HAMMER VFS - Major retooling of the refcount mechanics, and fix a deadlock
[dragonfly.git] / bin / sh / parser.c
blob10ca2c0f6fd9cc50dbfd71d2fce757fcacb5d1c2
1 /*-
2 * Copyright (c) 1991, 1993
3 * The Regents of the University of California. All rights reserved.
5 * This code is derived from software contributed to Berkeley by
6 * Kenneth Almquist.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. All advertising materials mentioning features or use of this software
17 * must display the following acknowledgement:
18 * This product includes software developed by the University of
19 * California, Berkeley and its contributors.
20 * 4. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * SUCH DAMAGE.
36 * @(#)parser.c 8.7 (Berkeley) 5/16/95
37 * $FreeBSD: src/bin/sh/parser.c,v 1.58 2006/11/05 18:36:05 stefanf Exp $
38 * $DragonFly: src/bin/sh/parser.c,v 1.12 2007/01/18 17:03:18 corecode Exp $
41 #include <stdio.h>
42 #include <stdlib.h>
43 #include <unistd.h>
45 #include "shell.h"
46 #include "parser.h"
47 #include "nodes.h"
48 #include "expand.h" /* defines rmescapes() */
49 #include "syntax.h"
50 #include "options.h"
51 #include "input.h"
52 #include "output.h"
53 #include "var.h"
54 #include "error.h"
55 #include "memalloc.h"
56 #include "mystring.h"
57 #include "alias.h"
58 #include "show.h"
59 #include "eval.h"
60 #ifndef NO_HISTORY
61 #include "myhistedit.h"
62 #endif
65 * Shell command parser.
68 #define EOFMARKLEN 79
69 #define PROMPTLEN 128
71 /* values returned by readtoken */
72 #include "token.h"
76 struct heredoc {
77 struct heredoc *next; /* next here document in list */
78 union node *here; /* redirection node */
79 char *eofmark; /* string indicating end of input */
80 int striptabs; /* if set, strip leading tabs */
85 STATIC struct heredoc *heredoclist; /* list of here documents to read */
86 STATIC int parsebackquote; /* nonzero if we are inside backquotes */
87 STATIC int doprompt; /* if set, prompt the user */
88 STATIC int needprompt; /* true if interactive and at start of line */
89 STATIC int lasttoken; /* last token read */
90 MKINIT int tokpushback; /* last token pushed back */
91 STATIC char *wordtext; /* text of last word returned by readtoken */
93 * 1 == check for kwds
94 * 2 == also eat newlines
95 * 3 == check for TNOT
97 MKINIT int checkkwd;
98 STATIC struct nodelist *backquotelist;
99 STATIC union node *redirnode;
100 STATIC struct heredoc *heredoc;
101 STATIC int quoteflag; /* set if (part of) last token was quoted */
102 STATIC int startlinno; /* line # where last token started */
103 STATIC int funclinno; /* line # where the current function started */
105 /* XXX When 'noaliases' is set to one, no alias expansion takes place. */
106 static int noaliases = 0;
109 STATIC union node *list(int);
110 STATIC union node *andor(void);
111 STATIC union node *pipeline(void);
112 STATIC union node *command(void);
113 STATIC union node *simplecmd(union node **, union node *);
114 STATIC union node *makename(void);
115 STATIC void parsefname(void);
116 STATIC void parseheredoc(void);
117 STATIC int peektoken(void);
118 STATIC int readtoken(void);
119 STATIC int xxreadtoken(void);
120 STATIC int readtoken1(int, char const *, char *, int);
121 STATIC int noexpand(char *);
122 STATIC void synexpect(int);
123 STATIC void synerror(const char *);
124 STATIC void setprompt(int);
128 * Read and parse a command. Returns NEOF on end of file. (NULL is a
129 * valid parse tree indicating a blank line.)
132 union node *
133 parsecmd(int interact)
135 int t;
137 tokpushback = 0;
138 doprompt = interact;
139 if (doprompt)
140 setprompt(1);
141 else
142 setprompt(0);
143 needprompt = 0;
144 t = readtoken();
145 if (t == TEOF)
146 return NEOF;
147 if (t == TNL)
148 return NULL;
149 tokpushback++;
150 return list(1);
154 STATIC union node *
155 list(int nlflag)
157 union node *n1, *n2, *n3;
158 int tok;
160 checkkwd = 2;
161 if (nlflag == 0 && tokendlist[peektoken()])
162 return NULL;
163 n1 = NULL;
164 for (;;) {
165 n2 = andor();
166 tok = readtoken();
167 if (tok == TBACKGND) {
168 if (n2->type == NCMD || n2->type == NPIPE) {
169 n2->ncmd.backgnd = 1;
170 } else if (n2->type == NREDIR) {
171 n2->type = NBACKGND;
172 } else {
173 n3 = (union node *)stalloc(sizeof (struct nredir));
174 n3->type = NBACKGND;
175 n3->nredir.n = n2;
176 n3->nredir.redirect = NULL;
177 n2 = n3;
180 if (n1 == NULL) {
181 n1 = n2;
183 else {
184 n3 = (union node *)stalloc(sizeof (struct nbinary));
185 n3->type = NSEMI;
186 n3->nbinary.ch1 = n1;
187 n3->nbinary.ch2 = n2;
188 n1 = n3;
190 switch (tok) {
191 case TBACKGND:
192 case TSEMI:
193 tok = readtoken();
194 /* FALLTHROUGH */
195 case TNL:
196 if (tok == TNL) {
197 parseheredoc();
198 if (nlflag)
199 return n1;
200 } else {
201 tokpushback++;
203 checkkwd = 2;
204 if (tokendlist[peektoken()])
205 return n1;
206 break;
207 case TEOF:
208 if (heredoclist)
209 parseheredoc();
210 else
211 pungetc(); /* push back EOF on input */
212 return n1;
213 default:
214 if (nlflag)
215 synexpect(-1);
216 tokpushback++;
217 return n1;
224 STATIC union node *
225 andor(void)
227 union node *n1, *n2, *n3;
228 int t;
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;
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;
251 STATIC union node *
252 pipeline(void)
254 union node *n1, *n2, *pipenode;
255 struct nodelist *lp, *prev;
256 int negate;
258 negate = 0;
259 TRACE(("pipeline: entered\n"));
261 checkkwd = 3;
262 while (readtoken() == TNOT)
263 negate = !negate;
264 tokpushback++;
266 n1 = command();
267 if (readtoken() == TPIPE) {
268 pipenode = (union node *)stalloc(sizeof (struct npipe));
269 pipenode->type = NPIPE;
270 pipenode->npipe.backgnd = 0;
271 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
272 pipenode->npipe.cmdlist = lp;
273 lp->n = n1;
274 do {
275 int innernegate = 0;
277 checkkwd = 3;
278 while (readtoken() == TNOT)
279 innernegate = !innernegate;
280 tokpushback++;
282 prev = lp;
283 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
284 lp->n = command();
285 prev->next = lp;
287 if (innernegate) {
288 n2 = (union node *)stalloc(sizeof (struct nnot));
289 n2->type = NNOT;
290 n2->nnot.com = lp->n;
291 lp->n = n2;
293 } while (readtoken() == TPIPE);
294 lp->next = NULL;
295 n1 = pipenode;
297 tokpushback++;
298 if (negate) {
299 n2 = (union node *)stalloc(sizeof (struct nnot));
300 n2->type = NNOT;
301 n2->nnot.com = n1;
302 return n2;
303 } else
304 return n1;
309 STATIC union node *
310 command(void)
312 union node *n1, *n2;
313 union node *ap, **app;
314 union node *cp, **cpp;
315 union node *redir, **rpp;
316 int t;
318 checkkwd = 2;
319 redir = NULL;
320 n1 = NULL;
321 rpp = &redir;
323 /* Check for redirection which may precede command */
324 while (readtoken() == TREDIR) {
325 *rpp = n2 = redirnode;
326 rpp = &n2->nfile.next;
327 parsefname();
329 tokpushback++;
331 switch (readtoken()) {
332 case TIF:
333 n1 = (union node *)stalloc(sizeof (struct nif));
334 n1->type = NIF;
335 if ((n1->nif.test = list(0)) == NULL)
336 synexpect(-1);
337 if (readtoken() != TTHEN)
338 synexpect(TTHEN);
339 n1->nif.ifpart = list(0);
340 n2 = n1;
341 while (readtoken() == TELIF) {
342 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif));
343 n2 = n2->nif.elsepart;
344 n2->type = NIF;
345 if ((n2->nif.test = list(0)) == NULL)
346 synexpect(-1);
347 if (readtoken() != TTHEN)
348 synexpect(TTHEN);
349 n2->nif.ifpart = list(0);
351 if (lasttoken == TELSE)
352 n2->nif.elsepart = list(0);
353 else {
354 n2->nif.elsepart = NULL;
355 tokpushback++;
357 if (readtoken() != TFI)
358 synexpect(TFI);
359 checkkwd = 1;
360 break;
361 case TWHILE:
362 case TUNTIL: {
363 int got;
364 n1 = (union node *)stalloc(sizeof (struct nbinary));
365 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL;
366 if ((n1->nbinary.ch1 = list(0)) == NULL)
367 synexpect(-1);
368 if ((got=readtoken()) != TDO) {
369 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : ""));
370 synexpect(TDO);
372 n1->nbinary.ch2 = list(0);
373 if (readtoken() != TDONE)
374 synexpect(TDONE);
375 checkkwd = 1;
376 break;
378 case TFOR:
379 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext))
380 synerror("Bad for loop variable");
381 n1 = (union node *)stalloc(sizeof (struct nfor));
382 n1->type = NFOR;
383 n1->nfor.var = wordtext;
384 if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) {
385 app = &ap;
386 while (readtoken() == TWORD) {
387 n2 = (union node *)stalloc(sizeof (struct narg));
388 n2->type = NARG;
389 n2->narg.text = wordtext;
390 n2->narg.backquote = backquotelist;
391 *app = n2;
392 app = &n2->narg.next;
394 *app = NULL;
395 n1->nfor.args = ap;
396 if (lasttoken != TNL && lasttoken != TSEMI)
397 synexpect(-1);
398 } else {
399 static char argvars[5] = {
400 CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'
402 n2 = (union node *)stalloc(sizeof (struct narg));
403 n2->type = NARG;
404 n2->narg.text = argvars;
405 n2->narg.backquote = NULL;
406 n2->narg.next = NULL;
407 n1->nfor.args = n2;
409 * Newline or semicolon here is optional (but note
410 * that the original Bourne shell only allowed NL).
412 if (lasttoken != TNL && lasttoken != TSEMI)
413 tokpushback++;
415 checkkwd = 2;
416 if ((t = readtoken()) == TDO)
417 t = TDONE;
418 else if (t == TBEGIN)
419 t = TEND;
420 else
421 synexpect(-1);
422 n1->nfor.body = list(0);
423 if (readtoken() != t)
424 synexpect(t);
425 checkkwd = 1;
426 break;
427 case TCASE:
428 n1 = (union node *)stalloc(sizeof (struct ncase));
429 n1->type = NCASE;
430 if (readtoken() != TWORD)
431 synexpect(TWORD);
432 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg));
433 n2->type = NARG;
434 n2->narg.text = wordtext;
435 n2->narg.backquote = backquotelist;
436 n2->narg.next = NULL;
437 while (readtoken() == TNL);
438 if (lasttoken != TWORD || ! equal(wordtext, "in"))
439 synerror("expecting \"in\"");
440 cpp = &n1->ncase.cases;
441 noaliases = 1; /* turn off alias expansion */
442 checkkwd = 2, readtoken();
443 while (lasttoken != TESAC) {
444 *cpp = cp = (union node *)stalloc(sizeof (struct nclist));
445 cp->type = NCLIST;
446 app = &cp->nclist.pattern;
447 if (lasttoken == TLP)
448 readtoken();
449 for (;;) {
450 *app = ap = (union node *)stalloc(sizeof (struct narg));
451 ap->type = NARG;
452 ap->narg.text = wordtext;
453 ap->narg.backquote = backquotelist;
454 if (checkkwd = 2, readtoken() != TPIPE)
455 break;
456 app = &ap->narg.next;
457 readtoken();
459 ap->narg.next = NULL;
460 if (lasttoken != TRP)
461 noaliases = 0, synexpect(TRP);
462 cp->nclist.body = list(0);
464 checkkwd = 2;
465 if ((t = readtoken()) != TESAC) {
466 if (t != TENDCASE)
467 noaliases = 0, synexpect(TENDCASE);
468 else
469 checkkwd = 2, readtoken();
471 cpp = &cp->nclist.next;
473 noaliases = 0; /* reset alias expansion */
474 *cpp = NULL;
475 checkkwd = 1;
476 break;
477 case TLP:
478 n1 = (union node *)stalloc(sizeof (struct nredir));
479 n1->type = NSUBSHELL;
480 n1->nredir.n = list(0);
481 n1->nredir.redirect = NULL;
482 if (readtoken() != TRP)
483 synexpect(TRP);
484 checkkwd = 1;
485 break;
486 case TBEGIN:
487 n1 = list(0);
488 if (readtoken() != TEND)
489 synexpect(TEND);
490 checkkwd = 1;
491 break;
492 /* Handle an empty command like other simple commands. */
493 case TSEMI:
494 case TAND:
495 case TOR:
497 * An empty command before a ; doesn't make much sense, and
498 * should certainly be disallowed in the case of `if ;'.
500 if (!redir)
501 synexpect(-1);
502 case TNL:
503 case TEOF:
504 case TWORD:
505 case TRP:
506 tokpushback++;
507 return simplecmd(rpp, redir);
508 default:
509 synexpect(-1);
512 /* Now check for redirection which may follow command */
513 while (readtoken() == TREDIR) {
514 *rpp = n2 = redirnode;
515 rpp = &n2->nfile.next;
516 parsefname();
518 tokpushback++;
519 *rpp = NULL;
520 if (redir) {
521 if (n1->type != NSUBSHELL) {
522 n2 = (union node *)stalloc(sizeof (struct nredir));
523 n2->type = NREDIR;
524 n2->nredir.n = n1;
525 n1 = n2;
527 n1->nredir.redirect = redir;
529 return n1;
533 STATIC union node *
534 simplecmd(union node **rpp, union node *redir)
536 union node *args, **app;
537 union node **orig_rpp = rpp;
538 union node *n = NULL;
540 /* If we don't have any redirections already, then we must reset */
541 /* rpp to be the address of the local redir variable. */
542 if (redir == 0)
543 rpp = &redir;
545 args = NULL;
546 app = &args;
548 * We save the incoming value, because we need this for shell
549 * functions. There can not be a redirect or an argument between
550 * the function name and the open parenthesis.
552 orig_rpp = rpp;
554 for (;;) {
555 if (readtoken() == TWORD) {
556 n = (union node *)stalloc(sizeof (struct narg));
557 n->type = NARG;
558 n->narg.text = wordtext;
559 n->narg.backquote = backquotelist;
560 *app = n;
561 app = &n->narg.next;
562 } else if (lasttoken == TREDIR) {
563 *rpp = n = redirnode;
564 rpp = &n->nfile.next;
565 parsefname(); /* read name of redirection file */
566 } else if (lasttoken == TLP && app == &args->narg.next
567 && rpp == orig_rpp) {
568 /* We have a function */
569 if (readtoken() != TRP)
570 synexpect(TRP);
571 funclinno = plinno;
572 #ifdef notdef
573 if (! goodname(n->narg.text))
574 synerror("Bad function name");
575 #endif
576 n->type = NDEFUN;
577 n->narg.next = command();
578 funclinno = 0;
579 return n;
580 } else {
581 tokpushback++;
582 break;
585 *app = NULL;
586 *rpp = NULL;
587 n = (union node *)stalloc(sizeof (struct ncmd));
588 n->type = NCMD;
589 n->ncmd.backgnd = 0;
590 n->ncmd.args = args;
591 n->ncmd.redirect = redir;
592 return n;
595 STATIC union node *
596 makename(void)
598 union node *n;
600 n = (union node *)stalloc(sizeof (struct narg));
601 n->type = NARG;
602 n->narg.next = NULL;
603 n->narg.text = wordtext;
604 n->narg.backquote = backquotelist;
605 return n;
608 void
609 fixredir(union node *n, const char *text, int err)
611 TRACE(("Fix redir %s %d\n", text, err));
612 if (!err)
613 n->ndup.vname = NULL;
615 if (is_digit(text[0]) && text[1] == '\0')
616 n->ndup.dupfd = digit_val(text[0]);
617 else if (text[0] == '-' && text[1] == '\0')
618 n->ndup.dupfd = -1;
619 else {
621 if (err)
622 synerror("Bad fd number");
623 else
624 n->ndup.vname = makename();
629 STATIC void
630 parsefname(void)
632 union node *n = redirnode;
634 if (readtoken() != TWORD)
635 synexpect(-1);
636 if (n->type == NHERE) {
637 struct heredoc *here = heredoc;
638 struct heredoc *p;
639 int i;
641 if (quoteflag == 0)
642 n->type = NXHERE;
643 TRACE(("Here document %d\n", n->type));
644 if (here->striptabs) {
645 while (*wordtext == '\t')
646 wordtext++;
648 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN)
649 synerror("Illegal eof marker for << redirection");
650 rmescapes(wordtext);
651 here->eofmark = wordtext;
652 here->next = NULL;
653 if (heredoclist == NULL)
654 heredoclist = here;
655 else {
656 for (p = heredoclist ; p->next ; p = p->next);
657 p->next = here;
659 } else if (n->type == NTOFD || n->type == NFROMFD) {
660 fixredir(n, wordtext, 0);
661 } else {
662 n->nfile.fname = makename();
668 * Input any here documents.
671 STATIC void
672 parseheredoc(void)
674 struct heredoc *here;
675 union node *n;
677 while (heredoclist) {
678 here = heredoclist;
679 heredoclist = here->next;
680 if (needprompt) {
681 setprompt(2);
682 needprompt = 0;
684 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX,
685 here->eofmark, here->striptabs);
686 n = (union node *)stalloc(sizeof (struct narg));
687 n->narg.type = NARG;
688 n->narg.next = NULL;
689 n->narg.text = wordtext;
690 n->narg.backquote = backquotelist;
691 here->here->nhere.doc = n;
695 STATIC int
696 peektoken(void)
698 int t;
700 t = readtoken();
701 tokpushback++;
702 return (t);
705 STATIC int
706 readtoken(void)
708 int t;
709 int savecheckkwd = checkkwd;
710 struct alias *ap;
711 #ifdef DEBUG
712 int alreadyseen = tokpushback;
713 #endif
715 top:
716 t = xxreadtoken();
718 if (checkkwd) {
720 * eat newlines
722 if (checkkwd > 1) {
723 checkkwd = 0;
724 while (t == TNL) {
725 parseheredoc();
726 t = xxreadtoken();
728 } else
729 checkkwd = 0;
731 * check for keywords and aliases
733 if (t == TWORD && !quoteflag)
735 const char * const *pp;
737 for (pp = parsekwd; *pp; pp++) {
738 if (**pp == *wordtext && equal(*pp, wordtext))
740 lasttoken = t = pp - parsekwd + KWDOFFSET;
741 TRACE(("keyword %s recognized\n", tokname[t]));
742 goto out;
745 if (noaliases == 0 &&
746 (ap = lookupalias(wordtext, 1)) != NULL) {
747 pushstring(ap->val, strlen(ap->val), ap);
748 checkkwd = savecheckkwd;
749 goto top;
752 out:
753 checkkwd = 0;
755 #ifdef DEBUG
756 if (!alreadyseen)
757 TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
758 else
759 TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
760 #endif
761 return (t);
766 * Read the next input token.
767 * If the token is a word, we set backquotelist to the list of cmds in
768 * backquotes. We set quoteflag to true if any part of the word was
769 * quoted.
770 * If the token is TREDIR, then we set redirnode to a structure containing
771 * the redirection.
772 * In all cases, the variable startlinno is set to the number of the line
773 * on which the token starts.
775 * [Change comment: here documents and internal procedures]
776 * [Readtoken shouldn't have any arguments. Perhaps we should make the
777 * word parsing code into a separate routine. In this case, readtoken
778 * doesn't need to have any internal procedures, but parseword does.
779 * We could also make parseoperator in essence the main routine, and
780 * have parseword (readtoken1?) handle both words and redirection.]
783 #define RETURN(token) return lasttoken = token
785 STATIC int
786 xxreadtoken(void)
788 int c;
790 if (tokpushback) {
791 tokpushback = 0;
792 return lasttoken;
794 if (needprompt) {
795 setprompt(2);
796 needprompt = 0;
798 startlinno = plinno;
799 for (;;) { /* until token or start of word found */
800 c = pgetc_macro();
801 if (c == ' ' || c == '\t')
802 continue; /* quick check for white space first */
803 switch (c) {
804 case ' ': case '\t':
805 continue;
806 case '#':
807 while ((c = pgetc()) != '\n' && c != PEOF);
808 pungetc();
809 continue;
810 case '\\':
811 if (pgetc() == '\n') {
812 startlinno = ++plinno;
813 if (doprompt)
814 setprompt(2);
815 else
816 setprompt(0);
817 continue;
819 pungetc();
820 goto breakloop;
821 case '\n':
822 plinno++;
823 needprompt = doprompt;
824 RETURN(TNL);
825 case PEOF:
826 RETURN(TEOF);
827 case '&':
828 if (pgetc() == '&')
829 RETURN(TAND);
830 pungetc();
831 RETURN(TBACKGND);
832 case '|':
833 if (pgetc() == '|')
834 RETURN(TOR);
835 pungetc();
836 RETURN(TPIPE);
837 case ';':
838 if (pgetc() == ';')
839 RETURN(TENDCASE);
840 pungetc();
841 RETURN(TSEMI);
842 case '(':
843 RETURN(TLP);
844 case ')':
845 RETURN(TRP);
846 case '!':
847 if (checkkwd == 3)
848 RETURN(TNOT);
849 /* else FALLTHROUGH */
850 default:
851 goto breakloop;
854 breakloop:
855 return readtoken1(c, BASESYNTAX, NULL, 0);
856 #undef RETURN
862 * If eofmark is NULL, read a word or a redirection symbol. If eofmark
863 * is not NULL, read a here document. In the latter case, eofmark is the
864 * word which marks the end of the document and striptabs is true if
865 * leading tabs should be stripped from the document. The argument firstc
866 * is the first character of the input token or document.
868 * Because C does not have internal subroutines, I have simulated them
869 * using goto's to implement the subroutine linkage. The following macros
870 * will run code that appears at the end of readtoken1.
873 #define CHECKEND() {goto checkend; checkend_return:;}
874 #define PARSEREDIR() {goto parseredir; parseredir_return:;}
875 #define PARSESUB() {goto parsesub; parsesub_return:;}
876 #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;}
877 #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;}
878 #define PARSEARITH() {goto parsearith; parsearith_return:;}
880 STATIC int
881 readtoken1(int firstc, char const *syn, char *eofmark, int striptabs)
883 int c = firstc;
884 char const * volatile syntax = syn;
885 char * volatile out;
886 int len;
887 char line[EOFMARKLEN + 1];
888 struct nodelist *bqlist;
889 volatile int quotef;
890 volatile int dblquote;
891 volatile int varnest; /* levels of variables expansion */
892 volatile int arinest; /* levels of arithmetic expansion */
893 volatile int parenlevel; /* levels of parens in arithmetic */
894 volatile int oldstyle;
895 char const * volatile prevsyntax = NULL; /* syntax before arithmetic */
896 int synentry;
898 startlinno = plinno;
899 dblquote = 0;
900 if (syntax == DQSYNTAX)
901 dblquote = 1;
902 quotef = 0;
903 bqlist = NULL;
904 varnest = 0;
905 arinest = 0;
906 parenlevel = 0;
908 STARTSTACKSTR(out);
909 loop: { /* for each line, until end of word */
910 CHECKEND(); /* set c to PEOF if at end of here document */
911 for (;;) { /* until end of line or end of word */
912 CHECKSTRSPACE(3, out); /* permit 3 calls to USTPUTC */
914 synentry = syntax[c];
916 switch(synentry) {
917 case CNL: /* '\n' */
918 if (syntax == BASESYNTAX)
919 goto endword; /* exit outer loop */
920 USTPUTC(c, out);
921 plinno++;
922 if (doprompt)
923 setprompt(2);
924 else
925 setprompt(0);
926 c = pgetc();
927 goto loop; /* continue outer loop */
928 case CWORD:
929 USTPUTC(c, out);
930 break;
931 case CCTL:
932 if (eofmark == NULL || dblquote)
933 USTPUTC(CTLESC, out);
934 USTPUTC(c, out);
935 break;
936 case CBACK: /* backslash */
937 c = pgetc();
938 if (c == PEOF) {
939 USTPUTC('\\', out);
940 pungetc();
941 } else if (c == '\n') {
942 plinno++;
943 if (doprompt)
944 setprompt(2);
945 else
946 setprompt(0);
947 } else {
948 if (dblquote && c != '\\' &&
949 c != '`' && c != '$' &&
950 (c != '"' || eofmark != NULL))
951 USTPUTC('\\', out);
952 if (SQSYNTAX[c] == CCTL)
953 USTPUTC(CTLESC, out);
954 else if (eofmark == NULL)
955 USTPUTC(CTLQUOTEMARK, out);
956 USTPUTC(c, out);
957 quotef++;
959 break;
960 case CSQUOTE:
961 if (eofmark == NULL)
962 USTPUTC(CTLQUOTEMARK, out);
963 syntax = SQSYNTAX;
964 break;
965 case CDQUOTE:
966 if (eofmark == NULL)
967 USTPUTC(CTLQUOTEMARK, out);
968 syntax = DQSYNTAX;
969 dblquote = 1;
970 break;
971 case CENDQUOTE:
972 if (eofmark != NULL && arinest == 0 &&
973 varnest == 0) {
974 USTPUTC(c, out);
975 } else {
976 if (arinest) {
977 syntax = ARISYNTAX;
978 dblquote = 0;
979 } else if (eofmark == NULL) {
980 syntax = BASESYNTAX;
981 dblquote = 0;
983 quotef++;
985 break;
986 case CVAR: /* '$' */
987 PARSESUB(); /* parse substitution */
988 break;
989 case CENDVAR: /* '}' */
990 if (varnest > 0) {
991 varnest--;
992 USTPUTC(CTLENDVAR, out);
993 } else {
994 USTPUTC(c, out);
996 break;
997 case CLP: /* '(' in arithmetic */
998 parenlevel++;
999 USTPUTC(c, out);
1000 break;
1001 case CRP: /* ')' in arithmetic */
1002 if (parenlevel > 0) {
1003 USTPUTC(c, out);
1004 --parenlevel;
1005 } else {
1006 if (pgetc() == ')') {
1007 if (--arinest == 0) {
1008 USTPUTC(CTLENDARI, out);
1009 syntax = prevsyntax;
1010 if (syntax == DQSYNTAX)
1011 dblquote = 1;
1012 else
1013 dblquote = 0;
1014 } else
1015 USTPUTC(')', out);
1016 } else {
1018 * unbalanced parens
1019 * (don't 2nd guess - no error)
1021 pungetc();
1022 USTPUTC(')', out);
1025 break;
1026 case CBQUOTE: /* '`' */
1027 PARSEBACKQOLD();
1028 break;
1029 case CEOF:
1030 goto endword; /* exit outer loop */
1031 default:
1032 if (varnest == 0)
1033 goto endword; /* exit outer loop */
1034 USTPUTC(c, out);
1036 c = pgetc_macro();
1039 endword:
1040 if (syntax == ARISYNTAX)
1041 synerror("Missing '))'");
1042 if (syntax != BASESYNTAX && ! parsebackquote && eofmark == NULL)
1043 synerror("Unterminated quoted string");
1044 if (varnest != 0) {
1045 startlinno = plinno;
1046 synerror("Missing '}'");
1048 USTPUTC('\0', out);
1049 len = out - stackblock();
1050 out = stackblock();
1051 if (eofmark == NULL) {
1052 if ((c == '>' || c == '<')
1053 && quotef == 0
1054 && len <= 2
1055 && (*out == '\0' || is_digit(*out))) {
1056 PARSEREDIR();
1057 return lasttoken = TREDIR;
1058 } else {
1059 pungetc();
1062 quoteflag = quotef;
1063 backquotelist = bqlist;
1064 grabstackblock(len);
1065 wordtext = out;
1066 return lasttoken = TWORD;
1067 /* end of readtoken routine */
1072 * Check to see whether we are at the end of the here document. When this
1073 * is called, c is set to the first character of the next input line. If
1074 * we are at the end of the here document, this routine sets the c to PEOF.
1077 checkend: {
1078 if (eofmark) {
1079 if (striptabs) {
1080 while (c == '\t')
1081 c = pgetc();
1083 if (c == *eofmark) {
1084 if (pfgets(line, sizeof line) != NULL) {
1085 char *p, *q;
1087 p = line;
1088 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++);
1089 if (*p == '\n' && *q == '\0') {
1090 c = PEOF;
1091 plinno++;
1092 needprompt = doprompt;
1093 } else {
1094 pushstring(line, strlen(line), NULL);
1099 goto checkend_return;
1104 * Parse a redirection operator. The variable "out" points to a string
1105 * specifying the fd to be redirected. The variable "c" contains the
1106 * first character of the redirection operator.
1109 parseredir: {
1110 char fd = *out;
1111 union node *np;
1113 np = (union node *)stalloc(sizeof (struct nfile));
1114 if (c == '>') {
1115 np->nfile.fd = 1;
1116 c = pgetc();
1117 if (c == '>')
1118 np->type = NAPPEND;
1119 else if (c == '&')
1120 np->type = NTOFD;
1121 else if (c == '|')
1122 np->type = NCLOBBER;
1123 else {
1124 np->type = NTO;
1125 pungetc();
1127 } else { /* c == '<' */
1128 np->nfile.fd = 0;
1129 c = pgetc();
1130 if (c == '<') {
1131 if (sizeof (struct nfile) != sizeof (struct nhere)) {
1132 np = (union node *)stalloc(sizeof (struct nhere));
1133 np->nfile.fd = 0;
1135 np->type = NHERE;
1136 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc));
1137 heredoc->here = np;
1138 if ((c = pgetc()) == '-') {
1139 heredoc->striptabs = 1;
1140 } else {
1141 heredoc->striptabs = 0;
1142 pungetc();
1144 } else if (c == '&')
1145 np->type = NFROMFD;
1146 else if (c == '>')
1147 np->type = NFROMTO;
1148 else {
1149 np->type = NFROM;
1150 pungetc();
1153 if (fd != '\0')
1154 np->nfile.fd = digit_val(fd);
1155 redirnode = np;
1156 goto parseredir_return;
1161 * Parse a substitution. At this point, we have read the dollar sign
1162 * and nothing else.
1165 parsesub: {
1166 char buf[10];
1167 int subtype;
1168 int typeloc;
1169 int flags;
1170 char *p;
1171 static const char types[] = "}-+?=";
1172 int bracketed_name = 0; /* used to handle ${[0-9]*} variables */
1173 int i;
1174 int linno;
1175 int length;
1177 c = pgetc();
1178 if (c != '(' && c != '{' && (is_eof(c) || !is_name(c)) &&
1179 !is_special(c)) {
1180 USTPUTC('$', out);
1181 pungetc();
1182 } else if (c == '(') { /* $(command) or $((arith)) */
1183 if (pgetc() == '(') {
1184 PARSEARITH();
1185 } else {
1186 pungetc();
1187 PARSEBACKQNEW();
1189 } else {
1190 USTPUTC(CTLVAR, out);
1191 typeloc = out - stackblock();
1192 USTPUTC(VSNORMAL, out);
1193 subtype = VSNORMAL;
1194 flags = 0;
1195 if (c == '{') {
1196 bracketed_name = 1;
1197 c = pgetc();
1198 if (c == '#') {
1199 if ((c = pgetc()) == '}')
1200 c = '#';
1201 else
1202 subtype = VSLENGTH;
1204 else
1205 subtype = 0;
1207 if (!is_eof(c) && is_name(c)) {
1208 length = 0;
1209 do {
1210 STPUTC(c, out);
1211 c = pgetc();
1212 length++;
1213 } while (!is_eof(c) && is_in_name(c));
1214 if (length == 6 &&
1215 strncmp(out - length, "LINENO", length) == 0) {
1216 /* Replace the variable name with the
1217 * current line number. */
1218 linno = plinno;
1219 if (funclinno != 0)
1220 linno -= funclinno - 1;
1221 snprintf(buf, sizeof(buf), "%d", linno);
1222 STADJUST(-6, out);
1223 for (i = 0; buf[i] != '\0'; i++)
1224 STPUTC(buf[i], out);
1225 flags |= VSLINENO;
1227 } else if (is_digit(c)) {
1228 if (bracketed_name) {
1229 do {
1230 STPUTC(c, out);
1231 c = pgetc();
1232 } while (is_digit(c));
1233 } else {
1234 STPUTC(c, out);
1235 c = pgetc();
1237 } else {
1238 if (! is_special(c)) {
1239 subtype = VSERROR;
1240 if (c == '}')
1241 pungetc();
1242 else
1243 USTPUTC(c, out);
1244 } else {
1245 USTPUTC(c, out);
1246 c = pgetc();
1249 if (subtype == 0) {
1250 switch (c) {
1251 case ':':
1252 flags |= VSNUL;
1253 c = pgetc();
1254 /*FALLTHROUGH*/
1255 default:
1256 p = strchr(types, c);
1257 if (p == NULL) {
1258 if (flags == VSNUL)
1259 STPUTC(':', out);
1260 STPUTC(c, out);
1261 subtype = VSERROR;
1262 } else
1263 subtype = p - types + VSNORMAL;
1264 break;
1265 case '%':
1266 case '#':
1268 int cc = c;
1269 subtype = c == '#' ? VSTRIMLEFT :
1270 VSTRIMRIGHT;
1271 c = pgetc();
1272 if (c == cc)
1273 subtype++;
1274 else
1275 pungetc();
1276 break;
1279 } else if (subtype != VSERROR) {
1280 pungetc();
1282 STPUTC('=', out);
1283 if (subtype != VSLENGTH && (dblquote || arinest))
1284 flags |= VSQUOTE;
1285 *(stackblock() + typeloc) = subtype | flags;
1286 if (subtype != VSNORMAL)
1287 varnest++;
1289 goto parsesub_return;
1294 * Called to parse command substitutions. Newstyle is set if the command
1295 * is enclosed inside $(...); nlpp is a pointer to the head of the linked
1296 * list of commands (passed by reference), and savelen is the number of
1297 * characters on the top of the stack which must be preserved.
1300 parsebackq: {
1301 struct nodelist **nlpp;
1302 int savepbq;
1303 union node *n;
1304 char *volatile str;
1305 struct jmploc jmploc;
1306 struct jmploc *volatile savehandler;
1307 int savelen;
1308 volatile int saveprompt;
1310 savepbq = parsebackquote;
1311 if (setjmp(jmploc.loc)) {
1312 if (str)
1313 ckfree(str);
1314 parsebackquote = 0;
1315 handler = savehandler;
1316 longjmp(handler->loc, 1);
1318 INTOFF;
1319 str = NULL;
1320 savelen = out - stackblock();
1321 if (savelen > 0) {
1322 str = ckmalloc(savelen);
1323 memcpy(str, stackblock(), savelen);
1325 savehandler = handler;
1326 handler = &jmploc;
1327 INTON;
1328 if (oldstyle) {
1329 /* We must read until the closing backquote, giving special
1330 treatment to some slashes, and then push the string and
1331 reread it as input, interpreting it normally. */
1332 char *pout;
1333 int pc;
1334 int psavelen;
1335 char *pstr;
1338 STARTSTACKSTR(pout);
1339 for (;;) {
1340 if (needprompt) {
1341 setprompt(2);
1342 needprompt = 0;
1344 switch (pc = pgetc()) {
1345 case '`':
1346 goto done;
1348 case '\\':
1349 if ((pc = pgetc()) == '\n') {
1350 plinno++;
1351 if (doprompt)
1352 setprompt(2);
1353 else
1354 setprompt(0);
1356 * If eating a newline, avoid putting
1357 * the newline into the new character
1358 * stream (via the STPUTC after the
1359 * switch).
1361 continue;
1363 if (pc != '\\' && pc != '`' && pc != '$'
1364 && (!dblquote || pc != '"'))
1365 STPUTC('\\', pout);
1366 break;
1368 case '\n':
1369 plinno++;
1370 needprompt = doprompt;
1371 break;
1373 case PEOF:
1374 startlinno = plinno;
1375 synerror("EOF in backquote substitution");
1376 break;
1378 default:
1379 break;
1381 STPUTC(pc, pout);
1383 done:
1384 STPUTC('\0', pout);
1385 psavelen = pout - stackblock();
1386 if (psavelen > 0) {
1387 pstr = ckmalloc(psavelen);
1388 memcpy(pstr, stackblock(), psavelen);
1389 setinputstring(pstr, 1);
1392 nlpp = &bqlist;
1393 while (*nlpp)
1394 nlpp = &(*nlpp)->next;
1395 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist));
1396 (*nlpp)->next = NULL;
1397 parsebackquote = oldstyle;
1399 if (oldstyle) {
1400 saveprompt = doprompt;
1401 doprompt = 0;
1404 n = list(0);
1406 if (oldstyle)
1407 doprompt = saveprompt;
1408 else {
1409 if (readtoken() != TRP)
1410 synexpect(TRP);
1413 (*nlpp)->n = n;
1414 if (oldstyle) {
1416 * Start reading from old file again, ignoring any pushed back
1417 * tokens left from the backquote parsing
1419 popfile();
1420 tokpushback = 0;
1422 while (stackblocksize() <= savelen)
1423 growstackblock();
1424 STARTSTACKSTR(out);
1425 if (str) {
1426 memcpy(out, str, savelen);
1427 STADJUST(savelen, out);
1428 INTOFF;
1429 ckfree(str);
1430 str = NULL;
1431 INTON;
1433 parsebackquote = savepbq;
1434 handler = savehandler;
1435 if (arinest || dblquote)
1436 USTPUTC(CTLBACKQ | CTLQUOTE, out);
1437 else
1438 USTPUTC(CTLBACKQ, out);
1439 if (oldstyle)
1440 goto parsebackq_oldreturn;
1441 else
1442 goto parsebackq_newreturn;
1446 * Parse an arithmetic expansion (indicate start of one and set state)
1448 parsearith: {
1450 if (++arinest == 1) {
1451 prevsyntax = syntax;
1452 syntax = ARISYNTAX;
1453 USTPUTC(CTLARI, out);
1454 if (dblquote)
1455 USTPUTC('"',out);
1456 else
1457 USTPUTC(' ',out);
1458 } else {
1460 * we collapse embedded arithmetic expansion to
1461 * parenthesis, which should be equivalent
1463 USTPUTC('(', out);
1465 goto parsearith_return;
1468 } /* end of readtoken */
1472 #ifdef mkinit
1473 RESET {
1474 tokpushback = 0;
1475 checkkwd = 0;
1477 #endif
1480 * Returns true if the text contains nothing to expand (no dollar signs
1481 * or backquotes).
1484 STATIC int
1485 noexpand(char *text)
1487 char *p;
1488 char c;
1490 p = text;
1491 while ((c = *p++) != '\0') {
1492 if ( c == CTLQUOTEMARK)
1493 continue;
1494 if (c == CTLESC)
1495 p++;
1496 else if (BASESYNTAX[(int)c] == CCTL)
1497 return 0;
1499 return 1;
1504 * Return true if the argument is a legal variable name (a letter or
1505 * underscore followed by zero or more letters, underscores, and digits).
1509 goodname(char *name)
1511 char *p;
1513 p = name;
1514 if (! is_name(*p))
1515 return 0;
1516 while (*++p) {
1517 if (! is_in_name(*p))
1518 return 0;
1520 return 1;
1525 * Called when an unexpected token is read during the parse. The argument
1526 * is the token that is expected, or -1 if more than one type of token can
1527 * occur at this point.
1530 STATIC void
1531 synexpect(int token)
1533 char msg[64];
1535 if (token >= 0) {
1536 fmtstr(msg, 64, "%s unexpected (expecting %s)",
1537 tokname[lasttoken], tokname[token]);
1538 } else {
1539 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]);
1541 synerror(msg);
1545 STATIC void
1546 synerror(const char *msg)
1548 if (commandname)
1549 outfmt(&errout, "%s: %d: ", commandname, startlinno);
1550 outfmt(&errout, "Syntax error: %s\n", msg);
1551 error(NULL);
1554 STATIC void
1555 setprompt(int which)
1557 whichprompt = which;
1559 #ifndef NO_HISTORY
1560 if (!el)
1561 #endif
1562 out2str(getprompt(NULL));
1566 * called by editline -- any expansions to the prompt
1567 * should be added here.
1569 const char *
1570 getprompt(void *unused __unused)
1572 static char ps[PROMPTLEN];
1573 const char *fmt;
1574 int i, j, trim;
1577 * Select prompt format.
1579 switch (whichprompt) {
1580 case 0:
1581 fmt = "";
1582 break;
1583 case 1:
1584 fmt = ps1val();
1585 break;
1586 case 2:
1587 fmt = ps2val();
1588 break;
1589 default:
1590 return "<internal prompt error>";
1594 * Format prompt string.
1596 for (i = 0; (i < 127) && (*fmt != '\0'); i++, fmt++)
1597 if (*fmt == '\\')
1598 switch (*++fmt) {
1601 * Hostname.
1603 * \h specifies just the local hostname,
1604 * \H specifies fully-qualified hostname.
1606 case 'h':
1607 case 'H':
1608 ps[i] = '\0';
1609 gethostname(&ps[i], PROMPTLEN - i);
1610 /* Skip to end of hostname. */
1611 trim = (*fmt == 'h') ? '.' : '\0';
1612 while ((ps[i+1] != '\0') && (ps[i+1] != trim))
1613 i++;
1614 break;
1617 * Working directory.
1619 * \W specifies just the final component,
1620 * \w specifies the entire path.
1622 case 'W':
1623 case 'w':
1624 ps[i] = '\0';
1625 getcwd(&ps[i], PROMPTLEN - i);
1626 if (*fmt == 'W') {
1627 /* Final path component only. */
1628 trim = 1;
1629 for (j = i; ps[j] != '\0'; j++)
1630 if (ps[j] == '/')
1631 trim = j + 1;
1632 memmove(&ps[i], &ps[trim],
1633 j - trim + 1);
1635 /* Skip to end of path. */
1636 while (ps[i + 1] != '\0')
1637 i++;
1638 break;
1641 * Superuser status.
1643 * '$' for normal users, '#' for root.
1645 case '$':
1646 ps[i] = (geteuid() != 0) ? '$' : '#';
1647 break;
1650 * A literal \.
1652 case '\\':
1653 ps[i] = '\\';
1654 break;
1657 * Emit unrecognized formats verbatim.
1659 default:
1660 ps[i++] = '\\';
1661 ps[i] = *fmt;
1662 break;
1664 else
1665 ps[i] = *fmt;
1666 ps[i] = '\0';
1667 return (ps);