xref: /original-bsd/bin/csh/glob.c (revision 1e9806ef)
1 static	char *sccsid = "@(#)glob.c 4.6 08/31/84";
2 
3 #include "sh.h"
4 #include <sys/dir.h>
5 
6 /*
7  * C Shell
8  */
9 
10 int	globcnt;
11 
12 char	*globchars =	"`{[*?";
13 
14 char	*gpath, *gpathp, *lastgpathp;
15 int	globbed;
16 bool	noglob;
17 bool	nonomatch;
18 char	*entp;
19 char	**sortbas;
20 
21 char **
22 glob(v)
23 	register char **v;
24 {
25 	char agpath[BUFSIZ];
26 	char *agargv[GAVSIZ];
27 
28 	gpath = agpath; gpathp = gpath; *gpathp = 0;
29 	lastgpathp = &gpath[sizeof agpath - 2];
30 	ginit(agargv); globcnt = 0;
31 #ifdef GDEBUG
32 	printf("glob entered: "); blkpr(v); printf("\n");
33 #endif
34 	noglob = adrof("noglob") != 0;
35 	nonomatch = adrof("nonomatch") != 0;
36 	globcnt = noglob | nonomatch;
37 	while (*v)
38 		collect(*v++);
39 #ifdef GDEBUG
40 	printf("glob done, globcnt=%d, gflag=%d: ", globcnt, gflag); blkpr(gargv); printf("\n");
41 #endif
42 	if (globcnt == 0 && (gflag&1)) {
43 		blkfree(gargv), gargv = 0;
44 		return (0);
45 	} else
46 		return (gargv = copyblk(gargv));
47 }
48 
49 ginit(agargv)
50 	char **agargv;
51 {
52 
53 	agargv[0] = 0; gargv = agargv; sortbas = agargv; gargc = 0;
54 	gnleft = NCARGS - 4;
55 }
56 
57 collect(as)
58 	register char *as;
59 {
60 	register int i;
61 
62 	if (any('`', as)) {
63 #ifdef GDEBUG
64 		printf("doing backp of %s\n", as);
65 #endif
66 		dobackp(as, 0);
67 #ifdef GDEBUG
68 		printf("backp done, acollect'ing\n");
69 #endif
70 		for (i = 0; i < pargc; i++)
71 			if (noglob) {
72 				Gcat(pargv[i], "");
73 				sortbas = &gargv[gargc];
74 			} else
75 				acollect(pargv[i]);
76 		if (pargv)
77 			blkfree(pargv), pargv = 0;
78 #ifdef GDEBUG
79 		printf("acollect done\n");
80 #endif
81 	} else if (noglob || eq(as, "{") || eq(as, "{}")) {
82 		Gcat(as, "");
83 		sort();
84 	} else
85 		acollect(as);
86 }
87 
88 acollect(as)
89 	register char *as;
90 {
91 	register int ogargc = gargc;
92 
93 	gpathp = gpath; *gpathp = 0; globbed = 0;
94 	expand(as);
95 	if (gargc == ogargc) {
96 		if (nonomatch) {
97 			Gcat(as, "");
98 			sort();
99 		}
100 	} else
101 		sort();
102 }
103 
104 sort()
105 {
106 	register char **p1, **p2, *c;
107 	char **Gvp = &gargv[gargc];
108 
109 	p1 = sortbas;
110 	while (p1 < Gvp-1) {
111 		p2 = p1;
112 		while (++p2 < Gvp)
113 			if (strcmp(*p1, *p2) > 0)
114 				c = *p1, *p1 = *p2, *p2 = c;
115 		p1++;
116 	}
117 	sortbas = Gvp;
118 }
119 
120 expand(as)
121 	char *as;
122 {
123 	register char *cs;
124 	register char *sgpathp, *oldcs;
125 	struct stat stb;
126 
127 	sgpathp = gpathp;
128 	cs = as;
129 	if (*cs == '~' && gpathp == gpath) {
130 		addpath('~');
131 		for (cs++; letter(*cs) || digit(*cs) || *cs == '-';)
132 			addpath(*cs++);
133 		if (!*cs || *cs == '/') {
134 			if (gpathp != gpath + 1) {
135 				*gpathp = 0;
136 				if (gethdir(gpath + 1))
137 					error("Unknown user: %s", gpath + 1);
138 				strcpy(gpath, gpath + 1);
139 			} else
140 				strcpy(gpath, value("home"));
141 			gpathp = strend(gpath);
142 		}
143 	}
144 	while (!any(*cs, globchars)) {
145 		if (*cs == 0) {
146 			if (!globbed)
147 				Gcat(gpath, "");
148 			else if (stat(gpath, &stb) >= 0) {
149 				Gcat(gpath, "");
150 				globcnt++;
151 			}
152 			goto endit;
153 		}
154 		addpath(*cs++);
155 	}
156 	oldcs = cs;
157 	while (cs > as && *cs != '/')
158 		cs--, gpathp--;
159 	if (*cs == '/')
160 		cs++, gpathp++;
161 	*gpathp = 0;
162 	if (*oldcs == '{') {
163 		execbrc(cs, NOSTR);
164 		return;
165 	}
166 	matchdir(cs);
167 endit:
168 	gpathp = sgpathp;
169 	*gpathp = 0;
170 }
171 
172 matchdir(pattern)
173 	char *pattern;
174 {
175 	struct stat stb;
176 	register struct direct *dp;
177 	DIR *dirp;
178 	register int cnt;
179 
180 	dirp = opendir(gpath);
181 	if (dirp == NULL) {
182 		if (globbed)
183 			return;
184 		goto patherr2;
185 	}
186 	if (fstat(dirp->dd_fd, &stb) < 0)
187 		goto patherr1;
188 	if (!isdir(stb)) {
189 		errno = ENOTDIR;
190 		goto patherr1;
191 	}
192 	while ((dp = readdir(dirp)) != NULL) {
193 		if (dp->d_ino == 0)
194 			continue;
195 		if (match(dp->d_name, pattern)) {
196 			Gcat(gpath, dp->d_name);
197 			globcnt++;
198 		}
199 	}
200 	closedir(dirp);
201 	return;
202 
203 patherr1:
204 	closedir(dirp);
205 patherr2:
206 	Perror(gpath);
207 }
208 
209 execbrc(p, s)
210 	char *p, *s;
211 {
212 	char restbuf[BUFSIZ + 2];
213 	register char *pe, *pm, *pl;
214 	int brclev = 0;
215 	char *lm, savec, *sgpathp;
216 
217 	for (lm = restbuf; *p != '{'; *lm++ = *p++)
218 		continue;
219 	for (pe = ++p; *pe; pe++)
220 	switch (*pe) {
221 
222 	case '{':
223 		brclev++;
224 		continue;
225 
226 	case '}':
227 		if (brclev == 0)
228 			goto pend;
229 		brclev--;
230 		continue;
231 
232 	case '[':
233 		for (pe++; *pe && *pe != ']'; pe++)
234 			continue;
235 		if (!*pe)
236 			error("Missing ]");
237 		continue;
238 	}
239 pend:
240 	if (brclev || !*pe)
241 		error("Missing }");
242 	for (pl = pm = p; pm <= pe; pm++)
243 	switch (*pm & (QUOTE|TRIM)) {
244 
245 	case '{':
246 		brclev++;
247 		continue;
248 
249 	case '}':
250 		if (brclev) {
251 			brclev--;
252 			continue;
253 		}
254 		goto doit;
255 
256 	case ','|QUOTE:
257 	case ',':
258 		if (brclev)
259 			continue;
260 doit:
261 		savec = *pm;
262 		*pm = 0;
263 		strcpy(lm, pl);
264 		strcat(restbuf, pe + 1);
265 		*pm = savec;
266 		if (s == 0) {
267 			sgpathp = gpathp;
268 			expand(restbuf);
269 			gpathp = sgpathp;
270 			*gpathp = 0;
271 		} else if (amatch(s, restbuf))
272 			return (1);
273 		sort();
274 		pl = pm + 1;
275 		continue;
276 
277 	case '[':
278 		for (pm++; *pm && *pm != ']'; pm++)
279 			continue;
280 		if (!*pm)
281 			error("Missing ]");
282 		continue;
283 	}
284 	return (0);
285 }
286 
287 match(s, p)
288 	char *s, *p;
289 {
290 	register int c;
291 	register char *sentp;
292 	char sglobbed = globbed;
293 
294 	if (*s == '.' && *p != '.')
295 		return (0);
296 	sentp = entp;
297 	entp = s;
298 	c = amatch(s, p);
299 	entp = sentp;
300 	globbed = sglobbed;
301 	return (c);
302 }
303 
304 amatch(s, p)
305 	register char *s, *p;
306 {
307 	register int scc;
308 	int ok, lc;
309 	char *sgpathp;
310 	struct stat stb;
311 	int c, cc;
312 
313 	globbed = 1;
314 	for (;;) {
315 		scc = *s++ & TRIM;
316 		switch (c = *p++) {
317 
318 		case '{':
319 			return (execbrc(p - 1, s - 1));
320 
321 		case '[':
322 			ok = 0;
323 			lc = 077777;
324 			while (cc = *p++) {
325 				if (cc == ']') {
326 					if (ok)
327 						break;
328 					return (0);
329 				}
330 				if (cc == '-') {
331 					if (lc <= scc && scc <= *p++)
332 						ok++;
333 				} else
334 					if (scc == (lc = cc))
335 						ok++;
336 			}
337 			if (cc == 0)
338 				error("Missing ]");
339 			continue;
340 
341 		case '*':
342 			if (!*p)
343 				return (1);
344 			if (*p == '/') {
345 				p++;
346 				goto slash;
347 			}
348 			for (s--; *s; s++)
349 				if (amatch(s, p))
350 					return (1);
351 			return (0);
352 
353 		case 0:
354 			return (scc == 0);
355 
356 		default:
357 			if ((c & TRIM) != scc)
358 				return (0);
359 			continue;
360 
361 		case '?':
362 			if (scc == 0)
363 				return (0);
364 			continue;
365 
366 		case '/':
367 			if (scc)
368 				return (0);
369 slash:
370 			s = entp;
371 			sgpathp = gpathp;
372 			while (*s)
373 				addpath(*s++);
374 			addpath('/');
375 			if (stat(gpath, &stb) == 0 && isdir(stb))
376 				if (*p == 0) {
377 					Gcat(gpath, "");
378 					globcnt++;
379 				} else
380 					expand(p);
381 			gpathp = sgpathp;
382 			*gpathp = 0;
383 			return (0);
384 		}
385 	}
386 }
387 
388 Gmatch(s, p)
389 	register char *s, *p;
390 {
391 	register int scc;
392 	int ok, lc;
393 	int c, cc;
394 
395 	for (;;) {
396 		scc = *s++ & TRIM;
397 		switch (c = *p++) {
398 
399 		case '[':
400 			ok = 0;
401 			lc = 077777;
402 			while (cc = *p++) {
403 				if (cc == ']') {
404 					if (ok)
405 						break;
406 					return (0);
407 				}
408 				if (cc == '-') {
409 					if (lc <= scc && scc <= *p++)
410 						ok++;
411 				} else
412 					if (scc == (lc = cc))
413 						ok++;
414 			}
415 			if (cc == 0)
416 				bferr("Missing ]");
417 			continue;
418 
419 		case '*':
420 			if (!*p)
421 				return (1);
422 			for (s--; *s; s++)
423 				if (Gmatch(s, p))
424 					return (1);
425 			return (0);
426 
427 		case 0:
428 			return (scc == 0);
429 
430 		default:
431 			if ((c & TRIM) != scc)
432 				return (0);
433 			continue;
434 
435 		case '?':
436 			if (scc == 0)
437 				return (0);
438 			continue;
439 
440 		}
441 	}
442 }
443 
444 Gcat(s1, s2)
445 	register char *s1, *s2;
446 {
447 
448 	gnleft -= strlen(s1) + strlen(s2) + 1;
449 	if (gnleft <= 0 || ++gargc >= GAVSIZ)
450 		error("Arguments too long");
451 	gargv[gargc] = 0;
452 	gargv[gargc - 1] = strspl(s1, s2);
453 }
454 
455 addpath(c)
456 	char c;
457 {
458 
459 	if (gpathp >= lastgpathp)
460 		error("Pathname too long");
461 	*gpathp++ = c;
462 	*gpathp = 0;
463 }
464 
465 rscan(t, f)
466 	register char **t;
467 	int (*f)();
468 {
469 	register char *p, c;
470 
471 	while (p = *t++) {
472 		if (f == tglob)
473 			if (*p == '~')
474 				gflag |= 2;
475 			else if (eq(p, "{") || eq(p, "{}"))
476 				continue;
477 		while (c = *p++)
478 			(*f)(c);
479 	}
480 }
481 
482 scan(t, f)
483 	register char **t;
484 	int (*f)();
485 {
486 	register char *p, c;
487 
488 	while (p = *t++)
489 		while (c = *p)
490 			*p++ = (*f)(c);
491 }
492 
493 tglob(c)
494 	register char c;
495 {
496 
497 	if (any(c, globchars))
498 		gflag |= c == '{' ? 2 : 1;
499 	return (c);
500 }
501 
502 trim(c)
503 	char c;
504 {
505 
506 	return (c & TRIM);
507 }
508 
509 tback(c)
510 	char c;
511 {
512 
513 	if (c == '`')
514 		gflag = 1;
515 }
516 
517 char *
518 globone(str)
519 	register char *str;
520 {
521 	char *gv[2];
522 	register char **gvp;
523 	register char *cp;
524 
525 	gv[0] = str;
526 	gv[1] = 0;
527 	gflag = 0;
528 	rscan(gv, tglob);
529 	if (gflag) {
530 		gvp = glob(gv);
531 		if (gvp == 0) {
532 			setname(str);
533 			bferr("No match");
534 		}
535 		cp = *gvp++;
536 		if (cp == 0)
537 			cp = "";
538 		else if (*gvp) {
539 			setname(str);
540 			bferr("Ambiguous");
541 		} else
542 			cp = strip(cp);
543 /*
544 		if (cp == 0 || *gvp) {
545 			setname(str);
546 			bferr(cp ? "Ambiguous" : "No output");
547 		}
548 */
549 		xfree((char *)gargv); gargv = 0;
550 	} else {
551 		scan(gv, trim);
552 		cp = savestr(gv[0]);
553 	}
554 	return (cp);
555 }
556 
557 /*
558  * Command substitute cp.  If literal, then this is
559  * a substitution from a << redirection, and so we should
560  * not crunch blanks and tabs, separating words only at newlines.
561  */
562 char **
563 dobackp(cp, literal)
564 	char *cp;
565 	bool literal;
566 {
567 	register char *lp, *rp;
568 	char *ep;
569 	char word[BUFSIZ];
570 	char *apargv[GAVSIZ + 2];
571 
572 	if (pargv) {
573 		abort();
574 		blkfree(pargv);
575 	}
576 	pargv = apargv;
577 	pargv[0] = NOSTR;
578 	pargcp = pargs = word;
579 	pargc = 0;
580 	pnleft = BUFSIZ - 4;
581 	for (;;) {
582 		for (lp = cp; *lp != '`'; lp++) {
583 			if (*lp == 0) {
584 				if (pargcp != pargs)
585 					pword();
586 #ifdef GDEBUG
587 				printf("leaving dobackp\n");
588 #endif
589 				return (pargv = copyblk(pargv));
590 			}
591 			psave(*lp);
592 		}
593 		lp++;
594 		for (rp = lp; *rp && *rp != '`'; rp++)
595 			if (*rp == '\\') {
596 				rp++;
597 				if (!*rp)
598 					goto oops;
599 			}
600 		if (!*rp)
601 oops:
602 			error("Unmatched `");
603 		ep = savestr(lp);
604 		ep[rp - lp] = 0;
605 		backeval(ep, literal);
606 #ifdef GDEBUG
607 		printf("back from backeval\n");
608 #endif
609 		cp = rp + 1;
610 	}
611 }
612 
613 backeval(cp, literal)
614 	char *cp;
615 	bool literal;
616 {
617 	int pvec[2];
618 	int quoted = (literal || (cp[0] & QUOTE)) ? QUOTE : 0;
619 	char ibuf[BUFSIZ];
620 	register int icnt = 0, c;
621 	register char *ip;
622 	bool hadnl = 0;
623 	char *fakecom[2];
624 	struct command faket;
625 
626 	faket.t_dtyp = TCOM;
627 	faket.t_dflg = 0;
628 	faket.t_dlef = 0;
629 	faket.t_drit = 0;
630 	faket.t_dspr = 0;
631 	faket.t_dcom = fakecom;
632 	fakecom[0] = "` ... `";
633 	fakecom[1] = 0;
634 	/*
635 	 * We do the psave job to temporarily change the current job
636 	 * so that the following fork is considered a separate job.
637 	 * This is so that when backquotes are used in a
638 	 * builtin function that calls glob the "current job" is not corrupted.
639 	 * We only need one level of pushed jobs as long as we are sure to
640 	 * fork here.
641 	 */
642 	psavejob();
643 	/*
644 	 * It would be nicer if we could integrate this redirection more
645 	 * with the routines in sh.sem.c by doing a fake execute on a builtin
646 	 * function that was piped out.
647 	 */
648 	mypipe(pvec);
649 	if (pfork(&faket, -1) == 0) {
650 		struct wordent paraml;
651 		struct command *t;
652 
653 		close(pvec[0]);
654 		dmove(pvec[1], 1);
655 		dmove(SHDIAG, 2);
656 		initdesc();
657 		arginp = cp;
658 		while (*cp)
659 			*cp++ &= TRIM;
660 		lex(&paraml);
661 		if (err)
662 			error(err);
663 		alias(&paraml);
664 		t = syntax(paraml.next, &paraml, 0);
665 		if (err)
666 			error(err);
667 		if (t)
668 			t->t_dflg |= FPAR;
669 		signal(SIGTSTP, SIG_IGN);
670 		signal(SIGTTIN, SIG_IGN);
671 		signal(SIGTTOU, SIG_IGN);
672 		execute(t, -1);
673 		exitstat();
674 	}
675 	xfree(cp);
676 	close(pvec[1]);
677 	do {
678 		int cnt = 0;
679 		for (;;) {
680 			if (icnt == 0) {
681 				ip = ibuf;
682 				icnt = read(pvec[0], ip, BUFSIZ);
683 				if (icnt <= 0) {
684 					c = -1;
685 					break;
686 				}
687 			}
688 			if (hadnl)
689 				break;
690 			--icnt;
691 			c = (*ip++ & TRIM);
692 			if (c == 0)
693 				break;
694 			if (c == '\n') {
695 				/*
696 				 * Continue around the loop one
697 				 * more time, so that we can eat
698 				 * the last newline without terminating
699 				 * this word.
700 				 */
701 				hadnl = 1;
702 				continue;
703 			}
704 			if (!quoted && (c == ' ' || c == '\t'))
705 				break;
706 			cnt++;
707 			psave(c | quoted);
708 		}
709 		/*
710 		 * Unless at end-of-file, we will form a new word
711 		 * here if there were characters in the word, or in
712 		 * any case when we take text literally.  If
713 		 * we didn't make empty words here when literal was
714 		 * set then we would lose blank lines.
715 		 */
716 		if (c != -1 && (cnt || literal))
717 			pword();
718 		hadnl = 0;
719 	} while (c >= 0);
720 #ifdef GDEBUG
721 	printf("done in backeval, pvec: %d %d\n", pvec[0], pvec[1]);
722 	printf("also c = %c <%o>\n", c, c);
723 #endif
724 	close(pvec[0]);
725 	pwait();
726 	prestjob();
727 }
728 
729 psave(c)
730 	char c;
731 {
732 
733 	if (--pnleft <= 0)
734 		error("Word too long");
735 	*pargcp++ = c;
736 }
737 
738 pword()
739 {
740 
741 	psave(0);
742 	if (pargc == GAVSIZ)
743 		error("Too many words from ``");
744 	pargv[pargc++] = savestr(pargs);
745 	pargv[pargc] = NOSTR;
746 #ifdef GDEBUG
747 	printf("got word %s\n", pargv[pargc-1]);
748 #endif
749 	pargcp = pargs;
750 	pnleft = BUFSIZ - 4;
751 }
752