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 SUFFIX(".tbz", ".tar"), 137 SUFFIX(".tbz2", ".tar"), 138 #endif 139 #ifndef NO_COMPRESS_SUPPORT 140 SUFFIX(Z_SUFFIX, ""), 141 #endif 142 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 143 #endif /* SMALL */ 144 #undef SUFFIX 145 }; 146 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 147 148 static const char gzip_version[] = "FreeBSD gzip 20070711"; 149 150 #ifndef SMALL 151 static const char gzip_copyright[] = \ 152 " Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green\n" 153 " All rights reserved.\n" 154 "\n" 155 " Redistribution and use in source and binary forms, with or without\n" 156 " modification, are permitted provided that the following conditions\n" 157 " are met:\n" 158 " 1. Redistributions of source code must retain the above copyright\n" 159 " notice, this list of conditions and the following disclaimer.\n" 160 " 2. Redistributions in binary form must reproduce the above copyright\n" 161 " notice, this list of conditions and the following disclaimer in the\n" 162 " documentation and/or other materials provided with the distribution.\n" 163 " 3. The name of the author may not be used to endorse or promote products\n" 164 " derived from this software without specific prior written permission.\n" 165 "\n" 166 " THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR\n" 167 " IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES\n" 168 " OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.\n" 169 " IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,\n" 170 " INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,\n" 171 " BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;\n" 172 " LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED\n" 173 " AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,\n" 174 " OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY\n" 175 " OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF\n" 176 " SUCH DAMAGE."; 177 #endif 178 179 static int cflag; /* stdout mode */ 180 static int dflag; /* decompress mode */ 181 static int lflag; /* list mode */ 182 static int numflag = 6; /* gzip -1..-9 value */ 183 184 #ifndef SMALL 185 static int fflag; /* force mode */ 186 static int kflag; /* don't delete input files */ 187 static int nflag; /* don't save name/timestamp */ 188 static int Nflag; /* don't restore name/timestamp */ 189 static int qflag; /* quiet mode */ 190 static int rflag; /* recursive mode */ 191 static int tflag; /* test */ 192 static int vflag; /* verbose mode */ 193 #else 194 #define qflag 0 195 #define tflag 0 196 #endif 197 198 static int exit_value = 0; /* exit value */ 199 200 static char *infile; /* name of file coming in */ 201 202 static void maybe_err(const char *fmt, ...) 203 __attribute__((__format__(__printf__, 1, 2))); 204 #ifndef NO_BZIP2_SUPPORT 205 static void maybe_errx(const char *fmt, ...) 206 __attribute__((__format__(__printf__, 1, 2))); 207 #endif 208 static void maybe_warn(const char *fmt, ...) 209 __attribute__((__format__(__printf__, 1, 2))); 210 static void maybe_warnx(const char *fmt, ...) 211 __attribute__((__format__(__printf__, 1, 2))); 212 static enum filetype file_gettype(u_char *); 213 #ifdef SMALL 214 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 215 #endif 216 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 217 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 218 static off_t file_compress(char *, char *, size_t); 219 static off_t file_uncompress(char *, char *, size_t); 220 static void handle_pathname(char *); 221 static void handle_file(char *, struct stat *); 222 static void handle_stdin(void); 223 static void handle_stdout(void); 224 static void print_ratio(off_t, off_t, FILE *); 225 static void print_list(int fd, off_t, const char *, time_t); 226 static void usage(void); 227 static void display_version(void); 228 #ifndef SMALL 229 static void display_license(void); 230 #endif 231 static const suffixes_t *check_suffix(char *, int); 232 static ssize_t read_retry(int, void *, size_t); 233 234 #ifdef SMALL 235 #define unlink_input(f, sb) unlink(f) 236 #else 237 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 238 static void prepend_gzip(char *, int *, char ***); 239 static void handle_dir(char *); 240 static void print_verbage(const char *, const char *, off_t, off_t); 241 static void print_test(const char *, int); 242 static void copymodes(int fd, const struct stat *, const char *file); 243 static int check_outfile(const char *outfile); 244 #endif 245 246 #ifndef NO_BZIP2_SUPPORT 247 static off_t unbzip2(int, int, char *, size_t, off_t *); 248 #endif 249 250 #ifndef NO_COMPRESS_SUPPORT 251 static FILE *zdopen(int); 252 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 253 #endif 254 255 int main(int, char **p); 256 257 #ifdef SMALL 258 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 259 #else 260 static const struct option longopts[] = { 261 { "stdout", no_argument, 0, 'c' }, 262 { "to-stdout", no_argument, 0, 'c' }, 263 { "decompress", no_argument, 0, 'd' }, 264 { "uncompress", no_argument, 0, 'd' }, 265 { "force", no_argument, 0, 'f' }, 266 { "help", no_argument, 0, 'h' }, 267 { "keep", no_argument, 0, 'k' }, 268 { "list", no_argument, 0, 'l' }, 269 { "no-name", no_argument, 0, 'n' }, 270 { "name", no_argument, 0, 'N' }, 271 { "quiet", no_argument, 0, 'q' }, 272 { "recursive", no_argument, 0, 'r' }, 273 { "suffix", required_argument, 0, 'S' }, 274 { "test", no_argument, 0, 't' }, 275 { "verbose", no_argument, 0, 'v' }, 276 { "version", no_argument, 0, 'V' }, 277 { "fast", no_argument, 0, '1' }, 278 { "best", no_argument, 0, '9' }, 279 { "ascii", no_argument, 0, 'a' }, 280 { "license", no_argument, 0, 'L' }, 281 { NULL, no_argument, 0, 0 }, 282 }; 283 #endif 284 285 int 286 main(int argc, char **argv) 287 { 288 const char *progname = getprogname(); 289 #ifndef SMALL 290 char *gzip; 291 int len; 292 #endif 293 int ch; 294 295 /* XXX set up signals */ 296 297 #ifndef SMALL 298 if ((gzip = getenv("GZIP")) != NULL) 299 prepend_gzip(gzip, &argc, &argv); 300 #endif 301 302 /* 303 * XXX 304 * handle being called `gunzip', `zcat' and `gzcat' 305 */ 306 if (strcmp(progname, "gunzip") == 0) 307 dflag = 1; 308 else if (strcmp(progname, "zcat") == 0 || 309 strcmp(progname, "gzcat") == 0) 310 dflag = cflag = 1; 311 312 #ifdef SMALL 313 #define OPT_LIST "123456789cdhltV" 314 #else 315 #define OPT_LIST "123456789acdfhklLNnqrS:tVv" 316 #endif 317 318 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 319 switch (ch) { 320 case '1': case '2': case '3': 321 case '4': case '5': case '6': 322 case '7': case '8': case '9': 323 numflag = ch - '0'; 324 break; 325 case 'c': 326 cflag = 1; 327 break; 328 case 'd': 329 dflag = 1; 330 break; 331 case 'l': 332 lflag = 1; 333 dflag = 1; 334 break; 335 case 'V': 336 display_version(); 337 /* NOTREACHED */ 338 #ifndef SMALL 339 case 'a': 340 fprintf(stderr, "%s: option --ascii ignored on this system\n", progname); 341 break; 342 case 'f': 343 fflag = 1; 344 break; 345 case 'k': 346 kflag = 1; 347 break; 348 case 'L': 349 display_license(); 350 /* NOT REACHED */ 351 case 'N': 352 nflag = 0; 353 Nflag = 1; 354 break; 355 case 'n': 356 nflag = 1; 357 Nflag = 0; 358 break; 359 case 'q': 360 qflag = 1; 361 break; 362 case 'r': 363 rflag = 1; 364 break; 365 case 'S': 366 len = strlen(optarg); 367 if (len != 0) { 368 suffixes[0].zipped = optarg; 369 suffixes[0].ziplen = len; 370 } else { 371 suffixes[NUM_SUFFIXES - 1].zipped = ""; 372 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 373 } 374 break; 375 case 't': 376 cflag = 1; 377 tflag = 1; 378 dflag = 1; 379 break; 380 case 'v': 381 vflag = 1; 382 break; 383 #endif 384 default: 385 usage(); 386 /* NOTREACHED */ 387 } 388 } 389 argv += optind; 390 argc -= optind; 391 392 if (argc == 0) { 393 if (dflag) /* stdin mode */ 394 handle_stdin(); 395 else /* stdout mode */ 396 handle_stdout(); 397 } else { 398 do { 399 handle_pathname(argv[0]); 400 } while (*++argv); 401 } 402 #ifndef SMALL 403 if (qflag == 0 && lflag && argc > 1) 404 print_list(-1, 0, "(totals)", 0); 405 #endif 406 exit(exit_value); 407 } 408 409 /* maybe print a warning */ 410 void 411 maybe_warn(const char *fmt, ...) 412 { 413 va_list ap; 414 415 if (qflag == 0) { 416 va_start(ap, fmt); 417 vwarn(fmt, ap); 418 va_end(ap); 419 } 420 if (exit_value == 0) 421 exit_value = 1; 422 } 423 424 /* ... without an errno. */ 425 void 426 maybe_warnx(const char *fmt, ...) 427 { 428 va_list ap; 429 430 if (qflag == 0) { 431 va_start(ap, fmt); 432 vwarnx(fmt, ap); 433 va_end(ap); 434 } 435 if (exit_value == 0) 436 exit_value = 1; 437 } 438 439 /* maybe print an error */ 440 void 441 maybe_err(const char *fmt, ...) 442 { 443 va_list ap; 444 445 if (qflag == 0) { 446 va_start(ap, fmt); 447 vwarn(fmt, ap); 448 va_end(ap); 449 } 450 exit(2); 451 } 452 453 #ifndef NO_BZIP2_SUPPORT 454 /* ... without an errno. */ 455 void 456 maybe_errx(const char *fmt, ...) 457 { 458 va_list ap; 459 460 if (qflag == 0) { 461 va_start(ap, fmt); 462 vwarnx(fmt, ap); 463 va_end(ap); 464 } 465 exit(2); 466 } 467 #endif 468 469 #ifndef SMALL 470 /* split up $GZIP and prepend it to the argument list */ 471 static void 472 prepend_gzip(char *gzip, int *argc, char ***argv) 473 { 474 char *s, **nargv, **ac; 475 int nenvarg = 0, i; 476 477 /* scan how many arguments there are */ 478 for (s = gzip;;) { 479 while (*s == ' ' || *s == '\t') 480 s++; 481 if (*s == 0) 482 goto count_done; 483 nenvarg++; 484 while (*s != ' ' && *s != '\t') 485 if (*s++ == 0) 486 goto count_done; 487 } 488 count_done: 489 /* punt early */ 490 if (nenvarg == 0) 491 return; 492 493 *argc += nenvarg; 494 ac = *argv; 495 496 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 497 if (nargv == NULL) 498 maybe_err("malloc"); 499 500 /* stash this away */ 501 *argv = nargv; 502 503 /* copy the program name first */ 504 i = 0; 505 nargv[i++] = *(ac++); 506 507 /* take a copy of $GZIP and add it to the array */ 508 s = strdup(gzip); 509 if (s == NULL) 510 maybe_err("strdup"); 511 for (;;) { 512 /* Skip whitespaces. */ 513 while (*s == ' ' || *s == '\t') 514 s++; 515 if (*s == 0) 516 goto copy_done; 517 nargv[i++] = s; 518 /* Find the end of this argument. */ 519 while (*s != ' ' && *s != '\t') 520 if (*s++ == 0) 521 /* Argument followed by NUL. */ 522 goto copy_done; 523 /* Terminate by overwriting ' ' or '\t' with NUL. */ 524 *s++ = 0; 525 } 526 copy_done: 527 528 /* copy the original arguments and a NULL */ 529 while (*ac) 530 nargv[i++] = *(ac++); 531 nargv[i] = NULL; 532 } 533 #endif 534 535 /* compress input to output. Return bytes read, -1 on error */ 536 static off_t 537 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 538 { 539 z_stream z; 540 char *outbufp, *inbufp; 541 off_t in_tot = 0, out_tot = 0; 542 ssize_t in_size; 543 int i, error; 544 uLong crc; 545 #ifdef SMALL 546 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 547 0, 0, 0, 0, 548 0, OS_CODE }; 549 #endif 550 551 outbufp = malloc(BUFLEN); 552 inbufp = malloc(BUFLEN); 553 if (outbufp == NULL || inbufp == NULL) { 554 maybe_err("malloc failed"); 555 goto out; 556 } 557 558 memset(&z, 0, sizeof z); 559 z.zalloc = Z_NULL; 560 z.zfree = Z_NULL; 561 z.opaque = 0; 562 563 #ifdef SMALL 564 memcpy(outbufp, header, sizeof header); 565 i = sizeof header; 566 #else 567 if (nflag != 0) { 568 mtime = 0; 569 origname = ""; 570 } 571 572 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 573 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 574 *origname ? ORIG_NAME : 0, 575 mtime & 0xff, 576 (mtime >> 8) & 0xff, 577 (mtime >> 16) & 0xff, 578 (mtime >> 24) & 0xff, 579 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 580 OS_CODE, origname); 581 if (i >= BUFLEN) 582 /* this need PATH_MAX > BUFLEN ... */ 583 maybe_err("snprintf"); 584 if (*origname) 585 i++; 586 #endif 587 588 z.next_out = (unsigned char *)outbufp + i; 589 z.avail_out = BUFLEN - i; 590 591 error = deflateInit2(&z, numflag, Z_DEFLATED, 592 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 593 if (error != Z_OK) { 594 maybe_warnx("deflateInit2 failed"); 595 in_tot = -1; 596 goto out; 597 } 598 599 crc = crc32(0L, Z_NULL, 0); 600 for (;;) { 601 if (z.avail_out == 0) { 602 if (write(out, outbufp, BUFLEN) != BUFLEN) { 603 maybe_warn("write"); 604 out_tot = -1; 605 goto out; 606 } 607 608 out_tot += BUFLEN; 609 z.next_out = (unsigned char *)outbufp; 610 z.avail_out = BUFLEN; 611 } 612 613 if (z.avail_in == 0) { 614 in_size = read(in, inbufp, BUFLEN); 615 if (in_size < 0) { 616 maybe_warn("read"); 617 in_tot = -1; 618 goto out; 619 } 620 if (in_size == 0) 621 break; 622 623 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 624 in_tot += in_size; 625 z.next_in = (unsigned char *)inbufp; 626 z.avail_in = in_size; 627 } 628 629 error = deflate(&z, Z_NO_FLUSH); 630 if (error != Z_OK && error != Z_STREAM_END) { 631 maybe_warnx("deflate failed"); 632 in_tot = -1; 633 goto out; 634 } 635 } 636 637 /* clean up */ 638 for (;;) { 639 size_t len; 640 ssize_t w; 641 642 error = deflate(&z, Z_FINISH); 643 if (error != Z_OK && error != Z_STREAM_END) { 644 maybe_warnx("deflate failed"); 645 in_tot = -1; 646 goto out; 647 } 648 649 len = (char *)z.next_out - outbufp; 650 651 w = write(out, outbufp, len); 652 if (w == -1 || (size_t)w != len) { 653 maybe_warn("write"); 654 out_tot = -1; 655 goto out; 656 } 657 out_tot += len; 658 z.next_out = (unsigned char *)outbufp; 659 z.avail_out = BUFLEN; 660 661 if (error == Z_STREAM_END) 662 break; 663 } 664 665 if (deflateEnd(&z) != Z_OK) { 666 maybe_warnx("deflateEnd failed"); 667 in_tot = -1; 668 goto out; 669 } 670 671 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 672 (int)crc & 0xff, 673 (int)(crc >> 8) & 0xff, 674 (int)(crc >> 16) & 0xff, 675 (int)(crc >> 24) & 0xff, 676 (int)in_tot & 0xff, 677 (int)(in_tot >> 8) & 0xff, 678 (int)(in_tot >> 16) & 0xff, 679 (int)(in_tot >> 24) & 0xff); 680 if (i != 8) 681 maybe_err("snprintf"); 682 if (write(out, outbufp, i) != i) { 683 maybe_warn("write"); 684 in_tot = -1; 685 } else 686 out_tot += i; 687 688 out: 689 if (inbufp != NULL) 690 free(inbufp); 691 if (outbufp != NULL) 692 free(outbufp); 693 if (gsizep) 694 *gsizep = out_tot; 695 return in_tot; 696 } 697 698 /* 699 * uncompress input to output then close the input. return the 700 * uncompressed size written, and put the compressed sized read 701 * into `*gsizep'. 702 */ 703 static off_t 704 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 705 const char *filename) 706 { 707 z_stream z; 708 char *outbufp, *inbufp; 709 off_t out_tot = -1, in_tot = 0; 710 uint32_t out_sub_tot = 0; 711 enum { 712 GZSTATE_MAGIC0, 713 GZSTATE_MAGIC1, 714 GZSTATE_METHOD, 715 GZSTATE_FLAGS, 716 GZSTATE_SKIPPING, 717 GZSTATE_EXTRA, 718 GZSTATE_EXTRA2, 719 GZSTATE_EXTRA3, 720 GZSTATE_ORIGNAME, 721 GZSTATE_COMMENT, 722 GZSTATE_HEAD_CRC1, 723 GZSTATE_HEAD_CRC2, 724 GZSTATE_INIT, 725 GZSTATE_READ, 726 GZSTATE_CRC, 727 GZSTATE_LEN, 728 } state = GZSTATE_MAGIC0; 729 int flags = 0, skip_count = 0; 730 int error = Z_STREAM_ERROR, done_reading = 0; 731 uLong crc = 0; 732 ssize_t wr; 733 int needmore = 0; 734 735 #define ADVANCE() { z.next_in++; z.avail_in--; } 736 737 if ((outbufp = malloc(BUFLEN)) == NULL) { 738 maybe_err("malloc failed"); 739 goto out2; 740 } 741 if ((inbufp = malloc(BUFLEN)) == NULL) { 742 maybe_err("malloc failed"); 743 goto out1; 744 } 745 746 memset(&z, 0, sizeof z); 747 z.avail_in = prelen; 748 z.next_in = (unsigned char *)pre; 749 z.avail_out = BUFLEN; 750 z.next_out = (unsigned char *)outbufp; 751 z.zalloc = NULL; 752 z.zfree = NULL; 753 z.opaque = 0; 754 755 in_tot = prelen; 756 out_tot = 0; 757 758 for (;;) { 759 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 760 ssize_t in_size; 761 762 if (z.avail_in > 0) { 763 memmove(inbufp, z.next_in, z.avail_in); 764 } 765 z.next_in = (unsigned char *)inbufp; 766 in_size = read(in, z.next_in + z.avail_in, 767 BUFLEN - z.avail_in); 768 769 if (in_size == -1) { 770 maybe_warn("failed to read stdin"); 771 goto stop_and_fail; 772 } else if (in_size == 0) { 773 done_reading = 1; 774 } 775 776 z.avail_in += in_size; 777 needmore = 0; 778 779 in_tot += in_size; 780 } 781 if (z.avail_in == 0) { 782 if (done_reading && state != GZSTATE_MAGIC0) { 783 maybe_warnx("%s: unexpected end of file", 784 filename); 785 goto stop_and_fail; 786 } 787 goto stop; 788 } 789 switch (state) { 790 case GZSTATE_MAGIC0: 791 if (*z.next_in != GZIP_MAGIC0) { 792 if (in_tot > 0) { 793 maybe_warnx("%s: trailing garbage " 794 "ignored", filename); 795 goto stop; 796 } 797 maybe_warnx("input not gziped (MAGIC0)"); 798 goto stop_and_fail; 799 } 800 ADVANCE(); 801 state++; 802 out_sub_tot = 0; 803 crc = crc32(0L, Z_NULL, 0); 804 break; 805 806 case GZSTATE_MAGIC1: 807 if (*z.next_in != GZIP_MAGIC1 && 808 *z.next_in != GZIP_OMAGIC1) { 809 maybe_warnx("input not gziped (MAGIC1)"); 810 goto stop_and_fail; 811 } 812 ADVANCE(); 813 state++; 814 break; 815 816 case GZSTATE_METHOD: 817 if (*z.next_in != Z_DEFLATED) { 818 maybe_warnx("unknown compression method"); 819 goto stop_and_fail; 820 } 821 ADVANCE(); 822 state++; 823 break; 824 825 case GZSTATE_FLAGS: 826 flags = *z.next_in; 827 ADVANCE(); 828 skip_count = 6; 829 state++; 830 break; 831 832 case GZSTATE_SKIPPING: 833 if (skip_count > 0) { 834 skip_count--; 835 ADVANCE(); 836 } else 837 state++; 838 break; 839 840 case GZSTATE_EXTRA: 841 if ((flags & EXTRA_FIELD) == 0) { 842 state = GZSTATE_ORIGNAME; 843 break; 844 } 845 skip_count = *z.next_in; 846 ADVANCE(); 847 state++; 848 break; 849 850 case GZSTATE_EXTRA2: 851 skip_count |= ((*z.next_in) << 8); 852 ADVANCE(); 853 state++; 854 break; 855 856 case GZSTATE_EXTRA3: 857 if (skip_count > 0) { 858 skip_count--; 859 ADVANCE(); 860 } else 861 state++; 862 break; 863 864 case GZSTATE_ORIGNAME: 865 if ((flags & ORIG_NAME) == 0) { 866 state++; 867 break; 868 } 869 if (*z.next_in == 0) 870 state++; 871 ADVANCE(); 872 break; 873 874 case GZSTATE_COMMENT: 875 if ((flags & COMMENT) == 0) { 876 state++; 877 break; 878 } 879 if (*z.next_in == 0) 880 state++; 881 ADVANCE(); 882 break; 883 884 case GZSTATE_HEAD_CRC1: 885 if (flags & HEAD_CRC) 886 skip_count = 2; 887 else 888 skip_count = 0; 889 state++; 890 break; 891 892 case GZSTATE_HEAD_CRC2: 893 if (skip_count > 0) { 894 skip_count--; 895 ADVANCE(); 896 } else 897 state++; 898 break; 899 900 case GZSTATE_INIT: 901 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 902 maybe_warnx("failed to inflateInit"); 903 goto stop_and_fail; 904 } 905 state++; 906 break; 907 908 case GZSTATE_READ: 909 error = inflate(&z, Z_FINISH); 910 switch (error) { 911 /* Z_BUF_ERROR goes with Z_FINISH... */ 912 case Z_BUF_ERROR: 913 case Z_STREAM_END: 914 case Z_OK: 915 break; 916 917 case Z_NEED_DICT: 918 maybe_warnx("Z_NEED_DICT error"); 919 goto stop_and_fail; 920 case Z_DATA_ERROR: 921 maybe_warnx("data stream error"); 922 goto stop_and_fail; 923 case Z_STREAM_ERROR: 924 maybe_warnx("internal stream error"); 925 goto stop_and_fail; 926 case Z_MEM_ERROR: 927 maybe_warnx("memory allocation error"); 928 goto stop_and_fail; 929 930 default: 931 maybe_warn("unknown error from inflate(): %d", 932 error); 933 } 934 wr = BUFLEN - z.avail_out; 935 936 if (wr != 0) { 937 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 938 if ( 939 #ifndef SMALL 940 /* don't write anything with -t */ 941 tflag == 0 && 942 #endif 943 write(out, outbufp, wr) != wr) { 944 maybe_warn("error writing to output"); 945 goto stop_and_fail; 946 } 947 948 out_tot += wr; 949 out_sub_tot += wr; 950 } 951 952 if (error == Z_STREAM_END) { 953 inflateEnd(&z); 954 state++; 955 } 956 957 z.next_out = (unsigned char *)outbufp; 958 z.avail_out = BUFLEN; 959 960 break; 961 case GZSTATE_CRC: 962 { 963 uLong origcrc; 964 965 if (z.avail_in < 4) { 966 if (!done_reading) { 967 needmore = 1; 968 continue; 969 } 970 maybe_warnx("truncated input"); 971 goto stop_and_fail; 972 } 973 origcrc = ((unsigned)z.next_in[0] & 0xff) | 974 ((unsigned)z.next_in[1] & 0xff) << 8 | 975 ((unsigned)z.next_in[2] & 0xff) << 16 | 976 ((unsigned)z.next_in[3] & 0xff) << 24; 977 if (origcrc != crc) { 978 maybe_warnx("invalid compressed" 979 " data--crc error"); 980 goto stop_and_fail; 981 } 982 } 983 984 z.avail_in -= 4; 985 z.next_in += 4; 986 987 if (!z.avail_in && done_reading) { 988 goto stop; 989 } 990 state++; 991 break; 992 case GZSTATE_LEN: 993 { 994 uLong origlen; 995 996 if (z.avail_in < 4) { 997 if (!done_reading) { 998 needmore = 1; 999 continue; 1000 } 1001 maybe_warnx("truncated input"); 1002 goto stop_and_fail; 1003 } 1004 origlen = ((unsigned)z.next_in[0] & 0xff) | 1005 ((unsigned)z.next_in[1] & 0xff) << 8 | 1006 ((unsigned)z.next_in[2] & 0xff) << 16 | 1007 ((unsigned)z.next_in[3] & 0xff) << 24; 1008 1009 if (origlen != out_sub_tot) { 1010 maybe_warnx("invalid compressed" 1011 " data--length error"); 1012 goto stop_and_fail; 1013 } 1014 } 1015 1016 z.avail_in -= 4; 1017 z.next_in += 4; 1018 1019 if (error < 0) { 1020 maybe_warnx("decompression error"); 1021 goto stop_and_fail; 1022 } 1023 state = GZSTATE_MAGIC0; 1024 break; 1025 } 1026 continue; 1027 stop_and_fail: 1028 out_tot = -1; 1029 stop: 1030 break; 1031 } 1032 if (state > GZSTATE_INIT) 1033 inflateEnd(&z); 1034 1035 free(inbufp); 1036 out1: 1037 free(outbufp); 1038 out2: 1039 if (gsizep) 1040 *gsizep = in_tot; 1041 return (out_tot); 1042 } 1043 1044 #ifndef SMALL 1045 /* 1046 * set the owner, mode, flags & utimes using the given file descriptor. 1047 * file is only used in possible warning messages. 1048 */ 1049 static void 1050 copymodes(int fd, const struct stat *sbp, const char *file) 1051 { 1052 struct timeval times[2]; 1053 struct stat sb; 1054 1055 /* 1056 * If we have no info on the input, give this file some 1057 * default values and return.. 1058 */ 1059 if (sbp == NULL) { 1060 mode_t mask = umask(022); 1061 1062 (void)fchmod(fd, DEFFILEMODE & ~mask); 1063 (void)umask(mask); 1064 return; 1065 } 1066 sb = *sbp; 1067 1068 /* if the chown fails, remove set-id bits as-per compress(1) */ 1069 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1070 if (errno != EPERM) 1071 maybe_warn("couldn't fchown: %s", file); 1072 sb.st_mode &= ~(S_ISUID|S_ISGID); 1073 } 1074 1075 /* we only allow set-id and the 9 normal permission bits */ 1076 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1077 if (fchmod(fd, sb.st_mode) < 0) 1078 maybe_warn("couldn't fchmod: %s", file); 1079 1080 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1081 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1082 if (futimes(fd, times) < 0) 1083 maybe_warn("couldn't utimes: %s", file); 1084 1085 /* only try flags if they exist already */ 1086 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1087 maybe_warn("couldn't fchflags: %s", file); 1088 } 1089 #endif 1090 1091 /* what sort of file is this? */ 1092 static enum filetype 1093 file_gettype(u_char *buf) 1094 { 1095 1096 if (buf[0] == GZIP_MAGIC0 && 1097 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1098 return FT_GZIP; 1099 else 1100 #ifndef NO_BZIP2_SUPPORT 1101 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1102 buf[3] >= '0' && buf[3] <= '9') 1103 return FT_BZIP2; 1104 else 1105 #endif 1106 #ifndef NO_COMPRESS_SUPPORT 1107 if (memcmp(buf, Z_MAGIC, 2) == 0) 1108 return FT_Z; 1109 else 1110 #endif 1111 return FT_UNKNOWN; 1112 } 1113 1114 #ifndef SMALL 1115 /* check the outfile is OK. */ 1116 static int 1117 check_outfile(const char *outfile) 1118 { 1119 struct stat sb; 1120 int ok = 1; 1121 1122 if (lflag == 0 && stat(outfile, &sb) == 0) { 1123 if (fflag) 1124 unlink(outfile); 1125 else if (isatty(STDIN_FILENO)) { 1126 char ans[10] = { 'n', '\0' }; /* default */ 1127 1128 fprintf(stderr, "%s already exists -- do you wish to " 1129 "overwrite (y or n)? " , outfile); 1130 (void)fgets(ans, sizeof(ans) - 1, stdin); 1131 if (ans[0] != 'y' && ans[0] != 'Y') { 1132 fprintf(stderr, "\tnot overwriting\n"); 1133 ok = 0; 1134 } else 1135 unlink(outfile); 1136 } else { 1137 maybe_warnx("%s already exists -- skipping", outfile); 1138 ok = 0; 1139 } 1140 } 1141 return ok; 1142 } 1143 1144 static void 1145 unlink_input(const char *file, const struct stat *sb) 1146 { 1147 struct stat nsb; 1148 1149 if (kflag) 1150 return; 1151 if (stat(file, &nsb) != 0) 1152 /* Must be gone alrady */ 1153 return; 1154 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1155 /* Definitely a different file */ 1156 return; 1157 unlink(file); 1158 } 1159 #endif 1160 1161 static const suffixes_t * 1162 check_suffix(char *file, int xlate) 1163 { 1164 const suffixes_t *s; 1165 int len = strlen(file); 1166 char *sp; 1167 1168 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1169 /* if it doesn't fit in "a.suf", don't bother */ 1170 if (s->ziplen >= len) 1171 continue; 1172 sp = file + len - s->ziplen; 1173 if (strcmp(s->zipped, sp) != 0) 1174 continue; 1175 if (xlate) 1176 strcpy(sp, s->normal); 1177 return s; 1178 } 1179 return NULL; 1180 } 1181 1182 /* 1183 * compress the given file: create a corresponding .gz file and remove the 1184 * original. 1185 */ 1186 static off_t 1187 file_compress(char *file, char *outfile, size_t outsize) 1188 { 1189 int in; 1190 int out; 1191 off_t size, insize; 1192 #ifndef SMALL 1193 struct stat isb, osb; 1194 const suffixes_t *suff; 1195 #endif 1196 1197 in = open(file, O_RDONLY); 1198 if (in == -1) { 1199 maybe_warn("can't open %s", file); 1200 return -1; 1201 } 1202 1203 if (cflag == 0) { 1204 #ifndef SMALL 1205 if (fstat(in, &isb) == 0) { 1206 if (isb.st_nlink > 1 && fflag == 0) { 1207 maybe_warnx("%s has %d other link%s -- " 1208 "skipping", file, isb.st_nlink - 1, 1209 isb.st_nlink == 1 ? "" : "s"); 1210 close(in); 1211 return -1; 1212 } 1213 } 1214 1215 if (fflag == 0 && (suff = check_suffix(file, 0)) 1216 && suff->zipped[0] != 0) { 1217 maybe_warnx("%s already has %s suffix -- unchanged", 1218 file, suff->zipped); 1219 close(in); 1220 return -1; 1221 } 1222 #endif 1223 1224 /* Add (usually) .gz to filename */ 1225 if ((size_t)snprintf(outfile, outsize, "%s%s", 1226 file, suffixes[0].zipped) >= outsize) 1227 memcpy(outfile - suffixes[0].ziplen - 1, 1228 suffixes[0].zipped, suffixes[0].ziplen + 1); 1229 1230 #ifndef SMALL 1231 if (check_outfile(outfile) == 0) { 1232 close(in); 1233 return -1; 1234 } 1235 #endif 1236 } 1237 1238 if (cflag == 0) { 1239 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1240 if (out == -1) { 1241 maybe_warn("could not create output: %s", outfile); 1242 fclose(stdin); 1243 return -1; 1244 } 1245 } else 1246 out = STDOUT_FILENO; 1247 1248 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1249 1250 (void)close(in); 1251 1252 /* 1253 * If there was an error, insize will be -1. 1254 * If we compressed to stdout, just return the size. 1255 * Otherwise stat the file and check it is the correct size. 1256 * We only blow away the file if we can stat the output and it 1257 * has the expected size. 1258 */ 1259 if (cflag != 0) 1260 return insize == -1 ? -1 : size; 1261 1262 #ifndef SMALL 1263 if (fstat(out, &osb) != 0) { 1264 maybe_warn("couldn't stat: %s", outfile); 1265 goto bad_outfile; 1266 } 1267 1268 if (osb.st_size != size) { 1269 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1270 " != %" PRIdOFF "), deleting", 1271 outfile, osb.st_size, size); 1272 goto bad_outfile; 1273 } 1274 1275 copymodes(out, &isb, outfile); 1276 #endif 1277 if (close(out) == -1) 1278 maybe_warn("couldn't close output"); 1279 1280 /* output is good, ok to delete input */ 1281 unlink_input(file, &isb); 1282 return size; 1283 1284 #ifndef SMALL 1285 bad_outfile: 1286 if (close(out) == -1) 1287 maybe_warn("couldn't close output"); 1288 1289 maybe_warnx("leaving original %s", file); 1290 unlink(outfile); 1291 return size; 1292 #endif 1293 } 1294 1295 /* uncompress the given file and remove the original */ 1296 static off_t 1297 file_uncompress(char *file, char *outfile, size_t outsize) 1298 { 1299 struct stat isb, osb; 1300 off_t size; 1301 ssize_t rbytes; 1302 unsigned char header1[4]; 1303 enum filetype method; 1304 int rv, fd, ofd, zfd = -1; 1305 #ifndef SMALL 1306 time_t timestamp = 0; 1307 unsigned char name[PATH_MAX + 1]; 1308 #endif 1309 1310 /* gather the old name info */ 1311 1312 fd = open(file, O_RDONLY); 1313 if (fd < 0) { 1314 maybe_warn("can't open %s", file); 1315 goto lose; 1316 } 1317 1318 strlcpy(outfile, file, outsize); 1319 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1320 maybe_warnx("%s: unknown suffix -- ignored", file); 1321 goto lose; 1322 } 1323 1324 rbytes = read(fd, header1, sizeof header1); 1325 if (rbytes != sizeof header1) { 1326 /* we don't want to fail here. */ 1327 #ifndef SMALL 1328 if (fflag) 1329 goto lose; 1330 #endif 1331 if (rbytes == -1) 1332 maybe_warn("can't read %s", file); 1333 else 1334 goto unexpected_EOF; 1335 goto lose; 1336 } 1337 1338 method = file_gettype(header1); 1339 1340 #ifndef SMALL 1341 if (fflag == 0 && method == FT_UNKNOWN) { 1342 maybe_warnx("%s: not in gzip format", file); 1343 goto lose; 1344 } 1345 1346 #endif 1347 1348 #ifndef SMALL 1349 if (method == FT_GZIP && Nflag) { 1350 unsigned char ts[4]; /* timestamp */ 1351 1352 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1353 if (rv >= 0 && (size_t)rv < sizeof ts) 1354 goto unexpected_EOF; 1355 if (rv == -1) { 1356 if (!fflag) 1357 maybe_warn("can't read %s", file); 1358 goto lose; 1359 } 1360 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1361 1362 if (header1[3] & ORIG_NAME) { 1363 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1364 if (rbytes < 0) { 1365 maybe_warn("can't read %s", file); 1366 goto lose; 1367 } 1368 if (name[0] != 0) { 1369 /* preserve original directory name */ 1370 char *dp = strrchr(file, '/'); 1371 if (dp == NULL) 1372 dp = file; 1373 else 1374 dp++; 1375 snprintf(outfile, outsize, "%.*s%.*s", 1376 (int) (dp - file), 1377 file, (int) rbytes, name); 1378 } 1379 } 1380 } 1381 #endif 1382 lseek(fd, 0, SEEK_SET); 1383 1384 if (cflag == 0 || lflag) { 1385 if (fstat(fd, &isb) != 0) 1386 goto lose; 1387 #ifndef SMALL 1388 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1389 maybe_warnx("%s has %d other links -- skipping", 1390 file, isb.st_nlink - 1); 1391 goto lose; 1392 } 1393 if (nflag == 0 && timestamp) 1394 isb.st_mtime = timestamp; 1395 if (check_outfile(outfile) == 0) 1396 goto lose; 1397 #endif 1398 } 1399 1400 if (cflag == 0 && lflag == 0) { 1401 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1402 if (zfd == STDOUT_FILENO) { 1403 /* We won't close STDOUT_FILENO later... */ 1404 zfd = dup(zfd); 1405 close(STDOUT_FILENO); 1406 } 1407 if (zfd == -1) { 1408 maybe_warn("can't open %s", outfile); 1409 goto lose; 1410 } 1411 } else 1412 zfd = STDOUT_FILENO; 1413 1414 #ifndef NO_BZIP2_SUPPORT 1415 if (method == FT_BZIP2) { 1416 1417 /* XXX */ 1418 if (lflag) { 1419 maybe_warnx("no -l with bzip2 files"); 1420 goto lose; 1421 } 1422 1423 size = unbzip2(fd, zfd, NULL, 0, NULL); 1424 } else 1425 #endif 1426 1427 #ifndef NO_COMPRESS_SUPPORT 1428 if (method == FT_Z) { 1429 FILE *in, *out; 1430 1431 /* XXX */ 1432 if (lflag) { 1433 maybe_warnx("no -l with Lempel-Ziv files"); 1434 goto lose; 1435 } 1436 1437 if ((in = zdopen(fd)) == NULL) { 1438 maybe_warn("zdopen for read: %s", file); 1439 goto lose; 1440 } 1441 1442 out = fdopen(dup(zfd), "w"); 1443 if (out == NULL) { 1444 maybe_warn("fdopen for write: %s", outfile); 1445 fclose(in); 1446 goto lose; 1447 } 1448 1449 size = zuncompress(in, out, NULL, 0, NULL); 1450 /* need to fclose() if ferror() is true... */ 1451 if (ferror(in) | fclose(in)) { 1452 maybe_warn("failed infile fclose"); 1453 unlink(outfile); 1454 (void)fclose(out); 1455 } 1456 if (fclose(out) != 0) { 1457 maybe_warn("failed outfile fclose"); 1458 unlink(outfile); 1459 goto lose; 1460 } 1461 } else 1462 #endif 1463 1464 #ifndef SMALL 1465 if (method == FT_UNKNOWN) { 1466 if (lflag) { 1467 maybe_warnx("no -l for unknown filetypes"); 1468 goto lose; 1469 } 1470 size = cat_fd(NULL, 0, NULL, fd); 1471 } else 1472 #endif 1473 { 1474 if (lflag) { 1475 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1476 close(fd); 1477 return -1; /* XXX */ 1478 } 1479 1480 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1481 } 1482 1483 if (close(fd) != 0) 1484 maybe_warn("couldn't close input"); 1485 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1486 maybe_warn("couldn't close output"); 1487 1488 if (size == -1) { 1489 if (cflag == 0) 1490 unlink(outfile); 1491 maybe_warnx("%s: uncompress failed", file); 1492 return -1; 1493 } 1494 1495 /* if testing, or we uncompressed to stdout, this is all we need */ 1496 #ifndef SMALL 1497 if (tflag) 1498 return size; 1499 #endif 1500 /* if we are uncompressing to stdin, don't remove the file. */ 1501 if (cflag) 1502 return size; 1503 1504 /* 1505 * if we create a file... 1506 */ 1507 /* 1508 * if we can't stat the file don't remove the file. 1509 */ 1510 1511 ofd = open(outfile, O_RDWR, 0); 1512 if (ofd == -1) { 1513 maybe_warn("couldn't open (leaving original): %s", 1514 outfile); 1515 return -1; 1516 } 1517 if (fstat(ofd, &osb) != 0) { 1518 maybe_warn("couldn't stat (leaving original): %s", 1519 outfile); 1520 close(ofd); 1521 return -1; 1522 } 1523 if (osb.st_size != size) { 1524 maybe_warnx("stat gave different size: %" PRIdOFF 1525 " != %" PRIdOFF " (leaving original)", 1526 size, osb.st_size); 1527 close(ofd); 1528 unlink(outfile); 1529 return -1; 1530 } 1531 unlink_input(file, &isb); 1532 #ifndef SMALL 1533 copymodes(ofd, &isb, outfile); 1534 #endif 1535 close(ofd); 1536 return size; 1537 1538 unexpected_EOF: 1539 maybe_warnx("%s: unexpected end of file", file); 1540 lose: 1541 if (fd != -1) 1542 close(fd); 1543 if (zfd != -1 && zfd != STDOUT_FILENO) 1544 close(fd); 1545 return -1; 1546 } 1547 1548 #ifndef SMALL 1549 static off_t 1550 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1551 { 1552 char buf[BUFLEN]; 1553 off_t in_tot; 1554 ssize_t w; 1555 1556 in_tot = count; 1557 w = write(STDOUT_FILENO, prepend, count); 1558 if (w == -1 || (size_t)w != count) { 1559 maybe_warn("write to stdout"); 1560 return -1; 1561 } 1562 for (;;) { 1563 ssize_t rv; 1564 1565 rv = read(fd, buf, sizeof buf); 1566 if (rv == 0) 1567 break; 1568 if (rv < 0) { 1569 maybe_warn("read from fd %d", fd); 1570 break; 1571 } 1572 1573 if (write(STDOUT_FILENO, buf, rv) != rv) { 1574 maybe_warn("write to stdout"); 1575 break; 1576 } 1577 in_tot += rv; 1578 } 1579 1580 if (gsizep) 1581 *gsizep = in_tot; 1582 return (in_tot); 1583 } 1584 #endif 1585 1586 static void 1587 handle_stdin(void) 1588 { 1589 unsigned char header1[4]; 1590 off_t usize, gsize; 1591 enum filetype method; 1592 ssize_t bytes_read; 1593 #ifndef NO_COMPRESS_SUPPORT 1594 FILE *in; 1595 #endif 1596 1597 #ifndef SMALL 1598 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1599 maybe_warnx("standard input is a terminal -- ignoring"); 1600 return; 1601 } 1602 #endif 1603 1604 if (lflag) { 1605 struct stat isb; 1606 1607 /* XXX could read the whole file, etc. */ 1608 if (fstat(STDIN_FILENO, &isb) < 0) { 1609 maybe_warn("fstat"); 1610 return; 1611 } 1612 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1613 return; 1614 } 1615 1616 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1617 if (bytes_read == -1) { 1618 maybe_warn("can't read stdin"); 1619 return; 1620 } else if (bytes_read != sizeof(header1)) { 1621 maybe_warnx("(stdin): unexpected end of file"); 1622 return; 1623 } 1624 1625 method = file_gettype(header1); 1626 switch (method) { 1627 default: 1628 #ifndef SMALL 1629 if (fflag == 0) { 1630 maybe_warnx("unknown compression format"); 1631 return; 1632 } 1633 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1634 break; 1635 #endif 1636 case FT_GZIP: 1637 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1638 (char *)header1, sizeof header1, &gsize, "(stdin)"); 1639 break; 1640 #ifndef NO_BZIP2_SUPPORT 1641 case FT_BZIP2: 1642 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1643 (char *)header1, sizeof header1, &gsize); 1644 break; 1645 #endif 1646 #ifndef NO_COMPRESS_SUPPORT 1647 case FT_Z: 1648 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1649 maybe_warnx("zopen of stdin"); 1650 return; 1651 } 1652 1653 usize = zuncompress(in, stdout, (char *)header1, sizeof header1, &gsize); 1654 fclose(in); 1655 break; 1656 #endif 1657 } 1658 1659 #ifndef SMALL 1660 if (vflag && !tflag && usize != -1 && gsize != -1) 1661 print_verbage(NULL, NULL, usize, gsize); 1662 if (vflag && tflag) 1663 print_test("(stdin)", usize != -1); 1664 #endif 1665 1666 } 1667 1668 static void 1669 handle_stdout(void) 1670 { 1671 off_t gsize, usize; 1672 struct stat sb; 1673 time_t systime; 1674 uint32_t mtime; 1675 int ret; 1676 1677 #ifndef SMALL 1678 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1679 maybe_warnx("standard output is a terminal -- ignoring"); 1680 return; 1681 } 1682 #endif 1683 /* If stdin is a file use it's mtime, otherwise use current time */ 1684 ret = fstat(STDIN_FILENO, &sb); 1685 1686 #ifndef SMALL 1687 if (ret < 0) { 1688 maybe_warn("Can't stat stdin"); 1689 return; 1690 } 1691 #endif 1692 1693 if (S_ISREG(sb.st_mode)) 1694 mtime = (uint32_t)sb.st_mtime; 1695 else { 1696 systime = time(NULL); 1697 #ifndef SMALL 1698 if (systime == -1) { 1699 maybe_warn("time"); 1700 return; 1701 } 1702 #endif 1703 mtime = (uint32_t)systime; 1704 } 1705 1706 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1707 #ifndef SMALL 1708 if (vflag && !tflag && usize != -1 && gsize != -1) 1709 print_verbage(NULL, NULL, usize, gsize); 1710 #endif 1711 } 1712 1713 /* do what is asked for, for the path name */ 1714 static void 1715 handle_pathname(char *path) 1716 { 1717 char *opath = path, *s = NULL; 1718 ssize_t len; 1719 int slen; 1720 struct stat sb; 1721 1722 /* check for stdout/stdin */ 1723 if (path[0] == '-' && path[1] == '\0') { 1724 if (dflag) 1725 handle_stdin(); 1726 else 1727 handle_stdout(); 1728 return; 1729 } 1730 1731 retry: 1732 if (stat(path, &sb) != 0) { 1733 /* lets try <path>.gz if we're decompressing */ 1734 if (dflag && s == NULL && errno == ENOENT) { 1735 len = strlen(path); 1736 slen = suffixes[0].ziplen; 1737 s = malloc(len + slen + 1); 1738 if (s == NULL) 1739 maybe_err("malloc"); 1740 memcpy(s, path, len); 1741 memcpy(s + len, suffixes[0].zipped, slen + 1); 1742 path = s; 1743 goto retry; 1744 } 1745 maybe_warn("can't stat: %s", opath); 1746 goto out; 1747 } 1748 1749 if (S_ISDIR(sb.st_mode)) { 1750 #ifndef SMALL 1751 if (rflag) 1752 handle_dir(path); 1753 else 1754 #endif 1755 maybe_warnx("%s is a directory", path); 1756 goto out; 1757 } 1758 1759 if (S_ISREG(sb.st_mode)) 1760 handle_file(path, &sb); 1761 else 1762 maybe_warnx("%s is not a regular file", path); 1763 1764 out: 1765 if (s) 1766 free(s); 1767 } 1768 1769 /* compress/decompress a file */ 1770 static void 1771 handle_file(char *file, struct stat *sbp) 1772 { 1773 off_t usize, gsize; 1774 char outfile[PATH_MAX]; 1775 1776 infile = file; 1777 if (dflag) { 1778 usize = file_uncompress(file, outfile, sizeof(outfile)); 1779 #ifndef SMALL 1780 if (vflag && tflag) 1781 print_test(file, usize != -1); 1782 #endif 1783 if (usize == -1) 1784 return; 1785 gsize = sbp->st_size; 1786 } else { 1787 gsize = file_compress(file, outfile, sizeof(outfile)); 1788 if (gsize == -1) 1789 return; 1790 usize = sbp->st_size; 1791 } 1792 1793 1794 #ifndef SMALL 1795 if (vflag && !tflag) 1796 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1797 #endif 1798 } 1799 1800 #ifndef SMALL 1801 /* this is used with -r to recursively descend directories */ 1802 static void 1803 handle_dir(char *dir) 1804 { 1805 char *path_argv[2]; 1806 FTS *fts; 1807 FTSENT *entry; 1808 1809 path_argv[0] = dir; 1810 path_argv[1] = 0; 1811 fts = fts_open(path_argv, FTS_PHYSICAL | FTS_NOCHDIR, NULL); 1812 if (fts == NULL) { 1813 warn("couldn't fts_open %s", dir); 1814 return; 1815 } 1816 1817 while ((entry = fts_read(fts))) { 1818 switch(entry->fts_info) { 1819 case FTS_D: 1820 case FTS_DP: 1821 continue; 1822 1823 case FTS_DNR: 1824 case FTS_ERR: 1825 case FTS_NS: 1826 maybe_warn("%s", entry->fts_path); 1827 continue; 1828 case FTS_F: 1829 handle_file(entry->fts_path, entry->fts_statp); 1830 } 1831 } 1832 (void)fts_close(fts); 1833 } 1834 #endif 1835 1836 /* print a ratio - size reduction as a fraction of uncompressed size */ 1837 static void 1838 print_ratio(off_t in, off_t out, FILE *where) 1839 { 1840 int percent10; /* 10 * percent */ 1841 off_t diff; 1842 char buff[8]; 1843 int len; 1844 1845 diff = in - out/2; 1846 if (diff <= 0) 1847 /* 1848 * Output is more than double size of input! print -99.9% 1849 * Quite possibly we've failed to get the original size. 1850 */ 1851 percent10 = -999; 1852 else { 1853 /* 1854 * We only need 12 bits of result from the final division, 1855 * so reduce the values until a 32bit division will suffice. 1856 */ 1857 while (in > 0x100000) { 1858 diff >>= 1; 1859 in >>= 1; 1860 } 1861 if (in != 0) 1862 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1863 else 1864 percent10 = 0; 1865 } 1866 1867 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1868 /* Move the '.' to before the last digit */ 1869 buff[len - 1] = buff[len - 2]; 1870 buff[len - 2] = '.'; 1871 fprintf(where, "%5s%%", buff); 1872 } 1873 1874 #ifndef SMALL 1875 /* print compression statistics, and the new name (if there is one!) */ 1876 static void 1877 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1878 { 1879 if (file) 1880 fprintf(stderr, "%s:%s ", file, 1881 strlen(file) < 7 ? "\t\t" : "\t"); 1882 print_ratio(usize, gsize, stderr); 1883 if (nfile) 1884 fprintf(stderr, " -- replaced with %s", nfile); 1885 fprintf(stderr, "\n"); 1886 fflush(stderr); 1887 } 1888 1889 /* print test results */ 1890 static void 1891 print_test(const char *file, int ok) 1892 { 1893 1894 if (exit_value == 0 && ok == 0) 1895 exit_value = 1; 1896 fprintf(stderr, "%s:%s %s\n", file, 1897 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1898 fflush(stderr); 1899 } 1900 #endif 1901 1902 /* print a file's info ala --list */ 1903 /* eg: 1904 compressed uncompressed ratio uncompressed_name 1905 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1906 */ 1907 static void 1908 print_list(int fd, off_t out, const char *outfile, time_t ts) 1909 { 1910 static int first = 1; 1911 #ifndef SMALL 1912 static off_t in_tot, out_tot; 1913 uint32_t crc = 0; 1914 #endif 1915 off_t in = 0, rv; 1916 1917 if (first) { 1918 #ifndef SMALL 1919 if (vflag) 1920 printf("method crc date time "); 1921 #endif 1922 if (qflag == 0) 1923 printf(" compressed uncompressed " 1924 "ratio uncompressed_name\n"); 1925 } 1926 first = 0; 1927 1928 /* print totals? */ 1929 #ifndef SMALL 1930 if (fd == -1) { 1931 in = in_tot; 1932 out = out_tot; 1933 } else 1934 #endif 1935 { 1936 /* read the last 4 bytes - this is the uncompressed size */ 1937 rv = lseek(fd, (off_t)(-8), SEEK_END); 1938 if (rv != -1) { 1939 unsigned char buf[8]; 1940 uint32_t usize; 1941 1942 rv = read(fd, (char *)buf, sizeof(buf)); 1943 if (rv == -1) 1944 maybe_warn("read of uncompressed size"); 1945 else if (rv != sizeof(buf)) 1946 maybe_warnx("read of uncompressed size"); 1947 1948 else { 1949 usize = buf[4] | buf[5] << 8 | 1950 buf[6] << 16 | buf[7] << 24; 1951 in = (off_t)usize; 1952 #ifndef SMALL 1953 crc = buf[0] | buf[1] << 8 | 1954 buf[2] << 16 | buf[3] << 24; 1955 #endif 1956 } 1957 } 1958 } 1959 1960 #ifndef SMALL 1961 if (vflag && fd == -1) 1962 printf(" "); 1963 else if (vflag) { 1964 char *date = ctime(&ts); 1965 1966 /* skip the day, 1/100th second, and year */ 1967 date += 4; 1968 date[12] = 0; 1969 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1970 } 1971 in_tot += in; 1972 out_tot += out; 1973 #endif 1974 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1975 print_ratio(in, out, stdout); 1976 printf(" %s\n", outfile); 1977 } 1978 1979 /* display the usage of NetBSD gzip */ 1980 static void 1981 usage(void) 1982 { 1983 1984 fprintf(stderr, "%s\n", gzip_version); 1985 fprintf(stderr, 1986 #ifdef SMALL 1987 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n", 1988 #else 1989 "usage: %s [-123456789acdfhklLNnqrtVv] [-S .suffix] [<file> [<file> ...]]\n" 1990 " -1 --fast fastest (worst) compression\n" 1991 " -2 .. -8 set compression level\n" 1992 " -9 --best best (slowest) compression\n" 1993 " -c --stdout write to stdout, keep original files\n" 1994 " --to-stdout\n" 1995 " -d --decompress uncompress files\n" 1996 " --uncompress\n" 1997 " -f --force force overwriting & compress links\n" 1998 " -h --help display this help\n" 1999 " -k --keep don't delete input files during operation\n" 2000 " -l --list list compressed file contents\n" 2001 " -N --name save or restore original file name and time stamp\n" 2002 " -n --no-name don't save original file name or time stamp\n" 2003 " -q --quiet output no warnings\n" 2004 " -r --recursive recursively compress files in directories\n" 2005 " -S .suf use suffix .suf instead of .gz\n" 2006 " --suffix .suf\n" 2007 " -t --test test compressed file\n" 2008 " -V --version display program version\n" 2009 " -v --verbose print extra statistics\n", 2010 #endif 2011 getprogname()); 2012 exit(0); 2013 } 2014 2015 #ifndef SMALL 2016 /* display the license information of FreeBSD gzip */ 2017 static void 2018 display_license(void) 2019 { 2020 2021 fprintf(stderr, "%s (based on NetBSD gzip 20060927)\n", gzip_version); 2022 fprintf(stderr, "%s\n", gzip_copyright); 2023 exit(0); 2024 } 2025 #endif 2026 2027 /* display the version of NetBSD gzip */ 2028 static void 2029 display_version(void) 2030 { 2031 2032 fprintf(stderr, "%s\n", gzip_version); 2033 exit(0); 2034 } 2035 2036 #ifndef NO_BZIP2_SUPPORT 2037 #include "unbzip2.c" 2038 #endif 2039 #ifndef NO_COMPRESS_SUPPORT 2040 #include "zuncompress.c" 2041 #endif 2042 2043 static ssize_t 2044 read_retry(int fd, void *buf, size_t sz) 2045 { 2046 char *cp = buf; 2047 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2048 2049 while (left > 0) { 2050 ssize_t ret; 2051 2052 ret = read(fd, cp, left); 2053 if (ret == -1) { 2054 return ret; 2055 } else if (ret == 0) { 2056 break; /* EOF */ 2057 } 2058 cp += ret; 2059 left -= ret; 2060 } 2061 2062 return sz - left; 2063 } 2064 2065