gzip.c revision 1.109
1 /* $NetBSD: gzip.c,v 1.109 2015/10/27 07:36:18 mrg 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 #include <sys/cdefs.h> 30 #ifndef lint 31 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006\ 32 Matthew R. Green. All rights reserved."); 33 __RCSID("$NetBSD: gzip.c,v 1.109 2015/10/27 07:36:18 mrg Exp $"); 34 #endif /* not lint */ 35 36 /* 37 * gzip.c -- GPL free gzip using zlib. 38 * 39 * RFC 1950 covers the zlib format 40 * RFC 1951 covers the deflate format 41 * RFC 1952 covers the gzip format 42 * 43 * TODO: 44 * - use mmap where possible 45 * - handle some signals better (remove outfile?) 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 #ifndef PRIdOFF 69 #define PRIdOFF PRId64 70 #endif 71 72 /* what type of file are we dealing with */ 73 enum filetype { 74 FT_GZIP, 75 #ifndef NO_BZIP2_SUPPORT 76 FT_BZIP2, 77 #endif 78 #ifndef NO_COMPRESS_SUPPORT 79 FT_Z, 80 #endif 81 #ifndef NO_PACK_SUPPORT 82 FT_PACK, 83 #endif 84 #ifndef NO_XZ_SUPPORT 85 FT_XZ, 86 #endif 87 FT_LAST, 88 FT_UNKNOWN 89 }; 90 91 #ifndef NO_BZIP2_SUPPORT 92 #include <bzlib.h> 93 94 #define BZ2_SUFFIX ".bz2" 95 #define BZIP2_MAGIC "\102\132\150" 96 #endif 97 98 #ifndef NO_COMPRESS_SUPPORT 99 #define Z_SUFFIX ".Z" 100 #define Z_MAGIC "\037\235" 101 #endif 102 103 #ifndef NO_PACK_SUPPORT 104 #define PACK_MAGIC "\037\036" 105 #endif 106 107 #ifndef NO_XZ_SUPPORT 108 #include <lzma.h> 109 #define XZ_SUFFIX ".xz" 110 #define XZ_MAGIC "\3757zXZ" 111 #endif 112 113 #define GZ_SUFFIX ".gz" 114 115 #define BUFLEN (64 * 1024) 116 117 #define GZIP_MAGIC0 0x1F 118 #define GZIP_MAGIC1 0x8B 119 #define GZIP_OMAGIC1 0x9E 120 121 #define GZIP_TIMESTAMP (off_t)4 122 #define GZIP_ORIGNAME (off_t)10 123 124 #define HEAD_CRC 0x02 125 #define EXTRA_FIELD 0x04 126 #define ORIG_NAME 0x08 127 #define COMMENT 0x10 128 129 #define OS_CODE 3 /* Unix */ 130 131 typedef struct { 132 const char *zipped; 133 int ziplen; 134 const char *normal; /* for unzip - must not be longer than zipped */ 135 } suffixes_t; 136 static suffixes_t suffixes[] = { 137 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 138 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 139 #ifndef SMALL 140 SUFFIX(GZ_SUFFIX, ""), 141 SUFFIX(".z", ""), 142 SUFFIX("-gz", ""), 143 SUFFIX("-z", ""), 144 SUFFIX("_z", ""), 145 SUFFIX(".taz", ".tar"), 146 SUFFIX(".tgz", ".tar"), 147 #ifndef NO_BZIP2_SUPPORT 148 SUFFIX(BZ2_SUFFIX, ""), 149 #endif 150 #ifndef NO_COMPRESS_SUPPORT 151 SUFFIX(Z_SUFFIX, ""), 152 #endif 153 #ifndef NO_XZ_SUPPORT 154 SUFFIX(XZ_SUFFIX, ""), 155 #endif 156 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 157 #endif /* SMALL */ 158 #undef SUFFIX 159 }; 160 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 161 #define SUFFIX_MAXLEN 30 162 163 static const char gzip_version[] = "NetBSD gzip 20150113"; 164 165 static int cflag; /* stdout mode */ 166 static int dflag; /* decompress mode */ 167 static int lflag; /* list mode */ 168 static int numflag = 6; /* gzip -1..-9 value */ 169 170 #ifndef SMALL 171 static int fflag; /* force mode */ 172 static int kflag; /* don't delete input files */ 173 static int nflag; /* don't save name/timestamp */ 174 static int Nflag; /* don't restore name/timestamp */ 175 static int qflag; /* quiet mode */ 176 static int rflag; /* recursive mode */ 177 static int tflag; /* test */ 178 static int vflag; /* verbose mode */ 179 #else 180 #define qflag 0 181 #define tflag 0 182 #endif 183 184 static int exit_value = 0; /* exit value */ 185 186 static char *infile; /* name of file coming in */ 187 188 static void maybe_err(const char *fmt, ...) __printflike(1, 2) __dead; 189 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \ 190 !defined(NO_XZ_SUPPORT) 191 static void maybe_errx(const char *fmt, ...) __printflike(1, 2) __dead; 192 #endif 193 static void maybe_warn(const char *fmt, ...) __printflike(1, 2); 194 static void maybe_warnx(const char *fmt, ...) __printflike(1, 2); 195 static enum filetype file_gettype(u_char *); 196 #ifdef SMALL 197 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 198 #endif 199 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 200 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 201 static off_t file_compress(char *, char *, size_t); 202 static off_t file_uncompress(char *, char *, size_t); 203 static void handle_pathname(char *); 204 static void handle_file(char *, struct stat *); 205 static void handle_stdin(void); 206 static void handle_stdout(void); 207 static void print_ratio(off_t, off_t, FILE *); 208 static void print_list(int fd, off_t, const char *, time_t); 209 __dead static void usage(void); 210 __dead static void display_version(void); 211 static const suffixes_t *check_suffix(char *, int); 212 static ssize_t read_retry(int, void *, size_t); 213 214 #ifdef SMALL 215 #define unlink_input(f, sb) unlink(f) 216 #else 217 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 218 static void prepend_gzip(char *, int *, char ***); 219 static void handle_dir(char *); 220 static void print_verbage(const char *, const char *, off_t, off_t); 221 static void print_test(const char *, int); 222 static void copymodes(int fd, const struct stat *, const char *file); 223 static int check_outfile(const char *outfile); 224 #endif 225 226 #ifndef NO_BZIP2_SUPPORT 227 static off_t unbzip2(int, int, char *, size_t, off_t *); 228 #endif 229 230 #ifndef NO_COMPRESS_SUPPORT 231 static FILE *zdopen(int); 232 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 233 #endif 234 235 #ifndef NO_PACK_SUPPORT 236 static off_t unpack(int, int, char *, size_t, off_t *); 237 #endif 238 239 #ifndef NO_XZ_SUPPORT 240 static off_t unxz(int, int, char *, size_t, off_t *); 241 #endif 242 243 #ifdef SMALL 244 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 245 #else 246 static const struct option longopts[] = { 247 { "stdout", no_argument, 0, 'c' }, 248 { "to-stdout", no_argument, 0, 'c' }, 249 { "decompress", no_argument, 0, 'd' }, 250 { "uncompress", no_argument, 0, 'd' }, 251 { "force", no_argument, 0, 'f' }, 252 { "help", no_argument, 0, 'h' }, 253 { "keep", no_argument, 0, 'k' }, 254 { "list", no_argument, 0, 'l' }, 255 { "no-name", no_argument, 0, 'n' }, 256 { "name", no_argument, 0, 'N' }, 257 { "quiet", no_argument, 0, 'q' }, 258 { "recursive", no_argument, 0, 'r' }, 259 { "suffix", required_argument, 0, 'S' }, 260 { "test", no_argument, 0, 't' }, 261 { "verbose", no_argument, 0, 'v' }, 262 { "version", no_argument, 0, 'V' }, 263 { "fast", no_argument, 0, '1' }, 264 { "best", no_argument, 0, '9' }, 265 #if 0 266 /* 267 * This is what else GNU gzip implements. --ascii isn't useful 268 * on NetBSD, and I don't care to have a --license. 269 */ 270 { "ascii", no_argument, 0, 'a' }, 271 { "license", no_argument, 0, 'L' }, 272 #endif 273 { NULL, no_argument, 0, 0 }, 274 }; 275 #endif 276 277 int 278 main(int argc, char **argv) 279 { 280 const char *progname = getprogname(); 281 #ifndef SMALL 282 char *gzip; 283 int len; 284 #endif 285 int ch; 286 287 /* XXX set up signals */ 288 289 #ifndef SMALL 290 if ((gzip = getenv("GZIP")) != NULL) 291 prepend_gzip(gzip, &argc, &argv); 292 #endif 293 294 /* 295 * XXX 296 * handle being called `gunzip', `zcat' and `gzcat' 297 */ 298 if (strcmp(progname, "gunzip") == 0) 299 dflag = 1; 300 else if (strcmp(progname, "zcat") == 0 || 301 strcmp(progname, "gzcat") == 0) 302 dflag = cflag = 1; 303 304 #ifdef SMALL 305 #define OPT_LIST "123456789cdhlV" 306 #else 307 #define OPT_LIST "123456789cdfhklNnqrS:tVv" 308 #endif 309 310 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 311 switch (ch) { 312 case '1': case '2': case '3': 313 case '4': case '5': case '6': 314 case '7': case '8': case '9': 315 numflag = ch - '0'; 316 break; 317 case 'c': 318 cflag = 1; 319 break; 320 case 'd': 321 dflag = 1; 322 break; 323 case 'l': 324 lflag = 1; 325 dflag = 1; 326 break; 327 case 'V': 328 display_version(); 329 /* NOTREACHED */ 330 #ifndef SMALL 331 case 'f': 332 fflag = 1; 333 break; 334 case 'k': 335 kflag = 1; 336 break; 337 case 'N': 338 nflag = 0; 339 Nflag = 1; 340 break; 341 case 'n': 342 nflag = 1; 343 Nflag = 0; 344 break; 345 case 'q': 346 qflag = 1; 347 break; 348 case 'r': 349 rflag = 1; 350 break; 351 case 'S': 352 len = strlen(optarg); 353 if (len != 0) { 354 if (len > SUFFIX_MAXLEN) 355 errx(1, "incorrect suffix: '%s'", optarg); 356 suffixes[0].zipped = optarg; 357 suffixes[0].ziplen = len; 358 } else { 359 suffixes[NUM_SUFFIXES - 1].zipped = ""; 360 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 361 } 362 break; 363 case 't': 364 cflag = 1; 365 tflag = 1; 366 dflag = 1; 367 break; 368 case 'v': 369 vflag = 1; 370 break; 371 #endif 372 default: 373 usage(); 374 /* NOTREACHED */ 375 } 376 } 377 argv += optind; 378 argc -= optind; 379 380 if (argc == 0) { 381 if (dflag) /* stdin mode */ 382 handle_stdin(); 383 else /* stdout mode */ 384 handle_stdout(); 385 } else { 386 do { 387 handle_pathname(argv[0]); 388 } while (*++argv); 389 } 390 #ifndef SMALL 391 if (qflag == 0 && lflag && argc > 1) 392 print_list(-1, 0, "(totals)", 0); 393 #endif 394 exit(exit_value); 395 } 396 397 /* maybe print a warning */ 398 void 399 maybe_warn(const char *fmt, ...) 400 { 401 va_list ap; 402 403 if (qflag == 0) { 404 va_start(ap, fmt); 405 vwarn(fmt, ap); 406 va_end(ap); 407 } 408 if (exit_value == 0) 409 exit_value = 1; 410 } 411 412 /* ... without an errno. */ 413 void 414 maybe_warnx(const char *fmt, ...) 415 { 416 va_list ap; 417 418 if (qflag == 0) { 419 va_start(ap, fmt); 420 vwarnx(fmt, ap); 421 va_end(ap); 422 } 423 if (exit_value == 0) 424 exit_value = 1; 425 } 426 427 /* maybe print an error */ 428 void 429 maybe_err(const char *fmt, ...) 430 { 431 va_list ap; 432 433 if (qflag == 0) { 434 va_start(ap, fmt); 435 vwarn(fmt, ap); 436 va_end(ap); 437 } 438 exit(2); 439 } 440 441 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \ 442 !defined(NO_XZ_SUPPORT) 443 /* ... without an errno. */ 444 void 445 maybe_errx(const char *fmt, ...) 446 { 447 va_list ap; 448 449 if (qflag == 0) { 450 va_start(ap, fmt); 451 vwarnx(fmt, ap); 452 va_end(ap); 453 } 454 exit(2); 455 } 456 #endif 457 458 #ifndef SMALL 459 /* split up $GZIP and prepend it to the argument list */ 460 static void 461 prepend_gzip(char *gzip, int *argc, char ***argv) 462 { 463 char *s, **nargv, **ac; 464 int nenvarg = 0, i; 465 466 /* scan how many arguments there are */ 467 for (s = gzip;;) { 468 while (*s == ' ' || *s == '\t') 469 s++; 470 if (*s == 0) 471 goto count_done; 472 nenvarg++; 473 while (*s != ' ' && *s != '\t') 474 if (*s++ == 0) 475 goto count_done; 476 } 477 count_done: 478 /* punt early */ 479 if (nenvarg == 0) 480 return; 481 482 *argc += nenvarg; 483 ac = *argv; 484 485 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 486 if (nargv == NULL) 487 maybe_err("malloc"); 488 489 /* stash this away */ 490 *argv = nargv; 491 492 /* copy the program name first */ 493 i = 0; 494 nargv[i++] = *(ac++); 495 496 /* take a copy of $GZIP and add it to the array */ 497 s = strdup(gzip); 498 if (s == NULL) 499 maybe_err("strdup"); 500 for (;;) { 501 /* Skip whitespaces. */ 502 while (*s == ' ' || *s == '\t') 503 s++; 504 if (*s == 0) 505 goto copy_done; 506 nargv[i++] = s; 507 /* Find the end of this argument. */ 508 while (*s != ' ' && *s != '\t') 509 if (*s++ == 0) 510 /* Argument followed by NUL. */ 511 goto copy_done; 512 /* Terminate by overwriting ' ' or '\t' with NUL. */ 513 *s++ = 0; 514 } 515 copy_done: 516 517 /* copy the original arguments and a NULL */ 518 while (*ac) 519 nargv[i++] = *(ac++); 520 nargv[i] = NULL; 521 } 522 #endif 523 524 /* compress input to output. Return bytes read, -1 on error */ 525 static off_t 526 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 527 { 528 z_stream z; 529 char *outbufp, *inbufp; 530 off_t in_tot = 0, out_tot = 0; 531 ssize_t in_size; 532 int i, error; 533 uLong crc; 534 #ifdef SMALL 535 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 536 0, 0, 0, 0, 537 0, OS_CODE }; 538 #endif 539 540 outbufp = malloc(BUFLEN); 541 inbufp = malloc(BUFLEN); 542 if (outbufp == NULL || inbufp == NULL) { 543 maybe_err("malloc failed"); 544 goto out; 545 } 546 547 memset(&z, 0, sizeof z); 548 z.zalloc = Z_NULL; 549 z.zfree = Z_NULL; 550 z.opaque = 0; 551 552 #ifdef SMALL 553 memcpy(outbufp, header, sizeof header); 554 i = sizeof header; 555 #else 556 if (nflag != 0) { 557 mtime = 0; 558 origname = ""; 559 } 560 561 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 562 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 563 *origname ? ORIG_NAME : 0, 564 mtime & 0xff, 565 (mtime >> 8) & 0xff, 566 (mtime >> 16) & 0xff, 567 (mtime >> 24) & 0xff, 568 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 569 OS_CODE, origname); 570 if (i >= BUFLEN) 571 /* this need PATH_MAX > BUFLEN ... */ 572 maybe_err("snprintf"); 573 if (*origname) 574 i++; 575 #endif 576 577 z.next_out = (unsigned char *)outbufp + i; 578 z.avail_out = BUFLEN - i; 579 580 error = deflateInit2(&z, numflag, Z_DEFLATED, 581 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 582 if (error != Z_OK) { 583 maybe_warnx("deflateInit2 failed"); 584 in_tot = -1; 585 goto out; 586 } 587 588 crc = crc32(0L, Z_NULL, 0); 589 for (;;) { 590 if (z.avail_out == 0) { 591 if (write(out, outbufp, BUFLEN) != BUFLEN) { 592 maybe_warn("write"); 593 out_tot = -1; 594 goto out; 595 } 596 597 out_tot += BUFLEN; 598 z.next_out = (unsigned char *)outbufp; 599 z.avail_out = BUFLEN; 600 } 601 602 if (z.avail_in == 0) { 603 in_size = read(in, inbufp, BUFLEN); 604 if (in_size < 0) { 605 maybe_warn("read"); 606 in_tot = -1; 607 goto out; 608 } 609 if (in_size == 0) 610 break; 611 612 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 613 in_tot += in_size; 614 z.next_in = (unsigned char *)inbufp; 615 z.avail_in = in_size; 616 } 617 618 error = deflate(&z, Z_NO_FLUSH); 619 if (error != Z_OK && error != Z_STREAM_END) { 620 maybe_warnx("deflate failed"); 621 in_tot = -1; 622 goto out; 623 } 624 } 625 626 /* clean up */ 627 for (;;) { 628 size_t len; 629 ssize_t w; 630 631 error = deflate(&z, Z_FINISH); 632 if (error != Z_OK && error != Z_STREAM_END) { 633 maybe_warnx("deflate failed"); 634 in_tot = -1; 635 goto out; 636 } 637 638 len = (char *)z.next_out - outbufp; 639 640 w = write(out, outbufp, len); 641 if (w == -1 || (size_t)w != len) { 642 maybe_warn("write"); 643 out_tot = -1; 644 goto out; 645 } 646 out_tot += len; 647 z.next_out = (unsigned char *)outbufp; 648 z.avail_out = BUFLEN; 649 650 if (error == Z_STREAM_END) 651 break; 652 } 653 654 if (deflateEnd(&z) != Z_OK) { 655 maybe_warnx("deflateEnd failed"); 656 in_tot = -1; 657 goto out; 658 } 659 660 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 661 (int)crc & 0xff, 662 (int)(crc >> 8) & 0xff, 663 (int)(crc >> 16) & 0xff, 664 (int)(crc >> 24) & 0xff, 665 (int)in_tot & 0xff, 666 (int)(in_tot >> 8) & 0xff, 667 (int)(in_tot >> 16) & 0xff, 668 (int)(in_tot >> 24) & 0xff); 669 if (i != 8) 670 maybe_err("snprintf"); 671 #if 0 672 if (in_tot > 0xffffffff) 673 maybe_warn("input file size >= 4GB cannot be saved"); 674 #endif 675 if (write(out, outbufp, i) != i) { 676 maybe_warn("write"); 677 in_tot = -1; 678 } else 679 out_tot += i; 680 681 out: 682 if (inbufp != NULL) 683 free(inbufp); 684 if (outbufp != NULL) 685 free(outbufp); 686 if (gsizep) 687 *gsizep = out_tot; 688 return in_tot; 689 } 690 691 /* 692 * uncompress input to output then close the input. return the 693 * uncompressed size written, and put the compressed sized read 694 * into `*gsizep'. 695 */ 696 static off_t 697 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 698 const char *filename) 699 { 700 z_stream z; 701 char *outbufp, *inbufp; 702 off_t out_tot = -1, in_tot = 0; 703 uint32_t out_sub_tot = 0; 704 enum { 705 GZSTATE_MAGIC0, 706 GZSTATE_MAGIC1, 707 GZSTATE_METHOD, 708 GZSTATE_FLAGS, 709 GZSTATE_SKIPPING, 710 GZSTATE_EXTRA, 711 GZSTATE_EXTRA2, 712 GZSTATE_EXTRA3, 713 GZSTATE_ORIGNAME, 714 GZSTATE_COMMENT, 715 GZSTATE_HEAD_CRC1, 716 GZSTATE_HEAD_CRC2, 717 GZSTATE_INIT, 718 GZSTATE_READ, 719 GZSTATE_CRC, 720 GZSTATE_LEN, 721 } state = GZSTATE_MAGIC0; 722 int flags = 0, skip_count = 0; 723 int error = Z_STREAM_ERROR, done_reading = 0; 724 uLong crc = 0; 725 ssize_t wr; 726 int needmore = 0; 727 728 #define ADVANCE() { z.next_in++; z.avail_in--; } 729 730 if ((outbufp = malloc(BUFLEN)) == NULL) { 731 maybe_err("malloc failed"); 732 goto out2; 733 } 734 if ((inbufp = malloc(BUFLEN)) == NULL) { 735 maybe_err("malloc failed"); 736 goto out1; 737 } 738 739 memset(&z, 0, sizeof z); 740 z.avail_in = prelen; 741 z.next_in = (unsigned char *)pre; 742 z.avail_out = BUFLEN; 743 z.next_out = (unsigned char *)outbufp; 744 z.zalloc = NULL; 745 z.zfree = NULL; 746 z.opaque = 0; 747 748 in_tot = prelen; 749 out_tot = 0; 750 751 for (;;) { 752 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 753 ssize_t in_size; 754 755 if (z.avail_in > 0) { 756 memmove(inbufp, z.next_in, z.avail_in); 757 } 758 z.next_in = (unsigned char *)inbufp; 759 in_size = read(in, z.next_in + z.avail_in, 760 BUFLEN - z.avail_in); 761 762 if (in_size == -1) { 763 maybe_warn("failed to read stdin"); 764 goto stop_and_fail; 765 } else if (in_size == 0) { 766 done_reading = 1; 767 } 768 769 z.avail_in += in_size; 770 needmore = 0; 771 772 in_tot += in_size; 773 } 774 if (z.avail_in == 0) { 775 if (done_reading && state != GZSTATE_MAGIC0) { 776 maybe_warnx("%s: unexpected end of file", 777 filename); 778 goto stop_and_fail; 779 } 780 goto stop; 781 } 782 switch (state) { 783 case GZSTATE_MAGIC0: 784 if (*z.next_in != GZIP_MAGIC0) { 785 if (in_tot > 0) { 786 maybe_warnx("%s: trailing garbage " 787 "ignored", filename); 788 goto stop; 789 } 790 maybe_warnx("input not gziped (MAGIC0)"); 791 exit_value = 2; 792 goto stop_and_fail; 793 } 794 ADVANCE(); 795 state++; 796 out_sub_tot = 0; 797 crc = crc32(0L, Z_NULL, 0); 798 break; 799 800 case GZSTATE_MAGIC1: 801 if (*z.next_in != GZIP_MAGIC1 && 802 *z.next_in != GZIP_OMAGIC1) { 803 maybe_warnx("input not gziped (MAGIC1)"); 804 goto stop_and_fail; 805 } 806 ADVANCE(); 807 state++; 808 break; 809 810 case GZSTATE_METHOD: 811 if (*z.next_in != Z_DEFLATED) { 812 maybe_warnx("unknown compression method"); 813 goto stop_and_fail; 814 } 815 ADVANCE(); 816 state++; 817 break; 818 819 case GZSTATE_FLAGS: 820 flags = *z.next_in; 821 ADVANCE(); 822 skip_count = 6; 823 state++; 824 break; 825 826 case GZSTATE_SKIPPING: 827 if (skip_count > 0) { 828 skip_count--; 829 ADVANCE(); 830 } else 831 state++; 832 break; 833 834 case GZSTATE_EXTRA: 835 if ((flags & EXTRA_FIELD) == 0) { 836 state = GZSTATE_ORIGNAME; 837 break; 838 } 839 skip_count = *z.next_in; 840 ADVANCE(); 841 state++; 842 break; 843 844 case GZSTATE_EXTRA2: 845 skip_count |= ((*z.next_in) << 8); 846 ADVANCE(); 847 state++; 848 break; 849 850 case GZSTATE_EXTRA3: 851 if (skip_count > 0) { 852 skip_count--; 853 ADVANCE(); 854 } else 855 state++; 856 break; 857 858 case GZSTATE_ORIGNAME: 859 if ((flags & ORIG_NAME) == 0) { 860 state++; 861 break; 862 } 863 if (*z.next_in == 0) 864 state++; 865 ADVANCE(); 866 break; 867 868 case GZSTATE_COMMENT: 869 if ((flags & COMMENT) == 0) { 870 state++; 871 break; 872 } 873 if (*z.next_in == 0) 874 state++; 875 ADVANCE(); 876 break; 877 878 case GZSTATE_HEAD_CRC1: 879 if (flags & HEAD_CRC) 880 skip_count = 2; 881 else 882 skip_count = 0; 883 state++; 884 break; 885 886 case GZSTATE_HEAD_CRC2: 887 if (skip_count > 0) { 888 skip_count--; 889 ADVANCE(); 890 } else 891 state++; 892 break; 893 894 case GZSTATE_INIT: 895 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 896 maybe_warnx("failed to inflateInit"); 897 goto stop_and_fail; 898 } 899 state++; 900 break; 901 902 case GZSTATE_READ: 903 error = inflate(&z, Z_FINISH); 904 switch (error) { 905 /* Z_BUF_ERROR goes with Z_FINISH... */ 906 case Z_BUF_ERROR: 907 if (z.avail_out > 0 && !done_reading) 908 continue; 909 910 case Z_STREAM_END: 911 case Z_OK: 912 break; 913 914 case Z_NEED_DICT: 915 maybe_warnx("Z_NEED_DICT error"); 916 goto stop_and_fail; 917 case Z_DATA_ERROR: 918 maybe_warnx("data stream error"); 919 goto stop_and_fail; 920 case Z_STREAM_ERROR: 921 maybe_warnx("internal stream error"); 922 goto stop_and_fail; 923 case Z_MEM_ERROR: 924 maybe_warnx("memory allocation error"); 925 goto stop_and_fail; 926 927 default: 928 maybe_warn("unknown error from inflate(): %d", 929 error); 930 } 931 wr = BUFLEN - z.avail_out; 932 933 if (wr != 0) { 934 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 935 if ( 936 #ifndef SMALL 937 /* don't write anything with -t */ 938 tflag == 0 && 939 #endif 940 write(out, outbufp, wr) != wr) { 941 maybe_warn("error writing to output"); 942 goto stop_and_fail; 943 } 944 945 out_tot += wr; 946 out_sub_tot += wr; 947 } 948 949 if (error == Z_STREAM_END) { 950 inflateEnd(&z); 951 state++; 952 } 953 954 z.next_out = (unsigned char *)outbufp; 955 z.avail_out = BUFLEN; 956 957 break; 958 case GZSTATE_CRC: 959 { 960 uLong origcrc; 961 962 if (z.avail_in < 4) { 963 if (!done_reading) { 964 needmore = 1; 965 continue; 966 } 967 maybe_warnx("truncated input"); 968 goto stop_and_fail; 969 } 970 origcrc = ((unsigned)z.next_in[0] & 0xff) | 971 ((unsigned)z.next_in[1] & 0xff) << 8 | 972 ((unsigned)z.next_in[2] & 0xff) << 16 | 973 ((unsigned)z.next_in[3] & 0xff) << 24; 974 if (origcrc != crc) { 975 maybe_warnx("invalid compressed" 976 " data--crc error"); 977 goto stop_and_fail; 978 } 979 } 980 981 z.avail_in -= 4; 982 z.next_in += 4; 983 984 if (!z.avail_in && done_reading) { 985 goto stop; 986 } 987 state++; 988 break; 989 case GZSTATE_LEN: 990 { 991 uLong origlen; 992 993 if (z.avail_in < 4) { 994 if (!done_reading) { 995 needmore = 1; 996 continue; 997 } 998 maybe_warnx("truncated input"); 999 goto stop_and_fail; 1000 } 1001 origlen = ((unsigned)z.next_in[0] & 0xff) | 1002 ((unsigned)z.next_in[1] & 0xff) << 8 | 1003 ((unsigned)z.next_in[2] & 0xff) << 16 | 1004 ((unsigned)z.next_in[3] & 0xff) << 24; 1005 1006 if (origlen != out_sub_tot) { 1007 maybe_warnx("invalid compressed" 1008 " data--length error"); 1009 goto stop_and_fail; 1010 } 1011 } 1012 1013 z.avail_in -= 4; 1014 z.next_in += 4; 1015 1016 if (error < 0) { 1017 maybe_warnx("decompression error"); 1018 goto stop_and_fail; 1019 } 1020 state = GZSTATE_MAGIC0; 1021 break; 1022 } 1023 continue; 1024 stop_and_fail: 1025 out_tot = -1; 1026 stop: 1027 break; 1028 } 1029 if (state > GZSTATE_INIT) 1030 inflateEnd(&z); 1031 1032 free(inbufp); 1033 out1: 1034 free(outbufp); 1035 out2: 1036 if (gsizep) 1037 *gsizep = in_tot; 1038 return (out_tot); 1039 } 1040 1041 #ifndef SMALL 1042 /* 1043 * set the owner, mode, flags & utimes using the given file descriptor. 1044 * file is only used in possible warning messages. 1045 */ 1046 static void 1047 copymodes(int fd, const struct stat *sbp, const char *file) 1048 { 1049 struct timeval times[2]; 1050 struct stat sb; 1051 1052 /* 1053 * If we have no info on the input, give this file some 1054 * default values and return.. 1055 */ 1056 if (sbp == NULL) { 1057 mode_t mask = umask(022); 1058 1059 (void)fchmod(fd, DEFFILEMODE & ~mask); 1060 (void)umask(mask); 1061 return; 1062 } 1063 sb = *sbp; 1064 1065 /* if the chown fails, remove set-id bits as-per compress(1) */ 1066 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1067 if (errno != EPERM) 1068 maybe_warn("couldn't fchown: %s", file); 1069 sb.st_mode &= ~(S_ISUID|S_ISGID); 1070 } 1071 1072 /* we only allow set-id and the 9 normal permission bits */ 1073 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1074 if (fchmod(fd, sb.st_mode) < 0) 1075 maybe_warn("couldn't fchmod: %s", file); 1076 1077 /* only try flags if they exist already */ 1078 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1079 maybe_warn("couldn't fchflags: %s", file); 1080 1081 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1082 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1083 if (futimes(fd, times) < 0) 1084 maybe_warn("couldn't utimes: %s", file); 1085 } 1086 #endif 1087 1088 /* what sort of file is this? */ 1089 static enum filetype 1090 file_gettype(u_char *buf) 1091 { 1092 1093 if (buf[0] == GZIP_MAGIC0 && 1094 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1095 return FT_GZIP; 1096 else 1097 #ifndef NO_BZIP2_SUPPORT 1098 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1099 buf[3] >= '0' && buf[3] <= '9') 1100 return FT_BZIP2; 1101 else 1102 #endif 1103 #ifndef NO_COMPRESS_SUPPORT 1104 if (memcmp(buf, Z_MAGIC, 2) == 0) 1105 return FT_Z; 1106 else 1107 #endif 1108 #ifndef NO_PACK_SUPPORT 1109 if (memcmp(buf, PACK_MAGIC, 2) == 0) 1110 return FT_PACK; 1111 else 1112 #endif 1113 #ifndef NO_XZ_SUPPORT 1114 if (memcmp(buf, XZ_MAGIC, 4) == 0) /* XXX: We only have 4 bytes */ 1115 return FT_XZ; 1116 else 1117 #endif 1118 return FT_UNKNOWN; 1119 } 1120 1121 #ifndef SMALL 1122 /* check the outfile is OK. */ 1123 static int 1124 check_outfile(const char *outfile) 1125 { 1126 struct stat sb; 1127 int ok = 1; 1128 1129 if (lflag == 0 && stat(outfile, &sb) == 0) { 1130 if (fflag) 1131 unlink(outfile); 1132 else if (isatty(STDIN_FILENO)) { 1133 char ans[10] = { 'n', '\0' }; /* default */ 1134 1135 fprintf(stderr, "%s already exists -- do you wish to " 1136 "overwrite (y or n)? " , outfile); 1137 (void)fgets(ans, sizeof(ans) - 1, stdin); 1138 if (ans[0] != 'y' && ans[0] != 'Y') { 1139 fprintf(stderr, "\tnot overwriting\n"); 1140 ok = 0; 1141 } else 1142 unlink(outfile); 1143 } else { 1144 maybe_warnx("%s already exists -- skipping", outfile); 1145 ok = 0; 1146 } 1147 } 1148 return ok; 1149 } 1150 1151 static void 1152 unlink_input(const char *file, const struct stat *sb) 1153 { 1154 struct stat nsb; 1155 1156 if (kflag) 1157 return; 1158 if (stat(file, &nsb) != 0) 1159 /* Must be gone already */ 1160 return; 1161 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1162 /* Definitely a different file */ 1163 return; 1164 unlink(file); 1165 } 1166 #endif 1167 1168 static const suffixes_t * 1169 check_suffix(char *file, int xlate) 1170 { 1171 const suffixes_t *s; 1172 int len = strlen(file); 1173 char *sp; 1174 1175 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1176 /* if it doesn't fit in "a.suf", don't bother */ 1177 if (s->ziplen >= len) 1178 continue; 1179 sp = file + len - s->ziplen; 1180 if (strcmp(s->zipped, sp) != 0) 1181 continue; 1182 if (xlate) 1183 strcpy(sp, s->normal); 1184 return s; 1185 } 1186 return NULL; 1187 } 1188 1189 /* 1190 * compress the given file: create a corresponding .gz file and remove the 1191 * original. 1192 */ 1193 static off_t 1194 file_compress(char *file, char *outfile, size_t outsize) 1195 { 1196 int in; 1197 int out; 1198 off_t size, insize; 1199 #ifndef SMALL 1200 struct stat isb, osb; 1201 const suffixes_t *suff; 1202 #endif 1203 1204 in = open(file, O_RDONLY); 1205 if (in == -1) { 1206 maybe_warn("can't open %s", file); 1207 return -1; 1208 } 1209 1210 if (cflag == 0) { 1211 #ifndef SMALL 1212 if (fstat(in, &isb) == 0) { 1213 if (isb.st_nlink > 1 && fflag == 0) { 1214 maybe_warnx("%s has %d other link%s -- " 1215 "skipping", file, isb.st_nlink - 1, 1216 isb.st_nlink == 1 ? "" : "s"); 1217 close(in); 1218 return -1; 1219 } 1220 } 1221 1222 if (fflag == 0 && (suff = check_suffix(file, 0)) 1223 && suff->zipped[0] != 0) { 1224 maybe_warnx("%s already has %s suffix -- unchanged", 1225 file, suff->zipped); 1226 close(in); 1227 return -1; 1228 } 1229 #endif 1230 1231 /* Add (usually) .gz to filename */ 1232 if ((size_t)snprintf(outfile, outsize, "%s%s", 1233 file, suffixes[0].zipped) >= outsize) 1234 memcpy(outfile + outsize - suffixes[0].ziplen - 1, 1235 suffixes[0].zipped, suffixes[0].ziplen + 1); 1236 1237 #ifndef SMALL 1238 if (check_outfile(outfile) == 0) { 1239 close(in); 1240 return -1; 1241 } 1242 #endif 1243 } 1244 1245 if (cflag == 0) { 1246 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1247 if (out == -1) { 1248 maybe_warn("could not create output: %s", outfile); 1249 fclose(stdin); 1250 return -1; 1251 } 1252 } else 1253 out = STDOUT_FILENO; 1254 1255 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1256 1257 (void)close(in); 1258 1259 /* 1260 * If there was an error, insize will be -1. 1261 * If we compressed to stdout, just return the size. 1262 * Otherwise stat the file and check it is the correct size. 1263 * We only blow away the file if we can stat the output and it 1264 * has the expected size. 1265 */ 1266 if (cflag != 0) 1267 return insize == -1 ? -1 : size; 1268 1269 #ifndef SMALL 1270 if (fstat(out, &osb) != 0) { 1271 maybe_warn("couldn't stat: %s", outfile); 1272 goto bad_outfile; 1273 } 1274 1275 if (osb.st_size != size) { 1276 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1277 " != %" PRIdOFF "), deleting", 1278 outfile, osb.st_size, size); 1279 goto bad_outfile; 1280 } 1281 1282 copymodes(out, &isb, outfile); 1283 #endif 1284 if (close(out) == -1) 1285 maybe_warn("couldn't close output"); 1286 1287 /* output is good, ok to delete input */ 1288 unlink_input(file, &isb); 1289 return size; 1290 1291 #ifndef SMALL 1292 bad_outfile: 1293 if (close(out) == -1) 1294 maybe_warn("couldn't close output"); 1295 1296 maybe_warnx("leaving original %s", file); 1297 unlink(outfile); 1298 return size; 1299 #endif 1300 } 1301 1302 /* uncompress the given file and remove the original */ 1303 static off_t 1304 file_uncompress(char *file, char *outfile, size_t outsize) 1305 { 1306 struct stat isb, osb; 1307 off_t size; 1308 ssize_t rbytes; 1309 unsigned char header1[4]; 1310 enum filetype method; 1311 int fd, ofd, zfd = -1; 1312 #ifndef SMALL 1313 ssize_t rv; 1314 time_t timestamp = 0; 1315 char name[PATH_MAX + 1]; 1316 #endif 1317 1318 /* gather the old name info */ 1319 1320 fd = open(file, O_RDONLY); 1321 if (fd < 0) { 1322 maybe_warn("can't open %s", file); 1323 goto lose; 1324 } 1325 1326 strlcpy(outfile, file, outsize); 1327 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1328 maybe_warnx("%s: unknown suffix -- ignored", file); 1329 goto lose; 1330 } 1331 1332 rbytes = read(fd, header1, sizeof header1); 1333 if (rbytes != sizeof header1) { 1334 /* we don't want to fail here. */ 1335 #ifndef SMALL 1336 if (fflag) 1337 goto lose; 1338 #endif 1339 if (rbytes == -1) 1340 maybe_warn("can't read %s", file); 1341 else 1342 goto unexpected_EOF; 1343 goto lose; 1344 } 1345 1346 method = file_gettype(header1); 1347 #ifndef SMALL 1348 if (fflag == 0 && method == FT_UNKNOWN) { 1349 maybe_warnx("%s: not in gzip format", file); 1350 goto lose; 1351 } 1352 1353 #endif 1354 1355 #ifndef SMALL 1356 if (method == FT_GZIP && Nflag) { 1357 unsigned char ts[4]; /* timestamp */ 1358 1359 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1360 if (rv >= 0 && rv < (ssize_t)(sizeof ts)) 1361 goto unexpected_EOF; 1362 if (rv == -1) { 1363 if (!fflag) 1364 maybe_warn("can't read %s", file); 1365 goto lose; 1366 } 1367 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1368 1369 if (header1[3] & ORIG_NAME) { 1370 rbytes = pread(fd, name, sizeof(name) - 1, GZIP_ORIGNAME); 1371 if (rbytes < 0) { 1372 maybe_warn("can't read %s", file); 1373 goto lose; 1374 } 1375 if (name[0] != '\0') { 1376 char *dp, *nf; 1377 1378 /* Make sure that name is NUL-terminated */ 1379 name[rbytes] = '\0'; 1380 1381 /* strip saved directory name */ 1382 nf = strrchr(name, '/'); 1383 if (nf == NULL) 1384 nf = name; 1385 else 1386 nf++; 1387 1388 /* preserve original directory name */ 1389 dp = strrchr(file, '/'); 1390 if (dp == NULL) 1391 dp = file; 1392 else 1393 dp++; 1394 snprintf(outfile, outsize, "%.*s%.*s", 1395 (int) (dp - file), 1396 file, (int) rbytes, nf); 1397 } 1398 } 1399 } 1400 #endif 1401 lseek(fd, 0, SEEK_SET); 1402 1403 if (cflag == 0 || lflag) { 1404 if (fstat(fd, &isb) != 0) 1405 goto lose; 1406 #ifndef SMALL 1407 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1408 maybe_warnx("%s has %d other links -- skipping", 1409 file, isb.st_nlink - 1); 1410 goto lose; 1411 } 1412 if (nflag == 0 && timestamp) 1413 isb.st_mtime = timestamp; 1414 if (check_outfile(outfile) == 0) 1415 goto lose; 1416 #endif 1417 } 1418 1419 if (cflag == 0 && lflag == 0) { 1420 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1421 if (zfd == STDOUT_FILENO) { 1422 /* We won't close STDOUT_FILENO later... */ 1423 zfd = dup(zfd); 1424 close(STDOUT_FILENO); 1425 } 1426 if (zfd == -1) { 1427 maybe_warn("can't open %s", outfile); 1428 goto lose; 1429 } 1430 } else 1431 zfd = STDOUT_FILENO; 1432 1433 switch (method) { 1434 #ifndef NO_BZIP2_SUPPORT 1435 case FT_BZIP2: 1436 /* XXX */ 1437 if (lflag) { 1438 maybe_warnx("no -l with bzip2 files"); 1439 goto lose; 1440 } 1441 1442 size = unbzip2(fd, zfd, NULL, 0, NULL); 1443 break; 1444 #endif 1445 1446 #ifndef NO_COMPRESS_SUPPORT 1447 case FT_Z: { 1448 FILE *in, *out; 1449 1450 /* XXX */ 1451 if (lflag) { 1452 maybe_warnx("no -l with Lempel-Ziv files"); 1453 goto lose; 1454 } 1455 1456 if ((in = zdopen(fd)) == NULL) { 1457 maybe_warn("zdopen for read: %s", file); 1458 goto lose; 1459 } 1460 1461 out = fdopen(dup(zfd), "w"); 1462 if (out == NULL) { 1463 maybe_warn("fdopen for write: %s", outfile); 1464 fclose(in); 1465 goto lose; 1466 } 1467 1468 size = zuncompress(in, out, NULL, 0, NULL); 1469 /* need to fclose() if ferror() is true... */ 1470 if (ferror(in) | fclose(in)) { 1471 maybe_warn("failed infile fclose"); 1472 unlink(outfile); 1473 (void)fclose(out); 1474 } 1475 if (fclose(out) != 0) { 1476 maybe_warn("failed outfile fclose"); 1477 unlink(outfile); 1478 goto lose; 1479 } 1480 break; 1481 } 1482 #endif 1483 1484 #ifndef NO_PACK_SUPPORT 1485 case FT_PACK: 1486 if (lflag) { 1487 maybe_warnx("no -l with packed files"); 1488 goto lose; 1489 } 1490 1491 size = unpack(fd, zfd, NULL, 0, NULL); 1492 break; 1493 #endif 1494 1495 #ifndef NO_XZ_SUPPORT 1496 case FT_XZ: 1497 if (lflag) { 1498 maybe_warnx("no -l with xz files"); 1499 goto lose; 1500 } 1501 1502 size = unxz(fd, zfd, NULL, 0, NULL); 1503 break; 1504 #endif 1505 1506 #ifndef SMALL 1507 case FT_UNKNOWN: 1508 if (lflag) { 1509 maybe_warnx("no -l for unknown filetypes"); 1510 goto lose; 1511 } 1512 size = cat_fd(NULL, 0, NULL, fd); 1513 break; 1514 #endif 1515 default: 1516 if (lflag) { 1517 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1518 close(fd); 1519 return -1; /* XXX */ 1520 } 1521 1522 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1523 break; 1524 } 1525 1526 if (close(fd) != 0) 1527 maybe_warn("couldn't close input"); 1528 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1529 maybe_warn("couldn't close output"); 1530 1531 if (size == -1) { 1532 if (cflag == 0) 1533 unlink(outfile); 1534 maybe_warnx("%s: uncompress failed", file); 1535 return -1; 1536 } 1537 1538 /* if testing, or we uncompressed to stdout, this is all we need */ 1539 #ifndef SMALL 1540 if (tflag) 1541 return size; 1542 #endif 1543 /* if we are uncompressing to stdin, don't remove the file. */ 1544 if (cflag) 1545 return size; 1546 1547 /* 1548 * if we create a file... 1549 */ 1550 /* 1551 * if we can't stat the file don't remove the file. 1552 */ 1553 1554 ofd = open(outfile, O_RDWR, 0); 1555 if (ofd == -1) { 1556 maybe_warn("couldn't open (leaving original): %s", 1557 outfile); 1558 return -1; 1559 } 1560 if (fstat(ofd, &osb) != 0) { 1561 maybe_warn("couldn't stat (leaving original): %s", 1562 outfile); 1563 close(ofd); 1564 return -1; 1565 } 1566 if (osb.st_size != size) { 1567 maybe_warnx("stat gave different size: %" PRIdOFF 1568 " != %" PRIdOFF " (leaving original)", 1569 size, osb.st_size); 1570 close(ofd); 1571 unlink(outfile); 1572 return -1; 1573 } 1574 unlink_input(file, &isb); 1575 #ifndef SMALL 1576 copymodes(ofd, &isb, outfile); 1577 #endif 1578 close(ofd); 1579 return size; 1580 1581 unexpected_EOF: 1582 maybe_warnx("%s: unexpected end of file", file); 1583 lose: 1584 if (fd != -1) 1585 close(fd); 1586 if (zfd != -1 && zfd != STDOUT_FILENO) 1587 close(fd); 1588 return -1; 1589 } 1590 1591 #ifndef SMALL 1592 static off_t 1593 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1594 { 1595 char buf[BUFLEN]; 1596 off_t in_tot; 1597 ssize_t w; 1598 1599 in_tot = count; 1600 w = write(STDOUT_FILENO, prepend, count); 1601 if (w == -1 || (size_t)w != count) { 1602 maybe_warn("write to stdout"); 1603 return -1; 1604 } 1605 for (;;) { 1606 ssize_t rv; 1607 1608 rv = read(fd, buf, sizeof buf); 1609 if (rv == 0) 1610 break; 1611 if (rv < 0) { 1612 maybe_warn("read from fd %d", fd); 1613 break; 1614 } 1615 1616 if (write(STDOUT_FILENO, buf, rv) != rv) { 1617 maybe_warn("write to stdout"); 1618 break; 1619 } 1620 in_tot += rv; 1621 } 1622 1623 if (gsizep) 1624 *gsizep = in_tot; 1625 return (in_tot); 1626 } 1627 #endif 1628 1629 static void 1630 handle_stdin(void) 1631 { 1632 unsigned char header1[4]; 1633 off_t usize, gsize; 1634 enum filetype method; 1635 ssize_t bytes_read; 1636 #ifndef NO_COMPRESS_SUPPORT 1637 FILE *in; 1638 #endif 1639 1640 #ifndef SMALL 1641 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1642 maybe_warnx("standard input is a terminal -- ignoring"); 1643 return; 1644 } 1645 #endif 1646 1647 if (lflag) { 1648 struct stat isb; 1649 1650 /* XXX could read the whole file, etc. */ 1651 if (fstat(STDIN_FILENO, &isb) < 0) { 1652 maybe_warn("fstat"); 1653 return; 1654 } 1655 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1656 return; 1657 } 1658 1659 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1660 if (bytes_read == -1) { 1661 maybe_warn("can't read stdin"); 1662 return; 1663 } else if (bytes_read != sizeof(header1)) { 1664 maybe_warnx("(stdin): unexpected end of file"); 1665 return; 1666 } 1667 1668 method = file_gettype(header1); 1669 switch (method) { 1670 default: 1671 #ifndef SMALL 1672 if (fflag == 0) { 1673 maybe_warnx("unknown compression format"); 1674 return; 1675 } 1676 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1677 break; 1678 #endif 1679 case FT_GZIP: 1680 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1681 (char *)header1, sizeof header1, &gsize, "(stdin)"); 1682 break; 1683 #ifndef NO_BZIP2_SUPPORT 1684 case FT_BZIP2: 1685 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1686 (char *)header1, sizeof header1, &gsize); 1687 break; 1688 #endif 1689 #ifndef NO_COMPRESS_SUPPORT 1690 case FT_Z: 1691 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1692 maybe_warnx("zopen of stdin"); 1693 return; 1694 } 1695 1696 usize = zuncompress(in, stdout, (char *)header1, 1697 sizeof header1, &gsize); 1698 fclose(in); 1699 break; 1700 #endif 1701 #ifndef NO_PACK_SUPPORT 1702 case FT_PACK: 1703 usize = unpack(STDIN_FILENO, STDOUT_FILENO, 1704 (char *)header1, sizeof header1, &gsize); 1705 break; 1706 #endif 1707 #ifndef NO_XZ_SUPPORT 1708 case FT_XZ: 1709 usize = unxz(STDIN_FILENO, STDOUT_FILENO, 1710 (char *)header1, sizeof header1, &gsize); 1711 break; 1712 #endif 1713 } 1714 1715 #ifndef SMALL 1716 if (vflag && !tflag && usize != -1 && gsize != -1) 1717 print_verbage(NULL, NULL, usize, gsize); 1718 if (vflag && tflag) 1719 print_test("(stdin)", usize != -1); 1720 #endif 1721 1722 } 1723 1724 static void 1725 handle_stdout(void) 1726 { 1727 off_t gsize, usize; 1728 struct stat sb; 1729 time_t systime; 1730 uint32_t mtime; 1731 int ret; 1732 1733 #ifndef SMALL 1734 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1735 maybe_warnx("standard output is a terminal -- ignoring"); 1736 return; 1737 } 1738 #endif 1739 /* If stdin is a file use its mtime, otherwise use current time */ 1740 ret = fstat(STDIN_FILENO, &sb); 1741 1742 #ifndef SMALL 1743 if (ret < 0) { 1744 maybe_warn("Can't stat stdin"); 1745 return; 1746 } 1747 #endif 1748 1749 if (S_ISREG(sb.st_mode)) 1750 mtime = (uint32_t)sb.st_mtime; 1751 else { 1752 systime = time(NULL); 1753 #ifndef SMALL 1754 if (systime == -1) { 1755 maybe_warn("time"); 1756 return; 1757 } 1758 #endif 1759 mtime = (uint32_t)systime; 1760 } 1761 1762 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1763 #ifndef SMALL 1764 if (vflag && !tflag && usize != -1 && gsize != -1) 1765 print_verbage(NULL, NULL, usize, gsize); 1766 #endif 1767 } 1768 1769 /* do what is asked for, for the path name */ 1770 static void 1771 handle_pathname(char *path) 1772 { 1773 char *opath = path, *s = NULL; 1774 ssize_t len; 1775 int slen; 1776 struct stat sb; 1777 1778 /* check for stdout/stdin */ 1779 if (path[0] == '-' && path[1] == '\0') { 1780 if (dflag) 1781 handle_stdin(); 1782 else 1783 handle_stdout(); 1784 return; 1785 } 1786 1787 retry: 1788 if (stat(path, &sb) != 0) { 1789 /* lets try <path>.gz if we're decompressing */ 1790 if (dflag && s == NULL && errno == ENOENT) { 1791 len = strlen(path); 1792 slen = suffixes[0].ziplen; 1793 s = malloc(len + slen + 1); 1794 if (s == NULL) 1795 maybe_err("malloc"); 1796 memcpy(s, path, len); 1797 memcpy(s + len, suffixes[0].zipped, slen + 1); 1798 path = s; 1799 goto retry; 1800 } 1801 maybe_warn("can't stat: %s", opath); 1802 goto out; 1803 } 1804 1805 if (S_ISDIR(sb.st_mode)) { 1806 #ifndef SMALL 1807 if (rflag) 1808 handle_dir(path); 1809 else 1810 #endif 1811 maybe_warnx("%s is a directory", path); 1812 goto out; 1813 } 1814 1815 if (S_ISREG(sb.st_mode)) 1816 handle_file(path, &sb); 1817 else 1818 maybe_warnx("%s is not a regular file", path); 1819 1820 out: 1821 if (s) 1822 free(s); 1823 } 1824 1825 /* compress/decompress a file */ 1826 static void 1827 handle_file(char *file, struct stat *sbp) 1828 { 1829 off_t usize, gsize; 1830 char outfile[PATH_MAX]; 1831 1832 infile = file; 1833 if (dflag) { 1834 usize = file_uncompress(file, outfile, sizeof(outfile)); 1835 #ifndef SMALL 1836 if (vflag && tflag) 1837 print_test(file, usize != -1); 1838 #endif 1839 if (usize == -1) 1840 return; 1841 gsize = sbp->st_size; 1842 } else { 1843 gsize = file_compress(file, outfile, sizeof(outfile)); 1844 if (gsize == -1) 1845 return; 1846 usize = sbp->st_size; 1847 } 1848 1849 1850 #ifndef SMALL 1851 if (vflag && !tflag) 1852 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1853 #endif 1854 } 1855 1856 #ifndef SMALL 1857 /* this is used with -r to recursively descend directories */ 1858 static void 1859 handle_dir(char *dir) 1860 { 1861 char *path_argv[2]; 1862 FTS *fts; 1863 FTSENT *entry; 1864 1865 path_argv[0] = dir; 1866 path_argv[1] = 0; 1867 fts = fts_open(path_argv, FTS_PHYSICAL, NULL); 1868 if (fts == NULL) { 1869 warn("couldn't fts_open %s", dir); 1870 return; 1871 } 1872 1873 while ((entry = fts_read(fts))) { 1874 switch(entry->fts_info) { 1875 case FTS_D: 1876 case FTS_DP: 1877 continue; 1878 1879 case FTS_DNR: 1880 case FTS_ERR: 1881 case FTS_NS: 1882 maybe_warn("%s", entry->fts_path); 1883 continue; 1884 case FTS_F: 1885 handle_file(entry->fts_name, entry->fts_statp); 1886 } 1887 } 1888 (void)fts_close(fts); 1889 } 1890 #endif 1891 1892 /* print a ratio - size reduction as a fraction of uncompressed size */ 1893 static void 1894 print_ratio(off_t in, off_t out, FILE *where) 1895 { 1896 int percent10; /* 10 * percent */ 1897 off_t diff; 1898 char buff[8]; 1899 int len; 1900 1901 diff = in - out/2; 1902 if (diff <= 0) 1903 /* 1904 * Output is more than double size of input! print -99.9% 1905 * Quite possibly we've failed to get the original size. 1906 */ 1907 percent10 = -999; 1908 else { 1909 /* 1910 * We only need 12 bits of result from the final division, 1911 * so reduce the values until a 32bit division will suffice. 1912 */ 1913 while (in > 0x100000) { 1914 diff >>= 1; 1915 in >>= 1; 1916 } 1917 if (in != 0) 1918 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1919 else 1920 percent10 = 0; 1921 } 1922 1923 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1924 /* Move the '.' to before the last digit */ 1925 buff[len - 1] = buff[len - 2]; 1926 buff[len - 2] = '.'; 1927 fprintf(where, "%5s%%", buff); 1928 } 1929 1930 #ifndef SMALL 1931 /* print compression statistics, and the new name (if there is one!) */ 1932 static void 1933 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1934 { 1935 if (file) 1936 fprintf(stderr, "%s:%s ", file, 1937 strlen(file) < 7 ? "\t\t" : "\t"); 1938 print_ratio(usize, gsize, stderr); 1939 if (nfile) 1940 fprintf(stderr, " -- replaced with %s", nfile); 1941 fprintf(stderr, "\n"); 1942 fflush(stderr); 1943 } 1944 1945 /* print test results */ 1946 static void 1947 print_test(const char *file, int ok) 1948 { 1949 1950 if (exit_value == 0 && ok == 0) 1951 exit_value = 1; 1952 fprintf(stderr, "%s:%s %s\n", file, 1953 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1954 fflush(stderr); 1955 } 1956 #endif 1957 1958 /* print a file's info ala --list */ 1959 /* eg: 1960 compressed uncompressed ratio uncompressed_name 1961 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1962 */ 1963 static void 1964 print_list(int fd, off_t out, const char *outfile, time_t ts) 1965 { 1966 static int first = 1; 1967 #ifndef SMALL 1968 static off_t in_tot, out_tot; 1969 uint32_t crc = 0; 1970 #endif 1971 off_t in = 0, rv; 1972 1973 if (first) { 1974 #ifndef SMALL 1975 if (vflag) 1976 printf("method crc date time "); 1977 #endif 1978 if (qflag == 0) 1979 printf(" compressed uncompressed " 1980 "ratio uncompressed_name\n"); 1981 } 1982 first = 0; 1983 1984 /* print totals? */ 1985 #ifndef SMALL 1986 if (fd == -1) { 1987 in = in_tot; 1988 out = out_tot; 1989 } else 1990 #endif 1991 { 1992 /* read the last 4 bytes - this is the uncompressed size */ 1993 rv = lseek(fd, (off_t)(-8), SEEK_END); 1994 if (rv != -1) { 1995 unsigned char buf[8]; 1996 uint32_t usize; 1997 1998 rv = read(fd, (char *)buf, sizeof(buf)); 1999 if (rv == -1) 2000 maybe_warn("read of uncompressed size"); 2001 else if (rv != sizeof(buf)) 2002 maybe_warnx("read of uncompressed size"); 2003 2004 else { 2005 usize = buf[4] | buf[5] << 8 | 2006 buf[6] << 16 | buf[7] << 24; 2007 in = (off_t)usize; 2008 #ifndef SMALL 2009 crc = buf[0] | buf[1] << 8 | 2010 buf[2] << 16 | buf[3] << 24; 2011 #endif 2012 } 2013 } 2014 } 2015 2016 #ifndef SMALL 2017 if (vflag && fd == -1) 2018 printf(" "); 2019 else if (vflag) { 2020 char *date = ctime(&ts); 2021 2022 /* skip the day, 1/100th second, and year */ 2023 date += 4; 2024 date[12] = 0; 2025 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 2026 } 2027 in_tot += in; 2028 out_tot += out; 2029 #endif 2030 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 2031 print_ratio(in, out, stdout); 2032 printf(" %s\n", outfile); 2033 } 2034 2035 /* display the usage of NetBSD gzip */ 2036 static void 2037 usage(void) 2038 { 2039 2040 fprintf(stderr, "%s\n", gzip_version); 2041 fprintf(stderr, 2042 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n" 2043 #ifndef SMALL 2044 " -1 --fast fastest (worst) compression\n" 2045 " -2 .. -8 set compression level\n" 2046 " -9 --best best (slowest) compression\n" 2047 " -c --stdout write to stdout, keep original files\n" 2048 " --to-stdout\n" 2049 " -d --decompress uncompress files\n" 2050 " --uncompress\n" 2051 " -f --force force overwriting & compress links\n" 2052 " -h --help display this help\n" 2053 " -k --keep don't delete input files during operation\n" 2054 " -l --list list compressed file contents\n" 2055 " -N --name save or restore original file name and time stamp\n" 2056 " -n --no-name don't save original file name or time stamp\n" 2057 " -q --quiet output no warnings\n" 2058 " -r --recursive recursively compress files in directories\n" 2059 " -S .suf use suffix .suf instead of .gz\n" 2060 " --suffix .suf\n" 2061 " -t --test test compressed file\n" 2062 " -V --version display program version\n" 2063 " -v --verbose print extra statistics\n", 2064 #else 2065 , 2066 #endif 2067 getprogname()); 2068 exit(0); 2069 } 2070 2071 /* display the version of NetBSD gzip */ 2072 static void 2073 display_version(void) 2074 { 2075 2076 fprintf(stderr, "%s\n", gzip_version); 2077 exit(0); 2078 } 2079 2080 #ifndef NO_BZIP2_SUPPORT 2081 #include "unbzip2.c" 2082 #endif 2083 #ifndef NO_COMPRESS_SUPPORT 2084 #include "zuncompress.c" 2085 #endif 2086 #ifndef NO_PACK_SUPPORT 2087 #include "unpack.c" 2088 #endif 2089 #ifndef NO_XZ_SUPPORT 2090 #include "unxz.c" 2091 #endif 2092 2093 static ssize_t 2094 read_retry(int fd, void *buf, size_t sz) 2095 { 2096 char *cp = buf; 2097 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2098 2099 while (left > 0) { 2100 ssize_t ret; 2101 2102 ret = read(fd, cp, left); 2103 if (ret == -1) { 2104 return ret; 2105 } else if (ret == 0) { 2106 break; /* EOF */ 2107 } 2108 cp += ret; 2109 left -= ret; 2110 } 2111 2112 return sz - left; 2113 } 2114
Indexes created Tue Sep 23 07:09:52 GMT 2025