xref: /netbsd/bin/csh/lex.c (revision bf9ec67e)
1 /* $NetBSD: lex.c,v 1.21 2002/05/25 23:29:16 wiz Exp $ */
2 
3 /*-
4  * Copyright (c) 1980, 1991, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. All advertising materials mentioning features or use of this software
16  *    must display the following acknowledgement:
17  *	This product includes software developed by the University of
18  *	California, Berkeley and its contributors.
19  * 4. Neither the name of the University nor the names of its contributors
20  *    may be used to endorse or promote products derived from this software
21  *    without specific prior written permission.
22  *
23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33  * SUCH DAMAGE.
34  */
35 
36 #include <sys/cdefs.h>
37 #ifndef lint
38 #if 0
39 static char sccsid[] = "@(#)lex.c	8.1 (Berkeley) 5/31/93";
40 #else
41 __RCSID("$NetBSD: lex.c,v 1.21 2002/05/25 23:29:16 wiz Exp $");
42 #endif
43 #endif /* not lint */
44 
45 #include <sys/ioctl.h>
46 #include <sys/types.h>
47 
48 #include <errno.h>
49 #include <stdarg.h>
50 #include <stdlib.h>
51 #include <string.h>
52 #include <termios.h>
53 #include <unistd.h>
54 
55 #include "csh.h"
56 #include "extern.h"
57 
58 /*
59  * These lexical routines read input and form lists of words.
60  * There is some involved processing here, because of the complications
61  * of input buffering, and especially because of history substitution.
62  */
63 
64 static Char *word(void);
65 static int getC1(int);
66 static void getdol(void);
67 static void getexcl(int);
68 static struct Hist *findev(Char *, bool);
69 static void setexclp(Char *);
70 static int bgetc(void);
71 static void bfree(void);
72 static struct wordent *gethent(int);
73 static int matchs(Char *, Char *);
74 static int getsel(int *, int *, int);
75 static struct wordent *getsub(struct wordent *);
76 static Char *subword(Char *, int, bool *);
77 static struct wordent *dosub(int, struct wordent *, bool);
78 
79 /*
80  * Peekc is a peek character for getC, peekread for readc.
81  * There is a subtlety here in many places... history routines
82  * will read ahead and then insert stuff into the input stream.
83  * If they push back a character then they must push it behind
84  * the text substituted by the history substitution.  On the other
85  * hand in several places we need 2 peek characters.  To make this
86  * all work, the history routines read with getC, and make use both
87  * of ungetC and unreadc.  The key observation is that the state
88  * of getC at the call of a history reference is such that calls
89  * to getC from the history routines will always yield calls of
90  * readc, unless this peeking is involved.  That is to say that during
91  * getexcl the variables lap, exclp, and exclnxt are all zero.
92  *
93  * Getdol invokes history substitution, hence the extra peek, peekd,
94  * which it can ungetD to be before history substitutions.
95  */
96 static Char peekc = 0, peekd = 0;
97 static Char peekread = 0;
98 
99 /* (Tail of) current word from ! subst */
100 static Char *exclp = NULL;
101 
102 /* The rest of the ! subst words */
103 static struct wordent *exclnxt = NULL;
104 
105 /* Count of remaining words in ! subst */
106 static int exclc = 0;
107 
108 /* "Globp" for alias resubstitution */
109 Char **alvec, *alvecp;
110 int aret = F_SEEK;
111 
112 /*
113  * Labuf implements a general buffer for lookahead during lexical operations.
114  * Text which is to be placed in the input stream can be stuck here.
115  * We stick parsed ahead $ constructs during initial input,
116  * process id's from `$$', and modified variable values (from qualifiers
117  * during expansion in sh.dol.c) here.
118  */
119 static Char labuf[BUFSIZE];
120 
121 /*
122  * Lex returns to its caller not only a wordlist (as a "var" parameter)
123  * but also whether a history substitution occurred.  This is used in
124  * the main (process) routine to determine whether to echo, and also
125  * when called by the alias routine to determine whether to keep the
126  * argument list.
127  */
128 static bool hadhist = 0;
129 
130 /*
131  * Avoid alias expansion recursion via \!#
132  */
133 int     hleft;
134 
135 static Char getCtmp;
136 
137 #define getC(f) ((getCtmp = peekc) ? (peekc = 0, getCtmp) : getC1(f))
138 #define	ungetC(c) peekc = c
139 #define	ungetD(c) peekd = c
140 
141 int
142 lex(struct wordent *hp)
143 {
144     struct wordent *wdp;
145     int c;
146 
147     btell(&lineloc);
148     hp->next = hp->prev = hp;
149     hp->word = STRNULL;
150     hadhist = 0;
151     do
152 	c = readc(0);
153     while (c == ' ' || c == '\t');
154     if (c == HISTSUB && intty)
155 	/* ^lef^rit	from tty is short !:s^lef^rit */
156 	getexcl(c);
157     else
158 	unreadc(c);
159     wdp = hp;
160     /*
161      * The following loop is written so that the links needed by freelex will
162      * be ready and rarin to go even if it is interrupted.
163      */
164     do {
165 	struct wordent *new;
166 
167 	new = (struct wordent *)xmalloc((size_t)sizeof(*wdp));
168 	new->word = 0;
169 	new->prev = wdp;
170 	new->next = hp;
171 	wdp->next = new;
172 	wdp = new;
173 	wdp->word = word();
174     } while (wdp->word[0] != '\n');
175     hp->prev = wdp;
176     return (hadhist);
177 }
178 
179 void
180 prlex(FILE *fp, struct wordent *sp0)
181 {
182     struct wordent *sp;
183 
184     sp = sp0->next;
185     for (;;) {
186 	(void)fprintf(fp, "%s", vis_str(sp->word));
187 	sp = sp->next;
188 	if (sp == sp0)
189 	    break;
190 	if (sp->word[0] != '\n')
191 	    (void) fputc(' ', fp);
192     }
193 }
194 
195 void
196 copylex(struct wordent *hp, struct wordent *fp)
197 {
198     struct wordent *wdp;
199 
200     wdp = hp;
201     fp = fp->next;
202     do {
203 	struct wordent *new;
204 
205 	new = (struct wordent *)xmalloc((size_t)sizeof(*wdp));
206 	new->prev = wdp;
207 	new->next = hp;
208 	wdp->next = new;
209 	wdp = new;
210 	wdp->word = Strsave(fp->word);
211 	fp = fp->next;
212     } while (wdp->word[0] != '\n');
213     hp->prev = wdp;
214 }
215 
216 void
217 freelex(struct wordent *vp)
218 {
219     struct wordent *fp;
220 
221     while (vp->next != vp) {
222 	fp = vp->next;
223 	vp->next = fp->next;
224 	xfree((ptr_t) fp->word);
225 	xfree((ptr_t) fp);
226     }
227     vp->prev = vp;
228 }
229 
230 static Char *
231 word(void)
232 {
233     Char wbuf[BUFSIZE], *wp;
234     int i;
235     Char c, c1;
236     bool dolflg;
237 
238     wp = wbuf;
239     i = BUFSIZE - 4;
240 loop:
241     while ((c = getC(DOALL)) == ' ' || c == '\t')
242 	continue;
243     if (cmap(c, _META | _ESC))
244 	switch (c) {
245 	case '&':
246 	case '|':
247 	case '<':
248 	case '>':
249 	    *wp++ = c;
250 	    c1 = getC(DOALL);
251 	    if (c1 == c)
252 		*wp++ = c1;
253 	    else
254 		ungetC(c1);
255 	    goto ret;
256 
257 	case '#':
258 	    if (intty)
259 		break;
260 	    c = 0;
261 	    do {
262 		c1 = c;
263 		c = getC(0);
264 	    } while (c != '\n');
265 	    if (c1 == '\\')
266 		goto loop;
267 	    /* FALLTHROUGH */
268 
269 	case ';':
270 	case '(':
271 	case ')':
272 	case '\n':
273 	    *wp++ = c;
274 	    goto ret;
275 
276 	case '\\':
277 	    c = getC(0);
278 	    if (c == '\n') {
279 		if (onelflg == 1)
280 		    onelflg = 2;
281 		goto loop;
282 	    }
283 	    if (c != HIST)
284 		*wp++ = '\\', --i;
285 	    c |= QUOTE;
286 	    break;
287 	}
288     c1 = 0;
289     dolflg = DOALL;
290     for (;;) {
291 	if (c1) {
292 	    if (c == c1) {
293 		c1 = 0;
294 		dolflg = DOALL;
295 	    }
296 	    else if (c == '\\') {
297 		c = getC(0);
298 		if (c == HIST)
299 		    c |= QUOTE;
300 		else {
301 		    if (c == '\n')
302 			/*
303 			 * if (c1 == '`') c = ' '; else
304 			 */
305 			c |= QUOTE;
306 		    ungetC(c);
307 		    c = '\\';
308 		}
309 	    }
310 	    else if (c == '\n') {
311 		seterror(ERR_UNMATCHED, c1);
312 		ungetC(c);
313 		break;
314 	    }
315 	}
316 	else if (cmap(c, _META | _QF | _QB | _ESC)) {
317 	    if (c == '\\') {
318 		c = getC(0);
319 		if (c == '\n') {
320 		    if (onelflg == 1)
321 			onelflg = 2;
322 		    break;
323 		}
324 		if (c != HIST)
325 		    *wp++ = '\\', --i;
326 		c |= QUOTE;
327 	    }
328 	    else if (cmap(c, _QF | _QB)) {	/* '"` */
329 		c1 = c;
330 		dolflg = c == '"' ? DOALL : DOEXCL;
331 	    }
332 	    else if (c != '#' || !intty) {
333 		ungetC(c);
334 		break;
335 	    }
336 	}
337 	if (--i > 0) {
338 	    *wp++ = c;
339 	    c = getC(dolflg);
340 	}
341 	else {
342 	    seterror(ERR_WTOOLONG);
343 	    wp = &wbuf[1];
344 	    break;
345 	}
346     }
347 ret:
348     *wp = 0;
349     return (Strsave(wbuf));
350 }
351 
352 static int
353 getC1(int flag)
354 {
355     Char c;
356 
357     for (;;) {
358 	if ((c = peekc) != '\0') {
359 	    peekc = 0;
360 	    return (c);
361 	}
362 	if (lap) {
363 	    if ((c = *lap++) == 0)
364 		lap = 0;
365 	    else {
366 		if (cmap(c, _META | _QF | _QB))
367 		    c |= QUOTE;
368 		return (c);
369 	    }
370 	}
371 	if ((c = peekd) != '\0') {
372 	    peekd = 0;
373 	    return (c);
374 	}
375 	if (exclp) {
376 	    if ((c = *exclp++) != '\0')
377 		return (c);
378 	    if (exclnxt && --exclc >= 0) {
379 		exclnxt = exclnxt->next;
380 		setexclp(exclnxt->word);
381 		return (' ');
382 	    }
383 	    exclp = 0;
384 	    exclnxt = 0;
385 	}
386 	if (exclnxt) {
387 	    exclnxt = exclnxt->next;
388 	    if (--exclc < 0)
389 		exclnxt = 0;
390 	    else
391 		setexclp(exclnxt->word);
392 	    continue;
393 	}
394 	c = readc(0);
395 	if (c == '$' && (flag & DODOL)) {
396 	    getdol();
397 	    continue;
398 	}
399 	if (c == HIST && (flag & DOEXCL)) {
400 	    getexcl(0);
401 	    continue;
402 	}
403 	break;
404     }
405     return (c);
406 }
407 
408 static void
409 getdol(void)
410 {
411     Char name[4*MAXVARLEN+1], *ep, *np;
412     int c, sc;
413     bool special, toolong;
414 
415     special = 0;
416     np = name, *np++ = '$';
417     c = sc = getC(DOEXCL);
418     if (any("\t \n", c)) {
419 	ungetD(c);
420 	ungetC('$' | QUOTE);
421 	return;
422     }
423     if (c == '{')
424 	*np++ = c, c = getC(DOEXCL);
425     if (c == '#' || c == '?')
426 	special++, *np++ = c, c = getC(DOEXCL);
427     *np++ = c;
428     switch (c) {
429     case '<':
430     case '$':
431     case '!':
432 	if (special)
433 	    seterror(ERR_SPDOLLT);
434 	*np = 0;
435 	addla(name);
436 	return;
437     case '\n':
438 	ungetD(c);
439 	np--;
440 	seterror(ERR_NEWLINE);
441 	*np = 0;
442 	addla(name);
443 	return;
444     case '*':
445 	if (special)
446 	    seterror(ERR_SPSTAR);
447 	*np = 0;
448 	addla(name);
449 	return;
450     default:
451 	toolong = 0;
452 	if (Isdigit(c)) {
453 #ifdef notdef
454 	    /* let $?0 pass for now */
455 	    if (special) {
456 		seterror(ERR_DIGIT);
457 		*np = 0;
458 		addla(name);
459 		return;
460 	    }
461 #endif
462 	    /* we know that np < &name[4] */
463 	    ep = &np[MAXVARLEN];
464 	    while ((c = getC(DOEXCL)) != '\0'){
465 		if (!Isdigit(c))
466 		    break;
467 		if (np < ep)
468 		    *np++ = c;
469 		else
470 		    toolong = 1;
471 	    }
472 	}
473 	else if (letter(c)) {
474 	    /* we know that np < &name[4] */
475 	    ep = &np[MAXVARLEN];
476 	    toolong = 0;
477 	    while ((c = getC(DOEXCL)) != '\0') {
478 		/* Bugfix for ${v123x} from Chris Torek, DAS DEC-90. */
479 		if (!letter(c) && !Isdigit(c))
480 		    break;
481 		if (np < ep)
482 		    *np++ = c;
483 		else
484 		    toolong = 1;
485 	    }
486 	}
487 	else {
488 	    *np = 0;
489 	    seterror(ERR_VARILL);
490 	    addla(name);
491 	    return;
492 	}
493 	if (toolong) {
494 	    seterror(ERR_VARTOOLONG);
495 	    *np = 0;
496 	    addla(name);
497 	    return;
498 	}
499 	break;
500     }
501     if (c == '[') {
502 	*np++ = c;
503 	/*
504 	 * Name up to here is a max of MAXVARLEN + 8.
505 	 */
506 	ep = &np[2 * MAXVARLEN + 8];
507 	do {
508 	    /*
509 	     * Michael Greim: Allow $ expansion to take place in selector
510 	     * expressions. (limits the number of characters returned)
511 	     */
512 	    c = getC(DOEXCL | DODOL);
513 	    if (c == '\n') {
514 		ungetD(c);
515 		np--;
516 		seterror(ERR_NLINDEX);
517 		*np = 0;
518 		addla(name);
519 		return;
520 	    }
521 	    if (np < ep)
522 		*np++ = c;
523 	} while (c != ']');
524 	*np = '\0';
525 	if (np >= ep) {
526 	    seterror(ERR_SELOVFL);
527 	    addla(name);
528 	    return;
529 	}
530 	c = getC(DOEXCL);
531     }
532     /*
533      * Name up to here is a max of 2 * MAXVARLEN + 8.
534      */
535     if (c == ':') {
536 	/*
537 	 * if the :g modifier is followed by a newline, then error right away!
538 	 * -strike
539 	 */
540 	int amodflag, gmodflag;
541 
542 	amodflag = 0;
543 	gmodflag = 0;
544 	do {
545 	    *np++ = c, c = getC(DOEXCL);
546 	    if (c == 'g' || c == 'a') {
547 		if (c == 'g')
548 		    gmodflag++;
549 		else
550 		    amodflag++;
551 		*np++ = c; c = getC(DOEXCL);
552 	    }
553 	    if ((c == 'g' && !gmodflag) || (c == 'a' && !amodflag)) {
554 		if (c == 'g')
555 		    gmodflag++;
556 		else
557 		    amodflag++;
558 		*np++ = c; c = getC(DOEXCL);
559 	    }
560 	    *np++ = c;
561 	    /* scan s// [eichin:19910926.0512EST] */
562 	    if (c == 's') {
563 		int delimcnt = 2;
564 		int delim = getC(0);
565 		*np++ = delim;
566 
567 		if (!delim || letter(delim)
568 		    || Isdigit(delim) || any(" \t\n", delim)) {
569 		    seterror(ERR_BADSUBST);
570 		    break;
571 		}
572 		while ((c = getC(0)) != (-1)) {
573 		    *np++ = c;
574 		    if(c == delim) delimcnt--;
575 		    if(!delimcnt) break;
576 		}
577 		if(delimcnt) {
578 		    seterror(ERR_BADSUBST);
579 		    break;
580 		}
581 		c = 's';
582 	    }
583 	    if (!any("htrqxes", c)) {
584 		if ((amodflag || gmodflag) && c == '\n')
585 		    stderror(ERR_VARSYN);	/* strike */
586 		seterror(ERR_VARMOD, c);
587 		*np = 0;
588 		addla(name);
589 		return;
590 	    }
591 	}
592 	while ((c = getC(DOEXCL)) == ':');
593 	ungetD(c);
594     }
595     else
596 	ungetD(c);
597     if (sc == '{') {
598 	c = getC(DOEXCL);
599 	if (c != '}') {
600 	    ungetD(c);
601 	    seterror(ERR_MISSING, '}');
602 	    *np = 0;
603 	    addla(name);
604 	    return;
605 	}
606 	*np++ = c;
607     }
608     *np = 0;
609     addla(name);
610     return;
611 }
612 
613 void
614 addla(Char *cp)
615 {
616     Char buf[BUFSIZE];
617 
618     if (Strlen(cp) + (lap ? Strlen(lap) : 0) >=
619 	(sizeof(labuf) - 4) / sizeof(Char)) {
620 	seterror(ERR_EXPOVFL);
621 	return;
622     }
623     if (lap)
624 	(void)Strcpy(buf, lap);
625     (void)Strcpy(labuf, cp);
626     if (lap)
627 	(void)Strcat(labuf, buf);
628     lap = labuf;
629 }
630 
631 static Char lhsb[32];
632 static Char slhs[32];
633 static Char rhsb[64];
634 static int quesarg;
635 
636 static void
637 getexcl(int sc)
638 {
639     struct wordent *hp, *ip;
640     int c, dol, left, right;
641 
642     if (sc == 0) {
643 	sc = getC(0);
644 	if (sc != '{') {
645 	    ungetC(sc);
646 	    sc = 0;
647 	}
648     }
649     quesarg = -1;
650     lastev = eventno;
651     hp = gethent(sc);
652     if (hp == 0)
653 	return;
654     hadhist = 1;
655     dol = 0;
656     if (hp == alhistp)
657 	for (ip = hp->next->next; ip != alhistt; ip = ip->next)
658 	    dol++;
659     else
660 	for (ip = hp->next->next; ip != hp->prev; ip = ip->next)
661 	    dol++;
662     left = 0, right = dol;
663     if (sc == HISTSUB) {
664 	ungetC('s'), unreadc(HISTSUB), c = ':';
665 	goto subst;
666     }
667     c = getC(0);
668     if (!any(":^$*-%", c))
669 	goto subst;
670     left = right = -1;
671     if (c == ':') {
672 	c = getC(0);
673 	unreadc(c);
674 	if (letter(c) || c == '&') {
675 	    c = ':';
676 	    left = 0, right = dol;
677 	    goto subst;
678 	}
679     }
680     else
681 	ungetC(c);
682     if (!getsel(&left, &right, dol))
683 	return;
684     c = getC(0);
685     if (c == '*')
686 	ungetC(c), c = '-';
687     if (c == '-') {
688 	if (!getsel(&left, &right, dol))
689 	    return;
690 	c = getC(0);
691     }
692 subst:
693     exclc = right - left + 1;
694     while (--left >= 0)
695 	hp = hp->next;
696     if (sc == HISTSUB || c == ':') {
697 	do {
698 	    hp = getsub(hp);
699 	    c = getC(0);
700 	} while (c == ':');
701     }
702     unreadc(c);
703     if (sc == '{') {
704 	c = getC(0);
705 	if (c != '}')
706 	    seterror(ERR_BADBANG);
707     }
708     exclnxt = hp;
709 }
710 
711 static struct wordent *
712 getsub(struct wordent *en)
713 {
714     Char orhsb[sizeof(rhsb) / sizeof(Char)];
715     Char *cp;
716     int c, delim, sc;
717     bool global;
718 
719     do {
720 	exclnxt = 0;
721 	global = 0;
722 	sc = c = getC(0);
723 	if (c == 'g' || c == 'a') {
724 	    global |= (c == 'g') ? 1 : 2;
725 	    sc = c = getC(0);
726 	}
727 	if (((c =='g') && !(global & 1)) || ((c == 'a') && !(global & 2))) {
728 	    global |= (c == 'g') ? 1 : 2;
729 	    sc = c = getC(0);
730 	}
731 
732 	switch (c) {
733 	case 'p':
734 	    justpr++;
735 	    return (en);
736 	case 'x':
737 	case 'q':
738 	    global |= 1;
739 	    /* FALLTHROUGH */
740 	case 'h':
741 	case 'r':
742 	case 't':
743 	case 'e':
744 	    break;
745 	case '&':
746 	    if (slhs[0] == 0) {
747 		seterror(ERR_NOSUBST);
748 		return (en);
749 	    }
750 	    (void) Strcpy(lhsb, slhs);
751 	    break;
752 #ifdef notdef
753 	case '~':
754 	    if (lhsb[0] == 0)
755 		goto badlhs;
756 	    break;
757 #endif
758 	case 's':
759 	    delim = getC(0);
760 	    if (letter(delim) || Isdigit(delim) || any(" \t\n", delim)) {
761 		unreadc(delim);
762 		lhsb[0] = 0;
763 		seterror(ERR_BADSUBST);
764 		return (en);
765 	    }
766 	    cp = lhsb;
767 	    for (;;) {
768 		c = getC(0);
769 		if (c == '\n') {
770 		    unreadc(c);
771 		    break;
772 		}
773 		if (c == delim)
774 		    break;
775 		if (cp > &lhsb[sizeof(lhsb) / sizeof(Char) - 2]) {
776 		    lhsb[0] = 0;
777 		    seterror(ERR_BADSUBST);
778 		    return (en);
779 		}
780 		if (c == '\\') {
781 		    c = getC(0);
782 		    if (c != delim && c != '\\')
783 			*cp++ = '\\';
784 		}
785 		*cp++ = c;
786 	    }
787 	    if (cp != lhsb)
788 		*cp++ = 0;
789 	    else if (lhsb[0] == 0) {
790 		seterror(ERR_LHS);
791 		return (en);
792 	    }
793 	    cp = rhsb;
794 	    (void)Strcpy(orhsb, cp);
795 	    for (;;) {
796 		c = getC(0);
797 		if (c == '\n') {
798 		    unreadc(c);
799 		    break;
800 		}
801 		if (c == delim)
802 		    break;
803 #ifdef notdef
804 		if (c == '~') {
805 		    if (&cp[Strlen(orhsb)] > &rhsb[sizeof(rhsb) /
806 						   sizeof(Char) - 2])
807 			goto toorhs;
808 		    (void)Strcpy(cp, orhsb);
809 		    cp = Strend(cp);
810 		    continue;
811 		}
812 #endif
813 		if (cp > &rhsb[sizeof(rhsb) / sizeof(Char) - 2]) {
814 		    seterror(ERR_RHSLONG);
815 		    return (en);
816 		}
817 		if (c == '\\') {
818 		    c = getC(0);
819 		    if (c != delim /* && c != '~' */ )
820 			*cp++ = '\\';
821 		}
822 		*cp++ = c;
823 	    }
824 	    *cp++ = 0;
825 	    break;
826 	default:
827 	    if (c == '\n')
828 		unreadc(c);
829 	    seterror(ERR_BADBANGMOD, c);
830 	    return (en);
831 	}
832 	(void)Strcpy(slhs, lhsb);
833 	if (exclc)
834 	    en = dosub(sc, en, global);
835     }
836     while ((c = getC(0)) == ':');
837     unreadc(c);
838     return (en);
839 }
840 
841 static struct wordent *
842 dosub(int sc, struct wordent *en, bool global)
843 {
844     struct wordent lexi, *hp, *wdp;
845     int i;
846     bool didone, didsub;
847 
848     didone = 0;
849     didsub = 0;
850     i = exclc;
851     hp = &lexi;
852 
853     wdp = hp;
854     while (--i >= 0) {
855 	struct wordent *new = (struct wordent *)xcalloc(1, sizeof *wdp);
856 
857 	new->word = 0;
858 	new->prev = wdp;
859 	new->next = hp;
860 	wdp->next = new;
861 	wdp = new;
862 	en = en->next;
863 	if (en->word) {
864 	    Char *tword, *otword;
865 
866 	    if ((global & 1) || didsub == 0) {
867 		tword = subword(en->word, sc, &didone);
868 		if (didone)
869 		    didsub = 1;
870 		if (global & 2) {
871 		    while (didone && tword != STRNULL) {
872 			otword = tword;
873 			tword = subword(otword, sc, &didone);
874 			if (Strcmp(tword, otword) == 0) {
875 			    xfree((ptr_t) otword);
876 			    break;
877 			}
878 			else
879 			    xfree((ptr_t)otword);
880 		    }
881 		}
882 	    }
883 	    else
884 		tword = Strsave(en->word);
885 	    wdp->word = tword;
886 	}
887     }
888     if (didsub == 0)
889 	seterror(ERR_MODFAIL);
890     hp->prev = wdp;
891     return (&enthist(-1000, &lexi, 0)->Hlex);
892 }
893 
894 static Char *
895 subword(Char *cp, int type, bool *adid)
896 {
897     Char wbuf[BUFSIZE];
898     Char *mp, *np, *wp;
899     int i;
900 
901     *adid = 0;
902     switch (type) {
903     case 'r':
904     case 'e':
905     case 'h':
906     case 't':
907     case 'q':
908     case 'x':
909 	wp = domod(cp, type);
910 	if (wp == 0)
911 	    return (Strsave(cp));
912 	*adid = 1;
913 	return (wp);
914     default:
915 	wp = wbuf;
916 	i = BUFSIZE - 4;
917 	for (mp = cp; *mp; mp++)
918 	    if (matchs(mp, lhsb)) {
919 		for (np = cp; np < mp;)
920 		    *wp++ = *np++, --i;
921 		for (np = rhsb; *np; np++)
922 		    switch (*np) {
923 		    case '\\':
924 			if (np[1] == '&')
925 			    np++;
926 			/* FALLTHROUGH */
927 		    default:
928 			if (--i < 0) {
929 			    seterror(ERR_SUBOVFL);
930 			    return (STRNULL);
931 			}
932 			*wp++ = *np;
933 			continue;
934 		    case '&':
935 			i -= Strlen(lhsb);
936 			if (i < 0) {
937 			    seterror(ERR_SUBOVFL);
938 			    return (STRNULL);
939 			}
940 			*wp = 0;
941 			(void) Strcat(wp, lhsb);
942 			wp = Strend(wp);
943 			continue;
944 		    }
945 		mp += Strlen(lhsb);
946 		i -= Strlen(mp);
947 		if (i < 0) {
948 		    seterror(ERR_SUBOVFL);
949 		    return (STRNULL);
950 		}
951 		*wp = 0;
952 		(void) Strcat(wp, mp);
953 		*adid = 1;
954 		return (Strsave(wbuf));
955 	    }
956 	return (Strsave(cp));
957     }
958 }
959 
960 Char *
961 domod(Char *cp, int type)
962 {
963     Char *wp, *xp;
964     int c;
965 
966     switch (type) {
967     case 'x':
968     case 'q':
969 	wp = Strsave(cp);
970 	for (xp = wp; (c = *xp) != '\0'; xp++)
971 	    if ((c != ' ' && c != '\t') || type == 'q')
972 		*xp |= QUOTE;
973 	return (wp);
974     case 'h':
975     case 't':
976 	if (!any(short2str(cp), '/'))
977 	    return (type == 't' ? Strsave(cp) : 0);
978 	wp = Strend(cp);
979 	while (*--wp != '/')
980 	    continue;
981 	if (type == 'h')
982 	    xp = Strsave(cp), xp[wp - cp] = 0;
983 	else
984 	    xp = Strsave(wp + 1);
985 	return (xp);
986     case 'e':
987     case 'r':
988 	wp = Strend(cp);
989 	for (wp--; wp >= cp && *wp != '/'; wp--)
990 	    if (*wp == '.') {
991 		if (type == 'e')
992 		    xp = Strsave(wp + 1);
993 		else
994 		    xp = Strsave(cp), xp[wp - cp] = 0;
995 		return (xp);
996 	    }
997 	return (Strsave(type == 'e' ? STRNULL : cp));
998     default:
999 	break;
1000     }
1001     return (0);
1002 }
1003 
1004 static int
1005 matchs(Char *str, Char *pat)
1006 {
1007     while (*str && *pat && *str == *pat)
1008 	str++, pat++;
1009     return (*pat == 0);
1010 }
1011 
1012 static int
1013 getsel(int *al, int *ar, int dol)
1014 {
1015     int c, i;
1016     bool first;
1017 
1018     c = getC(0);
1019     first = *al < 0;
1020 
1021     switch (c) {
1022     case '%':
1023 	if (quesarg == -1) {
1024 	    seterror(ERR_BADBANGARG);
1025 	    return (0);
1026 	}
1027 	if (*al < 0)
1028 	    *al = quesarg;
1029 	*ar = quesarg;
1030 	break;
1031     case '-':
1032 	if (*al < 0) {
1033 	    *al = 0;
1034 	    *ar = dol - 1;
1035 	    unreadc(c);
1036 	}
1037 	return (1);
1038     case '^':
1039 	if (*al < 0)
1040 	    *al = 1;
1041 	*ar = 1;
1042 	break;
1043     case '$':
1044 	if (*al < 0)
1045 	    *al = dol;
1046 	*ar = dol;
1047 	break;
1048     case '*':
1049 	if (*al < 0)
1050 	    *al = 1;
1051 	*ar = dol;
1052 	if (*ar < *al) {
1053 	    *ar = 0;
1054 	    *al = 1;
1055 	    return (1);
1056 	}
1057 	break;
1058     default:
1059 	if (Isdigit(c)) {
1060 	    i = 0;
1061 	    while (Isdigit(c)) {
1062 		i = i * 10 + c - '0';
1063 		c = getC(0);
1064 	    }
1065 	    if (i < 0)
1066 		i = dol + 1;
1067 	    if (*al < 0)
1068 		*al = i;
1069 	    *ar = i;
1070 	}
1071 	else if (*al < 0)
1072 	    *al = 0, *ar = dol;
1073 	else
1074 	    *ar = dol - 1;
1075 	unreadc(c);
1076 	break;
1077     }
1078     if (first) {
1079 	c = getC(0);
1080 	unreadc(c);
1081 	if (any("-$*", c))
1082 	    return (1);
1083     }
1084     if (*al > *ar || *ar > dol) {
1085 	seterror(ERR_BADBANGARG);
1086 	return (0);
1087     }
1088     return (1);
1089 
1090 }
1091 
1092 static struct wordent *
1093 gethent(int sc)
1094 {
1095     struct Hist *hp;
1096     Char *np;
1097     int c, event;
1098     bool back;
1099 
1100     back = 0;
1101     c = sc == HISTSUB ? HIST : getC(0);
1102     if (c == HIST) {
1103 	if (alhistp)
1104 	    return (alhistp);
1105 	event = eventno;
1106     }
1107     else
1108 	switch (c) {
1109 	case ':':
1110 	case '^':
1111 	case '$':
1112 	case '*':
1113 	case '%':
1114 	    ungetC(c);
1115 	    if (lastev == eventno && alhistp)
1116 		return (alhistp);
1117 	    event = lastev;
1118 	    break;
1119 	case '#':		/* !# is command being typed in (mrh) */
1120 	    if (--hleft == 0) {
1121 		seterror(ERR_HISTLOOP);
1122 		return (0);
1123 	    }
1124 	    else
1125 		return (&paraml);
1126 	    /* NOTREACHED */
1127 	case '-':
1128 	    back = 1;
1129 	    c = getC(0);
1130 	    /* FALLTHROUGH */
1131 	default:
1132 	    if (any("(=~", c)) {
1133 		unreadc(c);
1134 		ungetC(HIST);
1135 		return (0);
1136 	    }
1137 	    np = lhsb;
1138 	    event = 0;
1139 	    while (!cmap(c, _ESC | _META | _QF | _QB) && !any("${}:", c)) {
1140 		if (event != -1 && Isdigit(c))
1141 		    event = event * 10 + c - '0';
1142 		else
1143 		    event = -1;
1144 		if (np < &lhsb[sizeof(lhsb) / sizeof(Char) - 2])
1145 		    *np++ = c;
1146 		c = getC(0);
1147 	    }
1148 	    unreadc(c);
1149 	    if (np == lhsb) {
1150 		ungetC(HIST);
1151 		return (0);
1152 	    }
1153 	    *np++ = 0;
1154 	    if (event != -1) {
1155 		/*
1156 		 * History had only digits
1157 		 */
1158 		if (back)
1159 		    event = eventno + (alhistp == 0) - (event ? event : 0);
1160 		break;
1161 	    }
1162 	    hp = findev(lhsb, 0);
1163 	    if (hp)
1164 		lastev = hp->Hnum;
1165 	    return (&hp->Hlex);
1166 	case '?':
1167 	    np = lhsb;
1168 	    for (;;) {
1169 		c = getC(0);
1170 		if (c == '\n') {
1171 		    unreadc(c);
1172 		    break;
1173 		}
1174 		if (c == '?')
1175 		    break;
1176 		if (np < &lhsb[sizeof(lhsb) / sizeof(Char) - 2])
1177 		    *np++ = c;
1178 	    }
1179 	    if (np == lhsb) {
1180 		if (lhsb[0] == 0) {
1181 		    seterror(ERR_NOSEARCH);
1182 		    return (0);
1183 		}
1184 	    }
1185 	    else
1186 		*np++ = 0;
1187 	    hp = findev(lhsb, 1);
1188 	    if (hp)
1189 		lastev = hp->Hnum;
1190 	    return (&hp->Hlex);
1191 	}
1192 
1193     for (hp = Histlist.Hnext; hp; hp = hp->Hnext)
1194 	if (hp->Hnum == event) {
1195 	    hp->Href = eventno;
1196 	    lastev = hp->Hnum;
1197 	    return (&hp->Hlex);
1198 	}
1199     np = putn(event);
1200     seterror(ERR_NOEVENT, vis_str(np));
1201     return (0);
1202 }
1203 
1204 static struct Hist *
1205 findev(Char *cp, bool anyarg)
1206 {
1207     struct Hist *hp;
1208 
1209     for (hp = Histlist.Hnext; hp; hp = hp->Hnext) {
1210 	Char *dp, *p, *q;
1211 	struct wordent *lp;
1212 	int argno;
1213 
1214 	lp = hp->Hlex.next;
1215 	argno = 0;
1216 
1217 	/*
1218 	 * The entries added by alias substitution don't have a newline but do
1219 	 * have a negative event number. Savehist() trims off these entries,
1220 	 * but it happens before alias expansion, too early to delete those
1221 	 * from the previous command.
1222 	 */
1223 	if (hp->Hnum < 0)
1224 	    continue;
1225 	if (lp->word[0] == '\n')
1226 	    continue;
1227 	if (!anyarg) {
1228 	    p = cp;
1229 	    q = lp->word;
1230 	    do
1231 		if (!*p)
1232 		    return (hp);
1233 	    while (*p++ == *q++);
1234 	    continue;
1235 	}
1236 	do {
1237 	    for (dp = lp->word; *dp; dp++) {
1238 		p = cp;
1239 		q = dp;
1240 		do
1241 		    if (!*p) {
1242 			quesarg = argno;
1243 			return (hp);
1244 		    }
1245 		while (*p++ == *q++);
1246 	    }
1247 	    lp = lp->next;
1248 	    argno++;
1249 	} while (lp->word[0] != '\n');
1250     }
1251     seterror(ERR_NOEVENT, vis_str(cp));
1252     return (0);
1253 }
1254 
1255 
1256 static void
1257 setexclp(Char *cp)
1258 {
1259     if (cp && cp[0] == '\n')
1260 	return;
1261     exclp = cp;
1262 }
1263 
1264 void
1265 unreadc(int c)
1266 {
1267     peekread = c;
1268 }
1269 
1270 int
1271 readc(bool wanteof)
1272 {
1273     static int sincereal;
1274     int c;
1275 
1276     aret = F_SEEK;
1277     if ((c = peekread) != '\0') {
1278 	peekread = 0;
1279 	return (c);
1280     }
1281 top:
1282     aret = F_SEEK;
1283     if (alvecp) {
1284 	aret = A_SEEK;
1285 	if ((c = *alvecp++) != '\0')
1286 	    return (c);
1287 	if (alvec && *alvec) {
1288 		alvecp = *alvec++;
1289 		return (' ');
1290 	}
1291 	else {
1292 	    aret = F_SEEK;
1293 	    alvecp = NULL;
1294 	    return('\n');
1295 	}
1296     }
1297     if (alvec) {
1298 	if ((alvecp = *alvec) != '\0') {
1299 	    alvec++;
1300 	    goto top;
1301 	}
1302 	/* Infinite source! */
1303 	return ('\n');
1304     }
1305     if (evalp) {
1306 	aret = E_SEEK;
1307 	if ((c = *evalp++) != '\0')
1308 	    return (c);
1309 	if (evalvec && *evalvec) {
1310 	    evalp = *evalvec++;
1311 	    return (' ');
1312 	}
1313 	aret = F_SEEK;
1314 	evalp = 0;
1315     }
1316     if (evalvec) {
1317 	if (evalvec == (Char **) 1) {
1318 	    doneinp = 1;
1319 	    reset();
1320 	}
1321 	if ((evalp = *evalvec) != '\0') {
1322 	    evalvec++;
1323 	    goto top;
1324 	}
1325 	evalvec = (Char **) 1;
1326 	return ('\n');
1327     }
1328     do {
1329 	if (arginp == (Char *) 1 || onelflg == 1) {
1330 	    if (wanteof)
1331 		return (-1);
1332 	    exitstat();
1333 	}
1334 	if (arginp) {
1335 	    if ((c = *arginp++) == 0) {
1336 		arginp = (Char *) 1;
1337 		return ('\n');
1338 	    }
1339 	    return (c);
1340 	}
1341 reread:
1342 	c = bgetc();
1343 	if (c < 0) {
1344 	    struct termios tty;
1345 	    if (wanteof)
1346 		return (-1);
1347 	    /* was isatty but raw with ignoreeof yields problems */
1348 	    if (tcgetattr(SHIN, &tty) == 0 && (tty.c_lflag & ICANON))
1349 	    {
1350 		/* was 'short' for FILEC */
1351 		pid_t     ctpgrp;
1352 
1353 		if (++sincereal > 25)
1354 		    goto oops;
1355 		if (tpgrp != -1 &&
1356 		    (ctpgrp = tcgetpgrp(FSHTTY)) != -1 &&
1357 		    tpgrp != ctpgrp) {
1358 		    (void)tcsetpgrp(FSHTTY, tpgrp);
1359 		    (void)kill(-ctpgrp, SIGHUP);
1360 		    (void)fprintf(csherr, "Reset tty pgrp from %ld to %ld\n",
1361 				   (long)ctpgrp, (long)tpgrp);
1362 		    goto reread;
1363 		}
1364 		if (adrof(STRignoreeof)) {
1365 		    if (loginsh)
1366 			(void)fprintf(csherr,"\nUse \"logout\" to logout.\n");
1367 		    else
1368 			(void)fprintf(csherr,"\nUse \"exit\" to leave csh.\n");
1369 		    reset();
1370 		}
1371 		if (chkstop == 0)
1372 		    panystop(1);
1373 	    }
1374     oops:
1375 	    doneinp = 1;
1376 	    reset();
1377 	}
1378 	sincereal = 0;
1379 	if (c == '\n' && onelflg)
1380 	    onelflg--;
1381     } while (c == 0);
1382     return (c);
1383 }
1384 
1385 static int
1386 bgetc(void)
1387 {
1388 #ifdef FILEC
1389     char tbuf[BUFSIZE + 1];
1390     Char ttyline[BUFSIZE];
1391     int c, buf, numleft, off, roomleft;
1392 
1393     numleft = 0;
1394 #else /* FILEC */
1395     char tbuf[BUFSIZE + 1];
1396     int c, buf, off;
1397 #endif /* !FILEC */
1398 
1399     if (cantell) {
1400 	if (fseekp < fbobp || fseekp > feobp) {
1401 	    fbobp = feobp = fseekp;
1402 	    (void)lseek(SHIN, fseekp, SEEK_SET);
1403 	}
1404 	if (fseekp == feobp) {
1405 	    int i;
1406 
1407 	    fbobp = feobp;
1408 	    do
1409 		c = read(SHIN, tbuf, BUFSIZE);
1410 	    while (c < 0 && errno == EINTR);
1411 	    if (c <= 0)
1412 		return (-1);
1413 	    for (i = 0; i < c; i++)
1414 		fbuf[0][i] = (unsigned char) tbuf[i];
1415 	    feobp += c;
1416 	}
1417 	c = fbuf[0][fseekp - fbobp];
1418 	fseekp++;
1419 	return (c);
1420     }
1421 
1422 again:
1423     buf = (int) fseekp / BUFSIZE;
1424     if (buf >= fblocks) {
1425 	Char **nfbuf;
1426 
1427 	nfbuf = (Char **)xcalloc((size_t) (fblocks + 2), sizeof(char **));
1428 	if (fbuf) {
1429 	    (void)blkcpy(nfbuf, fbuf);
1430 	    xfree((ptr_t) fbuf);
1431 	}
1432 	fbuf = nfbuf;
1433 	fbuf[fblocks] = (Char *)xcalloc(BUFSIZE, sizeof(Char));
1434 	fblocks++;
1435 	if (!intty)
1436 	    goto again;
1437     }
1438     if (fseekp >= feobp) {
1439 	buf = (int) feobp / BUFSIZE;
1440 	off = (int) feobp % BUFSIZE;
1441 	roomleft = BUFSIZE - off;
1442 
1443 #ifdef FILEC
1444 	roomleft = BUFSIZE - off;
1445 	for (;;) {
1446 	    if (filec && intty) {
1447 		c = numleft ? numleft : tenex(ttyline, BUFSIZE);
1448 		if (c > roomleft) {
1449 		    /* start with fresh buffer */
1450 		    feobp = fseekp = fblocks * BUFSIZE;
1451 		    numleft = c;
1452 		    goto again;
1453 		}
1454 		if (c > 0)
1455 		    (void)memcpy(fbuf[buf] + off, ttyline, c * sizeof(Char));
1456 		numleft = 0;
1457 	    }
1458 	    else {
1459 #endif
1460 		c = read(SHIN, tbuf, roomleft);
1461 		if (c > 0) {
1462 		    int     i;
1463 		    Char   *ptr = fbuf[buf] + off;
1464 
1465 		    for (i = 0; i < c; i++)
1466 			ptr[i] = (unsigned char) tbuf[i];
1467 		}
1468 #ifdef FILEC
1469 	    }
1470 #endif
1471 	    if (c >= 0)
1472 		break;
1473 	    if (errno == EWOULDBLOCK) {
1474 		int     iooff = 0;
1475 
1476 		(void)ioctl(SHIN, FIONBIO, (ioctl_t) & iooff);
1477 	    }
1478 	    else if (errno != EINTR)
1479 		break;
1480 	}
1481 	if (c <= 0)
1482 	    return (-1);
1483 	feobp += c;
1484 #ifndef FILEC
1485 	goto again;
1486 #else
1487 	if (filec && !intty)
1488 	    goto again;
1489 #endif
1490     }
1491     c = fbuf[buf][(int)fseekp % BUFSIZE];
1492     fseekp++;
1493     return (c);
1494 }
1495 
1496 static void
1497 bfree(void)
1498 {
1499     int i, sb;
1500 
1501     if (cantell)
1502 	return;
1503     if (whyles)
1504 	return;
1505     sb = (int)(fseekp - 1) / BUFSIZE;
1506     if (sb > 0) {
1507 	for (i = 0; i < sb; i++)
1508 	    xfree((ptr_t) fbuf[i]);
1509 	(void)blkcpy(fbuf, &fbuf[sb]);
1510 	fseekp -= BUFSIZE * sb;
1511 	feobp -= BUFSIZE * sb;
1512 	fblocks -= sb;
1513     }
1514 }
1515 
1516 void
1517 bseek(struct Ain *l)
1518 {
1519     switch (aret = l->type) {
1520     case A_SEEK:
1521 	alvec = l->a_seek;
1522 	alvecp = l->c_seek;
1523 	return;
1524     case E_SEEK:
1525 	evalvec = l->a_seek;
1526 	evalp = l->c_seek;
1527 	return;
1528     case F_SEEK:
1529 	fseekp = l->f_seek;
1530 	return;
1531     default:
1532 	(void)fprintf(csherr, "Bad seek type %d\n", aret);
1533 	abort();
1534     }
1535 }
1536 
1537 void
1538 btell(struct Ain *l)
1539 {
1540     switch (l->type = aret) {
1541     case A_SEEK:
1542 	l->a_seek = alvec;
1543 	l->c_seek = alvecp;
1544 	return;
1545     case E_SEEK:
1546 	l->a_seek = evalvec;
1547 	l->c_seek = evalp;
1548 	return;
1549     case F_SEEK:
1550 	l->f_seek = fseekp;
1551 	l->a_seek = NULL;
1552 	return;
1553     default:
1554 	(void)fprintf(csherr, "Bad seek type %d\n", aret);
1555 	abort();
1556     }
1557 }
1558 
1559 void
1560 btoeof(void)
1561 {
1562     (void)lseek(SHIN, (off_t) 0, SEEK_END);
1563     aret = F_SEEK;
1564     fseekp = feobp;
1565     alvec = NULL;
1566     alvecp = NULL;
1567     evalvec = NULL;
1568     evalp = NULL;
1569     wfree();
1570     bfree();
1571 }
1572 
1573 void
1574 settell(void)
1575 {
1576     cantell = 0;
1577     if (arginp || onelflg || intty)
1578 	return;
1579     if (lseek(SHIN, (off_t) 0, SEEK_CUR) < 0 || errno == ESPIPE)
1580 	return;
1581     fbuf = (Char **)xcalloc(2, sizeof(Char **));
1582     fblocks = 1;
1583     fbuf[0] = (Char *)xcalloc(BUFSIZE, sizeof(Char));
1584     fseekp = fbobp = feobp = lseek(SHIN, (off_t) 0, SEEK_CUR);
1585     cantell = 1;
1586 }
1587