1 #include <u.h>
2 #include <libc.h>
3 #include <bio.h>
4 #include <flate.h>
5 #include "zip.h"
6 
7 enum
8 {
9 	BufSize	= 4096
10 };
11 
12 static	int	cheader(Biobuf *bin, ZipHead *zh);
13 static	int	copyout(int ofd, Biobuf *bin, long len);
14 static	int	crcwrite(void *ofd, void *buf, int n);
15 static	int	findCDir(Biobuf *bin, char *file);
16 static	int	get1(Biobuf *b);
17 static	int	get2(Biobuf *b);
18 static	u32int	get4(Biobuf *b);
19 static	char	*getname(Biobuf *b, int len);
20 static	int	header(Biobuf *bin, ZipHead *zh);
21 static	long	msdos2time(int time, int date);
22 static	int	sunzip(Biobuf *bin);
23 static	int	sunztable(Biobuf *bin);
24 static	void	trailer(Biobuf *bin, ZipHead *zh);
25 static	int	unzip(Biobuf *bin, char *file);
26 static	int	unzipEntry(Biobuf *bin, ZipHead *czh);
27 static	int	unztable(Biobuf *bin, char *file);
28 static	int	wantFile(char *file);
29 
30 static	void	*emalloc(u32int);
31 static	void	error(char*, ...);
32 /* #pragma	varargck	argpos	error	1 */
33 
34 static	Biobuf	bin;
35 static	u32int	crc;
36 static	u32int	*crctab;
37 static	int	debug;
38 static	char	*delfile;
39 static	int	lower;
40 static	int	nwant;
41 static	u32int	rlen;
42 static	int	settimes;
43 static	int	stdout;
44 static	int	verbose;
45 static	char	**want;
46 static	int	wbad;
47 static	u32int	wlen;
48 static	jmp_buf	zjmp;
49 
50 static void
51 usage(void)
52 {
53 	fprint(2, "usage: unzip [-tsv] [-f zipfile] [file ...]\n");
54 	exits("usage");
55 }
56 
57 void
58 main(int argc, char *argv[])
59 {
60 	char *zfile;
61 	int fd, ok, table, stream;
62 
63 	table = 0;
64 	stream = 0;
65 	zfile = nil;
66 	ARGBEGIN{
67 	case 'D':
68 		debug++;
69 		break;
70 	case 'c':
71 		stdout++;
72 		break;
73 	case 'i':
74 		lower++;
75 		break;
76 	case 'f':
77 		zfile = ARGF();
78 		if(zfile == nil)
79 			usage();
80 		break;
81 	case 's':
82 		stream++;
83 		break;
84 	case 't':
85 		table++;
86 		break;
87 	case 'T':
88 		settimes++;
89 		break;
90 	case 'v':
91 		verbose++;
92 		break;
93 	default:
94 		usage();
95 		break;
96 	}ARGEND
97 
98 	nwant = argc;
99 	want = argv;
100 
101 	crctab = mkcrctab(ZCrcPoly);
102 	ok = inflateinit();
103 	if(ok != FlateOk)
104 		sysfatal("inflateinit failed: %s\n", flateerr(ok));
105 
106 	if(zfile == nil){
107 		Binit(&bin, 0, OREAD);
108 		zfile = "<stdin>";
109 	}else{
110 		fd = open(zfile, OREAD);
111 		if(fd < 0)
112 			sysfatal("can't open %s: %r", zfile);
113 		Binit(&bin, fd, OREAD);
114 	}
115 
116 	if(table){
117 		if(stream)
118 			ok = sunztable(&bin);
119 		else
120 			ok = unztable(&bin, zfile);
121 	}else{
122 		if(stream)
123 			ok = sunzip(&bin);
124 		else
125 			ok = unzip(&bin, zfile);
126 	}
127 
128 	exits(ok ? nil: "errors");
129 }
130 
131 /*
132  * print the table of contents from the "central directory structure"
133  */
134 static int
135 unztable(Biobuf *bin, char *file)
136 {
137 	ZipHead zh;
138 	int volatile entries;
139 
140 	entries = findCDir(bin, file);
141 	if(entries < 0)
142 		return 0;
143 
144 	if(verbose > 1)
145 		print("%d items in the archive\n", entries);
146 	while(entries-- > 0){
147 		if(setjmp(zjmp)){
148 			free(zh.file);
149 			return 0;
150 		}
151 
152 		memset(&zh, 0, sizeof(zh));
153 		if(!cheader(bin, &zh))
154 			return 1;
155 
156 		if(wantFile(zh.file)){
157 			if(verbose)
158 				print("%-32s %10lud %s", zh.file, zh.uncsize, ctime(msdos2time(zh.modtime, zh.moddate)));
159 			else
160 				print("%s\n", zh.file);
161 
162 			if(verbose > 1){
163 				print("\tmade by os %d vers %d.%d\n", zh.madeos, zh.madevers/10, zh.madevers % 10);
164 				print("\textract by os %d vers %d.%d\n", zh.extos, zh.extvers/10, zh.extvers % 10);
165 				print("\tflags %x\n", zh.flags);
166 				print("\tmethod %d\n", zh.meth);
167 				print("\tmod time %d\n", zh.modtime);
168 				print("\tmod date %d\n", zh.moddate);
169 				print("\tcrc %lux\n", zh.crc);
170 				print("\tcompressed size %lud\n", zh.csize);
171 				print("\tuncompressed size %lud\n", zh.uncsize);
172 				print("\tinternal attributes %ux\n", zh.iattr);
173 				print("\texternal attributes %lux\n", zh.eattr);
174 				print("\tstarts at %ld\n", zh.off);
175 			}
176 		}
177 
178 		free(zh.file);
179 		zh.file = nil;
180 	}
181 
182 	return 1;
183 }
184 
185 /*
186  * print the "local file header" table of contents
187  */
188 static int
189 sunztable(Biobuf *bin)
190 {
191 	ZipHead zh;
192 	vlong off;
193 	u32int hcrc, hcsize, huncsize;
194 	int ok, err;
195 
196 	ok = 1;
197 	for(;;){
198 		if(setjmp(zjmp)){
199 			free(zh.file);
200 			return 0;
201 		}
202 
203 		memset(&zh, 0, sizeof(zh));
204 		if(!header(bin, &zh))
205 			return ok;
206 
207 		hcrc = zh.crc;
208 		hcsize = zh.csize;
209 		huncsize = zh.uncsize;
210 
211 		wlen = 0;
212 		rlen = 0;
213 		crc = 0;
214 		wbad = 0;
215 
216 		if(zh.meth == 0){
217 			if(!copyout(-1, bin, zh.csize))
218 				error("reading data for %s failed: %r", zh.file);
219 		}else if(zh.meth == 8){
220 			off = Boffset(bin);
221 			err = inflate((void*)-1, crcwrite, bin, (int(*)(void*))Bgetc);
222 			if(err != FlateOk)
223 				error("inflate %s failed: %s", zh.file, flateerr(err));
224 			rlen = Boffset(bin) - off;
225 		}else
226 			error("can't handle compression method %d for %s", zh.meth, zh.file);
227 
228 		trailer(bin, &zh);
229 
230 		if(wantFile(zh.file)){
231 			if(verbose)
232 				print("%-32s %10lud %s", zh.file, zh.uncsize, ctime(msdos2time(zh.modtime, zh.moddate)));
233 			else
234 				print("%s\n", zh.file);
235 
236 			if(verbose > 1){
237 				print("\textract by os %d vers %d.%d\n", zh.extos, zh.extvers / 10, zh.extvers % 10);
238 				print("\tflags %x\n", zh.flags);
239 				print("\tmethod %d\n", zh.meth);
240 				print("\tmod time %d\n", zh.modtime);
241 				print("\tmod date %d\n", zh.moddate);
242 				print("\tcrc %lux\n", zh.crc);
243 				print("\tcompressed size %lud\n", zh.csize);
244 				print("\tuncompressed size %lud\n", zh.uncsize);
245 				if((zh.flags & ZTrailInfo) && (hcrc || hcsize || huncsize)){
246 					print("\theader crc %lux\n", zh.crc);
247 					print("\theader compressed size %lud\n", zh.csize);
248 					print("\theader uncompressed size %lud\n", zh.uncsize);
249 				}
250 			}
251 		}
252 
253 		if(zh.crc != crc)
254 			error("crc mismatch for %s", zh.file);
255 		if(zh.uncsize != wlen)
256 			error("output size mismatch for %s", zh.file);
257 		if(zh.csize != rlen)
258 			error("input size mismatch for %s", zh.file);
259 
260 
261 		free(zh.file);
262 		zh.file = nil;
263 	}
264 }
265 
266 /*
267  * extract files using the info in the central directory structure
268  */
269 static int
270 unzip(Biobuf *bin, char *file)
271 {
272 	ZipHead zh;
273 	vlong off;
274 	int volatile ok, eok, entries;
275 
276 	entries = findCDir(bin, file);
277 	if(entries < 0)
278 		return 0;
279 
280 	ok = 1;
281 	while(entries-- > 0){
282 		if(setjmp(zjmp)){
283 			free(zh.file);
284 			return 0;
285 		}
286 		memset(&zh, 0, sizeof(zh));
287 		if(!cheader(bin, &zh))
288 			return ok;
289 
290 
291 		off = Boffset(bin);
292 		if(wantFile(zh.file)){
293 			if(Bseek(bin, zh.off, 0) < 0){
294 				fprint(2, "unzip: can't seek to start of %s, skipping\n", zh.file);
295 				ok = 0;
296 			}else{
297 				eok = unzipEntry(bin, &zh);
298 				if(eok <= 0){
299 					fprint(2, "unzip: skipping %s\n", zh.file);
300 					ok = 0;
301 				}
302 			}
303 		}
304 
305 		free(zh.file);
306 		zh.file = nil;
307 
308 		if(Bseek(bin, off, 0) < 0){
309 			fprint(2, "unzip: can't seek to start of next entry, terminating extraction\n");
310 			return 0;
311 		}
312 	}
313 
314 	return ok;
315 }
316 
317 /*
318  * extract files using the info the "local file headers"
319  */
320 static int
321 sunzip(Biobuf *bin)
322 {
323 	int eok;
324 
325 	for(;;){
326 		eok = unzipEntry(bin, nil);
327 		if(eok == 0)
328 			return 1;
329 		if(eok < 0)
330 			return 0;
331 	}
332 }
333 
334 static int
335 makedir(char *s)
336 {
337 	int f;
338 
339 	if (access(s, AEXIST) == 0)
340 		return -1;
341 	f = create(s, OREAD, DMDIR | 0777);
342 	if (f >= 0)
343 		close(f);
344 	return f;
345 }
346 
347 static void
348 mkpdirs(char *s)
349 {
350 	int done = 0;
351 	char *p = s;
352 
353 	while (!done && (p = strchr(p + 1, '/')) != nil) {
354 		*p = '\0';
355 		done = (access(s, AEXIST) < 0 && makedir(s) < 0);
356 		*p = '/';
357 	}
358 }
359 
360 /*
361  * extracts a single entry from a zip file
362  * czh is the optional corresponding central directory entry
363  */
364 static int
365 unzipEntry(Biobuf *bin, ZipHead *czh)
366 {
367 	Dir *d;
368 	ZipHead zh;
369 	char *p;
370 	vlong off;
371 	int fd, isdir, ok, err;
372 
373 	zh.file = nil;
374 	if(setjmp(zjmp)){
375 		delfile = nil;
376 		free(zh.file);
377 		return -1;
378 	}
379 
380 	memset(&zh, 0, sizeof(zh));
381 	if(!header(bin, &zh))
382 		return 0;
383 
384 	ok = 1;
385 	isdir = 0;
386 
387 	fd = -1;
388 	if(wantFile(zh.file)){
389 		if(verbose)
390 			fprint(2, "extracting %s\n", zh.file);
391 
392 		if(czh != nil && czh->extos == ZDos){
393 			isdir = czh->eattr & ZDDir;
394 			if(isdir && zh.uncsize != 0)
395 				fprint(2, "unzip: ignoring directory data for %s\n", zh.file);
396 		}
397 		if(zh.meth == 0 && zh.uncsize == 0){
398 			p = strchr(zh.file, '\0');
399 			if(p > zh.file && p[-1] == '/')
400 				isdir = 1;
401 		}
402 
403 		if(stdout){
404 			if(ok && !isdir)
405 				fd = 1;
406 		}else if(isdir){
407 			fd = create(zh.file, OREAD, DMDIR | 0775);
408 			if(fd < 0){
409 				mkpdirs(zh.file);
410 				fd = create(zh.file, OREAD, DMDIR | 0775);
411 			}
412 			if(fd < 0){
413 				d = dirstat(zh.file);
414 				if(d == nil || (d->mode & DMDIR) != DMDIR){
415 					fprint(2, "unzip: can't create directory %s: %r\n", zh.file);
416 					ok = 0;
417 				}
418 				free(d);
419 			}
420 		}else if(ok){
421 			fd = create(zh.file, OWRITE, 0664);
422 			if(fd < 0){
423 				mkpdirs(zh.file);
424 				fd = create(zh.file, OWRITE, 0664);
425 			}
426 			if(fd < 0){
427 				fprint(2, "unzip: can't create %s: %r\n", zh.file);
428 				ok = 0;
429 			}else
430 				delfile = zh.file;
431 		}
432 	}
433 
434 	wlen = 0;
435 	rlen = 0;
436 	crc = 0;
437 	wbad = 0;
438 
439 	if(zh.meth == 0){
440 		if(!copyout(fd, bin, zh.csize))
441 			error("copying data for %s failed: %r", zh.file);
442 	}else if(zh.meth == 8){
443 		off = Boffset(bin);
444 		err = inflate((void*)(uintptr)fd, crcwrite, bin, (int(*)(void*))Bgetc);
445 		if(err != FlateOk)
446 			error("inflate failed: %s", flateerr(err));
447 		rlen = Boffset(bin) - off;
448 	}else
449 		error("can't handle compression method %d for %s", zh.meth, zh.file);
450 
451 	trailer(bin, &zh);
452 
453 	if(zh.crc != crc)
454 		error("crc mismatch for %s", zh.file);
455 	if(zh.uncsize != wlen)
456 		error("output size mismatch for %s", zh.file);
457 	if(zh.csize != rlen)
458 		error("input size mismatch for %s", zh.file);
459 
460 	delfile = nil;
461 	free(zh.file);
462 
463 	if(fd >= 0 && !stdout){
464 		if(settimes){
465 			d = dirfstat(fd);
466 			if(d != nil){
467 				d->mtime = msdos2time(zh.modtime, zh.moddate);
468 				if(d->mtime)
469 					dirfwstat(fd, d);
470 			}
471 		}
472 		close(fd);
473 	}
474 
475 	return ok;
476 }
477 
478 static int
479 wantFile(char *file)
480 {
481 	int i, n;
482 
483 	if(nwant == 0)
484 		return 1;
485 	for(i = 0; i < nwant; i++){
486 		if(strcmp(want[i], file) == 0)
487 			return 1;
488 		n = strlen(want[i]);
489 		if(strncmp(want[i], file, n) == 0 && file[n] == '/')
490 			return 1;
491 	}
492 	return 0;
493 }
494 
495 /*
496  * find the start of the central directory
497  * returns the number of entries in the directory,
498  * or -1 if there was an error
499  */
500 static int
501 findCDir(Biobuf *bin, char *file)
502 {
503 	vlong ecoff;
504 	long off, size;
505 	int entries, zclen, dn, ds, de;
506 
507 	ecoff = Bseek(bin, -ZECHeadSize, 2);
508 	if(ecoff < 0){
509 		fprint(2, "unzip: can't seek to contents of %s; try adding -s\n", file);
510 		return -1;
511 	}
512 	if(setjmp(zjmp))
513 		return -1;
514 
515 	if(get4(bin) != ZECHeader){
516 		fprint(2, "unzip: bad magic number for contents of %s\n", file);
517 		return -1;
518 	}
519 	dn = get2(bin);
520 	ds = get2(bin);
521 	de = get2(bin);
522 	entries = get2(bin);
523 	size = get4(bin);
524 	off = get4(bin);
525 	zclen = get2(bin);
526 	while(zclen-- > 0)
527 		get1(bin);
528 
529 	if(verbose > 1){
530 		print("table starts at %ld for %ld bytes\n", off, size);
531 		if(ecoff - size != off)
532 			print("\ttable should start at %lld-%ld=%lld\n", ecoff, size, ecoff-size);
533 		if(dn || ds || de != entries)
534 			print("\tcurrent disk=%d start disk=%d table entries on this disk=%d\n", dn, ds, de);
535 	}
536 
537 	if(Bseek(bin, off, 0) != off){
538 		fprint(2, "unzip: can't seek to start of contents of %s\n", file);
539 		return -1;
540 	}
541 
542 	return entries;
543 }
544 
545 static int
546 cheader(Biobuf *bin, ZipHead *zh)
547 {
548 	u32int v;
549 	int flen, xlen, fclen;
550 
551 	v = get4(bin);
552 	if(v != ZCHeader){
553 		if(v == ZECHeader)
554 			return 0;
555 		error("bad magic number %lux", v);
556 	}
557 	zh->madevers = get1(bin);
558 	zh->madeos = get1(bin);
559 	zh->extvers = get1(bin);
560 	zh->extos = get1(bin);
561 	zh->flags = get2(bin);
562 	zh->meth = get2(bin);
563 	zh->modtime = get2(bin);
564 	zh->moddate = get2(bin);
565 	zh->crc = get4(bin);
566 	zh->csize = get4(bin);
567 	zh->uncsize = get4(bin);
568 	flen = get2(bin);
569 	xlen = get2(bin);
570 	fclen = get2(bin);
571 	get2(bin);		/* disk number start */
572 	zh->iattr = get2(bin);
573 	zh->eattr = get4(bin);
574 	zh->off = get4(bin);
575 
576 	zh->file = getname(bin, flen);
577 
578 	while(xlen-- > 0)
579 		get1(bin);
580 
581 	while(fclen-- > 0)
582 		get1(bin);
583 
584 	return 1;
585 }
586 
587 static int
588 header(Biobuf *bin, ZipHead *zh)
589 {
590 	u32int v;
591 	int flen, xlen;
592 
593 	v = get4(bin);
594 	if(v != ZHeader){
595 		if(v == ZCHeader)
596 			return 0;
597 		error("bad magic number %lux at %lld", v, Boffset(bin)-4);
598 	}
599 	zh->extvers = get1(bin);
600 	zh->extos = get1(bin);
601 	zh->flags = get2(bin);
602 	zh->meth = get2(bin);
603 	zh->modtime = get2(bin);
604 	zh->moddate = get2(bin);
605 	zh->crc = get4(bin);
606 	zh->csize = get4(bin);
607 	zh->uncsize = get4(bin);
608 	flen = get2(bin);
609 	xlen = get2(bin);
610 
611 	zh->file = getname(bin, flen);
612 
613 	while(xlen-- > 0)
614 		get1(bin);
615 
616 	return 1;
617 }
618 
619 static void
620 trailer(Biobuf *bin, ZipHead *zh)
621 {
622 	if(zh->flags & ZTrailInfo){
623 		zh->crc = get4(bin);
624 		zh->csize = get4(bin);
625 		zh->uncsize = get4(bin);
626 	}
627 }
628 
629 static char*
630 getname(Biobuf *bin, int len)
631 {
632 	char *s;
633 	int i, c;
634 
635 	s = emalloc(len + 1);
636 	for(i = 0; i < len; i++){
637 		c = get1(bin);
638 		if(lower)
639 			c = tolower(c);
640 		s[i] = c;
641 	}
642 	s[i] = '\0';
643 	return s;
644 }
645 
646 static int
647 crcwrite(void *out, void *buf, int n)
648 {
649 	int fd, nw;
650 
651 	wlen += n;
652 	crc = blockcrc(crctab, crc, buf, n);
653 	fd = (int)(uintptr)out;
654 	if(fd < 0)
655 		return n;
656 	nw = write(fd, buf, n);
657 	if(nw != n)
658 		wbad = 1;
659 	return nw;
660 }
661 
662 static int
663 copyout(int ofd, Biobuf *bin, long len)
664 {
665 	char buf[BufSize];
666 	int n;
667 
668 	for(; len > 0; len -= n){
669 		n = len;
670 		if(n > BufSize)
671 			n = BufSize;
672 		n = Bread(bin, buf, n);
673 		if(n <= 0)
674 			return 0;
675 		rlen += n;
676 		if(crcwrite((void*)(uintptr)ofd, buf, n) != n)
677 			return 0;
678 	}
679 	return 1;
680 }
681 
682 static u32int
683 get4(Biobuf *b)
684 {
685 	u32int v;
686 	int i, c;
687 
688 	v = 0;
689 	for(i = 0; i < 4; i++){
690 		c = Bgetc(b);
691 		if(c < 0)
692 			error("unexpected eof reading file information");
693 		v |= c << (i * 8);
694 	}
695 	return v;
696 }
697 
698 static int
699 get2(Biobuf *b)
700 {
701 	int i, c, v;
702 
703 	v = 0;
704 	for(i = 0; i < 2; i++){
705 		c = Bgetc(b);
706 		if(c < 0)
707 			error("unexpected eof reading file information");
708 		v |= c << (i * 8);
709 	}
710 	return v;
711 }
712 
713 static int
714 get1(Biobuf *b)
715 {
716 	int c;
717 
718 	c = Bgetc(b);
719 	if(c < 0)
720 		error("unexpected eof reading file information");
721 	return c;
722 }
723 
724 static long
725 msdos2time(int time, int date)
726 {
727 	Tm tm;
728 
729 	tm.hour = time >> 11;
730 	tm.min = (time >> 5) & 63;
731 	tm.sec = (time & 31) << 1;
732 	tm.year = 80 + (date >> 9);
733 	tm.mon = ((date >> 5) & 15) - 1;
734 	tm.mday = date & 31;
735 	tm.zone[0] = '\0';
736 	tm.yday = 0;
737 
738 	return tm2sec(&tm);
739 }
740 
741 static void*
742 emalloc(u32int n)
743 {
744 	void *p;
745 
746 	p = malloc(n);
747 	if(p == nil)
748 		sysfatal("out of memory");
749 	return p;
750 }
751 
752 static void
753 error(char *fmt, ...)
754 {
755 	va_list arg;
756 
757 	fprint(2, "unzip: ");
758 	va_start(arg, fmt);
759 	vfprint(2, fmt, arg);
760 	va_end(arg);
761 	fprint(2, "\n");
762 
763 	if(delfile != nil){
764 		fprint(2, "unzip: removing output file %s\n", delfile);
765 		remove(delfile);
766 		delfile = nil;
767 	}
768 
769 	longjmp(zjmp, 1);
770 }