1 /* $NetBSD: gzip.c,v 1.94 2009/04/12 10:31:14 lukem Exp $ */ 2 3 /*- 4 * Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 21 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 22 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 23 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 24 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 26 * SUCH DAMAGE. 27 * 28 */ 29 30 #include <sys/cdefs.h> 31 #ifndef lint 32 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006\ 33 Matthew R. Green. All rights reserved."); 34 __RCSID("$FreeBSD$"); 35 #endif /* not lint */ 36 37 /* 38 * gzip.c -- GPL free gzip using zlib. 39 * 40 * RFC 1950 covers the zlib format 41 * RFC 1951 covers the deflate format 42 * RFC 1952 covers the gzip format 43 * 44 * TODO: 45 * - use mmap where possible 46 * - handle some signals better (remove outfile?) 47 * - make bzip2/compress -v/-t/-l support work as well as possible 48 */ 49 50 #include <sys/param.h> 51 #include <sys/stat.h> 52 #include <sys/time.h> 53 54 #include <inttypes.h> 55 #include <unistd.h> 56 #include <stdio.h> 57 #include <string.h> 58 #include <stdlib.h> 59 #include <err.h> 60 #include <errno.h> 61 #include <fcntl.h> 62 #include <zlib.h> 63 #include <fts.h> 64 #include <libgen.h> 65 #include <stdarg.h> 66 #include <getopt.h> 67 #include <time.h> 68 69 #ifndef PRIdOFF 70 #define PRIdOFF PRId64 71 #endif 72 73 /* what type of file are we dealing with */ 74 enum filetype { 75 FT_GZIP, 76 #ifndef NO_BZIP2_SUPPORT 77 FT_BZIP2, 78 #endif 79 #ifndef NO_COMPRESS_SUPPORT 80 FT_Z, 81 #endif 82 #ifndef NO_PACK_SUPPORT 83 FT_PACK, 84 #endif 85 FT_LAST, 86 FT_UNKNOWN 87 }; 88 89 #ifndef NO_BZIP2_SUPPORT 90 #include <bzlib.h> 91 92 #define BZ2_SUFFIX ".bz2" 93 #define BZIP2_MAGIC "\102\132\150" 94 #endif 95 96 #ifndef NO_COMPRESS_SUPPORT 97 #define Z_SUFFIX ".Z" 98 #define Z_MAGIC "\037\235" 99 #endif 100 101 #ifndef NO_PACK_SUPPORT 102 #define PACK_MAGIC "\037\036" 103 #endif 104 105 #define GZ_SUFFIX ".gz" 106 107 #define BUFLEN (64 * 1024) 108 109 #define GZIP_MAGIC0 0x1F 110 #define GZIP_MAGIC1 0x8B 111 #define GZIP_OMAGIC1 0x9E 112 113 #define GZIP_TIMESTAMP (off_t)4 114 #define GZIP_ORIGNAME (off_t)10 115 116 #define HEAD_CRC 0x02 117 #define EXTRA_FIELD 0x04 118 #define ORIG_NAME 0x08 119 #define COMMENT 0x10 120 121 #define OS_CODE 3 /* Unix */ 122 123 typedef struct { 124 const char *zipped; 125 int ziplen; 126 const char *normal; /* for unzip - must not be longer than zipped */ 127 } suffixes_t; 128 static suffixes_t suffixes[] = { 129 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 130 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 131 #ifndef SMALL 132 SUFFIX(GZ_SUFFIX, ""), 133 SUFFIX(".z", ""), 134 SUFFIX("-gz", ""), 135 SUFFIX("-z", ""), 136 SUFFIX("_z", ""), 137 SUFFIX(".taz", ".tar"), 138 SUFFIX(".tgz", ".tar"), 139 #ifndef NO_BZIP2_SUPPORT 140 SUFFIX(BZ2_SUFFIX, ""), 141 SUFFIX(".tbz", ".tar"), 142 SUFFIX(".tbz2", ".tar"), 143 #endif 144 #ifndef NO_COMPRESS_SUPPORT 145 SUFFIX(Z_SUFFIX, ""), 146 #endif 147 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 148 #endif /* SMALL */ 149 #undef SUFFIX 150 }; 151 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 152 153 #define SUFFIX_MAXLEN 30 154 155 static const char gzip_version[] = "FreeBSD gzip 20090621"; 156 157 #ifndef SMALL 158 static const char gzip_copyright[] = \ 159 " Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green\n" 160 " All rights reserved.\n" 161 "\n" 162 " Redistribution and use in source and binary forms, with or without\n" 163 " modification, are permitted provided that the following conditions\n" 164 " are met:\n" 165 " 1. Redistributions of source code must retain the above copyright\n" 166 " notice, this list of conditions and the following disclaimer.\n" 167 " 2. Redistributions in binary form must reproduce the above copyright\n" 168 " notice, this list of conditions and the following disclaimer in the\n" 169 " documentation and/or other materials provided with the distribution.\n" 170 "\n" 171 " THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR\n" 172 " IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES\n" 173 " OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.\n" 174 " IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,\n" 175 " INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,\n" 176 " BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;\n" 177 " LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED\n" 178 " AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,\n" 179 " OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY\n" 180 " OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF\n" 181 " SUCH DAMAGE."; 182 #endif 183 184 static int cflag; /* stdout mode */ 185 static int dflag; /* decompress mode */ 186 static int lflag; /* list mode */ 187 static int numflag = 6; /* gzip -1..-9 value */ 188 189 #ifndef SMALL 190 static int fflag; /* force mode */ 191 static int kflag; /* don't delete input files */ 192 static int nflag; /* don't save name/timestamp */ 193 static int Nflag; /* don't restore name/timestamp */ 194 static int qflag; /* quiet mode */ 195 static int rflag; /* recursive mode */ 196 static int tflag; /* test */ 197 static int vflag; /* verbose mode */ 198 #else 199 #define qflag 0 200 #define tflag 0 201 #endif 202 203 static int exit_value = 0; /* exit value */ 204 205 static char *infile; /* name of file coming in */ 206 207 static void maybe_err(const char *fmt, ...) __dead2 208 __attribute__((__format__(__printf__, 1, 2))); 209 #ifndef NO_BZIP2_SUPPORT 210 static void maybe_errx(const char *fmt, ...) __dead2 211 __attribute__((__format__(__printf__, 1, 2))); 212 #endif 213 static void maybe_warn(const char *fmt, ...) 214 __attribute__((__format__(__printf__, 1, 2))); 215 static void maybe_warnx(const char *fmt, ...) 216 __attribute__((__format__(__printf__, 1, 2))); 217 static enum filetype file_gettype(u_char *); 218 #ifdef SMALL 219 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 220 #endif 221 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 222 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 223 static off_t file_compress(char *, char *, size_t); 224 static off_t file_uncompress(char *, char *, size_t); 225 static void handle_pathname(char *); 226 static void handle_file(char *, struct stat *); 227 static void handle_stdin(void); 228 static void handle_stdout(void); 229 static void print_ratio(off_t, off_t, FILE *); 230 static void print_list(int fd, off_t, const char *, time_t); 231 static void usage(void); 232 static void display_version(void); 233 #ifndef SMALL 234 static void display_license(void); 235 #endif 236 static const suffixes_t *check_suffix(char *, int); 237 static ssize_t read_retry(int, void *, size_t); 238 239 #ifdef SMALL 240 #define unlink_input(f, sb) unlink(f) 241 #else 242 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 243 static void prepend_gzip(char *, int *, char ***); 244 static void handle_dir(char *); 245 static void print_verbage(const char *, const char *, off_t, off_t); 246 static void print_test(const char *, int); 247 static void copymodes(int fd, const struct stat *, const char *file); 248 static int check_outfile(const char *outfile); 249 #endif 250 251 #ifndef NO_BZIP2_SUPPORT 252 static off_t unbzip2(int, int, char *, size_t, off_t *); 253 #endif 254 255 #ifndef NO_COMPRESS_SUPPORT 256 static FILE *zdopen(int); 257 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 258 #endif 259 260 #ifndef NO_PACK_SUPPORT 261 static off_t unpack(int, int, char *, size_t, off_t *); 262 #endif 263 264 int main(int, char **p); 265 266 #ifdef SMALL 267 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 268 #else 269 static const struct option longopts[] = { 270 { "stdout", no_argument, 0, 'c' }, 271 { "to-stdout", no_argument, 0, 'c' }, 272 { "decompress", no_argument, 0, 'd' }, 273 { "uncompress", no_argument, 0, 'd' }, 274 { "force", no_argument, 0, 'f' }, 275 { "help", no_argument, 0, 'h' }, 276 { "keep", no_argument, 0, 'k' }, 277 { "list", no_argument, 0, 'l' }, 278 { "no-name", no_argument, 0, 'n' }, 279 { "name", no_argument, 0, 'N' }, 280 { "quiet", no_argument, 0, 'q' }, 281 { "recursive", no_argument, 0, 'r' }, 282 { "suffix", required_argument, 0, 'S' }, 283 { "test", no_argument, 0, 't' }, 284 { "verbose", no_argument, 0, 'v' }, 285 { "version", no_argument, 0, 'V' }, 286 { "fast", no_argument, 0, '1' }, 287 { "best", no_argument, 0, '9' }, 288 { "ascii", no_argument, 0, 'a' }, 289 { "license", no_argument, 0, 'L' }, 290 { NULL, no_argument, 0, 0 }, 291 }; 292 #endif 293 294 int 295 main(int argc, char **argv) 296 { 297 const char *progname = getprogname(); 298 #ifndef SMALL 299 char *gzip; 300 int len; 301 #endif 302 int ch; 303 304 /* XXX set up signals */ 305 306 #ifndef SMALL 307 if ((gzip = getenv("GZIP")) != NULL) 308 prepend_gzip(gzip, &argc, &argv); 309 #endif 310 311 /* 312 * XXX 313 * handle being called `gunzip', `zcat' and `gzcat' 314 */ 315 if (strcmp(progname, "gunzip") == 0) 316 dflag = 1; 317 else if (strcmp(progname, "zcat") == 0 || 318 strcmp(progname, "gzcat") == 0) 319 dflag = cflag = 1; 320 321 #ifdef SMALL 322 #define OPT_LIST "123456789cdhltV" 323 #else 324 #define OPT_LIST "123456789acdfhklLNnqrS:tVv" 325 #endif 326 327 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 328 switch (ch) { 329 case '1': case '2': case '3': 330 case '4': case '5': case '6': 331 case '7': case '8': case '9': 332 numflag = ch - '0'; 333 break; 334 case 'c': 335 cflag = 1; 336 break; 337 case 'd': 338 dflag = 1; 339 break; 340 case 'l': 341 lflag = 1; 342 dflag = 1; 343 break; 344 case 'V': 345 display_version(); 346 /* NOTREACHED */ 347 #ifndef SMALL 348 case 'a': 349 fprintf(stderr, "%s: option --ascii ignored on this system\n", progname); 350 break; 351 case 'f': 352 fflag = 1; 353 break; 354 case 'k': 355 kflag = 1; 356 break; 357 case 'L': 358 display_license(); 359 /* NOT REACHED */ 360 case 'N': 361 nflag = 0; 362 Nflag = 1; 363 break; 364 case 'n': 365 nflag = 1; 366 Nflag = 0; 367 break; 368 case 'q': 369 qflag = 1; 370 break; 371 case 'r': 372 rflag = 1; 373 break; 374 case 'S': 375 len = strlen(optarg); 376 if (len != 0) { 377 if (len > SUFFIX_MAXLEN) 378 errx(1, "incorrect suffix: '%s': too long", optarg); 379 suffixes[0].zipped = optarg; 380 suffixes[0].ziplen = len; 381 } else { 382 suffixes[NUM_SUFFIXES - 1].zipped = ""; 383 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 384 } 385 break; 386 case 't': 387 cflag = 1; 388 tflag = 1; 389 dflag = 1; 390 break; 391 case 'v': 392 vflag = 1; 393 break; 394 #endif 395 default: 396 usage(); 397 /* NOTREACHED */ 398 } 399 } 400 argv += optind; 401 argc -= optind; 402 403 if (argc == 0) { 404 if (dflag) /* stdin mode */ 405 handle_stdin(); 406 else /* stdout mode */ 407 handle_stdout(); 408 } else { 409 do { 410 handle_pathname(argv[0]); 411 } while (*++argv); 412 } 413 #ifndef SMALL 414 if (qflag == 0 && lflag && argc > 1) 415 print_list(-1, 0, "(totals)", 0); 416 #endif 417 exit(exit_value); 418 } 419 420 /* maybe print a warning */ 421 void 422 maybe_warn(const char *fmt, ...) 423 { 424 va_list ap; 425 426 if (qflag == 0) { 427 va_start(ap, fmt); 428 vwarn(fmt, ap); 429 va_end(ap); 430 } 431 if (exit_value == 0) 432 exit_value = 1; 433 } 434 435 /* ... without an errno. */ 436 void 437 maybe_warnx(const char *fmt, ...) 438 { 439 va_list ap; 440 441 if (qflag == 0) { 442 va_start(ap, fmt); 443 vwarnx(fmt, ap); 444 va_end(ap); 445 } 446 if (exit_value == 0) 447 exit_value = 1; 448 } 449 450 /* maybe print an error */ 451 void 452 maybe_err(const char *fmt, ...) 453 { 454 va_list ap; 455 456 if (qflag == 0) { 457 va_start(ap, fmt); 458 vwarn(fmt, ap); 459 va_end(ap); 460 } 461 exit(2); 462 } 463 464 #ifndef NO_BZIP2_SUPPORT 465 /* ... without an errno. */ 466 void 467 maybe_errx(const char *fmt, ...) 468 { 469 va_list ap; 470 471 if (qflag == 0) { 472 va_start(ap, fmt); 473 vwarnx(fmt, ap); 474 va_end(ap); 475 } 476 exit(2); 477 } 478 #endif 479 480 #ifndef SMALL 481 /* split up $GZIP and prepend it to the argument list */ 482 static void 483 prepend_gzip(char *gzip, int *argc, char ***argv) 484 { 485 char *s, **nargv, **ac; 486 int nenvarg = 0, i; 487 488 /* scan how many arguments there are */ 489 for (s = gzip;;) { 490 while (*s == ' ' || *s == '\t') 491 s++; 492 if (*s == 0) 493 goto count_done; 494 nenvarg++; 495 while (*s != ' ' && *s != '\t') 496 if (*s++ == 0) 497 goto count_done; 498 } 499 count_done: 500 /* punt early */ 501 if (nenvarg == 0) 502 return; 503 504 *argc += nenvarg; 505 ac = *argv; 506 507 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 508 if (nargv == NULL) 509 maybe_err("malloc"); 510 511 /* stash this away */ 512 *argv = nargv; 513 514 /* copy the program name first */ 515 i = 0; 516 nargv[i++] = *(ac++); 517 518 /* take a copy of $GZIP and add it to the array */ 519 s = strdup(gzip); 520 if (s == NULL) 521 maybe_err("strdup"); 522 for (;;) { 523 /* Skip whitespaces. */ 524 while (*s == ' ' || *s == '\t') 525 s++; 526 if (*s == 0) 527 goto copy_done; 528 nargv[i++] = s; 529 /* Find the end of this argument. */ 530 while (*s != ' ' && *s != '\t') 531 if (*s++ == 0) 532 /* Argument followed by NUL. */ 533 goto copy_done; 534 /* Terminate by overwriting ' ' or '\t' with NUL. */ 535 *s++ = 0; 536 } 537 copy_done: 538 539 /* copy the original arguments and a NULL */ 540 while (*ac) 541 nargv[i++] = *(ac++); 542 nargv[i] = NULL; 543 } 544 #endif 545 546 /* compress input to output. Return bytes read, -1 on error */ 547 static off_t 548 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 549 { 550 z_stream z; 551 char *outbufp, *inbufp; 552 off_t in_tot = 0, out_tot = 0; 553 ssize_t in_size; 554 int i, error; 555 uLong crc; 556 #ifdef SMALL 557 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 558 0, 0, 0, 0, 559 0, OS_CODE }; 560 #endif 561 562 outbufp = malloc(BUFLEN); 563 inbufp = malloc(BUFLEN); 564 if (outbufp == NULL || inbufp == NULL) { 565 maybe_err("malloc failed"); 566 goto out; 567 } 568 569 memset(&z, 0, sizeof z); 570 z.zalloc = Z_NULL; 571 z.zfree = Z_NULL; 572 z.opaque = 0; 573 574 #ifdef SMALL 575 memcpy(outbufp, header, sizeof header); 576 i = sizeof header; 577 #else 578 if (nflag != 0) { 579 mtime = 0; 580 origname = ""; 581 } 582 583 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 584 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 585 *origname ? ORIG_NAME : 0, 586 mtime & 0xff, 587 (mtime >> 8) & 0xff, 588 (mtime >> 16) & 0xff, 589 (mtime >> 24) & 0xff, 590 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 591 OS_CODE, origname); 592 if (i >= BUFLEN) 593 /* this need PATH_MAX > BUFLEN ... */ 594 maybe_err("snprintf"); 595 if (*origname) 596 i++; 597 #endif 598 599 z.next_out = (unsigned char *)outbufp + i; 600 z.avail_out = BUFLEN - i; 601 602 error = deflateInit2(&z, numflag, Z_DEFLATED, 603 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 604 if (error != Z_OK) { 605 maybe_warnx("deflateInit2 failed"); 606 in_tot = -1; 607 goto out; 608 } 609 610 crc = crc32(0L, Z_NULL, 0); 611 for (;;) { 612 if (z.avail_out == 0) { 613 if (write(out, outbufp, BUFLEN) != BUFLEN) { 614 maybe_warn("write"); 615 out_tot = -1; 616 goto out; 617 } 618 619 out_tot += BUFLEN; 620 z.next_out = (unsigned char *)outbufp; 621 z.avail_out = BUFLEN; 622 } 623 624 if (z.avail_in == 0) { 625 in_size = read(in, inbufp, BUFLEN); 626 if (in_size < 0) { 627 maybe_warn("read"); 628 in_tot = -1; 629 goto out; 630 } 631 if (in_size == 0) 632 break; 633 634 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 635 in_tot += in_size; 636 z.next_in = (unsigned char *)inbufp; 637 z.avail_in = in_size; 638 } 639 640 error = deflate(&z, Z_NO_FLUSH); 641 if (error != Z_OK && error != Z_STREAM_END) { 642 maybe_warnx("deflate failed"); 643 in_tot = -1; 644 goto out; 645 } 646 } 647 648 /* clean up */ 649 for (;;) { 650 size_t len; 651 ssize_t w; 652 653 error = deflate(&z, Z_FINISH); 654 if (error != Z_OK && error != Z_STREAM_END) { 655 maybe_warnx("deflate failed"); 656 in_tot = -1; 657 goto out; 658 } 659 660 len = (char *)z.next_out - outbufp; 661 662 w = write(out, outbufp, len); 663 if (w == -1 || (size_t)w != len) { 664 maybe_warn("write"); 665 out_tot = -1; 666 goto out; 667 } 668 out_tot += len; 669 z.next_out = (unsigned char *)outbufp; 670 z.avail_out = BUFLEN; 671 672 if (error == Z_STREAM_END) 673 break; 674 } 675 676 if (deflateEnd(&z) != Z_OK) { 677 maybe_warnx("deflateEnd failed"); 678 in_tot = -1; 679 goto out; 680 } 681 682 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 683 (int)crc & 0xff, 684 (int)(crc >> 8) & 0xff, 685 (int)(crc >> 16) & 0xff, 686 (int)(crc >> 24) & 0xff, 687 (int)in_tot & 0xff, 688 (int)(in_tot >> 8) & 0xff, 689 (int)(in_tot >> 16) & 0xff, 690 (int)(in_tot >> 24) & 0xff); 691 if (i != 8) 692 maybe_err("snprintf"); 693 if (write(out, outbufp, i) != i) { 694 maybe_warn("write"); 695 in_tot = -1; 696 } else 697 out_tot += i; 698 699 out: 700 if (inbufp != NULL) 701 free(inbufp); 702 if (outbufp != NULL) 703 free(outbufp); 704 if (gsizep) 705 *gsizep = out_tot; 706 return in_tot; 707 } 708 709 /* 710 * uncompress input to output then close the input. return the 711 * uncompressed size written, and put the compressed sized read 712 * into `*gsizep'. 713 */ 714 static off_t 715 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 716 const char *filename) 717 { 718 z_stream z; 719 char *outbufp, *inbufp; 720 off_t out_tot = -1, in_tot = 0; 721 uint32_t out_sub_tot = 0; 722 enum { 723 GZSTATE_MAGIC0, 724 GZSTATE_MAGIC1, 725 GZSTATE_METHOD, 726 GZSTATE_FLAGS, 727 GZSTATE_SKIPPING, 728 GZSTATE_EXTRA, 729 GZSTATE_EXTRA2, 730 GZSTATE_EXTRA3, 731 GZSTATE_ORIGNAME, 732 GZSTATE_COMMENT, 733 GZSTATE_HEAD_CRC1, 734 GZSTATE_HEAD_CRC2, 735 GZSTATE_INIT, 736 GZSTATE_READ, 737 GZSTATE_CRC, 738 GZSTATE_LEN, 739 } state = GZSTATE_MAGIC0; 740 int flags = 0, skip_count = 0; 741 int error = Z_STREAM_ERROR, done_reading = 0; 742 uLong crc = 0; 743 ssize_t wr; 744 int needmore = 0; 745 746 #define ADVANCE() { z.next_in++; z.avail_in--; } 747 748 if ((outbufp = malloc(BUFLEN)) == NULL) { 749 maybe_err("malloc failed"); 750 goto out2; 751 } 752 if ((inbufp = malloc(BUFLEN)) == NULL) { 753 maybe_err("malloc failed"); 754 goto out1; 755 } 756 757 memset(&z, 0, sizeof z); 758 z.avail_in = prelen; 759 z.next_in = (unsigned char *)pre; 760 z.avail_out = BUFLEN; 761 z.next_out = (unsigned char *)outbufp; 762 z.zalloc = NULL; 763 z.zfree = NULL; 764 z.opaque = 0; 765 766 in_tot = prelen; 767 out_tot = 0; 768 769 for (;;) { 770 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 771 ssize_t in_size; 772 773 if (z.avail_in > 0) { 774 memmove(inbufp, z.next_in, z.avail_in); 775 } 776 z.next_in = (unsigned char *)inbufp; 777 in_size = read(in, z.next_in + z.avail_in, 778 BUFLEN - z.avail_in); 779 780 if (in_size == -1) { 781 maybe_warn("failed to read stdin"); 782 goto stop_and_fail; 783 } else if (in_size == 0) { 784 done_reading = 1; 785 } 786 787 z.avail_in += in_size; 788 needmore = 0; 789 790 in_tot += in_size; 791 } 792 if (z.avail_in == 0) { 793 if (done_reading && state != GZSTATE_MAGIC0) { 794 maybe_warnx("%s: unexpected end of file", 795 filename); 796 goto stop_and_fail; 797 } 798 goto stop; 799 } 800 switch (state) { 801 case GZSTATE_MAGIC0: 802 if (*z.next_in != GZIP_MAGIC0) { 803 if (in_tot > 0) { 804 maybe_warnx("%s: trailing garbage " 805 "ignored", filename); 806 goto stop; 807 } 808 maybe_warnx("input not gziped (MAGIC0)"); 809 goto stop_and_fail; 810 } 811 ADVANCE(); 812 state++; 813 out_sub_tot = 0; 814 crc = crc32(0L, Z_NULL, 0); 815 break; 816 817 case GZSTATE_MAGIC1: 818 if (*z.next_in != GZIP_MAGIC1 && 819 *z.next_in != GZIP_OMAGIC1) { 820 maybe_warnx("input not gziped (MAGIC1)"); 821 goto stop_and_fail; 822 } 823 ADVANCE(); 824 state++; 825 break; 826 827 case GZSTATE_METHOD: 828 if (*z.next_in != Z_DEFLATED) { 829 maybe_warnx("unknown compression method"); 830 goto stop_and_fail; 831 } 832 ADVANCE(); 833 state++; 834 break; 835 836 case GZSTATE_FLAGS: 837 flags = *z.next_in; 838 ADVANCE(); 839 skip_count = 6; 840 state++; 841 break; 842 843 case GZSTATE_SKIPPING: 844 if (skip_count > 0) { 845 skip_count--; 846 ADVANCE(); 847 } else 848 state++; 849 break; 850 851 case GZSTATE_EXTRA: 852 if ((flags & EXTRA_FIELD) == 0) { 853 state = GZSTATE_ORIGNAME; 854 break; 855 } 856 skip_count = *z.next_in; 857 ADVANCE(); 858 state++; 859 break; 860 861 case GZSTATE_EXTRA2: 862 skip_count |= ((*z.next_in) << 8); 863 ADVANCE(); 864 state++; 865 break; 866 867 case GZSTATE_EXTRA3: 868 if (skip_count > 0) { 869 skip_count--; 870 ADVANCE(); 871 } else 872 state++; 873 break; 874 875 case GZSTATE_ORIGNAME: 876 if ((flags & ORIG_NAME) == 0) { 877 state++; 878 break; 879 } 880 if (*z.next_in == 0) 881 state++; 882 ADVANCE(); 883 break; 884 885 case GZSTATE_COMMENT: 886 if ((flags & COMMENT) == 0) { 887 state++; 888 break; 889 } 890 if (*z.next_in == 0) 891 state++; 892 ADVANCE(); 893 break; 894 895 case GZSTATE_HEAD_CRC1: 896 if (flags & HEAD_CRC) 897 skip_count = 2; 898 else 899 skip_count = 0; 900 state++; 901 break; 902 903 case GZSTATE_HEAD_CRC2: 904 if (skip_count > 0) { 905 skip_count--; 906 ADVANCE(); 907 } else 908 state++; 909 break; 910 911 case GZSTATE_INIT: 912 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 913 maybe_warnx("failed to inflateInit"); 914 goto stop_and_fail; 915 } 916 state++; 917 break; 918 919 case GZSTATE_READ: 920 error = inflate(&z, Z_FINISH); 921 switch (error) { 922 /* Z_BUF_ERROR goes with Z_FINISH... */ 923 case Z_BUF_ERROR: 924 case Z_STREAM_END: 925 case Z_OK: 926 break; 927 928 case Z_NEED_DICT: 929 maybe_warnx("Z_NEED_DICT error"); 930 goto stop_and_fail; 931 case Z_DATA_ERROR: 932 maybe_warnx("data stream error"); 933 goto stop_and_fail; 934 case Z_STREAM_ERROR: 935 maybe_warnx("internal stream error"); 936 goto stop_and_fail; 937 case Z_MEM_ERROR: 938 maybe_warnx("memory allocation error"); 939 goto stop_and_fail; 940 941 default: 942 maybe_warn("unknown error from inflate(): %d", 943 error); 944 } 945 wr = BUFLEN - z.avail_out; 946 947 if (wr != 0) { 948 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 949 if ( 950 #ifndef SMALL 951 /* don't write anything with -t */ 952 tflag == 0 && 953 #endif 954 write(out, outbufp, wr) != wr) { 955 maybe_warn("error writing to output"); 956 goto stop_and_fail; 957 } 958 959 out_tot += wr; 960 out_sub_tot += wr; 961 } 962 963 if (error == Z_STREAM_END) { 964 inflateEnd(&z); 965 state++; 966 } 967 968 z.next_out = (unsigned char *)outbufp; 969 z.avail_out = BUFLEN; 970 971 break; 972 case GZSTATE_CRC: 973 { 974 uLong origcrc; 975 976 if (z.avail_in < 4) { 977 if (!done_reading) { 978 needmore = 1; 979 continue; 980 } 981 maybe_warnx("truncated input"); 982 goto stop_and_fail; 983 } 984 origcrc = ((unsigned)z.next_in[0] & 0xff) | 985 ((unsigned)z.next_in[1] & 0xff) << 8 | 986 ((unsigned)z.next_in[2] & 0xff) << 16 | 987 ((unsigned)z.next_in[3] & 0xff) << 24; 988 if (origcrc != crc) { 989 maybe_warnx("invalid compressed" 990 " data--crc error"); 991 goto stop_and_fail; 992 } 993 } 994 995 z.avail_in -= 4; 996 z.next_in += 4; 997 998 if (!z.avail_in && done_reading) { 999 goto stop; 1000 } 1001 state++; 1002 break; 1003 case GZSTATE_LEN: 1004 { 1005 uLong origlen; 1006 1007 if (z.avail_in < 4) { 1008 if (!done_reading) { 1009 needmore = 1; 1010 continue; 1011 } 1012 maybe_warnx("truncated input"); 1013 goto stop_and_fail; 1014 } 1015 origlen = ((unsigned)z.next_in[0] & 0xff) | 1016 ((unsigned)z.next_in[1] & 0xff) << 8 | 1017 ((unsigned)z.next_in[2] & 0xff) << 16 | 1018 ((unsigned)z.next_in[3] & 0xff) << 24; 1019 1020 if (origlen != out_sub_tot) { 1021 maybe_warnx("invalid compressed" 1022 " data--length error"); 1023 goto stop_and_fail; 1024 } 1025 } 1026 1027 z.avail_in -= 4; 1028 z.next_in += 4; 1029 1030 if (error < 0) { 1031 maybe_warnx("decompression error"); 1032 goto stop_and_fail; 1033 } 1034 state = GZSTATE_MAGIC0; 1035 break; 1036 } 1037 continue; 1038 stop_and_fail: 1039 out_tot = -1; 1040 stop: 1041 break; 1042 } 1043 if (state > GZSTATE_INIT) 1044 inflateEnd(&z); 1045 1046 free(inbufp); 1047 out1: 1048 free(outbufp); 1049 out2: 1050 if (gsizep) 1051 *gsizep = in_tot; 1052 return (out_tot); 1053 } 1054 1055 #ifndef SMALL 1056 /* 1057 * set the owner, mode, flags & utimes using the given file descriptor. 1058 * file is only used in possible warning messages. 1059 */ 1060 static void 1061 copymodes(int fd, const struct stat *sbp, const char *file) 1062 { 1063 struct timeval times[2]; 1064 struct stat sb; 1065 1066 /* 1067 * If we have no info on the input, give this file some 1068 * default values and return.. 1069 */ 1070 if (sbp == NULL) { 1071 mode_t mask = umask(022); 1072 1073 (void)fchmod(fd, DEFFILEMODE & ~mask); 1074 (void)umask(mask); 1075 return; 1076 } 1077 sb = *sbp; 1078 1079 /* if the chown fails, remove set-id bits as-per compress(1) */ 1080 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1081 if (errno != EPERM) 1082 maybe_warn("couldn't fchown: %s", file); 1083 sb.st_mode &= ~(S_ISUID|S_ISGID); 1084 } 1085 1086 /* we only allow set-id and the 9 normal permission bits */ 1087 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1088 if (fchmod(fd, sb.st_mode) < 0) 1089 maybe_warn("couldn't fchmod: %s", file); 1090 1091 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atim); 1092 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtim); 1093 if (futimes(fd, times) < 0) 1094 maybe_warn("couldn't utimes: %s", file); 1095 1096 /* only try flags if they exist already */ 1097 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1098 maybe_warn("couldn't fchflags: %s", file); 1099 } 1100 #endif 1101 1102 /* what sort of file is this? */ 1103 static enum filetype 1104 file_gettype(u_char *buf) 1105 { 1106 1107 if (buf[0] == GZIP_MAGIC0 && 1108 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1109 return FT_GZIP; 1110 else 1111 #ifndef NO_BZIP2_SUPPORT 1112 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1113 buf[3] >= '0' && buf[3] <= '9') 1114 return FT_BZIP2; 1115 else 1116 #endif 1117 #ifndef NO_COMPRESS_SUPPORT 1118 if (memcmp(buf, Z_MAGIC, 2) == 0) 1119 return FT_Z; 1120 else 1121 #endif 1122 #ifndef NO_PACK_SUPPORT 1123 if (memcmp(buf, PACK_MAGIC, 2) == 0) 1124 return FT_PACK; 1125 else 1126 #endif 1127 return FT_UNKNOWN; 1128 } 1129 1130 #ifndef SMALL 1131 /* check the outfile is OK. */ 1132 static int 1133 check_outfile(const char *outfile) 1134 { 1135 struct stat sb; 1136 int ok = 1; 1137 1138 if (lflag == 0 && stat(outfile, &sb) == 0) { 1139 if (fflag) 1140 unlink(outfile); 1141 else if (isatty(STDIN_FILENO)) { 1142 char ans[10] = { 'n', '\0' }; /* default */ 1143 1144 fprintf(stderr, "%s already exists -- do you wish to " 1145 "overwrite (y or n)? " , outfile); 1146 (void)fgets(ans, sizeof(ans) - 1, stdin); 1147 if (ans[0] != 'y' && ans[0] != 'Y') { 1148 fprintf(stderr, "\tnot overwriting\n"); 1149 ok = 0; 1150 } else 1151 unlink(outfile); 1152 } else { 1153 maybe_warnx("%s already exists -- skipping", outfile); 1154 ok = 0; 1155 } 1156 } 1157 return ok; 1158 } 1159 1160 static void 1161 unlink_input(const char *file, const struct stat *sb) 1162 { 1163 struct stat nsb; 1164 1165 if (kflag) 1166 return; 1167 if (stat(file, &nsb) != 0) 1168 /* Must be gone alrady */ 1169 return; 1170 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1171 /* Definitely a different file */ 1172 return; 1173 unlink(file); 1174 } 1175 #endif 1176 1177 static const suffixes_t * 1178 check_suffix(char *file, int xlate) 1179 { 1180 const suffixes_t *s; 1181 int len = strlen(file); 1182 char *sp; 1183 1184 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1185 /* if it doesn't fit in "a.suf", don't bother */ 1186 if (s->ziplen >= len) 1187 continue; 1188 sp = file + len - s->ziplen; 1189 if (strcmp(s->zipped, sp) != 0) 1190 continue; 1191 if (xlate) 1192 strcpy(sp, s->normal); 1193 return s; 1194 } 1195 return NULL; 1196 } 1197 1198 /* 1199 * compress the given file: create a corresponding .gz file and remove the 1200 * original. 1201 */ 1202 static off_t 1203 file_compress(char *file, char *outfile, size_t outsize) 1204 { 1205 int in; 1206 int out; 1207 off_t size, insize; 1208 #ifndef SMALL 1209 struct stat isb, osb; 1210 const suffixes_t *suff; 1211 #endif 1212 1213 in = open(file, O_RDONLY); 1214 if (in == -1) { 1215 maybe_warn("can't open %s", file); 1216 return -1; 1217 } 1218 1219 if (cflag == 0) { 1220 #ifndef SMALL 1221 if (fstat(in, &isb) == 0) { 1222 if (isb.st_nlink > 1 && fflag == 0) { 1223 maybe_warnx("%s has %d other link%s -- " 1224 "skipping", file, isb.st_nlink - 1, 1225 isb.st_nlink == 1 ? "" : "s"); 1226 close(in); 1227 return -1; 1228 } 1229 } 1230 1231 if (fflag == 0 && (suff = check_suffix(file, 0)) 1232 && suff->zipped[0] != 0) { 1233 maybe_warnx("%s already has %s suffix -- unchanged", 1234 file, suff->zipped); 1235 close(in); 1236 return -1; 1237 } 1238 #endif 1239 1240 /* Add (usually) .gz to filename */ 1241 if ((size_t)snprintf(outfile, outsize, "%s%s", 1242 file, suffixes[0].zipped) >= outsize) 1243 memcpy(outfile + outsize - suffixes[0].ziplen - 1, 1244 suffixes[0].zipped, suffixes[0].ziplen + 1); 1245 1246 #ifndef SMALL 1247 if (check_outfile(outfile) == 0) { 1248 close(in); 1249 return -1; 1250 } 1251 #endif 1252 } 1253 1254 if (cflag == 0) { 1255 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1256 if (out == -1) { 1257 maybe_warn("could not create output: %s", outfile); 1258 fclose(stdin); 1259 return -1; 1260 } 1261 } else 1262 out = STDOUT_FILENO; 1263 1264 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1265 1266 (void)close(in); 1267 1268 /* 1269 * If there was an error, insize will be -1. 1270 * If we compressed to stdout, just return the size. 1271 * Otherwise stat the file and check it is the correct size. 1272 * We only blow away the file if we can stat the output and it 1273 * has the expected size. 1274 */ 1275 if (cflag != 0) 1276 return insize == -1 ? -1 : size; 1277 1278 #ifndef SMALL 1279 if (fstat(out, &osb) != 0) { 1280 maybe_warn("couldn't stat: %s", outfile); 1281 goto bad_outfile; 1282 } 1283 1284 if (osb.st_size != size) { 1285 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1286 " != %" PRIdOFF "), deleting", 1287 outfile, osb.st_size, size); 1288 goto bad_outfile; 1289 } 1290 1291 copymodes(out, &isb, outfile); 1292 #endif 1293 if (close(out) == -1) 1294 maybe_warn("couldn't close output"); 1295 1296 /* output is good, ok to delete input */ 1297 unlink_input(file, &isb); 1298 return size; 1299 1300 #ifndef SMALL 1301 bad_outfile: 1302 if (close(out) == -1) 1303 maybe_warn("couldn't close output"); 1304 1305 maybe_warnx("leaving original %s", file); 1306 unlink(outfile); 1307 return size; 1308 #endif 1309 } 1310 1311 /* uncompress the given file and remove the original */ 1312 static off_t 1313 file_uncompress(char *file, char *outfile, size_t outsize) 1314 { 1315 struct stat isb, osb; 1316 off_t size; 1317 ssize_t rbytes; 1318 unsigned char header1[4]; 1319 enum filetype method; 1320 int fd, ofd, zfd = -1; 1321 #ifndef SMALL 1322 time_t timestamp = 0; 1323 unsigned char name[PATH_MAX + 1]; 1324 #endif 1325 1326 /* gather the old name info */ 1327 1328 fd = open(file, O_RDONLY); 1329 if (fd < 0) { 1330 maybe_warn("can't open %s", file); 1331 goto lose; 1332 } 1333 1334 strlcpy(outfile, file, outsize); 1335 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1336 maybe_warnx("%s: unknown suffix -- ignored", file); 1337 goto lose; 1338 } 1339 1340 rbytes = read(fd, header1, sizeof header1); 1341 if (rbytes != sizeof header1) { 1342 /* we don't want to fail here. */ 1343 #ifndef SMALL 1344 if (fflag) 1345 goto lose; 1346 #endif 1347 if (rbytes == -1) 1348 maybe_warn("can't read %s", file); 1349 else 1350 goto unexpected_EOF; 1351 goto lose; 1352 } 1353 1354 method = file_gettype(header1); 1355 1356 #ifndef SMALL 1357 if (fflag == 0 && method == FT_UNKNOWN) { 1358 maybe_warnx("%s: not in gzip format", file); 1359 goto lose; 1360 } 1361 1362 #endif 1363 1364 #ifndef SMALL 1365 if (method == FT_GZIP && Nflag) { 1366 unsigned char ts[4]; /* timestamp */ 1367 ssize_t rv; 1368 1369 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1370 if (rv >= 0 && rv < (ssize_t)(sizeof ts)) 1371 goto unexpected_EOF; 1372 if (rv == -1) { 1373 if (!fflag) 1374 maybe_warn("can't read %s", file); 1375 goto lose; 1376 } 1377 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1378 1379 if (header1[3] & ORIG_NAME) { 1380 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1381 if (rbytes < 0) { 1382 maybe_warn("can't read %s", file); 1383 goto lose; 1384 } 1385 if (name[0] != 0) { 1386 /* preserve original directory name */ 1387 char *dp = strrchr(file, '/'); 1388 if (dp == NULL) 1389 dp = file; 1390 else 1391 dp++; 1392 snprintf(outfile, outsize, "%.*s%.*s", 1393 (int) (dp - file), 1394 file, (int) rbytes, name); 1395 } 1396 } 1397 } 1398 #endif 1399 lseek(fd, 0, SEEK_SET); 1400 1401 if (cflag == 0 || lflag) { 1402 if (fstat(fd, &isb) != 0) 1403 goto lose; 1404 #ifndef SMALL 1405 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1406 maybe_warnx("%s has %d other links -- skipping", 1407 file, isb.st_nlink - 1); 1408 goto lose; 1409 } 1410 if (nflag == 0 && timestamp) 1411 isb.st_mtime = timestamp; 1412 if (check_outfile(outfile) == 0) 1413 goto lose; 1414 #endif 1415 } 1416 1417 if (cflag == 0 && lflag == 0) { 1418 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1419 if (zfd == STDOUT_FILENO) { 1420 /* We won't close STDOUT_FILENO later... */ 1421 zfd = dup(zfd); 1422 close(STDOUT_FILENO); 1423 } 1424 if (zfd == -1) { 1425 maybe_warn("can't open %s", outfile); 1426 goto lose; 1427 } 1428 } else 1429 zfd = STDOUT_FILENO; 1430 1431 #ifndef NO_BZIP2_SUPPORT 1432 if (method == FT_BZIP2) { 1433 1434 /* XXX */ 1435 if (lflag) { 1436 maybe_warnx("no -l with bzip2 files"); 1437 goto lose; 1438 } 1439 1440 size = unbzip2(fd, zfd, NULL, 0, NULL); 1441 } else 1442 #endif 1443 1444 #ifndef NO_COMPRESS_SUPPORT 1445 if (method == FT_Z) { 1446 FILE *in, *out; 1447 1448 /* XXX */ 1449 if (lflag) { 1450 maybe_warnx("no -l with Lempel-Ziv files"); 1451 goto lose; 1452 } 1453 1454 if ((in = zdopen(fd)) == NULL) { 1455 maybe_warn("zdopen for read: %s", file); 1456 goto lose; 1457 } 1458 1459 out = fdopen(dup(zfd), "w"); 1460 if (out == NULL) { 1461 maybe_warn("fdopen for write: %s", outfile); 1462 fclose(in); 1463 goto lose; 1464 } 1465 1466 size = zuncompress(in, out, NULL, 0, NULL); 1467 /* need to fclose() if ferror() is true... */ 1468 if (ferror(in) | fclose(in)) { 1469 maybe_warn("failed infile fclose"); 1470 unlink(outfile); 1471 (void)fclose(out); 1472 } 1473 if (fclose(out) != 0) { 1474 maybe_warn("failed outfile fclose"); 1475 unlink(outfile); 1476 goto lose; 1477 } 1478 } else 1479 #endif 1480 1481 #ifndef NO_PACK_SUPPORT 1482 if (method == FT_PACK) { 1483 if (lflag) { 1484 maybe_warnx("no -l with packed files"); 1485 goto lose; 1486 } 1487 1488 size = unpack(fd, zfd, NULL, 0, NULL); 1489 } else 1490 #endif 1491 1492 #ifndef SMALL 1493 if (method == FT_UNKNOWN) { 1494 if (lflag) { 1495 maybe_warnx("no -l for unknown filetypes"); 1496 goto lose; 1497 } 1498 size = cat_fd(NULL, 0, NULL, fd); 1499 } else 1500 #endif 1501 { 1502 if (lflag) { 1503 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1504 close(fd); 1505 return -1; /* XXX */ 1506 } 1507 1508 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1509 } 1510 1511 if (close(fd) != 0) 1512 maybe_warn("couldn't close input"); 1513 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1514 maybe_warn("couldn't close output"); 1515 1516 if (size == -1) { 1517 if (cflag == 0) 1518 unlink(outfile); 1519 maybe_warnx("%s: uncompress failed", file); 1520 return -1; 1521 } 1522 1523 /* if testing, or we uncompressed to stdout, this is all we need */ 1524 #ifndef SMALL 1525 if (tflag) 1526 return size; 1527 #endif 1528 /* if we are uncompressing to stdin, don't remove the file. */ 1529 if (cflag) 1530 return size; 1531 1532 /* 1533 * if we create a file... 1534 */ 1535 /* 1536 * if we can't stat the file don't remove the file. 1537 */ 1538 1539 ofd = open(outfile, O_RDWR, 0); 1540 if (ofd == -1) { 1541 maybe_warn("couldn't open (leaving original): %s", 1542 outfile); 1543 return -1; 1544 } 1545 if (fstat(ofd, &osb) != 0) { 1546 maybe_warn("couldn't stat (leaving original): %s", 1547 outfile); 1548 close(ofd); 1549 return -1; 1550 } 1551 if (osb.st_size != size) { 1552 maybe_warnx("stat gave different size: %" PRIdOFF 1553 " != %" PRIdOFF " (leaving original)", 1554 size, osb.st_size); 1555 close(ofd); 1556 unlink(outfile); 1557 return -1; 1558 } 1559 unlink_input(file, &isb); 1560 #ifndef SMALL 1561 copymodes(ofd, &isb, outfile); 1562 #endif 1563 close(ofd); 1564 return size; 1565 1566 unexpected_EOF: 1567 maybe_warnx("%s: unexpected end of file", file); 1568 lose: 1569 if (fd != -1) 1570 close(fd); 1571 if (zfd != -1 && zfd != STDOUT_FILENO) 1572 close(fd); 1573 return -1; 1574 } 1575 1576 #ifndef SMALL 1577 static off_t 1578 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1579 { 1580 char buf[BUFLEN]; 1581 off_t in_tot; 1582 ssize_t w; 1583 1584 in_tot = count; 1585 w = write(STDOUT_FILENO, prepend, count); 1586 if (w == -1 || (size_t)w != count) { 1587 maybe_warn("write to stdout"); 1588 return -1; 1589 } 1590 for (;;) { 1591 ssize_t rv; 1592 1593 rv = read(fd, buf, sizeof buf); 1594 if (rv == 0) 1595 break; 1596 if (rv < 0) { 1597 maybe_warn("read from fd %d", fd); 1598 break; 1599 } 1600 1601 if (write(STDOUT_FILENO, buf, rv) != rv) { 1602 maybe_warn("write to stdout"); 1603 break; 1604 } 1605 in_tot += rv; 1606 } 1607 1608 if (gsizep) 1609 *gsizep = in_tot; 1610 return (in_tot); 1611 } 1612 #endif 1613 1614 static void 1615 handle_stdin(void) 1616 { 1617 unsigned char header1[4]; 1618 off_t usize, gsize; 1619 enum filetype method; 1620 ssize_t bytes_read; 1621 #ifndef NO_COMPRESS_SUPPORT 1622 FILE *in; 1623 #endif 1624 1625 #ifndef SMALL 1626 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1627 maybe_warnx("standard input is a terminal -- ignoring"); 1628 return; 1629 } 1630 #endif 1631 1632 if (lflag) { 1633 struct stat isb; 1634 1635 /* XXX could read the whole file, etc. */ 1636 if (fstat(STDIN_FILENO, &isb) < 0) { 1637 maybe_warn("fstat"); 1638 return; 1639 } 1640 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1641 return; 1642 } 1643 1644 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1645 if (bytes_read == -1) { 1646 maybe_warn("can't read stdin"); 1647 return; 1648 } else if (bytes_read != sizeof(header1)) { 1649 maybe_warnx("(stdin): unexpected end of file"); 1650 return; 1651 } 1652 1653 method = file_gettype(header1); 1654 switch (method) { 1655 default: 1656 #ifndef SMALL 1657 if (fflag == 0) { 1658 maybe_warnx("unknown compression format"); 1659 return; 1660 } 1661 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1662 break; 1663 #endif 1664 case FT_GZIP: 1665 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1666 (char *)header1, sizeof header1, &gsize, "(stdin)"); 1667 break; 1668 #ifndef NO_BZIP2_SUPPORT 1669 case FT_BZIP2: 1670 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1671 (char *)header1, sizeof header1, &gsize); 1672 break; 1673 #endif 1674 #ifndef NO_COMPRESS_SUPPORT 1675 case FT_Z: 1676 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1677 maybe_warnx("zopen of stdin"); 1678 return; 1679 } 1680 1681 usize = zuncompress(in, stdout, (char *)header1, sizeof header1, &gsize); 1682 fclose(in); 1683 break; 1684 #endif 1685 #ifndef NO_PACK_SUPPORT 1686 case FT_PACK: 1687 usize = unpack(STDIN_FILENO, STDOUT_FILENO, 1688 (char *)header1, sizeof header1, &gsize); 1689 break; 1690 #endif 1691 } 1692 1693 #ifndef SMALL 1694 if (vflag && !tflag && usize != -1 && gsize != -1) 1695 print_verbage(NULL, NULL, usize, gsize); 1696 if (vflag && tflag) 1697 print_test("(stdin)", usize != -1); 1698 #endif 1699 1700 } 1701 1702 static void 1703 handle_stdout(void) 1704 { 1705 off_t gsize, usize; 1706 struct stat sb; 1707 time_t systime; 1708 uint32_t mtime; 1709 int ret; 1710 1711 #ifndef SMALL 1712 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1713 maybe_warnx("standard output is a terminal -- ignoring"); 1714 return; 1715 } 1716 #endif 1717 /* If stdin is a file use it's mtime, otherwise use current time */ 1718 ret = fstat(STDIN_FILENO, &sb); 1719 1720 #ifndef SMALL 1721 if (ret < 0) { 1722 maybe_warn("Can't stat stdin"); 1723 return; 1724 } 1725 #endif 1726 1727 if (S_ISREG(sb.st_mode)) 1728 mtime = (uint32_t)sb.st_mtime; 1729 else { 1730 systime = time(NULL); 1731 #ifndef SMALL 1732 if (systime == -1) { 1733 maybe_warn("time"); 1734 return; 1735 } 1736 #endif 1737 mtime = (uint32_t)systime; 1738 } 1739 1740 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1741 #ifndef SMALL 1742 if (vflag && !tflag && usize != -1 && gsize != -1) 1743 print_verbage(NULL, NULL, usize, gsize); 1744 #endif 1745 } 1746 1747 /* do what is asked for, for the path name */ 1748 static void 1749 handle_pathname(char *path) 1750 { 1751 char *opath = path, *s = NULL; 1752 ssize_t len; 1753 int slen; 1754 struct stat sb; 1755 1756 /* check for stdout/stdin */ 1757 if (path[0] == '-' && path[1] == '\0') { 1758 if (dflag) 1759 handle_stdin(); 1760 else 1761 handle_stdout(); 1762 return; 1763 } 1764 1765 retry: 1766 if (stat(path, &sb) != 0) { 1767 /* lets try <path>.gz if we're decompressing */ 1768 if (dflag && s == NULL && errno == ENOENT) { 1769 len = strlen(path); 1770 slen = suffixes[0].ziplen; 1771 s = malloc(len + slen + 1); 1772 if (s == NULL) 1773 maybe_err("malloc"); 1774 memcpy(s, path, len); 1775 memcpy(s + len, suffixes[0].zipped, slen + 1); 1776 path = s; 1777 goto retry; 1778 } 1779 maybe_warn("can't stat: %s", opath); 1780 goto out; 1781 } 1782 1783 if (S_ISDIR(sb.st_mode)) { 1784 #ifndef SMALL 1785 if (rflag) 1786 handle_dir(path); 1787 else 1788 #endif 1789 maybe_warnx("%s is a directory", path); 1790 goto out; 1791 } 1792 1793 if (S_ISREG(sb.st_mode)) 1794 handle_file(path, &sb); 1795 else 1796 maybe_warnx("%s is not a regular file", path); 1797 1798 out: 1799 if (s) 1800 free(s); 1801 } 1802 1803 /* compress/decompress a file */ 1804 static void 1805 handle_file(char *file, struct stat *sbp) 1806 { 1807 off_t usize, gsize; 1808 char outfile[PATH_MAX]; 1809 1810 infile = file; 1811 if (dflag) { 1812 usize = file_uncompress(file, outfile, sizeof(outfile)); 1813 #ifndef SMALL 1814 if (vflag && tflag) 1815 print_test(file, usize != -1); 1816 #endif 1817 if (usize == -1) 1818 return; 1819 gsize = sbp->st_size; 1820 } else { 1821 gsize = file_compress(file, outfile, sizeof(outfile)); 1822 if (gsize == -1) 1823 return; 1824 usize = sbp->st_size; 1825 } 1826 1827 1828 #ifndef SMALL 1829 if (vflag && !tflag) 1830 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1831 #endif 1832 } 1833 1834 #ifndef SMALL 1835 /* this is used with -r to recursively descend directories */ 1836 static void 1837 handle_dir(char *dir) 1838 { 1839 char *path_argv[2]; 1840 FTS *fts; 1841 FTSENT *entry; 1842 1843 path_argv[0] = dir; 1844 path_argv[1] = 0; 1845 fts = fts_open(path_argv, FTS_PHYSICAL | FTS_NOCHDIR, NULL); 1846 if (fts == NULL) { 1847 warn("couldn't fts_open %s", dir); 1848 return; 1849 } 1850 1851 while ((entry = fts_read(fts))) { 1852 switch(entry->fts_info) { 1853 case FTS_D: 1854 case FTS_DP: 1855 continue; 1856 1857 case FTS_DNR: 1858 case FTS_ERR: 1859 case FTS_NS: 1860 maybe_warn("%s", entry->fts_path); 1861 continue; 1862 case FTS_F: 1863 handle_file(entry->fts_path, entry->fts_statp); 1864 } 1865 } 1866 (void)fts_close(fts); 1867 } 1868 #endif 1869 1870 /* print a ratio - size reduction as a fraction of uncompressed size */ 1871 static void 1872 print_ratio(off_t in, off_t out, FILE *where) 1873 { 1874 int percent10; /* 10 * percent */ 1875 off_t diff; 1876 char buff[8]; 1877 int len; 1878 1879 diff = in - out/2; 1880 if (diff <= 0) 1881 /* 1882 * Output is more than double size of input! print -99.9% 1883 * Quite possibly we've failed to get the original size. 1884 */ 1885 percent10 = -999; 1886 else { 1887 /* 1888 * We only need 12 bits of result from the final division, 1889 * so reduce the values until a 32bit division will suffice. 1890 */ 1891 while (in > 0x100000) { 1892 diff >>= 1; 1893 in >>= 1; 1894 } 1895 if (in != 0) 1896 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1897 else 1898 percent10 = 0; 1899 } 1900 1901 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1902 /* Move the '.' to before the last digit */ 1903 buff[len - 1] = buff[len - 2]; 1904 buff[len - 2] = '.'; 1905 fprintf(where, "%5s%%", buff); 1906 } 1907 1908 #ifndef SMALL 1909 /* print compression statistics, and the new name (if there is one!) */ 1910 static void 1911 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1912 { 1913 if (file) 1914 fprintf(stderr, "%s:%s ", file, 1915 strlen(file) < 7 ? "\t\t" : "\t"); 1916 print_ratio(usize, gsize, stderr); 1917 if (nfile) 1918 fprintf(stderr, " -- replaced with %s", nfile); 1919 fprintf(stderr, "\n"); 1920 fflush(stderr); 1921 } 1922 1923 /* print test results */ 1924 static void 1925 print_test(const char *file, int ok) 1926 { 1927 1928 if (exit_value == 0 && ok == 0) 1929 exit_value = 1; 1930 fprintf(stderr, "%s:%s %s\n", file, 1931 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1932 fflush(stderr); 1933 } 1934 #endif 1935 1936 /* print a file's info ala --list */ 1937 /* eg: 1938 compressed uncompressed ratio uncompressed_name 1939 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1940 */ 1941 static void 1942 print_list(int fd, off_t out, const char *outfile, time_t ts) 1943 { 1944 static int first = 1; 1945 #ifndef SMALL 1946 static off_t in_tot, out_tot; 1947 uint32_t crc = 0; 1948 #endif 1949 off_t in = 0, rv; 1950 1951 if (first) { 1952 #ifndef SMALL 1953 if (vflag) 1954 printf("method crc date time "); 1955 #endif 1956 if (qflag == 0) 1957 printf(" compressed uncompressed " 1958 "ratio uncompressed_name\n"); 1959 } 1960 first = 0; 1961 1962 /* print totals? */ 1963 #ifndef SMALL 1964 if (fd == -1) { 1965 in = in_tot; 1966 out = out_tot; 1967 } else 1968 #endif 1969 { 1970 /* read the last 4 bytes - this is the uncompressed size */ 1971 rv = lseek(fd, (off_t)(-8), SEEK_END); 1972 if (rv != -1) { 1973 unsigned char buf[8]; 1974 uint32_t usize; 1975 1976 rv = read(fd, (char *)buf, sizeof(buf)); 1977 if (rv == -1) 1978 maybe_warn("read of uncompressed size"); 1979 else if (rv != sizeof(buf)) 1980 maybe_warnx("read of uncompressed size"); 1981 1982 else { 1983 usize = buf[4] | buf[5] << 8 | 1984 buf[6] << 16 | buf[7] << 24; 1985 in = (off_t)usize; 1986 #ifndef SMALL 1987 crc = buf[0] | buf[1] << 8 | 1988 buf[2] << 16 | buf[3] << 24; 1989 #endif 1990 } 1991 } 1992 } 1993 1994 #ifndef SMALL 1995 if (vflag && fd == -1) 1996 printf(" "); 1997 else if (vflag) { 1998 char *date = ctime(&ts); 1999 2000 /* skip the day, 1/100th second, and year */ 2001 date += 4; 2002 date[12] = 0; 2003 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 2004 } 2005 in_tot += in; 2006 out_tot += out; 2007 #else 2008 (void)&ts; /* XXX */ 2009 #endif 2010 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 2011 print_ratio(in, out, stdout); 2012 printf(" %s\n", outfile); 2013 } 2014 2015 /* display the usage of NetBSD gzip */ 2016 static void 2017 usage(void) 2018 { 2019 2020 fprintf(stderr, "%s\n", gzip_version); 2021 fprintf(stderr, 2022 #ifdef SMALL 2023 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n", 2024 #else 2025 "usage: %s [-123456789acdfhklLNnqrtVv] [-S .suffix] [<file> [<file> ...]]\n" 2026 " -1 --fast fastest (worst) compression\n" 2027 " -2 .. -8 set compression level\n" 2028 " -9 --best best (slowest) compression\n" 2029 " -c --stdout write to stdout, keep original files\n" 2030 " --to-stdout\n" 2031 " -d --decompress uncompress files\n" 2032 " --uncompress\n" 2033 " -f --force force overwriting & compress links\n" 2034 " -h --help display this help\n" 2035 " -k --keep don't delete input files during operation\n" 2036 " -l --list list compressed file contents\n" 2037 " -N --name save or restore original file name and time stamp\n" 2038 " -n --no-name don't save original file name or time stamp\n" 2039 " -q --quiet output no warnings\n" 2040 " -r --recursive recursively compress files in directories\n" 2041 " -S .suf use suffix .suf instead of .gz\n" 2042 " --suffix .suf\n" 2043 " -t --test test compressed file\n" 2044 " -V --version display program version\n" 2045 " -v --verbose print extra statistics\n", 2046 #endif 2047 getprogname()); 2048 exit(0); 2049 } 2050 2051 #ifndef SMALL 2052 /* display the license information of FreeBSD gzip */ 2053 static void 2054 display_license(void) 2055 { 2056 2057 fprintf(stderr, "%s (based on NetBSD gzip 20060927)\n", gzip_version); 2058 fprintf(stderr, "%s\n", gzip_copyright); 2059 exit(0); 2060 } 2061 #endif 2062 2063 /* display the version of NetBSD gzip */ 2064 static void 2065 display_version(void) 2066 { 2067 2068 fprintf(stderr, "%s\n", gzip_version); 2069 exit(0); 2070 } 2071 2072 #ifndef NO_BZIP2_SUPPORT 2073 #include "unbzip2.c" 2074 #endif 2075 #ifndef NO_COMPRESS_SUPPORT 2076 #include "zuncompress.c" 2077 #endif 2078 #ifndef NO_PACK_SUPPORT 2079 #include "unpack.c" 2080 #endif 2081 2082 static ssize_t 2083 read_retry(int fd, void *buf, size_t sz) 2084 { 2085 char *cp = buf; 2086 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2087 2088 while (left > 0) { 2089 ssize_t ret; 2090 2091 ret = read(fd, cp, left); 2092 if (ret == -1) { 2093 return ret; 2094 } else if (ret == 0) { 2095 break; /* EOF */ 2096 } 2097 cp += ret; 2098 left -= ret; 2099 } 2100 2101 return sz - left; 2102 } 2103 2104