1 /* $NetBSD: gzip.c,v 1.89 2006/11/13 21:57:59 mrg Exp $ */ 2 3 /*- 4 * Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. The name of the author may not be used to endorse or promote products 16 * derived from this software without specific prior written permission. 17 * 18 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 19 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 20 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 21 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 22 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 23 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 24 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 25 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 26 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 28 * SUCH DAMAGE. 29 * 30 */ 31 32 #include <sys/cdefs.h> 33 #ifndef lint 34 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green\n\ 35 All rights reserved.\n"); 36 __RCSID("$FreeBSD$"); 37 #endif /* not lint */ 38 39 /* 40 * gzip.c -- GPL free gzip using zlib. 41 * 42 * RFC 1950 covers the zlib format 43 * RFC 1951 covers the deflate format 44 * RFC 1952 covers the gzip format 45 * 46 * TODO: 47 * - use mmap where possible 48 * - handle some signals better (remove outfile?) 49 * - make bzip2/compress -v/-t/-l support work as well as possible 50 */ 51 52 #include <sys/param.h> 53 #include <sys/stat.h> 54 #include <sys/time.h> 55 56 #include <inttypes.h> 57 #include <unistd.h> 58 #include <stdio.h> 59 #include <string.h> 60 #include <stdlib.h> 61 #include <err.h> 62 #include <errno.h> 63 #include <fcntl.h> 64 #include <zlib.h> 65 #include <fts.h> 66 #include <libgen.h> 67 #include <stdarg.h> 68 #include <getopt.h> 69 #include <time.h> 70 71 #ifndef PRIdOFF 72 #define PRIdOFF PRId64 73 #endif 74 75 /* what type of file are we dealing with */ 76 enum filetype { 77 FT_GZIP, 78 #ifndef NO_BZIP2_SUPPORT 79 FT_BZIP2, 80 #endif 81 #ifndef NO_COMPRESS_SUPPORT 82 FT_Z, 83 #endif 84 FT_LAST, 85 FT_UNKNOWN 86 }; 87 88 #ifndef NO_BZIP2_SUPPORT 89 #include <bzlib.h> 90 91 #define BZ2_SUFFIX ".bz2" 92 #define BZIP2_MAGIC "\102\132\150" 93 #endif 94 95 #ifndef NO_COMPRESS_SUPPORT 96 #define Z_SUFFIX ".Z" 97 #define Z_MAGIC "\037\235" 98 #endif 99 100 #define GZ_SUFFIX ".gz" 101 102 #define BUFLEN (64 * 1024) 103 104 #define GZIP_MAGIC0 0x1F 105 #define GZIP_MAGIC1 0x8B 106 #define GZIP_OMAGIC1 0x9E 107 108 #define GZIP_TIMESTAMP (off_t)4 109 #define GZIP_ORIGNAME (off_t)10 110 111 #define HEAD_CRC 0x02 112 #define EXTRA_FIELD 0x04 113 #define ORIG_NAME 0x08 114 #define COMMENT 0x10 115 116 #define OS_CODE 3 /* Unix */ 117 118 typedef struct { 119 const char *zipped; 120 int ziplen; 121 const char *normal; /* for unzip - must not be longer than zipped */ 122 } suffixes_t; 123 static suffixes_t suffixes[] = { 124 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 125 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 126 #ifndef SMALL 127 SUFFIX(GZ_SUFFIX, ""), 128 SUFFIX(".z", ""), 129 SUFFIX("-gz", ""), 130 SUFFIX("-z", ""), 131 SUFFIX("_z", ""), 132 SUFFIX(".taz", ".tar"), 133 SUFFIX(".tgz", ".tar"), 134 #ifndef NO_BZIP2_SUPPORT 135 SUFFIX(BZ2_SUFFIX, ""), 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 " 3. The name of the author may not be used to endorse or promote products\n" 162 " derived from this software without specific prior written permission.\n" 163 "\n" 164 " THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR\n" 165 " IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES\n" 166 " OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.\n" 167 " IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,\n" 168 " INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,\n" 169 " BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;\n" 170 " LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED\n" 171 " AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,\n" 172 " OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY\n" 173 " OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF\n" 174 " SUCH DAMAGE."; 175 #endif 176 177 static int cflag; /* stdout mode */ 178 static int dflag; /* decompress mode */ 179 static int lflag; /* list mode */ 180 static int numflag = 6; /* gzip -1..-9 value */ 181 182 #ifndef SMALL 183 static int fflag; /* force mode */ 184 static int kflag; /* don't delete input files */ 185 static int nflag; /* don't save name/timestamp */ 186 static int Nflag; /* don't restore name/timestamp */ 187 static int qflag; /* quiet mode */ 188 static int rflag; /* recursive mode */ 189 static int tflag; /* test */ 190 static int vflag; /* verbose mode */ 191 #else 192 #define qflag 0 193 #define tflag 0 194 #endif 195 196 static int exit_value = 0; /* exit value */ 197 198 static char *infile; /* name of file coming in */ 199 200 static void maybe_err(const char *fmt, ...) 201 __attribute__((__format__(__printf__, 1, 2))); 202 #ifndef NO_BZIP2_SUPPORT 203 static void maybe_errx(const char *fmt, ...) 204 __attribute__((__format__(__printf__, 1, 2))); 205 #endif 206 static void maybe_warn(const char *fmt, ...) 207 __attribute__((__format__(__printf__, 1, 2))); 208 static void maybe_warnx(const char *fmt, ...) 209 __attribute__((__format__(__printf__, 1, 2))); 210 static enum filetype file_gettype(u_char *); 211 #ifdef SMALL 212 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 213 #endif 214 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 215 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 216 static off_t file_compress(char *, char *, size_t); 217 static off_t file_uncompress(char *, char *, size_t); 218 static void handle_pathname(char *); 219 static void handle_file(char *, struct stat *); 220 static void handle_stdin(void); 221 static void handle_stdout(void); 222 static void print_ratio(off_t, off_t, FILE *); 223 static void print_list(int fd, off_t, const char *, time_t); 224 static void usage(void); 225 static void display_version(void); 226 #ifndef SMALL 227 static void display_license(void); 228 #endif 229 static const suffixes_t *check_suffix(char *, int); 230 static ssize_t read_retry(int, void *, size_t); 231 232 #ifdef SMALL 233 #define unlink_input(f, sb) unlink(f) 234 #else 235 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 236 static void prepend_gzip(char *, int *, char ***); 237 static void handle_dir(char *); 238 static void print_verbage(const char *, const char *, off_t, off_t); 239 static void print_test(const char *, int); 240 static void copymodes(int fd, const struct stat *, const char *file); 241 static int check_outfile(const char *outfile); 242 #endif 243 244 #ifndef NO_BZIP2_SUPPORT 245 static off_t unbzip2(int, int, char *, size_t, off_t *); 246 #endif 247 248 #ifndef NO_COMPRESS_SUPPORT 249 static FILE *zdopen(int); 250 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 251 #endif 252 253 int main(int, char **p); 254 255 #ifdef SMALL 256 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 257 #else 258 static const struct option longopts[] = { 259 { "stdout", no_argument, 0, 'c' }, 260 { "to-stdout", no_argument, 0, 'c' }, 261 { "decompress", no_argument, 0, 'd' }, 262 { "uncompress", no_argument, 0, 'd' }, 263 { "force", no_argument, 0, 'f' }, 264 { "help", no_argument, 0, 'h' }, 265 { "keep", no_argument, 0, 'k' }, 266 { "list", no_argument, 0, 'l' }, 267 { "no-name", no_argument, 0, 'n' }, 268 { "name", no_argument, 0, 'N' }, 269 { "quiet", no_argument, 0, 'q' }, 270 { "recursive", no_argument, 0, 'r' }, 271 { "suffix", required_argument, 0, 'S' }, 272 { "test", no_argument, 0, 't' }, 273 { "verbose", no_argument, 0, 'v' }, 274 { "version", no_argument, 0, 'V' }, 275 { "fast", no_argument, 0, '1' }, 276 { "best", no_argument, 0, '9' }, 277 { "ascii", no_argument, 0, 'a' }, 278 { "license", no_argument, 0, 'L' }, 279 { NULL, no_argument, 0, 0 }, 280 }; 281 #endif 282 283 int 284 main(int argc, char **argv) 285 { 286 const char *progname = getprogname(); 287 #ifndef SMALL 288 char *gzip; 289 int len; 290 #endif 291 int ch; 292 293 /* XXX set up signals */ 294 295 #ifndef SMALL 296 if ((gzip = getenv("GZIP")) != NULL) 297 prepend_gzip(gzip, &argc, &argv); 298 #endif 299 300 /* 301 * XXX 302 * handle being called `gunzip', `zcat' and `gzcat' 303 */ 304 if (strcmp(progname, "gunzip") == 0) 305 dflag = 1; 306 else if (strcmp(progname, "zcat") == 0 || 307 strcmp(progname, "gzcat") == 0) 308 dflag = cflag = 1; 309 310 #ifdef SMALL 311 #define OPT_LIST "123456789cdhltV" 312 #else 313 #define OPT_LIST "123456789acdfhklLNnqrS:tVv" 314 #endif 315 316 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 317 switch (ch) { 318 case '1': case '2': case '3': 319 case '4': case '5': case '6': 320 case '7': case '8': case '9': 321 numflag = ch - '0'; 322 break; 323 case 'c': 324 cflag = 1; 325 break; 326 case 'd': 327 dflag = 1; 328 break; 329 case 'l': 330 lflag = 1; 331 dflag = 1; 332 break; 333 case 'V': 334 display_version(); 335 /* NOTREACHED */ 336 #ifndef SMALL 337 case 'a': 338 fprintf(stderr, "%s: option --ascii ignored on this system\n", progname); 339 break; 340 case 'f': 341 fflag = 1; 342 break; 343 case 'k': 344 kflag = 1; 345 break; 346 case 'L': 347 display_license(); 348 /* NOT REACHED */ 349 case 'N': 350 nflag = 0; 351 Nflag = 1; 352 break; 353 case 'n': 354 nflag = 1; 355 Nflag = 0; 356 break; 357 case 'q': 358 qflag = 1; 359 break; 360 case 'r': 361 rflag = 1; 362 break; 363 case 'S': 364 len = strlen(optarg); 365 if (len != 0) { 366 suffixes[0].zipped = optarg; 367 suffixes[0].ziplen = len; 368 } else { 369 suffixes[NUM_SUFFIXES - 1].zipped = ""; 370 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 371 } 372 break; 373 case 't': 374 cflag = 1; 375 tflag = 1; 376 dflag = 1; 377 break; 378 case 'v': 379 vflag = 1; 380 break; 381 #endif 382 default: 383 usage(); 384 /* NOTREACHED */ 385 } 386 } 387 argv += optind; 388 argc -= optind; 389 390 if (argc == 0) { 391 if (dflag) /* stdin mode */ 392 handle_stdin(); 393 else /* stdout mode */ 394 handle_stdout(); 395 } else { 396 do { 397 handle_pathname(argv[0]); 398 } while (*++argv); 399 } 400 #ifndef SMALL 401 if (qflag == 0 && lflag && argc > 1) 402 print_list(-1, 0, "(totals)", 0); 403 #endif 404 exit(exit_value); 405 } 406 407 /* maybe print a warning */ 408 void 409 maybe_warn(const char *fmt, ...) 410 { 411 va_list ap; 412 413 if (qflag == 0) { 414 va_start(ap, fmt); 415 vwarn(fmt, ap); 416 va_end(ap); 417 } 418 if (exit_value == 0) 419 exit_value = 1; 420 } 421 422 /* ... without an errno. */ 423 void 424 maybe_warnx(const char *fmt, ...) 425 { 426 va_list ap; 427 428 if (qflag == 0) { 429 va_start(ap, fmt); 430 vwarnx(fmt, ap); 431 va_end(ap); 432 } 433 if (exit_value == 0) 434 exit_value = 1; 435 } 436 437 /* maybe print an error */ 438 void 439 maybe_err(const char *fmt, ...) 440 { 441 va_list ap; 442 443 if (qflag == 0) { 444 va_start(ap, fmt); 445 vwarn(fmt, ap); 446 va_end(ap); 447 } 448 exit(2); 449 } 450 451 #ifndef NO_BZIP2_SUPPORT 452 /* ... without an errno. */ 453 void 454 maybe_errx(const char *fmt, ...) 455 { 456 va_list ap; 457 458 if (qflag == 0) { 459 va_start(ap, fmt); 460 vwarnx(fmt, ap); 461 va_end(ap); 462 } 463 exit(2); 464 } 465 #endif 466 467 #ifndef SMALL 468 /* split up $GZIP and prepend it to the argument list */ 469 static void 470 prepend_gzip(char *gzip, int *argc, char ***argv) 471 { 472 char *s, **nargv, **ac; 473 int nenvarg = 0, i; 474 475 /* scan how many arguments there are */ 476 for (s = gzip;;) { 477 while (*s == ' ' || *s == '\t') 478 s++; 479 if (*s == 0) 480 goto count_done; 481 nenvarg++; 482 while (*s != ' ' && *s != '\t') 483 if (*s++ == 0) 484 goto count_done; 485 } 486 count_done: 487 /* punt early */ 488 if (nenvarg == 0) 489 return; 490 491 *argc += nenvarg; 492 ac = *argv; 493 494 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 495 if (nargv == NULL) 496 maybe_err("malloc"); 497 498 /* stash this away */ 499 *argv = nargv; 500 501 /* copy the program name first */ 502 i = 0; 503 nargv[i++] = *(ac++); 504 505 /* take a copy of $GZIP and add it to the array */ 506 s = strdup(gzip); 507 if (s == NULL) 508 maybe_err("strdup"); 509 for (;;) { 510 /* Skip whitespaces. */ 511 while (*s == ' ' || *s == '\t') 512 s++; 513 if (*s == 0) 514 goto copy_done; 515 nargv[i++] = s; 516 /* Find the end of this argument. */ 517 while (*s != ' ' && *s != '\t') 518 if (*s++ == 0) 519 /* Argument followed by NUL. */ 520 goto copy_done; 521 /* Terminate by overwriting ' ' or '\t' with NUL. */ 522 *s++ = 0; 523 } 524 copy_done: 525 526 /* copy the original arguments and a NULL */ 527 while (*ac) 528 nargv[i++] = *(ac++); 529 nargv[i] = NULL; 530 } 531 #endif 532 533 /* compress input to output. Return bytes read, -1 on error */ 534 static off_t 535 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 536 { 537 z_stream z; 538 char *outbufp, *inbufp; 539 off_t in_tot = 0, out_tot = 0; 540 ssize_t in_size; 541 int i, error; 542 uLong crc; 543 #ifdef SMALL 544 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 545 0, 0, 0, 0, 546 0, OS_CODE }; 547 #endif 548 549 outbufp = malloc(BUFLEN); 550 inbufp = malloc(BUFLEN); 551 if (outbufp == NULL || inbufp == NULL) { 552 maybe_err("malloc failed"); 553 goto out; 554 } 555 556 memset(&z, 0, sizeof z); 557 z.zalloc = Z_NULL; 558 z.zfree = Z_NULL; 559 z.opaque = 0; 560 561 #ifdef SMALL 562 memcpy(outbufp, header, sizeof header); 563 i = sizeof header; 564 #else 565 if (nflag != 0) { 566 mtime = 0; 567 origname = ""; 568 } 569 570 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 571 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 572 *origname ? ORIG_NAME : 0, 573 mtime & 0xff, 574 (mtime >> 8) & 0xff, 575 (mtime >> 16) & 0xff, 576 (mtime >> 24) & 0xff, 577 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 578 OS_CODE, origname); 579 if (i >= BUFLEN) 580 /* this need PATH_MAX > BUFLEN ... */ 581 maybe_err("snprintf"); 582 if (*origname) 583 i++; 584 #endif 585 586 z.next_out = (unsigned char *)outbufp + i; 587 z.avail_out = BUFLEN - i; 588 589 error = deflateInit2(&z, numflag, Z_DEFLATED, 590 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 591 if (error != Z_OK) { 592 maybe_warnx("deflateInit2 failed"); 593 in_tot = -1; 594 goto out; 595 } 596 597 crc = crc32(0L, Z_NULL, 0); 598 for (;;) { 599 if (z.avail_out == 0) { 600 if (write(out, outbufp, BUFLEN) != BUFLEN) { 601 maybe_warn("write"); 602 out_tot = -1; 603 goto out; 604 } 605 606 out_tot += BUFLEN; 607 z.next_out = (unsigned char *)outbufp; 608 z.avail_out = BUFLEN; 609 } 610 611 if (z.avail_in == 0) { 612 in_size = read(in, inbufp, BUFLEN); 613 if (in_size < 0) { 614 maybe_warn("read"); 615 in_tot = -1; 616 goto out; 617 } 618 if (in_size == 0) 619 break; 620 621 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 622 in_tot += in_size; 623 z.next_in = (unsigned char *)inbufp; 624 z.avail_in = in_size; 625 } 626 627 error = deflate(&z, Z_NO_FLUSH); 628 if (error != Z_OK && error != Z_STREAM_END) { 629 maybe_warnx("deflate failed"); 630 in_tot = -1; 631 goto out; 632 } 633 } 634 635 /* clean up */ 636 for (;;) { 637 size_t len; 638 ssize_t w; 639 640 error = deflate(&z, Z_FINISH); 641 if (error != Z_OK && error != Z_STREAM_END) { 642 maybe_warnx("deflate failed"); 643 in_tot = -1; 644 goto out; 645 } 646 647 len = (char *)z.next_out - outbufp; 648 649 w = write(out, outbufp, len); 650 if (w == -1 || (size_t)w != len) { 651 maybe_warn("write"); 652 out_tot = -1; 653 goto out; 654 } 655 out_tot += len; 656 z.next_out = (unsigned char *)outbufp; 657 z.avail_out = BUFLEN; 658 659 if (error == Z_STREAM_END) 660 break; 661 } 662 663 if (deflateEnd(&z) != Z_OK) { 664 maybe_warnx("deflateEnd failed"); 665 in_tot = -1; 666 goto out; 667 } 668 669 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 670 (int)crc & 0xff, 671 (int)(crc >> 8) & 0xff, 672 (int)(crc >> 16) & 0xff, 673 (int)(crc >> 24) & 0xff, 674 (int)in_tot & 0xff, 675 (int)(in_tot >> 8) & 0xff, 676 (int)(in_tot >> 16) & 0xff, 677 (int)(in_tot >> 24) & 0xff); 678 if (i != 8) 679 maybe_err("snprintf"); 680 if (write(out, outbufp, i) != i) { 681 maybe_warn("write"); 682 in_tot = -1; 683 } else 684 out_tot += i; 685 686 out: 687 if (inbufp != NULL) 688 free(inbufp); 689 if (outbufp != NULL) 690 free(outbufp); 691 if (gsizep) 692 *gsizep = out_tot; 693 return in_tot; 694 } 695 696 /* 697 * uncompress input to output then close the input. return the 698 * uncompressed size written, and put the compressed sized read 699 * into `*gsizep'. 700 */ 701 static off_t 702 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 703 const char *filename) 704 { 705 z_stream z; 706 char *outbufp, *inbufp; 707 off_t out_tot = -1, in_tot = 0; 708 uint32_t out_sub_tot = 0; 709 enum { 710 GZSTATE_MAGIC0, 711 GZSTATE_MAGIC1, 712 GZSTATE_METHOD, 713 GZSTATE_FLAGS, 714 GZSTATE_SKIPPING, 715 GZSTATE_EXTRA, 716 GZSTATE_EXTRA2, 717 GZSTATE_EXTRA3, 718 GZSTATE_ORIGNAME, 719 GZSTATE_COMMENT, 720 GZSTATE_HEAD_CRC1, 721 GZSTATE_HEAD_CRC2, 722 GZSTATE_INIT, 723 GZSTATE_READ, 724 GZSTATE_CRC, 725 GZSTATE_LEN, 726 } state = GZSTATE_MAGIC0; 727 int flags = 0, skip_count = 0; 728 int error = Z_STREAM_ERROR, done_reading = 0; 729 uLong crc = 0; 730 ssize_t wr; 731 int needmore = 0; 732 733 #define ADVANCE() { z.next_in++; z.avail_in--; } 734 735 if ((outbufp = malloc(BUFLEN)) == NULL) { 736 maybe_err("malloc failed"); 737 goto out2; 738 } 739 if ((inbufp = malloc(BUFLEN)) == NULL) { 740 maybe_err("malloc failed"); 741 goto out1; 742 } 743 744 memset(&z, 0, sizeof z); 745 z.avail_in = prelen; 746 z.next_in = (unsigned char *)pre; 747 z.avail_out = BUFLEN; 748 z.next_out = (unsigned char *)outbufp; 749 z.zalloc = NULL; 750 z.zfree = NULL; 751 z.opaque = 0; 752 753 in_tot = prelen; 754 out_tot = 0; 755 756 for (;;) { 757 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 758 ssize_t in_size; 759 760 if (z.avail_in > 0) { 761 memmove(inbufp, z.next_in, z.avail_in); 762 } 763 z.next_in = (unsigned char *)inbufp; 764 in_size = read(in, z.next_in + z.avail_in, 765 BUFLEN - z.avail_in); 766 767 if (in_size == -1) { 768 maybe_warn("failed to read stdin"); 769 goto stop_and_fail; 770 } else if (in_size == 0) { 771 done_reading = 1; 772 } 773 774 z.avail_in += in_size; 775 needmore = 0; 776 777 in_tot += in_size; 778 } 779 if (z.avail_in == 0) { 780 if (done_reading && state != GZSTATE_MAGIC0) { 781 maybe_warnx("%s: unexpected end of file", 782 filename); 783 goto stop_and_fail; 784 } 785 goto stop; 786 } 787 switch (state) { 788 case GZSTATE_MAGIC0: 789 if (*z.next_in != GZIP_MAGIC0) { 790 if (in_tot > 0) { 791 maybe_warnx("%s: trailing garbage " 792 "ignored", filename); 793 goto stop; 794 } 795 maybe_warnx("input not gziped (MAGIC0)"); 796 goto stop_and_fail; 797 } 798 ADVANCE(); 799 state++; 800 out_sub_tot = 0; 801 crc = crc32(0L, Z_NULL, 0); 802 break; 803 804 case GZSTATE_MAGIC1: 805 if (*z.next_in != GZIP_MAGIC1 && 806 *z.next_in != GZIP_OMAGIC1) { 807 maybe_warnx("input not gziped (MAGIC1)"); 808 goto stop_and_fail; 809 } 810 ADVANCE(); 811 state++; 812 break; 813 814 case GZSTATE_METHOD: 815 if (*z.next_in != Z_DEFLATED) { 816 maybe_warnx("unknown compression method"); 817 goto stop_and_fail; 818 } 819 ADVANCE(); 820 state++; 821 break; 822 823 case GZSTATE_FLAGS: 824 flags = *z.next_in; 825 ADVANCE(); 826 skip_count = 6; 827 state++; 828 break; 829 830 case GZSTATE_SKIPPING: 831 if (skip_count > 0) { 832 skip_count--; 833 ADVANCE(); 834 } else 835 state++; 836 break; 837 838 case GZSTATE_EXTRA: 839 if ((flags & EXTRA_FIELD) == 0) { 840 state = GZSTATE_ORIGNAME; 841 break; 842 } 843 skip_count = *z.next_in; 844 ADVANCE(); 845 state++; 846 break; 847 848 case GZSTATE_EXTRA2: 849 skip_count |= ((*z.next_in) << 8); 850 ADVANCE(); 851 state++; 852 break; 853 854 case GZSTATE_EXTRA3: 855 if (skip_count > 0) { 856 skip_count--; 857 ADVANCE(); 858 } else 859 state++; 860 break; 861 862 case GZSTATE_ORIGNAME: 863 if ((flags & ORIG_NAME) == 0) { 864 state++; 865 break; 866 } 867 if (*z.next_in == 0) 868 state++; 869 ADVANCE(); 870 break; 871 872 case GZSTATE_COMMENT: 873 if ((flags & COMMENT) == 0) { 874 state++; 875 break; 876 } 877 if (*z.next_in == 0) 878 state++; 879 ADVANCE(); 880 break; 881 882 case GZSTATE_HEAD_CRC1: 883 if (flags & HEAD_CRC) 884 skip_count = 2; 885 else 886 skip_count = 0; 887 state++; 888 break; 889 890 case GZSTATE_HEAD_CRC2: 891 if (skip_count > 0) { 892 skip_count--; 893 ADVANCE(); 894 } else 895 state++; 896 break; 897 898 case GZSTATE_INIT: 899 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 900 maybe_warnx("failed to inflateInit"); 901 goto stop_and_fail; 902 } 903 state++; 904 break; 905 906 case GZSTATE_READ: 907 error = inflate(&z, Z_FINISH); 908 switch (error) { 909 /* Z_BUF_ERROR goes with Z_FINISH... */ 910 case Z_BUF_ERROR: 911 case Z_STREAM_END: 912 case Z_OK: 913 break; 914 915 case Z_NEED_DICT: 916 maybe_warnx("Z_NEED_DICT error"); 917 goto stop_and_fail; 918 case Z_DATA_ERROR: 919 maybe_warnx("data stream error"); 920 goto stop_and_fail; 921 case Z_STREAM_ERROR: 922 maybe_warnx("internal stream error"); 923 goto stop_and_fail; 924 case Z_MEM_ERROR: 925 maybe_warnx("memory allocation error"); 926 goto stop_and_fail; 927 928 default: 929 maybe_warn("unknown error from inflate(): %d", 930 error); 931 } 932 wr = BUFLEN - z.avail_out; 933 934 if (wr != 0) { 935 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 936 if ( 937 #ifndef SMALL 938 /* don't write anything with -t */ 939 tflag == 0 && 940 #endif 941 write(out, outbufp, wr) != wr) { 942 maybe_warn("error writing to output"); 943 goto stop_and_fail; 944 } 945 946 out_tot += wr; 947 out_sub_tot += wr; 948 } 949 950 if (error == Z_STREAM_END) { 951 inflateEnd(&z); 952 state++; 953 } 954 955 z.next_out = (unsigned char *)outbufp; 956 z.avail_out = BUFLEN; 957 958 break; 959 case GZSTATE_CRC: 960 { 961 uLong origcrc; 962 963 if (z.avail_in < 4) { 964 if (!done_reading) { 965 needmore = 1; 966 continue; 967 } 968 maybe_warnx("truncated input"); 969 goto stop_and_fail; 970 } 971 origcrc = ((unsigned)z.next_in[0] & 0xff) | 972 ((unsigned)z.next_in[1] & 0xff) << 8 | 973 ((unsigned)z.next_in[2] & 0xff) << 16 | 974 ((unsigned)z.next_in[3] & 0xff) << 24; 975 if (origcrc != crc) { 976 maybe_warnx("invalid compressed" 977 " data--crc error"); 978 goto stop_and_fail; 979 } 980 } 981 982 z.avail_in -= 4; 983 z.next_in += 4; 984 985 if (!z.avail_in && done_reading) { 986 goto stop; 987 } 988 state++; 989 break; 990 case GZSTATE_LEN: 991 { 992 uLong origlen; 993 994 if (z.avail_in < 4) { 995 if (!done_reading) { 996 needmore = 1; 997 continue; 998 } 999 maybe_warnx("truncated input"); 1000 goto stop_and_fail; 1001 } 1002 origlen = ((unsigned)z.next_in[0] & 0xff) | 1003 ((unsigned)z.next_in[1] & 0xff) << 8 | 1004 ((unsigned)z.next_in[2] & 0xff) << 16 | 1005 ((unsigned)z.next_in[3] & 0xff) << 24; 1006 1007 if (origlen != out_sub_tot) { 1008 maybe_warnx("invalid compressed" 1009 " data--length error"); 1010 goto stop_and_fail; 1011 } 1012 } 1013 1014 z.avail_in -= 4; 1015 z.next_in += 4; 1016 1017 if (error < 0) { 1018 maybe_warnx("decompression error"); 1019 goto stop_and_fail; 1020 } 1021 state = GZSTATE_MAGIC0; 1022 break; 1023 } 1024 continue; 1025 stop_and_fail: 1026 out_tot = -1; 1027 stop: 1028 break; 1029 } 1030 if (state > GZSTATE_INIT) 1031 inflateEnd(&z); 1032 1033 free(inbufp); 1034 out1: 1035 free(outbufp); 1036 out2: 1037 if (gsizep) 1038 *gsizep = in_tot; 1039 return (out_tot); 1040 } 1041 1042 #ifndef SMALL 1043 /* 1044 * set the owner, mode, flags & utimes using the given file descriptor. 1045 * file is only used in possible warning messages. 1046 */ 1047 static void 1048 copymodes(int fd, const struct stat *sbp, const char *file) 1049 { 1050 struct timeval times[2]; 1051 struct stat sb; 1052 1053 /* 1054 * If we have no info on the input, give this file some 1055 * default values and return.. 1056 */ 1057 if (sbp == NULL) { 1058 mode_t mask = umask(022); 1059 1060 (void)fchmod(fd, DEFFILEMODE & ~mask); 1061 (void)umask(mask); 1062 return; 1063 } 1064 sb = *sbp; 1065 1066 /* if the chown fails, remove set-id bits as-per compress(1) */ 1067 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1068 if (errno != EPERM) 1069 maybe_warn("couldn't fchown: %s", file); 1070 sb.st_mode &= ~(S_ISUID|S_ISGID); 1071 } 1072 1073 /* we only allow set-id and the 9 normal permission bits */ 1074 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1075 if (fchmod(fd, sb.st_mode) < 0) 1076 maybe_warn("couldn't fchmod: %s", file); 1077 1078 /* only try flags if they exist already */ 1079 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1080 maybe_warn("couldn't fchflags: %s", file); 1081 1082 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1083 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1084 if (futimes(fd, times) < 0) 1085 maybe_warn("couldn't utimes: %s", file); 1086 } 1087 #endif 1088 1089 /* what sort of file is this? */ 1090 static enum filetype 1091 file_gettype(u_char *buf) 1092 { 1093 1094 if (buf[0] == GZIP_MAGIC0 && 1095 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1096 return FT_GZIP; 1097 else 1098 #ifndef NO_BZIP2_SUPPORT 1099 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1100 buf[3] >= '0' && buf[3] <= '9') 1101 return FT_BZIP2; 1102 else 1103 #endif 1104 #ifndef NO_COMPRESS_SUPPORT 1105 if (memcmp(buf, Z_MAGIC, 2) == 0) 1106 return FT_Z; 1107 else 1108 #endif 1109 return FT_UNKNOWN; 1110 } 1111 1112 #ifndef SMALL 1113 /* check the outfile is OK. */ 1114 static int 1115 check_outfile(const char *outfile) 1116 { 1117 struct stat sb; 1118 int ok = 1; 1119 1120 if (lflag == 0 && stat(outfile, &sb) == 0) { 1121 if (fflag) 1122 unlink(outfile); 1123 else if (isatty(STDIN_FILENO)) { 1124 char ans[10] = { 'n', '\0' }; /* default */ 1125 1126 fprintf(stderr, "%s already exists -- do you wish to " 1127 "overwrite (y or n)? " , outfile); 1128 (void)fgets(ans, sizeof(ans) - 1, stdin); 1129 if (ans[0] != 'y' && ans[0] != 'Y') { 1130 fprintf(stderr, "\tnot overwriting\n"); 1131 ok = 0; 1132 } else 1133 unlink(outfile); 1134 } else { 1135 maybe_warnx("%s already exists -- skipping", outfile); 1136 ok = 0; 1137 } 1138 } 1139 return ok; 1140 } 1141 1142 static void 1143 unlink_input(const char *file, const struct stat *sb) 1144 { 1145 struct stat nsb; 1146 1147 if (kflag) 1148 return; 1149 if (stat(file, &nsb) != 0) 1150 /* Must be gone alrady */ 1151 return; 1152 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1153 /* Definitely a different file */ 1154 return; 1155 unlink(file); 1156 } 1157 #endif 1158 1159 static const suffixes_t * 1160 check_suffix(char *file, int xlate) 1161 { 1162 const suffixes_t *s; 1163 int len = strlen(file); 1164 char *sp; 1165 1166 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1167 /* if it doesn't fit in "a.suf", don't bother */ 1168 if (s->ziplen >= len) 1169 continue; 1170 sp = file + len - s->ziplen; 1171 if (strcmp(s->zipped, sp) != 0) 1172 continue; 1173 if (xlate) 1174 strcpy(sp, s->normal); 1175 return s; 1176 } 1177 return NULL; 1178 } 1179 1180 /* 1181 * compress the given file: create a corresponding .gz file and remove the 1182 * original. 1183 */ 1184 static off_t 1185 file_compress(char *file, char *outfile, size_t outsize) 1186 { 1187 int in; 1188 int out; 1189 off_t size, insize; 1190 #ifndef SMALL 1191 struct stat isb, osb; 1192 const suffixes_t *suff; 1193 #endif 1194 1195 in = open(file, O_RDONLY); 1196 if (in == -1) { 1197 maybe_warn("can't open %s", file); 1198 return -1; 1199 } 1200 1201 if (cflag == 0) { 1202 #ifndef SMALL 1203 if (fstat(in, &isb) == 0) { 1204 if (isb.st_nlink > 1 && fflag == 0) { 1205 maybe_warnx("%s has %d other link%s -- " 1206 "skipping", file, isb.st_nlink - 1, 1207 isb.st_nlink == 1 ? "" : "s"); 1208 close(in); 1209 return -1; 1210 } 1211 } 1212 1213 if (fflag == 0 && (suff = check_suffix(file, 0)) 1214 && suff->zipped[0] != 0) { 1215 maybe_warnx("%s already has %s suffix -- unchanged", 1216 file, suff->zipped); 1217 close(in); 1218 return -1; 1219 } 1220 #endif 1221 1222 /* Add (usually) .gz to filename */ 1223 if ((size_t)snprintf(outfile, outsize, "%s%s", 1224 file, suffixes[0].zipped) >= outsize) 1225 memcpy(outfile - suffixes[0].ziplen - 1, 1226 suffixes[0].zipped, suffixes[0].ziplen + 1); 1227 1228 #ifndef SMALL 1229 if (check_outfile(outfile) == 0) { 1230 close(in); 1231 return -1; 1232 } 1233 #endif 1234 } 1235 1236 if (cflag == 0) { 1237 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1238 if (out == -1) { 1239 maybe_warn("could not create output: %s", outfile); 1240 fclose(stdin); 1241 return -1; 1242 } 1243 } else 1244 out = STDOUT_FILENO; 1245 1246 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1247 1248 (void)close(in); 1249 1250 /* 1251 * If there was an error, insize will be -1. 1252 * If we compressed to stdout, just return the size. 1253 * Otherwise stat the file and check it is the correct size. 1254 * We only blow away the file if we can stat the output and it 1255 * has the expected size. 1256 */ 1257 if (cflag != 0) 1258 return insize == -1 ? -1 : size; 1259 1260 #ifndef SMALL 1261 if (fstat(out, &osb) != 0) { 1262 maybe_warn("couldn't stat: %s", outfile); 1263 goto bad_outfile; 1264 } 1265 1266 if (osb.st_size != size) { 1267 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1268 " != %" PRIdOFF "), deleting", 1269 outfile, osb.st_size, size); 1270 goto bad_outfile; 1271 } 1272 1273 copymodes(out, &isb, outfile); 1274 #endif 1275 if (close(out) == -1) 1276 maybe_warn("couldn't close output"); 1277 1278 /* output is good, ok to delete input */ 1279 unlink_input(file, &isb); 1280 return size; 1281 1282 #ifndef SMALL 1283 bad_outfile: 1284 if (close(out) == -1) 1285 maybe_warn("couldn't close output"); 1286 1287 maybe_warnx("leaving original %s", file); 1288 unlink(outfile); 1289 return size; 1290 #endif 1291 } 1292 1293 /* uncompress the given file and remove the original */ 1294 static off_t 1295 file_uncompress(char *file, char *outfile, size_t outsize) 1296 { 1297 struct stat isb, osb; 1298 off_t size; 1299 ssize_t rbytes; 1300 unsigned char header1[4]; 1301 enum filetype method; 1302 int rv, fd, ofd, zfd = -1; 1303 #ifndef SMALL 1304 time_t timestamp = 0; 1305 unsigned char name[PATH_MAX + 1]; 1306 #endif 1307 1308 /* gather the old name info */ 1309 1310 fd = open(file, O_RDONLY); 1311 if (fd < 0) { 1312 maybe_warn("can't open %s", file); 1313 goto lose; 1314 } 1315 1316 strlcpy(outfile, file, outsize); 1317 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1318 maybe_warnx("%s: unknown suffix -- ignored", file); 1319 goto lose; 1320 } 1321 1322 rbytes = read(fd, header1, sizeof header1); 1323 if (rbytes != sizeof header1) { 1324 /* we don't want to fail here. */ 1325 #ifndef SMALL 1326 if (fflag) 1327 goto lose; 1328 #endif 1329 if (rbytes == -1) 1330 maybe_warn("can't read %s", file); 1331 else 1332 goto unexpected_EOF; 1333 goto lose; 1334 } 1335 1336 method = file_gettype(header1); 1337 1338 #ifndef SMALL 1339 if (fflag == 0 && method == FT_UNKNOWN) { 1340 maybe_warnx("%s: not in gzip format", file); 1341 goto lose; 1342 } 1343 1344 #endif 1345 1346 #ifndef SMALL 1347 if (method == FT_GZIP && Nflag) { 1348 unsigned char ts[4]; /* timestamp */ 1349 1350 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1351 if (rv >= 0 && (size_t)rv < sizeof ts) 1352 goto unexpected_EOF; 1353 if (rv == -1) { 1354 if (!fflag) 1355 maybe_warn("can't read %s", file); 1356 goto lose; 1357 } 1358 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1359 1360 if (header1[3] & ORIG_NAME) { 1361 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1362 if (rbytes < 0) { 1363 maybe_warn("can't read %s", file); 1364 goto lose; 1365 } 1366 if (name[0] != 0) { 1367 /* preserve original directory name */ 1368 char *dp = strrchr(file, '/'); 1369 if (dp == NULL) 1370 dp = file; 1371 else 1372 dp++; 1373 snprintf(outfile, outsize, "%.*s%.*s", 1374 (int) (dp - file), 1375 file, (int) rbytes, name); 1376 } 1377 } 1378 } 1379 #endif 1380 lseek(fd, 0, SEEK_SET); 1381 1382 if (cflag == 0 || lflag) { 1383 if (fstat(fd, &isb) != 0) 1384 goto lose; 1385 #ifndef SMALL 1386 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1387 maybe_warnx("%s has %d other links -- skipping", 1388 file, isb.st_nlink - 1); 1389 goto lose; 1390 } 1391 if (nflag == 0 && timestamp) 1392 isb.st_mtime = timestamp; 1393 if (check_outfile(outfile) == 0) 1394 goto lose; 1395 #endif 1396 } 1397 1398 if (cflag == 0 && lflag == 0) { 1399 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1400 if (zfd == STDOUT_FILENO) { 1401 /* We won't close STDOUT_FILENO later... */ 1402 zfd = dup(zfd); 1403 close(STDOUT_FILENO); 1404 } 1405 if (zfd == -1) { 1406 maybe_warn("can't open %s", outfile); 1407 goto lose; 1408 } 1409 } else 1410 zfd = STDOUT_FILENO; 1411 1412 #ifndef NO_BZIP2_SUPPORT 1413 if (method == FT_BZIP2) { 1414 1415 /* XXX */ 1416 if (lflag) { 1417 maybe_warnx("no -l with bzip2 files"); 1418 goto lose; 1419 } 1420 1421 size = unbzip2(fd, zfd, NULL, 0, NULL); 1422 } else 1423 #endif 1424 1425 #ifndef NO_COMPRESS_SUPPORT 1426 if (method == FT_Z) { 1427 FILE *in, *out; 1428 1429 /* XXX */ 1430 if (lflag) { 1431 maybe_warnx("no -l with Lempel-Ziv files"); 1432 goto lose; 1433 } 1434 1435 if ((in = zdopen(fd)) == NULL) { 1436 maybe_warn("zdopen for read: %s", file); 1437 goto lose; 1438 } 1439 1440 out = fdopen(dup(zfd), "w"); 1441 if (out == NULL) { 1442 maybe_warn("fdopen for write: %s", outfile); 1443 fclose(in); 1444 goto lose; 1445 } 1446 1447 size = zuncompress(in, out, NULL, 0, NULL); 1448 /* need to fclose() if ferror() is true... */ 1449 if (ferror(in) | fclose(in)) { 1450 maybe_warn("failed infile fclose"); 1451 unlink(outfile); 1452 (void)fclose(out); 1453 } 1454 if (fclose(out) != 0) { 1455 maybe_warn("failed outfile fclose"); 1456 unlink(outfile); 1457 goto lose; 1458 } 1459 } else 1460 #endif 1461 1462 #ifndef SMALL 1463 if (method == FT_UNKNOWN) { 1464 if (lflag) { 1465 maybe_warnx("no -l for unknown filetypes"); 1466 goto lose; 1467 } 1468 size = cat_fd(NULL, 0, NULL, fd); 1469 } else 1470 #endif 1471 { 1472 if (lflag) { 1473 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1474 close(fd); 1475 return -1; /* XXX */ 1476 } 1477 1478 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1479 } 1480 1481 if (close(fd) != 0) 1482 maybe_warn("couldn't close input"); 1483 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1484 maybe_warn("couldn't close output"); 1485 1486 if (size == -1) { 1487 if (cflag == 0) 1488 unlink(outfile); 1489 maybe_warnx("%s: uncompress failed", file); 1490 return -1; 1491 } 1492 1493 /* if testing, or we uncompressed to stdout, this is all we need */ 1494 #ifndef SMALL 1495 if (tflag) 1496 return size; 1497 #endif 1498 /* if we are uncompressing to stdin, don't remove the file. */ 1499 if (cflag) 1500 return size; 1501 1502 /* 1503 * if we create a file... 1504 */ 1505 /* 1506 * if we can't stat the file don't remove the file. 1507 */ 1508 1509 ofd = open(outfile, O_RDWR, 0); 1510 if (ofd == -1) { 1511 maybe_warn("couldn't open (leaving original): %s", 1512 outfile); 1513 return -1; 1514 } 1515 if (fstat(ofd, &osb) != 0) { 1516 maybe_warn("couldn't stat (leaving original): %s", 1517 outfile); 1518 close(ofd); 1519 return -1; 1520 } 1521 if (osb.st_size != size) { 1522 maybe_warnx("stat gave different size: %" PRIdOFF 1523 " != %" PRIdOFF " (leaving original)", 1524 size, osb.st_size); 1525 close(ofd); 1526 unlink(outfile); 1527 return -1; 1528 } 1529 unlink_input(file, &isb); 1530 #ifndef SMALL 1531 copymodes(ofd, &isb, outfile); 1532 #endif 1533 close(ofd); 1534 return size; 1535 1536 unexpected_EOF: 1537 maybe_warnx("%s: unexpected end of file", file); 1538 lose: 1539 if (fd != -1) 1540 close(fd); 1541 if (zfd != -1 && zfd != STDOUT_FILENO) 1542 close(fd); 1543 return -1; 1544 } 1545 1546 #ifndef SMALL 1547 static off_t 1548 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1549 { 1550 char buf[BUFLEN]; 1551 off_t in_tot; 1552 ssize_t w; 1553 1554 in_tot = count; 1555 w = write(STDOUT_FILENO, prepend, count); 1556 if (w == -1 || (size_t)w != count) { 1557 maybe_warn("write to stdout"); 1558 return -1; 1559 } 1560 for (;;) { 1561 ssize_t rv; 1562 1563 rv = read(fd, buf, sizeof buf); 1564 if (rv == 0) 1565 break; 1566 if (rv < 0) { 1567 maybe_warn("read from fd %d", fd); 1568 break; 1569 } 1570 1571 if (write(STDOUT_FILENO, buf, rv) != rv) { 1572 maybe_warn("write to stdout"); 1573 break; 1574 } 1575 in_tot += rv; 1576 } 1577 1578 if (gsizep) 1579 *gsizep = in_tot; 1580 return (in_tot); 1581 } 1582 #endif 1583 1584 static void 1585 handle_stdin(void) 1586 { 1587 unsigned char header1[4]; 1588 off_t usize, gsize; 1589 enum filetype method; 1590 ssize_t bytes_read; 1591 #ifndef NO_COMPRESS_SUPPORT 1592 FILE *in; 1593 #endif 1594 1595 #ifndef SMALL 1596 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1597 maybe_warnx("standard input is a terminal -- ignoring"); 1598 return; 1599 } 1600 #endif 1601 1602 if (lflag) { 1603 struct stat isb; 1604 1605 /* XXX could read the whole file, etc. */ 1606 if (fstat(STDIN_FILENO, &isb) < 0) { 1607 maybe_warn("fstat"); 1608 return; 1609 } 1610 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1611 return; 1612 } 1613 1614 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1615 if (bytes_read == -1) { 1616 maybe_warn("can't read stdin"); 1617 return; 1618 } else if (bytes_read != sizeof(header1)) { 1619 maybe_warnx("(stdin): unexpected end of file"); 1620 return; 1621 } 1622 1623 method = file_gettype(header1); 1624 switch (method) { 1625 default: 1626 #ifndef SMALL 1627 if (fflag == 0) { 1628 maybe_warnx("unknown compression format"); 1629 return; 1630 } 1631 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1632 break; 1633 #endif 1634 case FT_GZIP: 1635 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1636 (char *)header1, sizeof header1, &gsize, "(stdin)"); 1637 break; 1638 #ifndef NO_BZIP2_SUPPORT 1639 case FT_BZIP2: 1640 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1641 (char *)header1, sizeof header1, &gsize); 1642 break; 1643 #endif 1644 #ifndef NO_COMPRESS_SUPPORT 1645 case FT_Z: 1646 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1647 maybe_warnx("zopen of stdin"); 1648 return; 1649 } 1650 1651 usize = zuncompress(in, stdout, (char *)header1, sizeof header1, &gsize); 1652 fclose(in); 1653 break; 1654 #endif 1655 } 1656 1657 #ifndef SMALL 1658 if (vflag && !tflag && usize != -1 && gsize != -1) 1659 print_verbage(NULL, NULL, usize, gsize); 1660 if (vflag && tflag) 1661 print_test("(stdin)", usize != -1); 1662 #endif 1663 1664 } 1665 1666 static void 1667 handle_stdout(void) 1668 { 1669 off_t gsize, usize; 1670 struct stat sb; 1671 time_t systime; 1672 uint32_t mtime; 1673 int ret; 1674 1675 #ifndef SMALL 1676 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1677 maybe_warnx("standard output is a terminal -- ignoring"); 1678 return; 1679 } 1680 #endif 1681 /* If stdin is a file use it's mtime, otherwise use current time */ 1682 ret = fstat(STDIN_FILENO, &sb); 1683 1684 #ifndef SMALL 1685 if (ret < 0) { 1686 maybe_warn("Can't stat stdin"); 1687 return; 1688 } 1689 #endif 1690 1691 if (S_ISREG(sb.st_mode)) 1692 mtime = (uint32_t)sb.st_mtime; 1693 else { 1694 systime = time(NULL); 1695 #ifndef SMALL 1696 if (systime == -1) { 1697 maybe_warn("time"); 1698 return; 1699 } 1700 #endif 1701 mtime = (uint32_t)systime; 1702 } 1703 1704 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1705 #ifndef SMALL 1706 if (vflag && !tflag && usize != -1 && gsize != -1) 1707 print_verbage(NULL, NULL, usize, gsize); 1708 #endif 1709 } 1710 1711 /* do what is asked for, for the path name */ 1712 static void 1713 handle_pathname(char *path) 1714 { 1715 char *opath = path, *s = NULL; 1716 ssize_t len; 1717 int slen; 1718 struct stat sb; 1719 1720 /* check for stdout/stdin */ 1721 if (path[0] == '-' && path[1] == '\0') { 1722 if (dflag) 1723 handle_stdin(); 1724 else 1725 handle_stdout(); 1726 return; 1727 } 1728 1729 retry: 1730 if (stat(path, &sb) != 0) { 1731 /* lets try <path>.gz if we're decompressing */ 1732 if (dflag && s == NULL && errno == ENOENT) { 1733 len = strlen(path); 1734 slen = suffixes[0].ziplen; 1735 s = malloc(len + slen + 1); 1736 if (s == NULL) 1737 maybe_err("malloc"); 1738 memcpy(s, path, len); 1739 memcpy(s + len, suffixes[0].zipped, slen + 1); 1740 path = s; 1741 goto retry; 1742 } 1743 maybe_warn("can't stat: %s", opath); 1744 goto out; 1745 } 1746 1747 if (S_ISDIR(sb.st_mode)) { 1748 #ifndef SMALL 1749 if (rflag) 1750 handle_dir(path); 1751 else 1752 #endif 1753 maybe_warnx("%s is a directory", path); 1754 goto out; 1755 } 1756 1757 if (S_ISREG(sb.st_mode)) 1758 handle_file(path, &sb); 1759 else 1760 maybe_warnx("%s is not a regular file", path); 1761 1762 out: 1763 if (s) 1764 free(s); 1765 } 1766 1767 /* compress/decompress a file */ 1768 static void 1769 handle_file(char *file, struct stat *sbp) 1770 { 1771 off_t usize, gsize; 1772 char outfile[PATH_MAX]; 1773 1774 infile = file; 1775 if (dflag) { 1776 usize = file_uncompress(file, outfile, sizeof(outfile)); 1777 #ifndef SMALL 1778 if (vflag && tflag) 1779 print_test(file, usize != -1); 1780 #endif 1781 if (usize == -1) 1782 return; 1783 gsize = sbp->st_size; 1784 } else { 1785 gsize = file_compress(file, outfile, sizeof(outfile)); 1786 if (gsize == -1) 1787 return; 1788 usize = sbp->st_size; 1789 } 1790 1791 1792 #ifndef SMALL 1793 if (vflag && !tflag) 1794 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1795 #endif 1796 } 1797 1798 #ifndef SMALL 1799 /* this is used with -r to recursively descend directories */ 1800 static void 1801 handle_dir(char *dir) 1802 { 1803 char *path_argv[2]; 1804 FTS *fts; 1805 FTSENT *entry; 1806 1807 path_argv[0] = dir; 1808 path_argv[1] = 0; 1809 fts = fts_open(path_argv, FTS_PHYSICAL | FTS_NOCHDIR, NULL); 1810 if (fts == NULL) { 1811 warn("couldn't fts_open %s", dir); 1812 return; 1813 } 1814 1815 while ((entry = fts_read(fts))) { 1816 switch(entry->fts_info) { 1817 case FTS_D: 1818 case FTS_DP: 1819 continue; 1820 1821 case FTS_DNR: 1822 case FTS_ERR: 1823 case FTS_NS: 1824 maybe_warn("%s", entry->fts_path); 1825 continue; 1826 case FTS_F: 1827 handle_file(entry->fts_path, entry->fts_statp); 1828 } 1829 } 1830 (void)fts_close(fts); 1831 } 1832 #endif 1833 1834 /* print a ratio - size reduction as a fraction of uncompressed size */ 1835 static void 1836 print_ratio(off_t in, off_t out, FILE *where) 1837 { 1838 int percent10; /* 10 * percent */ 1839 off_t diff; 1840 char buff[8]; 1841 int len; 1842 1843 diff = in - out/2; 1844 if (diff <= 0) 1845 /* 1846 * Output is more than double size of input! print -99.9% 1847 * Quite possibly we've failed to get the original size. 1848 */ 1849 percent10 = -999; 1850 else { 1851 /* 1852 * We only need 12 bits of result from the final division, 1853 * so reduce the values until a 32bit division will suffice. 1854 */ 1855 while (in > 0x100000) { 1856 diff >>= 1; 1857 in >>= 1; 1858 } 1859 if (in != 0) 1860 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1861 else 1862 percent10 = 0; 1863 } 1864 1865 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1866 /* Move the '.' to before the last digit */ 1867 buff[len - 1] = buff[len - 2]; 1868 buff[len - 2] = '.'; 1869 fprintf(where, "%5s%%", buff); 1870 } 1871 1872 #ifndef SMALL 1873 /* print compression statistics, and the new name (if there is one!) */ 1874 static void 1875 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1876 { 1877 if (file) 1878 fprintf(stderr, "%s:%s ", file, 1879 strlen(file) < 7 ? "\t\t" : "\t"); 1880 print_ratio(usize, gsize, stderr); 1881 if (nfile) 1882 fprintf(stderr, " -- replaced with %s", nfile); 1883 fprintf(stderr, "\n"); 1884 fflush(stderr); 1885 } 1886 1887 /* print test results */ 1888 static void 1889 print_test(const char *file, int ok) 1890 { 1891 1892 if (exit_value == 0 && ok == 0) 1893 exit_value = 1; 1894 fprintf(stderr, "%s:%s %s\n", file, 1895 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1896 fflush(stderr); 1897 } 1898 #endif 1899 1900 /* print a file's info ala --list */ 1901 /* eg: 1902 compressed uncompressed ratio uncompressed_name 1903 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1904 */ 1905 static void 1906 print_list(int fd, off_t out, const char *outfile, time_t ts) 1907 { 1908 static int first = 1; 1909 #ifndef SMALL 1910 static off_t in_tot, out_tot; 1911 uint32_t crc = 0; 1912 #endif 1913 off_t in = 0, rv; 1914 1915 if (first) { 1916 #ifndef SMALL 1917 if (vflag) 1918 printf("method crc date time "); 1919 #endif 1920 if (qflag == 0) 1921 printf(" compressed uncompressed " 1922 "ratio uncompressed_name\n"); 1923 } 1924 first = 0; 1925 1926 /* print totals? */ 1927 #ifndef SMALL 1928 if (fd == -1) { 1929 in = in_tot; 1930 out = out_tot; 1931 } else 1932 #endif 1933 { 1934 /* read the last 4 bytes - this is the uncompressed size */ 1935 rv = lseek(fd, (off_t)(-8), SEEK_END); 1936 if (rv != -1) { 1937 unsigned char buf[8]; 1938 uint32_t usize; 1939 1940 rv = read(fd, (char *)buf, sizeof(buf)); 1941 if (rv == -1) 1942 maybe_warn("read of uncompressed size"); 1943 else if (rv != sizeof(buf)) 1944 maybe_warnx("read of uncompressed size"); 1945 1946 else { 1947 usize = buf[4] | buf[5] << 8 | 1948 buf[6] << 16 | buf[7] << 24; 1949 in = (off_t)usize; 1950 #ifndef SMALL 1951 crc = buf[0] | buf[1] << 8 | 1952 buf[2] << 16 | buf[3] << 24; 1953 #endif 1954 } 1955 } 1956 } 1957 1958 #ifndef SMALL 1959 if (vflag && fd == -1) 1960 printf(" "); 1961 else if (vflag) { 1962 char *date = ctime(&ts); 1963 1964 /* skip the day, 1/100th second, and year */ 1965 date += 4; 1966 date[12] = 0; 1967 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1968 } 1969 in_tot += in; 1970 out_tot += out; 1971 #endif 1972 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1973 print_ratio(in, out, stdout); 1974 printf(" %s\n", outfile); 1975 } 1976 1977 /* display the usage of NetBSD gzip */ 1978 static void 1979 usage(void) 1980 { 1981 1982 fprintf(stderr, "%s\n", gzip_version); 1983 fprintf(stderr, 1984 #ifdef SMALL 1985 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n", 1986 #else 1987 "usage: %s [-123456789acdfhklLNnqrtVv] [-S .suffix] [<file> [<file> ...]]\n" 1988 " -1 --fast fastest (worst) compression\n" 1989 " -2 .. -8 set compression level\n" 1990 " -9 --best best (slowest) compression\n" 1991 " -c --stdout write to stdout, keep original files\n" 1992 " --to-stdout\n" 1993 " -d --decompress uncompress files\n" 1994 " --uncompress\n" 1995 " -f --force force overwriting & compress links\n" 1996 " -h --help display this help\n" 1997 " -k --keep don't delete input files during operation\n" 1998 " -l --list list compressed file contents\n" 1999 " -N --name save or restore original file name and time stamp\n" 2000 " -n --no-name don't save original file name or time stamp\n" 2001 " -q --quiet output no warnings\n" 2002 " -r --recursive recursively compress files in directories\n" 2003 " -S .suf use suffix .suf instead of .gz\n" 2004 " --suffix .suf\n" 2005 " -t --test test compressed file\n" 2006 " -V --version display program version\n" 2007 " -v --verbose print extra statistics\n", 2008 #endif 2009 getprogname()); 2010 exit(0); 2011 } 2012 2013 #ifndef SMALL 2014 /* display the license information of FreeBSD gzip */ 2015 static void 2016 display_license(void) 2017 { 2018 2019 fprintf(stderr, "%s (based on NetBSD gzip 20060927)\n", gzip_version); 2020 fprintf(stderr, "%s\n", gzip_copyright); 2021 exit(0); 2022 } 2023 #endif 2024 2025 /* display the version of NetBSD gzip */ 2026 static void 2027 display_version(void) 2028 { 2029 2030 fprintf(stderr, "%s\n", gzip_version); 2031 exit(0); 2032 } 2033 2034 #ifndef NO_BZIP2_SUPPORT 2035 #include "unbzip2.c" 2036 #endif 2037 #ifndef NO_COMPRESS_SUPPORT 2038 #include "zuncompress.c" 2039 #endif 2040 2041 static ssize_t 2042 read_retry(int fd, void *buf, size_t sz) 2043 { 2044 char *cp = buf; 2045 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2046 2047 while (left > 0) { 2048 ssize_t ret; 2049 2050 ret = read(fd, cp, left); 2051 if (ret == -1) { 2052 return ret; 2053 } else if (ret == 0) { 2054 break; /* EOF */ 2055 } 2056 cp += ret; 2057 left -= ret; 2058 } 2059 2060 return sz - left; 2061 } 2062 2063