parser.c revision 197691 
1/*-
2 * Copyright (c) 1991, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * This code is derived from software contributed to Berkeley by
6 * Kenneth Almquist.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 *    notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 *    notice, this list of conditions and the following disclaimer in the
15 *    documentation and/or other materials provided with the distribution.
16 * 4. Neither the name of the University nor the names of its contributors
17 *    may be used to endorse or promote products derived from this software
18 *    without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 */
32
33#ifndef lint
34#if 0
35static char sccsid[] = "@(#)parser.c	8.7 (Berkeley) 5/16/95";
36#endif
37#endif /* not lint */
38#include <sys/cdefs.h>
39__FBSDID("$FreeBSD: head/bin/sh/parser.c 197691 2009-10-01 21:40:08Z jilles $");
40
41#include <stdlib.h>
42#include <unistd.h>
43
44#include "shell.h"
45#include "parser.h"
46#include "nodes.h"
47#include "expand.h"	/* defines rmescapes() */
48#include "syntax.h"
49#include "options.h"
50#include "input.h"
51#include "output.h"
52#include "var.h"
53#include "error.h"
54#include "memalloc.h"
55#include "mystring.h"
56#include "alias.h"
57#include "show.h"
58#include "eval.h"
59#ifndef NO_HISTORY
60#include "myhistedit.h"
61#endif
62
63/*
64 * Shell command parser.
65 */
66
67#define	EOFMARKLEN	79
68#define	PROMPTLEN	128
69
70/* values returned by readtoken */
71#include "token.h"
72
73
74
75struct heredoc {
76	struct heredoc *next;	/* next here document in list */
77	union node *here;		/* redirection node */
78	char *eofmark;		/* string indicating end of input */
79	int striptabs;		/* if set, strip leading tabs */
80};
81
82
83
84STATIC struct heredoc *heredoclist;	/* list of here documents to read */
85STATIC int doprompt;		/* if set, prompt the user */
86STATIC int needprompt;		/* true if interactive and at start of line */
87STATIC int lasttoken;		/* last token read */
88MKINIT int tokpushback;		/* last token pushed back */
89STATIC char *wordtext;		/* text of last word returned by readtoken */
90MKINIT int checkkwd;            /* 1 == check for kwds, 2 == also eat newlines */
91STATIC struct nodelist *backquotelist;
92STATIC union node *redirnode;
93STATIC struct heredoc *heredoc;
94STATIC int quoteflag;		/* set if (part of) last token was quoted */
95STATIC int startlinno;		/* line # where last token started */
96STATIC int funclinno;		/* line # where the current function started */
97
98/* XXX When 'noaliases' is set to one, no alias expansion takes place. */
99static int noaliases = 0;
100
101
102STATIC union node *list(int);
103STATIC union node *andor(void);
104STATIC union node *pipeline(void);
105STATIC union node *command(void);
106STATIC union node *simplecmd(union node **, union node *);
107STATIC union node *makename(void);
108STATIC void parsefname(void);
109STATIC void parseheredoc(void);
110STATIC int peektoken(void);
111STATIC int readtoken(void);
112STATIC int xxreadtoken(void);
113STATIC int readtoken1(int, char const *, char *, int);
114STATIC int noexpand(char *);
115STATIC void synexpect(int);
116STATIC void synerror(char *);
117STATIC void setprompt(int);
118
119
120/*
121 * Read and parse a command.  Returns NEOF on end of file.  (NULL is a
122 * valid parse tree indicating a blank line.)
123 */
124
125union node *
126parsecmd(int interact)
127{
128	int t;
129
130	tokpushback = 0;
131	doprompt = interact;
132	if (doprompt)
133		setprompt(1);
134	else
135		setprompt(0);
136	needprompt = 0;
137	t = readtoken();
138	if (t == TEOF)
139		return NEOF;
140	if (t == TNL)
141		return NULL;
142	tokpushback++;
143	return list(1);
144}
145
146
147STATIC union node *
148list(int nlflag)
149{
150	union node *n1, *n2, *n3;
151	int tok;
152
153	checkkwd = 2;
154	if (nlflag == 0 && tokendlist[peektoken()])
155		return NULL;
156	n1 = NULL;
157	for (;;) {
158		n2 = andor();
159		tok = readtoken();
160		if (tok == TBACKGND) {
161			if (n2->type == NCMD || n2->type == NPIPE) {
162				n2->ncmd.backgnd = 1;
163			} else if (n2->type == NREDIR) {
164				n2->type = NBACKGND;
165			} else {
166				n3 = (union node *)stalloc(sizeof (struct nredir));
167				n3->type = NBACKGND;
168				n3->nredir.n = n2;
169				n3->nredir.redirect = NULL;
170				n2 = n3;
171			}
172		}
173		if (n1 == NULL) {
174			n1 = n2;
175		}
176		else {
177			n3 = (union node *)stalloc(sizeof (struct nbinary));
178			n3->type = NSEMI;
179			n3->nbinary.ch1 = n1;
180			n3->nbinary.ch2 = n2;
181			n1 = n3;
182		}
183		switch (tok) {
184		case TBACKGND:
185		case TSEMI:
186			tok = readtoken();
187			/* FALLTHROUGH */
188		case TNL:
189			if (tok == TNL) {
190				parseheredoc();
191				if (nlflag)
192					return n1;
193			} else {
194				tokpushback++;
195			}
196			checkkwd = 2;
197			if (tokendlist[peektoken()])
198				return n1;
199			break;
200		case TEOF:
201			if (heredoclist)
202				parseheredoc();
203			else
204				pungetc();		/* push back EOF on input */
205			return n1;
206		default:
207			if (nlflag)
208				synexpect(-1);
209			tokpushback++;
210			return n1;
211		}
212	}
213}
214
215
216
217STATIC union node *
218andor(void)
219{
220	union node *n1, *n2, *n3;
221	int t;
222
223	n1 = pipeline();
224	for (;;) {
225		if ((t = readtoken()) == TAND) {
226			t = NAND;
227		} else if (t == TOR) {
228			t = NOR;
229		} else {
230			tokpushback++;
231			return n1;
232		}
233		n2 = pipeline();
234		n3 = (union node *)stalloc(sizeof (struct nbinary));
235		n3->type = t;
236		n3->nbinary.ch1 = n1;
237		n3->nbinary.ch2 = n2;
238		n1 = n3;
239	}
240}
241
242
243
244STATIC union node *
245pipeline(void)
246{
247	union node *n1, *n2, *pipenode;
248	struct nodelist *lp, *prev;
249	int negate;
250
251	negate = 0;
252	checkkwd = 2;
253	TRACE(("pipeline: entered\n"));
254	while (readtoken() == TNOT)
255		negate = !negate;
256	tokpushback++;
257	n1 = command();
258	if (readtoken() == TPIPE) {
259		pipenode = (union node *)stalloc(sizeof (struct npipe));
260		pipenode->type = NPIPE;
261		pipenode->npipe.backgnd = 0;
262		lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
263		pipenode->npipe.cmdlist = lp;
264		lp->n = n1;
265		do {
266			prev = lp;
267			lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
268			lp->n = command();
269			prev->next = lp;
270		} while (readtoken() == TPIPE);
271		lp->next = NULL;
272		n1 = pipenode;
273	}
274	tokpushback++;
275	if (negate) {
276		n2 = (union node *)stalloc(sizeof (struct nnot));
277		n2->type = NNOT;
278		n2->nnot.com = n1;
279		return n2;
280	} else
281		return n1;
282}
283
284
285
286STATIC union node *
287command(void)
288{
289	union node *n1, *n2;
290	union node *ap, **app;
291	union node *cp, **cpp;
292	union node *redir, **rpp;
293	int t, negate = 0;
294
295	checkkwd = 2;
296	redir = NULL;
297	n1 = NULL;
298	rpp = &redir;
299
300	/* Check for redirection which may precede command */
301	while (readtoken() == TREDIR) {
302		*rpp = n2 = redirnode;
303		rpp = &n2->nfile.next;
304		parsefname();
305	}
306	tokpushback++;
307
308	while (readtoken() == TNOT) {
309		TRACE(("command: TNOT recognized\n"));
310		negate = !negate;
311	}
312	tokpushback++;
313
314	switch (readtoken()) {
315	case TIF:
316		n1 = (union node *)stalloc(sizeof (struct nif));
317		n1->type = NIF;
318		if ((n1->nif.test = list(0)) == NULL)
319			synexpect(-1);
320		if (readtoken() != TTHEN)
321			synexpect(TTHEN);
322		n1->nif.ifpart = list(0);
323		n2 = n1;
324		while (readtoken() == TELIF) {
325			n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif));
326			n2 = n2->nif.elsepart;
327			n2->type = NIF;
328			if ((n2->nif.test = list(0)) == NULL)
329				synexpect(-1);
330			if (readtoken() != TTHEN)
331				synexpect(TTHEN);
332			n2->nif.ifpart = list(0);
333		}
334		if (lasttoken == TELSE)
335			n2->nif.elsepart = list(0);
336		else {
337			n2->nif.elsepart = NULL;
338			tokpushback++;
339		}
340		if (readtoken() != TFI)
341			synexpect(TFI);
342		checkkwd = 1;
343		break;
344	case TWHILE:
345	case TUNTIL: {
346		int got;
347		n1 = (union node *)stalloc(sizeof (struct nbinary));
348		n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL;
349		if ((n1->nbinary.ch1 = list(0)) == NULL)
350			synexpect(-1);
351		if ((got=readtoken()) != TDO) {
352TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : ""));
353			synexpect(TDO);
354		}
355		n1->nbinary.ch2 = list(0);
356		if (readtoken() != TDONE)
357			synexpect(TDONE);
358		checkkwd = 1;
359		break;
360	}
361	case TFOR:
362		if (readtoken() != TWORD || quoteflag || ! goodname(wordtext))
363			synerror("Bad for loop variable");
364		n1 = (union node *)stalloc(sizeof (struct nfor));
365		n1->type = NFOR;
366		n1->nfor.var = wordtext;
367		if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) {
368			app = &ap;
369			while (readtoken() == TWORD) {
370				n2 = (union node *)stalloc(sizeof (struct narg));
371				n2->type = NARG;
372				n2->narg.text = wordtext;
373				n2->narg.backquote = backquotelist;
374				*app = n2;
375				app = &n2->narg.next;
376			}
377			*app = NULL;
378			n1->nfor.args = ap;
379			if (lasttoken != TNL && lasttoken != TSEMI)
380				synexpect(-1);
381		} else {
382			static char argvars[5] = {
383				CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'
384			};
385			n2 = (union node *)stalloc(sizeof (struct narg));
386			n2->type = NARG;
387			n2->narg.text = argvars;
388			n2->narg.backquote = NULL;
389			n2->narg.next = NULL;
390			n1->nfor.args = n2;
391			/*
392			 * Newline or semicolon here is optional (but note
393			 * that the original Bourne shell only allowed NL).
394			 */
395			if (lasttoken != TNL && lasttoken != TSEMI)
396				tokpushback++;
397		}
398		checkkwd = 2;
399		if ((t = readtoken()) == TDO)
400			t = TDONE;
401		else if (t == TBEGIN)
402			t = TEND;
403		else
404			synexpect(-1);
405		n1->nfor.body = list(0);
406		if (readtoken() != t)
407			synexpect(t);
408		checkkwd = 1;
409		break;
410	case TCASE:
411		n1 = (union node *)stalloc(sizeof (struct ncase));
412		n1->type = NCASE;
413		if (readtoken() != TWORD)
414			synexpect(TWORD);
415		n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg));
416		n2->type = NARG;
417		n2->narg.text = wordtext;
418		n2->narg.backquote = backquotelist;
419		n2->narg.next = NULL;
420		while (readtoken() == TNL);
421		if (lasttoken != TWORD || ! equal(wordtext, "in"))
422			synerror("expecting \"in\"");
423		cpp = &n1->ncase.cases;
424		noaliases = 1;	/* turn off alias expansion */
425		checkkwd = 2, readtoken();
426		while (lasttoken != TESAC) {
427			*cpp = cp = (union node *)stalloc(sizeof (struct nclist));
428			cp->type = NCLIST;
429			app = &cp->nclist.pattern;
430			if (lasttoken == TLP)
431				readtoken();
432			for (;;) {
433				*app = ap = (union node *)stalloc(sizeof (struct narg));
434				ap->type = NARG;
435				ap->narg.text = wordtext;
436				ap->narg.backquote = backquotelist;
437				if (checkkwd = 2, readtoken() != TPIPE)
438					break;
439				app = &ap->narg.next;
440				readtoken();
441			}
442			ap->narg.next = NULL;
443			if (lasttoken != TRP)
444				noaliases = 0, synexpect(TRP);
445			cp->nclist.body = list(0);
446
447			checkkwd = 2;
448			if ((t = readtoken()) != TESAC) {
449				if (t != TENDCASE)
450					noaliases = 0, synexpect(TENDCASE);
451				else
452					checkkwd = 2, readtoken();
453			}
454			cpp = &cp->nclist.next;
455		}
456		noaliases = 0;	/* reset alias expansion */
457		*cpp = NULL;
458		checkkwd = 1;
459		break;
460	case TLP:
461		n1 = (union node *)stalloc(sizeof (struct nredir));
462		n1->type = NSUBSHELL;
463		n1->nredir.n = list(0);
464		n1->nredir.redirect = NULL;
465		if (readtoken() != TRP)
466			synexpect(TRP);
467		checkkwd = 1;
468		break;
469	case TBEGIN:
470		n1 = list(0);
471		if (readtoken() != TEND)
472			synexpect(TEND);
473		checkkwd = 1;
474		break;
475	/* Handle an empty command like other simple commands.  */
476	case TSEMI:
477	case TAND:
478	case TOR:
479		/*
480		 * An empty command before a ; doesn't make much sense, and
481		 * should certainly be disallowed in the case of `if ;'.
482		 */
483		if (!redir)
484			synexpect(-1);
485	case TNL:
486	case TEOF:
487	case TWORD:
488	case TRP:
489		tokpushback++;
490		n1 = simplecmd(rpp, redir);
491		goto checkneg;
492	default:
493		synexpect(-1);
494	}
495
496	/* Now check for redirection which may follow command */
497	while (readtoken() == TREDIR) {
498		*rpp = n2 = redirnode;
499		rpp = &n2->nfile.next;
500		parsefname();
501	}
502	tokpushback++;
503	*rpp = NULL;
504	if (redir) {
505		if (n1->type != NSUBSHELL) {
506			n2 = (union node *)stalloc(sizeof (struct nredir));
507			n2->type = NREDIR;
508			n2->nredir.n = n1;
509			n1 = n2;
510		}
511		n1->nredir.redirect = redir;
512	}
513
514checkneg:
515	if (negate) {
516		n2 = (union node *)stalloc(sizeof (struct nnot));
517		n2->type = NNOT;
518		n2->nnot.com = n1;
519		return n2;
520	}
521	else
522		return n1;
523}
524
525
526STATIC union node *
527simplecmd(union node **rpp, union node *redir)
528{
529	union node *args, **app;
530	union node **orig_rpp = rpp;
531	union node *n = NULL, *n2;
532	int negate = 0;
533
534	/* If we don't have any redirections already, then we must reset */
535	/* rpp to be the address of the local redir variable.  */
536	if (redir == 0)
537		rpp = &redir;
538
539	args = NULL;
540	app = &args;
541	/*
542	 * We save the incoming value, because we need this for shell
543	 * functions.  There can not be a redirect or an argument between
544	 * the function name and the open parenthesis.
545	 */
546	orig_rpp = rpp;
547
548	while (readtoken() == TNOT) {
549		TRACE(("command: TNOT recognized\n"));
550		negate = !negate;
551	}
552	tokpushback++;
553
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			goto checkneg;
580		} else {
581			tokpushback++;
582			break;
583		}
584	}
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
593checkneg:
594	if (negate) {
595		n2 = (union node *)stalloc(sizeof (struct nnot));
596		n2->type = NNOT;
597		n2->nnot.com = n;
598		return n2;
599	}
600	else
601		return n;
602}
603
604STATIC union node *
605makename(void)
606{
607	union node *n;
608
609	n = (union node *)stalloc(sizeof (struct narg));
610	n->type = NARG;
611	n->narg.next = NULL;
612	n->narg.text = wordtext;
613	n->narg.backquote = backquotelist;
614	return n;
615}
616
617void fixredir(union node *n, const char *text, int err)
618{
619	TRACE(("Fix redir %s %d\n", text, err));
620	if (!err)
621		n->ndup.vname = NULL;
622
623	if (is_digit(text[0]) && text[1] == '\0')
624		n->ndup.dupfd = digit_val(text[0]);
625	else if (text[0] == '-' && text[1] == '\0')
626		n->ndup.dupfd = -1;
627	else {
628
629		if (err)
630			synerror("Bad fd number");
631		else
632			n->ndup.vname = makename();
633	}
634}
635
636
637STATIC void
638parsefname(void)
639{
640	union node *n = redirnode;
641
642	if (readtoken() != TWORD)
643		synexpect(-1);
644	if (n->type == NHERE) {
645		struct heredoc *here = heredoc;
646		struct heredoc *p;
647		int i;
648
649		if (quoteflag == 0)
650			n->type = NXHERE;
651		TRACE(("Here document %d\n", n->type));
652		if (here->striptabs) {
653			while (*wordtext == '\t')
654				wordtext++;
655		}
656		if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN)
657			synerror("Illegal eof marker for << redirection");
658		rmescapes(wordtext);
659		here->eofmark = wordtext;
660		here->next = NULL;
661		if (heredoclist == NULL)
662			heredoclist = here;
663		else {
664			for (p = heredoclist ; p->next ; p = p->next);
665			p->next = here;
666		}
667	} else if (n->type == NTOFD || n->type == NFROMFD) {
668		fixredir(n, wordtext, 0);
669	} else {
670		n->nfile.fname = makename();
671	}
672}
673
674
675/*
676 * Input any here documents.
677 */
678
679STATIC void
680parseheredoc(void)
681{
682	struct heredoc *here;
683	union node *n;
684
685	while (heredoclist) {
686		here = heredoclist;
687		heredoclist = here->next;
688		if (needprompt) {
689			setprompt(2);
690			needprompt = 0;
691		}
692		readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX,
693				here->eofmark, here->striptabs);
694		n = (union node *)stalloc(sizeof (struct narg));
695		n->narg.type = NARG;
696		n->narg.next = NULL;
697		n->narg.text = wordtext;
698		n->narg.backquote = backquotelist;
699		here->here->nhere.doc = n;
700	}
701}
702
703STATIC int
704peektoken(void)
705{
706	int t;
707
708	t = readtoken();
709	tokpushback++;
710	return (t);
711}
712
713STATIC int
714readtoken(void)
715{
716	int t;
717	int savecheckkwd = checkkwd;
718	struct alias *ap;
719#ifdef DEBUG
720	int alreadyseen = tokpushback;
721#endif
722
723	top:
724	t = xxreadtoken();
725
726	if (checkkwd) {
727		/*
728		 * eat newlines
729		 */
730		if (checkkwd == 2) {
731			checkkwd = 0;
732			while (t == TNL) {
733				parseheredoc();
734				t = xxreadtoken();
735			}
736		} else
737			checkkwd = 0;
738		/*
739		 * check for keywords and aliases
740		 */
741		if (t == TWORD && !quoteflag)
742		{
743			const char * const *pp;
744
745			for (pp = parsekwd; *pp; pp++) {
746				if (**pp == *wordtext && equal(*pp, wordtext))
747				{
748					lasttoken = t = pp - parsekwd + KWDOFFSET;
749					TRACE(("keyword %s recognized\n", tokname[t]));
750					goto out;
751				}
752			}
753			if (noaliases == 0 &&
754			    (ap = lookupalias(wordtext, 1)) != NULL) {
755				pushstring(ap->val, strlen(ap->val), ap);
756				checkkwd = savecheckkwd;
757				goto top;
758			}
759		}
760out:
761		checkkwd = (t == TNOT) ? savecheckkwd : 0;
762	}
763#ifdef DEBUG
764	if (!alreadyseen)
765	    TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
766	else
767	    TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
768#endif
769	return (t);
770}
771
772
773/*
774 * Read the next input token.
775 * If the token is a word, we set backquotelist to the list of cmds in
776 *	backquotes.  We set quoteflag to true if any part of the word was
777 *	quoted.
778 * If the token is TREDIR, then we set redirnode to a structure containing
779 *	the redirection.
780 * In all cases, the variable startlinno is set to the number of the line
781 *	on which the token starts.
782 *
783 * [Change comment:  here documents and internal procedures]
784 * [Readtoken shouldn't have any arguments.  Perhaps we should make the
785 *  word parsing code into a separate routine.  In this case, readtoken
786 *  doesn't need to have any internal procedures, but parseword does.
787 *  We could also make parseoperator in essence the main routine, and
788 *  have parseword (readtoken1?) handle both words and redirection.]
789 */
790
791#define RETURN(token)	return lasttoken = token
792
793STATIC int
794xxreadtoken(void)
795{
796	int c;
797
798	if (tokpushback) {
799		tokpushback = 0;
800		return lasttoken;
801	}
802	if (needprompt) {
803		setprompt(2);
804		needprompt = 0;
805	}
806	startlinno = plinno;
807	for (;;) {	/* until token or start of word found */
808		c = pgetc_macro();
809		if (c == ' ' || c == '\t')
810			continue;		/* quick check for white space first */
811		switch (c) {
812		case ' ': case '\t':
813			continue;
814		case '#':
815			while ((c = pgetc()) != '\n' && c != PEOF);
816			pungetc();
817			continue;
818		case '\\':
819			if (pgetc() == '\n') {
820				startlinno = ++plinno;
821				if (doprompt)
822					setprompt(2);
823				else
824					setprompt(0);
825				continue;
826			}
827			pungetc();
828			goto breakloop;
829		case '\n':
830			plinno++;
831			needprompt = doprompt;
832			RETURN(TNL);
833		case PEOF:
834			RETURN(TEOF);
835		case '&':
836			if (pgetc() == '&')
837				RETURN(TAND);
838			pungetc();
839			RETURN(TBACKGND);
840		case '|':
841			if (pgetc() == '|')
842				RETURN(TOR);
843			pungetc();
844			RETURN(TPIPE);
845		case ';':
846			if (pgetc() == ';')
847				RETURN(TENDCASE);
848			pungetc();
849			RETURN(TSEMI);
850		case '(':
851			RETURN(TLP);
852		case ')':
853			RETURN(TRP);
854		default:
855			goto breakloop;
856		}
857	}
858breakloop:
859	return readtoken1(c, BASESYNTAX, (char *)NULL, 0);
860#undef RETURN
861}
862
863
864
865/*
866 * If eofmark is NULL, read a word or a redirection symbol.  If eofmark
867 * is not NULL, read a here document.  In the latter case, eofmark is the
868 * word which marks the end of the document and striptabs is true if
869 * leading tabs should be stripped from the document.  The argument firstc
870 * is the first character of the input token or document.
871 *
872 * Because C does not have internal subroutines, I have simulated them
873 * using goto's to implement the subroutine linkage.  The following macros
874 * will run code that appears at the end of readtoken1.
875 */
876
877#define CHECKEND()	{goto checkend; checkend_return:;}
878#define PARSEREDIR()	{goto parseredir; parseredir_return:;}
879#define PARSESUB()	{goto parsesub; parsesub_return:;}
880#define PARSEBACKQOLD()	{oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;}
881#define PARSEBACKQNEW()	{oldstyle = 0; goto parsebackq; parsebackq_newreturn:;}
882#define	PARSEARITH()	{goto parsearith; parsearith_return:;}
883
884STATIC int
885readtoken1(int firstc, char const *syntax, char *eofmark, int striptabs)
886{
887	int c = firstc;
888	char *out;
889	int len;
890	char line[EOFMARKLEN + 1];
891	struct nodelist *bqlist;
892	int quotef;
893	int dblquote;
894	int varnest;	/* levels of variables expansion */
895	int arinest;	/* levels of arithmetic expansion */
896	int parenlevel;	/* levels of parens in arithmetic */
897	int oldstyle;
898	char const *prevsyntax;	/* syntax before arithmetic */
899	int synentry;
900
901	startlinno = plinno;
902	dblquote = 0;
903	if (syntax == DQSYNTAX)
904		dblquote = 1;
905	quotef = 0;
906	bqlist = NULL;
907	varnest = 0;
908	arinest = 0;
909	parenlevel = 0;
910
911	STARTSTACKSTR(out);
912	loop: {	/* for each line, until end of word */
913		CHECKEND();	/* set c to PEOF if at end of here document */
914		for (;;) {	/* until end of line or end of word */
915			CHECKSTRSPACE(3, out);	/* permit 3 calls to USTPUTC */
916
917			synentry = syntax[c];
918
919			switch(synentry) {
920			case CNL:	/* '\n' */
921				if (syntax == BASESYNTAX)
922					goto endword;	/* exit outer loop */
923				USTPUTC(c, out);
924				plinno++;
925				if (doprompt)
926					setprompt(2);
927				else
928					setprompt(0);
929				c = pgetc();
930				goto loop;		/* continue outer loop */
931			case CWORD:
932				USTPUTC(c, out);
933				break;
934			case CCTL:
935				if (eofmark == NULL || dblquote)
936					USTPUTC(CTLESC, out);
937				USTPUTC(c, out);
938				break;
939			case CBACK:	/* backslash */
940				c = pgetc();
941				if (c == PEOF) {
942					USTPUTC('\\', out);
943					pungetc();
944				} else if (c == '\n') {
945					plinno++;
946					if (doprompt)
947						setprompt(2);
948					else
949						setprompt(0);
950				} else {
951					if (dblquote && c != '\\' &&
952					    c != '`' && c != '$' &&
953					    (c != '"' || eofmark != NULL))
954						USTPUTC('\\', out);
955					if (SQSYNTAX[c] == CCTL)
956						USTPUTC(CTLESC, out);
957					else if (eofmark == NULL)
958						USTPUTC(CTLQUOTEMARK, out);
959					USTPUTC(c, out);
960					quotef++;
961				}
962				break;
963			case CSQUOTE:
964				if (eofmark == NULL)
965					USTPUTC(CTLQUOTEMARK, out);
966				syntax = SQSYNTAX;
967				break;
968			case CDQUOTE:
969				if (eofmark == NULL)
970					USTPUTC(CTLQUOTEMARK, out);
971				syntax = DQSYNTAX;
972				dblquote = 1;
973				break;
974			case CENDQUOTE:
975				if (eofmark != NULL && arinest == 0 &&
976				    varnest == 0) {
977					USTPUTC(c, out);
978				} else {
979					if (arinest) {
980						syntax = ARISYNTAX;
981						dblquote = 0;
982					} else if (eofmark == NULL) {
983						syntax = BASESYNTAX;
984						dblquote = 0;
985					}
986					quotef++;
987				}
988				break;
989			case CVAR:	/* '$' */
990				PARSESUB();		/* parse substitution */
991				break;
992			case CENDVAR:	/* '}' */
993				if (varnest > 0) {
994					varnest--;
995					USTPUTC(CTLENDVAR, out);
996				} else {
997					USTPUTC(c, out);
998				}
999				break;
1000			case CLP:	/* '(' in arithmetic */
1001				parenlevel++;
1002				USTPUTC(c, out);
1003				break;
1004			case CRP:	/* ')' in arithmetic */
1005				if (parenlevel > 0) {
1006					USTPUTC(c, out);
1007					--parenlevel;
1008				} else {
1009					if (pgetc() == ')') {
1010						if (--arinest == 0) {
1011							USTPUTC(CTLENDARI, out);
1012							syntax = prevsyntax;
1013							if (syntax == DQSYNTAX)
1014								dblquote = 1;
1015							else
1016								dblquote = 0;
1017						} else
1018							USTPUTC(')', out);
1019					} else {
1020						/*
1021						 * unbalanced parens
1022						 *  (don't 2nd guess - no error)
1023						 */
1024						pungetc();
1025						USTPUTC(')', out);
1026					}
1027				}
1028				break;
1029			case CBQUOTE:	/* '`' */
1030				PARSEBACKQOLD();
1031				break;
1032			case CEOF:
1033				goto endword;		/* exit outer loop */
1034			default:
1035				if (varnest == 0)
1036					goto endword;	/* exit outer loop */
1037				USTPUTC(c, out);
1038			}
1039			c = pgetc_macro();
1040		}
1041	}
1042endword:
1043	if (syntax == ARISYNTAX)
1044		synerror("Missing '))'");
1045	if (syntax != BASESYNTAX && eofmark == NULL)
1046		synerror("Unterminated quoted string");
1047	if (varnest != 0) {
1048		startlinno = plinno;
1049		synerror("Missing '}'");
1050	}
1051	USTPUTC('\0', out);
1052	len = out - stackblock();
1053	out = stackblock();
1054	if (eofmark == NULL) {
1055		if ((c == '>' || c == '<')
1056		 && quotef == 0
1057		 && len <= 2
1058		 && (*out == '\0' || is_digit(*out))) {
1059			PARSEREDIR();
1060			return lasttoken = TREDIR;
1061		} else {
1062			pungetc();
1063		}
1064	}
1065	quoteflag = quotef;
1066	backquotelist = bqlist;
1067	grabstackblock(len);
1068	wordtext = out;
1069	return lasttoken = TWORD;
1070/* end of readtoken routine */
1071
1072
1073
1074/*
1075 * Check to see whether we are at the end of the here document.  When this
1076 * is called, c is set to the first character of the next input line.  If
1077 * we are at the end of the here document, this routine sets the c to PEOF.
1078 */
1079
1080checkend: {
1081	if (eofmark) {
1082		if (striptabs) {
1083			while (c == '\t')
1084				c = pgetc();
1085		}
1086		if (c == *eofmark) {
1087			if (pfgets(line, sizeof line) != NULL) {
1088				char *p, *q;
1089
1090				p = line;
1091				for (q = eofmark + 1 ; *q && *p == *q ; p++, q++);
1092				if (*p == '\n' && *q == '\0') {
1093					c = PEOF;
1094					plinno++;
1095					needprompt = doprompt;
1096				} else {
1097					pushstring(line, strlen(line), NULL);
1098				}
1099			}
1100		}
1101	}
1102	goto checkend_return;
1103}
1104
1105
1106/*
1107 * Parse a redirection operator.  The variable "out" points to a string
1108 * specifying the fd to be redirected.  The variable "c" contains the
1109 * first character of the redirection operator.
1110 */
1111
1112parseredir: {
1113	char fd = *out;
1114	union node *np;
1115
1116	np = (union node *)stalloc(sizeof (struct nfile));
1117	if (c == '>') {
1118		np->nfile.fd = 1;
1119		c = pgetc();
1120		if (c == '>')
1121			np->type = NAPPEND;
1122		else if (c == '&')
1123			np->type = NTOFD;
1124		else if (c == '|')
1125			np->type = NCLOBBER;
1126		else {
1127			np->type = NTO;
1128			pungetc();
1129		}
1130	} else {	/* c == '<' */
1131		np->nfile.fd = 0;
1132		c = pgetc();
1133		if (c == '<') {
1134			if (sizeof (struct nfile) != sizeof (struct nhere)) {
1135				np = (union node *)stalloc(sizeof (struct nhere));
1136				np->nfile.fd = 0;
1137			}
1138			np->type = NHERE;
1139			heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc));
1140			heredoc->here = np;
1141			if ((c = pgetc()) == '-') {
1142				heredoc->striptabs = 1;
1143			} else {
1144				heredoc->striptabs = 0;
1145				pungetc();
1146			}
1147		} else if (c == '&')
1148			np->type = NFROMFD;
1149		else if (c == '>')
1150			np->type = NFROMTO;
1151		else {
1152			np->type = NFROM;
1153			pungetc();
1154		}
1155	}
1156	if (fd != '\0')
1157		np->nfile.fd = digit_val(fd);
1158	redirnode = np;
1159	goto parseredir_return;
1160}
1161
1162
1163/*
1164 * Parse a substitution.  At this point, we have read the dollar sign
1165 * and nothing else.
1166 */
1167
1168parsesub: {
1169	char buf[10];
1170	int subtype;
1171	int typeloc;
1172	int flags;
1173	char *p;
1174	static const char types[] = "}-+?=";
1175	int bracketed_name = 0; /* used to handle ${[0-9]*} variables */
1176	int i;
1177	int linno;
1178	int length;
1179
1180	c = pgetc();
1181	if (c != '(' && c != '{' && (is_eof(c) || !is_name(c)) &&
1182	    !is_special(c)) {
1183		USTPUTC('$', out);
1184		pungetc();
1185	} else if (c == '(') {	/* $(command) or $((arith)) */
1186		if (pgetc() == '(') {
1187			PARSEARITH();
1188		} else {
1189			pungetc();
1190			PARSEBACKQNEW();
1191		}
1192	} else {
1193		USTPUTC(CTLVAR, out);
1194		typeloc = out - stackblock();
1195		USTPUTC(VSNORMAL, out);
1196		subtype = VSNORMAL;
1197		flags = 0;
1198		if (c == '{') {
1199			bracketed_name = 1;
1200			c = pgetc();
1201			if (c == '#') {
1202				if ((c = pgetc()) == '}')
1203					c = '#';
1204				else
1205					subtype = VSLENGTH;
1206			}
1207			else
1208				subtype = 0;
1209		}
1210		if (!is_eof(c) && is_name(c)) {
1211			length = 0;
1212			do {
1213				STPUTC(c, out);
1214				c = pgetc();
1215				length++;
1216			} while (!is_eof(c) && is_in_name(c));
1217			if (length == 6 &&
1218			    strncmp(out - length, "LINENO", length) == 0) {
1219				/* Replace the variable name with the
1220				 * current line number. */
1221				linno = plinno;
1222				if (funclinno != 0)
1223					linno -= funclinno - 1;
1224				snprintf(buf, sizeof(buf), "%d", linno);
1225				STADJUST(-6, out);
1226				for (i = 0; buf[i] != '\0'; i++)
1227					STPUTC(buf[i], out);
1228				flags |= VSLINENO;
1229			}
1230		} else if (is_digit(c)) {
1231			if (bracketed_name) {
1232				do {
1233					STPUTC(c, out);
1234					c = pgetc();
1235				} while (is_digit(c));
1236			} else {
1237				STPUTC(c, out);
1238				c = pgetc();
1239			}
1240		} else {
1241			if (! is_special(c)) {
1242				subtype = VSERROR;
1243				if (c == '}')
1244					pungetc();
1245				else
1246					USTPUTC(c, out);
1247			} else {
1248				USTPUTC(c, out);
1249				c = pgetc();
1250			}
1251		}
1252		if (subtype == 0) {
1253			switch (c) {
1254			case ':':
1255				flags |= VSNUL;
1256				c = pgetc();
1257				/*FALLTHROUGH*/
1258			default:
1259				p = strchr(types, c);
1260				if (p == NULL) {
1261					if (flags == VSNUL)
1262						STPUTC(':', out);
1263					STPUTC(c, out);
1264					subtype = VSERROR;
1265				} else
1266					subtype = p - types + VSNORMAL;
1267				break;
1268			case '%':
1269			case '#':
1270				{
1271					int cc = c;
1272					subtype = c == '#' ? VSTRIMLEFT :
1273							     VSTRIMRIGHT;
1274					c = pgetc();
1275					if (c == cc)
1276						subtype++;
1277					else
1278						pungetc();
1279					break;
1280				}
1281			}
1282		} else if (subtype != VSERROR) {
1283			pungetc();
1284		}
1285		STPUTC('=', out);
1286		if (subtype != VSLENGTH && (dblquote || arinest))
1287			flags |= VSQUOTE;
1288		*(stackblock() + typeloc) = subtype | flags;
1289		if (subtype != VSNORMAL)
1290			varnest++;
1291	}
1292	goto parsesub_return;
1293}
1294
1295
1296/*
1297 * Called to parse command substitutions.  Newstyle is set if the command
1298 * is enclosed inside $(...); nlpp is a pointer to the head of the linked
1299 * list of commands (passed by reference), and savelen is the number of
1300 * characters on the top of the stack which must be preserved.
1301 */
1302
1303parsebackq: {
1304	struct nodelist **nlpp;
1305	union node *n;
1306	char *volatile str;
1307	struct jmploc jmploc;
1308	struct jmploc *const savehandler = handler;
1309	int savelen;
1310	int saveprompt;
1311
1312	if (setjmp(jmploc.loc)) {
1313		if (str)
1314			ckfree(str);
1315		handler = savehandler;
1316		longjmp(handler->loc, 1);
1317	}
1318	INTOFF;
1319	str = NULL;
1320	savelen = out - stackblock();
1321	if (savelen > 0) {
1322		str = ckmalloc(savelen);
1323		memcpy(str, stackblock(), savelen);
1324	}
1325	handler = &jmploc;
1326	INTON;
1327        if (oldstyle) {
1328                /* We must read until the closing backquote, giving special
1329                   treatment to some slashes, and then push the string and
1330                   reread it as input, interpreting it normally.  */
1331                char *out;
1332                int c;
1333                int savelen;
1334                char *str;
1335
1336
1337                STARTSTACKSTR(out);
1338		for (;;) {
1339			if (needprompt) {
1340				setprompt(2);
1341				needprompt = 0;
1342			}
1343			switch (c = pgetc()) {
1344			case '`':
1345				goto done;
1346
1347			case '\\':
1348                                if ((c = pgetc()) == '\n') {
1349					plinno++;
1350					if (doprompt)
1351						setprompt(2);
1352					else
1353						setprompt(0);
1354					/*
1355					 * If eating a newline, avoid putting
1356					 * the newline into the new character
1357					 * stream (via the STPUTC after the
1358					 * switch).
1359					 */
1360					continue;
1361				}
1362                                if (c != '\\' && c != '`' && c != '$'
1363                                    && (!dblquote || c != '"'))
1364                                        STPUTC('\\', out);
1365				break;
1366
1367			case '\n':
1368				plinno++;
1369				needprompt = doprompt;
1370				break;
1371
1372			case PEOF:
1373			        startlinno = plinno;
1374				synerror("EOF in backquote substitution");
1375 				break;
1376
1377			default:
1378				break;
1379			}
1380			STPUTC(c, out);
1381                }
1382done:
1383                STPUTC('\0', out);
1384                savelen = out - stackblock();
1385                if (savelen > 0) {
1386                        str = ckmalloc(savelen);
1387                        memcpy(str, stackblock(), savelen);
1388			setinputstring(str, 1);
1389                }
1390        }
1391	nlpp = &bqlist;
1392	while (*nlpp)
1393		nlpp = &(*nlpp)->next;
1394	*nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist));
1395	(*nlpp)->next = NULL;
1396
1397	if (oldstyle) {
1398		saveprompt = doprompt;
1399		doprompt = 0;
1400	}
1401
1402	n = list(0);
1403
1404	if (oldstyle)
1405		doprompt = saveprompt;
1406	else {
1407		if (readtoken() != TRP)
1408			synexpect(TRP);
1409	}
1410
1411	(*nlpp)->n = n;
1412        if (oldstyle) {
1413		/*
1414		 * Start reading from old file again, ignoring any pushed back
1415		 * tokens left from the backquote parsing
1416		 */
1417                popfile();
1418		tokpushback = 0;
1419	}
1420	while (stackblocksize() <= savelen)
1421		growstackblock();
1422	STARTSTACKSTR(out);
1423	if (str) {
1424		memcpy(out, str, savelen);
1425		STADJUST(savelen, out);
1426		INTOFF;
1427		ckfree(str);
1428		str = NULL;
1429		INTON;
1430	}
1431	handler = savehandler;
1432	if (arinest || dblquote)
1433		USTPUTC(CTLBACKQ | CTLQUOTE, out);
1434	else
1435		USTPUTC(CTLBACKQ, out);
1436	if (oldstyle)
1437		goto parsebackq_oldreturn;
1438	else
1439		goto parsebackq_newreturn;
1440}
1441
1442/*
1443 * Parse an arithmetic expansion (indicate start of one and set state)
1444 */
1445parsearith: {
1446
1447	if (++arinest == 1) {
1448		prevsyntax = syntax;
1449		syntax = ARISYNTAX;
1450		USTPUTC(CTLARI, out);
1451		if (dblquote)
1452			USTPUTC('"',out);
1453		else
1454			USTPUTC(' ',out);
1455	} else {
1456		/*
1457		 * we collapse embedded arithmetic expansion to
1458		 * parenthesis, which should be equivalent
1459		 */
1460		USTPUTC('(', out);
1461	}
1462	goto parsearith_return;
1463}
1464
1465} /* end of readtoken */
1466
1467
1468
1469#ifdef mkinit
1470RESET {
1471	tokpushback = 0;
1472	checkkwd = 0;
1473}
1474#endif
1475
1476/*
1477 * Returns true if the text contains nothing to expand (no dollar signs
1478 * or backquotes).
1479 */
1480
1481STATIC int
1482noexpand(char *text)
1483{
1484	char *p;
1485	char c;
1486
1487	p = text;
1488	while ((c = *p++) != '\0') {
1489		if ( c == CTLQUOTEMARK)
1490			continue;
1491		if (c == CTLESC)
1492			p++;
1493		else if (BASESYNTAX[(int)c] == CCTL)
1494			return 0;
1495	}
1496	return 1;
1497}
1498
1499
1500/*
1501 * Return true if the argument is a legal variable name (a letter or
1502 * underscore followed by zero or more letters, underscores, and digits).
1503 */
1504
1505int
1506goodname(char *name)
1507{
1508	char *p;
1509
1510	p = name;
1511	if (! is_name(*p))
1512		return 0;
1513	while (*++p) {
1514		if (! is_in_name(*p))
1515			return 0;
1516	}
1517	return 1;
1518}
1519
1520
1521/*
1522 * Called when an unexpected token is read during the parse.  The argument
1523 * is the token that is expected, or -1 if more than one type of token can
1524 * occur at this point.
1525 */
1526
1527STATIC void
1528synexpect(int token)
1529{
1530	char msg[64];
1531
1532	if (token >= 0) {
1533		fmtstr(msg, 64, "%s unexpected (expecting %s)",
1534			tokname[lasttoken], tokname[token]);
1535	} else {
1536		fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]);
1537	}
1538	synerror(msg);
1539}
1540
1541
1542STATIC void
1543synerror(char *msg)
1544{
1545	if (commandname)
1546		outfmt(&errout, "%s: %d: ", commandname, startlinno);
1547	outfmt(&errout, "Syntax error: %s\n", msg);
1548	error((char *)NULL);
1549}
1550
1551STATIC void
1552setprompt(int which)
1553{
1554	whichprompt = which;
1555
1556#ifndef NO_HISTORY
1557	if (!el)
1558#endif
1559		out2str(getprompt(NULL));
1560}
1561
1562/*
1563 * called by editline -- any expansions to the prompt
1564 *    should be added here.
1565 */
1566char *
1567getprompt(void *unused __unused)
1568{
1569	static char ps[PROMPTLEN];
1570	char *fmt;
1571	int i, j, trim;
1572
1573	/*
1574	 * Select prompt format.
1575	 */
1576	switch (whichprompt) {
1577	case 0:
1578		fmt = "";
1579		break;
1580	case 1:
1581		fmt = ps1val();
1582		break;
1583	case 2:
1584		fmt = ps2val();
1585		break;
1586	default:
1587		return "<internal prompt error>";
1588	}
1589
1590	/*
1591	 * Format prompt string.
1592	 */
1593	for (i = 0; (i < 127) && (*fmt != '\0'); i++, fmt++)
1594		if (*fmt == '\\')
1595			switch (*++fmt) {
1596
1597				/*
1598				 * Hostname.
1599				 *
1600				 * \h specifies just the local hostname,
1601				 * \H specifies fully-qualified hostname.
1602				 */
1603			case 'h':
1604			case 'H':
1605				ps[i] = '\0';
1606				gethostname(&ps[i], PROMPTLEN - i);
1607				/* Skip to end of hostname. */
1608				trim = (*fmt == 'h') ? '.' : '\0';
1609				while ((ps[i+1] != '\0') && (ps[i+1] != trim))
1610					i++;
1611				break;
1612
1613				/*
1614				 * Working directory.
1615				 *
1616				 * \W specifies just the final component,
1617				 * \w specifies the entire path.
1618				 */
1619			case 'W':
1620			case 'w':
1621				ps[i] = '\0';
1622				getcwd(&ps[i], PROMPTLEN - i);
1623				if (*fmt == 'W') {
1624					/* Final path component only. */
1625					trim = 1;
1626					for (j = i; ps[j] != '\0'; j++)
1627					  if (ps[j] == '/')
1628						trim = j + 1;
1629					memmove(&ps[i], &ps[trim],
1630					    j - trim + 1);
1631				}
1632				/* Skip to end of path. */
1633				while (ps[i + 1] != '\0')
1634					i++;
1635				break;
1636
1637				/*
1638				 * Superuser status.
1639				 *
1640				 * '$' for normal users, '#' for root.
1641				 */
1642			case '$':
1643				ps[i] = (geteuid() != 0) ? '$' : '#';
1644				break;
1645
1646				/*
1647				 * A literal \.
1648				 */
1649			case '\\':
1650				ps[i] = '\\';
1651				break;
1652
1653				/*
1654				 * Emit unrecognized formats verbatim.
1655				 */
1656			default:
1657				ps[i++] = '\\';
1658				ps[i] = *fmt;
1659				break;
1660			}
1661		else
1662			ps[i] = *fmt;
1663	ps[i] = '\0';
1664	return (ps);
1665}
1666