc(), *q != '\0' && c2 == *q; q++)
1001                         ;
1002                 if ((c2 == PEOF || c2 == '\n') && *q == '\0') {
1003                         c = PEOF;
1004                         if (c2 == '\n') {
1005                                 plinno++;
1006                                 needprompt = doprompt;
1007                         }
1008                 } else {
1009                         pungetc();
1010                         pushstring(eofmark + 1, q - (eofmark + 1), NULL);
1011                 }
1012         } else if (c == '\n' && *eofmark == '\0') {
1013                 c = PEOF;
1014                 plinno++;
1015                 needprompt = doprompt;
1016         }
1017         return (c);
1018 }
1019
1020
1021 /*
1022  * Parse a redirection operator.  The variable "out" points to a string
1023  * specifying the fd to be redirected.  The variable "c" contains the
1024  * first character of the redirection operator.
1025  */
1026
1027 static void
1028 parseredir(char *out, int c)
1029 {
1030         char fd = *out;
1031         union node *np;
1032
1033         np = (union node *)stalloc(sizeof (struct nfile));
1034         if (c == '>') {
1035                 np->nfile.fd = 1;
1036                 c = pgetc_linecont();
1037                 if (c == '>')
1038                         np->type = NAPPEND;
1039                 else if (c == '&')
1040                         np->type = NTOFD;
1041                 else if (c == '|')
1042                         np->type = NCLOBBER;
1043                 else {
1044                         np->type = NTO;
1045                         pungetc();
1046                 }
1047         } else {        /* c == '<' */
1048                 np->nfile.fd = 0;
1049                 c = pgetc_linecont();
1050                 if (c == '<') {
1051                         if (sizeof (struct nfile) != sizeof (struct nhere)) {
1052                                 np = (union node *)stalloc(sizeof (struct nhere));
1053                                 np->nfile.fd = 0;
1054                         }
1055                         np->type = NHERE;
1056                         heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc));
1057                         heredoc->here = np;
1058                         if ((c = pgetc_linecont()) == '-') {
1059                                 heredoc->striptabs = 1;
1060                         } else {
1061                                 heredoc->striptabs = 0;
1062                                 pungetc();
1063                         }
1064                 } else if (c == '&')
1065                         np->type = NFROMFD;
1066                 else if (c == '>')
1067                         np->type = NFROMTO;
1068                 else {
1069                         np->type = NFROM;
1070                         pungetc();
1071                 }
1072         }
1073         if (fd != '\0')
1074                 np->nfile.fd = digit_val(fd);
1075         redirnode = np;
1076 }
1077
1078 /*
1079  * Called to parse command substitutions.
1080  */
1081
1082 static char *
1083 parsebackq(char *out, struct nodelist **pbqlist,
1084                 int oldstyle, int dblquote, int quoted)
1085 {
1086         struct nodelist **nlpp;
1087         union node *n;
1088         char *volatile str;
1089         struct jmploc jmploc;
1090         struct jmploc *const savehandler = handler;
1091         size_t savelen;
1092         int saveprompt;
1093         const int bq_startlinno = plinno;
1094         char *volatile ostr = NULL;
1095         struct parsefile *const savetopfile = getcurrentfile();
1096         struct heredoc *const saveheredoclist = heredoclist;
1097         struct heredoc *here;
1098
1099         str = NULL;
1100         if (setjmp(jmploc.loc)) {
1101                 popfilesupto(savetopfile);
1102                 if (str)
1103                         ckfree(str);
1104                 if (ostr)
1105                         ckfree(ostr);
1106                 heredoclist = saveheredoclist;
1107                 handler = savehandler;
1108                 if (exception == EXERROR) {
1109                         startlinno = bq_startlinno;
1110                         synerror("Error in command substitution");
1111                 }
1112                 longjmp(handler->loc, 1);
1113         }
1114         INTOFF;
1115         savelen = out - stackblock();
1116         if (savelen > 0) {
1117                 str = ckmalloc(savelen);
1118                 memcpy(str, stackblock(), savelen);
1119         }
1120         handler = &jmploc;
1121         heredoclist = NULL;
1122         INTON;
1123         if (oldstyle) {
1124                 /* We must read until the closing backquote, giving special
1125                    treatment to some slashes, and then push the string and
1126                    reread it as input, interpreting it normally.  */
1127                 char *oout;
1128                 int c;
1129                 int olen;
1130
1131
1132                 STARTSTACKSTR(oout);
1133                 for (;;) {
1134                         if (needprompt) {
1135                                 setprompt(2);
1136                                 needprompt = 0;
1137                         }
1138                         CHECKSTRSPACE(2, oout);
1139                         c = pgetc_linecont();
1140                         if (c == '`')
1141                                 break;
1142                         switch (c) {
1143                         case '\\':
1144                                 c = pgetc();
1145                                 if (c != '\\' && c != '`' && c != '$'
1146                                     && (!dblquote || c != '"'))
1147                                         USTPUTC('\\', oout);
1148                                 break;
1149
1150                         case '\n':
1151                                 plinno++;
1152                                 needprompt = doprompt;
1153                                 break;
1154
1155                         case PEOF:
1156                                 startlinno = plinno;
1157                                 synerror("EOF in backquote substitution");
1158                                 break;
1159
1160                         default:
1161                                 break;
1162                         }
1163                         USTPUTC(c, oout);
1164                 }
1165                 USTPUTC('\0', oout);
1166                 olen = oout - stackblock();
1167                 INTOFF;
1168                 ostr = ckmalloc(olen);
1169                 memcpy(ostr, stackblock(), olen);
1170                 setinputstring(ostr, 1);
1171                 INTON;
1172         }
1173         nlpp = pbqlist;
1174         while (*nlpp)
1175                 nlpp = &(*nlpp)->next;
1176         *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist));
1177         (*nlpp)->next = NULL;
1178
1179         if (oldstyle) {
1180                 saveprompt = doprompt;
1181                 doprompt = 0;
1182         }
1183
1184         n = list(0);
1185
1186         if (oldstyle) {
1187                 if (peektoken() != TEOF)
1188                         synexpect(-1);
1189                 doprompt = saveprompt;
1190         } else
1191                 consumetoken(TRP);
1192
1193         (*nlpp)->n = n;
1194         if (oldstyle) {
1195                 /*
1196                  * Start reading from old file again, ignoring any pushed back
1197                  * tokens left from the backquote parsing
1198                  */
1199                 popfile();
1200                 tokpushback = 0;
1201         }
1202         STARTSTACKSTR(out);
1203         CHECKSTRSPACE(savelen + 1, out);
1204         INTOFF;
1205         if (str) {
1206                 memcpy(out, str, savelen);
1207                 STADJUST(savelen, out);
1208                 ckfree(str);
1209                 str = NULL;
1210         }
1211         if (ostr) {
1212                 ckfree(ostr);
1213                 ostr = NULL;
1214         }
1215         here = saveheredoclist;
1216         if (here != NULL) {
1217                 while (here->next != NULL)
1218                         here = here->next;
1219                 here->next = heredoclist;
1220                 heredoclist = saveheredoclist;
1221         }
1222         handler = savehandler;
1223         INTON;
1224         if (quoted)
1225                 USTPUTC(CTLBACKQ | CTLQUOTE, out);
1226         else
1227                 USTPUTC(CTLBACKQ, out);
1228         return out;
1229 }
1230
1231
1232 /*
1233  * Called to parse a backslash escape sequence inside $'...'.
1234  * The backslash has already been read.
1235  */
1236 static char *
1237 readcstyleesc(char *out)
1238 {
1239         int c, vc, i, n;
1240         unsigned int v;
1241
1242         c = pgetc();
1243         switch (c) {
1244         case '\0':
1245                 synerror("Unterminated quoted string");
1246         case '\n':
1247                 plinno++;
1248                 if (doprompt)
1249                         setprompt(2);
1250                 else
1251                         setprompt(0);
1252                 return out;
1253         case '\\':
1254         case '\'':
1255         case '"':
1256                 v = c;
1257                 break;
1258         case 'a': v = '\a'; break;
1259         case 'b': v = '\b'; break;
1260         case 'e': v = '\033'; break;
1261         case 'f': v = '\f'; break;
1262         case 'n': v = '\n'; break;
1263         case 'r': v = '\r'; break;
1264         case 't': v = '\t'; break;
1265         case 'v': v = '\v'; break;
1266         case 'x':
1267                   v = 0;
1268                   for (;;) {
1269                           c = pgetc();
1270                           if (c >= '0' && c <= '9')
1271                                   v = (v << 4) + c - '0';
1272                           else if (c >= 'A' && c <= 'F')
1273                                   v = (v << 4) + c - 'A' + 10;
1274                           else if (c >= 'a' && c <= 'f')
1275                                   v = (v << 4) + c - 'a' + 10;
1276                           else
1277                                   break;
1278                   }
1279                   pungetc();
1280                   break;
1281         case '0': case '1': case '2': case '3':
1282         case '4': case '5': case '6': case '7':
1283                   v = c - '0';
1284                   c = pgetc();
1285                   if (c >= '0' && c <= '7') {
1286                           v <<= 3;
1287                           v += c - '0';
1288                           c = pgetc();
1289                           if (c >= '0' && c <= '7') {
1290                                   v <<= 3;
1291                                   v += c - '0';
1292                           } else
1293                                   pungetc();
1294                   } else
1295                           pungetc();
1296                   break;
1297         case 'c':
1298                   c = pgetc();
1299                   if (c < 0x3f || c > 0x7a || c == 0x60)
1300                           synerror("Bad escape sequence");
1301                   if (c == '\\' && pgetc() != '\\')
1302                           synerror("Bad escape sequence");
1303                   if (c == '?')
1304                           v = 127;
1305                   else
1306                           v = c & 0x1f;
1307                   break;
1308         case 'u':
1309         case 'U':
1310                   n = c == 'U' ? 8 : 4;
1311                   v = 0;
1312                   for (i = 0; i < n; i++) {
1313                           c = pgetc();
1314                           if (c >= '0' && c <= '9')
1315                                   v = (v << 4) + c - '0';
1316                           else if (c >= 'A' && c <= 'F')
1317                                   v = (v << 4) + c - 'A' + 10;
1318                           else if (c >= 'a' && c <= 'f')
1319                                   v = (v << 4) + c - 'a' + 10;
1320                           else
1321                                   synerror("Bad escape sequence");
1322                   }
1323                   if (v == 0 || (v >= 0xd800 && v <= 0xdfff))
1324                           synerror("Bad escape sequence");
1325                   /* We really need iconv here. */
1326                   if (initial_localeisutf8 && v > 127) {
1327                           CHECKSTRSPACE(4, out);
1328                           /*
1329                            * We cannot use wctomb() as the locale may have
1330                            * changed.
1331                            */
1332                           if (v <= 0x7ff) {
1333                                   USTPUTC(0xc0 | v >> 6, out);
1334                                   USTPUTC(0x80 | (v & 0x3f), out);
1335                                   return out;
1336                           } else if (v <= 0xffff) {
1337                                   USTPUTC(0xe0 | v >> 12, out);
1338                                   USTPUTC(0x80 | ((v >> 6) & 0x3f), out);
1339                                   USTPUTC(0x80 | (v & 0x3f), out);
1340                                   return out;
1341                           } else if (v <= 0x10ffff) {
1342                                   USTPUTC(0xf0 | v >> 18, out);
1343                                   USTPUTC(0x80 | ((v >> 12) & 0x3f), out);
1344                                   USTPUTC(0x80 | ((v >> 6) & 0x3f), out);
1345                                   USTPUTC(0x80 | (v & 0x3f), out);
1346                                   return out;
1347                           }
1348                   }
1349                   if (v > 127)
1350                           v = '?';
1351                   break;
1352         default:
1353                   synerror("Bad escape sequence");
1354         }
1355         vc = (char)v;
1356         /*
1357          * We can't handle NUL bytes.
1358          * POSIX says we should skip till the closing quote.
1359          */
1360         if (vc == '\0') {
1361                 while ((c = pgetc()) != '\'') {
1362                         if (c == '\\')
1363                                 c = pgetc();
1364                         if (c == PEOF)
1365                                 synerror("Unterminated quoted string");
1366                         if (c == '\n') {
1367                                 plinno++;
1368                                 if (doprompt)
1369                                         setprompt(2);
1370                                 else
1371                                         setprompt(0);
1372                         }
1373                 }
1374                 pungetc();
1375                 return out;
1376         }
1377         if (SQSYNTAX[vc] == CCTL)
1378                 USTPUTC(CTLESC, out);
1379         USTPUTC(vc, out);
1380         return out;
1381 }
1382
1383
1384 /*
1385  * If eofmark is NULL, read a word or a redirection symbol.  If eofmark
1386  * is not NULL, read a here document.  In the latter case, eofmark is the
1387  * word which marks the end of the document and striptabs is true if
1388  * leading tabs should be stripped from the document.  The argument firstc
1389  * is the first character of the input token or document.
1390  *
1391  * Because C does not have internal subroutines, I have simulated them
1392  * using goto's to implement the subroutine linkage.  The following macros
1393  * will run code that appears at the end of readtoken1.
1394  */
1395
1396 #define PARSESUB()      {goto parsesub; parsesub_return:;}
1397 #define PARSEARITH()    {goto parsearith; parsearith_return:;}
1398
1399 static int
1400 readtoken1(int firstc, char const *initialsyntax, const char *eofmark,
1401     int striptabs)
1402 {
1403         int c = firstc;
1404         char *out;
1405         int len;
1406         struct nodelist *bqlist;
1407         int quotef;
1408         int newvarnest;
1409         int level;
1410         int synentry;
1411         struct tokenstate state_static[MAXNEST_static];
1412         int maxnest = MAXNEST_static;
1413         struct tokenstate *state = state_static;
1414         int sqiscstyle = 0;
1415
1416         startlinno = plinno;
1417         quotef = 0;
1418         bqlist = NULL;
1419         newvarnest = 0;
1420         level = 0;
1421         state[level].syntax = initialsyntax;
1422         state[level].parenlevel = 0;
1423         state[level].category = TSTATE_TOP;
1424
1425         STARTSTACKSTR(out);
1426         loop: { /* for each line, until end of word */
1427                 if (eofmark && eofmark != NOEOFMARK)
1428                         /* set c to PEOF if at end of here document */
1429                         c = checkend(c, eofmark, striptabs);
1430                 for (;;) {      /* until end of line or end of word */
1431                         CHECKSTRSPACE(4, out);  /* permit 4 calls to USTPUTC */
1432
1433                         synentry = state[level].syntax[c];
1434
1435                         switch(synentry) {
1436                         case CNL:       /* '\n' */
1437                                 if (level == 0 &&
1438                                     state[level].syntax == BASESYNTAX)
1439                                         goto endword;   /* exit outer loop */
1440                                 USTPUTC(c, out);
1441                                 plinno++;
1442                                 if (doprompt)
1443                                         setprompt(2);
1444                                 else
1445                                         setprompt(0);
1446                                 c = pgetc();
1447                                 goto loop;              /* continue outer loop */
1448                         case CSBACK:
1449                                 if (sqiscstyle) {
1450                                         out = readcstyleesc(out);
1451                                         break;
1452                                 }
1453                                 /* FALLTHROUGH */
1454                         case CWORD:
1455                                 USTPUTC(c, out);
1456                                 break;
1457                         case CCTL:
1458                                 if (eofmark == NULL || initialsyntax != SQSYNTAX)
1459                                         USTPUTC(CTLESC, out);
1460                                 USTPUTC(c, out);
1461                                 break;
1462                         case CBACK:     /* backslash */
1463                                 c = pgetc();
1464                                 if (c == PEOF) {
1465                                         USTPUTC('\\', out);
1466                                         pungetc();
1467                                 } else if (c == '\n') {
1468                                         plinno++;
1469                                         if (doprompt)
1470                                                 setprompt(2);
1471                                         else
1472                                                 setprompt(0);
1473                                 } else {
1474                                         if (state[level].syntax == DQSYNTAX &&
1475                                             c != '\\' && c != '`' && c != '$' &&
1476                                             (c != '"' || (eofmark != NULL &&
1477                                                 newvarnest == 0)) &&
1478                                             (c != '}' || state[level].category != TSTATE_VAR_OLD))
1479                                                 USTPUTC('\\', out);
1480                                         if ((eofmark == NULL ||
1481                                             newvarnest > 0) &&
1482                                             state[level].syntax == BASESYNTAX)
1483                                                 USTPUTC(CTLQUOTEMARK, out);
1484                                         if (SQSYNTAX[c] == CCTL)
1485                                                 USTPUTC(CTLESC, out);
1486                                         USTPUTC(c, out);
1487                                         if ((eofmark == NULL ||
1488                                             newvarnest > 0) &&
1489                                             state[level].syntax == BASESYNTAX &&
1490                                             state[level].category == TSTATE_VAR_OLD)
1491                                                 USTPUTC(CTLQUOTEEND, out);
1492                                         quotef++;
1493                                 }
1494                                 break;
1495                         case CSQUOTE:
1496                                 USTPUTC(CTLQUOTEMARK, out);
1497                                 state[level].syntax = SQSYNTAX;
1498                                 sqiscstyle = 0;
1499                                 break;
1500                         case CDQUOTE:
1501                                 USTPUTC(CTLQUOTEMARK, out);
1502                                 state[level].syntax = DQSYNTAX;
1503                                 break;
1504                         case CENDQUOTE:
1505                                 if (eofmark != NULL && newvarnest == 0)
1506                                         USTPUTC(c, out);
1507                                 else {
1508                                         if (state[level].category == TSTATE_VAR_OLD)
1509                                                 USTPUTC(CTLQUOTEEND, out);
1510                                         state[level].syntax = BASESYNTAX;
1511                                         quotef++;
1512                                 }
1513                                 break;
1514                         case CVAR:      /* '$' */
1515                                 PARSESUB();             /* parse substitution */
1516                                 break;
1517                         case CENDVAR:   /* '}' */
1518                                 if (level > 0 &&
1519                                     ((state[level].category == TSTATE_VAR_OLD &&
1520                                       state[level].syntax ==
1521                                       state[level - 1].syntax) ||
1522                                     (state[level].category == TSTATE_VAR_NEW &&
1523                                      state[level].syntax == BASESYNTAX))) {
1524                                         if (state[level].category == TSTATE_VAR_NEW)
1525                                                 newvarnest--;
1526                                         level--;
1527                                         USTPUTC(CTLENDVAR, out);
1528                                 } else {
1529                                         USTPUTC(c, out);
1530                                 }
1531                                 break;
1532                         case CLP:       /* '(' in arithmetic */
1533                                 state[level].parenlevel++;
1534                                 USTPUTC(c, out);
1535                                 break;
1536                         case CRP:       /* ')' in arithmetic */
1537                                 if (state[level].parenlevel > 0) {
1538                                         USTPUTC(c, out);
1539                                         --state[level].parenlevel;
1540                                 } else {
1541                                         if (pgetc_linecont() == ')') {
1542                                                 if (level > 0 &&
1543                                                     state[level].category == TSTATE_ARITH) {
1544                                                         level--;
1545                                                         USTPUTC(CTLENDARI, out);
1546                                                 } else
1547                                                         USTPUTC(')', out);
1548                                         } else {
1549                                                 /*
1550                                                  * unbalanced parens
1551                                                  *  (don't 2nd guess - no error)
1552                                                  */
1553                                                 pungetc();
1554                                                 USTPUTC(')', out);
1555                                         }
1556                                 }
1557                                 break;
1558                         case CBQUOTE:   /* '`' */
1559                                 out = parsebackq(out, &bqlist, 1,
1560                                     state[level].syntax == DQSYNTAX &&
1561                                     (eofmark == NULL || newvarnest > 0),
1562                                     state[level].syntax == DQSYNTAX || state[level].syntax == ARISYNTAX);
1563                                 break;
1564                         case CEOF:
1565                                 goto endword;           /* exit outer loop */
1566                         case CIGN:
1567                                 break;
1568                         default:
1569                                 if (level == 0)
1570                                         goto endword;   /* exit outer loop */
1571                                 USTPUTC(c, out);
1572                         }
1573                         c = pgetc_macro();
1574                 }
1575         }
1576 endword:
1577         if (state[level].syntax == ARISYNTAX)
1578                 synerror("Missing '))'");
1579         if (state[level].syntax != BASESYNTAX && eofmark == NULL)
1580                 synerror("Unterminated quoted string");
1581         if (state[level].category == TSTATE_VAR_OLD ||
1582             state[level].category == TSTATE_VAR_NEW) {
1583                 startlinno = plinno;
1584                 synerror("Missing '}'");
1585         }
1586         if (state != state_static)
1587                 parser_temp_free_upto(state);
1588         USTPUTC('\0', out);
1589         len = out - stackblock();
1590         out = stackblock();
1591         if (eofmark == NULL) {
1592                 if ((c == '>' || c == '<')
1593                  && quotef == 0
1594                  && len <= 2
1595                  && (*out == '\0' || is_digit(*out))) {
1596                         parseredir(out, c);
1597                         return lasttoken = TREDIR;
1598                 } else {
1599                         pungetc();
1600                 }
1601         }
1602         quoteflag = quotef;
1603         backquotelist = bqlist;
1604         grabstackblock(len);
1605         wordtext = out;
1606         return lasttoken = TWORD;
1607 /* end of readtoken routine */
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         int subtype;
1617         int typeloc;
1618         int flags;
1619         char *p;
1620         static const char types[] = "}-+?=";
1621         int linno;
1622         int length;
1623         int c1;
1624
1625         c = pgetc_linecont();
1626         if (c == '(') { /* $(command) or $((arith)) */
1627                 if (pgetc_linecont() == '(') {
1628                         PARSEARITH();
1629                 } else {
1630                         pungetc();
1631                         out = parsebackq(out, &bqlist, 0,
1632                             state[level].syntax == DQSYNTAX &&
1633                             (eofmark == NULL || newvarnest > 0),
1634                             state[level].syntax == DQSYNTAX ||
1635                             state[level].syntax == ARISYNTAX);
1636                 }
1637         } else if (c == '{' || is_name(c) || is_special(c)) {
1638                 USTPUTC(CTLVAR, out);
1639                 typeloc = out - stackblock();
1640                 USTPUTC(VSNORMAL, out);
1641                 subtype = VSNORMAL;
1642                 flags = 0;
1643                 if (c == '{') {
1644                         c = pgetc_linecont();
1645                         subtype = 0;
1646                 }
1647 varname:
1648                 if (!is_eof(c) && is_name(c)) {
1649                         length = 0;
1650                         do {
1651                                 STPUTC(c, out);
1652                                 c = pgetc_linecont();
1653                                 length++;
1654                         } while (!is_eof(c) && is_in_name(c));
1655                         if (length == 6 &&
1656                             strncmp(out - length, "LINENO", length) == 0) {
1657                                 /* Replace the variable name with the
1658                                  * current line number. */
1659                                 STADJUST(-6, out);
1660                                 CHECKSTRSPACE(11, out);
1661                                 linno = plinno;
1662                                 if (funclinno != 0)
1663                                         linno -= funclinno - 1;
1664                                 length = snprintf(out, 11, "%d", linno);
1665                                 if (length > 10)
1666                                         length = 10;
1667                                 out += length;
1668                                 flags |= VSLINENO;
1669                         }
1670                 } else if (is_digit(c)) {
1671                         if (subtype != VSNORMAL) {
1672                                 do {
1673                                         STPUTC(c, out);
1674                                         c = pgetc_linecont();
1675                                 } while (is_digit(c));
1676                         } else {
1677                                 USTPUTC(c, out);
1678                                 c = pgetc_linecont();
1679                         }
1680                 } else if (is_special(c)) {
1681                         c1 = c;
1682                         c = pgetc_linecont();
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_linecont();
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_linecont();
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_linecont();
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         else if (arg0)
1927                 outfmt(out2, "%s: ", arg0);
1928         outfmt(out2, "Syntax error: %s\n", msg);
1929         error((char *)NULL);
1930 }
1931
1932 static void
1933 setprompt(int which)
1934 {
1935         whichprompt = which;
1936         if (which == 0)
1937                 return;
1938
1939 #ifndef NO_HISTORY
1940         if (!el)
1941 #endif
1942         {
1943                 out2str(getprompt(NULL));
1944                 flushout(out2);
1945         }
1946 }
1947
1948 static int
1949 pgetc_linecont(void)
1950 {
1951         int c;
1952
1953         while ((c = pgetc_macro()) == '\\') {
1954                 c = pgetc();
1955                 if (c == '\n') {
1956                         plinno++;
1957                         if (doprompt)
1958                                 setprompt(2);
1959                         else
1960                                 setprompt(0);
1961                 } else {
1962                         pungetc();
1963                         /* Allow the backslash to be pushed back. */
1964                         pushstring("\\", 1, NULL);
1965                         return (pgetc());
1966                 }
1967         }
1968         return (c);
1969 }
1970
1971 /*
1972  * called by editline -- any expansions to the prompt
1973  *    should be added here.
1974  */
1975 char *
1976 getprompt(void *unused __unused)
1977 {
1978         static char ps[PROMPTLEN];
1979         const char *fmt;
1980         const char *pwd;
1981         int i, trim;
1982         static char internal_error[] = "??";
1983
1984         /*
1985          * Select prompt format.
1986          */
1987         switch (whichprompt) {
1988         case 0:
1989                 fmt = "";
1990                 break;
1991         case 1:
1992                 fmt = ps1val();
1993                 break;
1994         case 2:
1995                 fmt = ps2val();
1996                 break;
1997         default:
1998                 return internal_error;
1999         }
2000
2001         /*
2002          * Format prompt string.
2003          */
2004         for (i = 0; (i < PROMPTLEN - 1) && (*fmt != '\0'); i++, fmt++)
2005                 if (*fmt == '\\')
2006                         switch (*++fmt) {
2007
2008                                 /*
2009                                  * Hostname.
2010                                  *
2011                                  * \h specifies just the local hostname,
2012                                  * \H specifies fully-qualified hostname.
2013                                  */
2014                         case 'h':
2015                         case 'H':
2016                                 ps[i] = '\0';
2017                                 gethostname(&ps[i], PROMPTLEN - i - 1);
2018                                 ps[PROMPTLEN - 1] = '\0';
2019                                 /* Skip to end of hostname. */
2020                                 trim = (*fmt == 'h') ? '.' : '\0';
2021                                 while ((ps[i] != '\0') && (ps[i] != trim))
2022                                         i++;
2023                                 --i;
2024                                 break;
2025
2026                                 /*
2027                                  * Working directory.
2028                                  *
2029                                  * \W specifies just the final component,
2030                                  * \w specifies the entire path.
2031                                  */
2032                         case 'W':
2033                         case 'w':
2034                                 pwd = lookupvar("PWD");
2035                                 if (pwd == NULL || *pwd == '\0')
2036                                         pwd = "?";
2037                                 if (*fmt == 'W' &&
2038                                     *pwd == '/' && pwd[1] != '\0')
2039                                         strlcpy(&ps[i], strrchr(pwd, '/') + 1,
2040                                             PROMPTLEN - i);
2041                                 else
2042                                         strlcpy(&ps[i], pwd, PROMPTLEN - i);
2043                                 /* Skip to end of path. */
2044                                 while (ps[i + 1] != '\0')
2045                                         i++;
2046                                 break;
2047
2048                                 /*
2049                                  * Superuser status.
2050                                  *
2051                                  * '$' for normal users, '#' for root.
2052                                  */
2053                         case '$':
2054                                 ps[i] = (geteuid() != 0) ? '$' : '#';
2055                                 break;
2056
2057                                 /*
2058                                  * A literal \.
2059                                  */
2060                         case '\\':
2061                                 ps[i] = '\\';
2062                                 break;
2063
2064                                 /*
2065                                  * Emit unrecognized formats verbatim.
2066                                  */
2067                         default:
2068                                 ps[i] = '\\';
2069                                 if (i < PROMPTLEN - 2)
2070                                         ps[++i] = *fmt;
2071                                 break;
2072                         }
2073                 else
2074                         ps[i] = *fmt;
2075         ps[i] = '\0';
2076         return (ps);
2077 }
2078
2079
2080 const char *
2081 expandstr(const char *ps)
2082 {
2083         union node n;
2084         struct jmploc jmploc;
2085         struct jmploc *const savehandler = handler;
2086         const int saveprompt = doprompt;
2087         struct parsefile *const savetopfile = getcurrentfile();
2088         struct parser_temp *const saveparser_temp = parser_temp;
2089         const char *result = NULL;
2090
2091         if (!setjmp(jmploc.loc)) {
2092                 handler = &jmploc;
2093                 parser_temp = NULL;
2094                 setinputstring(ps, 1);
2095                 doprompt = 0;
2096                 readtoken1(pgetc(), DQSYNTAX, NOEOFMARK, 0);
2097                 if (backquotelist != NULL)
2098                         error("Command substitution not allowed here");
2099
2100                 n.narg.type = NARG;
2101                 n.narg.next = NULL;
2102                 n.narg.text = wordtext;
2103                 n.narg.backquote = backquotelist;
2104
2105                 expandarg(&n, NULL, 0);
2106                 result = stackblock();
2107                 INTOFF;
2108         }
2109         handler = savehandler;
2110         doprompt = saveprompt;
2111         popfilesupto(savetopfile);
2112         if (parser_temp != saveparser_temp) {
2113                 parser_temp_free_all();
2114                 parser_temp = saveparser_temp;
2115         }
2116         if (result != NULL) {
2117                 INTON;
2118         } else if (exception == EXINT)
2119                 raise(SIGINT);
2120         return result;
2121 }