parser.c revision 1.97 1 /* $NetBSD: parser.c,v 1.97 2016/02/22 19:42:46 christos Exp $ */
2
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 #include <sys/cdefs.h>
36 #ifndef lint
37 #if 0
38 static char sccsid[] = "@(#)parser.c 8.7 (Berkeley) 5/16/95";
39 #else
40 __RCSID("$NetBSD: parser.c,v 1.97 2016/02/22 19:42:46 christos Exp $");
41 #endif
42 #endif /* not lint */
43
44 #include <stdio.h>
45 #include <stdlib.h>
46 #include <limits.h>
47
48 #include "shell.h"
49 #include "parser.h"
50 #include "nodes.h"
51 #include "expand.h" /* defines rmescapes() */
52 #include "eval.h" /* defines commandname */
53 #include "redir.h" /* defines copyfd() */
54 #include "syntax.h"
55 #include "options.h"
56 #include "input.h"
57 #include "output.h"
58 #include "var.h"
59 #include "error.h"
60 #include "memalloc.h"
61 #include "mystring.h"
62 #include "alias.h"
63 #include "show.h"
64 #ifndef SMALL
65 #include "myhistedit.h"
66 #endif
67
68 /*
69 * Shell command parser.
70 */
71
72 #define EOFMARKLEN 79
73
74 /* values returned by readtoken */
75 #include "token.h"
76
77 #define OPENBRACE '{'
78 #define CLOSEBRACE '}'
79
80
81 struct heredoc {
82 struct heredoc *next; /* next here document in list */
83 union node *here; /* redirection node */
84 char *eofmark; /* string indicating end of input */
85 int striptabs; /* if set, strip leading tabs */
86 };
87
88
89
90 static int noalias = 0; /* when set, don't handle aliases */
91 struct heredoc *heredoclist; /* list of here documents to read */
92 int parsebackquote; /* nonzero if we are inside backquotes */
93 int doprompt; /* if set, prompt the user */
94 int needprompt; /* true if interactive and at start of line */
95 int lasttoken; /* last token read */
96 MKINIT int tokpushback; /* last token pushed back */
97 char *wordtext; /* text of last word returned by readtoken */
98 MKINIT int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */
99 struct nodelist *backquotelist;
100 union node *redirnode;
101 struct heredoc *heredoc;
102 int quoteflag; /* set if (part of) last token was quoted */
103 int startlinno; /* line # where last token started */
104 int funclinno; /* line # where the current function started */
105
106
107 STATIC union node *list(int, int);
108 STATIC union node *andor(void);
109 STATIC union node *pipeline(void);
110 STATIC union node *command(void);
111 STATIC union node *simplecmd(union node **, union node *);
112 STATIC union node *makename(void);
113 STATIC void parsefname(void);
114 STATIC void parseheredoc(void);
115 STATIC int peektoken(void);
116 STATIC int readtoken(void);
117 STATIC int xxreadtoken(void);
118 STATIC int readtoken1(int, char const *, char *, int);
119 STATIC int noexpand(char *);
120 STATIC void synexpect(int) __dead;
121 STATIC void synerror(const char *) __dead;
122 STATIC void setprompt(int);
123
124
125 /*
126 * Read and parse a command. Returns NEOF on end of file. (NULL is a
127 * valid parse tree indicating a blank line.)
128 */
129
130 union node *
131 parsecmd(int interact)
132 {
133 int t;
134
135 tokpushback = 0;
136 doprompt = interact;
137 if (doprompt)
138 setprompt(1);
139 else
140 setprompt(0);
141 needprompt = 0;
142 t = readtoken();
143 if (t == TEOF)
144 return NEOF;
145 if (t == TNL)
146 return NULL;
147 tokpushback++;
148 return list(1, 0);
149 }
150
151
152 STATIC union node *
153 list(int nlflag, int erflag)
154 {
155 union node *n1, *n2, *n3;
156 int tok;
157 TRACE(("list: entered\n"));
158
159 checkkwd = 2;
160 if (nlflag == 0 && tokendlist[peektoken()])
161 return NULL;
162 n1 = NULL;
163 for (;;) {
164 n2 = andor();
165 tok = readtoken();
166 if (tok == TBACKGND) {
167 if (n2->type == NCMD || n2->type == NPIPE) {
168 n2->ncmd.backgnd = 1;
169 } else if (n2->type == NREDIR) {
170 n2->type = NBACKGND;
171 } else {
172 n3 = (union node *)stalloc(sizeof (struct nredir));
173 n3->type = NBACKGND;
174 n3->nredir.n = n2;
175 n3->nredir.redirect = NULL;
176 n2 = n3;
177 }
178 }
179 if (n1 == NULL) {
180 n1 = n2;
181 }
182 else {
183 n3 = (union node *)stalloc(sizeof (struct nbinary));
184 n3->type = NSEMI;
185 n3->nbinary.ch1 = n1;
186 n3->nbinary.ch2 = n2;
187 n1 = n3;
188 }
189 switch (tok) {
190 case TBACKGND:
191 case TSEMI:
192 tok = readtoken();
193 /* fall through */
194 case TNL:
195 if (tok == TNL) {
196 parseheredoc();
197 if (nlflag)
198 return n1;
199 } else {
200 tokpushback++;
201 }
202 checkkwd = 2;
203 if (tokendlist[peektoken()])
204 return n1;
205 break;
206 case TEOF:
207 if (heredoclist)
208 parseheredoc();
209 else
210 pungetc(); /* push back EOF on input */
211 return n1;
212 default:
213 if (nlflag || erflag)
214 synexpect(-1);
215 tokpushback++;
216 return n1;
217 }
218 }
219 }
220
221
222
223 STATIC union node *
224 andor(void)
225 {
226 union node *n1, *n2, *n3;
227 int t;
228
229 TRACE(("andor: entered\n"));
230 n1 = pipeline();
231 for (;;) {
232 if ((t = readtoken()) == TAND) {
233 t = NAND;
234 } else if (t == TOR) {
235 t = NOR;
236 } else {
237 tokpushback++;
238 return n1;
239 }
240 n2 = pipeline();
241 n3 = (union node *)stalloc(sizeof (struct nbinary));
242 n3->type = t;
243 n3->nbinary.ch1 = n1;
244 n3->nbinary.ch2 = n2;
245 n1 = n3;
246 }
247 }
248
249
250
251 STATIC union node *
252 pipeline(void)
253 {
254 union node *n1, *n2, *pipenode;
255 struct nodelist *lp, *prev;
256 int negate;
257
258 TRACE(("pipeline: entered\n"));
259
260 negate = 0;
261 checkkwd = 2;
262 while (readtoken() == TNOT) {
263 TRACE(("pipeline: TNOT recognized\n"));
264 negate = !negate;
265 }
266 tokpushback++;
267 n1 = command();
268 if (readtoken() == TPIPE) {
269 pipenode = (union node *)stalloc(sizeof (struct npipe));
270 pipenode->type = NPIPE;
271 pipenode->npipe.backgnd = 0;
272 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
273 pipenode->npipe.cmdlist = lp;
274 lp->n = n1;
275 do {
276 prev = lp;
277 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
278 lp->n = command();
279 prev->next = lp;
280 } while (readtoken() == TPIPE);
281 lp->next = NULL;
282 n1 = pipenode;
283 }
284 tokpushback++;
285 if (negate) {
286 TRACE(("negate pipeline\n"));
287 n2 = (union node *)stalloc(sizeof (struct nnot));
288 n2->type = NNOT;
289 n2->nnot.com = n1;
290 return n2;
291 } else
292 return n1;
293 }
294
295
296
297 STATIC union node *
298 command(void)
299 {
300 union node *n1, *n2;
301 union node *ap, **app;
302 union node *cp, **cpp;
303 union node *redir, **rpp;
304 int t, negate = 0;
305
306 TRACE(("command: entered\n"));
307
308 checkkwd = 2;
309 redir = NULL;
310 n1 = NULL;
311 rpp = &redir;
312
313 /* Check for redirection which may precede command */
314 while (readtoken() == TREDIR) {
315 *rpp = n2 = redirnode;
316 rpp = &n2->nfile.next;
317 parsefname();
318 }
319 tokpushback++;
320
321 while (readtoken() == TNOT) {
322 TRACE(("command: TNOT recognized\n"));
323 negate = !negate;
324 }
325 tokpushback++;
326
327 switch (readtoken()) {
328 case TIF:
329 n1 = (union node *)stalloc(sizeof (struct nif));
330 n1->type = NIF;
331 n1->nif.test = list(0, 0);
332 if (readtoken() != TTHEN)
333 synexpect(TTHEN);
334 n1->nif.ifpart = list(0, 0);
335 n2 = n1;
336 while (readtoken() == TELIF) {
337 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif));
338 n2 = n2->nif.elsepart;
339 n2->type = NIF;
340 n2->nif.test = list(0, 0);
341 if (readtoken() != TTHEN)
342 synexpect(TTHEN);
343 n2->nif.ifpart = list(0, 0);
344 }
345 if (lasttoken == TELSE)
346 n2->nif.elsepart = list(0, 0);
347 else {
348 n2->nif.elsepart = NULL;
349 tokpushback++;
350 }
351 if (readtoken() != TFI)
352 synexpect(TFI);
353 checkkwd = 1;
354 break;
355 case TWHILE:
356 case TUNTIL: {
357 int got;
358 n1 = (union node *)stalloc(sizeof (struct nbinary));
359 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL;
360 n1->nbinary.ch1 = list(0, 0);
361 if ((got=readtoken()) != TDO) {
362 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : ""));
363 synexpect(TDO);
364 }
365 n1->nbinary.ch2 = list(0, 0);
366 if (readtoken() != TDONE)
367 synexpect(TDONE);
368 checkkwd = 1;
369 break;
370 }
371 case TFOR:
372 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext))
373 synerror("Bad for loop variable");
374 n1 = (union node *)stalloc(sizeof (struct nfor));
375 n1->type = NFOR;
376 n1->nfor.var = wordtext;
377 if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) {
378 app = ≈
379 while (readtoken() == TWORD) {
380 n2 = (union node *)stalloc(sizeof (struct narg));
381 n2->type = NARG;
382 n2->narg.text = wordtext;
383 n2->narg.backquote = backquotelist;
384 *app = n2;
385 app = &n2->narg.next;
386 }
387 *app = NULL;
388 n1->nfor.args = ap;
389 if (lasttoken != TNL && lasttoken != TSEMI)
390 synexpect(-1);
391 } else {
392 static char argvars[5] = {
393 CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'
394 };
395 n2 = (union node *)stalloc(sizeof (struct narg));
396 n2->type = NARG;
397 n2->narg.text = argvars;
398 n2->narg.backquote = NULL;
399 n2->narg.next = NULL;
400 n1->nfor.args = n2;
401 /*
402 * Newline or semicolon here is optional (but note
403 * that the original Bourne shell only allowed NL).
404 */
405 if (lasttoken != TNL && lasttoken != TSEMI)
406 tokpushback++;
407 }
408 checkkwd = 2;
409 if ((t = readtoken()) == TDO)
410 t = TDONE;
411 else if (t == TBEGIN)
412 t = TEND;
413 else
414 synexpect(-1);
415 n1->nfor.body = list(0, 0);
416 if (readtoken() != t)
417 synexpect(t);
418 checkkwd = 1;
419 break;
420 case TCASE:
421 n1 = (union node *)stalloc(sizeof (struct ncase));
422 n1->type = NCASE;
423 if (readtoken() != TWORD)
424 synexpect(TWORD);
425 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg));
426 n2->type = NARG;
427 n2->narg.text = wordtext;
428 n2->narg.backquote = backquotelist;
429 n2->narg.next = NULL;
430 while (readtoken() == TNL);
431 if (lasttoken != TWORD || ! equal(wordtext, "in"))
432 synerror("expecting \"in\"");
433 cpp = &n1->ncase.cases;
434 noalias = 1;
435 checkkwd = 2, readtoken();
436 /*
437 * Both ksh and bash accept 'case x in esac'
438 * so configure scripts started taking advantage of this.
439 * The page: http://pubs.opengroup.org/onlinepubs/\
440 * 009695399/utilities/xcu_chap02.html contradicts itself,
441 * as to if this is legal; the "Case Conditional Format"
442 * paragraph shows one case is required, but the "Grammar"
443 * section shows a grammar that explicitly allows the no
444 * case option.
445 */
446 while (lasttoken != TESAC) {
447 *cpp = cp = (union node *)stalloc(sizeof (struct nclist));
448 if (lasttoken == TLP)
449 readtoken();
450 cp->type = NCLIST;
451 app = &cp->nclist.pattern;
452 for (;;) {
453 *app = ap = (union node *)stalloc(sizeof (struct narg));
454 ap->type = NARG;
455 ap->narg.text = wordtext;
456 ap->narg.backquote = backquotelist;
457 if (checkkwd = 2, readtoken() != TPIPE)
458 break;
459 app = &ap->narg.next;
460 readtoken();
461 }
462 ap->narg.next = NULL;
463 noalias = 0;
464 if (lasttoken != TRP) {
465 synexpect(TRP);
466 }
467 cp->nclist.body = list(0, 0);
468
469 checkkwd = 2;
470 if ((t = readtoken()) != TESAC) {
471 if (t != TENDCASE) {
472 noalias = 0;
473 synexpect(TENDCASE);
474 } else {
475 noalias = 1;
476 checkkwd = 2;
477 readtoken();
478 }
479 }
480 cpp = &cp->nclist.next;
481 }
482 noalias = 0;
483 *cpp = NULL;
484 checkkwd = 1;
485 break;
486 case TLP:
487 n1 = (union node *)stalloc(sizeof (struct nredir));
488 n1->type = NSUBSHELL;
489 n1->nredir.n = list(0, 0);
490 n1->nredir.redirect = NULL;
491 if (readtoken() != TRP)
492 synexpect(TRP);
493 checkkwd = 1;
494 break;
495 case TBEGIN:
496 n1 = list(0, 0);
497 if (readtoken() != TEND)
498 synexpect(TEND);
499 checkkwd = 1;
500 break;
501 /* Handle an empty command like other simple commands. */
502 case TSEMI:
503 /*
504 * An empty command before a ; doesn't make much sense, and
505 * should certainly be disallowed in the case of `if ;'.
506 */
507 if (!redir)
508 synexpect(-1);
509 case TAND:
510 case TOR:
511 case TNL:
512 case TEOF:
513 case TWORD:
514 case TRP:
515 tokpushback++;
516 n1 = simplecmd(rpp, redir);
517 goto checkneg;
518 case TENDCASE:
519 if (redir) {
520 tokpushback++;
521 goto checkneg;
522 }
523 /* FALLTHROUGH */
524 default:
525 synexpect(-1);
526 /* NOTREACHED */
527 }
528
529 /* Now check for redirection which may follow command */
530 while (readtoken() == TREDIR) {
531 *rpp = n2 = redirnode;
532 rpp = &n2->nfile.next;
533 parsefname();
534 }
535 tokpushback++;
536 *rpp = NULL;
537 if (redir) {
538 if (n1->type != NSUBSHELL) {
539 n2 = (union node *)stalloc(sizeof (struct nredir));
540 n2->type = NREDIR;
541 n2->nredir.n = n1;
542 n1 = n2;
543 }
544 n1->nredir.redirect = redir;
545 }
546
547 checkneg:
548 if (negate) {
549 TRACE(("negate command\n"));
550 n2 = (union node *)stalloc(sizeof (struct nnot));
551 n2->type = NNOT;
552 n2->nnot.com = n1;
553 return n2;
554 }
555 else
556 return n1;
557 }
558
559
560 STATIC union node *
561 simplecmd(union node **rpp, union node *redir)
562 {
563 union node *args, **app;
564 union node **orig_rpp = rpp;
565 union node *n = NULL, *n2;
566 int negate = 0;
567
568 /* If we don't have any redirections already, then we must reset */
569 /* rpp to be the address of the local redir variable. */
570 if (redir == 0)
571 rpp = &redir;
572
573 args = NULL;
574 app = &args;
575 /*
576 * We save the incoming value, because we need this for shell
577 * functions. There can not be a redirect or an argument between
578 * the function name and the open parenthesis.
579 */
580 orig_rpp = rpp;
581
582 while (readtoken() == TNOT) {
583 TRACE(("simplcmd: TNOT recognized\n"));
584 negate = !negate;
585 }
586 tokpushback++;
587
588 for (;;) {
589 if (readtoken() == TWORD) {
590 n = (union node *)stalloc(sizeof (struct narg));
591 n->type = NARG;
592 n->narg.text = wordtext;
593 n->narg.backquote = backquotelist;
594 *app = n;
595 app = &n->narg.next;
596 } else if (lasttoken == TREDIR) {
597 *rpp = n = redirnode;
598 rpp = &n->nfile.next;
599 parsefname(); /* read name of redirection file */
600 } else if (lasttoken == TLP && app == &args->narg.next
601 && rpp == orig_rpp) {
602 /* We have a function */
603 if (readtoken() != TRP)
604 synexpect(TRP);
605 funclinno = plinno;
606 rmescapes(n->narg.text);
607 if (!goodname(n->narg.text))
608 synerror("Bad function name");
609 n->type = NDEFUN;
610 n->narg.next = command();
611 funclinno = 0;
612 goto checkneg;
613 } else {
614 tokpushback++;
615 break;
616 }
617 }
618 *app = NULL;
619 *rpp = NULL;
620 n = (union node *)stalloc(sizeof (struct ncmd));
621 n->type = NCMD;
622 n->ncmd.backgnd = 0;
623 n->ncmd.args = args;
624 n->ncmd.redirect = redir;
625
626 checkneg:
627 if (negate) {
628 TRACE(("negate simplecmd\n"));
629 n2 = (union node *)stalloc(sizeof (struct nnot));
630 n2->type = NNOT;
631 n2->nnot.com = n;
632 return n2;
633 }
634 else
635 return n;
636 }
637
638 STATIC union node *
639 makename(void)
640 {
641 union node *n;
642
643 n = (union node *)stalloc(sizeof (struct narg));
644 n->type = NARG;
645 n->narg.next = NULL;
646 n->narg.text = wordtext;
647 n->narg.backquote = backquotelist;
648 return n;
649 }
650
651 void
652 fixredir(union node *n, const char *text, int err)
653 {
654 TRACE(("Fix redir %s %d\n", text, err));
655 if (!err)
656 n->ndup.vname = NULL;
657
658 if (is_number(text))
659 n->ndup.dupfd = number(text);
660 else if (text[0] == '-' && text[1] == '\0')
661 n->ndup.dupfd = -1;
662 else {
663
664 if (err)
665 synerror("Bad fd number");
666 else
667 n->ndup.vname = makename();
668 }
669 }
670
671
672 STATIC void
673 parsefname(void)
674 {
675 union node *n = redirnode;
676
677 if (readtoken() != TWORD)
678 synexpect(-1);
679 if (n->type == NHERE) {
680 struct heredoc *here = heredoc;
681 struct heredoc *p;
682 int i;
683
684 if (quoteflag == 0)
685 n->type = NXHERE;
686 TRACE(("Here document %d\n", n->type));
687 if (here->striptabs) {
688 while (*wordtext == '\t')
689 wordtext++;
690 }
691 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN)
692 synerror("Illegal eof marker for << redirection");
693 rmescapes(wordtext);
694 here->eofmark = wordtext;
695 here->next = NULL;
696 if (heredoclist == NULL)
697 heredoclist = here;
698 else {
699 for (p = heredoclist ; p->next ; p = p->next)
700 continue;
701 p->next = here;
702 }
703 } else if (n->type == NTOFD || n->type == NFROMFD) {
704 fixredir(n, wordtext, 0);
705 } else {
706 n->nfile.fname = makename();
707 }
708 }
709
710
711 /*
712 * Input any here documents.
713 */
714
715 STATIC void
716 parseheredoc(void)
717 {
718 struct heredoc *here;
719 union node *n;
720
721 while (heredoclist) {
722 here = heredoclist;
723 heredoclist = here->next;
724 if (needprompt) {
725 setprompt(2);
726 needprompt = 0;
727 }
728 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX,
729 here->eofmark, here->striptabs);
730 n = (union node *)stalloc(sizeof (struct narg));
731 n->narg.type = NARG;
732 n->narg.next = NULL;
733 n->narg.text = wordtext;
734 n->narg.backquote = backquotelist;
735 here->here->nhere.doc = n;
736 }
737 }
738
739 STATIC int
740 peektoken(void)
741 {
742 int t;
743
744 t = readtoken();
745 tokpushback++;
746 return (t);
747 }
748
749 STATIC int
750 readtoken(void)
751 {
752 int t;
753 int savecheckkwd = checkkwd;
754 #ifdef DEBUG
755 int alreadyseen = tokpushback;
756 #endif
757 struct alias *ap;
758
759 top:
760 t = xxreadtoken();
761
762 if (checkkwd) {
763 /*
764 * eat newlines
765 */
766 if (checkkwd == 2) {
767 checkkwd = 0;
768 while (t == TNL) {
769 parseheredoc();
770 t = xxreadtoken();
771 }
772 } else
773 checkkwd = 0;
774 /*
775 * check for keywords and aliases
776 */
777 if (t == TWORD && !quoteflag) {
778 const char *const *pp;
779
780 for (pp = parsekwd; *pp; pp++) {
781 if (**pp == *wordtext && equal(*pp, wordtext)) {
782 lasttoken = t = pp -
783 parsekwd + KWDOFFSET;
784 TRACE(("keyword %s recognized\n", tokname[t]));
785 goto out;
786 }
787 }
788 if (!noalias &&
789 (ap = lookupalias(wordtext, 1)) != NULL) {
790 pushstring(ap->val, strlen(ap->val), ap);
791 checkkwd = savecheckkwd;
792 goto top;
793 }
794 }
795 out:
796 checkkwd = (t == TNOT) ? savecheckkwd : 0;
797 }
798 TRACE(("%stoken %s %s\n", alreadyseen ? "reread " : "", tokname[t], t == TWORD ? wordtext : ""));
799 return (t);
800 }
801
802
803 /*
804 * Read the next input token.
805 * If the token is a word, we set backquotelist to the list of cmds in
806 * backquotes. We set quoteflag to true if any part of the word was
807 * quoted.
808 * If the token is TREDIR, then we set redirnode to a structure containing
809 * the redirection.
810 * In all cases, the variable startlinno is set to the number of the line
811 * on which the token starts.
812 *
813 * [Change comment: here documents and internal procedures]
814 * [Readtoken shouldn't have any arguments. Perhaps we should make the
815 * word parsing code into a separate routine. In this case, readtoken
816 * doesn't need to have any internal procedures, but parseword does.
817 * We could also make parseoperator in essence the main routine, and
818 * have parseword (readtoken1?) handle both words and redirection.]
819 */
820
821 #define RETURN(token) return lasttoken = token
822
823 STATIC int
824 xxreadtoken(void)
825 {
826 int c;
827
828 if (tokpushback) {
829 tokpushback = 0;
830 return lasttoken;
831 }
832 if (needprompt) {
833 setprompt(2);
834 needprompt = 0;
835 }
836 startlinno = plinno;
837 for (;;) { /* until token or start of word found */
838 c = pgetc_macro();
839 switch (c) {
840 case ' ': case '\t':
841 continue;
842 case '#':
843 while ((c = pgetc()) != '\n' && c != PEOF)
844 continue;
845 pungetc();
846 continue;
847 case '\\':
848 switch (pgetc()) {
849 case '\n':
850 startlinno = ++plinno;
851 if (doprompt)
852 setprompt(2);
853 else
854 setprompt(0);
855 continue;
856 case PEOF:
857 RETURN(TEOF);
858 default:
859 pungetc();
860 break;
861 }
862 goto breakloop;
863 case '\n':
864 plinno++;
865 needprompt = doprompt;
866 RETURN(TNL);
867 case PEOF:
868 RETURN(TEOF);
869 case '&':
870 if (pgetc() == '&')
871 RETURN(TAND);
872 pungetc();
873 RETURN(TBACKGND);
874 case '|':
875 if (pgetc() == '|')
876 RETURN(TOR);
877 pungetc();
878 RETURN(TPIPE);
879 case ';':
880 if (pgetc() == ';')
881 RETURN(TENDCASE);
882 pungetc();
883 RETURN(TSEMI);
884 case '(':
885 RETURN(TLP);
886 case ')':
887 RETURN(TRP);
888 default:
889 goto breakloop;
890 }
891 }
892 breakloop:
893 return readtoken1(c, BASESYNTAX, NULL, 0);
894 #undef RETURN
895 }
896
897
898
899 /*
900 * If eofmark is NULL, read a word or a redirection symbol. If eofmark
901 * is not NULL, read a here document. In the latter case, eofmark is the
902 * word which marks the end of the document and striptabs is true if
903 * leading tabs should be stripped from the document. The argument firstc
904 * is the first character of the input token or document.
905 *
906 * Because C does not have internal subroutines, I have simulated them
907 * using goto's to implement the subroutine linkage. The following macros
908 * will run code that appears at the end of readtoken1.
909 */
910
911 /*
912 * We used to remember only the current syntax, variable nesting level,
913 * double quote state for each var nexting level, and arith nesting
914 * level (unrelated to var nesting) and one prev syntax when in arith
915 * syntax. This worked for simple cases, but can't handle arith inside
916 * var expansion inside arith inside var with some quoted and some not.
917 *
918 * Inspired by FreeBSD's implementation (though it was the obvious way)
919 * though implemented differently, we now have a stack that keeps track
920 * of what we are doing now, and what we were doing previously.
921 * Every time something changes, which will eventually end and should
922 * revert to the previous state, we push this stack, and then pop it
923 * again later (that is every ${} with an operator (to parse the word
924 * or pattern that follows) ${x} and $x are too * simple to need it)
925 * $(( )) $( ) and "...". Always. Really, always!
926 *
927 * The stack is implemented as one static (on the C stack) base block
928 * containing LEVELS_PER_BLOCK (8) stack entries, which should be
929 * enough for the vast majority of cases. For torture tests, we
930 * malloc more blocks as needed. All accesses through the inline
931 * functions below.
932 */
933
934 /*
935 * varnest & arinest will typically be 0 or 1
936 * (varnest can increment in usages like ${x=${y}} but probably
937 * does not really need to)
938 * parenlevel allows balancing parens inside a $(( )), it is reset
939 * at each new nesting level ( $(( ( x + 3 ${unset-)} )) does not work.
940 * quoted is special - we need to know 2 things ... are we inside "..."
941 * (even if inherited from some previous nesting level) and was there
942 * an opening '"' at this level (so the next will be closing).
943 * "..." can span nexting levels, but cannot be opened in one and
944 * closed in a different one.
945 * To handle this, "quoted" has two fields, the bottom 4 (really 2)
946 * bits are 0, 1, or 2, for un, single, and double quoted (single quoted
947 * is really so special that this setting is not very important)
948 * and 0x10 that indicates that an opening quote has been seen.
949 * The bottom 4 bits are inherited, the 0x10 bit is not.
950 */
951 struct tokenstate {
952 const char *ts_syntax;
953 unsigned short ts_parenlevel; /* counters */
954 unsigned short ts_varnest; /* 64000 levels should be enough! */
955 unsigned short ts_arinest;
956 unsigned short ts_quoted; /* 1 -> single, 2 -> double */
957 };
958
959 #define NQ 0x00
960 #define SQ 0x01
961 #define DQ 0x02
962 #define QF 0x0F
963 #define QS 0x10
964
965 #define LEVELS_PER_BLOCK 8
966 #define VSS volatile struct statestack
967
968 struct statestack {
969 VSS *prev; /* previous block in list */
970 int cur; /* which of our tokenstates is current */
971 struct tokenstate tokenstate[LEVELS_PER_BLOCK];
972 };
973
974 static inline volatile struct tokenstate *
975 currentstate(VSS *stack)
976 {
977 return &stack->tokenstate[stack->cur];
978 }
979
980 static inline volatile struct tokenstate *
981 prevstate(VSS *stack)
982 {
983 if (stack->cur != 0)
984 return &stack->tokenstate[stack->cur - 1];
985 if (stack->prev == NULL) /* cannot drop below base */
986 return &stack->tokenstate[0];
987 return &stack->prev->tokenstate[LEVELS_PER_BLOCK - 1];
988 }
989
990 static inline VSS *
991 bump_state_level(VSS *stack)
992 {
993 volatile struct tokenstate *os, *ts;
994
995 os = currentstate(stack);
996
997 if (++stack->cur >= LEVELS_PER_BLOCK) {
998 VSS *ss;
999
1000 ss = (VSS *)ckmalloc(sizeof (struct statestack));
1001 ss->cur = 0;
1002 ss->prev = stack;
1003 stack = ss;
1004 }
1005
1006 ts = currentstate(stack);
1007
1008 ts->ts_parenlevel = 0; /* parens inside never match outside */
1009
1010 ts->ts_quoted = os->ts_quoted & QF; /* these are default settings */
1011 ts->ts_varnest = os->ts_varnest;
1012 ts->ts_arinest = os->ts_arinest; /* when appropriate */
1013 ts->ts_syntax = os->ts_syntax; /* they will be altered */
1014
1015 return stack;
1016 }
1017
1018 static inline VSS *
1019 drop_state_level(VSS *stack)
1020 {
1021 if (stack->cur == 0) {
1022 VSS *ss;
1023
1024 ss = stack;
1025 stack = ss->prev;
1026 if (stack == NULL)
1027 return ss;
1028 ckfree(__UNVOLATILE(ss));
1029 }
1030 --stack->cur;
1031 return stack;
1032 }
1033
1034 static inline void
1035 cleanup_state_stack(VSS *stack)
1036 {
1037 while (stack->prev != NULL) {
1038 stack->cur = 0;
1039 stack = drop_state_level(stack);
1040 }
1041 }
1042
1043 #define CHECKEND() {goto checkend; checkend_return:;}
1044 #define PARSEREDIR() {goto parseredir; parseredir_return:;}
1045 #define PARSESUB() {goto parsesub; parsesub_return:;}
1046 #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;}
1047 #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;}
1048 #define PARSEARITH() {goto parsearith; parsearith_return:;}
1049
1050 /*
1051 * The following macros all assume the existance of a local var "stack"
1052 * which contains a pointer to the current struct stackstate
1053 */
1054
1055 /*
1056 * These are macros rather than inline funcs to avoid code churn as much
1057 * as possible - they replace macros of the same name used previously.
1058 */
1059 #define ISDBLQUOTE() (currentstate(stack)->ts_quoted & QS)
1060 #define SETDBLQUOTE() (currentstate(stack)->ts_quoted = QS | DQ)
1061 #define CLRDBLQUOTE() (currentstate(stack)->ts_quoted = \
1062 stack->cur != 0 || stack->prev ? \
1063 prevstate(stack)->ts_quoted & QF : 0)
1064
1065 /*
1066 * This set are just to avoid excess typing and line lengths...
1067 * The ones that "look like" var names must be implemented to be lvalues
1068 */
1069 #define syntax (currentstate(stack)->ts_syntax)
1070 #define parenlevel (currentstate(stack)->ts_parenlevel)
1071 #define varnest (currentstate(stack)->ts_varnest)
1072 #define arinest (currentstate(stack)->ts_arinest)
1073 #define quoted (currentstate(stack)->ts_quoted)
1074 #define TS_PUSH() (stack = bump_state_level(stack))
1075 #define TS_POP() (stack = drop_state_level(stack))
1076
1077 STATIC int
1078 readtoken1(int firstc, char const *syn, char *eofmark, int striptabs)
1079 {
1080 int c = firstc;
1081 char * volatile out;
1082 int len;
1083 char line[EOFMARKLEN + 1];
1084 struct nodelist *bqlist;
1085 volatile int quotef;
1086 volatile int oldstyle;
1087 VSS static_stack;
1088 VSS *stack = &static_stack;
1089
1090 stack->prev = NULL;
1091 stack->cur = 0;
1092
1093 syntax = syn;
1094
1095 startlinno = plinno;
1096 varnest = 0;
1097 if (syntax == DQSYNTAX) {
1098 SETDBLQUOTE();
1099 }
1100 quotef = 0;
1101 bqlist = NULL;
1102 arinest = 0;
1103 parenlevel = 0;
1104 quoted = 0;
1105
1106 STARTSTACKSTR(out);
1107 loop: { /* for each line, until end of word */
1108 #if ATTY
1109 if (c == '\034' && doprompt
1110 && attyset() && ! equal(termval(), "emacs")) {
1111 attyline();
1112 if (syntax == BASESYNTAX)
1113 return readtoken();
1114 c = pgetc();
1115 goto loop;
1116 }
1117 #endif
1118 CHECKEND(); /* set c to PEOF if at end of here document */
1119 for (;;) { /* until end of line or end of word */
1120 CHECKSTRSPACE(4, out); /* permit 4 calls to USTPUTC */
1121 switch(syntax[c]) {
1122 case CNL: /* '\n' */
1123 if (syntax == BASESYNTAX)
1124 goto endword; /* exit outer loop */
1125 USTPUTC(c, out);
1126 plinno++;
1127 if (doprompt)
1128 setprompt(2);
1129 else
1130 setprompt(0);
1131 c = pgetc();
1132 goto loop; /* continue outer loop */
1133 case CWORD:
1134 USTPUTC(c, out);
1135 break;
1136 case CCTL:
1137 if (eofmark == NULL || ISDBLQUOTE())
1138 USTPUTC(CTLESC, out);
1139 USTPUTC(c, out);
1140 break;
1141 case CBACK: /* backslash */
1142 c = pgetc();
1143 if (c == PEOF) {
1144 USTPUTC('\\', out);
1145 pungetc();
1146 break;
1147 }
1148 if (c == '\n') {
1149 plinno++;
1150 if (doprompt)
1151 setprompt(2);
1152 else
1153 setprompt(0);
1154 break;
1155 }
1156 quotef = 1;
1157 if (ISDBLQUOTE() && c != '\\' &&
1158 c != '`' && c != '$' &&
1159 (c != '"' || eofmark != NULL))
1160 USTPUTC('\\', out);
1161 if (SQSYNTAX[c] == CCTL)
1162 USTPUTC(CTLESC, out);
1163 else if (eofmark == NULL) {
1164 USTPUTC(CTLQUOTEMARK, out);
1165 USTPUTC(c, out);
1166 if (varnest != 0)
1167 USTPUTC(CTLQUOTEEND, out);
1168 break;
1169 }
1170 USTPUTC(c, out);
1171 break;
1172 case CSQUOTE:
1173 if (syntax != SQSYNTAX) {
1174 if (eofmark == NULL)
1175 USTPUTC(CTLQUOTEMARK, out);
1176 quotef = 1;
1177 TS_PUSH();
1178 syntax = SQSYNTAX;
1179 quoted = SQ;
1180 break;
1181 }
1182 if (eofmark != NULL && arinest == 0 &&
1183 varnest == 0) {
1184 /* Ignore inside quoted here document */
1185 USTPUTC(c, out);
1186 break;
1187 }
1188 /* End of single quotes... */
1189 TS_POP();
1190 if (syntax == BASESYNTAX && varnest != 0)
1191 USTPUTC(CTLQUOTEEND, out);
1192 break;
1193 case CDQUOTE:
1194 if (eofmark != NULL && arinest == 0 &&
1195 varnest == 0) {
1196 /* Ignore inside here document */
1197 USTPUTC(c, out);
1198 break;
1199 }
1200 quotef = 1;
1201 if (arinest) {
1202 if (ISDBLQUOTE()) {
1203 TS_POP();
1204 } else {
1205 TS_PUSH();
1206 syntax = DQSYNTAX;
1207 SETDBLQUOTE();
1208 USTPUTC(CTLQUOTEMARK, out);
1209 }
1210 break;
1211 }
1212 if (eofmark != NULL)
1213 break;
1214 if (ISDBLQUOTE()) {
1215 TS_POP();
1216 if (varnest != 0)
1217 USTPUTC(CTLQUOTEEND, out);
1218 } else {
1219 TS_PUSH();
1220 syntax = DQSYNTAX;
1221 SETDBLQUOTE();
1222 USTPUTC(CTLQUOTEMARK, out);
1223 }
1224 break;
1225 case CVAR: /* '$' */
1226 PARSESUB(); /* parse substitution */
1227 break;
1228 case CENDVAR: /* CLOSEBRACE */
1229 if (varnest > 0 && !ISDBLQUOTE()) {
1230 TS_POP();
1231 USTPUTC(CTLENDVAR, out);
1232 } else {
1233 USTPUTC(c, out);
1234 }
1235 break;
1236 case CLP: /* '(' in arithmetic */
1237 parenlevel++;
1238 USTPUTC(c, out);
1239 break;
1240 case CRP: /* ')' in arithmetic */
1241 if (parenlevel > 0) {
1242 USTPUTC(c, out);
1243 --parenlevel;
1244 } else {
1245 if (pgetc() == ')') {
1246 if (--arinest == 0) {
1247 TS_POP();
1248 USTPUTC(CTLENDARI, out);
1249 } else
1250 USTPUTC(')', out);
1251 } else {
1252 /*
1253 * unbalanced parens
1254 * (don't 2nd guess - no error)
1255 */
1256 pungetc();
1257 USTPUTC(')', out);
1258 }
1259 }
1260 break;
1261 case CBQUOTE: /* '`' */
1262 PARSEBACKQOLD();
1263 break;
1264 case CEOF:
1265 goto endword; /* exit outer loop */
1266 default:
1267 if (varnest == 0 && !ISDBLQUOTE())
1268 goto endword; /* exit outer loop */
1269 USTPUTC(c, out);
1270 }
1271 c = pgetc_macro();
1272 }
1273 }
1274 endword:
1275 if (syntax == ARISYNTAX) {
1276 cleanup_state_stack(stack);
1277 synerror("Missing '))'");
1278 }
1279 if (syntax != BASESYNTAX && /* ! parsebackquote && */ eofmark == NULL) {
1280 cleanup_state_stack(stack);
1281 synerror("Unterminated quoted string");
1282 }
1283 if (varnest != 0) {
1284 cleanup_state_stack(stack);
1285 startlinno = plinno;
1286 /* { */
1287 synerror("Missing '}'");
1288 }
1289 USTPUTC('\0', out);
1290 len = out - stackblock();
1291 out = stackblock();
1292 if (eofmark == NULL) {
1293 if ((c == '>' || c == '<')
1294 && quotef == 0
1295 && (*out == '\0' || is_number(out))) {
1296 PARSEREDIR();
1297 cleanup_state_stack(stack);
1298 return lasttoken = TREDIR;
1299 } else {
1300 pungetc();
1301 }
1302 }
1303 quoteflag = quotef;
1304 backquotelist = bqlist;
1305 grabstackblock(len);
1306 wordtext = out;
1307 cleanup_state_stack(stack);
1308 return lasttoken = TWORD;
1309 /* end of readtoken routine */
1310
1311
1312
1313 /*
1314 * Check to see whether we are at the end of the here document. When this
1315 * is called, c is set to the first character of the next input line. If
1316 * we are at the end of the here document, this routine sets the c to PEOF.
1317 */
1318
1319 checkend: {
1320 if (eofmark) {
1321 if (striptabs) {
1322 while (c == '\t')
1323 c = pgetc();
1324 }
1325 if (c == *eofmark) {
1326 if (pfgets(line, sizeof line) != NULL) {
1327 char *p, *q;
1328
1329 p = line;
1330 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++)
1331 continue;
1332 if ((*p == '\0' || *p == '\n') && *q == '\0') {
1333 c = PEOF;
1334 plinno++;
1335 needprompt = doprompt;
1336 } else {
1337 pushstring(line, strlen(line), NULL);
1338 }
1339 }
1340 }
1341 }
1342 goto checkend_return;
1343 }
1344
1345
1346 /*
1347 * Parse a redirection operator. The variable "out" points to a string
1348 * specifying the fd to be redirected. The variable "c" contains the
1349 * first character of the redirection operator.
1350 */
1351
1352 parseredir: {
1353 char fd[64];
1354 union node *np;
1355 strlcpy(fd, out, sizeof(fd));
1356
1357 np = (union node *)stalloc(sizeof (struct nfile));
1358 if (c == '>') {
1359 np->nfile.fd = 1;
1360 c = pgetc();
1361 if (c == '>')
1362 np->type = NAPPEND;
1363 else if (c == '|')
1364 np->type = NCLOBBER;
1365 else if (c == '&')
1366 np->type = NTOFD;
1367 else {
1368 np->type = NTO;
1369 pungetc();
1370 }
1371 } else { /* c == '<' */
1372 np->nfile.fd = 0;
1373 switch (c = pgetc()) {
1374 case '<':
1375 if (sizeof (struct nfile) != sizeof (struct nhere)) {
1376 np = (union node *)stalloc(sizeof (struct nhere));
1377 np->nfile.fd = 0;
1378 }
1379 np->type = NHERE;
1380 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc));
1381 heredoc->here = np;
1382 if ((c = pgetc()) == '-') {
1383 heredoc->striptabs = 1;
1384 } else {
1385 heredoc->striptabs = 0;
1386 pungetc();
1387 }
1388 break;
1389
1390 case '&':
1391 np->type = NFROMFD;
1392 break;
1393
1394 case '>':
1395 np->type = NFROMTO;
1396 break;
1397
1398 default:
1399 np->type = NFROM;
1400 pungetc();
1401 break;
1402 }
1403 }
1404 if (*fd != '\0')
1405 np->nfile.fd = number(fd);
1406 redirnode = np;
1407 goto parseredir_return;
1408 }
1409
1410
1411 /*
1412 * Parse a substitution. At this point, we have read the dollar sign
1413 * and nothing else.
1414 */
1415
1416 parsesub: {
1417 char buf[10];
1418 int subtype;
1419 int typeloc;
1420 int flags;
1421 char *p;
1422 static const char types[] = "}-+?=";
1423 int i;
1424 int linno;
1425
1426 c = pgetc();
1427 if (c != '(' && c != OPENBRACE && !is_name(c) && !is_special(c)) {
1428 USTPUTC('$', out);
1429 pungetc();
1430 } else if (c == '(') { /* $(command) or $((arith)) */
1431 if (pgetc() == '(') {
1432 PARSEARITH();
1433 } else {
1434 pungetc();
1435 PARSEBACKQNEW();
1436 }
1437 } else {
1438 USTPUTC(CTLVAR, out);
1439 typeloc = out - stackblock();
1440 USTPUTC(VSNORMAL, out);
1441 subtype = VSNORMAL;
1442 flags = 0;
1443 if (c == OPENBRACE) {
1444 c = pgetc();
1445 if (c == '#') {
1446 if ((c = pgetc()) == CLOSEBRACE)
1447 c = '#';
1448 else
1449 subtype = VSLENGTH;
1450 }
1451 else
1452 subtype = 0;
1453 }
1454 if (is_name(c)) {
1455 p = out;
1456 do {
1457 STPUTC(c, out);
1458 c = pgetc();
1459 } while (is_in_name(c));
1460 if (out - p == 6 && strncmp(p, "LINENO", 6) == 0) {
1461 /* Replace the variable name with the
1462 * current line number. */
1463 linno = plinno;
1464 if (funclinno != 0)
1465 linno -= funclinno - 1;
1466 snprintf(buf, sizeof(buf), "%d", linno);
1467 STADJUST(-6, out);
1468 for (i = 0; buf[i] != '\0'; i++)
1469 STPUTC(buf[i], out);
1470 flags |= VSLINENO;
1471 }
1472 } else if (is_digit(c)) {
1473 do {
1474 USTPUTC(c, out);
1475 c = pgetc();
1476 } while (is_digit(c));
1477 }
1478 else if (is_special(c)) {
1479 USTPUTC(c, out);
1480 c = pgetc();
1481 }
1482 else {
1483 badsub:
1484 cleanup_state_stack(stack);
1485 synerror("Bad substitution");
1486 }
1487
1488 STPUTC('=', out);
1489 if (subtype == 0) {
1490 switch (c) {
1491 case ':':
1492 flags |= VSNUL;
1493 c = pgetc();
1494 /*FALLTHROUGH*/
1495 default:
1496 p = strchr(types, c);
1497 if (p == NULL)
1498 goto badsub;
1499 subtype = p - types + VSNORMAL;
1500 break;
1501 case '%':
1502 case '#':
1503 {
1504 int cc = c;
1505 subtype = c == '#' ? VSTRIMLEFT :
1506 VSTRIMRIGHT;
1507 c = pgetc();
1508 if (c == cc)
1509 subtype++;
1510 else
1511 pungetc();
1512 break;
1513 }
1514 }
1515 } else {
1516 pungetc();
1517 }
1518 if (ISDBLQUOTE() || arinest)
1519 flags |= VSQUOTE;
1520 *(stackblock() + typeloc) = subtype | flags;
1521 if (subtype != VSNORMAL) {
1522 TS_PUSH();
1523 varnest++;
1524 arinest = 0;
1525 if (subtype > VSASSIGN) { /* # ## % %% */
1526 syntax = BASESYNTAX;
1527 CLRDBLQUOTE();
1528 }
1529 }
1530 }
1531 goto parsesub_return;
1532 }
1533
1534
1535 /*
1536 * Called to parse command substitutions. Newstyle is set if the command
1537 * is enclosed inside $(...); nlpp is a pointer to the head of the linked
1538 * list of commands (passed by reference), and savelen is the number of
1539 * characters on the top of the stack which must be preserved.
1540 */
1541
1542 parsebackq: {
1543 struct nodelist **nlpp;
1544 int savepbq;
1545 union node *n;
1546 char *volatile str = NULL;
1547 struct jmploc jmploc;
1548 struct jmploc *volatile savehandler = NULL;
1549 int savelen;
1550 int saveprompt;
1551
1552 savepbq = parsebackquote;
1553 if (setjmp(jmploc.loc)) {
1554 if (str)
1555 ckfree(str);
1556 cleanup_state_stack(stack);
1557 parsebackquote = 0;
1558 handler = savehandler;
1559 longjmp(handler->loc, 1);
1560 }
1561 INTOFF;
1562 str = NULL;
1563 savelen = out - stackblock();
1564 if (savelen > 0) {
1565 str = ckmalloc(savelen);
1566 memcpy(str, stackblock(), savelen);
1567 }
1568 savehandler = handler;
1569 handler = &jmploc;
1570 INTON;
1571 if (oldstyle) {
1572 /* We must read until the closing backquote, giving special
1573 treatment to some slashes, and then push the string and
1574 reread it as input, interpreting it normally. */
1575 char *pout;
1576 int pc;
1577 int psavelen;
1578 char *pstr;
1579
1580 /*
1581 * Because the entire `...` is read here, we don't
1582 * need to bother the state stack. That will be used
1583 * (as appropriate) when the processed string is re-read.
1584 */
1585 STARTSTACKSTR(pout);
1586 for (;;) {
1587 if (needprompt) {
1588 setprompt(2);
1589 needprompt = 0;
1590 }
1591 switch (pc = pgetc()) {
1592 case '`':
1593 goto done;
1594
1595 case '\\':
1596 if ((pc = pgetc()) == '\n') {
1597 plinno++;
1598 if (doprompt)
1599 setprompt(2);
1600 else
1601 setprompt(0);
1602 /*
1603 * If eating a newline, avoid putting
1604 * the newline into the new character
1605 * stream (via the STPUTC after the
1606 * switch).
1607 */
1608 continue;
1609 }
1610 if (pc != '\\' && pc != '`' && pc != '$'
1611 && (!ISDBLQUOTE() || pc != '"'))
1612 STPUTC('\\', pout);
1613 break;
1614
1615 case '\n':
1616 plinno++;
1617 needprompt = doprompt;
1618 break;
1619
1620 case PEOF:
1621 startlinno = plinno;
1622 synerror("EOF in backquote substitution");
1623 break;
1624
1625 default:
1626 break;
1627 }
1628 STPUTC(pc, pout);
1629 }
1630 done:
1631 STPUTC('\0', pout);
1632 psavelen = pout - stackblock();
1633 if (psavelen > 0) {
1634 pstr = grabstackstr(pout);
1635 setinputstring(pstr, 1);
1636 }
1637 }
1638 nlpp = &bqlist;
1639 while (*nlpp)
1640 nlpp = &(*nlpp)->next;
1641 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist));
1642 (*nlpp)->next = NULL;
1643 parsebackquote = oldstyle;
1644
1645 if (oldstyle) {
1646 saveprompt = doprompt;
1647 doprompt = 0;
1648 } else
1649 saveprompt = 0;
1650
1651 n = list(0, oldstyle);
1652
1653 if (oldstyle)
1654 doprompt = saveprompt;
1655 else {
1656 if (readtoken() != TRP) {
1657 cleanup_state_stack(stack);
1658 synexpect(TRP);
1659 }
1660 }
1661
1662 (*nlpp)->n = n;
1663 if (oldstyle) {
1664 /*
1665 * Start reading from old file again, ignoring any pushed back
1666 * tokens left from the backquote parsing
1667 */
1668 popfile();
1669 tokpushback = 0;
1670 }
1671 while (stackblocksize() <= savelen)
1672 growstackblock();
1673 STARTSTACKSTR(out);
1674 if (str) {
1675 memcpy(out, str, savelen);
1676 STADJUST(savelen, out);
1677 INTOFF;
1678 ckfree(str);
1679 str = NULL;
1680 INTON;
1681 }
1682 parsebackquote = savepbq;
1683 handler = savehandler;
1684 if (arinest || ISDBLQUOTE())
1685 USTPUTC(CTLBACKQ | CTLQUOTE, out);
1686 else
1687 USTPUTC(CTLBACKQ, out);
1688 if (oldstyle)
1689 goto parsebackq_oldreturn;
1690 else
1691 goto parsebackq_newreturn;
1692 }
1693
1694 /*
1695 * Parse an arithmetic expansion (indicate start of one and set state)
1696 */
1697 parsearith: {
1698
1699 if (syntax == ARISYNTAX) {
1700 /*
1701 * we collapse embedded arithmetic expansion to
1702 * parentheses, which should be equivalent
1703 */
1704 USTPUTC('(', out);
1705 USTPUTC('(', out);
1706 /*
1707 * Need 2 of them because there will (should be)
1708 * two closing ))'s to follow later.
1709 */
1710 parenlevel += 2;
1711 } else {
1712 TS_PUSH();
1713 syntax = ARISYNTAX;
1714 ++arinest;
1715 varnest = 0;
1716
1717 USTPUTC(CTLARI, out);
1718 if (ISDBLQUOTE())
1719 USTPUTC('"',out);
1720 else
1721 USTPUTC(' ',out);
1722 }
1723 goto parsearith_return;
1724 }
1725
1726 } /* end of readtoken */
1727
1728
1729
1730 #ifdef mkinit
1731 RESET {
1732 tokpushback = 0;
1733 checkkwd = 0;
1734 }
1735 #endif
1736
1737 /*
1738 * Returns true if the text contains nothing to expand (no dollar signs
1739 * or backquotes).
1740 */
1741
1742 STATIC int
1743 noexpand(char *text)
1744 {
1745 char *p;
1746 char c;
1747
1748 p = text;
1749 while ((c = *p++) != '\0') {
1750 if (c == CTLQUOTEMARK)
1751 continue;
1752 if (c == CTLESC)
1753 p++;
1754 else if (BASESYNTAX[(int)c] == CCTL)
1755 return 0;
1756 }
1757 return 1;
1758 }
1759
1760
1761 /*
1762 * Return true if the argument is a legal variable name (a letter or
1763 * underscore followed by zero or more letters, underscores, and digits).
1764 */
1765
1766 int
1767 goodname(char *name)
1768 {
1769 char *p;
1770
1771 p = name;
1772 if (! is_name(*p))
1773 return 0;
1774 while (*++p) {
1775 if (! is_in_name(*p))
1776 return 0;
1777 }
1778 return 1;
1779 }
1780
1781
1782 /*
1783 * Called when an unexpected token is read during the parse. The argument
1784 * is the token that is expected, or -1 if more than one type of token can
1785 * occur at this point.
1786 */
1787
1788 STATIC void
1789 synexpect(int token)
1790 {
1791 char msg[64];
1792
1793 if (token >= 0) {
1794 fmtstr(msg, 64, "%s unexpected (expecting %s)",
1795 tokname[lasttoken], tokname[token]);
1796 } else {
1797 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]);
1798 }
1799 synerror(msg);
1800 /* NOTREACHED */
1801 }
1802
1803
1804 STATIC void
1805 synerror(const char *msg)
1806 {
1807 if (commandname)
1808 outfmt(&errout, "%s: %d: ", commandname, startlinno);
1809 else
1810 outfmt(&errout, "%s: ", getprogname());
1811 outfmt(&errout, "Syntax error: %s\n", msg);
1812 error(NULL);
1813 /* NOTREACHED */
1814 }
1815
1816 STATIC void
1817 setprompt(int which)
1818 {
1819 whichprompt = which;
1820
1821 #ifndef SMALL
1822 if (!el)
1823 #endif
1824 out2str(getprompt(NULL));
1825 }
1826
1827 /*
1828 * called by editline -- any expansions to the prompt
1829 * should be added here.
1830 */
1831 const char *
1832 getprompt(void *unused)
1833 {
1834 switch (whichprompt) {
1835 case 0:
1836 return "";
1837 case 1:
1838 return ps1val();
1839 case 2:
1840 return ps2val();
1841 default:
1842 return "<internal prompt error>";
1843 }
1844 }
1845