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	ulong	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(ulong);
31 static	void	error(char*, ...);
32 /* #pragma	varargck	argpos	error	1 */
33 
34 static	Biobuf	bin;
35 static	ulong	crc;
36 static	ulong	*crctab;
37 static	int	debug;
38 static	char	*delfile;
39 static	int	lower;
40 static	int	nwant;
41 static	ulong	rlen;
42 static	int	settimes;
43 static	int	stdout;
44 static	int	verbose;
45 static	char	**want;
46 static	int	wbad;
47 static	ulong	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 	ulong 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 /*
335  * extracts a single entry from a zip file
336  * czh is the optional corresponding central directory entry
337  */
338 static int
339 unzipEntry(Biobuf *bin, ZipHead *czh)
340 {
341 	Dir *d;
342 	ZipHead zh;
343 	char *p;
344 	vlong off;
345 	int fd, isdir, ok, err;
346 
347 	zh.file = nil;
348 	if(setjmp(zjmp)){
349 		delfile = nil;
350 		free(zh.file);
351 		return -1;
352 	}
353 
354 	memset(&zh, 0, sizeof(zh));
355 	if(!header(bin, &zh))
356 		return 0;
357 
358 	ok = 1;
359 	isdir = 0;
360 
361 	fd = -1;
362 	if(wantFile(zh.file)){
363 		if(verbose)
364 			fprint(2, "extracting %s\n", zh.file);
365 
366 		if(czh != nil && czh->extos == ZDos){
367 			isdir = czh->eattr & ZDDir;
368 			if(isdir && zh.uncsize != 0)
369 				fprint(2, "unzip: ignoring directory data for %s\n", zh.file);
370 		}
371 		if(zh.meth == 0 && zh.uncsize == 0){
372 			p = strchr(zh.file, '\0');
373 			if(p > zh.file && p[-1] == '/')
374 				isdir = 1;
375 		}
376 
377 		if(stdout){
378 			if(ok && !isdir)
379 				fd = 1;
380 		}else if(isdir){
381 			fd = create(zh.file, OREAD, DMDIR | 0775);
382 			if(fd < 0){
383 				d = dirstat(zh.file);
384 				if(d == nil || (d->mode & DMDIR) != DMDIR){
385 					fprint(2, "unzip: can't create directory %s: %r\n", zh.file);
386 					ok = 0;
387 				}
388 				free(d);
389 			}
390 		}else if(ok){
391 			fd = create(zh.file, OWRITE, 0664);
392 			if(fd < 0){
393 				fprint(2, "unzip: can't create %s: %r\n", zh.file);
394 				ok = 0;
395 			}else
396 				delfile = zh.file;
397 		}
398 	}
399 
400 	wlen = 0;
401 	rlen = 0;
402 	crc = 0;
403 	wbad = 0;
404 
405 	if(zh.meth == 0){
406 		if(!copyout(fd, bin, zh.csize))
407 			error("copying data for %s failed: %r", zh.file);
408 	}else if(zh.meth == 8){
409 		off = Boffset(bin);
410 		err = inflate((void*)fd, crcwrite, bin, (int(*)(void*))Bgetc);
411 		if(err != FlateOk)
412 			error("inflate failed: %s", flateerr(err));
413 		rlen = Boffset(bin) - off;
414 	}else
415 		error("can't handle compression method %d for %s", zh.meth, zh.file);
416 
417 	trailer(bin, &zh);
418 
419 	if(zh.crc != crc)
420 		error("crc mismatch for %s", zh.file);
421 	if(zh.uncsize != wlen)
422 		error("output size mismatch for %s", zh.file);
423 	if(zh.csize != rlen)
424 		error("input size mismatch for %s", zh.file);
425 
426 	delfile = nil;
427 	free(zh.file);
428 
429 	if(fd >= 0 && !stdout){
430 		if(settimes){
431 			d = dirfstat(fd);
432 			if(d != nil){
433 				d->mtime = msdos2time(zh.modtime, zh.moddate);
434 				if(d->mtime)
435 					dirfwstat(fd, d);
436 			}
437 		}
438 		close(fd);
439 	}
440 
441 	return ok;
442 }
443 
444 static int
445 wantFile(char *file)
446 {
447 	int i, n;
448 
449 	if(nwant == 0)
450 		return 1;
451 	for(i = 0; i < nwant; i++){
452 		if(strcmp(want[i], file) == 0)
453 			return 1;
454 		n = strlen(want[i]);
455 		if(strncmp(want[i], file, n) == 0 && file[n] == '/')
456 			return 1;
457 	}
458 	return 0;
459 }
460 
461 /*
462  * find the start of the central directory
463  * returns the number of entries in the directory,
464  * or -1 if there was an error
465  */
466 static int
467 findCDir(Biobuf *bin, char *file)
468 {
469 	vlong ecoff;
470 	long off, size;
471 	int entries, zclen, dn, ds, de;
472 
473 	ecoff = Bseek(bin, -ZECHeadSize, 2);
474 	if(ecoff < 0){
475 		fprint(2, "unzip: can't seek to contents of %s; try adding -s\n", file);
476 		return -1;
477 	}
478 	if(setjmp(zjmp))
479 		return -1;
480 
481 	if(get4(bin) != ZECHeader){
482 		fprint(2, "unzip: bad magic number for contents of %s\n", file);
483 		return -1;
484 	}
485 	dn = get2(bin);
486 	ds = get2(bin);
487 	de = get2(bin);
488 	entries = get2(bin);
489 	size = get4(bin);
490 	off = get4(bin);
491 	zclen = get2(bin);
492 	while(zclen-- > 0)
493 		get1(bin);
494 
495 	if(verbose > 1){
496 		print("table starts at %ld for %ld bytes\n", off, size);
497 		if(ecoff - size != off)
498 			print("\ttable should start at %lld-%ld=%lld\n", ecoff, size, ecoff-size);
499 		if(dn || ds || de != entries)
500 			print("\tcurrent disk=%d start disk=%d table entries on this disk=%d\n", dn, ds, de);
501 	}
502 
503 	if(Bseek(bin, off, 0) != off){
504 		fprint(2, "unzip: can't seek to start of contents of %s\n", file);
505 		return -1;
506 	}
507 
508 	return entries;
509 }
510 
511 static int
512 cheader(Biobuf *bin, ZipHead *zh)
513 {
514 	ulong v;
515 	int flen, xlen, fclen;
516 
517 	v = get4(bin);
518 	if(v != ZCHeader){
519 		if(v == ZECHeader)
520 			return 0;
521 		error("bad magic number %lux", v);
522 	}
523 	zh->madevers = get1(bin);
524 	zh->madeos = get1(bin);
525 	zh->extvers = get1(bin);
526 	zh->extos = get1(bin);
527 	zh->flags = get2(bin);
528 	zh->meth = get2(bin);
529 	zh->modtime = get2(bin);
530 	zh->moddate = get2(bin);
531 	zh->crc = get4(bin);
532 	zh->csize = get4(bin);
533 	zh->uncsize = get4(bin);
534 	flen = get2(bin);
535 	xlen = get2(bin);
536 	fclen = get2(bin);
537 	get2(bin);		/* disk number start */
538 	zh->iattr = get2(bin);
539 	zh->eattr = get4(bin);
540 	zh->off = get4(bin);
541 
542 	zh->file = getname(bin, flen);
543 
544 	while(xlen-- > 0)
545 		get1(bin);
546 
547 	while(fclen-- > 0)
548 		get1(bin);
549 
550 	return 1;
551 }
552 
553 static int
554 header(Biobuf *bin, ZipHead *zh)
555 {
556 	ulong v;
557 	int flen, xlen;
558 
559 	v = get4(bin);
560 	if(v != ZHeader){
561 		if(v == ZCHeader)
562 			return 0;
563 		error("bad magic number %lux at %lld", v, Boffset(bin)-4);
564 	}
565 	zh->extvers = get1(bin);
566 	zh->extos = get1(bin);
567 	zh->flags = get2(bin);
568 	zh->meth = get2(bin);
569 	zh->modtime = get2(bin);
570 	zh->moddate = get2(bin);
571 	zh->crc = get4(bin);
572 	zh->csize = get4(bin);
573 	zh->uncsize = get4(bin);
574 	flen = get2(bin);
575 	xlen = get2(bin);
576 
577 	zh->file = getname(bin, flen);
578 
579 	while(xlen-- > 0)
580 		get1(bin);
581 
582 	return 1;
583 }
584 
585 static void
586 trailer(Biobuf *bin, ZipHead *zh)
587 {
588 	if(zh->flags & ZTrailInfo){
589 		zh->crc = get4(bin);
590 		zh->csize = get4(bin);
591 		zh->uncsize = get4(bin);
592 	}
593 }
594 
595 static char*
596 getname(Biobuf *bin, int len)
597 {
598 	char *s;
599 	int i, c;
600 
601 	s = emalloc(len + 1);
602 	for(i = 0; i < len; i++){
603 		c = get1(bin);
604 		if(lower)
605 			c = tolower(c);
606 		s[i] = c;
607 	}
608 	s[i] = '\0';
609 	return s;
610 }
611 
612 static int
613 crcwrite(void *out, void *buf, int n)
614 {
615 	int fd, nw;
616 
617 	wlen += n;
618 	crc = blockcrc(crctab, crc, buf, n);
619 	fd = (int)out;
620 	if(fd < 0)
621 		return n;
622 	nw = write(fd, buf, n);
623 	if(nw != n)
624 		wbad = 1;
625 	return nw;
626 }
627 
628 static int
629 copyout(int ofd, Biobuf *bin, long len)
630 {
631 	char buf[BufSize];
632 	int n;
633 
634 	for(; len > 0; len -= n){
635 		n = len;
636 		if(n > BufSize)
637 			n = BufSize;
638 		n = Bread(bin, buf, n);
639 		if(n <= 0)
640 			return 0;
641 		rlen += n;
642 		if(crcwrite((void*)ofd, buf, n) != n)
643 			return 0;
644 	}
645 	return 1;
646 }
647 
648 static ulong
649 get4(Biobuf *b)
650 {
651 	ulong v;
652 	int i, c;
653 
654 	v = 0;
655 	for(i = 0; i < 4; i++){
656 		c = Bgetc(b);
657 		if(c < 0)
658 			error("unexpected eof reading file information");
659 		v |= c << (i * 8);
660 	}
661 	return v;
662 }
663 
664 static int
665 get2(Biobuf *b)
666 {
667 	int i, c, v;
668 
669 	v = 0;
670 	for(i = 0; i < 2; i++){
671 		c = Bgetc(b);
672 		if(c < 0)
673 			error("unexpected eof reading file information");
674 		v |= c << (i * 8);
675 	}
676 	return v;
677 }
678 
679 static int
680 get1(Biobuf *b)
681 {
682 	int c;
683 
684 	c = Bgetc(b);
685 	if(c < 0)
686 		error("unexpected eof reading file information");
687 	return c;
688 }
689 
690 static long
691 msdos2time(int time, int date)
692 {
693 	Tm tm;
694 
695 	tm.hour = time >> 11;
696 	tm.min = (time >> 5) & 63;
697 	tm.sec = (time & 31) << 1;
698 	tm.year = 80 + (date >> 9);
699 	tm.mon = ((date >> 5) & 15) - 1;
700 	tm.mday = date & 31;
701 	tm.zone[0] = '\0';
702 	tm.yday = 0;
703 
704 	return tm2sec(&tm);
705 }
706 
707 static void*
708 emalloc(ulong n)
709 {
710 	void *p;
711 
712 	p = malloc(n);
713 	if(p == nil)
714 		sysfatal("out of memory");
715 	return p;
716 }
717 
718 static void
719 error(char *fmt, ...)
720 {
721 	va_list arg;
722 
723 	fprint(2, "unzip: ");
724 	va_start(arg, fmt);
725 	vfprint(2, fmt, arg);
726 	va_end(arg);
727 	fprint(2, "\n");
728 
729 	if(delfile != nil){
730 		fprint(2, "unzip: removing output file %s\n", delfile);
731 		remove(delfile);
732 		delfile = nil;
733 	}
734 
735 	longjmp(zjmp, 1);
736 }