xref: /netbsd/bin/csh/lex.c (revision 6550d01e)
1 /* $NetBSD: lex.c,v 1.27 2010/01/17 12:15:36 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. Neither the name of the University nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 #include <sys/cdefs.h>
33 #ifndef lint
34 #if 0
35 static char sccsid[] = "@(#)lex.c	8.1 (Berkeley) 5/31/93";
36 #else
37 __RCSID("$NetBSD: lex.c,v 1.27 2010/01/17 12:15:36 wiz Exp $");
38 #endif
39 #endif /* not lint */
40 
41 #include <sys/ioctl.h>
42 #include <sys/types.h>
43 
44 #include <errno.h>
45 #include <stdarg.h>
46 #include <stdlib.h>
47 #include <string.h>
48 #include <termios.h>
49 #include <unistd.h>
50 
51 #include "csh.h"
52 #include "extern.h"
53 
54 /*
55  * These lexical routines read input and form lists of words.
56  * There is some involved processing here, because of the complications
57  * of input buffering, and especially because of history substitution.
58  */
59 
60 static Char *word(void);
61 static int getC1(int);
62 static void getdol(void);
63 static void getexcl(int);
64 static struct Hist *findev(Char *, int);
65 static void setexclp(Char *);
66 static int bgetc(void);
67 static void bfree(void);
68 static struct wordent *gethent(int);
69 static int matchs(Char *, Char *);
70 static int getsel(int *, int *, int);
71 static struct wordent *getsub(struct wordent *);
72 static Char *subword(Char *, int, int *);
73 static struct wordent *dosub(int, struct wordent *, int);
74 
75 /*
76  * Peekc is a peek character for getC, peekread for readc.
77  * There is a subtlety here in many places... history routines
78  * will read ahead and then insert stuff into the input stream.
79  * If they push back a character then they must push it behind
80  * the text substituted by the history substitution.  On the other
81  * hand in several places we need 2 peek characters.  To make this
82  * all work, the history routines read with getC, and make use both
83  * of ungetC and unreadc.  The key observation is that the state
84  * of getC at the call of a history reference is such that calls
85  * to getC from the history routines will always yield calls of
86  * readc, unless this peeking is involved.  That is to say that during
87  * getexcl the variables lap, exclp, and exclnxt are all zero.
88  *
89  * Getdol invokes history substitution, hence the extra peek, peekd,
90  * which it can ungetD to be before history substitutions.
91  */
92 static Char peekc = 0, peekd = 0;
93 static Char peekread = 0;
94 
95 /* (Tail of) current word from ! subst */
96 static Char *exclp = NULL;
97 
98 /* The rest of the ! subst words */
99 static struct wordent *exclnxt = NULL;
100 
101 /* Count of remaining words in ! subst */
102 static int exclc = 0;
103 
104 /* "Globp" for alias resubstitution */
105 Char **alvec, *alvecp;
106 int aret = F_SEEK;
107 
108 /*
109  * Labuf implements a general buffer for lookahead during lexical operations.
110  * Text which is to be placed in the input stream can be stuck here.
111  * We stick parsed ahead $ constructs during initial input,
112  * process id's from `$$', and modified variable values (from qualifiers
113  * during expansion in sh.dol.c) here.
114  */
115 static Char labuf[BUFSIZE];
116 
117 /*
118  * Lex returns to its caller not only a wordlist (as a "var" parameter)
119  * but also whether a history substitution occurred.  This is used in
120  * the main (process) routine to determine whether to echo, and also
121  * when called by the alias routine to determine whether to keep the
122  * argument list.
123  */
124 static int hadhist = 0;
125 
126 /*
127  * Avoid alias expansion recursion via \!#
128  */
129 int     hleft;
130 
131 static Char getCtmp;
132 
133 #define getC(f) ((getCtmp = peekc) ? (peekc = 0, getCtmp) : getC1(f))
134 #define	ungetC(c) peekc = c
135 #define	ungetD(c) peekd = c
136 
137 int
138 lex(struct wordent *hp)
139 {
140     struct wordent *wdp;
141     int c;
142 
143     btell(&lineloc);
144     hp->next = hp->prev = hp;
145     hp->word = STRNULL;
146     hadhist = 0;
147     do
148 	c = readc(0);
149     while (c == ' ' || c == '\t');
150     if (c == HISTSUB && intty)
151 	/* ^lef^rit	from tty is short !:s^lef^rit */
152 	getexcl(c);
153     else
154 	unreadc(c);
155     wdp = hp;
156     /*
157      * The following loop is written so that the links needed by freelex will
158      * be ready and rarin to go even if it is interrupted.
159      */
160     do {
161 	struct wordent *new;
162 
163 	new = (struct wordent *)xmalloc((size_t)sizeof(*wdp));
164 	new->word = 0;
165 	new->prev = wdp;
166 	new->next = hp;
167 	wdp->next = new;
168 	wdp = new;
169 	wdp->word = word();
170     } while (wdp->word[0] != '\n');
171     hp->prev = wdp;
172     return (hadhist);
173 }
174 
175 void
176 prlex(FILE *fp, struct wordent *sp0)
177 {
178     struct wordent *sp;
179 
180     sp = sp0->next;
181     for (;;) {
182 	(void)fprintf(fp, "%s", vis_str(sp->word));
183 	sp = sp->next;
184 	if (sp == sp0)
185 	    break;
186 	if (sp->word[0] != '\n')
187 	    (void) fputc(' ', fp);
188     }
189 }
190 
191 void
192 copylex(struct wordent *hp, struct wordent *fp)
193 {
194     struct wordent *wdp;
195 
196     wdp = hp;
197     fp = fp->next;
198     do {
199 	struct wordent *new;
200 
201 	new = (struct wordent *)xmalloc((size_t)sizeof(*wdp));
202 	new->prev = wdp;
203 	new->next = hp;
204 	wdp->next = new;
205 	wdp = new;
206 	wdp->word = Strsave(fp->word);
207 	fp = fp->next;
208     } while (wdp->word[0] != '\n');
209     hp->prev = wdp;
210 }
211 
212 void
213 freelex(struct wordent *vp)
214 {
215     struct wordent *fp;
216 
217     while (vp->next != vp) {
218 	fp = vp->next;
219 	vp->next = fp->next;
220 	xfree((ptr_t) fp->word);
221 	xfree((ptr_t) fp);
222     }
223     vp->prev = vp;
224 }
225 
226 static Char *
227 word(void)
228 {
229     Char wbuf[BUFSIZE], *wp;
230     int i;
231     Char c, c1;
232     int dolflg;
233 
234     wp = wbuf;
235     i = BUFSIZE - 4;
236 loop:
237     while ((c = getC(DOALL)) == ' ' || c == '\t')
238 	continue;
239     if (cmap(c, _META | _ESC))
240 	switch (c) {
241 	case '&':
242 	case '|':
243 	case '<':
244 	case '>':
245 	    *wp++ = c;
246 	    c1 = getC(DOALL);
247 	    if (c1 == c)
248 		*wp++ = c1;
249 	    else
250 		ungetC(c1);
251 	    goto ret;
252 
253 	case '#':
254 	    if (intty)
255 		break;
256 	    c = 0;
257 	    do {
258 		c1 = c;
259 		c = getC(0);
260 	    } while (c != '\n');
261 	    if (c1 == '\\')
262 		goto loop;
263 	    /* FALLTHROUGH */
264 
265 	case ';':
266 	case '(':
267 	case ')':
268 	case '\n':
269 	    *wp++ = c;
270 	    goto ret;
271 
272 	case '\\':
273 	    c = getC(0);
274 	    if (c == '\n') {
275 		if (onelflg == 1)
276 		    onelflg = 2;
277 		goto loop;
278 	    }
279 	    if (c != HIST)
280 		*wp++ = '\\', --i;
281 	    c |= QUOTE;
282 	    break;
283 	}
284     c1 = 0;
285     dolflg = DOALL;
286     for (;;) {
287 	if (c1) {
288 	    if (c == c1) {
289 		c1 = 0;
290 		dolflg = DOALL;
291 	    }
292 	    else if (c == '\\') {
293 		c = getC(0);
294 		if (c == HIST)
295 		    c |= QUOTE;
296 		else {
297 		    if (c == '\n')
298 			/*
299 			 * if (c1 == '`') c = ' '; else
300 			 */
301 			c |= QUOTE;
302 		    ungetC(c);
303 		    c = '\\';
304 		}
305 	    }
306 	    else if (c == '\n') {
307 		seterror(ERR_UNMATCHED, c1);
308 		ungetC(c);
309 		break;
310 	    }
311 	}
312 	else if (cmap(c, _META | _QF | _QB | _ESC)) {
313 	    if (c == '\\') {
314 		c = getC(0);
315 		if (c == '\n') {
316 		    if (onelflg == 1)
317 			onelflg = 2;
318 		    break;
319 		}
320 		if (c != HIST)
321 		    *wp++ = '\\', --i;
322 		c |= QUOTE;
323 	    }
324 	    else if (cmap(c, _QF | _QB)) {	/* '"` */
325 		c1 = c;
326 		dolflg = c == '"' ? DOALL : DOEXCL;
327 	    }
328 	    else if (c != '#' || !intty) {
329 		ungetC(c);
330 		break;
331 	    }
332 	}
333 	if (--i > 0) {
334 	    *wp++ = c;
335 	    c = getC(dolflg);
336 	}
337 	else {
338 	    seterror(ERR_WTOOLONG);
339 	    wp = &wbuf[1];
340 	    break;
341 	}
342     }
343 ret:
344     *wp = 0;
345     return (Strsave(wbuf));
346 }
347 
348 static int
349 getC1(int flag)
350 {
351     Char c;
352 
353     for (;;) {
354 	if ((c = peekc) != '\0') {
355 	    peekc = 0;
356 	    return (c);
357 	}
358 	if (lap) {
359 	    if ((c = *lap++) == 0)
360 		lap = 0;
361 	    else {
362 		if (cmap(c, _META | _QF | _QB))
363 		    c |= QUOTE;
364 		return (c);
365 	    }
366 	}
367 	if ((c = peekd) != '\0') {
368 	    peekd = 0;
369 	    return (c);
370 	}
371 	if (exclp) {
372 	    if ((c = *exclp++) != '\0')
373 		return (c);
374 	    if (exclnxt && --exclc >= 0) {
375 		exclnxt = exclnxt->next;
376 		setexclp(exclnxt->word);
377 		return (' ');
378 	    }
379 	    exclp = 0;
380 	    exclnxt = 0;
381 	}
382 	if (exclnxt) {
383 	    exclnxt = exclnxt->next;
384 	    if (--exclc < 0)
385 		exclnxt = 0;
386 	    else
387 		setexclp(exclnxt->word);
388 	    continue;
389 	}
390 	c = readc(0);
391 	if (c == '$' && (flag & DODOL)) {
392 	    getdol();
393 	    continue;
394 	}
395 	if (c == HIST && (flag & DOEXCL)) {
396 	    getexcl(0);
397 	    continue;
398 	}
399 	break;
400     }
401     return (c);
402 }
403 
404 static void
405 getdol(void)
406 {
407     Char name[4*MAXVARLEN+1], *ep, *np;
408     int c, sc;
409     int special, toolong;
410 
411     special = 0;
412     np = name, *np++ = '$';
413     c = sc = getC(DOEXCL);
414     if (any("\t \n", c)) {
415 	ungetD(c);
416 	ungetC('$' | QUOTE);
417 	return;
418     }
419     if (c == '{')
420 	*np++ = c, c = getC(DOEXCL);
421     if (c == '#' || c == '?')
422 	special++, *np++ = c, c = getC(DOEXCL);
423     *np++ = c;
424     switch (c) {
425     case '<':
426     case '$':
427     case '!':
428 	if (special)
429 	    seterror(ERR_SPDOLLT);
430 	*np = 0;
431 	addla(name);
432 	return;
433     case '\n':
434 	ungetD(c);
435 	np--;
436 	seterror(ERR_NEWLINE);
437 	*np = 0;
438 	addla(name);
439 	return;
440     case '*':
441 	if (special)
442 	    seterror(ERR_SPSTAR);
443 	*np = 0;
444 	addla(name);
445 	return;
446     default:
447 	toolong = 0;
448 	if (Isdigit(c)) {
449 #ifdef notdef
450 	    /* let $?0 pass for now */
451 	    if (special) {
452 		seterror(ERR_DIGIT);
453 		*np = 0;
454 		addla(name);
455 		return;
456 	    }
457 #endif
458 	    /* we know that np < &name[4] */
459 	    ep = &np[MAXVARLEN];
460 	    while ((c = getC(DOEXCL)) != '\0'){
461 		if (!Isdigit(c))
462 		    break;
463 		if (np < ep)
464 		    *np++ = c;
465 		else
466 		    toolong = 1;
467 	    }
468 	}
469 	else if (letter(c)) {
470 	    /* we know that np < &name[4] */
471 	    ep = &np[MAXVARLEN];
472 	    toolong = 0;
473 	    while ((c = getC(DOEXCL)) != '\0') {
474 		/* Bugfix for ${v123x} from Chris Torek, DAS DEC-90. */
475 		if (!letter(c) && !Isdigit(c))
476 		    break;
477 		if (np < ep)
478 		    *np++ = c;
479 		else
480 		    toolong = 1;
481 	    }
482 	}
483 	else {
484 	    *np = 0;
485 	    seterror(ERR_VARILL);
486 	    addla(name);
487 	    return;
488 	}
489 	if (toolong) {
490 	    seterror(ERR_VARTOOLONG);
491 	    *np = 0;
492 	    addla(name);
493 	    return;
494 	}
495 	break;
496     }
497     if (c == '[') {
498 	*np++ = c;
499 	/*
500 	 * Name up to here is a max of MAXVARLEN + 8.
501 	 */
502 	ep = &np[2 * MAXVARLEN + 8];
503 	do {
504 	    /*
505 	     * Michael Greim: Allow $ expansion to take place in selector
506 	     * expressions. (limits the number of characters returned)
507 	     */
508 	    c = getC(DOEXCL | DODOL);
509 	    if (c == '\n') {
510 		ungetD(c);
511 		np--;
512 		seterror(ERR_NLINDEX);
513 		*np = 0;
514 		addla(name);
515 		return;
516 	    }
517 	    if (np < ep)
518 		*np++ = c;
519 	} while (c != ']');
520 	*np = '\0';
521 	if (np >= ep) {
522 	    seterror(ERR_SELOVFL);
523 	    addla(name);
524 	    return;
525 	}
526 	c = getC(DOEXCL);
527     }
528     /*
529      * Name up to here is a max of 2 * MAXVARLEN + 8.
530      */
531     if (c == ':') {
532 	/*
533 	 * if the :g modifier is followed by a newline, then error right away!
534 	 * -strike
535 	 */
536 	int amodflag, gmodflag;
537 
538 	amodflag = 0;
539 	gmodflag = 0;
540 	do {
541 	    *np++ = c, c = getC(DOEXCL);
542 	    if (c == 'g' || c == 'a') {
543 		if (c == 'g')
544 		    gmodflag++;
545 		else
546 		    amodflag++;
547 		*np++ = c; c = getC(DOEXCL);
548 	    }
549 	    if ((c == 'g' && !gmodflag) || (c == 'a' && !amodflag)) {
550 		if (c == 'g')
551 		    gmodflag++;
552 		else
553 		    amodflag++;
554 		*np++ = c; c = getC(DOEXCL);
555 	    }
556 	    *np++ = c;
557 	    /* scan s// [eichin:19910926.0512EST] */
558 	    if (c == 's') {
559 		int delimcnt = 2;
560 		int delim = getC(0);
561 		*np++ = delim;
562 
563 		if (!delim || letter(delim)
564 		    || Isdigit(delim) || any(" \t\n", delim)) {
565 		    seterror(ERR_BADSUBST);
566 		    break;
567 		}
568 		while ((c = getC(0)) != (-1)) {
569 		    *np++ = c;
570 		    if(c == delim) delimcnt--;
571 		    if(!delimcnt) break;
572 		}
573 		if(delimcnt) {
574 		    seterror(ERR_BADSUBST);
575 		    break;
576 		}
577 		c = 's';
578 	    }
579 	    if (!any("htrqxes", c)) {
580 		if ((amodflag || gmodflag) && c == '\n')
581 		    stderror(ERR_VARSYN);	/* strike */
582 		seterror(ERR_VARMOD, c);
583 		*np = 0;
584 		addla(name);
585 		return;
586 	    }
587 	}
588 	while ((c = getC(DOEXCL)) == ':');
589 	ungetD(c);
590     }
591     else
592 	ungetD(c);
593     if (sc == '{') {
594 	c = getC(DOEXCL);
595 	if (c != '}') {
596 	    ungetD(c);
597 	    seterror(ERR_MISSING, '}');
598 	    *np = 0;
599 	    addla(name);
600 	    return;
601 	}
602 	*np++ = c;
603     }
604     *np = 0;
605     addla(name);
606     return;
607 }
608 
609 void
610 addla(Char *cp)
611 {
612     Char buf[BUFSIZE];
613 
614     if (Strlen(cp) + (lap ? Strlen(lap) : 0) >=
615 	(sizeof(labuf) - 4) / sizeof(Char)) {
616 	seterror(ERR_EXPOVFL);
617 	return;
618     }
619     if (lap)
620 	(void)Strcpy(buf, lap);
621     (void)Strcpy(labuf, cp);
622     if (lap)
623 	(void)Strcat(labuf, buf);
624     lap = labuf;
625 }
626 
627 static Char lhsb[32];
628 static Char slhs[32];
629 static Char rhsb[64];
630 static int quesarg;
631 
632 static void
633 getexcl(int sc)
634 {
635     struct wordent *hp, *ip;
636     int c, dol, left, right;
637 
638     if (sc == 0) {
639 	sc = getC(0);
640 	if (sc != '{') {
641 	    ungetC(sc);
642 	    sc = 0;
643 	}
644     }
645     quesarg = -1;
646     lastev = eventno;
647     hp = gethent(sc);
648     if (hp == 0)
649 	return;
650     hadhist = 1;
651     dol = 0;
652     if (hp == alhistp)
653 	for (ip = hp->next->next; ip != alhistt; ip = ip->next)
654 	    dol++;
655     else
656 	for (ip = hp->next->next; ip != hp->prev; ip = ip->next)
657 	    dol++;
658     left = 0, right = dol;
659     if (sc == HISTSUB) {
660 	ungetC('s'), unreadc(HISTSUB), c = ':';
661 	goto subst;
662     }
663     c = getC(0);
664     if (!any(":^$*-%", c))
665 	goto subst;
666     left = right = -1;
667     if (c == ':') {
668 	c = getC(0);
669 	unreadc(c);
670 	if (letter(c) || c == '&') {
671 	    c = ':';
672 	    left = 0, right = dol;
673 	    goto subst;
674 	}
675     }
676     else
677 	ungetC(c);
678     if (!getsel(&left, &right, dol))
679 	return;
680     c = getC(0);
681     if (c == '*')
682 	ungetC(c), c = '-';
683     if (c == '-') {
684 	if (!getsel(&left, &right, dol))
685 	    return;
686 	c = getC(0);
687     }
688 subst:
689     exclc = right - left + 1;
690     while (--left >= 0)
691 	hp = hp->next;
692     if (sc == HISTSUB || c == ':') {
693 	do {
694 	    hp = getsub(hp);
695 	    c = getC(0);
696 	} while (c == ':');
697     }
698     unreadc(c);
699     if (sc == '{') {
700 	c = getC(0);
701 	if (c != '}')
702 	    seterror(ERR_BADBANG);
703     }
704     exclnxt = hp;
705 }
706 
707 static struct wordent *
708 getsub(struct wordent *en)
709 {
710     Char orhsb[sizeof(rhsb) / sizeof(Char)];
711     Char *cp;
712     int c, delim, sc;
713     int global;
714 
715     do {
716 	exclnxt = 0;
717 	global = 0;
718 	sc = c = getC(0);
719 	if (c == 'g' || c == 'a') {
720 	    global |= (c == 'g') ? 1 : 2;
721 	    sc = c = getC(0);
722 	}
723 	if (((c =='g') && !(global & 1)) || ((c == 'a') && !(global & 2))) {
724 	    global |= (c == 'g') ? 1 : 2;
725 	    sc = c = getC(0);
726 	}
727 
728 	switch (c) {
729 	case 'p':
730 	    justpr++;
731 	    return (en);
732 	case 'x':
733 	case 'q':
734 	    global |= 1;
735 	    /* FALLTHROUGH */
736 	case 'h':
737 	case 'r':
738 	case 't':
739 	case 'e':
740 	    break;
741 	case '&':
742 	    if (slhs[0] == 0) {
743 		seterror(ERR_NOSUBST);
744 		return (en);
745 	    }
746 	    (void) Strcpy(lhsb, slhs);
747 	    break;
748 #ifdef notdef
749 	case '~':
750 	    if (lhsb[0] == 0)
751 		goto badlhs;
752 	    break;
753 #endif
754 	case 's':
755 	    delim = getC(0);
756 	    if (letter(delim) || Isdigit(delim) || any(" \t\n", delim)) {
757 		unreadc(delim);
758 		lhsb[0] = 0;
759 		seterror(ERR_BADSUBST);
760 		return (en);
761 	    }
762 	    cp = lhsb;
763 	    for (;;) {
764 		c = getC(0);
765 		if (c == '\n') {
766 		    unreadc(c);
767 		    break;
768 		}
769 		if (c == delim)
770 		    break;
771 		if (cp > &lhsb[sizeof(lhsb) / sizeof(Char) - 2]) {
772 		    lhsb[0] = 0;
773 		    seterror(ERR_BADSUBST);
774 		    return (en);
775 		}
776 		if (c == '\\') {
777 		    c = getC(0);
778 		    if (c != delim && c != '\\')
779 			*cp++ = '\\';
780 		}
781 		*cp++ = c;
782 	    }
783 	    if (cp != lhsb)
784 		*cp++ = 0;
785 	    else if (lhsb[0] == 0) {
786 		seterror(ERR_LHS);
787 		return (en);
788 	    }
789 	    cp = rhsb;
790 	    (void)Strcpy(orhsb, cp);
791 	    for (;;) {
792 		c = getC(0);
793 		if (c == '\n') {
794 		    unreadc(c);
795 		    break;
796 		}
797 		if (c == delim)
798 		    break;
799 #ifdef notdef
800 		if (c == '~') {
801 		    if (&cp[Strlen(orhsb)] > &rhsb[sizeof(rhsb) /
802 						   sizeof(Char) - 2])
803 			goto toorhs;
804 		    (void)Strcpy(cp, orhsb);
805 		    cp = Strend(cp);
806 		    continue;
807 		}
808 #endif
809 		if (cp > &rhsb[sizeof(rhsb) / sizeof(Char) - 2]) {
810 		    seterror(ERR_RHSLONG);
811 		    return (en);
812 		}
813 		if (c == '\\') {
814 		    c = getC(0);
815 		    if (c != delim /* && c != '~' */ )
816 			*cp++ = '\\';
817 		}
818 		*cp++ = c;
819 	    }
820 	    *cp++ = 0;
821 	    break;
822 	default:
823 	    if (c == '\n')
824 		unreadc(c);
825 	    seterror(ERR_BADBANGMOD, c);
826 	    return (en);
827 	}
828 	(void)Strcpy(slhs, lhsb);
829 	if (exclc)
830 	    en = dosub(sc, en, global);
831     }
832     while ((c = getC(0)) == ':');
833     unreadc(c);
834     return (en);
835 }
836 
837 static struct wordent *
838 dosub(int sc, struct wordent *en, int global)
839 {
840     struct wordent lexi, *hp, *wdp;
841     int i;
842     int didone, didsub;
843 
844     didone = 0;
845     didsub = 0;
846     i = exclc;
847     hp = &lexi;
848 
849     wdp = hp;
850     while (--i >= 0) {
851 	struct wordent *new = (struct wordent *)xcalloc(1, sizeof *wdp);
852 
853 	new->word = 0;
854 	new->prev = wdp;
855 	new->next = hp;
856 	wdp->next = new;
857 	wdp = new;
858 	en = en->next;
859 	if (en->word) {
860 	    Char *tword, *otword;
861 
862 	    if ((global & 1) || didsub == 0) {
863 		tword = subword(en->word, sc, &didone);
864 		if (didone)
865 		    didsub = 1;
866 		if (global & 2) {
867 		    while (didone && tword != STRNULL) {
868 			otword = tword;
869 			tword = subword(otword, sc, &didone);
870 			if (Strcmp(tword, otword) == 0) {
871 			    xfree((ptr_t) otword);
872 			    break;
873 			}
874 			else
875 			    xfree((ptr_t)otword);
876 		    }
877 		}
878 	    }
879 	    else
880 		tword = Strsave(en->word);
881 	    wdp->word = tword;
882 	}
883     }
884     if (didsub == 0)
885 	seterror(ERR_MODFAIL);
886     hp->prev = wdp;
887     return (&enthist(-1000, &lexi, 0)->Hlex);
888 }
889 
890 static Char *
891 subword(Char *cp, int type, int *adid)
892 {
893     Char wbuf[BUFSIZE];
894     Char *mp, *np, *wp;
895     int i;
896 
897     *adid = 0;
898     switch (type) {
899     case 'r':
900     case 'e':
901     case 'h':
902     case 't':
903     case 'q':
904     case 'x':
905 	wp = domod(cp, type);
906 	if (wp == 0)
907 	    return (Strsave(cp));
908 	*adid = 1;
909 	return (wp);
910     default:
911 	wp = wbuf;
912 	i = BUFSIZE - 4;
913 	for (mp = cp; *mp; mp++)
914 	    if (matchs(mp, lhsb)) {
915 		for (np = cp; np < mp;)
916 		    *wp++ = *np++, --i;
917 		for (np = rhsb; *np; np++)
918 		    switch (*np) {
919 		    case '\\':
920 			if (np[1] == '&')
921 			    np++;
922 			/* FALLTHROUGH */
923 		    default:
924 			if (--i < 0) {
925 			    seterror(ERR_SUBOVFL);
926 			    return (STRNULL);
927 			}
928 			*wp++ = *np;
929 			continue;
930 		    case '&':
931 			i -= Strlen(lhsb);
932 			if (i < 0) {
933 			    seterror(ERR_SUBOVFL);
934 			    return (STRNULL);
935 			}
936 			*wp = 0;
937 			(void) Strcat(wp, lhsb);
938 			wp = Strend(wp);
939 			continue;
940 		    }
941 		mp += Strlen(lhsb);
942 		i -= Strlen(mp);
943 		if (i < 0) {
944 		    seterror(ERR_SUBOVFL);
945 		    return (STRNULL);
946 		}
947 		*wp = 0;
948 		(void) Strcat(wp, mp);
949 		*adid = 1;
950 		return (Strsave(wbuf));
951 	    }
952 	return (Strsave(cp));
953     }
954 }
955 
956 Char *
957 domod(Char *cp, int type)
958 {
959     Char *wp, *xp;
960     int c;
961 
962     switch (type) {
963     case 'x':
964     case 'q':
965 	wp = Strsave(cp);
966 	for (xp = wp; (c = *xp) != '\0'; xp++)
967 	    if ((c != ' ' && c != '\t') || type == 'q')
968 		*xp |= QUOTE;
969 	return (wp);
970     case 'h':
971     case 't':
972 	if (!any(short2str(cp), '/'))
973 	    return (type == 't' ? Strsave(cp) : 0);
974 	wp = Strend(cp);
975 	while (*--wp != '/')
976 	    continue;
977 	if (type == 'h')
978 	    xp = Strsave(cp), xp[wp - cp] = 0;
979 	else
980 	    xp = Strsave(wp + 1);
981 	return (xp);
982     case 'e':
983     case 'r':
984 	wp = Strend(cp);
985 	for (wp--; wp >= cp && *wp != '/'; wp--)
986 	    if (*wp == '.') {
987 		if (type == 'e')
988 		    xp = Strsave(wp + 1);
989 		else
990 		    xp = Strsave(cp), xp[wp - cp] = 0;
991 		return (xp);
992 	    }
993 	return (Strsave(type == 'e' ? STRNULL : cp));
994     default:
995 	break;
996     }
997     return (0);
998 }
999 
1000 static int
1001 matchs(Char *str, Char *pat)
1002 {
1003     while (*str && *pat && *str == *pat)
1004 	str++, pat++;
1005     return (*pat == 0);
1006 }
1007 
1008 static int
1009 getsel(int *al, int *ar, int dol)
1010 {
1011     int c, i;
1012     int first;
1013 
1014     c = getC(0);
1015     first = *al < 0;
1016 
1017     switch (c) {
1018     case '%':
1019 	if (quesarg == -1) {
1020 	    seterror(ERR_BADBANGARG);
1021 	    return (0);
1022 	}
1023 	if (*al < 0)
1024 	    *al = quesarg;
1025 	*ar = quesarg;
1026 	break;
1027     case '-':
1028 	if (*al < 0) {
1029 	    *al = 0;
1030 	    *ar = dol - 1;
1031 	    unreadc(c);
1032 	}
1033 	return (1);
1034     case '^':
1035 	if (*al < 0)
1036 	    *al = 1;
1037 	*ar = 1;
1038 	break;
1039     case '$':
1040 	if (*al < 0)
1041 	    *al = dol;
1042 	*ar = dol;
1043 	break;
1044     case '*':
1045 	if (*al < 0)
1046 	    *al = 1;
1047 	*ar = dol;
1048 	if (*ar < *al) {
1049 	    *ar = 0;
1050 	    *al = 1;
1051 	    return (1);
1052 	}
1053 	break;
1054     default:
1055 	if (Isdigit(c)) {
1056 	    i = 0;
1057 	    while (Isdigit(c)) {
1058 		i = i * 10 + c - '0';
1059 		c = getC(0);
1060 	    }
1061 	    if (i < 0)
1062 		i = dol + 1;
1063 	    if (*al < 0)
1064 		*al = i;
1065 	    *ar = i;
1066 	}
1067 	else if (*al < 0)
1068 	    *al = 0, *ar = dol;
1069 	else
1070 	    *ar = dol - 1;
1071 	unreadc(c);
1072 	break;
1073     }
1074     if (first) {
1075 	c = getC(0);
1076 	unreadc(c);
1077 	if (any("-$*", c))
1078 	    return (1);
1079     }
1080     if (*al > *ar || *ar > dol) {
1081 	seterror(ERR_BADBANGARG);
1082 	return (0);
1083     }
1084     return (1);
1085 
1086 }
1087 
1088 static struct wordent *
1089 gethent(int sc)
1090 {
1091     struct Hist *hp;
1092     Char *np;
1093     char *str;
1094     int c, event;
1095     int back;
1096 
1097     back = 0;
1098     c = sc == HISTSUB ? HIST : getC(0);
1099     if (c == HIST) {
1100 	if (alhistp)
1101 	    return (alhistp);
1102 	event = eventno;
1103     }
1104     else
1105 	switch (c) {
1106 	case ':':
1107 	case '^':
1108 	case '$':
1109 	case '*':
1110 	case '%':
1111 	    ungetC(c);
1112 	    if (lastev == eventno && alhistp)
1113 		return (alhistp);
1114 	    event = lastev;
1115 	    break;
1116 	case '#':		/* !# is command being typed in (mrh) */
1117 	    if (--hleft == 0) {
1118 		seterror(ERR_HISTLOOP);
1119 		return (0);
1120 	    }
1121 	    else
1122 		return (&paraml);
1123 	    /* NOTREACHED */
1124 	case '-':
1125 	    back = 1;
1126 	    c = getC(0);
1127 	    /* FALLTHROUGH */
1128 	default:
1129 	    if (any("(=~", c)) {
1130 		unreadc(c);
1131 		ungetC(HIST);
1132 		return (0);
1133 	    }
1134 	    np = lhsb;
1135 	    event = 0;
1136 	    while (!cmap(c, _ESC | _META | _QF | _QB) && !any("${}:", c)) {
1137 		if (event != -1 && Isdigit(c))
1138 		    event = event * 10 + c - '0';
1139 		else
1140 		    event = -1;
1141 		if (np < &lhsb[sizeof(lhsb) / sizeof(Char) - 2])
1142 		    *np++ = c;
1143 		c = getC(0);
1144 	    }
1145 	    unreadc(c);
1146 	    if (np == lhsb) {
1147 		ungetC(HIST);
1148 		return (0);
1149 	    }
1150 	    *np++ = 0;
1151 	    if (event != -1) {
1152 		/*
1153 		 * History had only digits
1154 		 */
1155 		if (back)
1156 		    event = eventno + (alhistp == 0) - (event ? event : 0);
1157 		break;
1158 	    }
1159 	    hp = findev(lhsb, 0);
1160 	    if (hp)
1161 		lastev = hp->Hnum;
1162 	    return (&hp->Hlex);
1163 	case '?':
1164 	    np = lhsb;
1165 	    for (;;) {
1166 		c = getC(0);
1167 		if (c == '\n') {
1168 		    unreadc(c);
1169 		    break;
1170 		}
1171 		if (c == '?')
1172 		    break;
1173 		if (np < &lhsb[sizeof(lhsb) / sizeof(Char) - 2])
1174 		    *np++ = c;
1175 	    }
1176 	    if (np == lhsb) {
1177 		if (lhsb[0] == 0) {
1178 		    seterror(ERR_NOSEARCH);
1179 		    return (0);
1180 		}
1181 	    }
1182 	    else
1183 		*np++ = 0;
1184 	    hp = findev(lhsb, 1);
1185 	    if (hp)
1186 		lastev = hp->Hnum;
1187 	    return (&hp->Hlex);
1188 	}
1189 
1190     for (hp = Histlist.Hnext; hp; hp = hp->Hnext)
1191 	if (hp->Hnum == event) {
1192 	    hp->Href = eventno;
1193 	    lastev = hp->Hnum;
1194 	    return (&hp->Hlex);
1195 	}
1196     np = putn(event);
1197     str = vis_str(np);
1198     xfree((ptr_t) np);
1199     seterror(ERR_NOEVENT, str);
1200     return (0);
1201 }
1202 
1203 static struct Hist *
1204 findev(Char *cp, int anyarg)
1205 {
1206     struct Hist *hp;
1207 
1208     for (hp = Histlist.Hnext; hp; hp = hp->Hnext) {
1209 	Char *dp, *p, *q;
1210 	struct wordent *lp;
1211 	int argno;
1212 
1213 	lp = hp->Hlex.next;
1214 	argno = 0;
1215 
1216 	/*
1217 	 * The entries added by alias substitution don't have a newline but do
1218 	 * have a negative event number. Savehist() trims off these entries,
1219 	 * but it happens before alias expansion, too early to delete those
1220 	 * from the previous command.
1221 	 */
1222 	if (hp->Hnum < 0)
1223 	    continue;
1224 	if (lp->word[0] == '\n')
1225 	    continue;
1226 	if (!anyarg) {
1227 	    p = cp;
1228 	    q = lp->word;
1229 	    do
1230 		if (!*p)
1231 		    return (hp);
1232 	    while (*p++ == *q++);
1233 	    continue;
1234 	}
1235 	do {
1236 	    for (dp = lp->word; *dp; dp++) {
1237 		p = cp;
1238 		q = dp;
1239 		do
1240 		    if (!*p) {
1241 			quesarg = argno;
1242 			return (hp);
1243 		    }
1244 		while (*p++ == *q++);
1245 	    }
1246 	    lp = lp->next;
1247 	    argno++;
1248 	} while (lp->word[0] != '\n');
1249     }
1250     seterror(ERR_NOEVENT, vis_str(cp));
1251     return (0);
1252 }
1253 
1254 
1255 static void
1256 setexclp(Char *cp)
1257 {
1258     if (cp && cp[0] == '\n')
1259 	return;
1260     exclp = cp;
1261 }
1262 
1263 void
1264 unreadc(int c)
1265 {
1266     peekread = c;
1267 }
1268 
1269 int
1270 readc(int wanteof)
1271 {
1272     static int sincereal;
1273     int c;
1274 
1275     aret = F_SEEK;
1276     if ((c = peekread) != '\0') {
1277 	peekread = 0;
1278 	return (c);
1279     }
1280 top:
1281     aret = F_SEEK;
1282     if (alvecp) {
1283 	aret = A_SEEK;
1284 	if ((c = *alvecp++) != '\0')
1285 	    return (c);
1286 	if (alvec && *alvec) {
1287 		alvecp = *alvec++;
1288 		return (' ');
1289 	}
1290 	else {
1291 	    aret = F_SEEK;
1292 	    alvecp = NULL;
1293 	    return('\n');
1294 	}
1295     }
1296     if (alvec) {
1297 	if ((alvecp = *alvec) != '\0') {
1298 	    alvec++;
1299 	    goto top;
1300 	}
1301 	/* Infinite source! */
1302 	return ('\n');
1303     }
1304     if (evalp) {
1305 	aret = E_SEEK;
1306 	if ((c = *evalp++) != '\0')
1307 	    return (c);
1308 	if (evalvec && *evalvec) {
1309 	    evalp = *evalvec++;
1310 	    return (' ');
1311 	}
1312 	aret = F_SEEK;
1313 	evalp = 0;
1314     }
1315     if (evalvec) {
1316 	if (evalvec == (Char **) 1) {
1317 	    doneinp = 1;
1318 	    reset();
1319 	}
1320 	if ((evalp = *evalvec) != '\0') {
1321 	    evalvec++;
1322 	    goto top;
1323 	}
1324 	evalvec = (Char **) 1;
1325 	return ('\n');
1326     }
1327     do {
1328 	if (arginp == (Char *) 1 || onelflg == 1) {
1329 	    if (wanteof)
1330 		return (-1);
1331 	    exitstat();
1332 	}
1333 	if (arginp) {
1334 	    if ((c = *arginp++) == 0) {
1335 		arginp = (Char *) 1;
1336 		return ('\n');
1337 	    }
1338 	    return (c);
1339 	}
1340 reread:
1341 	c = bgetc();
1342 	if (c < 0) {
1343 	    struct termios tty;
1344 	    if (wanteof)
1345 		return (-1);
1346 	    /* was isatty but raw with ignoreeof yields problems */
1347 	    if (tcgetattr(SHIN, &tty) == 0 && (tty.c_lflag & ICANON))
1348 	    {
1349 		/* was 'short' for FILEC */
1350 		pid_t     ctpgrp;
1351 
1352 		if (++sincereal > 25)
1353 		    goto oops;
1354 		if (tpgrp != -1 &&
1355 		    (ctpgrp = tcgetpgrp(FSHTTY)) != -1 &&
1356 		    tpgrp != ctpgrp) {
1357 		    (void)tcsetpgrp(FSHTTY, tpgrp);
1358 		    (void)kill(-ctpgrp, SIGHUP);
1359 		    (void)fprintf(csherr, "Reset tty pgrp from %ld to %ld\n",
1360 				   (long)ctpgrp, (long)tpgrp);
1361 		    goto reread;
1362 		}
1363 		if (adrof(STRignoreeof)) {
1364 		    if (loginsh)
1365 			(void)fprintf(csherr,"\nUse \"logout\" to logout.\n");
1366 		    else
1367 			(void)fprintf(csherr,"\nUse \"exit\" to leave csh.\n");
1368 		    reset();
1369 		}
1370 		if (chkstop == 0)
1371 		    panystop(1);
1372 	    }
1373     oops:
1374 	    doneinp = 1;
1375 	    reset();
1376 	}
1377 	sincereal = 0;
1378 	if (c == '\n' && onelflg)
1379 	    onelflg--;
1380     } while (c == 0);
1381     return (c);
1382 }
1383 
1384 static int
1385 bgetc(void)
1386 {
1387 #ifdef FILEC
1388     char tbuf[BUFSIZE + 1];
1389     Char ttyline[BUFSIZE];
1390     int c, buf, numleft, off, roomleft;
1391 
1392     numleft = 0;
1393 #else /* FILEC */
1394     char tbuf[BUFSIZE + 1];
1395     int c, buf, off;
1396 #endif /* !FILEC */
1397 
1398     if (cantell) {
1399 	if (fseekp < fbobp || fseekp > feobp) {
1400 	    fbobp = feobp = fseekp;
1401 	    (void)lseek(SHIN, fseekp, SEEK_SET);
1402 	}
1403 	if (fseekp == feobp) {
1404 	    int i;
1405 
1406 	    fbobp = feobp;
1407 	    do
1408 		c = read(SHIN, tbuf, BUFSIZE);
1409 	    while (c < 0 && errno == EINTR);
1410 	    if (c <= 0)
1411 		return (-1);
1412 	    for (i = 0; i < c; i++)
1413 		fbuf[0][i] = (unsigned char) tbuf[i];
1414 	    feobp += c;
1415 	}
1416 	c = fbuf[0][fseekp - fbobp];
1417 	fseekp++;
1418 	return (c);
1419     }
1420 
1421 again:
1422     buf = (int) fseekp / BUFSIZE;
1423     if (buf >= fblocks) {
1424 	Char **nfbuf;
1425 
1426 	nfbuf = (Char **)xcalloc((size_t) (fblocks + 2), sizeof(char **));
1427 	if (fbuf) {
1428 	    (void)blkcpy(nfbuf, fbuf);
1429 	    xfree((ptr_t) fbuf);
1430 	}
1431 	fbuf = nfbuf;
1432 	fbuf[fblocks] = (Char *)xcalloc(BUFSIZE, sizeof(Char));
1433 	fblocks++;
1434 	if (!intty)
1435 	    goto again;
1436     }
1437     if (fseekp >= feobp) {
1438 	buf = (int) feobp / BUFSIZE;
1439 	off = (int) feobp % BUFSIZE;
1440 	roomleft = BUFSIZE - off;
1441 
1442 #ifdef FILEC
1443 	roomleft = BUFSIZE - off;
1444 	for (;;) {
1445 	    if (filec && intty) {
1446 		c = numleft ? numleft : tenex(ttyline, BUFSIZE);
1447 		if (c > roomleft) {
1448 		    /* start with fresh buffer */
1449 		    feobp = fseekp = fblocks * BUFSIZE;
1450 		    numleft = c;
1451 		    goto again;
1452 		}
1453 		if (c > 0)
1454 		    (void)memcpy(fbuf[buf] + off, ttyline, c * sizeof(Char));
1455 		numleft = 0;
1456 	    }
1457 	    else {
1458 #endif
1459 		c = read(SHIN, tbuf, roomleft);
1460 		if (c > 0) {
1461 		    int     i;
1462 		    Char   *ptr = fbuf[buf] + off;
1463 
1464 		    for (i = 0; i < c; i++)
1465 			ptr[i] = (unsigned char) tbuf[i];
1466 		}
1467 #ifdef FILEC
1468 	    }
1469 #endif
1470 	    if (c >= 0)
1471 		break;
1472 	    if (errno == EWOULDBLOCK) {
1473 		int     iooff = 0;
1474 
1475 		(void)ioctl(SHIN, FIONBIO, (ioctl_t) & iooff);
1476 	    }
1477 	    else if (errno != EINTR)
1478 		break;
1479 #ifdef FILEC
1480 	}
1481 #endif
1482 	if (c <= 0)
1483 	    return (-1);
1484 	feobp += c;
1485 #ifndef FILEC
1486 	goto again;
1487 #else
1488 	if (filec && !intty)
1489 	    goto again;
1490 #endif
1491     }
1492     c = fbuf[buf][(int)fseekp % BUFSIZE];
1493     fseekp++;
1494     return (c);
1495 }
1496 
1497 static void
1498 bfree(void)
1499 {
1500     int i, sb;
1501 
1502     if (cantell)
1503 	return;
1504     if (whyles)
1505 	return;
1506     sb = (int)(fseekp - 1) / BUFSIZE;
1507     if (sb > 0) {
1508 	for (i = 0; i < sb; i++)
1509 	    xfree((ptr_t) fbuf[i]);
1510 	(void)blkcpy(fbuf, &fbuf[sb]);
1511 	fseekp -= BUFSIZE * sb;
1512 	feobp -= BUFSIZE * sb;
1513 	fblocks -= sb;
1514     }
1515 }
1516 
1517 void
1518 bseek(struct Ain *l)
1519 {
1520     switch (aret = l->type) {
1521     case A_SEEK:
1522 	alvec = l->a_seek;
1523 	alvecp = l->c_seek;
1524 	return;
1525     case E_SEEK:
1526 	evalvec = l->a_seek;
1527 	evalp = l->c_seek;
1528 	return;
1529     case F_SEEK:
1530 	fseekp = l->f_seek;
1531 	return;
1532     default:
1533 	(void)fprintf(csherr, "Bad seek type %d\n", aret);
1534 	abort();
1535     }
1536 }
1537 
1538 void
1539 btell(struct Ain *l)
1540 {
1541     switch (l->type = aret) {
1542     case A_SEEK:
1543 	l->a_seek = alvec;
1544 	l->c_seek = alvecp;
1545 	return;
1546     case E_SEEK:
1547 	l->a_seek = evalvec;
1548 	l->c_seek = evalp;
1549 	return;
1550     case F_SEEK:
1551 	l->f_seek = fseekp;
1552 	l->a_seek = NULL;
1553 	return;
1554     default:
1555 	(void)fprintf(csherr, "Bad seek type %d\n", aret);
1556 	abort();
1557     }
1558 }
1559 
1560 void
1561 btoeof(void)
1562 {
1563     (void)lseek(SHIN, (off_t) 0, SEEK_END);
1564     aret = F_SEEK;
1565     fseekp = feobp;
1566     alvec = NULL;
1567     alvecp = NULL;
1568     evalvec = NULL;
1569     evalp = NULL;
1570     wfree();
1571     bfree();
1572 }
1573 
1574 void
1575 settell(void)
1576 {
1577     cantell = 0;
1578     if (arginp || onelflg || intty)
1579 	return;
1580     if (lseek(SHIN, (off_t) 0, SEEK_CUR) < 0 || errno == ESPIPE)
1581 	return;
1582     fbuf = (Char **)xcalloc(2, sizeof(Char **));
1583     fblocks = 1;
1584     fbuf[0] = (Char *)xcalloc(BUFSIZE, sizeof(Char));
1585     fseekp = fbobp = feobp = lseek(SHIN, (off_t) 0, SEEK_CUR);
1586     cantell = 1;
1587 }
1588