gzip.c revision 1.108
1 /* $NetBSD: gzip.c,v 1.108 2015/04/15 02:29:12 christos 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 1.108 2015/04/15 02:29:12 christos 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 goto stop_and_fail; 792 } 793 ADVANCE(); 794 state++; 795 out_sub_tot = 0; 796 crc = crc32(0L, Z_NULL, 0); 797 break; 798 799 case GZSTATE_MAGIC1: 800 if (*z.next_in != GZIP_MAGIC1 && 801 *z.next_in != GZIP_OMAGIC1) { 802 maybe_warnx("input not gziped (MAGIC1)"); 803 goto stop_and_fail; 804 } 805 ADVANCE(); 806 state++; 807 break; 808 809 case GZSTATE_METHOD: 810 if (*z.next_in != Z_DEFLATED) { 811 maybe_warnx("unknown compression method"); 812 goto stop_and_fail; 813 } 814 ADVANCE(); 815 state++; 816 break; 817 818 case GZSTATE_FLAGS: 819 flags = *z.next_in; 820 ADVANCE(); 821 skip_count = 6; 822 state++; 823 break; 824 825 case GZSTATE_SKIPPING: 826 if (skip_count > 0) { 827 skip_count--; 828 ADVANCE(); 829 } else 830 state++; 831 break; 832 833 case GZSTATE_EXTRA: 834 if ((flags & EXTRA_FIELD) == 0) { 835 state = GZSTATE_ORIGNAME; 836 break; 837 } 838 skip_count = *z.next_in; 839 ADVANCE(); 840 state++; 841 break; 842 843 case GZSTATE_EXTRA2: 844 skip_count |= ((*z.next_in) << 8); 845 ADVANCE(); 846 state++; 847 break; 848 849 case GZSTATE_EXTRA3: 850 if (skip_count > 0) { 851 skip_count--; 852 ADVANCE(); 853 } else 854 state++; 855 break; 856 857 case GZSTATE_ORIGNAME: 858 if ((flags & ORIG_NAME) == 0) { 859 state++; 860 break; 861 } 862 if (*z.next_in == 0) 863 state++; 864 ADVANCE(); 865 break; 866 867 case GZSTATE_COMMENT: 868 if ((flags & COMMENT) == 0) { 869 state++; 870 break; 871 } 872 if (*z.next_in == 0) 873 state++; 874 ADVANCE(); 875 break; 876 877 case GZSTATE_HEAD_CRC1: 878 if (flags & HEAD_CRC) 879 skip_count = 2; 880 else 881 skip_count = 0; 882 state++; 883 break; 884 885 case GZSTATE_HEAD_CRC2: 886 if (skip_count > 0) { 887 skip_count--; 888 ADVANCE(); 889 } else 890 state++; 891 break; 892 893 case GZSTATE_INIT: 894 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 895 maybe_warnx("failed to inflateInit"); 896 goto stop_and_fail; 897 } 898 state++; 899 break; 900 901 case GZSTATE_READ: 902 error = inflate(&z, Z_FINISH); 903 switch (error) { 904 /* Z_BUF_ERROR goes with Z_FINISH... */ 905 case Z_BUF_ERROR: 906 if (z.avail_out > 0 && !done_reading) 907 continue; 908 909 case Z_STREAM_END: 910 case Z_OK: 911 break; 912 913 case Z_NEED_DICT: 914 maybe_warnx("Z_NEED_DICT error"); 915 goto stop_and_fail; 916 case Z_DATA_ERROR: 917 maybe_warnx("data stream error"); 918 goto stop_and_fail; 919 case Z_STREAM_ERROR: 920 maybe_warnx("internal stream error"); 921 goto stop_and_fail; 922 case Z_MEM_ERROR: 923 maybe_warnx("memory allocation error"); 924 goto stop_and_fail; 925 926 default: 927 maybe_warn("unknown error from inflate(): %d", 928 error); 929 } 930 wr = BUFLEN - z.avail_out; 931 932 if (wr != 0) { 933 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 934 if ( 935 #ifndef SMALL 936 /* don't write anything with -t */ 937 tflag == 0 && 938 #endif 939 write(out, outbufp, wr) != wr) { 940 maybe_warn("error writing to output"); 941 goto stop_and_fail; 942 } 943 944 out_tot += wr; 945 out_sub_tot += wr; 946 } 947 948 if (error == Z_STREAM_END) { 949 inflateEnd(&z); 950 state++; 951 } 952 953 z.next_out = (unsigned char *)outbufp; 954 z.avail_out = BUFLEN; 955 956 break; 957 case GZSTATE_CRC: 958 { 959 uLong origcrc; 960 961 if (z.avail_in < 4) { 962 if (!done_reading) { 963 needmore = 1; 964 continue; 965 } 966 maybe_warnx("truncated input"); 967 goto stop_and_fail; 968 } 969 origcrc = ((unsigned)z.next_in[0] & 0xff) | 970 ((unsigned)z.next_in[1] & 0xff) << 8 | 971 ((unsigned)z.next_in[2] & 0xff) << 16 | 972 ((unsigned)z.next_in[3] & 0xff) << 24; 973 if (origcrc != crc) { 974 maybe_warnx("invalid compressed" 975 " data--crc error"); 976 goto stop_and_fail; 977 } 978 } 979 980 z.avail_in -= 4; 981 z.next_in += 4; 982 983 if (!z.avail_in && done_reading) { 984 goto stop; 985 } 986 state++; 987 break; 988 case GZSTATE_LEN: 989 { 990 uLong origlen; 991 992 if (z.avail_in < 4) { 993 if (!done_reading) { 994 needmore = 1; 995 continue; 996 } 997 maybe_warnx("truncated input"); 998 goto stop_and_fail; 999 } 1000 origlen = ((unsigned)z.next_in[0] & 0xff) | 1001 ((unsigned)z.next_in[1] & 0xff) << 8 | 1002 ((unsigned)z.next_in[2] & 0xff) << 16 | 1003 ((unsigned)z.next_in[3] & 0xff) << 24; 1004 1005 if (origlen != out_sub_tot) { 1006 maybe_warnx("invalid compressed" 1007 " data--length error"); 1008 goto stop_and_fail; 1009 } 1010 } 1011 1012 z.avail_in -= 4; 1013 z.next_in += 4; 1014 1015 if (error < 0) { 1016 maybe_warnx("decompression error"); 1017 goto stop_and_fail; 1018 } 1019 state = GZSTATE_MAGIC0; 1020 break; 1021 } 1022 continue; 1023 stop_and_fail: 1024 out_tot = -1; 1025 stop: 1026 break; 1027 } 1028 if (state > GZSTATE_INIT) 1029 inflateEnd(&z); 1030 1031 free(inbufp); 1032 out1: 1033 free(outbufp); 1034 out2: 1035 if (gsizep) 1036 *gsizep = in_tot; 1037 return (out_tot); 1038 } 1039 1040 #ifndef SMALL 1041 /* 1042 * set the owner, mode, flags & utimes using the given file descriptor. 1043 * file is only used in possible warning messages. 1044 */ 1045 static void 1046 copymodes(int fd, const struct stat *sbp, const char *file) 1047 { 1048 struct timeval times[2]; 1049 struct stat sb; 1050 1051 /* 1052 * If we have no info on the input, give this file some 1053 * default values and return.. 1054 */ 1055 if (sbp == NULL) { 1056 mode_t mask = umask(022); 1057 1058 (void)fchmod(fd, DEFFILEMODE & ~mask); 1059 (void)umask(mask); 1060 return; 1061 } 1062 sb = *sbp; 1063 1064 /* if the chown fails, remove set-id bits as-per compress(1) */ 1065 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1066 if (errno != EPERM) 1067 maybe_warn("couldn't fchown: %s", file); 1068 sb.st_mode &= ~(S_ISUID|S_ISGID); 1069 } 1070 1071 /* we only allow set-id and the 9 normal permission bits */ 1072 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1073 if (fchmod(fd, sb.st_mode) < 0) 1074 maybe_warn("couldn't fchmod: %s", file); 1075 1076 /* only try flags if they exist already */ 1077 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1078 maybe_warn("couldn't fchflags: %s", file); 1079 1080 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1081 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1082 if (futimes(fd, times) < 0) 1083 maybe_warn("couldn't utimes: %s", file); 1084 } 1085 #endif 1086 1087 /* what sort of file is this? */ 1088 static enum filetype 1089 file_gettype(u_char *buf) 1090 { 1091 1092 if (buf[0] == GZIP_MAGIC0 && 1093 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1094 return FT_GZIP; 1095 else 1096 #ifndef NO_BZIP2_SUPPORT 1097 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1098 buf[3] >= '0' && buf[3] <= '9') 1099 return FT_BZIP2; 1100 else 1101 #endif 1102 #ifndef NO_COMPRESS_SUPPORT 1103 if (memcmp(buf, Z_MAGIC, 2) == 0) 1104 return FT_Z; 1105 else 1106 #endif 1107 #ifndef NO_PACK_SUPPORT 1108 if (memcmp(buf, PACK_MAGIC, 2) == 0) 1109 return FT_PACK; 1110 else 1111 #endif 1112 #ifndef NO_XZ_SUPPORT 1113 if (memcmp(buf, XZ_MAGIC, 4) == 0) /* XXX: We only have 4 bytes */ 1114 return FT_XZ; 1115 else 1116 #endif 1117 return FT_UNKNOWN; 1118 } 1119 1120 #ifndef SMALL 1121 /* check the outfile is OK. */ 1122 static int 1123 check_outfile(const char *outfile) 1124 { 1125 struct stat sb; 1126 int ok = 1; 1127 1128 if (lflag == 0 && stat(outfile, &sb) == 0) { 1129 if (fflag) 1130 unlink(outfile); 1131 else if (isatty(STDIN_FILENO)) { 1132 char ans[10] = { 'n', '\0' }; /* default */ 1133 1134 fprintf(stderr, "%s already exists -- do you wish to " 1135 "overwrite (y or n)? " , outfile); 1136 (void)fgets(ans, sizeof(ans) - 1, stdin); 1137 if (ans[0] != 'y' && ans[0] != 'Y') { 1138 fprintf(stderr, "\tnot overwriting\n"); 1139 ok = 0; 1140 } else 1141 unlink(outfile); 1142 } else { 1143 maybe_warnx("%s already exists -- skipping", outfile); 1144 ok = 0; 1145 } 1146 } 1147 return ok; 1148 } 1149 1150 static void 1151 unlink_input(const char *file, const struct stat *sb) 1152 { 1153 struct stat nsb; 1154 1155 if (kflag) 1156 return; 1157 if (stat(file, &nsb) != 0) 1158 /* Must be gone already */ 1159 return; 1160 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1161 /* Definitely a different file */ 1162 return; 1163 unlink(file); 1164 } 1165 #endif 1166 1167 static const suffixes_t * 1168 check_suffix(char *file, int xlate) 1169 { 1170 const suffixes_t *s; 1171 int len = strlen(file); 1172 char *sp; 1173 1174 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1175 /* if it doesn't fit in "a.suf", don't bother */ 1176 if (s->ziplen >= len) 1177 continue; 1178 sp = file + len - s->ziplen; 1179 if (strcmp(s->zipped, sp) != 0) 1180 continue; 1181 if (xlate) 1182 strcpy(sp, s->normal); 1183 return s; 1184 } 1185 return NULL; 1186 } 1187 1188 /* 1189 * compress the given file: create a corresponding .gz file and remove the 1190 * original. 1191 */ 1192 static off_t 1193 file_compress(char *file, char *outfile, size_t outsize) 1194 { 1195 int in; 1196 int out; 1197 off_t size, insize; 1198 #ifndef SMALL 1199 struct stat isb, osb; 1200 const suffixes_t *suff; 1201 #endif 1202 1203 in = open(file, O_RDONLY); 1204 if (in == -1) { 1205 maybe_warn("can't open %s", file); 1206 return -1; 1207 } 1208 1209 if (cflag == 0) { 1210 #ifndef SMALL 1211 if (fstat(in, &isb) == 0) { 1212 if (isb.st_nlink > 1 && fflag == 0) { 1213 maybe_warnx("%s has %d other link%s -- " 1214 "skipping", file, isb.st_nlink - 1, 1215 isb.st_nlink == 1 ? "" : "s"); 1216 close(in); 1217 return -1; 1218 } 1219 } 1220 1221 if (fflag == 0 && (suff = check_suffix(file, 0)) 1222 && suff->zipped[0] != 0) { 1223 maybe_warnx("%s already has %s suffix -- unchanged", 1224 file, suff->zipped); 1225 close(in); 1226 return -1; 1227 } 1228 #endif 1229 1230 /* Add (usually) .gz to filename */ 1231 if ((size_t)snprintf(outfile, outsize, "%s%s", 1232 file, suffixes[0].zipped) >= outsize) 1233 memcpy(outfile + outsize - suffixes[0].ziplen - 1, 1234 suffixes[0].zipped, suffixes[0].ziplen + 1); 1235 1236 #ifndef SMALL 1237 if (check_outfile(outfile) == 0) { 1238 close(in); 1239 return -1; 1240 } 1241 #endif 1242 } 1243 1244 if (cflag == 0) { 1245 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1246 if (out == -1) { 1247 maybe_warn("could not create output: %s", outfile); 1248 fclose(stdin); 1249 return -1; 1250 } 1251 } else 1252 out = STDOUT_FILENO; 1253 1254 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1255 1256 (void)close(in); 1257 1258 /* 1259 * If there was an error, insize will be -1. 1260 * If we compressed to stdout, just return the size. 1261 * Otherwise stat the file and check it is the correct size. 1262 * We only blow away the file if we can stat the output and it 1263 * has the expected size. 1264 */ 1265 if (cflag != 0) 1266 return insize == -1 ? -1 : size; 1267 1268 #ifndef SMALL 1269 if (fstat(out, &osb) != 0) { 1270 maybe_warn("couldn't stat: %s", outfile); 1271 goto bad_outfile; 1272 } 1273 1274 if (osb.st_size != size) { 1275 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1276 " != %" PRIdOFF "), deleting", 1277 outfile, osb.st_size, size); 1278 goto bad_outfile; 1279 } 1280 1281 copymodes(out, &isb, outfile); 1282 #endif 1283 if (close(out) == -1) 1284 maybe_warn("couldn't close output"); 1285 1286 /* output is good, ok to delete input */ 1287 unlink_input(file, &isb); 1288 return size; 1289 1290 #ifndef SMALL 1291 bad_outfile: 1292 if (close(out) == -1) 1293 maybe_warn("couldn't close output"); 1294 1295 maybe_warnx("leaving original %s", file); 1296 unlink(outfile); 1297 return size; 1298 #endif 1299 } 1300 1301 /* uncompress the given file and remove the original */ 1302 static off_t 1303 file_uncompress(char *file, char *outfile, size_t outsize) 1304 { 1305 struct stat isb, osb; 1306 off_t size; 1307 ssize_t rbytes; 1308 unsigned char header1[4]; 1309 enum filetype method; 1310 int fd, ofd, zfd = -1; 1311 #ifndef SMALL 1312 ssize_t rv; 1313 time_t timestamp = 0; 1314 char name[PATH_MAX + 1]; 1315 #endif 1316 1317 /* gather the old name info */ 1318 1319 fd = open(file, O_RDONLY); 1320 if (fd < 0) { 1321 maybe_warn("can't open %s", file); 1322 goto lose; 1323 } 1324 1325 strlcpy(outfile, file, outsize); 1326 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1327 maybe_warnx("%s: unknown suffix -- ignored", file); 1328 goto lose; 1329 } 1330 1331 rbytes = read(fd, header1, sizeof header1); 1332 if (rbytes != sizeof header1) { 1333 /* we don't want to fail here. */ 1334 #ifndef SMALL 1335 if (fflag) 1336 goto lose; 1337 #endif 1338 if (rbytes == -1) 1339 maybe_warn("can't read %s", file); 1340 else 1341 goto unexpected_EOF; 1342 goto lose; 1343 } 1344 1345 method = file_gettype(header1); 1346 #ifndef SMALL 1347 if (fflag == 0 && method == FT_UNKNOWN) { 1348 maybe_warnx("%s: not in gzip format", file); 1349 goto lose; 1350 } 1351 1352 #endif 1353 1354 #ifndef SMALL 1355 if (method == FT_GZIP && Nflag) { 1356 unsigned char ts[4]; /* timestamp */ 1357 1358 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1359 if (rv >= 0 && rv < (ssize_t)(sizeof ts)) 1360 goto unexpected_EOF; 1361 if (rv == -1) { 1362 if (!fflag) 1363 maybe_warn("can't read %s", file); 1364 goto lose; 1365 } 1366 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1367 1368 if (header1[3] & ORIG_NAME) { 1369 rbytes = pread(fd, name, sizeof(name) - 1, GZIP_ORIGNAME); 1370 if (rbytes < 0) { 1371 maybe_warn("can't read %s", file); 1372 goto lose; 1373 } 1374 if (name[0] != '\0') { 1375 char *dp, *nf; 1376 1377 /* Make sure that name is NUL-terminated */ 1378 name[rbytes] = '\0'; 1379 1380 /* strip saved directory name */ 1381 nf = strrchr(name, '/'); 1382 if (nf == NULL) 1383 nf = name; 1384 else 1385 nf++; 1386 1387 /* preserve original directory name */ 1388 dp = strrchr(file, '/'); 1389 if (dp == NULL) 1390 dp = file; 1391 else 1392 dp++; 1393 snprintf(outfile, outsize, "%.*s%.*s", 1394 (int) (dp - file), 1395 file, (int) rbytes, nf); 1396 } 1397 } 1398 } 1399 #endif 1400 lseek(fd, 0, SEEK_SET); 1401 1402 if (cflag == 0 || lflag) { 1403 if (fstat(fd, &isb) != 0) 1404 goto lose; 1405 #ifndef SMALL 1406 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1407 maybe_warnx("%s has %d other links -- skipping", 1408 file, isb.st_nlink - 1); 1409 goto lose; 1410 } 1411 if (nflag == 0 && timestamp) 1412 isb.st_mtime = timestamp; 1413 if (check_outfile(outfile) == 0) 1414 goto lose; 1415 #endif 1416 } 1417 1418 if (cflag == 0 && lflag == 0) { 1419 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1420 if (zfd == STDOUT_FILENO) { 1421 /* We won't close STDOUT_FILENO later... */ 1422 zfd = dup(zfd); 1423 close(STDOUT_FILENO); 1424 } 1425 if (zfd == -1) { 1426 maybe_warn("can't open %s", outfile); 1427 goto lose; 1428 } 1429 } else 1430 zfd = STDOUT_FILENO; 1431 1432 switch (method) { 1433 #ifndef NO_BZIP2_SUPPORT 1434 case FT_BZIP2: 1435 /* XXX */ 1436 if (lflag) { 1437 maybe_warnx("no -l with bzip2 files"); 1438 goto lose; 1439 } 1440 1441 size = unbzip2(fd, zfd, NULL, 0, NULL); 1442 break; 1443 #endif 1444 1445 #ifndef NO_COMPRESS_SUPPORT 1446 case FT_Z: { 1447 FILE *in, *out; 1448 1449 /* XXX */ 1450 if (lflag) { 1451 maybe_warnx("no -l with Lempel-Ziv files"); 1452 goto lose; 1453 } 1454 1455 if ((in = zdopen(fd)) == NULL) { 1456 maybe_warn("zdopen for read: %s", file); 1457 goto lose; 1458 } 1459 1460 out = fdopen(dup(zfd), "w"); 1461 if (out == NULL) { 1462 maybe_warn("fdopen for write: %s", outfile); 1463 fclose(in); 1464 goto lose; 1465 } 1466 1467 size = zuncompress(in, out, NULL, 0, NULL); 1468 /* need to fclose() if ferror() is true... */ 1469 if (ferror(in) | fclose(in)) { 1470 maybe_warn("failed infile fclose"); 1471 unlink(outfile); 1472 (void)fclose(out); 1473 } 1474 if (fclose(out) != 0) { 1475 maybe_warn("failed outfile fclose"); 1476 unlink(outfile); 1477 goto lose; 1478 } 1479 break; 1480 } 1481 #endif 1482 1483 #ifndef NO_PACK_SUPPORT 1484 case FT_PACK: 1485 if (lflag) { 1486 maybe_warnx("no -l with packed files"); 1487 goto lose; 1488 } 1489 1490 size = unpack(fd, zfd, NULL, 0, NULL); 1491 break; 1492 #endif 1493 1494 #ifndef NO_XZ_SUPPORT 1495 case FT_XZ: 1496 if (lflag) { 1497 maybe_warnx("no -l with xz files"); 1498 goto lose; 1499 } 1500 1501 size = unxz(fd, zfd, NULL, 0, NULL); 1502 break; 1503 #endif 1504 1505 #ifndef SMALL 1506 case FT_UNKNOWN: 1507 if (lflag) { 1508 maybe_warnx("no -l for unknown filetypes"); 1509 goto lose; 1510 } 1511 size = cat_fd(NULL, 0, NULL, fd); 1512 break; 1513 #endif 1514 default: 1515 if (lflag) { 1516 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1517 close(fd); 1518 return -1; /* XXX */ 1519 } 1520 1521 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1522 break; 1523 } 1524 1525 if (close(fd) != 0) 1526 maybe_warn("couldn't close input"); 1527 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1528 maybe_warn("couldn't close output"); 1529 1530 if (size == -1) { 1531 if (cflag == 0) 1532 unlink(outfile); 1533 maybe_warnx("%s: uncompress failed", file); 1534 return -1; 1535 } 1536 1537 /* if testing, or we uncompressed to stdout, this is all we need */ 1538 #ifndef SMALL 1539 if (tflag) 1540 return size; 1541 #endif 1542 /* if we are uncompressing to stdin, don't remove the file. */ 1543 if (cflag) 1544 return size; 1545 1546 /* 1547 * if we create a file... 1548 */ 1549 /* 1550 * if we can't stat the file don't remove the file. 1551 */ 1552 1553 ofd = open(outfile, O_RDWR, 0); 1554 if (ofd == -1) { 1555 maybe_warn("couldn't open (leaving original): %s", 1556 outfile); 1557 return -1; 1558 } 1559 if (fstat(ofd, &osb) != 0) { 1560 maybe_warn("couldn't stat (leaving original): %s", 1561 outfile); 1562 close(ofd); 1563 return -1; 1564 } 1565 if (osb.st_size != size) { 1566 maybe_warnx("stat gave different size: %" PRIdOFF 1567 " != %" PRIdOFF " (leaving original)", 1568 size, osb.st_size); 1569 close(ofd); 1570 unlink(outfile); 1571 return -1; 1572 } 1573 unlink_input(file, &isb); 1574 #ifndef SMALL 1575 copymodes(ofd, &isb, outfile); 1576 #endif 1577 close(ofd); 1578 return size; 1579 1580 unexpected_EOF: 1581 maybe_warnx("%s: unexpected end of file", file); 1582 lose: 1583 if (fd != -1) 1584 close(fd); 1585 if (zfd != -1 && zfd != STDOUT_FILENO) 1586 close(fd); 1587 return -1; 1588 } 1589 1590 #ifndef SMALL 1591 static off_t 1592 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1593 { 1594 char buf[BUFLEN]; 1595 off_t in_tot; 1596 ssize_t w; 1597 1598 in_tot = count; 1599 w = write(STDOUT_FILENO, prepend, count); 1600 if (w == -1 || (size_t)w != count) { 1601 maybe_warn("write to stdout"); 1602 return -1; 1603 } 1604 for (;;) { 1605 ssize_t rv; 1606 1607 rv = read(fd, buf, sizeof buf); 1608 if (rv == 0) 1609 break; 1610 if (rv < 0) { 1611 maybe_warn("read from fd %d", fd); 1612 break; 1613 } 1614 1615 if (write(STDOUT_FILENO, buf, rv) != rv) { 1616 maybe_warn("write to stdout"); 1617 break; 1618 } 1619 in_tot += rv; 1620 } 1621 1622 if (gsizep) 1623 *gsizep = in_tot; 1624 return (in_tot); 1625 } 1626 #endif 1627 1628 static void 1629 handle_stdin(void) 1630 { 1631 unsigned char header1[4]; 1632 off_t usize, gsize; 1633 enum filetype method; 1634 ssize_t bytes_read; 1635 #ifndef NO_COMPRESS_SUPPORT 1636 FILE *in; 1637 #endif 1638 1639 #ifndef SMALL 1640 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1641 maybe_warnx("standard input is a terminal -- ignoring"); 1642 return; 1643 } 1644 #endif 1645 1646 if (lflag) { 1647 struct stat isb; 1648 1649 /* XXX could read the whole file, etc. */ 1650 if (fstat(STDIN_FILENO, &isb) < 0) { 1651 maybe_warn("fstat"); 1652 return; 1653 } 1654 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1655 return; 1656 } 1657 1658 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1659 if (bytes_read == -1) { 1660 maybe_warn("can't read stdin"); 1661 return; 1662 } else if (bytes_read != sizeof(header1)) { 1663 maybe_warnx("(stdin): unexpected end of file"); 1664 return; 1665 } 1666 1667 method = file_gettype(header1); 1668 switch (method) { 1669 default: 1670 #ifndef SMALL 1671 if (fflag == 0) { 1672 maybe_warnx("unknown compression format"); 1673 return; 1674 } 1675 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1676 break; 1677 #endif 1678 case FT_GZIP: 1679 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1680 (char *)header1, sizeof header1, &gsize, "(stdin)"); 1681 break; 1682 #ifndef NO_BZIP2_SUPPORT 1683 case FT_BZIP2: 1684 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1685 (char *)header1, sizeof header1, &gsize); 1686 break; 1687 #endif 1688 #ifndef NO_COMPRESS_SUPPORT 1689 case FT_Z: 1690 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1691 maybe_warnx("zopen of stdin"); 1692 return; 1693 } 1694 1695 usize = zuncompress(in, stdout, (char *)header1, 1696 sizeof header1, &gsize); 1697 fclose(in); 1698 break; 1699 #endif 1700 #ifndef NO_PACK_SUPPORT 1701 case FT_PACK: 1702 usize = unpack(STDIN_FILENO, STDOUT_FILENO, 1703 (char *)header1, sizeof header1, &gsize); 1704 break; 1705 #endif 1706 #ifndef NO_XZ_SUPPORT 1707 case FT_XZ: 1708 usize = unxz(STDIN_FILENO, STDOUT_FILENO, 1709 (char *)header1, sizeof header1, &gsize); 1710 break; 1711 #endif 1712 } 1713 1714 #ifndef SMALL 1715 if (vflag && !tflag && usize != -1 && gsize != -1) 1716 print_verbage(NULL, NULL, usize, gsize); 1717 if (vflag && tflag) 1718 print_test("(stdin)", usize != -1); 1719 #endif 1720 1721 } 1722 1723 static void 1724 handle_stdout(void) 1725 { 1726 off_t gsize, usize; 1727 struct stat sb; 1728 time_t systime; 1729 uint32_t mtime; 1730 int ret; 1731 1732 #ifndef SMALL 1733 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1734 maybe_warnx("standard output is a terminal -- ignoring"); 1735 return; 1736 } 1737 #endif 1738 /* If stdin is a file use its mtime, otherwise use current time */ 1739 ret = fstat(STDIN_FILENO, &sb); 1740 1741 #ifndef SMALL 1742 if (ret < 0) { 1743 maybe_warn("Can't stat stdin"); 1744 return; 1745 } 1746 #endif 1747 1748 if (S_ISREG(sb.st_mode)) 1749 mtime = (uint32_t)sb.st_mtime; 1750 else { 1751 systime = time(NULL); 1752 #ifndef SMALL 1753 if (systime == -1) { 1754 maybe_warn("time"); 1755 return; 1756 } 1757 #endif 1758 mtime = (uint32_t)systime; 1759 } 1760 1761 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1762 #ifndef SMALL 1763 if (vflag && !tflag && usize != -1 && gsize != -1) 1764 print_verbage(NULL, NULL, usize, gsize); 1765 #endif 1766 } 1767 1768 /* do what is asked for, for the path name */ 1769 static void 1770 handle_pathname(char *path) 1771 { 1772 char *opath = path, *s = NULL; 1773 ssize_t len; 1774 int slen; 1775 struct stat sb; 1776 1777 /* check for stdout/stdin */ 1778 if (path[0] == '-' && path[1] == '\0') { 1779 if (dflag) 1780 handle_stdin(); 1781 else 1782 handle_stdout(); 1783 return; 1784 } 1785 1786 retry: 1787 if (stat(path, &sb) != 0) { 1788 /* lets try <path>.gz if we're decompressing */ 1789 if (dflag && s == NULL && errno == ENOENT) { 1790 len = strlen(path); 1791 slen = suffixes[0].ziplen; 1792 s = malloc(len + slen + 1); 1793 if (s == NULL) 1794 maybe_err("malloc"); 1795 memcpy(s, path, len); 1796 memcpy(s + len, suffixes[0].zipped, slen + 1); 1797 path = s; 1798 goto retry; 1799 } 1800 maybe_warn("can't stat: %s", opath); 1801 goto out; 1802 } 1803 1804 if (S_ISDIR(sb.st_mode)) { 1805 #ifndef SMALL 1806 if (rflag) 1807 handle_dir(path); 1808 else 1809 #endif 1810 maybe_warnx("%s is a directory", path); 1811 goto out; 1812 } 1813 1814 if (S_ISREG(sb.st_mode)) 1815 handle_file(path, &sb); 1816 else 1817 maybe_warnx("%s is not a regular file", path); 1818 1819 out: 1820 if (s) 1821 free(s); 1822 } 1823 1824 /* compress/decompress a file */ 1825 static void 1826 handle_file(char *file, struct stat *sbp) 1827 { 1828 off_t usize, gsize; 1829 char outfile[PATH_MAX]; 1830 1831 infile = file; 1832 if (dflag) { 1833 usize = file_uncompress(file, outfile, sizeof(outfile)); 1834 #ifndef SMALL 1835 if (vflag && tflag) 1836 print_test(file, usize != -1); 1837 #endif 1838 if (usize == -1) 1839 return; 1840 gsize = sbp->st_size; 1841 } else { 1842 gsize = file_compress(file, outfile, sizeof(outfile)); 1843 if (gsize == -1) 1844 return; 1845 usize = sbp->st_size; 1846 } 1847 1848 1849 #ifndef SMALL 1850 if (vflag && !tflag) 1851 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1852 #endif 1853 } 1854 1855 #ifndef SMALL 1856 /* this is used with -r to recursively descend directories */ 1857 static void 1858 handle_dir(char *dir) 1859 { 1860 char *path_argv[2]; 1861 FTS *fts; 1862 FTSENT *entry; 1863 1864 path_argv[0] = dir; 1865 path_argv[1] = 0; 1866 fts = fts_open(path_argv, FTS_PHYSICAL, NULL); 1867 if (fts == NULL) { 1868 warn("couldn't fts_open %s", dir); 1869 return; 1870 } 1871 1872 while ((entry = fts_read(fts))) { 1873 switch(entry->fts_info) { 1874 case FTS_D: 1875 case FTS_DP: 1876 continue; 1877 1878 case FTS_DNR: 1879 case FTS_ERR: 1880 case FTS_NS: 1881 maybe_warn("%s", entry->fts_path); 1882 continue; 1883 case FTS_F: 1884 handle_file(entry->fts_name, entry->fts_statp); 1885 } 1886 } 1887 (void)fts_close(fts); 1888 } 1889 #endif 1890 1891 /* print a ratio - size reduction as a fraction of uncompressed size */ 1892 static void 1893 print_ratio(off_t in, off_t out, FILE *where) 1894 { 1895 int percent10; /* 10 * percent */ 1896 off_t diff; 1897 char buff[8]; 1898 int len; 1899 1900 diff = in - out/2; 1901 if (diff <= 0) 1902 /* 1903 * Output is more than double size of input! print -99.9% 1904 * Quite possibly we've failed to get the original size. 1905 */ 1906 percent10 = -999; 1907 else { 1908 /* 1909 * We only need 12 bits of result from the final division, 1910 * so reduce the values until a 32bit division will suffice. 1911 */ 1912 while (in > 0x100000) { 1913 diff >>= 1; 1914 in >>= 1; 1915 } 1916 if (in != 0) 1917 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1918 else 1919 percent10 = 0; 1920 } 1921 1922 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1923 /* Move the '.' to before the last digit */ 1924 buff[len - 1] = buff[len - 2]; 1925 buff[len - 2] = '.'; 1926 fprintf(where, "%5s%%", buff); 1927 } 1928 1929 #ifndef SMALL 1930 /* print compression statistics, and the new name (if there is one!) */ 1931 static void 1932 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1933 { 1934 if (file) 1935 fprintf(stderr, "%s:%s ", file, 1936 strlen(file) < 7 ? "\t\t" : "\t"); 1937 print_ratio(usize, gsize, stderr); 1938 if (nfile) 1939 fprintf(stderr, " -- replaced with %s", nfile); 1940 fprintf(stderr, "\n"); 1941 fflush(stderr); 1942 } 1943 1944 /* print test results */ 1945 static void 1946 print_test(const char *file, int ok) 1947 { 1948 1949 if (exit_value == 0 && ok == 0) 1950 exit_value = 1; 1951 fprintf(stderr, "%s:%s %s\n", file, 1952 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1953 fflush(stderr); 1954 } 1955 #endif 1956 1957 /* print a file's info ala --list */ 1958 /* eg: 1959 compressed uncompressed ratio uncompressed_name 1960 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1961 */ 1962 static void 1963 print_list(int fd, off_t out, const char *outfile, time_t ts) 1964 { 1965 static int first = 1; 1966 #ifndef SMALL 1967 static off_t in_tot, out_tot; 1968 uint32_t crc = 0; 1969 #endif 1970 off_t in = 0, rv; 1971 1972 if (first) { 1973 #ifndef SMALL 1974 if (vflag) 1975 printf("method crc date time "); 1976 #endif 1977 if (qflag == 0) 1978 printf(" compressed uncompressed " 1979 "ratio uncompressed_name\n"); 1980 } 1981 first = 0; 1982 1983 /* print totals? */ 1984 #ifndef SMALL 1985 if (fd == -1) { 1986 in = in_tot; 1987 out = out_tot; 1988 } else 1989 #endif 1990 { 1991 /* read the last 4 bytes - this is the uncompressed size */ 1992 rv = lseek(fd, (off_t)(-8), SEEK_END); 1993 if (rv != -1) { 1994 unsigned char buf[8]; 1995 uint32_t usize; 1996 1997 rv = read(fd, (char *)buf, sizeof(buf)); 1998 if (rv == -1) 1999 maybe_warn("read of uncompressed size"); 2000 else if (rv != sizeof(buf)) 2001 maybe_warnx("read of uncompressed size"); 2002 2003 else { 2004 usize = buf[4] | buf[5] << 8 | 2005 buf[6] << 16 | buf[7] << 24; 2006 in = (off_t)usize; 2007 #ifndef SMALL 2008 crc = buf[0] | buf[1] << 8 | 2009 buf[2] << 16 | buf[3] << 24; 2010 #endif 2011 } 2012 } 2013 } 2014 2015 #ifndef SMALL 2016 if (vflag && fd == -1) 2017 printf(" "); 2018 else if (vflag) { 2019 char *date = ctime(&ts); 2020 2021 /* skip the day, 1/100th second, and year */ 2022 date += 4; 2023 date[12] = 0; 2024 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 2025 } 2026 in_tot += in; 2027 out_tot += out; 2028 #endif 2029 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 2030 print_ratio(in, out, stdout); 2031 printf(" %s\n", outfile); 2032 } 2033 2034 /* display the usage of NetBSD gzip */ 2035 static void 2036 usage(void) 2037 { 2038 2039 fprintf(stderr, "%s\n", gzip_version); 2040 fprintf(stderr, 2041 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n" 2042 #ifndef SMALL 2043 " -1 --fast fastest (worst) compression\n" 2044 " -2 .. -8 set compression level\n" 2045 " -9 --best best (slowest) compression\n" 2046 " -c --stdout write to stdout, keep original files\n" 2047 " --to-stdout\n" 2048 " -d --decompress uncompress files\n" 2049 " --uncompress\n" 2050 " -f --force force overwriting & compress links\n" 2051 " -h --help display this help\n" 2052 " -k --keep don't delete input files during operation\n" 2053 " -l --list list compressed file contents\n" 2054 " -N --name save or restore original file name and time stamp\n" 2055 " -n --no-name don't save original file name or time stamp\n" 2056 " -q --quiet output no warnings\n" 2057 " -r --recursive recursively compress files in directories\n" 2058 " -S .suf use suffix .suf instead of .gz\n" 2059 " --suffix .suf\n" 2060 " -t --test test compressed file\n" 2061 " -V --version display program version\n" 2062 " -v --verbose print extra statistics\n", 2063 #else 2064 , 2065 #endif 2066 getprogname()); 2067 exit(0); 2068 } 2069 2070 /* display the version of NetBSD gzip */ 2071 static void 2072 display_version(void) 2073 { 2074 2075 fprintf(stderr, "%s\n", gzip_version); 2076 exit(0); 2077 } 2078 2079 #ifndef NO_BZIP2_SUPPORT 2080 #include "unbzip2.c" 2081 #endif 2082 #ifndef NO_COMPRESS_SUPPORT 2083 #include "zuncompress.c" 2084 #endif 2085 #ifndef NO_PACK_SUPPORT 2086 #include "unpack.c" 2087 #endif 2088 #ifndef NO_XZ_SUPPORT 2089 #include "unxz.c" 2090 #endif 2091 2092 static ssize_t 2093 read_retry(int fd, void *buf, size_t sz) 2094 { 2095 char *cp = buf; 2096 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2097 2098 while (left > 0) { 2099 ssize_t ret; 2100 2101 ret = read(fd, cp, left); 2102 if (ret == -1) { 2103 return ret; 2104 } else if (ret == 0) { 2105 break; /* EOF */ 2106 } 2107 cp += ret; 2108 left -= ret; 2109 } 2110 2111 return sz - left; 2112 } 2113
Indexes created Tue Sep 23 14:10:03 GMT 2025