1 /* $NetBSD: gzip.c,v 1.92 2008/07/21 14:19:22 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 FT_LAST, 83 FT_UNKNOWN 84 }; 85 86 #ifndef NO_BZIP2_SUPPORT 87 #include <bzlib.h> 88 89 #define BZ2_SUFFIX ".bz2" 90 #define BZIP2_MAGIC "\102\132\150" 91 #endif 92 93 #ifndef NO_COMPRESS_SUPPORT 94 #define Z_SUFFIX ".Z" 95 #define Z_MAGIC "\037\235" 96 #endif 97 98 #define GZ_SUFFIX ".gz" 99 100 #define BUFLEN (64 * 1024) 101 102 #define GZIP_MAGIC0 0x1F 103 #define GZIP_MAGIC1 0x8B 104 #define GZIP_OMAGIC1 0x9E 105 106 #define GZIP_TIMESTAMP (off_t)4 107 #define GZIP_ORIGNAME (off_t)10 108 109 #define HEAD_CRC 0x02 110 #define EXTRA_FIELD 0x04 111 #define ORIG_NAME 0x08 112 #define COMMENT 0x10 113 114 #define OS_CODE 3 /* Unix */ 115 116 typedef struct { 117 const char *zipped; 118 int ziplen; 119 const char *normal; /* for unzip - must not be longer than zipped */ 120 } suffixes_t; 121 static suffixes_t suffixes[] = { 122 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 123 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 124 #ifndef SMALL 125 SUFFIX(GZ_SUFFIX, ""), 126 SUFFIX(".z", ""), 127 SUFFIX("-gz", ""), 128 SUFFIX("-z", ""), 129 SUFFIX("_z", ""), 130 SUFFIX(".taz", ".tar"), 131 SUFFIX(".tgz", ".tar"), 132 #ifndef NO_BZIP2_SUPPORT 133 SUFFIX(BZ2_SUFFIX, ""), 134 SUFFIX(".tbz", ".tar"), 135 SUFFIX(".tbz2", ".tar"), 136 #endif 137 #ifndef NO_COMPRESS_SUPPORT 138 SUFFIX(Z_SUFFIX, ""), 139 #endif 140 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 141 #endif /* SMALL */ 142 #undef SUFFIX 143 }; 144 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 145 146 static const char gzip_version[] = "FreeBSD gzip 20070711"; 147 148 #ifndef SMALL 149 static const char gzip_copyright[] = \ 150 " Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green\n" 151 " All rights reserved.\n" 152 "\n" 153 " Redistribution and use in source and binary forms, with or without\n" 154 " modification, are permitted provided that the following conditions\n" 155 " are met:\n" 156 " 1. Redistributions of source code must retain the above copyright\n" 157 " notice, this list of conditions and the following disclaimer.\n" 158 " 2. Redistributions in binary form must reproduce the above copyright\n" 159 " notice, this list of conditions and the following disclaimer in the\n" 160 " documentation and/or other materials provided with the distribution.\n" 161 "\n" 162 " THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR\n" 163 " IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES\n" 164 " OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.\n" 165 " IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,\n" 166 " INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,\n" 167 " BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;\n" 168 " LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED\n" 169 " AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,\n" 170 " OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY\n" 171 " OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF\n" 172 " SUCH DAMAGE."; 173 #endif 174 175 static int cflag; /* stdout mode */ 176 static int dflag; /* decompress mode */ 177 static int lflag; /* list mode */ 178 static int numflag = 6; /* gzip -1..-9 value */ 179 180 #ifndef SMALL 181 static int fflag; /* force mode */ 182 static int kflag; /* don't delete input files */ 183 static int nflag; /* don't save name/timestamp */ 184 static int Nflag; /* don't restore name/timestamp */ 185 static int qflag; /* quiet mode */ 186 static int rflag; /* recursive mode */ 187 static int tflag; /* test */ 188 static int vflag; /* verbose mode */ 189 #else 190 #define qflag 0 191 #define tflag 0 192 #endif 193 194 static int exit_value = 0; /* exit value */ 195 196 static char *infile; /* name of file coming in */ 197 198 static void maybe_err(const char *fmt, ...) __dead2 199 __attribute__((__format__(__printf__, 1, 2))); 200 #ifndef NO_BZIP2_SUPPORT 201 static void maybe_errx(const char *fmt, ...) __dead2 202 __attribute__((__format__(__printf__, 1, 2))); 203 #endif 204 static void maybe_warn(const char *fmt, ...) 205 __attribute__((__format__(__printf__, 1, 2))); 206 static void maybe_warnx(const char *fmt, ...) 207 __attribute__((__format__(__printf__, 1, 2))); 208 static enum filetype file_gettype(u_char *); 209 #ifdef SMALL 210 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 211 #endif 212 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 213 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 214 static off_t file_compress(char *, char *, size_t); 215 static off_t file_uncompress(char *, char *, size_t); 216 static void handle_pathname(char *); 217 static void handle_file(char *, struct stat *); 218 static void handle_stdin(void); 219 static void handle_stdout(void); 220 static void print_ratio(off_t, off_t, FILE *); 221 static void print_list(int fd, off_t, const char *, time_t); 222 static void usage(void); 223 static void display_version(void); 224 #ifndef SMALL 225 static void display_license(void); 226 #endif 227 static const suffixes_t *check_suffix(char *, int); 228 static ssize_t read_retry(int, void *, size_t); 229 230 #ifdef SMALL 231 #define unlink_input(f, sb) unlink(f) 232 #else 233 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 234 static void prepend_gzip(char *, int *, char ***); 235 static void handle_dir(char *); 236 static void print_verbage(const char *, const char *, off_t, off_t); 237 static void print_test(const char *, int); 238 static void copymodes(int fd, const struct stat *, const char *file); 239 static int check_outfile(const char *outfile); 240 #endif 241 242 #ifndef NO_BZIP2_SUPPORT 243 static off_t unbzip2(int, int, char *, size_t, off_t *); 244 #endif 245 246 #ifndef NO_COMPRESS_SUPPORT 247 static FILE *zdopen(int); 248 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 249 #endif 250 251 int main(int, char **p); 252 253 #ifdef SMALL 254 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 255 #else 256 static const struct option longopts[] = { 257 { "stdout", no_argument, 0, 'c' }, 258 { "to-stdout", no_argument, 0, 'c' }, 259 { "decompress", no_argument, 0, 'd' }, 260 { "uncompress", no_argument, 0, 'd' }, 261 { "force", no_argument, 0, 'f' }, 262 { "help", no_argument, 0, 'h' }, 263 { "keep", no_argument, 0, 'k' }, 264 { "list", no_argument, 0, 'l' }, 265 { "no-name", no_argument, 0, 'n' }, 266 { "name", no_argument, 0, 'N' }, 267 { "quiet", no_argument, 0, 'q' }, 268 { "recursive", no_argument, 0, 'r' }, 269 { "suffix", required_argument, 0, 'S' }, 270 { "test", no_argument, 0, 't' }, 271 { "verbose", no_argument, 0, 'v' }, 272 { "version", no_argument, 0, 'V' }, 273 { "fast", no_argument, 0, '1' }, 274 { "best", no_argument, 0, '9' }, 275 { "ascii", no_argument, 0, 'a' }, 276 { "license", no_argument, 0, 'L' }, 277 { NULL, no_argument, 0, 0 }, 278 }; 279 #endif 280 281 int 282 main(int argc, char **argv) 283 { 284 const char *progname = getprogname(); 285 #ifndef SMALL 286 char *gzip; 287 int len; 288 #endif 289 int ch; 290 291 /* XXX set up signals */ 292 293 #ifndef SMALL 294 if ((gzip = getenv("GZIP")) != NULL) 295 prepend_gzip(gzip, &argc, &argv); 296 #endif 297 298 /* 299 * XXX 300 * handle being called `gunzip', `zcat' and `gzcat' 301 */ 302 if (strcmp(progname, "gunzip") == 0) 303 dflag = 1; 304 else if (strcmp(progname, "zcat") == 0 || 305 strcmp(progname, "gzcat") == 0) 306 dflag = cflag = 1; 307 308 #ifdef SMALL 309 #define OPT_LIST "123456789cdhltV" 310 #else 311 #define OPT_LIST "123456789acdfhklLNnqrS:tVv" 312 #endif 313 314 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 315 switch (ch) { 316 case '1': case '2': case '3': 317 case '4': case '5': case '6': 318 case '7': case '8': case '9': 319 numflag = ch - '0'; 320 break; 321 case 'c': 322 cflag = 1; 323 break; 324 case 'd': 325 dflag = 1; 326 break; 327 case 'l': 328 lflag = 1; 329 dflag = 1; 330 break; 331 case 'V': 332 display_version(); 333 /* NOTREACHED */ 334 #ifndef SMALL 335 case 'a': 336 fprintf(stderr, "%s: option --ascii ignored on this system\n", progname); 337 break; 338 case 'f': 339 fflag = 1; 340 break; 341 case 'k': 342 kflag = 1; 343 break; 344 case 'L': 345 display_license(); 346 /* NOT REACHED */ 347 case 'N': 348 nflag = 0; 349 Nflag = 1; 350 break; 351 case 'n': 352 nflag = 1; 353 Nflag = 0; 354 break; 355 case 'q': 356 qflag = 1; 357 break; 358 case 'r': 359 rflag = 1; 360 break; 361 case 'S': 362 len = strlen(optarg); 363 if (len != 0) { 364 suffixes[0].zipped = optarg; 365 suffixes[0].ziplen = len; 366 } else { 367 suffixes[NUM_SUFFIXES - 1].zipped = ""; 368 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 369 } 370 break; 371 case 't': 372 cflag = 1; 373 tflag = 1; 374 dflag = 1; 375 break; 376 case 'v': 377 vflag = 1; 378 break; 379 #endif 380 default: 381 usage(); 382 /* NOTREACHED */ 383 } 384 } 385 argv += optind; 386 argc -= optind; 387 388 if (argc == 0) { 389 if (dflag) /* stdin mode */ 390 handle_stdin(); 391 else /* stdout mode */ 392 handle_stdout(); 393 } else { 394 do { 395 handle_pathname(argv[0]); 396 } while (*++argv); 397 } 398 #ifndef SMALL 399 if (qflag == 0 && lflag && argc > 1) 400 print_list(-1, 0, "(totals)", 0); 401 #endif 402 exit(exit_value); 403 } 404 405 /* maybe print a warning */ 406 void 407 maybe_warn(const char *fmt, ...) 408 { 409 va_list ap; 410 411 if (qflag == 0) { 412 va_start(ap, fmt); 413 vwarn(fmt, ap); 414 va_end(ap); 415 } 416 if (exit_value == 0) 417 exit_value = 1; 418 } 419 420 /* ... without an errno. */ 421 void 422 maybe_warnx(const char *fmt, ...) 423 { 424 va_list ap; 425 426 if (qflag == 0) { 427 va_start(ap, fmt); 428 vwarnx(fmt, ap); 429 va_end(ap); 430 } 431 if (exit_value == 0) 432 exit_value = 1; 433 } 434 435 /* maybe print an error */ 436 void 437 maybe_err(const char *fmt, ...) 438 { 439 va_list ap; 440 441 if (qflag == 0) { 442 va_start(ap, fmt); 443 vwarn(fmt, ap); 444 va_end(ap); 445 } 446 exit(2); 447 } 448 449 #ifndef NO_BZIP2_SUPPORT 450 /* ... without an errno. */ 451 void 452 maybe_errx(const char *fmt, ...) 453 { 454 va_list ap; 455 456 if (qflag == 0) { 457 va_start(ap, fmt); 458 vwarnx(fmt, ap); 459 va_end(ap); 460 } 461 exit(2); 462 } 463 #endif 464 465 #ifndef SMALL 466 /* split up $GZIP and prepend it to the argument list */ 467 static void 468 prepend_gzip(char *gzip, int *argc, char ***argv) 469 { 470 char *s, **nargv, **ac; 471 int nenvarg = 0, i; 472 473 /* scan how many arguments there are */ 474 for (s = gzip;;) { 475 while (*s == ' ' || *s == '\t') 476 s++; 477 if (*s == 0) 478 goto count_done; 479 nenvarg++; 480 while (*s != ' ' && *s != '\t') 481 if (*s++ == 0) 482 goto count_done; 483 } 484 count_done: 485 /* punt early */ 486 if (nenvarg == 0) 487 return; 488 489 *argc += nenvarg; 490 ac = *argv; 491 492 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 493 if (nargv == NULL) 494 maybe_err("malloc"); 495 496 /* stash this away */ 497 *argv = nargv; 498 499 /* copy the program name first */ 500 i = 0; 501 nargv[i++] = *(ac++); 502 503 /* take a copy of $GZIP and add it to the array */ 504 s = strdup(gzip); 505 if (s == NULL) 506 maybe_err("strdup"); 507 for (;;) { 508 /* Skip whitespaces. */ 509 while (*s == ' ' || *s == '\t') 510 s++; 511 if (*s == 0) 512 goto copy_done; 513 nargv[i++] = s; 514 /* Find the end of this argument. */ 515 while (*s != ' ' && *s != '\t') 516 if (*s++ == 0) 517 /* Argument followed by NUL. */ 518 goto copy_done; 519 /* Terminate by overwriting ' ' or '\t' with NUL. */ 520 *s++ = 0; 521 } 522 copy_done: 523 524 /* copy the original arguments and a NULL */ 525 while (*ac) 526 nargv[i++] = *(ac++); 527 nargv[i] = NULL; 528 } 529 #endif 530 531 /* compress input to output. Return bytes read, -1 on error */ 532 static off_t 533 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 534 { 535 z_stream z; 536 char *outbufp, *inbufp; 537 off_t in_tot = 0, out_tot = 0; 538 ssize_t in_size; 539 int i, error; 540 uLong crc; 541 #ifdef SMALL 542 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 543 0, 0, 0, 0, 544 0, OS_CODE }; 545 #endif 546 547 outbufp = malloc(BUFLEN); 548 inbufp = malloc(BUFLEN); 549 if (outbufp == NULL || inbufp == NULL) { 550 maybe_err("malloc failed"); 551 goto out; 552 } 553 554 memset(&z, 0, sizeof z); 555 z.zalloc = Z_NULL; 556 z.zfree = Z_NULL; 557 z.opaque = 0; 558 559 #ifdef SMALL 560 memcpy(outbufp, header, sizeof header); 561 i = sizeof header; 562 #else 563 if (nflag != 0) { 564 mtime = 0; 565 origname = ""; 566 } 567 568 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 569 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 570 *origname ? ORIG_NAME : 0, 571 mtime & 0xff, 572 (mtime >> 8) & 0xff, 573 (mtime >> 16) & 0xff, 574 (mtime >> 24) & 0xff, 575 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 576 OS_CODE, origname); 577 if (i >= BUFLEN) 578 /* this need PATH_MAX > BUFLEN ... */ 579 maybe_err("snprintf"); 580 if (*origname) 581 i++; 582 #endif 583 584 z.next_out = (unsigned char *)outbufp + i; 585 z.avail_out = BUFLEN - i; 586 587 error = deflateInit2(&z, numflag, Z_DEFLATED, 588 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 589 if (error != Z_OK) { 590 maybe_warnx("deflateInit2 failed"); 591 in_tot = -1; 592 goto out; 593 } 594 595 crc = crc32(0L, Z_NULL, 0); 596 for (;;) { 597 if (z.avail_out == 0) { 598 if (write(out, outbufp, BUFLEN) != BUFLEN) { 599 maybe_warn("write"); 600 out_tot = -1; 601 goto out; 602 } 603 604 out_tot += BUFLEN; 605 z.next_out = (unsigned char *)outbufp; 606 z.avail_out = BUFLEN; 607 } 608 609 if (z.avail_in == 0) { 610 in_size = read(in, inbufp, BUFLEN); 611 if (in_size < 0) { 612 maybe_warn("read"); 613 in_tot = -1; 614 goto out; 615 } 616 if (in_size == 0) 617 break; 618 619 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 620 in_tot += in_size; 621 z.next_in = (unsigned char *)inbufp; 622 z.avail_in = in_size; 623 } 624 625 error = deflate(&z, Z_NO_FLUSH); 626 if (error != Z_OK && error != Z_STREAM_END) { 627 maybe_warnx("deflate failed"); 628 in_tot = -1; 629 goto out; 630 } 631 } 632 633 /* clean up */ 634 for (;;) { 635 size_t len; 636 ssize_t w; 637 638 error = deflate(&z, Z_FINISH); 639 if (error != Z_OK && error != Z_STREAM_END) { 640 maybe_warnx("deflate failed"); 641 in_tot = -1; 642 goto out; 643 } 644 645 len = (char *)z.next_out - outbufp; 646 647 w = write(out, outbufp, len); 648 if (w == -1 || (size_t)w != len) { 649 maybe_warn("write"); 650 out_tot = -1; 651 goto out; 652 } 653 out_tot += len; 654 z.next_out = (unsigned char *)outbufp; 655 z.avail_out = BUFLEN; 656 657 if (error == Z_STREAM_END) 658 break; 659 } 660 661 if (deflateEnd(&z) != Z_OK) { 662 maybe_warnx("deflateEnd failed"); 663 in_tot = -1; 664 goto out; 665 } 666 667 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 668 (int)crc & 0xff, 669 (int)(crc >> 8) & 0xff, 670 (int)(crc >> 16) & 0xff, 671 (int)(crc >> 24) & 0xff, 672 (int)in_tot & 0xff, 673 (int)(in_tot >> 8) & 0xff, 674 (int)(in_tot >> 16) & 0xff, 675 (int)(in_tot >> 24) & 0xff); 676 if (i != 8) 677 maybe_err("snprintf"); 678 if (write(out, outbufp, i) != i) { 679 maybe_warn("write"); 680 in_tot = -1; 681 } else 682 out_tot += i; 683 684 out: 685 if (inbufp != NULL) 686 free(inbufp); 687 if (outbufp != NULL) 688 free(outbufp); 689 if (gsizep) 690 *gsizep = out_tot; 691 return in_tot; 692 } 693 694 /* 695 * uncompress input to output then close the input. return the 696 * uncompressed size written, and put the compressed sized read 697 * into `*gsizep'. 698 */ 699 static off_t 700 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 701 const char *filename) 702 { 703 z_stream z; 704 char *outbufp, *inbufp; 705 off_t out_tot = -1, in_tot = 0; 706 uint32_t out_sub_tot = 0; 707 enum { 708 GZSTATE_MAGIC0, 709 GZSTATE_MAGIC1, 710 GZSTATE_METHOD, 711 GZSTATE_FLAGS, 712 GZSTATE_SKIPPING, 713 GZSTATE_EXTRA, 714 GZSTATE_EXTRA2, 715 GZSTATE_EXTRA3, 716 GZSTATE_ORIGNAME, 717 GZSTATE_COMMENT, 718 GZSTATE_HEAD_CRC1, 719 GZSTATE_HEAD_CRC2, 720 GZSTATE_INIT, 721 GZSTATE_READ, 722 GZSTATE_CRC, 723 GZSTATE_LEN, 724 } state = GZSTATE_MAGIC0; 725 int flags = 0, skip_count = 0; 726 int error = Z_STREAM_ERROR, done_reading = 0; 727 uLong crc = 0; 728 ssize_t wr; 729 int needmore = 0; 730 731 #define ADVANCE() { z.next_in++; z.avail_in--; } 732 733 if ((outbufp = malloc(BUFLEN)) == NULL) { 734 maybe_err("malloc failed"); 735 goto out2; 736 } 737 if ((inbufp = malloc(BUFLEN)) == NULL) { 738 maybe_err("malloc failed"); 739 goto out1; 740 } 741 742 memset(&z, 0, sizeof z); 743 z.avail_in = prelen; 744 z.next_in = (unsigned char *)pre; 745 z.avail_out = BUFLEN; 746 z.next_out = (unsigned char *)outbufp; 747 z.zalloc = NULL; 748 z.zfree = NULL; 749 z.opaque = 0; 750 751 in_tot = prelen; 752 out_tot = 0; 753 754 for (;;) { 755 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 756 ssize_t in_size; 757 758 if (z.avail_in > 0) { 759 memmove(inbufp, z.next_in, z.avail_in); 760 } 761 z.next_in = (unsigned char *)inbufp; 762 in_size = read(in, z.next_in + z.avail_in, 763 BUFLEN - z.avail_in); 764 765 if (in_size == -1) { 766 maybe_warn("failed to read stdin"); 767 goto stop_and_fail; 768 } else if (in_size == 0) { 769 done_reading = 1; 770 } 771 772 z.avail_in += in_size; 773 needmore = 0; 774 775 in_tot += in_size; 776 } 777 if (z.avail_in == 0) { 778 if (done_reading && state != GZSTATE_MAGIC0) { 779 maybe_warnx("%s: unexpected end of file", 780 filename); 781 goto stop_and_fail; 782 } 783 goto stop; 784 } 785 switch (state) { 786 case GZSTATE_MAGIC0: 787 if (*z.next_in != GZIP_MAGIC0) { 788 if (in_tot > 0) { 789 maybe_warnx("%s: trailing garbage " 790 "ignored", filename); 791 goto stop; 792 } 793 maybe_warnx("input not gziped (MAGIC0)"); 794 goto stop_and_fail; 795 } 796 ADVANCE(); 797 state++; 798 out_sub_tot = 0; 799 crc = crc32(0L, Z_NULL, 0); 800 break; 801 802 case GZSTATE_MAGIC1: 803 if (*z.next_in != GZIP_MAGIC1 && 804 *z.next_in != GZIP_OMAGIC1) { 805 maybe_warnx("input not gziped (MAGIC1)"); 806 goto stop_and_fail; 807 } 808 ADVANCE(); 809 state++; 810 break; 811 812 case GZSTATE_METHOD: 813 if (*z.next_in != Z_DEFLATED) { 814 maybe_warnx("unknown compression method"); 815 goto stop_and_fail; 816 } 817 ADVANCE(); 818 state++; 819 break; 820 821 case GZSTATE_FLAGS: 822 flags = *z.next_in; 823 ADVANCE(); 824 skip_count = 6; 825 state++; 826 break; 827 828 case GZSTATE_SKIPPING: 829 if (skip_count > 0) { 830 skip_count--; 831 ADVANCE(); 832 } else 833 state++; 834 break; 835 836 case GZSTATE_EXTRA: 837 if ((flags & EXTRA_FIELD) == 0) { 838 state = GZSTATE_ORIGNAME; 839 break; 840 } 841 skip_count = *z.next_in; 842 ADVANCE(); 843 state++; 844 break; 845 846 case GZSTATE_EXTRA2: 847 skip_count |= ((*z.next_in) << 8); 848 ADVANCE(); 849 state++; 850 break; 851 852 case GZSTATE_EXTRA3: 853 if (skip_count > 0) { 854 skip_count--; 855 ADVANCE(); 856 } else 857 state++; 858 break; 859 860 case GZSTATE_ORIGNAME: 861 if ((flags & ORIG_NAME) == 0) { 862 state++; 863 break; 864 } 865 if (*z.next_in == 0) 866 state++; 867 ADVANCE(); 868 break; 869 870 case GZSTATE_COMMENT: 871 if ((flags & COMMENT) == 0) { 872 state++; 873 break; 874 } 875 if (*z.next_in == 0) 876 state++; 877 ADVANCE(); 878 break; 879 880 case GZSTATE_HEAD_CRC1: 881 if (flags & HEAD_CRC) 882 skip_count = 2; 883 else 884 skip_count = 0; 885 state++; 886 break; 887 888 case GZSTATE_HEAD_CRC2: 889 if (skip_count > 0) { 890 skip_count--; 891 ADVANCE(); 892 } else 893 state++; 894 break; 895 896 case GZSTATE_INIT: 897 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 898 maybe_warnx("failed to inflateInit"); 899 goto stop_and_fail; 900 } 901 state++; 902 break; 903 904 case GZSTATE_READ: 905 error = inflate(&z, Z_FINISH); 906 switch (error) { 907 /* Z_BUF_ERROR goes with Z_FINISH... */ 908 case Z_BUF_ERROR: 909 case Z_STREAM_END: 910 case Z_OK: 911 break; 912 913 case Z_NEED_DICT: 914 maybe_warnx("Z_NEED_DICT error"); 915 goto stop_and_fail; 916 case Z_DATA_ERROR: 917 maybe_warnx("data stream error"); 918 goto stop_and_fail; 919 case Z_STREAM_ERROR: 920 maybe_warnx("internal stream error"); 921 goto stop_and_fail; 922 case Z_MEM_ERROR: 923 maybe_warnx("memory allocation error"); 924 goto stop_and_fail; 925 926 default: 927 maybe_warn("unknown error from inflate(): %d", 928 error); 929 } 930 wr = BUFLEN - z.avail_out; 931 932 if (wr != 0) { 933 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 934 if ( 935 #ifndef SMALL 936 /* don't write anything with -t */ 937 tflag == 0 && 938 #endif 939 write(out, outbufp, wr) != wr) { 940 maybe_warn("error writing to output"); 941 goto stop_and_fail; 942 } 943 944 out_tot += wr; 945 out_sub_tot += wr; 946 } 947 948 if (error == Z_STREAM_END) { 949 inflateEnd(&z); 950 state++; 951 } 952 953 z.next_out = (unsigned char *)outbufp; 954 z.avail_out = BUFLEN; 955 956 break; 957 case GZSTATE_CRC: 958 { 959 uLong origcrc; 960 961 if (z.avail_in < 4) { 962 if (!done_reading) { 963 needmore = 1; 964 continue; 965 } 966 maybe_warnx("truncated input"); 967 goto stop_and_fail; 968 } 969 origcrc = ((unsigned)z.next_in[0] & 0xff) | 970 ((unsigned)z.next_in[1] & 0xff) << 8 | 971 ((unsigned)z.next_in[2] & 0xff) << 16 | 972 ((unsigned)z.next_in[3] & 0xff) << 24; 973 if (origcrc != crc) { 974 maybe_warnx("invalid compressed" 975 " data--crc error"); 976 goto stop_and_fail; 977 } 978 } 979 980 z.avail_in -= 4; 981 z.next_in += 4; 982 983 if (!z.avail_in && done_reading) { 984 goto stop; 985 } 986 state++; 987 break; 988 case GZSTATE_LEN: 989 { 990 uLong origlen; 991 992 if (z.avail_in < 4) { 993 if (!done_reading) { 994 needmore = 1; 995 continue; 996 } 997 maybe_warnx("truncated input"); 998 goto stop_and_fail; 999 } 1000 origlen = ((unsigned)z.next_in[0] & 0xff) | 1001 ((unsigned)z.next_in[1] & 0xff) << 8 | 1002 ((unsigned)z.next_in[2] & 0xff) << 16 | 1003 ((unsigned)z.next_in[3] & 0xff) << 24; 1004 1005 if (origlen != out_sub_tot) { 1006 maybe_warnx("invalid compressed" 1007 " data--length error"); 1008 goto stop_and_fail; 1009 } 1010 } 1011 1012 z.avail_in -= 4; 1013 z.next_in += 4; 1014 1015 if (error < 0) { 1016 maybe_warnx("decompression error"); 1017 goto stop_and_fail; 1018 } 1019 state = GZSTATE_MAGIC0; 1020 break; 1021 } 1022 continue; 1023 stop_and_fail: 1024 out_tot = -1; 1025 stop: 1026 break; 1027 } 1028 if (state > GZSTATE_INIT) 1029 inflateEnd(&z); 1030 1031 free(inbufp); 1032 out1: 1033 free(outbufp); 1034 out2: 1035 if (gsizep) 1036 *gsizep = in_tot; 1037 return (out_tot); 1038 } 1039 1040 #ifndef SMALL 1041 /* 1042 * set the owner, mode, flags & utimes using the given file descriptor. 1043 * file is only used in possible warning messages. 1044 */ 1045 static void 1046 copymodes(int fd, const struct stat *sbp, const char *file) 1047 { 1048 struct timeval times[2]; 1049 struct stat sb; 1050 1051 /* 1052 * If we have no info on the input, give this file some 1053 * default values and return.. 1054 */ 1055 if (sbp == NULL) { 1056 mode_t mask = umask(022); 1057 1058 (void)fchmod(fd, DEFFILEMODE & ~mask); 1059 (void)umask(mask); 1060 return; 1061 } 1062 sb = *sbp; 1063 1064 /* if the chown fails, remove set-id bits as-per compress(1) */ 1065 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1066 if (errno != EPERM) 1067 maybe_warn("couldn't fchown: %s", file); 1068 sb.st_mode &= ~(S_ISUID|S_ISGID); 1069 } 1070 1071 /* we only allow set-id and the 9 normal permission bits */ 1072 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1073 if (fchmod(fd, sb.st_mode) < 0) 1074 maybe_warn("couldn't fchmod: %s", file); 1075 1076 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1077 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1078 if (futimes(fd, times) < 0) 1079 maybe_warn("couldn't utimes: %s", file); 1080 1081 /* only try flags if they exist already */ 1082 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1083 maybe_warn("couldn't fchflags: %s", file); 1084 } 1085 #endif 1086 1087 /* what sort of file is this? */ 1088 static enum filetype 1089 file_gettype(u_char *buf) 1090 { 1091 1092 if (buf[0] == GZIP_MAGIC0 && 1093 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1094 return FT_GZIP; 1095 else 1096 #ifndef NO_BZIP2_SUPPORT 1097 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1098 buf[3] >= '0' && buf[3] <= '9') 1099 return FT_BZIP2; 1100 else 1101 #endif 1102 #ifndef NO_COMPRESS_SUPPORT 1103 if (memcmp(buf, Z_MAGIC, 2) == 0) 1104 return FT_Z; 1105 else 1106 #endif 1107 return FT_UNKNOWN; 1108 } 1109 1110 #ifndef SMALL 1111 /* check the outfile is OK. */ 1112 static int 1113 check_outfile(const char *outfile) 1114 { 1115 struct stat sb; 1116 int ok = 1; 1117 1118 if (lflag == 0 && stat(outfile, &sb) == 0) { 1119 if (fflag) 1120 unlink(outfile); 1121 else if (isatty(STDIN_FILENO)) { 1122 char ans[10] = { 'n', '\0' }; /* default */ 1123 1124 fprintf(stderr, "%s already exists -- do you wish to " 1125 "overwrite (y or n)? " , outfile); 1126 (void)fgets(ans, sizeof(ans) - 1, stdin); 1127 if (ans[0] != 'y' && ans[0] != 'Y') { 1128 fprintf(stderr, "\tnot overwriting\n"); 1129 ok = 0; 1130 } else 1131 unlink(outfile); 1132 } else { 1133 maybe_warnx("%s already exists -- skipping", outfile); 1134 ok = 0; 1135 } 1136 } 1137 return ok; 1138 } 1139 1140 static void 1141 unlink_input(const char *file, const struct stat *sb) 1142 { 1143 struct stat nsb; 1144 1145 if (kflag) 1146 return; 1147 if (stat(file, &nsb) != 0) 1148 /* Must be gone alrady */ 1149 return; 1150 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1151 /* Definitely a different file */ 1152 return; 1153 unlink(file); 1154 } 1155 #endif 1156 1157 static const suffixes_t * 1158 check_suffix(char *file, int xlate) 1159 { 1160 const suffixes_t *s; 1161 int len = strlen(file); 1162 char *sp; 1163 1164 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1165 /* if it doesn't fit in "a.suf", don't bother */ 1166 if (s->ziplen >= len) 1167 continue; 1168 sp = file + len - s->ziplen; 1169 if (strcmp(s->zipped, sp) != 0) 1170 continue; 1171 if (xlate) 1172 strcpy(sp, s->normal); 1173 return s; 1174 } 1175 return NULL; 1176 } 1177 1178 /* 1179 * compress the given file: create a corresponding .gz file and remove the 1180 * original. 1181 */ 1182 static off_t 1183 file_compress(char *file, char *outfile, size_t outsize) 1184 { 1185 int in; 1186 int out; 1187 off_t size, insize; 1188 #ifndef SMALL 1189 struct stat isb, osb; 1190 const suffixes_t *suff; 1191 #endif 1192 1193 in = open(file, O_RDONLY); 1194 if (in == -1) { 1195 maybe_warn("can't open %s", file); 1196 return -1; 1197 } 1198 1199 if (cflag == 0) { 1200 #ifndef SMALL 1201 if (fstat(in, &isb) == 0) { 1202 if (isb.st_nlink > 1 && fflag == 0) { 1203 maybe_warnx("%s has %d other link%s -- " 1204 "skipping", file, isb.st_nlink - 1, 1205 isb.st_nlink == 1 ? "" : "s"); 1206 close(in); 1207 return -1; 1208 } 1209 } 1210 1211 if (fflag == 0 && (suff = check_suffix(file, 0)) 1212 && suff->zipped[0] != 0) { 1213 maybe_warnx("%s already has %s suffix -- unchanged", 1214 file, suff->zipped); 1215 close(in); 1216 return -1; 1217 } 1218 #endif 1219 1220 /* Add (usually) .gz to filename */ 1221 if ((size_t)snprintf(outfile, outsize, "%s%s", 1222 file, suffixes[0].zipped) >= outsize) 1223 memcpy(outfile - suffixes[0].ziplen - 1, 1224 suffixes[0].zipped, suffixes[0].ziplen + 1); 1225 1226 #ifndef SMALL 1227 if (check_outfile(outfile) == 0) { 1228 close(in); 1229 return -1; 1230 } 1231 #endif 1232 } 1233 1234 if (cflag == 0) { 1235 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1236 if (out == -1) { 1237 maybe_warn("could not create output: %s", outfile); 1238 fclose(stdin); 1239 return -1; 1240 } 1241 } else 1242 out = STDOUT_FILENO; 1243 1244 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1245 1246 (void)close(in); 1247 1248 /* 1249 * If there was an error, insize will be -1. 1250 * If we compressed to stdout, just return the size. 1251 * Otherwise stat the file and check it is the correct size. 1252 * We only blow away the file if we can stat the output and it 1253 * has the expected size. 1254 */ 1255 if (cflag != 0) 1256 return insize == -1 ? -1 : size; 1257 1258 #ifndef SMALL 1259 if (fstat(out, &osb) != 0) { 1260 maybe_warn("couldn't stat: %s", outfile); 1261 goto bad_outfile; 1262 } 1263 1264 if (osb.st_size != size) { 1265 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1266 " != %" PRIdOFF "), deleting", 1267 outfile, osb.st_size, size); 1268 goto bad_outfile; 1269 } 1270 1271 copymodes(out, &isb, outfile); 1272 #endif 1273 if (close(out) == -1) 1274 maybe_warn("couldn't close output"); 1275 1276 /* output is good, ok to delete input */ 1277 unlink_input(file, &isb); 1278 return size; 1279 1280 #ifndef SMALL 1281 bad_outfile: 1282 if (close(out) == -1) 1283 maybe_warn("couldn't close output"); 1284 1285 maybe_warnx("leaving original %s", file); 1286 unlink(outfile); 1287 return size; 1288 #endif 1289 } 1290 1291 /* uncompress the given file and remove the original */ 1292 static off_t 1293 file_uncompress(char *file, char *outfile, size_t outsize) 1294 { 1295 struct stat isb, osb; 1296 off_t size; 1297 ssize_t rbytes; 1298 unsigned char header1[4]; 1299 enum filetype method; 1300 int rv, fd, ofd, zfd = -1; 1301 #ifndef SMALL 1302 time_t timestamp = 0; 1303 unsigned char name[PATH_MAX + 1]; 1304 #endif 1305 1306 /* gather the old name info */ 1307 1308 fd = open(file, O_RDONLY); 1309 if (fd < 0) { 1310 maybe_warn("can't open %s", file); 1311 goto lose; 1312 } 1313 1314 strlcpy(outfile, file, outsize); 1315 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1316 maybe_warnx("%s: unknown suffix -- ignored", file); 1317 goto lose; 1318 } 1319 1320 rbytes = read(fd, header1, sizeof header1); 1321 if (rbytes != sizeof header1) { 1322 /* we don't want to fail here. */ 1323 #ifndef SMALL 1324 if (fflag) 1325 goto lose; 1326 #endif 1327 if (rbytes == -1) 1328 maybe_warn("can't read %s", file); 1329 else 1330 goto unexpected_EOF; 1331 goto lose; 1332 } 1333 1334 method = file_gettype(header1); 1335 1336 #ifndef SMALL 1337 if (fflag == 0 && method == FT_UNKNOWN) { 1338 maybe_warnx("%s: not in gzip format", file); 1339 goto lose; 1340 } 1341 1342 #endif 1343 1344 #ifndef SMALL 1345 if (method == FT_GZIP && Nflag) { 1346 unsigned char ts[4]; /* timestamp */ 1347 1348 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1349 if (rv >= 0 && (size_t)rv < sizeof ts) 1350 goto unexpected_EOF; 1351 if (rv == -1) { 1352 if (!fflag) 1353 maybe_warn("can't read %s", file); 1354 goto lose; 1355 } 1356 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1357 1358 if (header1[3] & ORIG_NAME) { 1359 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1360 if (rbytes < 0) { 1361 maybe_warn("can't read %s", file); 1362 goto lose; 1363 } 1364 if (name[0] != 0) { 1365 /* preserve original directory name */ 1366 char *dp = strrchr(file, '/'); 1367 if (dp == NULL) 1368 dp = file; 1369 else 1370 dp++; 1371 snprintf(outfile, outsize, "%.*s%.*s", 1372 (int) (dp - file), 1373 file, (int) rbytes, name); 1374 } 1375 } 1376 } 1377 #endif 1378 lseek(fd, 0, SEEK_SET); 1379 1380 if (cflag == 0 || lflag) { 1381 if (fstat(fd, &isb) != 0) 1382 goto lose; 1383 #ifndef SMALL 1384 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1385 maybe_warnx("%s has %d other links -- skipping", 1386 file, isb.st_nlink - 1); 1387 goto lose; 1388 } 1389 if (nflag == 0 && timestamp) 1390 isb.st_mtime = timestamp; 1391 if (check_outfile(outfile) == 0) 1392 goto lose; 1393 #endif 1394 } 1395 1396 if (cflag == 0 && lflag == 0) { 1397 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1398 if (zfd == STDOUT_FILENO) { 1399 /* We won't close STDOUT_FILENO later... */ 1400 zfd = dup(zfd); 1401 close(STDOUT_FILENO); 1402 } 1403 if (zfd == -1) { 1404 maybe_warn("can't open %s", outfile); 1405 goto lose; 1406 } 1407 } else 1408 zfd = STDOUT_FILENO; 1409 1410 #ifndef NO_BZIP2_SUPPORT 1411 if (method == FT_BZIP2) { 1412 1413 /* XXX */ 1414 if (lflag) { 1415 maybe_warnx("no -l with bzip2 files"); 1416 goto lose; 1417 } 1418 1419 size = unbzip2(fd, zfd, NULL, 0, NULL); 1420 } else 1421 #endif 1422 1423 #ifndef NO_COMPRESS_SUPPORT 1424 if (method == FT_Z) { 1425 FILE *in, *out; 1426 1427 /* XXX */ 1428 if (lflag) { 1429 maybe_warnx("no -l with Lempel-Ziv files"); 1430 goto lose; 1431 } 1432 1433 if ((in = zdopen(fd)) == NULL) { 1434 maybe_warn("zdopen for read: %s", file); 1435 goto lose; 1436 } 1437 1438 out = fdopen(dup(zfd), "w"); 1439 if (out == NULL) { 1440 maybe_warn("fdopen for write: %s", outfile); 1441 fclose(in); 1442 goto lose; 1443 } 1444 1445 size = zuncompress(in, out, NULL, 0, NULL); 1446 /* need to fclose() if ferror() is true... */ 1447 if (ferror(in) | fclose(in)) { 1448 maybe_warn("failed infile fclose"); 1449 unlink(outfile); 1450 (void)fclose(out); 1451 } 1452 if (fclose(out) != 0) { 1453 maybe_warn("failed outfile fclose"); 1454 unlink(outfile); 1455 goto lose; 1456 } 1457 } else 1458 #endif 1459 1460 #ifndef SMALL 1461 if (method == FT_UNKNOWN) { 1462 if (lflag) { 1463 maybe_warnx("no -l for unknown filetypes"); 1464 goto lose; 1465 } 1466 size = cat_fd(NULL, 0, NULL, fd); 1467 } else 1468 #endif 1469 { 1470 if (lflag) { 1471 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1472 close(fd); 1473 return -1; /* XXX */ 1474 } 1475 1476 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1477 } 1478 1479 if (close(fd) != 0) 1480 maybe_warn("couldn't close input"); 1481 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1482 maybe_warn("couldn't close output"); 1483 1484 if (size == -1) { 1485 if (cflag == 0) 1486 unlink(outfile); 1487 maybe_warnx("%s: uncompress failed", file); 1488 return -1; 1489 } 1490 1491 /* if testing, or we uncompressed to stdout, this is all we need */ 1492 #ifndef SMALL 1493 if (tflag) 1494 return size; 1495 #endif 1496 /* if we are uncompressing to stdin, don't remove the file. */ 1497 if (cflag) 1498 return size; 1499 1500 /* 1501 * if we create a file... 1502 */ 1503 /* 1504 * if we can't stat the file don't remove the file. 1505 */ 1506 1507 ofd = open(outfile, O_RDWR, 0); 1508 if (ofd == -1) { 1509 maybe_warn("couldn't open (leaving original): %s", 1510 outfile); 1511 return -1; 1512 } 1513 if (fstat(ofd, &osb) != 0) { 1514 maybe_warn("couldn't stat (leaving original): %s", 1515 outfile); 1516 close(ofd); 1517 return -1; 1518 } 1519 if (osb.st_size != size) { 1520 maybe_warnx("stat gave different size: %" PRIdOFF 1521 " != %" PRIdOFF " (leaving original)", 1522 size, osb.st_size); 1523 close(ofd); 1524 unlink(outfile); 1525 return -1; 1526 } 1527 unlink_input(file, &isb); 1528 #ifndef SMALL 1529 copymodes(ofd, &isb, outfile); 1530 #endif 1531 close(ofd); 1532 return size; 1533 1534 unexpected_EOF: 1535 maybe_warnx("%s: unexpected end of file", file); 1536 lose: 1537 if (fd != -1) 1538 close(fd); 1539 if (zfd != -1 && zfd != STDOUT_FILENO) 1540 close(fd); 1541 return -1; 1542 } 1543 1544 #ifndef SMALL 1545 static off_t 1546 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1547 { 1548 char buf[BUFLEN]; 1549 off_t in_tot; 1550 ssize_t w; 1551 1552 in_tot = count; 1553 w = write(STDOUT_FILENO, prepend, count); 1554 if (w == -1 || (size_t)w != count) { 1555 maybe_warn("write to stdout"); 1556 return -1; 1557 } 1558 for (;;) { 1559 ssize_t rv; 1560 1561 rv = read(fd, buf, sizeof buf); 1562 if (rv == 0) 1563 break; 1564 if (rv < 0) { 1565 maybe_warn("read from fd %d", fd); 1566 break; 1567 } 1568 1569 if (write(STDOUT_FILENO, buf, rv) != rv) { 1570 maybe_warn("write to stdout"); 1571 break; 1572 } 1573 in_tot += rv; 1574 } 1575 1576 if (gsizep) 1577 *gsizep = in_tot; 1578 return (in_tot); 1579 } 1580 #endif 1581 1582 static void 1583 handle_stdin(void) 1584 { 1585 unsigned char header1[4]; 1586 off_t usize, gsize; 1587 enum filetype method; 1588 ssize_t bytes_read; 1589 #ifndef NO_COMPRESS_SUPPORT 1590 FILE *in; 1591 #endif 1592 1593 #ifndef SMALL 1594 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1595 maybe_warnx("standard input is a terminal -- ignoring"); 1596 return; 1597 } 1598 #endif 1599 1600 if (lflag) { 1601 struct stat isb; 1602 1603 /* XXX could read the whole file, etc. */ 1604 if (fstat(STDIN_FILENO, &isb) < 0) { 1605 maybe_warn("fstat"); 1606 return; 1607 } 1608 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1609 return; 1610 } 1611 1612 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1613 if (bytes_read == -1) { 1614 maybe_warn("can't read stdin"); 1615 return; 1616 } else if (bytes_read != sizeof(header1)) { 1617 maybe_warnx("(stdin): unexpected end of file"); 1618 return; 1619 } 1620 1621 method = file_gettype(header1); 1622 switch (method) { 1623 default: 1624 #ifndef SMALL 1625 if (fflag == 0) { 1626 maybe_warnx("unknown compression format"); 1627 return; 1628 } 1629 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1630 break; 1631 #endif 1632 case FT_GZIP: 1633 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1634 (char *)header1, sizeof header1, &gsize, "(stdin)"); 1635 break; 1636 #ifndef NO_BZIP2_SUPPORT 1637 case FT_BZIP2: 1638 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1639 (char *)header1, sizeof header1, &gsize); 1640 break; 1641 #endif 1642 #ifndef NO_COMPRESS_SUPPORT 1643 case FT_Z: 1644 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1645 maybe_warnx("zopen of stdin"); 1646 return; 1647 } 1648 1649 usize = zuncompress(in, stdout, (char *)header1, sizeof header1, &gsize); 1650 fclose(in); 1651 break; 1652 #endif 1653 } 1654 1655 #ifndef SMALL 1656 if (vflag && !tflag && usize != -1 && gsize != -1) 1657 print_verbage(NULL, NULL, usize, gsize); 1658 if (vflag && tflag) 1659 print_test("(stdin)", usize != -1); 1660 #endif 1661 1662 } 1663 1664 static void 1665 handle_stdout(void) 1666 { 1667 off_t gsize, usize; 1668 struct stat sb; 1669 time_t systime; 1670 uint32_t mtime; 1671 int ret; 1672 1673 #ifndef SMALL 1674 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1675 maybe_warnx("standard output is a terminal -- ignoring"); 1676 return; 1677 } 1678 #endif 1679 /* If stdin is a file use it's mtime, otherwise use current time */ 1680 ret = fstat(STDIN_FILENO, &sb); 1681 1682 #ifndef SMALL 1683 if (ret < 0) { 1684 maybe_warn("Can't stat stdin"); 1685 return; 1686 } 1687 #endif 1688 1689 if (S_ISREG(sb.st_mode)) 1690 mtime = (uint32_t)sb.st_mtime; 1691 else { 1692 systime = time(NULL); 1693 #ifndef SMALL 1694 if (systime == -1) { 1695 maybe_warn("time"); 1696 return; 1697 } 1698 #endif 1699 mtime = (uint32_t)systime; 1700 } 1701 1702 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1703 #ifndef SMALL 1704 if (vflag && !tflag && usize != -1 && gsize != -1) 1705 print_verbage(NULL, NULL, usize, gsize); 1706 #endif 1707 } 1708 1709 /* do what is asked for, for the path name */ 1710 static void 1711 handle_pathname(char *path) 1712 { 1713 char *opath = path, *s = NULL; 1714 ssize_t len; 1715 int slen; 1716 struct stat sb; 1717 1718 /* check for stdout/stdin */ 1719 if (path[0] == '-' && path[1] == '\0') { 1720 if (dflag) 1721 handle_stdin(); 1722 else 1723 handle_stdout(); 1724 return; 1725 } 1726 1727 retry: 1728 if (stat(path, &sb) != 0) { 1729 /* lets try <path>.gz if we're decompressing */ 1730 if (dflag && s == NULL && errno == ENOENT) { 1731 len = strlen(path); 1732 slen = suffixes[0].ziplen; 1733 s = malloc(len + slen + 1); 1734 if (s == NULL) 1735 maybe_err("malloc"); 1736 memcpy(s, path, len); 1737 memcpy(s + len, suffixes[0].zipped, slen + 1); 1738 path = s; 1739 goto retry; 1740 } 1741 maybe_warn("can't stat: %s", opath); 1742 goto out; 1743 } 1744 1745 if (S_ISDIR(sb.st_mode)) { 1746 #ifndef SMALL 1747 if (rflag) 1748 handle_dir(path); 1749 else 1750 #endif 1751 maybe_warnx("%s is a directory", path); 1752 goto out; 1753 } 1754 1755 if (S_ISREG(sb.st_mode)) 1756 handle_file(path, &sb); 1757 else 1758 maybe_warnx("%s is not a regular file", path); 1759 1760 out: 1761 if (s) 1762 free(s); 1763 } 1764 1765 /* compress/decompress a file */ 1766 static void 1767 handle_file(char *file, struct stat *sbp) 1768 { 1769 off_t usize, gsize; 1770 char outfile[PATH_MAX]; 1771 1772 infile = file; 1773 if (dflag) { 1774 usize = file_uncompress(file, outfile, sizeof(outfile)); 1775 #ifndef SMALL 1776 if (vflag && tflag) 1777 print_test(file, usize != -1); 1778 #endif 1779 if (usize == -1) 1780 return; 1781 gsize = sbp->st_size; 1782 } else { 1783 gsize = file_compress(file, outfile, sizeof(outfile)); 1784 if (gsize == -1) 1785 return; 1786 usize = sbp->st_size; 1787 } 1788 1789 1790 #ifndef SMALL 1791 if (vflag && !tflag) 1792 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1793 #endif 1794 } 1795 1796 #ifndef SMALL 1797 /* this is used with -r to recursively descend directories */ 1798 static void 1799 handle_dir(char *dir) 1800 { 1801 char *path_argv[2]; 1802 FTS *fts; 1803 FTSENT *entry; 1804 1805 path_argv[0] = dir; 1806 path_argv[1] = 0; 1807 fts = fts_open(path_argv, FTS_PHYSICAL | FTS_NOCHDIR, NULL); 1808 if (fts == NULL) { 1809 warn("couldn't fts_open %s", dir); 1810 return; 1811 } 1812 1813 while ((entry = fts_read(fts))) { 1814 switch(entry->fts_info) { 1815 case FTS_D: 1816 case FTS_DP: 1817 continue; 1818 1819 case FTS_DNR: 1820 case FTS_ERR: 1821 case FTS_NS: 1822 maybe_warn("%s", entry->fts_path); 1823 continue; 1824 case FTS_F: 1825 handle_file(entry->fts_path, entry->fts_statp); 1826 } 1827 } 1828 (void)fts_close(fts); 1829 } 1830 #endif 1831 1832 /* print a ratio - size reduction as a fraction of uncompressed size */ 1833 static void 1834 print_ratio(off_t in, off_t out, FILE *where) 1835 { 1836 int percent10; /* 10 * percent */ 1837 off_t diff; 1838 char buff[8]; 1839 int len; 1840 1841 diff = in - out/2; 1842 if (diff <= 0) 1843 /* 1844 * Output is more than double size of input! print -99.9% 1845 * Quite possibly we've failed to get the original size. 1846 */ 1847 percent10 = -999; 1848 else { 1849 /* 1850 * We only need 12 bits of result from the final division, 1851 * so reduce the values until a 32bit division will suffice. 1852 */ 1853 while (in > 0x100000) { 1854 diff >>= 1; 1855 in >>= 1; 1856 } 1857 if (in != 0) 1858 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1859 else 1860 percent10 = 0; 1861 } 1862 1863 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1864 /* Move the '.' to before the last digit */ 1865 buff[len - 1] = buff[len - 2]; 1866 buff[len - 2] = '.'; 1867 fprintf(where, "%5s%%", buff); 1868 } 1869 1870 #ifndef SMALL 1871 /* print compression statistics, and the new name (if there is one!) */ 1872 static void 1873 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1874 { 1875 if (file) 1876 fprintf(stderr, "%s:%s ", file, 1877 strlen(file) < 7 ? "\t\t" : "\t"); 1878 print_ratio(usize, gsize, stderr); 1879 if (nfile) 1880 fprintf(stderr, " -- replaced with %s", nfile); 1881 fprintf(stderr, "\n"); 1882 fflush(stderr); 1883 } 1884 1885 /* print test results */ 1886 static void 1887 print_test(const char *file, int ok) 1888 { 1889 1890 if (exit_value == 0 && ok == 0) 1891 exit_value = 1; 1892 fprintf(stderr, "%s:%s %s\n", file, 1893 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1894 fflush(stderr); 1895 } 1896 #endif 1897 1898 /* print a file's info ala --list */ 1899 /* eg: 1900 compressed uncompressed ratio uncompressed_name 1901 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1902 */ 1903 static void 1904 print_list(int fd, off_t out, const char *outfile, time_t ts) 1905 { 1906 static int first = 1; 1907 #ifndef SMALL 1908 static off_t in_tot, out_tot; 1909 uint32_t crc = 0; 1910 #endif 1911 off_t in = 0, rv; 1912 1913 if (first) { 1914 #ifndef SMALL 1915 if (vflag) 1916 printf("method crc date time "); 1917 #endif 1918 if (qflag == 0) 1919 printf(" compressed uncompressed " 1920 "ratio uncompressed_name\n"); 1921 } 1922 first = 0; 1923 1924 /* print totals? */ 1925 #ifndef SMALL 1926 if (fd == -1) { 1927 in = in_tot; 1928 out = out_tot; 1929 } else 1930 #endif 1931 { 1932 /* read the last 4 bytes - this is the uncompressed size */ 1933 rv = lseek(fd, (off_t)(-8), SEEK_END); 1934 if (rv != -1) { 1935 unsigned char buf[8]; 1936 uint32_t usize; 1937 1938 rv = read(fd, (char *)buf, sizeof(buf)); 1939 if (rv == -1) 1940 maybe_warn("read of uncompressed size"); 1941 else if (rv != sizeof(buf)) 1942 maybe_warnx("read of uncompressed size"); 1943 1944 else { 1945 usize = buf[4] | buf[5] << 8 | 1946 buf[6] << 16 | buf[7] << 24; 1947 in = (off_t)usize; 1948 #ifndef SMALL 1949 crc = buf[0] | buf[1] << 8 | 1950 buf[2] << 16 | buf[3] << 24; 1951 #endif 1952 } 1953 } 1954 } 1955 1956 #ifndef SMALL 1957 if (vflag && fd == -1) 1958 printf(" "); 1959 else if (vflag) { 1960 char *date = ctime(&ts); 1961 1962 /* skip the day, 1/100th second, and year */ 1963 date += 4; 1964 date[12] = 0; 1965 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1966 } 1967 in_tot += in; 1968 out_tot += out; 1969 #endif 1970 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1971 print_ratio(in, out, stdout); 1972 printf(" %s\n", outfile); 1973 } 1974 1975 /* display the usage of NetBSD gzip */ 1976 static void 1977 usage(void) 1978 { 1979 1980 fprintf(stderr, "%s\n", gzip_version); 1981 fprintf(stderr, 1982 #ifdef SMALL 1983 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n", 1984 #else 1985 "usage: %s [-123456789acdfhklLNnqrtVv] [-S .suffix] [<file> [<file> ...]]\n" 1986 " -1 --fast fastest (worst) compression\n" 1987 " -2 .. -8 set compression level\n" 1988 " -9 --best best (slowest) compression\n" 1989 " -c --stdout write to stdout, keep original files\n" 1990 " --to-stdout\n" 1991 " -d --decompress uncompress files\n" 1992 " --uncompress\n" 1993 " -f --force force overwriting & compress links\n" 1994 " -h --help display this help\n" 1995 " -k --keep don't delete input files during operation\n" 1996 " -l --list list compressed file contents\n" 1997 " -N --name save or restore original file name and time stamp\n" 1998 " -n --no-name don't save original file name or time stamp\n" 1999 " -q --quiet output no warnings\n" 2000 " -r --recursive recursively compress files in directories\n" 2001 " -S .suf use suffix .suf instead of .gz\n" 2002 " --suffix .suf\n" 2003 " -t --test test compressed file\n" 2004 " -V --version display program version\n" 2005 " -v --verbose print extra statistics\n", 2006 #endif 2007 getprogname()); 2008 exit(0); 2009 } 2010 2011 #ifndef SMALL 2012 /* display the license information of FreeBSD gzip */ 2013 static void 2014 display_license(void) 2015 { 2016 2017 fprintf(stderr, "%s (based on NetBSD gzip 20060927)\n", gzip_version); 2018 fprintf(stderr, "%s\n", gzip_copyright); 2019 exit(0); 2020 } 2021 #endif 2022 2023 /* display the version of NetBSD gzip */ 2024 static void 2025 display_version(void) 2026 { 2027 2028 fprintf(stderr, "%s\n", gzip_version); 2029 exit(0); 2030 } 2031 2032 #ifndef NO_BZIP2_SUPPORT 2033 #include "unbzip2.c" 2034 #endif 2035 #ifndef NO_COMPRESS_SUPPORT 2036 #include "zuncompress.c" 2037 #endif 2038 2039 static ssize_t 2040 read_retry(int fd, void *buf, size_t sz) 2041 { 2042 char *cp = buf; 2043 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2044 2045 while (left > 0) { 2046 ssize_t ret; 2047 2048 ret = read(fd, cp, left); 2049 if (ret == -1) { 2050 return ret; 2051 } else if (ret == 0) { 2052 break; /* EOF */ 2053 } 2054 cp += ret; 2055 left -= ret; 2056 } 2057 2058 return sz - left; 2059 } 2060 2061