gzip.c revision 273507
1/* $NetBSD: gzip.c,v 1.106 2014/10/18 08:33:30 snj Exp $ */ 2 3/*- 4 * Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green 5 * 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 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 21 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 22 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 23 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 24 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 26 * SUCH DAMAGE. 27 * 28 */ 29 30#include <sys/cdefs.h> 31#ifndef lint 32__COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006\ 33 Matthew R. Green. All rights reserved."); 34__FBSDID("$FreeBSD: head/usr.bin/gzip/gzip.c 273507 2014-10-23 01:22:29Z delphij $"); 35#endif /* not lint */ 36 37/* 38 * gzip.c -- GPL free gzip using zlib. 39 * 40 * RFC 1950 covers the zlib format 41 * RFC 1951 covers the deflate format 42 * RFC 1952 covers the gzip format 43 * 44 * TODO: 45 * - use mmap where possible 46 * - make bzip2/compress -v/-t/-l support work as well as possible 47 */ 48 49#include <sys/param.h> 50#include <sys/stat.h> 51#include <sys/time.h> 52 53#include <inttypes.h> 54#include <unistd.h> 55#include <stdio.h> 56#include <string.h> 57#include <stdlib.h> 58#include <err.h> 59#include <errno.h> 60#include <fcntl.h> 61#include <zlib.h> 62#include <fts.h> 63#include <libgen.h> 64#include <stdarg.h> 65#include <getopt.h> 66#include <time.h> 67 68/* what type of file are we dealing with */ 69enum filetype { 70 FT_GZIP, 71#ifndef NO_BZIP2_SUPPORT 72 FT_BZIP2, 73#endif 74#ifndef NO_COMPRESS_SUPPORT 75 FT_Z, 76#endif 77#ifndef NO_PACK_SUPPORT 78 FT_PACK, 79#endif 80#ifndef NO_XZ_SUPPORT 81 FT_XZ, 82#endif 83 FT_LAST, 84 FT_UNKNOWN 85}; 86 87#ifndef NO_BZIP2_SUPPORT 88#include <bzlib.h> 89 90#define BZ2_SUFFIX ".bz2" 91#define BZIP2_MAGIC "\102\132\150" 92#endif 93 94#ifndef NO_COMPRESS_SUPPORT 95#define Z_SUFFIX ".Z" 96#define Z_MAGIC "\037\235" 97#endif 98 99#ifndef NO_PACK_SUPPORT 100#define PACK_MAGIC "\037\036" 101#endif 102 103#ifndef NO_XZ_SUPPORT 104#include <lzma.h> 105#define XZ_SUFFIX ".xz" 106#define XZ_MAGIC "\3757zXZ" 107#endif 108 109#define GZ_SUFFIX ".gz" 110 111#define BUFLEN (64 * 1024) 112 113#define GZIP_MAGIC0 0x1F 114#define GZIP_MAGIC1 0x8B 115#define GZIP_OMAGIC1 0x9E 116 117#define GZIP_TIMESTAMP (off_t)4 118#define GZIP_ORIGNAME (off_t)10 119 120#define HEAD_CRC 0x02 121#define EXTRA_FIELD 0x04 122#define ORIG_NAME 0x08 123#define COMMENT 0x10 124 125#define OS_CODE 3 /* Unix */ 126 127typedef struct { 128 const char *zipped; 129 int ziplen; 130 const char *normal; /* for unzip - must not be longer than zipped */ 131} suffixes_t; 132static suffixes_t suffixes[] = { 133#define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 134 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 135#ifndef SMALL 136 SUFFIX(GZ_SUFFIX, ""), 137 SUFFIX(".z", ""), 138 SUFFIX("-gz", ""), 139 SUFFIX("-z", ""), 140 SUFFIX("_z", ""), 141 SUFFIX(".taz", ".tar"), 142 SUFFIX(".tgz", ".tar"), 143#ifndef NO_BZIP2_SUPPORT 144 SUFFIX(BZ2_SUFFIX, ""), 145 SUFFIX(".tbz", ".tar"), 146 SUFFIX(".tbz2", ".tar"), 147#endif 148#ifndef NO_COMPRESS_SUPPORT 149 SUFFIX(Z_SUFFIX, ""), 150#endif 151#ifndef NO_XZ_SUPPORT 152 SUFFIX(XZ_SUFFIX, ""), 153#endif 154 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 155#endif /* SMALL */ 156#undef SUFFIX 157}; 158#define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 159#define SUFFIX_MAXLEN 30 160 161static const char gzip_version[] = "FreeBSD gzip 20141022"; 162 163#ifndef SMALL 164static const char gzip_copyright[] = \ 165" Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green\n" 166" All rights reserved.\n" 167"\n" 168" Redistribution and use in source and binary forms, with or without\n" 169" modification, are permitted provided that the following conditions\n" 170" are met:\n" 171" 1. Redistributions of source code must retain the above copyright\n" 172" notice, this list of conditions and the following disclaimer.\n" 173" 2. Redistributions in binary form must reproduce the above copyright\n" 174" notice, this list of conditions and the following disclaimer in the\n" 175" documentation and/or other materials provided with the distribution.\n" 176"\n" 177" THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR\n" 178" IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES\n" 179" OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.\n" 180" IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,\n" 181" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,\n" 182" BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;\n" 183" LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED\n" 184" AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,\n" 185" OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY\n" 186" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF\n" 187" SUCH DAMAGE."; 188#endif 189 190static int cflag; /* stdout mode */ 191static int dflag; /* decompress mode */ 192static int lflag; /* list mode */ 193static int numflag = 6; /* gzip -1..-9 value */ 194 195#ifndef SMALL 196static int fflag; /* force mode */ 197static int kflag; /* don't delete input files */ 198static int nflag; /* don't save name/timestamp */ 199static int Nflag; /* don't restore name/timestamp */ 200static int qflag; /* quiet mode */ 201static int rflag; /* recursive mode */ 202static int tflag; /* test */ 203static int vflag; /* verbose mode */ 204static const char *remove_file = NULL; /* file to be removed upon SIGINT */ 205#else 206#define qflag 0 207#define tflag 0 208#endif 209 210static int exit_value = 0; /* exit value */ 211 212static char *infile; /* name of file coming in */ 213 214static void maybe_err(const char *fmt, ...) __printflike(1, 2) __dead2; 215#if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \ 216 !defined(NO_XZ_SUPPORT) 217static void maybe_errx(const char *fmt, ...) __printflike(1, 2) __dead2; 218#endif 219static void maybe_warn(const char *fmt, ...) __printflike(1, 2); 220static void maybe_warnx(const char *fmt, ...) __printflike(1, 2); 221static enum filetype file_gettype(u_char *); 222#ifdef SMALL 223#define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 224#endif 225static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 226static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 227static off_t file_compress(char *, char *, size_t); 228static off_t file_uncompress(char *, char *, size_t); 229static void handle_pathname(char *); 230static void handle_file(char *, struct stat *); 231static void handle_stdin(void); 232static void handle_stdout(void); 233static void print_ratio(off_t, off_t, FILE *); 234static void print_list(int fd, off_t, const char *, time_t); 235static void usage(void) __dead2; 236static void display_version(void) __dead2; 237#ifndef SMALL 238static void display_license(void); 239static void sigint_handler(int); 240#endif 241static const suffixes_t *check_suffix(char *, int); 242static ssize_t read_retry(int, void *, size_t); 243 244#ifdef SMALL 245#define unlink_input(f, sb) unlink(f) 246#else 247static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 248static void prepend_gzip(char *, int *, char ***); 249static void handle_dir(char *); 250static void print_verbage(const char *, const char *, off_t, off_t); 251static void print_test(const char *, int); 252static void copymodes(int fd, const struct stat *, const char *file); 253static int check_outfile(const char *outfile); 254#endif 255 256#ifndef NO_BZIP2_SUPPORT 257static off_t unbzip2(int, int, char *, size_t, off_t *); 258#endif 259 260#ifndef NO_COMPRESS_SUPPORT 261static FILE *zdopen(int); 262static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 263#endif 264 265#ifndef NO_PACK_SUPPORT 266static off_t unpack(int, int, char *, size_t, off_t *); 267#endif 268 269#ifndef NO_XZ_SUPPORT 270static off_t unxz(int, int, char *, size_t, off_t *); 271#endif 272 273#ifdef SMALL 274#define getopt_long(a,b,c,d,e) getopt(a,b,c) 275#else 276static const struct option longopts[] = { 277 { "stdout", no_argument, 0, 'c' }, 278 { "to-stdout", no_argument, 0, 'c' }, 279 { "decompress", no_argument, 0, 'd' }, 280 { "uncompress", no_argument, 0, 'd' }, 281 { "force", no_argument, 0, 'f' }, 282 { "help", no_argument, 0, 'h' }, 283 { "keep", no_argument, 0, 'k' }, 284 { "list", no_argument, 0, 'l' }, 285 { "no-name", no_argument, 0, 'n' }, 286 { "name", no_argument, 0, 'N' }, 287 { "quiet", no_argument, 0, 'q' }, 288 { "recursive", no_argument, 0, 'r' }, 289 { "suffix", required_argument, 0, 'S' }, 290 { "test", no_argument, 0, 't' }, 291 { "verbose", no_argument, 0, 'v' }, 292 { "version", no_argument, 0, 'V' }, 293 { "fast", no_argument, 0, '1' }, 294 { "best", no_argument, 0, '9' }, 295 { "ascii", no_argument, 0, 'a' }, 296 { "license", no_argument, 0, 'L' }, 297 { NULL, no_argument, 0, 0 }, 298}; 299#endif 300 301int 302main(int argc, char **argv) 303{ 304 const char *progname = getprogname(); 305#ifndef SMALL 306 char *gzip; 307 int len; 308#endif 309 int ch; 310 311#ifndef SMALL 312 if ((gzip = getenv("GZIP")) != NULL) 313 prepend_gzip(gzip, &argc, &argv); 314 signal(SIGINT, sigint_handler); 315#endif 316 317 /* 318 * XXX 319 * handle being called `gunzip', `zcat' and `gzcat' 320 */ 321 if (strcmp(progname, "gunzip") == 0) 322 dflag = 1; 323 else if (strcmp(progname, "zcat") == 0 || 324 strcmp(progname, "gzcat") == 0) 325 dflag = cflag = 1; 326 327#ifdef SMALL 328#define OPT_LIST "123456789cdhlV" 329#else 330#define OPT_LIST "123456789acdfhklLNnqrS:tVv" 331#endif 332 333 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 334 switch (ch) { 335 case '1': case '2': case '3': 336 case '4': case '5': case '6': 337 case '7': case '8': case '9': 338 numflag = ch - '0'; 339 break; 340 case 'c': 341 cflag = 1; 342 break; 343 case 'd': 344 dflag = 1; 345 break; 346 case 'l': 347 lflag = 1; 348 dflag = 1; 349 break; 350 case 'V': 351 display_version(); 352 /* NOTREACHED */ 353#ifndef SMALL 354 case 'a': 355 fprintf(stderr, "%s: option --ascii ignored on this system\n", progname); 356 break; 357 case 'f': 358 fflag = 1; 359 break; 360 case 'k': 361 kflag = 1; 362 break; 363 case 'L': 364 display_license(); 365 /* NOT REACHED */ 366 case 'N': 367 nflag = 0; 368 Nflag = 1; 369 break; 370 case 'n': 371 nflag = 1; 372 Nflag = 0; 373 break; 374 case 'q': 375 qflag = 1; 376 break; 377 case 'r': 378 rflag = 1; 379 break; 380 case 'S': 381 len = strlen(optarg); 382 if (len != 0) { 383 if (len > SUFFIX_MAXLEN) 384 errx(1, "incorrect suffix: '%s': too long", optarg); 385 suffixes[0].zipped = optarg; 386 suffixes[0].ziplen = len; 387 } else { 388 suffixes[NUM_SUFFIXES - 1].zipped = ""; 389 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 390 } 391 break; 392 case 't': 393 cflag = 1; 394 tflag = 1; 395 dflag = 1; 396 break; 397 case 'v': 398 vflag = 1; 399 break; 400#endif 401 default: 402 usage(); 403 /* NOTREACHED */ 404 } 405 } 406 argv += optind; 407 argc -= optind; 408 409 if (argc == 0) { 410 if (dflag) /* stdin mode */ 411 handle_stdin(); 412 else /* stdout mode */ 413 handle_stdout(); 414 } else { 415 do { 416 handle_pathname(argv[0]); 417 } while (*++argv); 418 } 419#ifndef SMALL 420 if (qflag == 0 && lflag && argc > 1) 421 print_list(-1, 0, "(totals)", 0); 422#endif 423 exit(exit_value); 424} 425 426/* maybe print a warning */ 427void 428maybe_warn(const char *fmt, ...) 429{ 430 va_list ap; 431 432 if (qflag == 0) { 433 va_start(ap, fmt); 434 vwarn(fmt, ap); 435 va_end(ap); 436 } 437 if (exit_value == 0) 438 exit_value = 1; 439} 440 441/* ... without an errno. */ 442void 443maybe_warnx(const char *fmt, ...) 444{ 445 va_list ap; 446 447 if (qflag == 0) { 448 va_start(ap, fmt); 449 vwarnx(fmt, ap); 450 va_end(ap); 451 } 452 if (exit_value == 0) 453 exit_value = 1; 454} 455 456/* maybe print an error */ 457void 458maybe_err(const char *fmt, ...) 459{ 460 va_list ap; 461 462 if (qflag == 0) { 463 va_start(ap, fmt); 464 vwarn(fmt, ap); 465 va_end(ap); 466 } 467 exit(2); 468} 469 470#if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \ 471 !defined(NO_XZ_SUPPORT) 472/* ... without an errno. */ 473void 474maybe_errx(const char *fmt, ...) 475{ 476 va_list ap; 477 478 if (qflag == 0) { 479 va_start(ap, fmt); 480 vwarnx(fmt, ap); 481 va_end(ap); 482 } 483 exit(2); 484} 485#endif 486 487#ifndef SMALL 488/* split up $GZIP and prepend it to the argument list */ 489static void 490prepend_gzip(char *gzip, int *argc, char ***argv) 491{ 492 char *s, **nargv, **ac; 493 int nenvarg = 0, i; 494 495 /* scan how many arguments there are */ 496 for (s = gzip;;) { 497 while (*s == ' ' || *s == '\t') 498 s++; 499 if (*s == 0) 500 goto count_done; 501 nenvarg++; 502 while (*s != ' ' && *s != '\t') 503 if (*s++ == 0) 504 goto count_done; 505 } 506count_done: 507 /* punt early */ 508 if (nenvarg == 0) 509 return; 510 511 *argc += nenvarg; 512 ac = *argv; 513 514 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 515 if (nargv == NULL) 516 maybe_err("malloc"); 517 518 /* stash this away */ 519 *argv = nargv; 520 521 /* copy the program name first */ 522 i = 0; 523 nargv[i++] = *(ac++); 524 525 /* take a copy of $GZIP and add it to the array */ 526 s = strdup(gzip); 527 if (s == NULL) 528 maybe_err("strdup"); 529 for (;;) { 530 /* Skip whitespaces. */ 531 while (*s == ' ' || *s == '\t') 532 s++; 533 if (*s == 0) 534 goto copy_done; 535 nargv[i++] = s; 536 /* Find the end of this argument. */ 537 while (*s != ' ' && *s != '\t') 538 if (*s++ == 0) 539 /* Argument followed by NUL. */ 540 goto copy_done; 541 /* Terminate by overwriting ' ' or '\t' with NUL. */ 542 *s++ = 0; 543 } 544copy_done: 545 546 /* copy the original arguments and a NULL */ 547 while (*ac) 548 nargv[i++] = *(ac++); 549 nargv[i] = NULL; 550} 551#endif 552 553/* compress input to output. Return bytes read, -1 on error */ 554static off_t 555gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 556{ 557 z_stream z; 558 char *outbufp, *inbufp; 559 off_t in_tot = 0, out_tot = 0; 560 ssize_t in_size; 561 int i, error; 562 uLong crc; 563#ifdef SMALL 564 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 565 0, 0, 0, 0, 566 0, OS_CODE }; 567#endif 568 569 outbufp = malloc(BUFLEN); 570 inbufp = malloc(BUFLEN); 571 if (outbufp == NULL || inbufp == NULL) { 572 maybe_err("malloc failed"); 573 goto out; 574 } 575 576 memset(&z, 0, sizeof z); 577 z.zalloc = Z_NULL; 578 z.zfree = Z_NULL; 579 z.opaque = 0; 580 581#ifdef SMALL 582 memcpy(outbufp, header, sizeof header); 583 i = sizeof header; 584#else 585 if (nflag != 0) { 586 mtime = 0; 587 origname = ""; 588 } 589 590 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 591 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 592 *origname ? ORIG_NAME : 0, 593 mtime & 0xff, 594 (mtime >> 8) & 0xff, 595 (mtime >> 16) & 0xff, 596 (mtime >> 24) & 0xff, 597 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 598 OS_CODE, origname); 599 if (i >= BUFLEN) 600 /* this need PATH_MAX > BUFLEN ... */ 601 maybe_err("snprintf"); 602 if (*origname) 603 i++; 604#endif 605 606 z.next_out = (unsigned char *)outbufp + i; 607 z.avail_out = BUFLEN - i; 608 609 error = deflateInit2(&z, numflag, Z_DEFLATED, 610 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 611 if (error != Z_OK) { 612 maybe_warnx("deflateInit2 failed"); 613 in_tot = -1; 614 goto out; 615 } 616 617 crc = crc32(0L, Z_NULL, 0); 618 for (;;) { 619 if (z.avail_out == 0) { 620 if (write(out, outbufp, BUFLEN) != BUFLEN) { 621 maybe_warn("write"); 622 out_tot = -1; 623 goto out; 624 } 625 626 out_tot += BUFLEN; 627 z.next_out = (unsigned char *)outbufp; 628 z.avail_out = BUFLEN; 629 } 630 631 if (z.avail_in == 0) { 632 in_size = read(in, inbufp, BUFLEN); 633 if (in_size < 0) { 634 maybe_warn("read"); 635 in_tot = -1; 636 goto out; 637 } 638 if (in_size == 0) 639 break; 640 641 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 642 in_tot += in_size; 643 z.next_in = (unsigned char *)inbufp; 644 z.avail_in = in_size; 645 } 646 647 error = deflate(&z, Z_NO_FLUSH); 648 if (error != Z_OK && error != Z_STREAM_END) { 649 maybe_warnx("deflate failed"); 650 in_tot = -1; 651 goto out; 652 } 653 } 654 655 /* clean up */ 656 for (;;) { 657 size_t len; 658 ssize_t w; 659 660 error = deflate(&z, Z_FINISH); 661 if (error != Z_OK && error != Z_STREAM_END) { 662 maybe_warnx("deflate failed"); 663 in_tot = -1; 664 goto out; 665 } 666 667 len = (char *)z.next_out - outbufp; 668 669 w = write(out, outbufp, len); 670 if (w == -1 || (size_t)w != len) { 671 maybe_warn("write"); 672 out_tot = -1; 673 goto out; 674 } 675 out_tot += len; 676 z.next_out = (unsigned char *)outbufp; 677 z.avail_out = BUFLEN; 678 679 if (error == Z_STREAM_END) 680 break; 681 } 682 683 if (deflateEnd(&z) != Z_OK) { 684 maybe_warnx("deflateEnd failed"); 685 in_tot = -1; 686 goto out; 687 } 688 689 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 690 (int)crc & 0xff, 691 (int)(crc >> 8) & 0xff, 692 (int)(crc >> 16) & 0xff, 693 (int)(crc >> 24) & 0xff, 694 (int)in_tot & 0xff, 695 (int)(in_tot >> 8) & 0xff, 696 (int)(in_tot >> 16) & 0xff, 697 (int)(in_tot >> 24) & 0xff); 698 if (i != 8) 699 maybe_err("snprintf"); 700 if (write(out, outbufp, i) != i) { 701 maybe_warn("write"); 702 in_tot = -1; 703 } else 704 out_tot += i; 705 706out: 707 if (inbufp != NULL) 708 free(inbufp); 709 if (outbufp != NULL) 710 free(outbufp); 711 if (gsizep) 712 *gsizep = out_tot; 713 return in_tot; 714} 715 716/* 717 * uncompress input to output then close the input. return the 718 * uncompressed size written, and put the compressed sized read 719 * into `*gsizep'. 720 */ 721static off_t 722gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 723 const char *filename) 724{ 725 z_stream z; 726 char *outbufp, *inbufp; 727 off_t out_tot = -1, in_tot = 0; 728 uint32_t out_sub_tot = 0; 729 enum { 730 GZSTATE_MAGIC0, 731 GZSTATE_MAGIC1, 732 GZSTATE_METHOD, 733 GZSTATE_FLAGS, 734 GZSTATE_SKIPPING, 735 GZSTATE_EXTRA, 736 GZSTATE_EXTRA2, 737 GZSTATE_EXTRA3, 738 GZSTATE_ORIGNAME, 739 GZSTATE_COMMENT, 740 GZSTATE_HEAD_CRC1, 741 GZSTATE_HEAD_CRC2, 742 GZSTATE_INIT, 743 GZSTATE_READ, 744 GZSTATE_CRC, 745 GZSTATE_LEN, 746 } state = GZSTATE_MAGIC0; 747 int flags = 0, skip_count = 0; 748 int error = Z_STREAM_ERROR, done_reading = 0; 749 uLong crc = 0; 750 ssize_t wr; 751 int needmore = 0; 752 753#define ADVANCE() { z.next_in++; z.avail_in--; } 754 755 if ((outbufp = malloc(BUFLEN)) == NULL) { 756 maybe_err("malloc failed"); 757 goto out2; 758 } 759 if ((inbufp = malloc(BUFLEN)) == NULL) { 760 maybe_err("malloc failed"); 761 goto out1; 762 } 763 764 memset(&z, 0, sizeof z); 765 z.avail_in = prelen; 766 z.next_in = (unsigned char *)pre; 767 z.avail_out = BUFLEN; 768 z.next_out = (unsigned char *)outbufp; 769 z.zalloc = NULL; 770 z.zfree = NULL; 771 z.opaque = 0; 772 773 in_tot = prelen; 774 out_tot = 0; 775 776 for (;;) { 777 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 778 ssize_t in_size; 779 780 if (z.avail_in > 0) { 781 memmove(inbufp, z.next_in, z.avail_in); 782 } 783 z.next_in = (unsigned char *)inbufp; 784 in_size = read(in, z.next_in + z.avail_in, 785 BUFLEN - z.avail_in); 786 787 if (in_size == -1) { 788 maybe_warn("failed to read stdin"); 789 goto stop_and_fail; 790 } else if (in_size == 0) { 791 done_reading = 1; 792 } 793 794 z.avail_in += in_size; 795 needmore = 0; 796 797 in_tot += in_size; 798 } 799 if (z.avail_in == 0) { 800 if (done_reading && state != GZSTATE_MAGIC0) { 801 maybe_warnx("%s: unexpected end of file", 802 filename); 803 goto stop_and_fail; 804 } 805 goto stop; 806 } 807 switch (state) { 808 case GZSTATE_MAGIC0: 809 if (*z.next_in != GZIP_MAGIC0) { 810 if (in_tot > 0) { 811 maybe_warnx("%s: trailing garbage " 812 "ignored", filename); 813 goto stop; 814 } 815 maybe_warnx("input not gziped (MAGIC0)"); 816 goto stop_and_fail; 817 } 818 ADVANCE(); 819 state++; 820 out_sub_tot = 0; 821 crc = crc32(0L, Z_NULL, 0); 822 break; 823 824 case GZSTATE_MAGIC1: 825 if (*z.next_in != GZIP_MAGIC1 && 826 *z.next_in != GZIP_OMAGIC1) { 827 maybe_warnx("input not gziped (MAGIC1)"); 828 goto stop_and_fail; 829 } 830 ADVANCE(); 831 state++; 832 break; 833 834 case GZSTATE_METHOD: 835 if (*z.next_in != Z_DEFLATED) { 836 maybe_warnx("unknown compression method"); 837 goto stop_and_fail; 838 } 839 ADVANCE(); 840 state++; 841 break; 842 843 case GZSTATE_FLAGS: 844 flags = *z.next_in; 845 ADVANCE(); 846 skip_count = 6; 847 state++; 848 break; 849 850 case GZSTATE_SKIPPING: 851 if (skip_count > 0) { 852 skip_count--; 853 ADVANCE(); 854 } else 855 state++; 856 break; 857 858 case GZSTATE_EXTRA: 859 if ((flags & EXTRA_FIELD) == 0) { 860 state = GZSTATE_ORIGNAME; 861 break; 862 } 863 skip_count = *z.next_in; 864 ADVANCE(); 865 state++; 866 break; 867 868 case GZSTATE_EXTRA2: 869 skip_count |= ((*z.next_in) << 8); 870 ADVANCE(); 871 state++; 872 break; 873 874 case GZSTATE_EXTRA3: 875 if (skip_count > 0) { 876 skip_count--; 877 ADVANCE(); 878 } else 879 state++; 880 break; 881 882 case GZSTATE_ORIGNAME: 883 if ((flags & ORIG_NAME) == 0) { 884 state++; 885 break; 886 } 887 if (*z.next_in == 0) 888 state++; 889 ADVANCE(); 890 break; 891 892 case GZSTATE_COMMENT: 893 if ((flags & COMMENT) == 0) { 894 state++; 895 break; 896 } 897 if (*z.next_in == 0) 898 state++; 899 ADVANCE(); 900 break; 901 902 case GZSTATE_HEAD_CRC1: 903 if (flags & HEAD_CRC) 904 skip_count = 2; 905 else 906 skip_count = 0; 907 state++; 908 break; 909 910 case GZSTATE_HEAD_CRC2: 911 if (skip_count > 0) { 912 skip_count--; 913 ADVANCE(); 914 } else 915 state++; 916 break; 917 918 case GZSTATE_INIT: 919 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 920 maybe_warnx("failed to inflateInit"); 921 goto stop_and_fail; 922 } 923 state++; 924 break; 925 926 case GZSTATE_READ: 927 error = inflate(&z, Z_FINISH); 928 switch (error) { 929 /* Z_BUF_ERROR goes with Z_FINISH... */ 930 case Z_BUF_ERROR: 931 if (z.avail_out > 0 && !done_reading) 932 continue; 933 934 case Z_STREAM_END: 935 case Z_OK: 936 break; 937 938 case Z_NEED_DICT: 939 maybe_warnx("Z_NEED_DICT error"); 940 goto stop_and_fail; 941 case Z_DATA_ERROR: 942 maybe_warnx("data stream error"); 943 goto stop_and_fail; 944 case Z_STREAM_ERROR: 945 maybe_warnx("internal stream error"); 946 goto stop_and_fail; 947 case Z_MEM_ERROR: 948 maybe_warnx("memory allocation error"); 949 goto stop_and_fail; 950 951 default: 952 maybe_warn("unknown error from inflate(): %d", 953 error); 954 } 955 wr = BUFLEN - z.avail_out; 956 957 if (wr != 0) { 958 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 959 if ( 960#ifndef SMALL 961 /* don't write anything with -t */ 962 tflag == 0 && 963#endif 964 write(out, outbufp, wr) != wr) { 965 maybe_warn("error writing to output"); 966 goto stop_and_fail; 967 } 968 969 out_tot += wr; 970 out_sub_tot += wr; 971 } 972 973 if (error == Z_STREAM_END) { 974 inflateEnd(&z); 975 state++; 976 } 977 978 z.next_out = (unsigned char *)outbufp; 979 z.avail_out = BUFLEN; 980 981 break; 982 case GZSTATE_CRC: 983 { 984 uLong origcrc; 985 986 if (z.avail_in < 4) { 987 if (!done_reading) { 988 needmore = 1; 989 continue; 990 } 991 maybe_warnx("truncated input"); 992 goto stop_and_fail; 993 } 994 origcrc = ((unsigned)z.next_in[0] & 0xff) | 995 ((unsigned)z.next_in[1] & 0xff) << 8 | 996 ((unsigned)z.next_in[2] & 0xff) << 16 | 997 ((unsigned)z.next_in[3] & 0xff) << 24; 998 if (origcrc != crc) { 999 maybe_warnx("invalid compressed" 1000 " data--crc error"); 1001 goto stop_and_fail; 1002 } 1003 } 1004 1005 z.avail_in -= 4; 1006 z.next_in += 4; 1007 1008 if (!z.avail_in && done_reading) { 1009 goto stop; 1010 } 1011 state++; 1012 break; 1013 case GZSTATE_LEN: 1014 { 1015 uLong origlen; 1016 1017 if (z.avail_in < 4) { 1018 if (!done_reading) { 1019 needmore = 1; 1020 continue; 1021 } 1022 maybe_warnx("truncated input"); 1023 goto stop_and_fail; 1024 } 1025 origlen = ((unsigned)z.next_in[0] & 0xff) | 1026 ((unsigned)z.next_in[1] & 0xff) << 8 | 1027 ((unsigned)z.next_in[2] & 0xff) << 16 | 1028 ((unsigned)z.next_in[3] & 0xff) << 24; 1029 1030 if (origlen != out_sub_tot) { 1031 maybe_warnx("invalid compressed" 1032 " data--length error"); 1033 goto stop_and_fail; 1034 } 1035 } 1036 1037 z.avail_in -= 4; 1038 z.next_in += 4; 1039 1040 if (error < 0) { 1041 maybe_warnx("decompression error"); 1042 goto stop_and_fail; 1043 } 1044 state = GZSTATE_MAGIC0; 1045 break; 1046 } 1047 continue; 1048stop_and_fail: 1049 out_tot = -1; 1050stop: 1051 break; 1052 } 1053 if (state > GZSTATE_INIT) 1054 inflateEnd(&z); 1055 1056 free(inbufp); 1057out1: 1058 free(outbufp); 1059out2: 1060 if (gsizep) 1061 *gsizep = in_tot; 1062 return (out_tot); 1063} 1064 1065#ifndef SMALL 1066/* 1067 * set the owner, mode, flags & utimes using the given file descriptor. 1068 * file is only used in possible warning messages. 1069 */ 1070static void 1071copymodes(int fd, const struct stat *sbp, const char *file) 1072{ 1073 struct timeval times[2]; 1074 struct stat sb; 1075 1076 /* 1077 * If we have no info on the input, give this file some 1078 * default values and return.. 1079 */ 1080 if (sbp == NULL) { 1081 mode_t mask = umask(022); 1082 1083 (void)fchmod(fd, DEFFILEMODE & ~mask); 1084 (void)umask(mask); 1085 return; 1086 } 1087 sb = *sbp; 1088 1089 /* if the chown fails, remove set-id bits as-per compress(1) */ 1090 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1091 if (errno != EPERM) 1092 maybe_warn("couldn't fchown: %s", file); 1093 sb.st_mode &= ~(S_ISUID|S_ISGID); 1094 } 1095 1096 /* we only allow set-id and the 9 normal permission bits */ 1097 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1098 if (fchmod(fd, sb.st_mode) < 0) 1099 maybe_warn("couldn't fchmod: %s", file); 1100 1101 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atim); 1102 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtim); 1103 if (futimes(fd, times) < 0) 1104 maybe_warn("couldn't utimes: %s", file); 1105 1106 /* only try flags if they exist already */ 1107 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1108 maybe_warn("couldn't fchflags: %s", file); 1109} 1110#endif 1111 1112/* what sort of file is this? */ 1113static enum filetype 1114file_gettype(u_char *buf) 1115{ 1116 1117 if (buf[0] == GZIP_MAGIC0 && 1118 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1119 return FT_GZIP; 1120 else 1121#ifndef NO_BZIP2_SUPPORT 1122 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1123 buf[3] >= '0' && buf[3] <= '9') 1124 return FT_BZIP2; 1125 else 1126#endif 1127#ifndef NO_COMPRESS_SUPPORT 1128 if (memcmp(buf, Z_MAGIC, 2) == 0) 1129 return FT_Z; 1130 else 1131#endif 1132#ifndef NO_PACK_SUPPORT 1133 if (memcmp(buf, PACK_MAGIC, 2) == 0) 1134 return FT_PACK; 1135 else 1136#endif 1137#ifndef NO_XZ_SUPPORT 1138 if (memcmp(buf, XZ_MAGIC, 4) == 0) /* XXX: We only have 4 bytes */ 1139 return FT_XZ; 1140 else 1141#endif 1142 return FT_UNKNOWN; 1143} 1144 1145#ifndef SMALL 1146/* check the outfile is OK. */ 1147static int 1148check_outfile(const char *outfile) 1149{ 1150 struct stat sb; 1151 int ok = 1; 1152 1153 if (lflag == 0 && stat(outfile, &sb) == 0) { 1154 if (fflag) 1155 unlink(outfile); 1156 else if (isatty(STDIN_FILENO)) { 1157 char ans[10] = { 'n', '\0' }; /* default */ 1158 1159 fprintf(stderr, "%s already exists -- do you wish to " 1160 "overwrite (y or n)? " , outfile); 1161 (void)fgets(ans, sizeof(ans) - 1, stdin); 1162 if (ans[0] != 'y' && ans[0] != 'Y') { 1163 fprintf(stderr, "\tnot overwriting\n"); 1164 ok = 0; 1165 } else 1166 unlink(outfile); 1167 } else { 1168 maybe_warnx("%s already exists -- skipping", outfile); 1169 ok = 0; 1170 } 1171 } 1172 return ok; 1173} 1174 1175static void 1176unlink_input(const char *file, const struct stat *sb) 1177{ 1178 struct stat nsb; 1179 1180 if (kflag) 1181 return; 1182 if (stat(file, &nsb) != 0) 1183 /* Must be gone already */ 1184 return; 1185 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1186 /* Definitely a different file */ 1187 return; 1188 unlink(file); 1189} 1190 1191static void 1192sigint_handler(int signo __unused) 1193{ 1194 1195 if (remove_file != NULL) 1196 unlink(remove_file); 1197 _exit(2); 1198} 1199#endif 1200 1201static const suffixes_t * 1202check_suffix(char *file, int xlate) 1203{ 1204 const suffixes_t *s; 1205 int len = strlen(file); 1206 char *sp; 1207 1208 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1209 /* if it doesn't fit in "a.suf", don't bother */ 1210 if (s->ziplen >= len) 1211 continue; 1212 sp = file + len - s->ziplen; 1213 if (strcmp(s->zipped, sp) != 0) 1214 continue; 1215 if (xlate) 1216 strcpy(sp, s->normal); 1217 return s; 1218 } 1219 return NULL; 1220} 1221 1222/* 1223 * compress the given file: create a corresponding .gz file and remove the 1224 * original. 1225 */ 1226static off_t 1227file_compress(char *file, char *outfile, size_t outsize) 1228{ 1229 int in; 1230 int out; 1231 off_t size, insize; 1232#ifndef SMALL 1233 struct stat isb, osb; 1234 const suffixes_t *suff; 1235#endif 1236 1237 in = open(file, O_RDONLY); 1238 if (in == -1) { 1239 maybe_warn("can't open %s", file); 1240 return (-1); 1241 } 1242 1243#ifndef SMALL 1244 if (fstat(in, &isb) != 0) { 1245 maybe_warn("couldn't stat: %s", file); 1246 close(in); 1247 return (-1); 1248 } 1249#endif 1250 1251 if (cflag == 0) { 1252#ifndef SMALL 1253 if (isb.st_nlink > 1 && fflag == 0) { 1254 maybe_warnx("%s has %d other link%s -- skipping", 1255 file, isb.st_nlink - 1, 1256 (isb.st_nlink - 1) == 1 ? "" : "s"); 1257 close(in); 1258 return (-1); 1259 } 1260 1261 if (fflag == 0 && (suff = check_suffix(file, 0)) && 1262 suff->zipped[0] != 0) { 1263 maybe_warnx("%s already has %s suffix -- unchanged", 1264 file, suff->zipped); 1265 close(in); 1266 return (-1); 1267 } 1268#endif 1269 1270 /* Add (usually) .gz to filename */ 1271 if ((size_t)snprintf(outfile, outsize, "%s%s", 1272 file, suffixes[0].zipped) >= outsize) 1273 memcpy(outfile + outsize - suffixes[0].ziplen - 1, 1274 suffixes[0].zipped, suffixes[0].ziplen + 1); 1275 1276#ifndef SMALL 1277 if (check_outfile(outfile) == 0) { 1278 close(in); 1279 return (-1); 1280 } 1281#endif 1282 } 1283 1284 if (cflag == 0) { 1285 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1286 if (out == -1) { 1287 maybe_warn("could not create output: %s", outfile); 1288 fclose(stdin); 1289 return (-1); 1290 } 1291#ifndef SMALL 1292 remove_file = outfile; 1293#endif 1294 } else 1295 out = STDOUT_FILENO; 1296 1297 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1298 1299 (void)close(in); 1300 1301 /* 1302 * If there was an error, insize will be -1. 1303 * If we compressed to stdout, just return the size. 1304 * Otherwise stat the file and check it is the correct size. 1305 * We only blow away the file if we can stat the output and it 1306 * has the expected size. 1307 */ 1308 if (cflag != 0) 1309 return (insize == -1 ? -1 : size); 1310 1311#ifndef SMALL 1312 if (fstat(out, &osb) != 0) { 1313 maybe_warn("couldn't stat: %s", outfile); 1314 goto bad_outfile; 1315 } 1316 1317 if (osb.st_size != size) { 1318 maybe_warnx("output file: %s wrong size (%ju != %ju), deleting", 1319 outfile, (uintmax_t)osb.st_size, (uintmax_t)size); 1320 goto bad_outfile; 1321 } 1322 1323 copymodes(out, &isb, outfile); 1324 remove_file = NULL; 1325#endif 1326 if (close(out) == -1) 1327 maybe_warn("couldn't close output"); 1328 1329 /* output is good, ok to delete input */ 1330 unlink_input(file, &isb); 1331 return (size); 1332 1333#ifndef SMALL 1334 bad_outfile: 1335 if (close(out) == -1) 1336 maybe_warn("couldn't close output"); 1337 1338 maybe_warnx("leaving original %s", file); 1339 unlink(outfile); 1340 return (size); 1341#endif 1342} 1343 1344/* uncompress the given file and remove the original */ 1345static off_t 1346file_uncompress(char *file, char *outfile, size_t outsize) 1347{ 1348 struct stat isb, osb; 1349 off_t size; 1350 ssize_t rbytes; 1351 unsigned char header1[4]; 1352 enum filetype method; 1353 int fd, ofd, zfd = -1; 1354#ifndef SMALL 1355 ssize_t rv; 1356 time_t timestamp = 0; 1357 unsigned char name[PATH_MAX + 1]; 1358#endif 1359 1360 /* gather the old name info */ 1361 1362 fd = open(file, O_RDONLY); 1363 if (fd < 0) { 1364 maybe_warn("can't open %s", file); 1365 goto lose; 1366 } 1367 1368 strlcpy(outfile, file, outsize); 1369 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1370 maybe_warnx("%s: unknown suffix -- ignored", file); 1371 goto lose; 1372 } 1373 1374 rbytes = read(fd, header1, sizeof header1); 1375 if (rbytes != sizeof header1) { 1376 /* we don't want to fail here. */ 1377#ifndef SMALL 1378 if (fflag) 1379 goto lose; 1380#endif 1381 if (rbytes == -1) 1382 maybe_warn("can't read %s", file); 1383 else 1384 goto unexpected_EOF; 1385 goto lose; 1386 } 1387 1388 method = file_gettype(header1); 1389#ifndef SMALL 1390 if (fflag == 0 && method == FT_UNKNOWN) { 1391 maybe_warnx("%s: not in gzip format", file); 1392 goto lose; 1393 } 1394 1395#endif 1396 1397#ifndef SMALL 1398 if (method == FT_GZIP && Nflag) { 1399 unsigned char ts[4]; /* timestamp */ 1400 1401 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1402 if (rv >= 0 && rv < (ssize_t)(sizeof ts)) 1403 goto unexpected_EOF; 1404 if (rv == -1) { 1405 if (!fflag) 1406 maybe_warn("can't read %s", file); 1407 goto lose; 1408 } 1409 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1410 1411 if (header1[3] & ORIG_NAME) { 1412 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1413 if (rbytes < 0) { 1414 maybe_warn("can't read %s", file); 1415 goto lose; 1416 } 1417 if (name[0] != 0) { 1418 /* preserve original directory name */ 1419 char *dp = strrchr(file, '/'); 1420 if (dp == NULL) 1421 dp = file; 1422 else 1423 dp++; 1424 snprintf(outfile, outsize, "%.*s%.*s", 1425 (int) (dp - file), 1426 file, (int) rbytes, name); 1427 } 1428 } 1429 } 1430#endif 1431 lseek(fd, 0, SEEK_SET); 1432 1433 if (cflag == 0 || lflag) { 1434 if (fstat(fd, &isb) != 0) 1435 goto lose; 1436#ifndef SMALL 1437 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1438 maybe_warnx("%s has %d other links -- skipping", 1439 file, isb.st_nlink - 1); 1440 goto lose; 1441 } 1442 if (nflag == 0 && timestamp) 1443 isb.st_mtime = timestamp; 1444 if (check_outfile(outfile) == 0) 1445 goto lose; 1446#endif 1447 } 1448 1449 if (cflag == 0 && lflag == 0) { 1450 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1451 if (zfd == STDOUT_FILENO) { 1452 /* We won't close STDOUT_FILENO later... */ 1453 zfd = dup(zfd); 1454 close(STDOUT_FILENO); 1455 } 1456 if (zfd == -1) { 1457 maybe_warn("can't open %s", outfile); 1458 goto lose; 1459 } 1460#ifndef SMALL 1461 remove_file = outfile; 1462#endif 1463 } else 1464 zfd = STDOUT_FILENO; 1465 1466 switch (method) { 1467#ifndef NO_BZIP2_SUPPORT 1468 case FT_BZIP2: 1469 /* XXX */ 1470 if (lflag) { 1471 maybe_warnx("no -l with bzip2 files"); 1472 goto lose; 1473 } 1474 1475 size = unbzip2(fd, zfd, NULL, 0, NULL); 1476 break; 1477#endif 1478 1479#ifndef NO_COMPRESS_SUPPORT 1480 case FT_Z: { 1481 FILE *in, *out; 1482 1483 /* XXX */ 1484 if (lflag) { 1485 maybe_warnx("no -l with Lempel-Ziv files"); 1486 goto lose; 1487 } 1488 1489 if ((in = zdopen(fd)) == NULL) { 1490 maybe_warn("zdopen for read: %s", file); 1491 goto lose; 1492 } 1493 1494 out = fdopen(dup(zfd), "w"); 1495 if (out == NULL) { 1496 maybe_warn("fdopen for write: %s", outfile); 1497 fclose(in); 1498 goto lose; 1499 } 1500 1501 size = zuncompress(in, out, NULL, 0, NULL); 1502 /* need to fclose() if ferror() is true... */ 1503 if (ferror(in) | fclose(in)) { 1504 maybe_warn("failed infile fclose"); 1505 unlink(outfile); 1506 (void)fclose(out); 1507 } 1508 if (fclose(out) != 0) { 1509 maybe_warn("failed outfile fclose"); 1510 unlink(outfile); 1511 goto lose; 1512 } 1513 break; 1514 } 1515#endif 1516 1517#ifndef NO_PACK_SUPPORT 1518 case FT_PACK: 1519 if (lflag) { 1520 maybe_warnx("no -l with packed files"); 1521 goto lose; 1522 } 1523 1524 size = unpack(fd, zfd, NULL, 0, NULL); 1525 break; 1526#endif 1527 1528#ifndef NO_XZ_SUPPORT 1529 case FT_XZ: 1530 if (lflag) { 1531 maybe_warnx("no -l with xz files"); 1532 goto lose; 1533 } 1534 1535 size = unxz(fd, zfd, NULL, 0, NULL); 1536 break; 1537#endif 1538 1539#ifndef SMALL 1540 case FT_UNKNOWN: 1541 if (lflag) { 1542 maybe_warnx("no -l for unknown filetypes"); 1543 goto lose; 1544 } 1545 size = cat_fd(NULL, 0, NULL, fd); 1546 break; 1547#endif 1548 default: 1549 if (lflag) { 1550 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1551 close(fd); 1552 return -1; /* XXX */ 1553 } 1554 1555 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1556 break; 1557 } 1558 1559 if (close(fd) != 0) 1560 maybe_warn("couldn't close input"); 1561 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1562 maybe_warn("couldn't close output"); 1563 1564 if (size == -1) { 1565 if (cflag == 0) 1566 unlink(outfile); 1567 maybe_warnx("%s: uncompress failed", file); 1568 return -1; 1569 } 1570 1571 /* if testing, or we uncompressed to stdout, this is all we need */ 1572#ifndef SMALL 1573 if (tflag) 1574 return size; 1575#endif 1576 /* if we are uncompressing to stdin, don't remove the file. */ 1577 if (cflag) 1578 return size; 1579 1580 /* 1581 * if we create a file... 1582 */ 1583 /* 1584 * if we can't stat the file don't remove the file. 1585 */ 1586 1587 ofd = open(outfile, O_RDWR, 0); 1588 if (ofd == -1) { 1589 maybe_warn("couldn't open (leaving original): %s", 1590 outfile); 1591 return -1; 1592 } 1593 if (fstat(ofd, &osb) != 0) { 1594 maybe_warn("couldn't stat (leaving original): %s", 1595 outfile); 1596 close(ofd); 1597 return -1; 1598 } 1599 if (osb.st_size != size) { 1600 maybe_warnx("stat gave different size: %ju != %ju (leaving original)", 1601 (uintmax_t)size, (uintmax_t)osb.st_size); 1602 close(ofd); 1603 unlink(outfile); 1604 return -1; 1605 } 1606#ifndef SMALL 1607 copymodes(ofd, &isb, outfile); 1608 remove_file = NULL; 1609#endif 1610 close(ofd); 1611 unlink_input(file, &isb); 1612 return size; 1613 1614 unexpected_EOF: 1615 maybe_warnx("%s: unexpected end of file", file); 1616 lose: 1617 if (fd != -1) 1618 close(fd); 1619 if (zfd != -1 && zfd != STDOUT_FILENO) 1620 close(fd); 1621 return -1; 1622} 1623 1624#ifndef SMALL 1625static off_t 1626cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1627{ 1628 char buf[BUFLEN]; 1629 off_t in_tot; 1630 ssize_t w; 1631 1632 in_tot = count; 1633 w = write(STDOUT_FILENO, prepend, count); 1634 if (w == -1 || (size_t)w != count) { 1635 maybe_warn("write to stdout"); 1636 return -1; 1637 } 1638 for (;;) { 1639 ssize_t rv; 1640 1641 rv = read(fd, buf, sizeof buf); 1642 if (rv == 0) 1643 break; 1644 if (rv < 0) { 1645 maybe_warn("read from fd %d", fd); 1646 break; 1647 } 1648 1649 if (write(STDOUT_FILENO, buf, rv) != rv) { 1650 maybe_warn("write to stdout"); 1651 break; 1652 } 1653 in_tot += rv; 1654 } 1655 1656 if (gsizep) 1657 *gsizep = in_tot; 1658 return (in_tot); 1659} 1660#endif 1661 1662static void 1663handle_stdin(void) 1664{ 1665 unsigned char header1[4]; 1666 off_t usize, gsize; 1667 enum filetype method; 1668 ssize_t bytes_read; 1669#ifndef NO_COMPRESS_SUPPORT 1670 FILE *in; 1671#endif 1672 1673#ifndef SMALL 1674 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1675 maybe_warnx("standard input is a terminal -- ignoring"); 1676 return; 1677 } 1678#endif 1679 1680 if (lflag) { 1681 struct stat isb; 1682 1683 /* XXX could read the whole file, etc. */ 1684 if (fstat(STDIN_FILENO, &isb) < 0) { 1685 maybe_warn("fstat"); 1686 return; 1687 } 1688 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1689 return; 1690 } 1691 1692 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1693 if (bytes_read == -1) { 1694 maybe_warn("can't read stdin"); 1695 return; 1696 } else if (bytes_read != sizeof(header1)) { 1697 maybe_warnx("(stdin): unexpected end of file"); 1698 return; 1699 } 1700 1701 method = file_gettype(header1); 1702 switch (method) { 1703 default: 1704#ifndef SMALL 1705 if (fflag == 0) { 1706 maybe_warnx("unknown compression format"); 1707 return; 1708 } 1709 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1710 break; 1711#endif 1712 case FT_GZIP: 1713 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1714 (char *)header1, sizeof header1, &gsize, "(stdin)"); 1715 break; 1716#ifndef NO_BZIP2_SUPPORT 1717 case FT_BZIP2: 1718 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1719 (char *)header1, sizeof header1, &gsize); 1720 break; 1721#endif 1722#ifndef NO_COMPRESS_SUPPORT 1723 case FT_Z: 1724 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1725 maybe_warnx("zopen of stdin"); 1726 return; 1727 } 1728 1729 usize = zuncompress(in, stdout, (char *)header1, 1730 sizeof header1, &gsize); 1731 fclose(in); 1732 break; 1733#endif 1734#ifndef NO_PACK_SUPPORT 1735 case FT_PACK: 1736 usize = unpack(STDIN_FILENO, STDOUT_FILENO, 1737 (char *)header1, sizeof header1, &gsize); 1738 break; 1739#endif 1740#ifndef NO_XZ_SUPPORT 1741 case FT_XZ: 1742 usize = unxz(STDIN_FILENO, STDOUT_FILENO, 1743 (char *)header1, sizeof header1, &gsize); 1744 break; 1745#endif 1746 } 1747 1748#ifndef SMALL 1749 if (vflag && !tflag && usize != -1 && gsize != -1) 1750 print_verbage(NULL, NULL, usize, gsize); 1751 if (vflag && tflag) 1752 print_test("(stdin)", usize != -1); 1753#endif 1754 1755} 1756 1757static void 1758handle_stdout(void) 1759{ 1760 off_t gsize, usize; 1761 struct stat sb; 1762 time_t systime; 1763 uint32_t mtime; 1764 int ret; 1765 1766#ifndef SMALL 1767 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1768 maybe_warnx("standard output is a terminal -- ignoring"); 1769 return; 1770 } 1771#endif 1772 /* If stdin is a file use its mtime, otherwise use current time */ 1773 ret = fstat(STDIN_FILENO, &sb); 1774 1775#ifndef SMALL 1776 if (ret < 0) { 1777 maybe_warn("Can't stat stdin"); 1778 return; 1779 } 1780#endif 1781 1782 if (S_ISREG(sb.st_mode)) 1783 mtime = (uint32_t)sb.st_mtime; 1784 else { 1785 systime = time(NULL); 1786#ifndef SMALL 1787 if (systime == -1) { 1788 maybe_warn("time"); 1789 return; 1790 } 1791#endif 1792 mtime = (uint32_t)systime; 1793 } 1794 1795 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1796#ifndef SMALL 1797 if (vflag && !tflag && usize != -1 && gsize != -1) 1798 print_verbage(NULL, NULL, usize, gsize); 1799#endif 1800} 1801 1802/* do what is asked for, for the path name */ 1803static void 1804handle_pathname(char *path) 1805{ 1806 char *opath = path, *s = NULL; 1807 ssize_t len; 1808 int slen; 1809 struct stat sb; 1810 1811 /* check for stdout/stdin */ 1812 if (path[0] == '-' && path[1] == '\0') { 1813 if (dflag) 1814 handle_stdin(); 1815 else 1816 handle_stdout(); 1817 return; 1818 } 1819 1820retry: 1821 if (stat(path, &sb) != 0 || (fflag == 0 && cflag == 0 && 1822 lstat(path, &sb) != 0)) { 1823 /* lets try <path>.gz if we're decompressing */ 1824 if (dflag && s == NULL && errno == ENOENT) { 1825 len = strlen(path); 1826 slen = suffixes[0].ziplen; 1827 s = malloc(len + slen + 1); 1828 if (s == NULL) 1829 maybe_err("malloc"); 1830 memcpy(s, path, len); 1831 memcpy(s + len, suffixes[0].zipped, slen + 1); 1832 path = s; 1833 goto retry; 1834 } 1835 maybe_warn("can't stat: %s", opath); 1836 goto out; 1837 } 1838 1839 if (S_ISDIR(sb.st_mode)) { 1840#ifndef SMALL 1841 if (rflag) 1842 handle_dir(path); 1843 else 1844#endif 1845 maybe_warnx("%s is a directory", path); 1846 goto out; 1847 } 1848 1849 if (S_ISREG(sb.st_mode)) 1850 handle_file(path, &sb); 1851 else 1852 maybe_warnx("%s is not a regular file", path); 1853 1854out: 1855 if (s) 1856 free(s); 1857} 1858 1859/* compress/decompress a file */ 1860static void 1861handle_file(char *file, struct stat *sbp) 1862{ 1863 off_t usize, gsize; 1864 char outfile[PATH_MAX]; 1865 1866 infile = file; 1867 if (dflag) { 1868 usize = file_uncompress(file, outfile, sizeof(outfile)); 1869#ifndef SMALL 1870 if (vflag && tflag) 1871 print_test(file, usize != -1); 1872#endif 1873 if (usize == -1) 1874 return; 1875 gsize = sbp->st_size; 1876 } else { 1877 gsize = file_compress(file, outfile, sizeof(outfile)); 1878 if (gsize == -1) 1879 return; 1880 usize = sbp->st_size; 1881 } 1882 1883 1884#ifndef SMALL 1885 if (vflag && !tflag) 1886 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1887#endif 1888} 1889 1890#ifndef SMALL 1891/* this is used with -r to recursively descend directories */ 1892static void 1893handle_dir(char *dir) 1894{ 1895 char *path_argv[2]; 1896 FTS *fts; 1897 FTSENT *entry; 1898 1899 path_argv[0] = dir; 1900 path_argv[1] = 0; 1901 fts = fts_open(path_argv, FTS_PHYSICAL | FTS_NOCHDIR, NULL); 1902 if (fts == NULL) { 1903 warn("couldn't fts_open %s", dir); 1904 return; 1905 } 1906 1907 while ((entry = fts_read(fts))) { 1908 switch(entry->fts_info) { 1909 case FTS_D: 1910 case FTS_DP: 1911 continue; 1912 1913 case FTS_DNR: 1914 case FTS_ERR: 1915 case FTS_NS: 1916 maybe_warn("%s", entry->fts_path); 1917 continue; 1918 case FTS_F: 1919 handle_file(entry->fts_path, entry->fts_statp); 1920 } 1921 } 1922 (void)fts_close(fts); 1923} 1924#endif 1925 1926/* print a ratio - size reduction as a fraction of uncompressed size */ 1927static void 1928print_ratio(off_t in, off_t out, FILE *where) 1929{ 1930 int percent10; /* 10 * percent */ 1931 off_t diff; 1932 char buff[8]; 1933 int len; 1934 1935 diff = in - out/2; 1936 if (diff <= 0) 1937 /* 1938 * Output is more than double size of input! print -99.9% 1939 * Quite possibly we've failed to get the original size. 1940 */ 1941 percent10 = -999; 1942 else { 1943 /* 1944 * We only need 12 bits of result from the final division, 1945 * so reduce the values until a 32bit division will suffice. 1946 */ 1947 while (in > 0x100000) { 1948 diff >>= 1; 1949 in >>= 1; 1950 } 1951 if (in != 0) 1952 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1953 else 1954 percent10 = 0; 1955 } 1956 1957 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1958 /* Move the '.' to before the last digit */ 1959 buff[len - 1] = buff[len - 2]; 1960 buff[len - 2] = '.'; 1961 fprintf(where, "%5s%%", buff); 1962} 1963 1964#ifndef SMALL 1965/* print compression statistics, and the new name (if there is one!) */ 1966static void 1967print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1968{ 1969 if (file) 1970 fprintf(stderr, "%s:%s ", file, 1971 strlen(file) < 7 ? "\t\t" : "\t"); 1972 print_ratio(usize, gsize, stderr); 1973 if (nfile) 1974 fprintf(stderr, " -- replaced with %s", nfile); 1975 fprintf(stderr, "\n"); 1976 fflush(stderr); 1977} 1978 1979/* print test results */ 1980static void 1981print_test(const char *file, int ok) 1982{ 1983 1984 if (exit_value == 0 && ok == 0) 1985 exit_value = 1; 1986 fprintf(stderr, "%s:%s %s\n", file, 1987 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1988 fflush(stderr); 1989} 1990#endif 1991 1992/* print a file's info ala --list */ 1993/* eg: 1994 compressed uncompressed ratio uncompressed_name 1995 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1996*/ 1997static void 1998print_list(int fd, off_t out, const char *outfile, time_t ts) 1999{ 2000 static int first = 1; 2001#ifndef SMALL 2002 static off_t in_tot, out_tot; 2003 uint32_t crc = 0; 2004#endif 2005 off_t in = 0, rv; 2006 2007 if (first) { 2008#ifndef SMALL 2009 if (vflag) 2010 printf("method crc date time "); 2011#endif 2012 if (qflag == 0) 2013 printf(" compressed uncompressed " 2014 "ratio uncompressed_name\n"); 2015 } 2016 first = 0; 2017 2018 /* print totals? */ 2019#ifndef SMALL 2020 if (fd == -1) { 2021 in = in_tot; 2022 out = out_tot; 2023 } else 2024#endif 2025 { 2026 /* read the last 4 bytes - this is the uncompressed size */ 2027 rv = lseek(fd, (off_t)(-8), SEEK_END); 2028 if (rv != -1) { 2029 unsigned char buf[8]; 2030 uint32_t usize; 2031 2032 rv = read(fd, (char *)buf, sizeof(buf)); 2033 if (rv == -1) 2034 maybe_warn("read of uncompressed size"); 2035 else if (rv != sizeof(buf)) 2036 maybe_warnx("read of uncompressed size"); 2037 2038 else { 2039 usize = buf[4] | buf[5] << 8 | 2040 buf[6] << 16 | buf[7] << 24; 2041 in = (off_t)usize; 2042#ifndef SMALL 2043 crc = buf[0] | buf[1] << 8 | 2044 buf[2] << 16 | buf[3] << 24; 2045#endif 2046 } 2047 } 2048 } 2049 2050#ifndef SMALL 2051 if (vflag && fd == -1) 2052 printf(" "); 2053 else if (vflag) { 2054 char *date = ctime(&ts); 2055 2056 /* skip the day, 1/100th second, and year */ 2057 date += 4; 2058 date[12] = 0; 2059 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 2060 } 2061 in_tot += in; 2062 out_tot += out; 2063#else 2064 (void)&ts; /* XXX */ 2065#endif 2066 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 2067 print_ratio(in, out, stdout); 2068 printf(" %s\n", outfile); 2069} 2070 2071/* display the usage of NetBSD gzip */ 2072static void 2073usage(void) 2074{ 2075 2076 fprintf(stderr, "%s\n", gzip_version); 2077 fprintf(stderr, 2078#ifdef SMALL 2079 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n", 2080#else 2081 "usage: %s [-123456789acdfhklLNnqrtVv] [-S .suffix] [<file> [<file> ...]]\n" 2082 " -1 --fast fastest (worst) compression\n" 2083 " -2 .. -8 set compression level\n" 2084 " -9 --best best (slowest) compression\n" 2085 " -c --stdout write to stdout, keep original files\n" 2086 " --to-stdout\n" 2087 " -d --decompress uncompress files\n" 2088 " --uncompress\n" 2089 " -f --force force overwriting & compress links\n" 2090 " -h --help display this help\n" 2091 " -k --keep don't delete input files during operation\n" 2092 " -l --list list compressed file contents\n" 2093 " -N --name save or restore original file name and time stamp\n" 2094 " -n --no-name don't save original file name or time stamp\n" 2095 " -q --quiet output no warnings\n" 2096 " -r --recursive recursively compress files in directories\n" 2097 " -S .suf use suffix .suf instead of .gz\n" 2098 " --suffix .suf\n" 2099 " -t --test test compressed file\n" 2100 " -V --version display program version\n" 2101 " -v --verbose print extra statistics\n", 2102#endif 2103 getprogname()); 2104 exit(0); 2105} 2106 2107#ifndef SMALL 2108/* display the license information of FreeBSD gzip */ 2109static void 2110display_license(void) 2111{ 2112 2113 fprintf(stderr, "%s (based on NetBSD gzip 20141018)\n", gzip_version); 2114 fprintf(stderr, "%s\n", gzip_copyright); 2115 exit(0); 2116} 2117#endif 2118 2119/* display the version of NetBSD gzip */ 2120static void 2121display_version(void) 2122{ 2123 2124 fprintf(stderr, "%s\n", gzip_version); 2125 exit(0); 2126} 2127 2128#ifndef NO_BZIP2_SUPPORT 2129#include "unbzip2.c" 2130#endif 2131#ifndef NO_COMPRESS_SUPPORT 2132#include "zuncompress.c" 2133#endif 2134#ifndef NO_PACK_SUPPORT 2135#include "unpack.c" 2136#endif 2137#ifndef NO_XZ_SUPPORT 2138#include "unxz.c" 2139#endif 2140 2141static ssize_t 2142read_retry(int fd, void *buf, size_t sz) 2143{ 2144 char *cp = buf; 2145 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2146 2147 while (left > 0) { 2148 ssize_t ret; 2149 2150 ret = read(fd, cp, left); 2151 if (ret == -1) { 2152 return ret; 2153 } else if (ret == 0) { 2154 break; /* EOF */ 2155 } 2156 cp += ret; 2157 left -= ret; 2158 } 2159 2160 return sz - left; 2161} 2162