gzip.c revision 1.78
1 /* $NetBSD: gzip.c,v 1.78 2005/11/22 09:05:29 mrg Exp $ */ 2 3 /* 4 * Copyright (c) 1997, 1998, 2003, 2004 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 #include <sys/cdefs.h> 32 #ifndef lint 33 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004 Matthew R. Green\n\ 34 All rights reserved.\n"); 35 __RCSID("$NetBSD: gzip.c,v 1.78 2005/11/22 09:05:29 mrg Exp $"); 36 #endif /* not lint */ 37 38 /* 39 * gzip.c -- GPL free gzip using zlib. 40 * 41 * RFC 1950 covers the zlib format 42 * RFC 1951 covers the deflate format 43 * RFC 1952 covers the gzip format 44 * 45 * TODO: 46 * - use mmap where possible 47 * - handle some signals better (remove outfile?) 48 * - make bzip2/compress -v/-t/-l support work as well as possible 49 */ 50 51 #include <sys/param.h> 52 #include <sys/stat.h> 53 #include <sys/time.h> 54 55 #include <inttypes.h> 56 #include <unistd.h> 57 #include <stdio.h> 58 #include <string.h> 59 #include <stdlib.h> 60 #include <err.h> 61 #include <errno.h> 62 #include <fcntl.h> 63 #include <zlib.h> 64 #include <fts.h> 65 #include <libgen.h> 66 #include <stdarg.h> 67 #include <getopt.h> 68 #include <time.h> 69 70 #ifndef PRIdOFF 71 #define PRIdOFF PRId64 72 #endif 73 74 /* what type of file are we dealing with */ 75 enum filetype { 76 FT_GZIP, 77 #ifndef NO_BZIP2_SUPPORT 78 FT_BZIP2, 79 #endif 80 #ifndef NO_COMPRESS_SUPPORT 81 FT_Z, 82 #endif 83 FT_LAST, 84 FT_UNKNOWN 85 }; 86 87 #ifndef NO_BZIP2_SUPPORT 88 #include <bzlib.h> 89 90 #define BZ2_SUFFIX ".bz2" 91 #define BZIP2_MAGIC "\102\132\150" 92 #endif 93 94 #ifndef NO_COMPRESS_SUPPORT 95 #define Z_SUFFIX ".Z" 96 #define Z_MAGIC "\037\235" 97 #endif 98 99 #define GZ_SUFFIX ".gz" 100 101 #define BUFLEN (64 * 1024) 102 103 #define GZIP_MAGIC0 0x1F 104 #define GZIP_MAGIC1 0x8B 105 #define GZIP_OMAGIC1 0x9E 106 107 #define GZIP_TIMESTAMP (off_t)4 108 #define GZIP_ORIGNAME (off_t)10 109 110 #define HEAD_CRC 0x02 111 #define EXTRA_FIELD 0x04 112 #define ORIG_NAME 0x08 113 #define COMMENT 0x10 114 115 #define OS_CODE 3 /* Unix */ 116 117 typedef struct { 118 const char *zipped; 119 int ziplen; 120 const char *normal; /* for unzip - must not be longer than zipped */ 121 } suffixes_t; 122 static suffixes_t suffixes[] = { 123 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 124 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 125 #ifndef SMALL 126 SUFFIX(GZ_SUFFIX, ""), 127 SUFFIX(".z", ""), 128 SUFFIX("-gz", ""), 129 SUFFIX("-z", ""), 130 SUFFIX("_z", ""), 131 SUFFIX(".taz", ".tar"), 132 SUFFIX(".tgz", ".tar"), 133 #ifndef NO_BZIP2_SUPPORT 134 SUFFIX(BZ2_SUFFIX, ""), 135 #endif 136 #ifndef NO_COMPRESS_SUPPORT 137 SUFFIX(Z_SUFFIX, ""), 138 #endif 139 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 140 #endif /* SMALL */ 141 #undef SUFFIX 142 }; 143 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 144 145 static const char gzip_version[] = "NetBSD gzip 20040830"; 146 147 static int cflag; /* stdout mode */ 148 static int dflag; /* decompress mode */ 149 static int lflag; /* list mode */ 150 static int numflag = 6; /* gzip -1..-9 value */ 151 152 #ifndef SMALL 153 static int fflag; /* force mode */ 154 static int nflag; /* don't save name/timestamp */ 155 static int Nflag; /* don't restore name/timestamp */ 156 static int qflag; /* quiet mode */ 157 static int rflag; /* recursive mode */ 158 static int tflag; /* test */ 159 static int vflag; /* verbose mode */ 160 #else 161 #define qflag 0 162 #define tflag 0 163 #endif 164 165 static int exit_value = 0; /* exit value */ 166 167 static char *infile; /* name of file coming in */ 168 169 static void maybe_err(const char *fmt, ...) 170 __attribute__((__format__(__printf__, 1, 2))); 171 #ifndef NO_BZIP2_SUPPORT 172 static void maybe_errx(const char *fmt, ...) 173 __attribute__((__format__(__printf__, 1, 2))); 174 #endif 175 static void maybe_warn(const char *fmt, ...) 176 __attribute__((__format__(__printf__, 1, 2))); 177 static void maybe_warnx(const char *fmt, ...) 178 __attribute__((__format__(__printf__, 1, 2))); 179 static enum filetype file_gettype(u_char *); 180 #ifdef SMALL 181 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 182 #endif 183 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 184 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 185 static off_t file_compress(char *, char *, size_t); 186 static off_t file_uncompress(char *, char *, size_t); 187 static void handle_pathname(char *); 188 static void handle_file(char *, struct stat *); 189 static void handle_stdin(void); 190 static void handle_stdout(void); 191 static void print_ratio(off_t, off_t, FILE *); 192 static void print_list(int fd, off_t, const char *, time_t); 193 static void usage(void); 194 static void display_version(void); 195 static const suffixes_t *check_suffix(char *, int); 196 static ssize_t read_retry(int, void *, size_t); 197 198 #ifdef SMALL 199 #define unlink_input(f, sb) unlink(f) 200 #else 201 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 202 static void prepend_gzip(char *, int *, char ***); 203 static void handle_dir(char *); 204 static void print_verbage(const char *, const char *, off_t, off_t); 205 static void print_test(const char *, int); 206 static void copymodes(int fd, const struct stat *, const char *file); 207 static int check_outfile(const char *outfile); 208 #endif 209 210 #ifndef NO_BZIP2_SUPPORT 211 static off_t unbzip2(int, int, char *, size_t, off_t *); 212 #endif 213 214 #ifndef NO_COMPRESS_SUPPORT 215 static FILE *zdopen(int); 216 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 217 #endif 218 219 int main(int, char *p[]); 220 221 #ifdef SMALL 222 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 223 #else 224 static const struct option longopts[] = { 225 { "stdout", no_argument, 0, 'c' }, 226 { "to-stdout", no_argument, 0, 'c' }, 227 { "decompress", no_argument, 0, 'd' }, 228 { "uncompress", no_argument, 0, 'd' }, 229 { "force", no_argument, 0, 'f' }, 230 { "help", no_argument, 0, 'h' }, 231 { "list", no_argument, 0, 'l' }, 232 { "no-name", no_argument, 0, 'n' }, 233 { "name", no_argument, 0, 'N' }, 234 { "quiet", no_argument, 0, 'q' }, 235 { "recursive", no_argument, 0, 'r' }, 236 { "suffix", required_argument, 0, 'S' }, 237 { "test", no_argument, 0, 't' }, 238 { "verbose", no_argument, 0, 'v' }, 239 { "version", no_argument, 0, 'V' }, 240 { "fast", no_argument, 0, '1' }, 241 { "best", no_argument, 0, '9' }, 242 #if 0 243 /* 244 * This is what else GNU gzip implements. --ascii isn't useful 245 * on NetBSD, and I don't care to have a --license. 246 */ 247 { "ascii", no_argument, 0, 'a' }, 248 { "license", no_argument, 0, 'L' }, 249 #endif 250 { NULL, no_argument, 0, 0 }, 251 }; 252 #endif 253 254 int 255 main(int argc, char **argv) 256 { 257 const char *progname = getprogname(); 258 #ifndef SMALL 259 char *gzip; 260 int len; 261 #endif 262 int ch; 263 264 /* XXX set up signals */ 265 266 #ifndef SMALL 267 if ((gzip = getenv("GZIP")) != NULL) 268 prepend_gzip(gzip, &argc, &argv); 269 #endif 270 271 /* 272 * XXX 273 * handle being called `gunzip', `zcat' and `gzcat' 274 */ 275 if (strcmp(progname, "gunzip") == 0) 276 dflag = 1; 277 else if (strcmp(progname, "zcat") == 0 || 278 strcmp(progname, "gzcat") == 0) 279 dflag = cflag = 1; 280 281 #ifdef SMALL 282 #define OPT_LIST "cdhHltV123456789" 283 #else 284 #define OPT_LIST "cdfhHlnNqrS:tvV123456789" 285 #endif 286 287 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 288 switch (ch) { 289 case 'c': 290 cflag = 1; 291 break; 292 case 'd': 293 dflag = 1; 294 break; 295 case 'l': 296 lflag = 1; 297 dflag = 1; 298 break; 299 case 'V': 300 display_version(); 301 /* NOTREACHED */ 302 case '1': case '2': case '3': 303 case '4': case '5': case '6': 304 case '7': case '8': case '9': 305 numflag = ch - '0'; 306 break; 307 #ifndef SMALL 308 case 'f': 309 fflag = 1; 310 break; 311 case 'n': 312 nflag = 1; 313 Nflag = 0; 314 break; 315 case 'N': 316 nflag = 0; 317 Nflag = 1; 318 break; 319 case 'q': 320 qflag = 1; 321 break; 322 case 'r': 323 rflag = 1; 324 break; 325 case 'S': 326 len = strlen(optarg); 327 if (len != 0) { 328 suffixes[0].zipped = optarg; 329 suffixes[0].ziplen = len; 330 } else { 331 suffixes[NUM_SUFFIXES - 1].zipped = ""; 332 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 333 } 334 break; 335 case 't': 336 cflag = 1; 337 tflag = 1; 338 dflag = 1; 339 break; 340 case 'v': 341 vflag = 1; 342 break; 343 #endif 344 default: 345 usage(); 346 /* NOTREACHED */ 347 } 348 } 349 argv += optind; 350 argc -= optind; 351 352 if (argc == 0) { 353 if (dflag) /* stdin mode */ 354 handle_stdin(); 355 else /* stdout mode */ 356 handle_stdout(); 357 } else { 358 do { 359 handle_pathname(argv[0]); 360 } while (*++argv); 361 } 362 #ifndef SMALL 363 if (qflag == 0 && lflag && argc > 1) 364 print_list(-1, 0, "(totals)", 0); 365 #endif 366 exit(exit_value); 367 } 368 369 /* maybe print a warning */ 370 void 371 maybe_warn(const char *fmt, ...) 372 { 373 va_list ap; 374 375 if (qflag == 0) { 376 va_start(ap, fmt); 377 vwarn(fmt, ap); 378 va_end(ap); 379 } 380 if (exit_value == 0) 381 exit_value = 1; 382 } 383 384 /* ... without an errno. */ 385 void 386 maybe_warnx(const char *fmt, ...) 387 { 388 va_list ap; 389 390 if (qflag == 0) { 391 va_start(ap, fmt); 392 vwarnx(fmt, ap); 393 va_end(ap); 394 } 395 if (exit_value == 0) 396 exit_value = 1; 397 } 398 399 /* maybe print an error */ 400 void 401 maybe_err(const char *fmt, ...) 402 { 403 va_list ap; 404 405 if (qflag == 0) { 406 va_start(ap, fmt); 407 vwarn(fmt, ap); 408 va_end(ap); 409 } 410 exit(2); 411 } 412 413 #ifndef NO_BZIP2_SUPPORT 414 /* ... without an errno. */ 415 void 416 maybe_errx(const char *fmt, ...) 417 { 418 va_list ap; 419 420 if (qflag == 0) { 421 va_start(ap, fmt); 422 vwarnx(fmt, ap); 423 va_end(ap); 424 } 425 exit(2); 426 } 427 #endif 428 429 #ifndef SMALL 430 /* split up $GZIP and prepend it to the argument list */ 431 static void 432 prepend_gzip(char *gzip, int *argc, char ***argv) 433 { 434 char *s, **nargv, **ac; 435 int nenvarg = 0, i; 436 437 /* scan how many arguments there are */ 438 for (s = gzip;;) { 439 while (*s == ' ' || *s == '\t') 440 s++; 441 if (*s == 0) 442 goto count_done; 443 nenvarg++; 444 while (*s != ' ' && *s != '\t') 445 if (*s++ == 0) 446 goto count_done; 447 } 448 count_done: 449 /* punt early */ 450 if (nenvarg == 0) 451 return; 452 453 *argc += nenvarg; 454 ac = *argv; 455 456 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 457 if (nargv == NULL) 458 maybe_err("malloc"); 459 460 /* stash this away */ 461 *argv = nargv; 462 463 /* copy the program name first */ 464 i = 0; 465 nargv[i++] = *(ac++); 466 467 /* take a copy of $GZIP and add it to the array */ 468 s = strdup(gzip); 469 if (s == NULL) 470 maybe_err("strdup"); 471 for (;;) { 472 /* Skip whitespaces. */ 473 while (*s == ' ' || *s == '\t') 474 s++; 475 if (*s == 0) 476 goto copy_done; 477 nargv[i++] = s; 478 /* Find the end of this argument. */ 479 while (*s != ' ' && *s != '\t') 480 if (*s++ == 0) 481 /* Argument followed by NUL. */ 482 goto copy_done; 483 /* Terminate by overwriting ' ' or '\t' with NUL. */ 484 *s++ = 0; 485 } 486 copy_done: 487 488 /* copy the original arguments and a NULL */ 489 while (*ac) 490 nargv[i++] = *(ac++); 491 nargv[i] = NULL; 492 } 493 #endif 494 495 /* compress input to output. Return bytes read, -1 on error */ 496 static off_t 497 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 498 { 499 z_stream z; 500 char *outbufp, *inbufp; 501 off_t in_tot = 0, out_tot = 0; 502 ssize_t in_size; 503 int i, error; 504 uLong crc; 505 #ifdef SMALL 506 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 507 0, 0, 0, 0, 508 0, OS_CODE }; 509 #endif 510 511 outbufp = malloc(BUFLEN); 512 inbufp = malloc(BUFLEN); 513 if (outbufp == NULL || inbufp == NULL) { 514 maybe_err("malloc failed"); 515 goto out; 516 } 517 518 memset(&z, 0, sizeof z); 519 z.zalloc = Z_NULL; 520 z.zfree = Z_NULL; 521 z.opaque = 0; 522 523 #ifdef SMALL 524 memcpy(outbufp, header, sizeof header); 525 i = sizeof header; 526 #else 527 if (nflag != 0) { 528 mtime = 0; 529 origname = ""; 530 } 531 532 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 533 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 534 *origname ? ORIG_NAME : 0, 535 mtime & 0xff, 536 (mtime >> 8) & 0xff, 537 (mtime >> 16) & 0xff, 538 (mtime >> 24) & 0xff, 539 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 540 OS_CODE, origname); 541 if (i >= BUFLEN) 542 /* this need PATH_MAX > BUFLEN ... */ 543 maybe_err("snprintf"); 544 if (*origname) 545 i++; 546 #endif 547 548 z.next_out = outbufp + i; 549 z.avail_out = BUFLEN - i; 550 551 error = deflateInit2(&z, numflag, Z_DEFLATED, 552 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 553 if (error != Z_OK) { 554 maybe_warnx("deflateInit2 failed"); 555 in_tot = -1; 556 goto out; 557 } 558 559 crc = crc32(0L, Z_NULL, 0); 560 for (;;) { 561 if (z.avail_out == 0) { 562 if (write(out, outbufp, BUFLEN) != BUFLEN) { 563 maybe_warn("write"); 564 in_tot = -1; 565 goto out; 566 } 567 568 out_tot += BUFLEN; 569 z.next_out = outbufp; 570 z.avail_out = BUFLEN; 571 } 572 573 if (z.avail_in == 0) { 574 in_size = read(in, inbufp, BUFLEN); 575 if (in_size < 0) { 576 maybe_warn("read"); 577 in_tot = -1; 578 goto out; 579 } 580 if (in_size == 0) 581 break; 582 583 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 584 in_tot += in_size; 585 z.next_in = inbufp; 586 z.avail_in = in_size; 587 } 588 589 error = deflate(&z, Z_NO_FLUSH); 590 if (error != Z_OK && error != Z_STREAM_END) { 591 maybe_warnx("deflate failed"); 592 in_tot = -1; 593 goto out; 594 } 595 } 596 597 /* clean up */ 598 for (;;) { 599 size_t len; 600 ssize_t w; 601 602 error = deflate(&z, Z_FINISH); 603 if (error != Z_OK && error != Z_STREAM_END) { 604 maybe_warnx("deflate failed"); 605 in_tot = -1; 606 goto out; 607 } 608 609 len = (char *)z.next_out - outbufp; 610 611 w = write(out, outbufp, len); 612 if (w == -1 || (size_t)w != len) { 613 maybe_warn("write"); 614 out_tot = -1; 615 goto out; 616 } 617 out_tot += len; 618 z.next_out = outbufp; 619 z.avail_out = BUFLEN; 620 621 if (error == Z_STREAM_END) 622 break; 623 } 624 625 if (deflateEnd(&z) != Z_OK) { 626 maybe_warnx("deflateEnd failed"); 627 in_tot = -1; 628 goto out; 629 } 630 631 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 632 (int)crc & 0xff, 633 (int)(crc >> 8) & 0xff, 634 (int)(crc >> 16) & 0xff, 635 (int)(crc >> 24) & 0xff, 636 (int)in_tot & 0xff, 637 (int)(in_tot >> 8) & 0xff, 638 (int)(in_tot >> 16) & 0xff, 639 (int)(in_tot >> 24) & 0xff); 640 if (i != 8) 641 maybe_err("snprintf"); 642 if (in_tot > 0xffffffff) 643 maybe_warn("input file size >= 4GB cannot be saved"); 644 if (write(out, outbufp, i) != i) { 645 maybe_warn("write"); 646 in_tot = -1; 647 } else 648 out_tot += i; 649 650 out: 651 if (inbufp != NULL) 652 free(inbufp); 653 if (outbufp != NULL) 654 free(outbufp); 655 if (gsizep) 656 *gsizep = out_tot; 657 return in_tot; 658 } 659 660 /* 661 * uncompress input to output then close the input. return the 662 * uncompressed size written, and put the compressed sized read 663 * into `*gsizep'. 664 */ 665 static off_t 666 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 667 const char *filename) 668 { 669 z_stream z; 670 char *outbufp, *inbufp; 671 off_t out_tot = -1, in_tot = 0; 672 uint32_t out_sub_tot = 0; 673 enum { 674 GZSTATE_MAGIC0, 675 GZSTATE_MAGIC1, 676 GZSTATE_METHOD, 677 GZSTATE_FLAGS, 678 GZSTATE_SKIPPING, 679 GZSTATE_EXTRA, 680 GZSTATE_EXTRA2, 681 GZSTATE_EXTRA3, 682 GZSTATE_ORIGNAME, 683 GZSTATE_COMMENT, 684 GZSTATE_HEAD_CRC1, 685 GZSTATE_HEAD_CRC2, 686 GZSTATE_INIT, 687 GZSTATE_READ, 688 GZSTATE_CRC, 689 GZSTATE_LEN, 690 } state = GZSTATE_MAGIC0; 691 int flags = 0, skip_count = 0; 692 int error = Z_STREAM_ERROR, done_reading = 0; 693 uLong crc = 0; 694 ssize_t wr; 695 int needmore = 0; 696 697 #define ADVANCE() { z.next_in++; z.avail_in--; } 698 699 if ((outbufp = malloc(BUFLEN)) == NULL) { 700 maybe_err("malloc failed"); 701 goto out2; 702 } 703 if ((inbufp = malloc(BUFLEN)) == NULL) { 704 maybe_err("malloc failed"); 705 goto out1; 706 } 707 708 memset(&z, 0, sizeof z); 709 z.avail_in = prelen; 710 z.next_in = pre; 711 z.avail_out = BUFLEN; 712 z.next_out = outbufp; 713 z.zalloc = NULL; 714 z.zfree = NULL; 715 z.opaque = 0; 716 717 in_tot = prelen; 718 out_tot = 0; 719 720 for (;;) { 721 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 722 ssize_t in_size; 723 724 if (z.avail_in > 0) { 725 memmove(inbufp, z.next_in, z.avail_in); 726 } 727 z.next_in = inbufp; 728 in_size = read(in, z.next_in + z.avail_in, 729 BUFLEN - z.avail_in); 730 731 if (in_size == -1) { 732 maybe_warn("failed to read stdin"); 733 goto stop_and_fail; 734 } else if (in_size == 0) { 735 done_reading = 1; 736 } 737 738 z.avail_in += in_size; 739 needmore = 0; 740 741 in_tot += in_size; 742 } 743 if (z.avail_in == 0) { 744 if (done_reading && state != GZSTATE_MAGIC0) 745 maybe_warnx("%s: unexpected end of file", 746 filename); 747 goto stop; 748 } 749 switch (state) { 750 case GZSTATE_MAGIC0: 751 if (*z.next_in != GZIP_MAGIC0) { 752 maybe_warnx("input not gziped (MAGIC0)"); 753 goto stop_and_fail; 754 } 755 ADVANCE(); 756 state++; 757 out_sub_tot = 0; 758 crc = crc32(0L, Z_NULL, 0); 759 break; 760 761 case GZSTATE_MAGIC1: 762 if (*z.next_in != GZIP_MAGIC1 && 763 *z.next_in != GZIP_OMAGIC1) { 764 maybe_warnx("input not gziped (MAGIC1)"); 765 goto stop_and_fail; 766 } 767 ADVANCE(); 768 state++; 769 break; 770 771 case GZSTATE_METHOD: 772 if (*z.next_in != Z_DEFLATED) { 773 maybe_warnx("unknown compression method"); 774 goto stop_and_fail; 775 } 776 ADVANCE(); 777 state++; 778 break; 779 780 case GZSTATE_FLAGS: 781 flags = *z.next_in; 782 ADVANCE(); 783 skip_count = 6; 784 state++; 785 break; 786 787 case GZSTATE_SKIPPING: 788 if (skip_count > 0) { 789 skip_count--; 790 ADVANCE(); 791 } else 792 state++; 793 break; 794 795 case GZSTATE_EXTRA: 796 if ((flags & EXTRA_FIELD) == 0) { 797 state = GZSTATE_ORIGNAME; 798 break; 799 } 800 skip_count = *z.next_in; 801 ADVANCE(); 802 state++; 803 break; 804 805 case GZSTATE_EXTRA2: 806 skip_count |= ((*z.next_in) << 8); 807 ADVANCE(); 808 state++; 809 break; 810 811 case GZSTATE_EXTRA3: 812 if (skip_count > 0) { 813 skip_count--; 814 ADVANCE(); 815 } else 816 state++; 817 break; 818 819 case GZSTATE_ORIGNAME: 820 if ((flags & ORIG_NAME) == 0) { 821 state++; 822 break; 823 } 824 if (*z.next_in == 0) 825 state++; 826 ADVANCE(); 827 break; 828 829 case GZSTATE_COMMENT: 830 if ((flags & COMMENT) == 0) { 831 state++; 832 break; 833 } 834 if (*z.next_in == 0) 835 state++; 836 ADVANCE(); 837 break; 838 839 case GZSTATE_HEAD_CRC1: 840 if (flags & HEAD_CRC) 841 skip_count = 2; 842 else 843 skip_count = 0; 844 state++; 845 break; 846 847 case GZSTATE_HEAD_CRC2: 848 if (skip_count > 0) { 849 skip_count--; 850 ADVANCE(); 851 } else 852 state++; 853 break; 854 855 case GZSTATE_INIT: 856 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 857 maybe_warnx("failed to inflateInit"); 858 goto stop_and_fail; 859 } 860 state++; 861 break; 862 863 case GZSTATE_READ: 864 error = inflate(&z, Z_FINISH); 865 switch (error) { 866 /* Z_BUF_ERROR goes with Z_FINISH... */ 867 case Z_BUF_ERROR: 868 case Z_STREAM_END: 869 case Z_OK: 870 break; 871 872 case Z_NEED_DICT: 873 maybe_warnx("Z_NEED_DICT error"); 874 goto stop_and_fail; 875 case Z_DATA_ERROR: 876 maybe_warnx("data stream error"); 877 goto stop_and_fail; 878 case Z_STREAM_ERROR: 879 maybe_warnx("internal stream error"); 880 goto stop_and_fail; 881 case Z_MEM_ERROR: 882 maybe_warnx("memory allocation error"); 883 goto stop_and_fail; 884 885 default: 886 maybe_warn("unknown error from inflate(): %d", 887 error); 888 } 889 wr = BUFLEN - z.avail_out; 890 891 if (wr != 0) { 892 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 893 if ( 894 #ifndef SMALL 895 /* don't write anything with -t */ 896 tflag == 0 && 897 #endif 898 write(out, outbufp, wr) != wr) { 899 maybe_warn("error writing to output"); 900 goto stop_and_fail; 901 } 902 903 out_tot += wr; 904 out_sub_tot += wr; 905 } 906 907 if (error == Z_STREAM_END) { 908 inflateEnd(&z); 909 state++; 910 } 911 912 z.next_out = outbufp; 913 z.avail_out = BUFLEN; 914 915 break; 916 case GZSTATE_CRC: 917 { 918 uLong origcrc; 919 920 if (z.avail_in < 4) { 921 if (!done_reading) { 922 needmore = 1; 923 continue; 924 } 925 maybe_warnx("truncated input"); 926 goto stop_and_fail; 927 } 928 origcrc = ((unsigned)z.next_in[0] & 0xff) | 929 ((unsigned)z.next_in[1] & 0xff) << 8 | 930 ((unsigned)z.next_in[2] & 0xff) << 16 | 931 ((unsigned)z.next_in[3] & 0xff) << 24; 932 if (origcrc != crc) { 933 maybe_warnx("invalid compressed" 934 " data--crc error"); 935 goto stop_and_fail; 936 } 937 } 938 939 z.avail_in -= 4; 940 z.next_in += 4; 941 942 if (!z.avail_in && done_reading) { 943 goto stop; 944 } 945 state++; 946 break; 947 case GZSTATE_LEN: 948 { 949 uLong origlen; 950 951 if (z.avail_in < 4) { 952 if (!done_reading) { 953 needmore = 1; 954 continue; 955 } 956 maybe_warnx("truncated input"); 957 goto stop_and_fail; 958 } 959 origlen = ((unsigned)z.next_in[0] & 0xff) | 960 ((unsigned)z.next_in[1] & 0xff) << 8 | 961 ((unsigned)z.next_in[2] & 0xff) << 16 | 962 ((unsigned)z.next_in[3] & 0xff) << 24; 963 964 if (origlen != out_sub_tot) { 965 maybe_warnx("invalid compressed" 966 " data--length error"); 967 goto stop_and_fail; 968 } 969 } 970 971 z.avail_in -= 4; 972 z.next_in += 4; 973 974 if (error < 0) { 975 maybe_warnx("decompression error"); 976 goto stop_and_fail; 977 } 978 state = GZSTATE_MAGIC0; 979 break; 980 } 981 continue; 982 stop_and_fail: 983 out_tot = -1; 984 stop: 985 break; 986 } 987 if (state > GZSTATE_INIT) 988 inflateEnd(&z); 989 990 free(inbufp); 991 out1: 992 free(outbufp); 993 out2: 994 if (gsizep) 995 *gsizep = in_tot; 996 return (out_tot); 997 } 998 999 #ifndef SMALL 1000 /* 1001 * set the owner, mode, flags & utimes using the given file descriptor. 1002 * file is only used in possible warning messages. 1003 */ 1004 static void 1005 copymodes(int fd, const struct stat *sbp, const char *file) 1006 { 1007 struct timeval times[2]; 1008 struct stat sb; 1009 1010 /* 1011 * If we have no info on the input, give this file some 1012 * default values and return.. 1013 */ 1014 if (sbp == NULL) { 1015 mode_t mask = umask(022); 1016 1017 (void)fchmod(fd, DEFFILEMODE & ~mask); 1018 (void)umask(mask); 1019 return; 1020 } 1021 sb = *sbp; 1022 1023 /* if the chown fails, remove set-id bits as-per compress(1) */ 1024 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1025 if (errno != EPERM) 1026 maybe_warn("couldn't fchown: %s", file); 1027 sb.st_mode &= ~(S_ISUID|S_ISGID); 1028 } 1029 1030 /* we only allow set-id and the 9 normal permission bits */ 1031 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1032 if (fchmod(fd, sb.st_mode) < 0) 1033 maybe_warn("couldn't fchmod: %s", file); 1034 1035 /* only try flags if they exist already */ 1036 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1037 maybe_warn("couldn't fchflags: %s", file); 1038 1039 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1040 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1041 if (futimes(fd, times) < 0) 1042 maybe_warn("couldn't utimes: %s", file); 1043 } 1044 #endif 1045 1046 /* what sort of file is this? */ 1047 static enum filetype 1048 file_gettype(u_char *buf) 1049 { 1050 1051 if (buf[0] == GZIP_MAGIC0 && 1052 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1053 return FT_GZIP; 1054 else 1055 #ifndef NO_BZIP2_SUPPORT 1056 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1057 buf[3] >= '0' && buf[3] <= '9') 1058 return FT_BZIP2; 1059 else 1060 #endif 1061 #ifndef NO_COMPRESS_SUPPORT 1062 if (memcmp(buf, Z_MAGIC, 2) == 0) 1063 return FT_Z; 1064 else 1065 #endif 1066 return FT_UNKNOWN; 1067 } 1068 1069 #ifndef SMALL 1070 /* check the outfile is OK. */ 1071 static int 1072 check_outfile(const char *outfile) 1073 { 1074 struct stat sb; 1075 int ok = 1; 1076 1077 if (lflag == 0 && stat(outfile, &sb) == 0) { 1078 if (fflag) 1079 unlink(outfile); 1080 else if (isatty(STDIN_FILENO)) { 1081 char ans[10] = { 'n', '\0' }; /* default */ 1082 1083 fprintf(stderr, "%s already exists -- do you wish to " 1084 "overwrite (y or n)? " , outfile); 1085 (void)fgets(ans, sizeof(ans) - 1, stdin); 1086 if (ans[0] != 'y' && ans[0] != 'Y') { 1087 fprintf(stderr, "\tnot overwritting\n"); 1088 ok = 0; 1089 } else 1090 unlink(outfile); 1091 } else { 1092 maybe_warnx("%s already exists -- skipping", outfile); 1093 ok = 0; 1094 } 1095 } 1096 return ok; 1097 } 1098 1099 static void 1100 unlink_input(const char *file, const struct stat *sb) 1101 { 1102 struct stat nsb; 1103 1104 if (stat(file, &nsb) != 0) 1105 /* Must be gone alrady */ 1106 return; 1107 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1108 /* Definitely a different file */ 1109 return; 1110 unlink(file); 1111 } 1112 #endif 1113 1114 static const suffixes_t * 1115 check_suffix(char *file, int xlate) 1116 { 1117 const suffixes_t *s; 1118 int len = strlen(file); 1119 char *sp; 1120 1121 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1122 /* if it doesn't fit in "a.suf", don't bother */ 1123 if (s->ziplen >= len) 1124 continue; 1125 sp = file + len - s->ziplen; 1126 if (strcmp(s->zipped, sp) != 0) 1127 continue; 1128 if (xlate) 1129 strcpy(sp, s->normal); 1130 return s; 1131 } 1132 return NULL; 1133 } 1134 1135 /* 1136 * compress the given file: create a corresponding .gz file and remove the 1137 * original. 1138 */ 1139 static off_t 1140 file_compress(char *file, char *outfile, size_t outsize) 1141 { 1142 int in; 1143 int out; 1144 off_t size, insize; 1145 #ifndef SMALL 1146 struct stat isb, osb; 1147 const suffixes_t *suff; 1148 #endif 1149 1150 in = open(file, O_RDONLY); 1151 if (in == -1) { 1152 maybe_warn("can't open %s", file); 1153 return -1; 1154 } 1155 1156 if (cflag == 0) { 1157 #ifndef SMALL 1158 if (fstat(in, &isb) == 0) { 1159 if (isb.st_nlink > 1 && fflag == 0) { 1160 maybe_warnx("%s has %d other link%s -- " 1161 "skipping", file, isb.st_nlink - 1, 1162 isb.st_nlink == 1 ? "" : "s"); 1163 close(in); 1164 return -1; 1165 } 1166 } 1167 1168 if (fflag == 0 && (suff = check_suffix(file, 0)) 1169 && suff->zipped[0] != 0) { 1170 maybe_warnx("%s already has %s suffix -- unchanged", 1171 file, suff->zipped); 1172 close(in); 1173 return -1; 1174 } 1175 #endif 1176 1177 /* Add (usually) .gz to filename */ 1178 if ((size_t)snprintf(outfile, outsize, "%s%s", 1179 file, suffixes[0].zipped) >= outsize) 1180 memcpy(outfile - suffixes[0].ziplen - 1, 1181 suffixes[0].zipped, suffixes[0].ziplen + 1); 1182 1183 #ifndef SMALL 1184 if (check_outfile(outfile) == 0) { 1185 close(in); 1186 return -1; 1187 } 1188 #endif 1189 } 1190 1191 if (cflag == 0) { 1192 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1193 if (out == -1) { 1194 maybe_warn("could not create output: %s", outfile); 1195 fclose(stdin); 1196 return -1; 1197 } 1198 } else 1199 out = STDOUT_FILENO; 1200 1201 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1202 1203 (void)close(in); 1204 1205 /* 1206 * If there was an error, insize will be -1. 1207 * If we compressed to stdout, just return the size. 1208 * Otherwise stat the file and check it is the correct size. 1209 * We only blow away the file if we can stat the output and it 1210 * has the expected size. 1211 */ 1212 if (cflag != 0) 1213 return insize == -1 ? -1 : size; 1214 1215 #ifndef SMALL 1216 if (fstat(out, &osb) != 0) { 1217 maybe_warn("couldn't stat: %s", outfile); 1218 goto bad_outfile; 1219 } 1220 1221 if (osb.st_size != size) { 1222 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1223 " != %" PRIdOFF "), deleting", 1224 outfile, osb.st_size, size); 1225 goto bad_outfile; 1226 } 1227 1228 copymodes(out, &isb, outfile); 1229 #endif 1230 if (close(out) == -1) 1231 maybe_warn("couldn't close output"); 1232 1233 /* output is good, ok to delete input */ 1234 unlink_input(file, &isb); 1235 return size; 1236 1237 #ifndef SMALL 1238 bad_outfile: 1239 if (close(out) == -1) 1240 maybe_warn("couldn't close output"); 1241 1242 maybe_warnx("leaving original %s", file); 1243 unlink(outfile); 1244 return size; 1245 #endif 1246 } 1247 1248 /* uncompress the given file and remove the original */ 1249 static off_t 1250 file_uncompress(char *file, char *outfile, size_t outsize) 1251 { 1252 struct stat isb, osb; 1253 off_t size; 1254 ssize_t rbytes; 1255 unsigned char header1[4]; 1256 enum filetype method; 1257 int fd, ofd, zfd = -1; 1258 #ifndef SMALL 1259 time_t timestamp = 0; 1260 unsigned char name[PATH_MAX + 1]; 1261 #endif 1262 1263 /* gather the old name info */ 1264 1265 fd = open(file, O_RDONLY); 1266 if (fd < 0) { 1267 maybe_warn("can't open %s", file); 1268 goto lose; 1269 } 1270 1271 strlcpy(outfile, file, outsize); 1272 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1273 maybe_warnx("%s: unknown suffix -- ignored", file); 1274 goto lose; 1275 } 1276 1277 rbytes = read(fd, header1, sizeof header1); 1278 if (rbytes != sizeof header1) { 1279 /* we don't want to fail here. */ 1280 #ifndef SMALL 1281 if (fflag) 1282 goto lose; 1283 #endif 1284 if (rbytes == -1) 1285 maybe_warn("can't read %s", file); 1286 else 1287 maybe_warnx("%s: unexpected end of file", file); 1288 goto lose; 1289 } 1290 1291 method = file_gettype(header1); 1292 1293 #ifndef SMALL 1294 if (fflag == 0 && method == FT_UNKNOWN) { 1295 maybe_warnx("%s: not in gzip format", file); 1296 goto lose; 1297 } 1298 1299 #endif 1300 1301 #ifndef SMALL 1302 if (method == FT_GZIP && Nflag) { 1303 unsigned char ts[4]; /* timestamp */ 1304 1305 if (pread(fd, ts, sizeof ts, GZIP_TIMESTAMP) != sizeof ts) { 1306 if (!fflag) 1307 maybe_warn("can't read %s", file); 1308 goto lose; 1309 } 1310 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1311 1312 if (header1[3] & ORIG_NAME) { 1313 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1314 if (rbytes < 0) { 1315 maybe_warn("can't read %s", file); 1316 goto lose; 1317 } 1318 if (name[0] != 0) { 1319 /* preserve original directory name */ 1320 char *dp = strrchr(file, '/'); 1321 if (dp == NULL) 1322 dp = file; 1323 else 1324 dp++; 1325 snprintf(outfile, outsize, "%.*s%.*s", 1326 (int) (dp - file), 1327 file, (int) rbytes, name); 1328 } 1329 } 1330 } 1331 #endif 1332 lseek(fd, 0, SEEK_SET); 1333 1334 if (cflag == 0 || lflag) { 1335 if (fstat(fd, &isb) != 0) 1336 goto lose; 1337 #ifndef SMALL 1338 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1339 maybe_warnx("%s has %d other links -- skipping", 1340 file, isb.st_nlink - 1); 1341 goto lose; 1342 } 1343 if (nflag == 0 && timestamp) 1344 isb.st_mtime = timestamp; 1345 if (check_outfile(outfile) == 0) 1346 goto lose; 1347 #endif 1348 } 1349 1350 if (cflag == 0 && lflag == 0) { 1351 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1352 if (zfd == STDOUT_FILENO) { 1353 /* We won't close STDOUT_FILENO later... */ 1354 zfd = dup(zfd); 1355 close(STDOUT_FILENO); 1356 } 1357 if (zfd == -1) { 1358 maybe_warn("can't open %s", outfile); 1359 goto lose; 1360 } 1361 } else 1362 zfd = STDOUT_FILENO; 1363 1364 #ifndef NO_BZIP2_SUPPORT 1365 if (method == FT_BZIP2) { 1366 1367 /* XXX */ 1368 if (lflag) { 1369 maybe_warnx("no -l with bzip2 files"); 1370 goto lose; 1371 } 1372 1373 size = unbzip2(fd, zfd, NULL, 0, NULL); 1374 } else 1375 #endif 1376 1377 #ifndef NO_COMPRESS_SUPPORT 1378 if (method == FT_Z) { 1379 FILE *in, *out; 1380 1381 /* XXX */ 1382 if (lflag) { 1383 maybe_warnx("no -l with Lempel-Ziv files"); 1384 goto lose; 1385 } 1386 1387 if ((in = zdopen(fd)) == NULL) { 1388 maybe_warn("zdopen for read: %s", file); 1389 goto lose; 1390 } 1391 1392 out = fdopen(dup(zfd), "w"); 1393 if (out == NULL) { 1394 maybe_warn("fdopen for write: %s", outfile); 1395 fclose(in); 1396 goto lose; 1397 } 1398 1399 size = zuncompress(in, out, NULL, 0, NULL); 1400 /* need to fclose() if ferror() is true... */ 1401 if (ferror(in) | fclose(in)) { 1402 maybe_warn("failed infile fclose"); 1403 unlink(outfile); 1404 (void)fclose(out); 1405 } 1406 if (fclose(out) != 0) { 1407 maybe_warn("failed outfile fclose"); 1408 unlink(outfile); 1409 goto lose; 1410 } 1411 } else 1412 #endif 1413 1414 #ifndef SMALL 1415 if (method == FT_UNKNOWN) { 1416 if (lflag) { 1417 maybe_warnx("no -l for unknown filetypes"); 1418 goto lose; 1419 } 1420 size = cat_fd(NULL, 0, NULL, fd); 1421 } else 1422 #endif 1423 { 1424 if (lflag) { 1425 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1426 close(fd); 1427 return -1; /* XXX */ 1428 } 1429 1430 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1431 } 1432 1433 if (close(fd) != 0) 1434 maybe_warn("couldn't close input"); 1435 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1436 maybe_warn("couldn't close output"); 1437 1438 if (size == -1) { 1439 if (cflag == 0) 1440 unlink(outfile); 1441 maybe_warnx("%s: uncompress failed", file); 1442 return -1; 1443 } 1444 1445 /* if testing, or we uncompressed to stdout, this is all we need */ 1446 #ifndef SMALL 1447 if (tflag) 1448 return size; 1449 #endif 1450 /* if we are uncompressing to stdin, don't remove the file. */ 1451 if (cflag) 1452 return size; 1453 1454 /* 1455 * if we create a file... 1456 */ 1457 /* 1458 * if we can't stat the file don't remove the file. 1459 */ 1460 1461 ofd = open(outfile, O_RDWR, 0); 1462 if (ofd == -1) { 1463 maybe_warn("couldn't open (leaving original): %s", 1464 outfile); 1465 return -1; 1466 } 1467 if (fstat(ofd, &osb) != 0) { 1468 maybe_warn("couldn't stat (leaving original): %s", 1469 outfile); 1470 close(ofd); 1471 return -1; 1472 } 1473 if (osb.st_size != size) { 1474 maybe_warnx("stat gave different size: %" PRIdOFF 1475 " != %" PRIdOFF " (leaving original)", 1476 size, osb.st_size); 1477 close(ofd); 1478 unlink(outfile); 1479 return -1; 1480 } 1481 unlink_input(file, &isb); 1482 #ifndef SMALL 1483 copymodes(ofd, &isb, outfile); 1484 #endif 1485 close(ofd); 1486 return size; 1487 1488 lose: 1489 if (fd != -1) 1490 close(fd); 1491 if (zfd != -1 && zfd != STDOUT_FILENO) 1492 close(fd); 1493 return -1; 1494 } 1495 1496 #ifndef SMALL 1497 static off_t 1498 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1499 { 1500 char buf[BUFLEN]; 1501 off_t in_tot; 1502 ssize_t w; 1503 1504 in_tot = count; 1505 w = write(STDOUT_FILENO, prepend, count); 1506 if (w == -1 || (size_t)w != count) { 1507 maybe_warn("write to stdout"); 1508 return -1; 1509 } 1510 for (;;) { 1511 ssize_t rv; 1512 1513 rv = read(fd, buf, sizeof buf); 1514 if (rv == 0) 1515 break; 1516 if (rv < 0) { 1517 maybe_warn("read from fd %d", fd); 1518 break; 1519 } 1520 1521 if (write(STDOUT_FILENO, buf, rv) != rv) { 1522 maybe_warn("write to stdout"); 1523 break; 1524 } 1525 in_tot += rv; 1526 } 1527 1528 if (gsizep) 1529 *gsizep = in_tot; 1530 return (in_tot); 1531 } 1532 #endif 1533 1534 static void 1535 handle_stdin(void) 1536 { 1537 unsigned char header1[4]; 1538 off_t usize, gsize; 1539 enum filetype method; 1540 ssize_t bytes_read; 1541 #ifndef NO_COMPRESS_SUPPORT 1542 FILE *in; 1543 #endif 1544 1545 #ifndef SMALL 1546 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1547 maybe_warnx("standard input is a terminal -- ignoring"); 1548 return; 1549 } 1550 #endif 1551 1552 if (lflag) { 1553 struct stat isb; 1554 1555 /* XXX could read the whole file, etc. */ 1556 if (fstat(STDIN_FILENO, &isb) < 0) { 1557 maybe_warn("fstat"); 1558 return; 1559 } 1560 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1561 return; 1562 } 1563 1564 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1565 if (bytes_read == -1) { 1566 maybe_warn("can't read stdin"); 1567 return; 1568 } else if (bytes_read != sizeof(header1)) { 1569 maybe_warnx("unexpected EOF"); 1570 return; 1571 } 1572 1573 method = file_gettype(header1); 1574 switch (method) { 1575 default: 1576 #ifndef SMALL 1577 if (fflag == 0) { 1578 maybe_warnx("unknown compression format"); 1579 return; 1580 } 1581 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1582 break; 1583 #endif 1584 case FT_GZIP: 1585 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1586 header1, sizeof header1, &gsize, "(stdin)"); 1587 break; 1588 #ifndef NO_BZIP2_SUPPORT 1589 case FT_BZIP2: 1590 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1591 header1, sizeof header1, &gsize); 1592 break; 1593 #endif 1594 #ifndef NO_COMPRESS_SUPPORT 1595 case FT_Z: 1596 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1597 maybe_warnx("zopen of stdin"); 1598 return; 1599 } 1600 1601 usize = zuncompress(in, stdout, header1, sizeof header1, &gsize); 1602 fclose(in); 1603 break; 1604 #endif 1605 } 1606 1607 #ifndef SMALL 1608 if (vflag && !tflag && usize != -1 && gsize != -1) 1609 print_verbage(NULL, NULL, usize, gsize); 1610 if (vflag && tflag) 1611 print_test("(stdin)", usize != -1); 1612 #endif 1613 1614 } 1615 1616 static void 1617 handle_stdout(void) 1618 { 1619 off_t gsize, usize; 1620 struct stat sb; 1621 time_t systime; 1622 uint32_t mtime; 1623 int ret; 1624 1625 #ifndef SMALL 1626 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1627 maybe_warnx("standard output is a terminal -- ignoring"); 1628 return; 1629 } 1630 #endif 1631 /* If stdin is a file use it's mtime, otherwise use current time */ 1632 ret = fstat(STDIN_FILENO, &sb); 1633 1634 #ifndef SMALL 1635 if (ret < 0) { 1636 maybe_warn("Can't stat stdin"); 1637 return; 1638 } 1639 #endif 1640 1641 if (S_ISREG(sb.st_mode)) 1642 mtime = (uint32_t)sb.st_mtime; 1643 else { 1644 systime = time(NULL); 1645 #ifndef SMALL 1646 if (systime == -1) { 1647 maybe_warn("time"); 1648 return; 1649 } 1650 #endif 1651 mtime = (uint32_t)systime; 1652 } 1653 1654 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1655 #ifndef SMALL 1656 if (vflag && !tflag && usize != -1 && gsize != -1) 1657 print_verbage(NULL, NULL, usize, gsize); 1658 #endif 1659 } 1660 1661 /* do what is asked for, for the path name */ 1662 static void 1663 handle_pathname(char *path) 1664 { 1665 char *opath = path, *s = NULL; 1666 ssize_t len; 1667 int slen; 1668 struct stat sb; 1669 1670 /* check for stdout/stdin */ 1671 if (path[0] == '-' && path[1] == '\0') { 1672 if (dflag) 1673 handle_stdin(); 1674 else 1675 handle_stdout(); 1676 return; 1677 } 1678 1679 retry: 1680 if (stat(path, &sb) != 0) { 1681 /* lets try <path>.gz if we're decompressing */ 1682 if (dflag && s == NULL && errno == ENOENT) { 1683 len = strlen(path); 1684 slen = suffixes[0].ziplen; 1685 s = malloc(len + slen + 1); 1686 if (s == NULL) 1687 maybe_err("malloc"); 1688 memcpy(s, path, len); 1689 memcpy(s + len, suffixes[0].zipped, slen + 1); 1690 path = s; 1691 goto retry; 1692 } 1693 maybe_warn("can't stat: %s", opath); 1694 goto out; 1695 } 1696 1697 if (S_ISDIR(sb.st_mode)) { 1698 #ifndef SMALL 1699 if (rflag) 1700 handle_dir(path); 1701 else 1702 #endif 1703 maybe_warnx("%s is a directory", path); 1704 goto out; 1705 } 1706 1707 if (S_ISREG(sb.st_mode)) 1708 handle_file(path, &sb); 1709 else 1710 maybe_warnx("%s is not a regular file", path); 1711 1712 out: 1713 if (s) 1714 free(s); 1715 } 1716 1717 /* compress/decompress a file */ 1718 static void 1719 handle_file(char *file, struct stat *sbp) 1720 { 1721 off_t usize, gsize; 1722 char outfile[PATH_MAX]; 1723 1724 infile = file; 1725 if (dflag) { 1726 usize = file_uncompress(file, outfile, sizeof(outfile)); 1727 if (vflag && tflag) 1728 print_test(file, usize != -1); 1729 if (usize == -1) 1730 return; 1731 gsize = sbp->st_size; 1732 } else { 1733 gsize = file_compress(file, outfile, sizeof(outfile)); 1734 if (gsize == -1) 1735 return; 1736 usize = sbp->st_size; 1737 } 1738 1739 1740 #ifndef SMALL 1741 if (vflag && !tflag) 1742 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1743 #endif 1744 } 1745 1746 #ifndef SMALL 1747 /* this is used with -r to recursively descend directories */ 1748 static void 1749 handle_dir(char *dir) 1750 { 1751 char *path_argv[2]; 1752 FTS *fts; 1753 FTSENT *entry; 1754 1755 path_argv[0] = dir; 1756 path_argv[1] = 0; 1757 fts = fts_open(path_argv, FTS_PHYSICAL, NULL); 1758 if (fts == NULL) { 1759 warn("couldn't fts_open %s", dir); 1760 return; 1761 } 1762 1763 while ((entry = fts_read(fts))) { 1764 switch(entry->fts_info) { 1765 case FTS_D: 1766 case FTS_DP: 1767 continue; 1768 1769 case FTS_DNR: 1770 case FTS_ERR: 1771 case FTS_NS: 1772 maybe_warn("%s", entry->fts_path); 1773 continue; 1774 case FTS_F: 1775 handle_file(entry->fts_name, entry->fts_statp); 1776 } 1777 } 1778 (void)fts_close(fts); 1779 } 1780 #endif 1781 1782 /* print a ratio - size reduction as a fraction of uncompressed size */ 1783 static void 1784 print_ratio(off_t in, off_t out, FILE *where) 1785 { 1786 int percent10; /* 10 * percent */ 1787 off_t diff; 1788 char buff[8]; 1789 int len; 1790 1791 diff = in - out/2; 1792 if (diff <= 0) 1793 /* 1794 * Output is more than double size of input! print -99.9% 1795 * Quite possibly we've failed to get the original size. 1796 */ 1797 percent10 = -999; 1798 else { 1799 /* 1800 * We only need 12 bits of result from the final division, 1801 * so reduce the values until a 32bit division will suffice. 1802 */ 1803 while (in > 0x100000) { 1804 diff >>= 1; 1805 in >>= 1; 1806 } 1807 if (in != 0) 1808 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1809 else 1810 percent10 = 0; 1811 } 1812 1813 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1814 /* Move the '.' to before the last digit */ 1815 buff[len - 1] = buff[len - 2]; 1816 buff[len - 2] = '.'; 1817 fprintf(where, "%5s%%", buff); 1818 } 1819 1820 #ifndef SMALL 1821 /* print compression statistics, and the new name (if there is one!) */ 1822 static void 1823 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1824 { 1825 if (file) 1826 fprintf(stderr, "%s:%s ", file, 1827 strlen(file) < 7 ? "\t\t" : "\t"); 1828 print_ratio(usize, gsize, stderr); 1829 if (nfile) 1830 fprintf(stderr, " -- replaced with %s", nfile); 1831 fprintf(stderr, "\n"); 1832 fflush(stderr); 1833 } 1834 1835 /* print test results */ 1836 static void 1837 print_test(const char *file, int ok) 1838 { 1839 1840 if (exit_value == 0 && ok == 0) 1841 exit_value = 1; 1842 fprintf(stderr, "%s:%s %s\n", file, 1843 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1844 fflush(stderr); 1845 } 1846 #endif 1847 1848 /* print a file's info ala --list */ 1849 /* eg: 1850 compressed uncompressed ratio uncompressed_name 1851 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1852 */ 1853 static void 1854 print_list(int fd, off_t out, const char *outfile, time_t ts) 1855 { 1856 static int first = 1; 1857 #ifndef SMALL 1858 static off_t in_tot, out_tot; 1859 uint32_t crc = 0; 1860 #endif 1861 off_t in = 0, rv; 1862 1863 if (first) { 1864 #ifndef SMALL 1865 if (vflag) 1866 printf("method crc date time "); 1867 #endif 1868 if (qflag == 0) 1869 printf(" compressed uncompressed " 1870 "ratio uncompressed_name\n"); 1871 } 1872 first = 0; 1873 1874 /* print totals? */ 1875 #ifndef SMALL 1876 if (fd == -1) { 1877 in = in_tot; 1878 out = out_tot; 1879 } else 1880 #endif 1881 { 1882 /* read the last 4 bytes - this is the uncompressed size */ 1883 rv = lseek(fd, (off_t)(-8), SEEK_END); 1884 if (rv != -1) { 1885 unsigned char buf[8]; 1886 uint32_t usize; 1887 1888 if (read(fd, (char *)buf, sizeof(buf)) != sizeof(buf)) 1889 maybe_warn("read of uncompressed size"); 1890 usize = buf[4] | buf[5] << 8 | buf[6] << 16 | buf[7] << 24; 1891 in = (off_t)usize; 1892 #ifndef SMALL 1893 crc = buf[0] | buf[1] << 8 | buf[2] << 16 | buf[3] << 24; 1894 #endif 1895 } 1896 } 1897 1898 #ifndef SMALL 1899 if (vflag && fd == -1) 1900 printf(" "); 1901 else if (vflag) { 1902 char *date = ctime(&ts); 1903 1904 /* skip the day, 1/100th second, and year */ 1905 date += 4; 1906 date[12] = 0; 1907 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1908 } 1909 in_tot += in; 1910 out_tot += out; 1911 #endif 1912 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1913 print_ratio(in, out, stdout); 1914 printf(" %s\n", outfile); 1915 } 1916 1917 /* display the usage of NetBSD gzip */ 1918 static void 1919 usage(void) 1920 { 1921 1922 fprintf(stderr, "%s\n", gzip_version); 1923 fprintf(stderr, 1924 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n" 1925 #ifndef SMALL 1926 " -c --stdout write to stdout, keep original files\n" 1927 " --to-stdout\n" 1928 " -d --decompress uncompress files\n" 1929 " --uncompress\n" 1930 " -f --force force overwriting & compress links\n" 1931 " -h --help display this help\n" 1932 " -n --no-name don't save original file name or time stamp\n" 1933 " -N --name save or restore original file name and time stamp\n" 1934 " -q --quiet output no warnings\n" 1935 " -r --recursive recursively compress files in directories\n" 1936 " -S .suf use suffix .suf instead of .gz\n" 1937 " --suffix .suf\n" 1938 " -t --test test compressed file\n" 1939 " -v --verbose print extra statistics\n" 1940 " -V --version display program version\n" 1941 " -1 --fast fastest (worst) compression\n" 1942 " -2 .. -8 set compression level\n" 1943 " -9 --best best (slowest) compression\n", 1944 #else 1945 , 1946 #endif 1947 getprogname()); 1948 exit(0); 1949 } 1950 1951 /* display the version of NetBSD gzip */ 1952 static void 1953 display_version(void) 1954 { 1955 1956 fprintf(stderr, "%s\n", gzip_version); 1957 exit(0); 1958 } 1959 1960 #ifndef NO_BZIP2_SUPPORT 1961 #include "unbzip2.c" 1962 #endif 1963 #ifndef NO_COMPRESS_SUPPORT 1964 #include "zuncompress.c" 1965 #endif 1966 1967 static ssize_t 1968 read_retry(int fd, void *buf, size_t sz) 1969 { 1970 char *cp = buf; 1971 size_t left = MIN(sz, (size_t) SSIZE_MAX); 1972 1973 while (left > 0) { 1974 ssize_t ret; 1975 1976 ret = read(fd, cp, left); 1977 if (ret == -1) { 1978 return ret; 1979 } else if (ret == 0) { 1980 break; /* EOF */ 1981 } 1982 cp += ret; 1983 left -= ret; 1984 } 1985 1986 return sz - left; 1987 } 1988
Indexes created Sun Sep 21 20:09:37 GMT 2025