gzip.c revision 1.103
1 /* $NetBSD: gzip.c,v 1.103 2011/06/21 13:25:45 joerg 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 1.103 2011/06/21 13:25:45 joerg 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 20101018"; 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, ...) 189 __attribute__((__format__(__printf__, 1, 2))); 190 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \ 191 !defined(NO_XZ_SUPPORT) 192 static void maybe_errx(const char *fmt, ...) 193 __attribute__((__format__(__printf__, 1, 2))); 194 #endif 195 static void maybe_warn(const char *fmt, ...) 196 __attribute__((__format__(__printf__, 1, 2))); 197 static void maybe_warnx(const char *fmt, ...) 198 __attribute__((__format__(__printf__, 1, 2))); 199 static enum filetype file_gettype(u_char *); 200 #ifdef SMALL 201 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 202 #endif 203 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 204 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 205 static off_t file_compress(char *, char *, size_t); 206 static off_t file_uncompress(char *, char *, size_t); 207 static void handle_pathname(char *); 208 static void handle_file(char *, struct stat *); 209 static void handle_stdin(void); 210 static void handle_stdout(void); 211 static void print_ratio(off_t, off_t, FILE *); 212 static void print_list(int fd, off_t, const char *, time_t); 213 static void usage(void); 214 static void display_version(void); 215 static const suffixes_t *check_suffix(char *, int); 216 static ssize_t read_retry(int, void *, size_t); 217 218 #ifdef SMALL 219 #define unlink_input(f, sb) unlink(f) 220 #else 221 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 222 static void prepend_gzip(char *, int *, char ***); 223 static void handle_dir(char *); 224 static void print_verbage(const char *, const char *, off_t, off_t); 225 static void print_test(const char *, int); 226 static void copymodes(int fd, const struct stat *, const char *file); 227 static int check_outfile(const char *outfile); 228 #endif 229 230 #ifndef NO_BZIP2_SUPPORT 231 static off_t unbzip2(int, int, char *, size_t, off_t *); 232 #endif 233 234 #ifndef NO_COMPRESS_SUPPORT 235 static FILE *zdopen(int); 236 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 237 #endif 238 239 #ifndef NO_PACK_SUPPORT 240 static off_t unpack(int, int, char *, size_t, off_t *); 241 #endif 242 243 #ifndef NO_XZ_SUPPORT 244 static off_t unxz(int, int, char *, size_t, off_t *); 245 #endif 246 247 int main(int, char *p[]); 248 249 #ifdef SMALL 250 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 251 #else 252 static const struct option longopts[] = { 253 { "stdout", no_argument, 0, 'c' }, 254 { "to-stdout", no_argument, 0, 'c' }, 255 { "decompress", no_argument, 0, 'd' }, 256 { "uncompress", no_argument, 0, 'd' }, 257 { "force", no_argument, 0, 'f' }, 258 { "help", no_argument, 0, 'h' }, 259 { "keep", no_argument, 0, 'k' }, 260 { "list", no_argument, 0, 'l' }, 261 { "no-name", no_argument, 0, 'n' }, 262 { "name", no_argument, 0, 'N' }, 263 { "quiet", no_argument, 0, 'q' }, 264 { "recursive", no_argument, 0, 'r' }, 265 { "suffix", required_argument, 0, 'S' }, 266 { "test", no_argument, 0, 't' }, 267 { "verbose", no_argument, 0, 'v' }, 268 { "version", no_argument, 0, 'V' }, 269 { "fast", no_argument, 0, '1' }, 270 { "best", no_argument, 0, '9' }, 271 #if 0 272 /* 273 * This is what else GNU gzip implements. --ascii isn't useful 274 * on NetBSD, and I don't care to have a --license. 275 */ 276 { "ascii", no_argument, 0, 'a' }, 277 { "license", no_argument, 0, 'L' }, 278 #endif 279 { NULL, no_argument, 0, 0 }, 280 }; 281 #endif 282 283 int 284 main(int argc, char **argv) 285 { 286 const char *progname = getprogname(); 287 #ifndef SMALL 288 char *gzip; 289 int len; 290 #endif 291 int ch; 292 293 /* XXX set up signals */ 294 295 #ifndef SMALL 296 if ((gzip = getenv("GZIP")) != NULL) 297 prepend_gzip(gzip, &argc, &argv); 298 #endif 299 300 /* 301 * XXX 302 * handle being called `gunzip', `zcat' and `gzcat' 303 */ 304 if (strcmp(progname, "gunzip") == 0) 305 dflag = 1; 306 else if (strcmp(progname, "zcat") == 0 || 307 strcmp(progname, "gzcat") == 0) 308 dflag = cflag = 1; 309 310 #ifdef SMALL 311 #define OPT_LIST "123456789cdhlV" 312 #else 313 #define OPT_LIST "123456789cdfhklNnqrS:tVv" 314 #endif 315 316 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 317 switch (ch) { 318 case '1': case '2': case '3': 319 case '4': case '5': case '6': 320 case '7': case '8': case '9': 321 numflag = ch - '0'; 322 break; 323 case 'c': 324 cflag = 1; 325 break; 326 case 'd': 327 dflag = 1; 328 break; 329 case 'l': 330 lflag = 1; 331 dflag = 1; 332 break; 333 case 'V': 334 display_version(); 335 /* NOTREACHED */ 336 #ifndef SMALL 337 case 'f': 338 fflag = 1; 339 break; 340 case 'k': 341 kflag = 1; 342 break; 343 case 'N': 344 nflag = 0; 345 Nflag = 1; 346 break; 347 case 'n': 348 nflag = 1; 349 Nflag = 0; 350 break; 351 case 'q': 352 qflag = 1; 353 break; 354 case 'r': 355 rflag = 1; 356 break; 357 case 'S': 358 len = strlen(optarg); 359 if (len != 0) { 360 if (len > SUFFIX_MAXLEN) 361 errx(1, "incorrect suffix: '%s'", optarg); 362 suffixes[0].zipped = optarg; 363 suffixes[0].ziplen = len; 364 } else { 365 suffixes[NUM_SUFFIXES - 1].zipped = ""; 366 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 367 } 368 break; 369 case 't': 370 cflag = 1; 371 tflag = 1; 372 dflag = 1; 373 break; 374 case 'v': 375 vflag = 1; 376 break; 377 #endif 378 default: 379 usage(); 380 /* NOTREACHED */ 381 } 382 } 383 argv += optind; 384 argc -= optind; 385 386 if (argc == 0) { 387 if (dflag) /* stdin mode */ 388 handle_stdin(); 389 else /* stdout mode */ 390 handle_stdout(); 391 } else { 392 do { 393 handle_pathname(argv[0]); 394 } while (*++argv); 395 } 396 #ifndef SMALL 397 if (qflag == 0 && lflag && argc > 1) 398 print_list(-1, 0, "(totals)", 0); 399 #endif 400 exit(exit_value); 401 } 402 403 /* maybe print a warning */ 404 void 405 maybe_warn(const char *fmt, ...) 406 { 407 va_list ap; 408 409 if (qflag == 0) { 410 va_start(ap, fmt); 411 vwarn(fmt, ap); 412 va_end(ap); 413 } 414 if (exit_value == 0) 415 exit_value = 1; 416 } 417 418 /* ... without an errno. */ 419 void 420 maybe_warnx(const char *fmt, ...) 421 { 422 va_list ap; 423 424 if (qflag == 0) { 425 va_start(ap, fmt); 426 vwarnx(fmt, ap); 427 va_end(ap); 428 } 429 if (exit_value == 0) 430 exit_value = 1; 431 } 432 433 /* maybe print an error */ 434 void 435 maybe_err(const char *fmt, ...) 436 { 437 va_list ap; 438 439 if (qflag == 0) { 440 va_start(ap, fmt); 441 vwarn(fmt, ap); 442 va_end(ap); 443 } 444 exit(2); 445 } 446 447 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \ 448 !defined(NO_XZ_SUPPORT) 449 /* ... without an errno. */ 450 void 451 maybe_errx(const char *fmt, ...) 452 { 453 va_list ap; 454 455 if (qflag == 0) { 456 va_start(ap, fmt); 457 vwarnx(fmt, ap); 458 va_end(ap); 459 } 460 exit(2); 461 } 462 #endif 463 464 #ifndef SMALL 465 /* split up $GZIP and prepend it to the argument list */ 466 static void 467 prepend_gzip(char *gzip, int *argc, char ***argv) 468 { 469 char *s, **nargv, **ac; 470 int nenvarg = 0, i; 471 472 /* scan how many arguments there are */ 473 for (s = gzip;;) { 474 while (*s == ' ' || *s == '\t') 475 s++; 476 if (*s == 0) 477 goto count_done; 478 nenvarg++; 479 while (*s != ' ' && *s != '\t') 480 if (*s++ == 0) 481 goto count_done; 482 } 483 count_done: 484 /* punt early */ 485 if (nenvarg == 0) 486 return; 487 488 *argc += nenvarg; 489 ac = *argv; 490 491 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 492 if (nargv == NULL) 493 maybe_err("malloc"); 494 495 /* stash this away */ 496 *argv = nargv; 497 498 /* copy the program name first */ 499 i = 0; 500 nargv[i++] = *(ac++); 501 502 /* take a copy of $GZIP and add it to the array */ 503 s = strdup(gzip); 504 if (s == NULL) 505 maybe_err("strdup"); 506 for (;;) { 507 /* Skip whitespaces. */ 508 while (*s == ' ' || *s == '\t') 509 s++; 510 if (*s == 0) 511 goto copy_done; 512 nargv[i++] = s; 513 /* Find the end of this argument. */ 514 while (*s != ' ' && *s != '\t') 515 if (*s++ == 0) 516 /* Argument followed by NUL. */ 517 goto copy_done; 518 /* Terminate by overwriting ' ' or '\t' with NUL. */ 519 *s++ = 0; 520 } 521 copy_done: 522 523 /* copy the original arguments and a NULL */ 524 while (*ac) 525 nargv[i++] = *(ac++); 526 nargv[i] = NULL; 527 } 528 #endif 529 530 /* compress input to output. Return bytes read, -1 on error */ 531 static off_t 532 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 533 { 534 z_stream z; 535 char *outbufp, *inbufp; 536 off_t in_tot = 0, out_tot = 0; 537 ssize_t in_size; 538 int i, error; 539 uLong crc; 540 #ifdef SMALL 541 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 542 0, 0, 0, 0, 543 0, OS_CODE }; 544 #endif 545 546 outbufp = malloc(BUFLEN); 547 inbufp = malloc(BUFLEN); 548 if (outbufp == NULL || inbufp == NULL) { 549 maybe_err("malloc failed"); 550 goto out; 551 } 552 553 memset(&z, 0, sizeof z); 554 z.zalloc = Z_NULL; 555 z.zfree = Z_NULL; 556 z.opaque = 0; 557 558 #ifdef SMALL 559 memcpy(outbufp, header, sizeof header); 560 i = sizeof header; 561 #else 562 if (nflag != 0) { 563 mtime = 0; 564 origname = ""; 565 } 566 567 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 568 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 569 *origname ? ORIG_NAME : 0, 570 mtime & 0xff, 571 (mtime >> 8) & 0xff, 572 (mtime >> 16) & 0xff, 573 (mtime >> 24) & 0xff, 574 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 575 OS_CODE, origname); 576 if (i >= BUFLEN) 577 /* this need PATH_MAX > BUFLEN ... */ 578 maybe_err("snprintf"); 579 if (*origname) 580 i++; 581 #endif 582 583 z.next_out = (unsigned char *)outbufp + i; 584 z.avail_out = BUFLEN - i; 585 586 error = deflateInit2(&z, numflag, Z_DEFLATED, 587 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 588 if (error != Z_OK) { 589 maybe_warnx("deflateInit2 failed"); 590 in_tot = -1; 591 goto out; 592 } 593 594 crc = crc32(0L, Z_NULL, 0); 595 for (;;) { 596 if (z.avail_out == 0) { 597 if (write(out, outbufp, BUFLEN) != BUFLEN) { 598 maybe_warn("write"); 599 out_tot = -1; 600 goto out; 601 } 602 603 out_tot += BUFLEN; 604 z.next_out = (unsigned char *)outbufp; 605 z.avail_out = BUFLEN; 606 } 607 608 if (z.avail_in == 0) { 609 in_size = read(in, inbufp, BUFLEN); 610 if (in_size < 0) { 611 maybe_warn("read"); 612 in_tot = -1; 613 goto out; 614 } 615 if (in_size == 0) 616 break; 617 618 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 619 in_tot += in_size; 620 z.next_in = (unsigned char *)inbufp; 621 z.avail_in = in_size; 622 } 623 624 error = deflate(&z, Z_NO_FLUSH); 625 if (error != Z_OK && error != Z_STREAM_END) { 626 maybe_warnx("deflate failed"); 627 in_tot = -1; 628 goto out; 629 } 630 } 631 632 /* clean up */ 633 for (;;) { 634 size_t len; 635 ssize_t w; 636 637 error = deflate(&z, Z_FINISH); 638 if (error != Z_OK && error != Z_STREAM_END) { 639 maybe_warnx("deflate failed"); 640 in_tot = -1; 641 goto out; 642 } 643 644 len = (char *)z.next_out - outbufp; 645 646 w = write(out, outbufp, len); 647 if (w == -1 || (size_t)w != len) { 648 maybe_warn("write"); 649 out_tot = -1; 650 goto out; 651 } 652 out_tot += len; 653 z.next_out = (unsigned char *)outbufp; 654 z.avail_out = BUFLEN; 655 656 if (error == Z_STREAM_END) 657 break; 658 } 659 660 if (deflateEnd(&z) != Z_OK) { 661 maybe_warnx("deflateEnd failed"); 662 in_tot = -1; 663 goto out; 664 } 665 666 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 667 (int)crc & 0xff, 668 (int)(crc >> 8) & 0xff, 669 (int)(crc >> 16) & 0xff, 670 (int)(crc >> 24) & 0xff, 671 (int)in_tot & 0xff, 672 (int)(in_tot >> 8) & 0xff, 673 (int)(in_tot >> 16) & 0xff, 674 (int)(in_tot >> 24) & 0xff); 675 if (i != 8) 676 maybe_err("snprintf"); 677 #if 0 678 if (in_tot > 0xffffffff) 679 maybe_warn("input file size >= 4GB cannot be saved"); 680 #endif 681 if (write(out, outbufp, i) != i) { 682 maybe_warn("write"); 683 in_tot = -1; 684 } else 685 out_tot += i; 686 687 out: 688 if (inbufp != NULL) 689 free(inbufp); 690 if (outbufp != NULL) 691 free(outbufp); 692 if (gsizep) 693 *gsizep = out_tot; 694 return in_tot; 695 } 696 697 /* 698 * uncompress input to output then close the input. return the 699 * uncompressed size written, and put the compressed sized read 700 * into `*gsizep'. 701 */ 702 static off_t 703 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 704 const char *filename) 705 { 706 z_stream z; 707 char *outbufp, *inbufp; 708 off_t out_tot = -1, in_tot = 0; 709 uint32_t out_sub_tot = 0; 710 enum { 711 GZSTATE_MAGIC0, 712 GZSTATE_MAGIC1, 713 GZSTATE_METHOD, 714 GZSTATE_FLAGS, 715 GZSTATE_SKIPPING, 716 GZSTATE_EXTRA, 717 GZSTATE_EXTRA2, 718 GZSTATE_EXTRA3, 719 GZSTATE_ORIGNAME, 720 GZSTATE_COMMENT, 721 GZSTATE_HEAD_CRC1, 722 GZSTATE_HEAD_CRC2, 723 GZSTATE_INIT, 724 GZSTATE_READ, 725 GZSTATE_CRC, 726 GZSTATE_LEN, 727 } state = GZSTATE_MAGIC0; 728 int flags = 0, skip_count = 0; 729 int error = Z_STREAM_ERROR, done_reading = 0; 730 uLong crc = 0; 731 ssize_t wr; 732 int needmore = 0; 733 734 #define ADVANCE() { z.next_in++; z.avail_in--; } 735 736 if ((outbufp = malloc(BUFLEN)) == NULL) { 737 maybe_err("malloc failed"); 738 goto out2; 739 } 740 if ((inbufp = malloc(BUFLEN)) == NULL) { 741 maybe_err("malloc failed"); 742 goto out1; 743 } 744 745 memset(&z, 0, sizeof z); 746 z.avail_in = prelen; 747 z.next_in = (unsigned char *)pre; 748 z.avail_out = BUFLEN; 749 z.next_out = (unsigned char *)outbufp; 750 z.zalloc = NULL; 751 z.zfree = NULL; 752 z.opaque = 0; 753 754 in_tot = prelen; 755 out_tot = 0; 756 757 for (;;) { 758 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 759 ssize_t in_size; 760 761 if (z.avail_in > 0) { 762 memmove(inbufp, z.next_in, z.avail_in); 763 } 764 z.next_in = (unsigned char *)inbufp; 765 in_size = read(in, z.next_in + z.avail_in, 766 BUFLEN - z.avail_in); 767 768 if (in_size == -1) { 769 maybe_warn("failed to read stdin"); 770 goto stop_and_fail; 771 } else if (in_size == 0) { 772 done_reading = 1; 773 } 774 775 z.avail_in += in_size; 776 needmore = 0; 777 778 in_tot += in_size; 779 } 780 if (z.avail_in == 0) { 781 if (done_reading && state != GZSTATE_MAGIC0) { 782 maybe_warnx("%s: unexpected end of file", 783 filename); 784 goto stop_and_fail; 785 } 786 goto stop; 787 } 788 switch (state) { 789 case GZSTATE_MAGIC0: 790 if (*z.next_in != GZIP_MAGIC0) { 791 if (in_tot > 0) { 792 maybe_warnx("%s: trailing garbage " 793 "ignored", filename); 794 goto stop; 795 } 796 maybe_warnx("input not gziped (MAGIC0)"); 797 goto stop_and_fail; 798 } 799 ADVANCE(); 800 state++; 801 out_sub_tot = 0; 802 crc = crc32(0L, Z_NULL, 0); 803 break; 804 805 case GZSTATE_MAGIC1: 806 if (*z.next_in != GZIP_MAGIC1 && 807 *z.next_in != GZIP_OMAGIC1) { 808 maybe_warnx("input not gziped (MAGIC1)"); 809 goto stop_and_fail; 810 } 811 ADVANCE(); 812 state++; 813 break; 814 815 case GZSTATE_METHOD: 816 if (*z.next_in != Z_DEFLATED) { 817 maybe_warnx("unknown compression method"); 818 goto stop_and_fail; 819 } 820 ADVANCE(); 821 state++; 822 break; 823 824 case GZSTATE_FLAGS: 825 flags = *z.next_in; 826 ADVANCE(); 827 skip_count = 6; 828 state++; 829 break; 830 831 case GZSTATE_SKIPPING: 832 if (skip_count > 0) { 833 skip_count--; 834 ADVANCE(); 835 } else 836 state++; 837 break; 838 839 case GZSTATE_EXTRA: 840 if ((flags & EXTRA_FIELD) == 0) { 841 state = GZSTATE_ORIGNAME; 842 break; 843 } 844 skip_count = *z.next_in; 845 ADVANCE(); 846 state++; 847 break; 848 849 case GZSTATE_EXTRA2: 850 skip_count |= ((*z.next_in) << 8); 851 ADVANCE(); 852 state++; 853 break; 854 855 case GZSTATE_EXTRA3: 856 if (skip_count > 0) { 857 skip_count--; 858 ADVANCE(); 859 } else 860 state++; 861 break; 862 863 case GZSTATE_ORIGNAME: 864 if ((flags & ORIG_NAME) == 0) { 865 state++; 866 break; 867 } 868 if (*z.next_in == 0) 869 state++; 870 ADVANCE(); 871 break; 872 873 case GZSTATE_COMMENT: 874 if ((flags & COMMENT) == 0) { 875 state++; 876 break; 877 } 878 if (*z.next_in == 0) 879 state++; 880 ADVANCE(); 881 break; 882 883 case GZSTATE_HEAD_CRC1: 884 if (flags & HEAD_CRC) 885 skip_count = 2; 886 else 887 skip_count = 0; 888 state++; 889 break; 890 891 case GZSTATE_HEAD_CRC2: 892 if (skip_count > 0) { 893 skip_count--; 894 ADVANCE(); 895 } else 896 state++; 897 break; 898 899 case GZSTATE_INIT: 900 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 901 maybe_warnx("failed to inflateInit"); 902 goto stop_and_fail; 903 } 904 state++; 905 break; 906 907 case GZSTATE_READ: 908 error = inflate(&z, Z_FINISH); 909 switch (error) { 910 /* Z_BUF_ERROR goes with Z_FINISH... */ 911 case Z_BUF_ERROR: 912 if (z.avail_out > 0 && !done_reading) 913 continue; 914 915 case Z_STREAM_END: 916 case Z_OK: 917 break; 918 919 case Z_NEED_DICT: 920 maybe_warnx("Z_NEED_DICT error"); 921 goto stop_and_fail; 922 case Z_DATA_ERROR: 923 maybe_warnx("data stream error"); 924 goto stop_and_fail; 925 case Z_STREAM_ERROR: 926 maybe_warnx("internal stream error"); 927 goto stop_and_fail; 928 case Z_MEM_ERROR: 929 maybe_warnx("memory allocation error"); 930 goto stop_and_fail; 931 932 default: 933 maybe_warn("unknown error from inflate(): %d", 934 error); 935 } 936 wr = BUFLEN - z.avail_out; 937 938 if (wr != 0) { 939 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 940 if ( 941 #ifndef SMALL 942 /* don't write anything with -t */ 943 tflag == 0 && 944 #endif 945 write(out, outbufp, wr) != wr) { 946 maybe_warn("error writing to output"); 947 goto stop_and_fail; 948 } 949 950 out_tot += wr; 951 out_sub_tot += wr; 952 } 953 954 if (error == Z_STREAM_END) { 955 inflateEnd(&z); 956 state++; 957 } 958 959 z.next_out = (unsigned char *)outbufp; 960 z.avail_out = BUFLEN; 961 962 break; 963 case GZSTATE_CRC: 964 { 965 uLong origcrc; 966 967 if (z.avail_in < 4) { 968 if (!done_reading) { 969 needmore = 1; 970 continue; 971 } 972 maybe_warnx("truncated input"); 973 goto stop_and_fail; 974 } 975 origcrc = ((unsigned)z.next_in[0] & 0xff) | 976 ((unsigned)z.next_in[1] & 0xff) << 8 | 977 ((unsigned)z.next_in[2] & 0xff) << 16 | 978 ((unsigned)z.next_in[3] & 0xff) << 24; 979 if (origcrc != crc) { 980 maybe_warnx("invalid compressed" 981 " data--crc error"); 982 goto stop_and_fail; 983 } 984 } 985 986 z.avail_in -= 4; 987 z.next_in += 4; 988 989 if (!z.avail_in && done_reading) { 990 goto stop; 991 } 992 state++; 993 break; 994 case GZSTATE_LEN: 995 { 996 uLong origlen; 997 998 if (z.avail_in < 4) { 999 if (!done_reading) { 1000 needmore = 1; 1001 continue; 1002 } 1003 maybe_warnx("truncated input"); 1004 goto stop_and_fail; 1005 } 1006 origlen = ((unsigned)z.next_in[0] & 0xff) | 1007 ((unsigned)z.next_in[1] & 0xff) << 8 | 1008 ((unsigned)z.next_in[2] & 0xff) << 16 | 1009 ((unsigned)z.next_in[3] & 0xff) << 24; 1010 1011 if (origlen != out_sub_tot) { 1012 maybe_warnx("invalid compressed" 1013 " data--length error"); 1014 goto stop_and_fail; 1015 } 1016 } 1017 1018 z.avail_in -= 4; 1019 z.next_in += 4; 1020 1021 if (error < 0) { 1022 maybe_warnx("decompression error"); 1023 goto stop_and_fail; 1024 } 1025 state = GZSTATE_MAGIC0; 1026 break; 1027 } 1028 continue; 1029 stop_and_fail: 1030 out_tot = -1; 1031 stop: 1032 break; 1033 } 1034 if (state > GZSTATE_INIT) 1035 inflateEnd(&z); 1036 1037 free(inbufp); 1038 out1: 1039 free(outbufp); 1040 out2: 1041 if (gsizep) 1042 *gsizep = in_tot; 1043 return (out_tot); 1044 } 1045 1046 #ifndef SMALL 1047 /* 1048 * set the owner, mode, flags & utimes using the given file descriptor. 1049 * file is only used in possible warning messages. 1050 */ 1051 static void 1052 copymodes(int fd, const struct stat *sbp, const char *file) 1053 { 1054 struct timeval times[2]; 1055 struct stat sb; 1056 1057 /* 1058 * If we have no info on the input, give this file some 1059 * default values and return.. 1060 */ 1061 if (sbp == NULL) { 1062 mode_t mask = umask(022); 1063 1064 (void)fchmod(fd, DEFFILEMODE & ~mask); 1065 (void)umask(mask); 1066 return; 1067 } 1068 sb = *sbp; 1069 1070 /* if the chown fails, remove set-id bits as-per compress(1) */ 1071 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1072 if (errno != EPERM) 1073 maybe_warn("couldn't fchown: %s", file); 1074 sb.st_mode &= ~(S_ISUID|S_ISGID); 1075 } 1076 1077 /* we only allow set-id and the 9 normal permission bits */ 1078 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1079 if (fchmod(fd, sb.st_mode) < 0) 1080 maybe_warn("couldn't fchmod: %s", file); 1081 1082 /* only try flags if they exist already */ 1083 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1084 maybe_warn("couldn't fchflags: %s", file); 1085 1086 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1087 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1088 if (futimes(fd, times) < 0) 1089 maybe_warn("couldn't utimes: %s", file); 1090 } 1091 #endif 1092 1093 /* what sort of file is this? */ 1094 static enum filetype 1095 file_gettype(u_char *buf) 1096 { 1097 1098 if (buf[0] == GZIP_MAGIC0 && 1099 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1100 return FT_GZIP; 1101 else 1102 #ifndef NO_BZIP2_SUPPORT 1103 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1104 buf[3] >= '0' && buf[3] <= '9') 1105 return FT_BZIP2; 1106 else 1107 #endif 1108 #ifndef NO_COMPRESS_SUPPORT 1109 if (memcmp(buf, Z_MAGIC, 2) == 0) 1110 return FT_Z; 1111 else 1112 #endif 1113 #ifndef NO_PACK_SUPPORT 1114 if (memcmp(buf, PACK_MAGIC, 2) == 0) 1115 return FT_PACK; 1116 else 1117 #endif 1118 #ifndef NO_XZ_SUPPORT 1119 if (memcmp(buf, XZ_MAGIC, 4) == 0) /* XXX: We only have 4 bytes */ 1120 return FT_XZ; 1121 else 1122 #endif 1123 return FT_UNKNOWN; 1124 } 1125 1126 #ifndef SMALL 1127 /* check the outfile is OK. */ 1128 static int 1129 check_outfile(const char *outfile) 1130 { 1131 struct stat sb; 1132 int ok = 1; 1133 1134 if (lflag == 0 && stat(outfile, &sb) == 0) { 1135 if (fflag) 1136 unlink(outfile); 1137 else if (isatty(STDIN_FILENO)) { 1138 char ans[10] = { 'n', '\0' }; /* default */ 1139 1140 fprintf(stderr, "%s already exists -- do you wish to " 1141 "overwrite (y or n)? " , outfile); 1142 (void)fgets(ans, sizeof(ans) - 1, stdin); 1143 if (ans[0] != 'y' && ans[0] != 'Y') { 1144 fprintf(stderr, "\tnot overwriting\n"); 1145 ok = 0; 1146 } else 1147 unlink(outfile); 1148 } else { 1149 maybe_warnx("%s already exists -- skipping", outfile); 1150 ok = 0; 1151 } 1152 } 1153 return ok; 1154 } 1155 1156 static void 1157 unlink_input(const char *file, const struct stat *sb) 1158 { 1159 struct stat nsb; 1160 1161 if (kflag) 1162 return; 1163 if (stat(file, &nsb) != 0) 1164 /* Must be gone already */ 1165 return; 1166 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1167 /* Definitely a different file */ 1168 return; 1169 unlink(file); 1170 } 1171 #endif 1172 1173 static const suffixes_t * 1174 check_suffix(char *file, int xlate) 1175 { 1176 const suffixes_t *s; 1177 int len = strlen(file); 1178 char *sp; 1179 1180 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1181 /* if it doesn't fit in "a.suf", don't bother */ 1182 if (s->ziplen >= len) 1183 continue; 1184 sp = file + len - s->ziplen; 1185 if (strcmp(s->zipped, sp) != 0) 1186 continue; 1187 if (xlate) 1188 strcpy(sp, s->normal); 1189 return s; 1190 } 1191 return NULL; 1192 } 1193 1194 /* 1195 * compress the given file: create a corresponding .gz file and remove the 1196 * original. 1197 */ 1198 static off_t 1199 file_compress(char *file, char *outfile, size_t outsize) 1200 { 1201 int in; 1202 int out; 1203 off_t size, insize; 1204 #ifndef SMALL 1205 struct stat isb, osb; 1206 const suffixes_t *suff; 1207 #endif 1208 1209 in = open(file, O_RDONLY); 1210 if (in == -1) { 1211 maybe_warn("can't open %s", file); 1212 return -1; 1213 } 1214 1215 if (cflag == 0) { 1216 #ifndef SMALL 1217 if (fstat(in, &isb) == 0) { 1218 if (isb.st_nlink > 1 && fflag == 0) { 1219 maybe_warnx("%s has %d other link%s -- " 1220 "skipping", file, isb.st_nlink - 1, 1221 isb.st_nlink == 1 ? "" : "s"); 1222 close(in); 1223 return -1; 1224 } 1225 } 1226 1227 if (fflag == 0 && (suff = check_suffix(file, 0)) 1228 && suff->zipped[0] != 0) { 1229 maybe_warnx("%s already has %s suffix -- unchanged", 1230 file, suff->zipped); 1231 close(in); 1232 return -1; 1233 } 1234 #endif 1235 1236 /* Add (usually) .gz to filename */ 1237 if ((size_t)snprintf(outfile, outsize, "%s%s", 1238 file, suffixes[0].zipped) >= outsize) 1239 memcpy(outfile + outsize - suffixes[0].ziplen - 1, 1240 suffixes[0].zipped, suffixes[0].ziplen + 1); 1241 1242 #ifndef SMALL 1243 if (check_outfile(outfile) == 0) { 1244 close(in); 1245 return -1; 1246 } 1247 #endif 1248 } 1249 1250 if (cflag == 0) { 1251 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1252 if (out == -1) { 1253 maybe_warn("could not create output: %s", outfile); 1254 fclose(stdin); 1255 return -1; 1256 } 1257 } else 1258 out = STDOUT_FILENO; 1259 1260 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1261 1262 (void)close(in); 1263 1264 /* 1265 * If there was an error, insize will be -1. 1266 * If we compressed to stdout, just return the size. 1267 * Otherwise stat the file and check it is the correct size. 1268 * We only blow away the file if we can stat the output and it 1269 * has the expected size. 1270 */ 1271 if (cflag != 0) 1272 return insize == -1 ? -1 : size; 1273 1274 #ifndef SMALL 1275 if (fstat(out, &osb) != 0) { 1276 maybe_warn("couldn't stat: %s", outfile); 1277 goto bad_outfile; 1278 } 1279 1280 if (osb.st_size != size) { 1281 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1282 " != %" PRIdOFF "), deleting", 1283 outfile, osb.st_size, size); 1284 goto bad_outfile; 1285 } 1286 1287 copymodes(out, &isb, outfile); 1288 #endif 1289 if (close(out) == -1) 1290 maybe_warn("couldn't close output"); 1291 1292 /* output is good, ok to delete input */ 1293 unlink_input(file, &isb); 1294 return size; 1295 1296 #ifndef SMALL 1297 bad_outfile: 1298 if (close(out) == -1) 1299 maybe_warn("couldn't close output"); 1300 1301 maybe_warnx("leaving original %s", file); 1302 unlink(outfile); 1303 return size; 1304 #endif 1305 } 1306 1307 /* uncompress the given file and remove the original */ 1308 static off_t 1309 file_uncompress(char *file, char *outfile, size_t outsize) 1310 { 1311 struct stat isb, osb; 1312 off_t size; 1313 ssize_t rbytes; 1314 unsigned char header1[4]; 1315 enum filetype method; 1316 int fd, ofd, zfd = -1; 1317 #ifndef SMALL 1318 ssize_t rv; 1319 time_t timestamp = 0; 1320 unsigned char name[PATH_MAX + 1]; 1321 #endif 1322 1323 /* gather the old name info */ 1324 1325 fd = open(file, O_RDONLY); 1326 if (fd < 0) { 1327 maybe_warn("can't open %s", file); 1328 goto lose; 1329 } 1330 1331 strlcpy(outfile, file, outsize); 1332 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1333 maybe_warnx("%s: unknown suffix -- ignored", file); 1334 goto lose; 1335 } 1336 1337 rbytes = read(fd, header1, sizeof header1); 1338 if (rbytes != sizeof header1) { 1339 /* we don't want to fail here. */ 1340 #ifndef SMALL 1341 if (fflag) 1342 goto lose; 1343 #endif 1344 if (rbytes == -1) 1345 maybe_warn("can't read %s", file); 1346 else 1347 goto unexpected_EOF; 1348 goto lose; 1349 } 1350 1351 method = file_gettype(header1); 1352 #ifndef SMALL 1353 if (fflag == 0 && method == FT_UNKNOWN) { 1354 maybe_warnx("%s: not in gzip format", file); 1355 goto lose; 1356 } 1357 1358 #endif 1359 1360 #ifndef SMALL 1361 if (method == FT_GZIP && Nflag) { 1362 unsigned char ts[4]; /* timestamp */ 1363 1364 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1365 if (rv >= 0 && rv < (ssize_t)(sizeof ts)) 1366 goto unexpected_EOF; 1367 if (rv == -1) { 1368 if (!fflag) 1369 maybe_warn("can't read %s", file); 1370 goto lose; 1371 } 1372 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1373 1374 if (header1[3] & ORIG_NAME) { 1375 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1376 if (rbytes < 0) { 1377 maybe_warn("can't read %s", file); 1378 goto lose; 1379 } 1380 if (name[0] != 0) { 1381 /* preserve original directory name */ 1382 char *dp = strrchr(file, '/'); 1383 if (dp == NULL) 1384 dp = file; 1385 else 1386 dp++; 1387 snprintf(outfile, outsize, "%.*s%.*s", 1388 (int) (dp - file), 1389 file, (int) rbytes, name); 1390 } 1391 } 1392 } 1393 #endif 1394 lseek(fd, 0, SEEK_SET); 1395 1396 if (cflag == 0 || lflag) { 1397 if (fstat(fd, &isb) != 0) 1398 goto lose; 1399 #ifndef SMALL 1400 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1401 maybe_warnx("%s has %d other links -- skipping", 1402 file, isb.st_nlink - 1); 1403 goto lose; 1404 } 1405 if (nflag == 0 && timestamp) 1406 isb.st_mtime = timestamp; 1407 if (check_outfile(outfile) == 0) 1408 goto lose; 1409 #endif 1410 } 1411 1412 if (cflag == 0 && lflag == 0) { 1413 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1414 if (zfd == STDOUT_FILENO) { 1415 /* We won't close STDOUT_FILENO later... */ 1416 zfd = dup(zfd); 1417 close(STDOUT_FILENO); 1418 } 1419 if (zfd == -1) { 1420 maybe_warn("can't open %s", outfile); 1421 goto lose; 1422 } 1423 } else 1424 zfd = STDOUT_FILENO; 1425 1426 switch (method) { 1427 #ifndef NO_BZIP2_SUPPORT 1428 case FT_BZIP2: 1429 /* XXX */ 1430 if (lflag) { 1431 maybe_warnx("no -l with bzip2 files"); 1432 goto lose; 1433 } 1434 1435 size = unbzip2(fd, zfd, NULL, 0, NULL); 1436 break; 1437 #endif 1438 1439 #ifndef NO_COMPRESS_SUPPORT 1440 case FT_Z: { 1441 FILE *in, *out; 1442 1443 /* XXX */ 1444 if (lflag) { 1445 maybe_warnx("no -l with Lempel-Ziv files"); 1446 goto lose; 1447 } 1448 1449 if ((in = zdopen(fd)) == NULL) { 1450 maybe_warn("zdopen for read: %s", file); 1451 goto lose; 1452 } 1453 1454 out = fdopen(dup(zfd), "w"); 1455 if (out == NULL) { 1456 maybe_warn("fdopen for write: %s", outfile); 1457 fclose(in); 1458 goto lose; 1459 } 1460 1461 size = zuncompress(in, out, NULL, 0, NULL); 1462 /* need to fclose() if ferror() is true... */ 1463 if (ferror(in) | fclose(in)) { 1464 maybe_warn("failed infile fclose"); 1465 unlink(outfile); 1466 (void)fclose(out); 1467 } 1468 if (fclose(out) != 0) { 1469 maybe_warn("failed outfile fclose"); 1470 unlink(outfile); 1471 goto lose; 1472 } 1473 break; 1474 } 1475 #endif 1476 1477 #ifndef NO_PACK_SUPPORT 1478 case FT_PACK: 1479 if (lflag) { 1480 maybe_warnx("no -l with packed files"); 1481 goto lose; 1482 } 1483 1484 size = unpack(fd, zfd, NULL, 0, NULL); 1485 break; 1486 #endif 1487 1488 #ifndef NO_XZ_SUPPORT 1489 case FT_XZ: 1490 if (lflag) { 1491 maybe_warnx("no -l with xz files"); 1492 goto lose; 1493 } 1494 1495 size = unxz(fd, zfd, NULL, 0, NULL); 1496 break; 1497 #endif 1498 1499 #ifndef SMALL 1500 case FT_UNKNOWN: 1501 if (lflag) { 1502 maybe_warnx("no -l for unknown filetypes"); 1503 goto lose; 1504 } 1505 size = cat_fd(NULL, 0, NULL, fd); 1506 break; 1507 #endif 1508 default: 1509 if (lflag) { 1510 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1511 close(fd); 1512 return -1; /* XXX */ 1513 } 1514 1515 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1516 break; 1517 } 1518 1519 if (close(fd) != 0) 1520 maybe_warn("couldn't close input"); 1521 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1522 maybe_warn("couldn't close output"); 1523 1524 if (size == -1) { 1525 if (cflag == 0) 1526 unlink(outfile); 1527 maybe_warnx("%s: uncompress failed", file); 1528 return -1; 1529 } 1530 1531 /* if testing, or we uncompressed to stdout, this is all we need */ 1532 #ifndef SMALL 1533 if (tflag) 1534 return size; 1535 #endif 1536 /* if we are uncompressing to stdin, don't remove the file. */ 1537 if (cflag) 1538 return size; 1539 1540 /* 1541 * if we create a file... 1542 */ 1543 /* 1544 * if we can't stat the file don't remove the file. 1545 */ 1546 1547 ofd = open(outfile, O_RDWR, 0); 1548 if (ofd == -1) { 1549 maybe_warn("couldn't open (leaving original): %s", 1550 outfile); 1551 return -1; 1552 } 1553 if (fstat(ofd, &osb) != 0) { 1554 maybe_warn("couldn't stat (leaving original): %s", 1555 outfile); 1556 close(ofd); 1557 return -1; 1558 } 1559 if (osb.st_size != size) { 1560 maybe_warnx("stat gave different size: %" PRIdOFF 1561 " != %" PRIdOFF " (leaving original)", 1562 size, osb.st_size); 1563 close(ofd); 1564 unlink(outfile); 1565 return -1; 1566 } 1567 unlink_input(file, &isb); 1568 #ifndef SMALL 1569 copymodes(ofd, &isb, outfile); 1570 #endif 1571 close(ofd); 1572 return size; 1573 1574 unexpected_EOF: 1575 maybe_warnx("%s: unexpected end of file", file); 1576 lose: 1577 if (fd != -1) 1578 close(fd); 1579 if (zfd != -1 && zfd != STDOUT_FILENO) 1580 close(fd); 1581 return -1; 1582 } 1583 1584 #ifndef SMALL 1585 static off_t 1586 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1587 { 1588 char buf[BUFLEN]; 1589 off_t in_tot; 1590 ssize_t w; 1591 1592 in_tot = count; 1593 w = write(STDOUT_FILENO, prepend, count); 1594 if (w == -1 || (size_t)w != count) { 1595 maybe_warn("write to stdout"); 1596 return -1; 1597 } 1598 for (;;) { 1599 ssize_t rv; 1600 1601 rv = read(fd, buf, sizeof buf); 1602 if (rv == 0) 1603 break; 1604 if (rv < 0) { 1605 maybe_warn("read from fd %d", fd); 1606 break; 1607 } 1608 1609 if (write(STDOUT_FILENO, buf, rv) != rv) { 1610 maybe_warn("write to stdout"); 1611 break; 1612 } 1613 in_tot += rv; 1614 } 1615 1616 if (gsizep) 1617 *gsizep = in_tot; 1618 return (in_tot); 1619 } 1620 #endif 1621 1622 static void 1623 handle_stdin(void) 1624 { 1625 unsigned char header1[4]; 1626 off_t usize, gsize; 1627 enum filetype method; 1628 ssize_t bytes_read; 1629 #ifndef NO_COMPRESS_SUPPORT 1630 FILE *in; 1631 #endif 1632 1633 #ifndef SMALL 1634 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1635 maybe_warnx("standard input is a terminal -- ignoring"); 1636 return; 1637 } 1638 #endif 1639 1640 if (lflag) { 1641 struct stat isb; 1642 1643 /* XXX could read the whole file, etc. */ 1644 if (fstat(STDIN_FILENO, &isb) < 0) { 1645 maybe_warn("fstat"); 1646 return; 1647 } 1648 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1649 return; 1650 } 1651 1652 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1653 if (bytes_read == -1) { 1654 maybe_warn("can't read stdin"); 1655 return; 1656 } else if (bytes_read != sizeof(header1)) { 1657 maybe_warnx("(stdin): unexpected end of file"); 1658 return; 1659 } 1660 1661 method = file_gettype(header1); 1662 switch (method) { 1663 default: 1664 #ifndef SMALL 1665 if (fflag == 0) { 1666 maybe_warnx("unknown compression format"); 1667 return; 1668 } 1669 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1670 break; 1671 #endif 1672 case FT_GZIP: 1673 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1674 (char *)header1, sizeof header1, &gsize, "(stdin)"); 1675 break; 1676 #ifndef NO_BZIP2_SUPPORT 1677 case FT_BZIP2: 1678 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1679 (char *)header1, sizeof header1, &gsize); 1680 break; 1681 #endif 1682 #ifndef NO_COMPRESS_SUPPORT 1683 case FT_Z: 1684 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1685 maybe_warnx("zopen of stdin"); 1686 return; 1687 } 1688 1689 usize = zuncompress(in, stdout, (char *)header1, 1690 sizeof header1, &gsize); 1691 fclose(in); 1692 break; 1693 #endif 1694 #ifndef NO_PACK_SUPPORT 1695 case FT_PACK: 1696 usize = unpack(STDIN_FILENO, STDOUT_FILENO, 1697 (char *)header1, sizeof header1, &gsize); 1698 break; 1699 #endif 1700 #ifndef NO_XZ_SUPPORT 1701 case FT_XZ: 1702 usize = unxz(STDIN_FILENO, STDOUT_FILENO, 1703 (char *)header1, sizeof header1, &gsize); 1704 break; 1705 #endif 1706 } 1707 1708 #ifndef SMALL 1709 if (vflag && !tflag && usize != -1 && gsize != -1) 1710 print_verbage(NULL, NULL, usize, gsize); 1711 if (vflag && tflag) 1712 print_test("(stdin)", usize != -1); 1713 #endif 1714 1715 } 1716 1717 static void 1718 handle_stdout(void) 1719 { 1720 off_t gsize, usize; 1721 struct stat sb; 1722 time_t systime; 1723 uint32_t mtime; 1724 int ret; 1725 1726 #ifndef SMALL 1727 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1728 maybe_warnx("standard output is a terminal -- ignoring"); 1729 return; 1730 } 1731 #endif 1732 /* If stdin is a file use it's mtime, otherwise use current time */ 1733 ret = fstat(STDIN_FILENO, &sb); 1734 1735 #ifndef SMALL 1736 if (ret < 0) { 1737 maybe_warn("Can't stat stdin"); 1738 return; 1739 } 1740 #endif 1741 1742 if (S_ISREG(sb.st_mode)) 1743 mtime = (uint32_t)sb.st_mtime; 1744 else { 1745 systime = time(NULL); 1746 #ifndef SMALL 1747 if (systime == -1) { 1748 maybe_warn("time"); 1749 return; 1750 } 1751 #endif 1752 mtime = (uint32_t)systime; 1753 } 1754 1755 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1756 #ifndef SMALL 1757 if (vflag && !tflag && usize != -1 && gsize != -1) 1758 print_verbage(NULL, NULL, usize, gsize); 1759 #endif 1760 } 1761 1762 /* do what is asked for, for the path name */ 1763 static void 1764 handle_pathname(char *path) 1765 { 1766 char *opath = path, *s = NULL; 1767 ssize_t len; 1768 int slen; 1769 struct stat sb; 1770 1771 /* check for stdout/stdin */ 1772 if (path[0] == '-' && path[1] == '\0') { 1773 if (dflag) 1774 handle_stdin(); 1775 else 1776 handle_stdout(); 1777 return; 1778 } 1779 1780 retry: 1781 if (stat(path, &sb) != 0) { 1782 /* lets try <path>.gz if we're decompressing */ 1783 if (dflag && s == NULL && errno == ENOENT) { 1784 len = strlen(path); 1785 slen = suffixes[0].ziplen; 1786 s = malloc(len + slen + 1); 1787 if (s == NULL) 1788 maybe_err("malloc"); 1789 memcpy(s, path, len); 1790 memcpy(s + len, suffixes[0].zipped, slen + 1); 1791 path = s; 1792 goto retry; 1793 } 1794 maybe_warn("can't stat: %s", opath); 1795 goto out; 1796 } 1797 1798 if (S_ISDIR(sb.st_mode)) { 1799 #ifndef SMALL 1800 if (rflag) 1801 handle_dir(path); 1802 else 1803 #endif 1804 maybe_warnx("%s is a directory", path); 1805 goto out; 1806 } 1807 1808 if (S_ISREG(sb.st_mode)) 1809 handle_file(path, &sb); 1810 else 1811 maybe_warnx("%s is not a regular file", path); 1812 1813 out: 1814 if (s) 1815 free(s); 1816 } 1817 1818 /* compress/decompress a file */ 1819 static void 1820 handle_file(char *file, struct stat *sbp) 1821 { 1822 off_t usize, gsize; 1823 char outfile[PATH_MAX]; 1824 1825 infile = file; 1826 if (dflag) { 1827 usize = file_uncompress(file, outfile, sizeof(outfile)); 1828 #ifndef SMALL 1829 if (vflag && tflag) 1830 print_test(file, usize != -1); 1831 #endif 1832 if (usize == -1) 1833 return; 1834 gsize = sbp->st_size; 1835 } else { 1836 gsize = file_compress(file, outfile, sizeof(outfile)); 1837 if (gsize == -1) 1838 return; 1839 usize = sbp->st_size; 1840 } 1841 1842 1843 #ifndef SMALL 1844 if (vflag && !tflag) 1845 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1846 #endif 1847 } 1848 1849 #ifndef SMALL 1850 /* this is used with -r to recursively descend directories */ 1851 static void 1852 handle_dir(char *dir) 1853 { 1854 char *path_argv[2]; 1855 FTS *fts; 1856 FTSENT *entry; 1857 1858 path_argv[0] = dir; 1859 path_argv[1] = 0; 1860 fts = fts_open(path_argv, FTS_PHYSICAL, NULL); 1861 if (fts == NULL) { 1862 warn("couldn't fts_open %s", dir); 1863 return; 1864 } 1865 1866 while ((entry = fts_read(fts))) { 1867 switch(entry->fts_info) { 1868 case FTS_D: 1869 case FTS_DP: 1870 continue; 1871 1872 case FTS_DNR: 1873 case FTS_ERR: 1874 case FTS_NS: 1875 maybe_warn("%s", entry->fts_path); 1876 continue; 1877 case FTS_F: 1878 handle_file(entry->fts_name, entry->fts_statp); 1879 } 1880 } 1881 (void)fts_close(fts); 1882 } 1883 #endif 1884 1885 /* print a ratio - size reduction as a fraction of uncompressed size */ 1886 static void 1887 print_ratio(off_t in, off_t out, FILE *where) 1888 { 1889 int percent10; /* 10 * percent */ 1890 off_t diff; 1891 char buff[8]; 1892 int len; 1893 1894 diff = in - out/2; 1895 if (diff <= 0) 1896 /* 1897 * Output is more than double size of input! print -99.9% 1898 * Quite possibly we've failed to get the original size. 1899 */ 1900 percent10 = -999; 1901 else { 1902 /* 1903 * We only need 12 bits of result from the final division, 1904 * so reduce the values until a 32bit division will suffice. 1905 */ 1906 while (in > 0x100000) { 1907 diff >>= 1; 1908 in >>= 1; 1909 } 1910 if (in != 0) 1911 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1912 else 1913 percent10 = 0; 1914 } 1915 1916 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1917 /* Move the '.' to before the last digit */ 1918 buff[len - 1] = buff[len - 2]; 1919 buff[len - 2] = '.'; 1920 fprintf(where, "%5s%%", buff); 1921 } 1922 1923 #ifndef SMALL 1924 /* print compression statistics, and the new name (if there is one!) */ 1925 static void 1926 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1927 { 1928 if (file) 1929 fprintf(stderr, "%s:%s ", file, 1930 strlen(file) < 7 ? "\t\t" : "\t"); 1931 print_ratio(usize, gsize, stderr); 1932 if (nfile) 1933 fprintf(stderr, " -- replaced with %s", nfile); 1934 fprintf(stderr, "\n"); 1935 fflush(stderr); 1936 } 1937 1938 /* print test results */ 1939 static void 1940 print_test(const char *file, int ok) 1941 { 1942 1943 if (exit_value == 0 && ok == 0) 1944 exit_value = 1; 1945 fprintf(stderr, "%s:%s %s\n", file, 1946 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1947 fflush(stderr); 1948 } 1949 #endif 1950 1951 /* print a file's info ala --list */ 1952 /* eg: 1953 compressed uncompressed ratio uncompressed_name 1954 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1955 */ 1956 static void 1957 print_list(int fd, off_t out, const char *outfile, time_t ts) 1958 { 1959 static int first = 1; 1960 #ifndef SMALL 1961 static off_t in_tot, out_tot; 1962 uint32_t crc = 0; 1963 #endif 1964 off_t in = 0, rv; 1965 1966 if (first) { 1967 #ifndef SMALL 1968 if (vflag) 1969 printf("method crc date time "); 1970 #endif 1971 if (qflag == 0) 1972 printf(" compressed uncompressed " 1973 "ratio uncompressed_name\n"); 1974 } 1975 first = 0; 1976 1977 /* print totals? */ 1978 #ifndef SMALL 1979 if (fd == -1) { 1980 in = in_tot; 1981 out = out_tot; 1982 } else 1983 #endif 1984 { 1985 /* read the last 4 bytes - this is the uncompressed size */ 1986 rv = lseek(fd, (off_t)(-8), SEEK_END); 1987 if (rv != -1) { 1988 unsigned char buf[8]; 1989 uint32_t usize; 1990 1991 rv = read(fd, (char *)buf, sizeof(buf)); 1992 if (rv == -1) 1993 maybe_warn("read of uncompressed size"); 1994 else if (rv != sizeof(buf)) 1995 maybe_warnx("read of uncompressed size"); 1996 1997 else { 1998 usize = buf[4] | buf[5] << 8 | 1999 buf[6] << 16 | buf[7] << 24; 2000 in = (off_t)usize; 2001 #ifndef SMALL 2002 crc = buf[0] | buf[1] << 8 | 2003 buf[2] << 16 | buf[3] << 24; 2004 #endif 2005 } 2006 } 2007 } 2008 2009 #ifndef SMALL 2010 if (vflag && fd == -1) 2011 printf(" "); 2012 else if (vflag) { 2013 char *date = ctime(&ts); 2014 2015 /* skip the day, 1/100th second, and year */ 2016 date += 4; 2017 date[12] = 0; 2018 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 2019 } 2020 in_tot += in; 2021 out_tot += out; 2022 #endif 2023 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 2024 print_ratio(in, out, stdout); 2025 printf(" %s\n", outfile); 2026 } 2027 2028 /* display the usage of NetBSD gzip */ 2029 static void 2030 usage(void) 2031 { 2032 2033 fprintf(stderr, "%s\n", gzip_version); 2034 fprintf(stderr, 2035 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n" 2036 #ifndef SMALL 2037 " -1 --fast fastest (worst) compression\n" 2038 " -2 .. -8 set compression level\n" 2039 " -9 --best best (slowest) compression\n" 2040 " -c --stdout write to stdout, keep original files\n" 2041 " --to-stdout\n" 2042 " -d --decompress uncompress files\n" 2043 " --uncompress\n" 2044 " -f --force force overwriting & compress links\n" 2045 " -h --help display this help\n" 2046 " -k --keep don't delete input files during operation\n" 2047 " -l --list list compressed file contents\n" 2048 " -N --name save or restore original file name and time stamp\n" 2049 " -n --no-name don't save original file name or time stamp\n" 2050 " -q --quiet output no warnings\n" 2051 " -r --recursive recursively compress files in directories\n" 2052 " -S .suf use suffix .suf instead of .gz\n" 2053 " --suffix .suf\n" 2054 " -t --test test compressed file\n" 2055 " -V --version display program version\n" 2056 " -v --verbose print extra statistics\n", 2057 #else 2058 , 2059 #endif 2060 getprogname()); 2061 exit(0); 2062 } 2063 2064 /* display the version of NetBSD gzip */ 2065 static void 2066 display_version(void) 2067 { 2068 2069 fprintf(stderr, "%s\n", gzip_version); 2070 exit(0); 2071 } 2072 2073 #ifndef NO_BZIP2_SUPPORT 2074 #include "unbzip2.c" 2075 #endif 2076 #ifndef NO_COMPRESS_SUPPORT 2077 #include "zuncompress.c" 2078 #endif 2079 #ifndef NO_PACK_SUPPORT 2080 #include "unpack.c" 2081 #endif 2082 #ifndef NO_XZ_SUPPORT 2083 #include "unxz.c" 2084 #endif 2085 2086 static ssize_t 2087 read_retry(int fd, void *buf, size_t sz) 2088 { 2089 char *cp = buf; 2090 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2091 2092 while (left > 0) { 2093 ssize_t ret; 2094 2095 ret = read(fd, cp, left); 2096 if (ret == -1) { 2097 return ret; 2098 } else if (ret == 0) { 2099 break; /* EOF */ 2100 } 2101 cp += ret; 2102 left -= ret; 2103 } 2104 2105 return sz - left; 2106 } 2107
Indexes created Tue Sep 23 02:09:52 GMT 2025