gzip.c revision 1.93
1 /* $NetBSD: gzip.c,v 1.93 2008/08/03 09:25:05 skrll Exp $ */ 2 3 /* 4 * Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 21 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 22 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 23 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 24 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 26 * SUCH DAMAGE. 27 */ 28 29 #include <sys/cdefs.h> 30 #ifndef lint 31 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006\ 32 Matthew R. Green. All rights reserved."); 33 __RCSID("$NetBSD: gzip.c,v 1.93 2008/08/03 09:25:05 skrll Exp $"); 34 #endif /* not lint */ 35 36 /* 37 * gzip.c -- GPL free gzip using zlib. 38 * 39 * RFC 1950 covers the zlib format 40 * RFC 1951 covers the deflate format 41 * RFC 1952 covers the gzip format 42 * 43 * TODO: 44 * - use mmap where possible 45 * - handle some signals better (remove outfile?) 46 * - make bzip2/compress -v/-t/-l support work as well as possible 47 */ 48 49 #include <sys/param.h> 50 #include <sys/stat.h> 51 #include <sys/time.h> 52 53 #include <inttypes.h> 54 #include <unistd.h> 55 #include <stdio.h> 56 #include <string.h> 57 #include <stdlib.h> 58 #include <err.h> 59 #include <errno.h> 60 #include <fcntl.h> 61 #include <zlib.h> 62 #include <fts.h> 63 #include <libgen.h> 64 #include <stdarg.h> 65 #include <getopt.h> 66 #include <time.h> 67 68 #ifndef PRIdOFF 69 #define PRIdOFF PRId64 70 #endif 71 72 /* what type of file are we dealing with */ 73 enum filetype { 74 FT_GZIP, 75 #ifndef NO_BZIP2_SUPPORT 76 FT_BZIP2, 77 #endif 78 #ifndef NO_COMPRESS_SUPPORT 79 FT_Z, 80 #endif 81 FT_LAST, 82 FT_UNKNOWN 83 }; 84 85 #ifndef NO_BZIP2_SUPPORT 86 #include <bzlib.h> 87 88 #define BZ2_SUFFIX ".bz2" 89 #define BZIP2_MAGIC "\102\132\150" 90 #endif 91 92 #ifndef NO_COMPRESS_SUPPORT 93 #define Z_SUFFIX ".Z" 94 #define Z_MAGIC "\037\235" 95 #endif 96 97 #define GZ_SUFFIX ".gz" 98 99 #define BUFLEN (64 * 1024) 100 101 #define GZIP_MAGIC0 0x1F 102 #define GZIP_MAGIC1 0x8B 103 #define GZIP_OMAGIC1 0x9E 104 105 #define GZIP_TIMESTAMP (off_t)4 106 #define GZIP_ORIGNAME (off_t)10 107 108 #define HEAD_CRC 0x02 109 #define EXTRA_FIELD 0x04 110 #define ORIG_NAME 0x08 111 #define COMMENT 0x10 112 113 #define OS_CODE 3 /* Unix */ 114 115 typedef struct { 116 const char *zipped; 117 int ziplen; 118 const char *normal; /* for unzip - must not be longer than zipped */ 119 } suffixes_t; 120 static suffixes_t suffixes[] = { 121 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 122 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 123 #ifndef SMALL 124 SUFFIX(GZ_SUFFIX, ""), 125 SUFFIX(".z", ""), 126 SUFFIX("-gz", ""), 127 SUFFIX("-z", ""), 128 SUFFIX("_z", ""), 129 SUFFIX(".taz", ".tar"), 130 SUFFIX(".tgz", ".tar"), 131 #ifndef NO_BZIP2_SUPPORT 132 SUFFIX(BZ2_SUFFIX, ""), 133 #endif 134 #ifndef NO_COMPRESS_SUPPORT 135 SUFFIX(Z_SUFFIX, ""), 136 #endif 137 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 138 #endif /* SMALL */ 139 #undef SUFFIX 140 }; 141 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 142 143 static const char gzip_version[] = "NetBSD gzip 20060927"; 144 145 static int cflag; /* stdout mode */ 146 static int dflag; /* decompress mode */ 147 static int lflag; /* list mode */ 148 static int numflag = 6; /* gzip -1..-9 value */ 149 150 #ifndef SMALL 151 static int fflag; /* force mode */ 152 static int nflag; /* don't save name/timestamp */ 153 static int Nflag; /* don't restore name/timestamp */ 154 static int qflag; /* quiet mode */ 155 static int rflag; /* recursive mode */ 156 static int tflag; /* test */ 157 static int vflag; /* verbose mode */ 158 #else 159 #define qflag 0 160 #define tflag 0 161 #endif 162 163 static int exit_value = 0; /* exit value */ 164 165 static char *infile; /* name of file coming in */ 166 167 static void maybe_err(const char *fmt, ...) 168 __attribute__((__format__(__printf__, 1, 2))); 169 #ifndef NO_BZIP2_SUPPORT 170 static void maybe_errx(const char *fmt, ...) 171 __attribute__((__format__(__printf__, 1, 2))); 172 #endif 173 static void maybe_warn(const char *fmt, ...) 174 __attribute__((__format__(__printf__, 1, 2))); 175 static void maybe_warnx(const char *fmt, ...) 176 __attribute__((__format__(__printf__, 1, 2))); 177 static enum filetype file_gettype(u_char *); 178 #ifdef SMALL 179 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 180 #endif 181 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 182 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 183 static off_t file_compress(char *, char *, size_t); 184 static off_t file_uncompress(char *, char *, size_t); 185 static void handle_pathname(char *); 186 static void handle_file(char *, struct stat *); 187 static void handle_stdin(void); 188 static void handle_stdout(void); 189 static void print_ratio(off_t, off_t, FILE *); 190 static void print_list(int fd, off_t, const char *, time_t); 191 static void usage(void); 192 static void display_version(void); 193 static const suffixes_t *check_suffix(char *, int); 194 static ssize_t read_retry(int, void *, size_t); 195 196 #ifdef SMALL 197 #define unlink_input(f, sb) unlink(f) 198 #else 199 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 200 static void prepend_gzip(char *, int *, char ***); 201 static void handle_dir(char *); 202 static void print_verbage(const char *, const char *, off_t, off_t); 203 static void print_test(const char *, int); 204 static void copymodes(int fd, const struct stat *, const char *file); 205 static int check_outfile(const char *outfile); 206 #endif 207 208 #ifndef NO_BZIP2_SUPPORT 209 static off_t unbzip2(int, int, char *, size_t, off_t *); 210 #endif 211 212 #ifndef NO_COMPRESS_SUPPORT 213 static FILE *zdopen(int); 214 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 215 #endif 216 217 int main(int, char *p[]); 218 219 #ifdef SMALL 220 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 221 #else 222 static const struct option longopts[] = { 223 { "stdout", no_argument, 0, 'c' }, 224 { "to-stdout", no_argument, 0, 'c' }, 225 { "decompress", no_argument, 0, 'd' }, 226 { "uncompress", no_argument, 0, 'd' }, 227 { "force", no_argument, 0, 'f' }, 228 { "help", no_argument, 0, 'h' }, 229 { "list", no_argument, 0, 'l' }, 230 { "no-name", no_argument, 0, 'n' }, 231 { "name", no_argument, 0, 'N' }, 232 { "quiet", no_argument, 0, 'q' }, 233 { "recursive", no_argument, 0, 'r' }, 234 { "suffix", required_argument, 0, 'S' }, 235 { "test", no_argument, 0, 't' }, 236 { "verbose", no_argument, 0, 'v' }, 237 { "version", no_argument, 0, 'V' }, 238 { "fast", no_argument, 0, '1' }, 239 { "best", no_argument, 0, '9' }, 240 #if 0 241 /* 242 * This is what else GNU gzip implements. --ascii isn't useful 243 * on NetBSD, and I don't care to have a --license. 244 */ 245 { "ascii", no_argument, 0, 'a' }, 246 { "license", no_argument, 0, 'L' }, 247 #endif 248 { NULL, no_argument, 0, 0 }, 249 }; 250 #endif 251 252 int 253 main(int argc, char **argv) 254 { 255 const char *progname = getprogname(); 256 #ifndef SMALL 257 char *gzip; 258 int len; 259 #endif 260 int ch; 261 262 /* XXX set up signals */ 263 264 #ifndef SMALL 265 if ((gzip = getenv("GZIP")) != NULL) 266 prepend_gzip(gzip, &argc, &argv); 267 #endif 268 269 /* 270 * XXX 271 * handle being called `gunzip', `zcat' and `gzcat' 272 */ 273 if (strcmp(progname, "gunzip") == 0) 274 dflag = 1; 275 else if (strcmp(progname, "zcat") == 0 || 276 strcmp(progname, "gzcat") == 0) 277 dflag = cflag = 1; 278 279 #ifdef SMALL 280 #define OPT_LIST "123456789cdhltV" 281 #else 282 #define OPT_LIST "123456789cdfhlNnqrS:tVv" 283 #endif 284 285 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 286 switch (ch) { 287 case '1': case '2': case '3': 288 case '4': case '5': case '6': 289 case '7': case '8': case '9': 290 numflag = ch - '0'; 291 break; 292 case 'c': 293 cflag = 1; 294 break; 295 case 'd': 296 dflag = 1; 297 break; 298 case 'l': 299 lflag = 1; 300 dflag = 1; 301 break; 302 case 'V': 303 display_version(); 304 /* NOTREACHED */ 305 #ifndef SMALL 306 case 'f': 307 fflag = 1; 308 break; 309 case 'N': 310 nflag = 0; 311 Nflag = 1; 312 break; 313 case 'n': 314 nflag = 1; 315 Nflag = 0; 316 break; 317 case 'q': 318 qflag = 1; 319 break; 320 case 'r': 321 rflag = 1; 322 break; 323 case 'S': 324 len = strlen(optarg); 325 if (len != 0) { 326 suffixes[0].zipped = optarg; 327 suffixes[0].ziplen = len; 328 } else { 329 suffixes[NUM_SUFFIXES - 1].zipped = ""; 330 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 331 } 332 break; 333 case 't': 334 cflag = 1; 335 tflag = 1; 336 dflag = 1; 337 break; 338 case 'v': 339 vflag = 1; 340 break; 341 #endif 342 default: 343 usage(); 344 /* NOTREACHED */ 345 } 346 } 347 argv += optind; 348 argc -= optind; 349 350 if (argc == 0) { 351 if (dflag) /* stdin mode */ 352 handle_stdin(); 353 else /* stdout mode */ 354 handle_stdout(); 355 } else { 356 do { 357 handle_pathname(argv[0]); 358 } while (*++argv); 359 } 360 #ifndef SMALL 361 if (qflag == 0 && lflag && argc > 1) 362 print_list(-1, 0, "(totals)", 0); 363 #endif 364 exit(exit_value); 365 } 366 367 /* maybe print a warning */ 368 void 369 maybe_warn(const char *fmt, ...) 370 { 371 va_list ap; 372 373 if (qflag == 0) { 374 va_start(ap, fmt); 375 vwarn(fmt, ap); 376 va_end(ap); 377 } 378 if (exit_value == 0) 379 exit_value = 1; 380 } 381 382 /* ... without an errno. */ 383 void 384 maybe_warnx(const char *fmt, ...) 385 { 386 va_list ap; 387 388 if (qflag == 0) { 389 va_start(ap, fmt); 390 vwarnx(fmt, ap); 391 va_end(ap); 392 } 393 if (exit_value == 0) 394 exit_value = 1; 395 } 396 397 /* maybe print an error */ 398 void 399 maybe_err(const char *fmt, ...) 400 { 401 va_list ap; 402 403 if (qflag == 0) { 404 va_start(ap, fmt); 405 vwarn(fmt, ap); 406 va_end(ap); 407 } 408 exit(2); 409 } 410 411 #ifndef NO_BZIP2_SUPPORT 412 /* ... without an errno. */ 413 void 414 maybe_errx(const char *fmt, ...) 415 { 416 va_list ap; 417 418 if (qflag == 0) { 419 va_start(ap, fmt); 420 vwarnx(fmt, ap); 421 va_end(ap); 422 } 423 exit(2); 424 } 425 #endif 426 427 #ifndef SMALL 428 /* split up $GZIP and prepend it to the argument list */ 429 static void 430 prepend_gzip(char *gzip, int *argc, char ***argv) 431 { 432 char *s, **nargv, **ac; 433 int nenvarg = 0, i; 434 435 /* scan how many arguments there are */ 436 for (s = gzip;;) { 437 while (*s == ' ' || *s == '\t') 438 s++; 439 if (*s == 0) 440 goto count_done; 441 nenvarg++; 442 while (*s != ' ' && *s != '\t') 443 if (*s++ == 0) 444 goto count_done; 445 } 446 count_done: 447 /* punt early */ 448 if (nenvarg == 0) 449 return; 450 451 *argc += nenvarg; 452 ac = *argv; 453 454 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 455 if (nargv == NULL) 456 maybe_err("malloc"); 457 458 /* stash this away */ 459 *argv = nargv; 460 461 /* copy the program name first */ 462 i = 0; 463 nargv[i++] = *(ac++); 464 465 /* take a copy of $GZIP and add it to the array */ 466 s = strdup(gzip); 467 if (s == NULL) 468 maybe_err("strdup"); 469 for (;;) { 470 /* Skip whitespaces. */ 471 while (*s == ' ' || *s == '\t') 472 s++; 473 if (*s == 0) 474 goto copy_done; 475 nargv[i++] = s; 476 /* Find the end of this argument. */ 477 while (*s != ' ' && *s != '\t') 478 if (*s++ == 0) 479 /* Argument followed by NUL. */ 480 goto copy_done; 481 /* Terminate by overwriting ' ' or '\t' with NUL. */ 482 *s++ = 0; 483 } 484 copy_done: 485 486 /* copy the original arguments and a NULL */ 487 while (*ac) 488 nargv[i++] = *(ac++); 489 nargv[i] = NULL; 490 } 491 #endif 492 493 /* compress input to output. Return bytes read, -1 on error */ 494 static off_t 495 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 496 { 497 z_stream z; 498 char *outbufp, *inbufp; 499 off_t in_tot = 0, out_tot = 0; 500 ssize_t in_size; 501 int i, error; 502 uLong crc; 503 #ifdef SMALL 504 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 505 0, 0, 0, 0, 506 0, OS_CODE }; 507 #endif 508 509 outbufp = malloc(BUFLEN); 510 inbufp = malloc(BUFLEN); 511 if (outbufp == NULL || inbufp == NULL) { 512 maybe_err("malloc failed"); 513 goto out; 514 } 515 516 memset(&z, 0, sizeof z); 517 z.zalloc = Z_NULL; 518 z.zfree = Z_NULL; 519 z.opaque = 0; 520 521 #ifdef SMALL 522 memcpy(outbufp, header, sizeof header); 523 i = sizeof header; 524 #else 525 if (nflag != 0) { 526 mtime = 0; 527 origname = ""; 528 } 529 530 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 531 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 532 *origname ? ORIG_NAME : 0, 533 mtime & 0xff, 534 (mtime >> 8) & 0xff, 535 (mtime >> 16) & 0xff, 536 (mtime >> 24) & 0xff, 537 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 538 OS_CODE, origname); 539 if (i >= BUFLEN) 540 /* this need PATH_MAX > BUFLEN ... */ 541 maybe_err("snprintf"); 542 if (*origname) 543 i++; 544 #endif 545 546 z.next_out = outbufp + i; 547 z.avail_out = BUFLEN - i; 548 549 error = deflateInit2(&z, numflag, Z_DEFLATED, 550 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 551 if (error != Z_OK) { 552 maybe_warnx("deflateInit2 failed"); 553 in_tot = -1; 554 goto out; 555 } 556 557 crc = crc32(0L, Z_NULL, 0); 558 for (;;) { 559 if (z.avail_out == 0) { 560 if (write(out, outbufp, BUFLEN) != BUFLEN) { 561 maybe_warn("write"); 562 out_tot = -1; 563 goto out; 564 } 565 566 out_tot += BUFLEN; 567 z.next_out = outbufp; 568 z.avail_out = BUFLEN; 569 } 570 571 if (z.avail_in == 0) { 572 in_size = read(in, inbufp, BUFLEN); 573 if (in_size < 0) { 574 maybe_warn("read"); 575 in_tot = -1; 576 goto out; 577 } 578 if (in_size == 0) 579 break; 580 581 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 582 in_tot += in_size; 583 z.next_in = inbufp; 584 z.avail_in = in_size; 585 } 586 587 error = deflate(&z, Z_NO_FLUSH); 588 if (error != Z_OK && error != Z_STREAM_END) { 589 maybe_warnx("deflate failed"); 590 in_tot = -1; 591 goto out; 592 } 593 } 594 595 /* clean up */ 596 for (;;) { 597 size_t len; 598 ssize_t w; 599 600 error = deflate(&z, Z_FINISH); 601 if (error != Z_OK && error != Z_STREAM_END) { 602 maybe_warnx("deflate failed"); 603 in_tot = -1; 604 goto out; 605 } 606 607 len = (char *)z.next_out - outbufp; 608 609 w = write(out, outbufp, len); 610 if (w == -1 || (size_t)w != len) { 611 maybe_warn("write"); 612 out_tot = -1; 613 goto out; 614 } 615 out_tot += len; 616 z.next_out = outbufp; 617 z.avail_out = BUFLEN; 618 619 if (error == Z_STREAM_END) 620 break; 621 } 622 623 if (deflateEnd(&z) != Z_OK) { 624 maybe_warnx("deflateEnd failed"); 625 in_tot = -1; 626 goto out; 627 } 628 629 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 630 (int)crc & 0xff, 631 (int)(crc >> 8) & 0xff, 632 (int)(crc >> 16) & 0xff, 633 (int)(crc >> 24) & 0xff, 634 (int)in_tot & 0xff, 635 (int)(in_tot >> 8) & 0xff, 636 (int)(in_tot >> 16) & 0xff, 637 (int)(in_tot >> 24) & 0xff); 638 if (i != 8) 639 maybe_err("snprintf"); 640 #if 0 641 if (in_tot > 0xffffffff) 642 maybe_warn("input file size >= 4GB cannot be saved"); 643 #endif 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_and_fail; 748 } 749 goto stop; 750 } 751 switch (state) { 752 case GZSTATE_MAGIC0: 753 if (*z.next_in != GZIP_MAGIC0) { 754 if (in_tot > 0) { 755 maybe_warnx("%s: trailing garbage " 756 "ignored", filename); 757 goto stop; 758 } 759 maybe_warnx("input not gziped (MAGIC0)"); 760 goto stop_and_fail; 761 } 762 ADVANCE(); 763 state++; 764 out_sub_tot = 0; 765 crc = crc32(0L, Z_NULL, 0); 766 break; 767 768 case GZSTATE_MAGIC1: 769 if (*z.next_in != GZIP_MAGIC1 && 770 *z.next_in != GZIP_OMAGIC1) { 771 maybe_warnx("input not gziped (MAGIC1)"); 772 goto stop_and_fail; 773 } 774 ADVANCE(); 775 state++; 776 break; 777 778 case GZSTATE_METHOD: 779 if (*z.next_in != Z_DEFLATED) { 780 maybe_warnx("unknown compression method"); 781 goto stop_and_fail; 782 } 783 ADVANCE(); 784 state++; 785 break; 786 787 case GZSTATE_FLAGS: 788 flags = *z.next_in; 789 ADVANCE(); 790 skip_count = 6; 791 state++; 792 break; 793 794 case GZSTATE_SKIPPING: 795 if (skip_count > 0) { 796 skip_count--; 797 ADVANCE(); 798 } else 799 state++; 800 break; 801 802 case GZSTATE_EXTRA: 803 if ((flags & EXTRA_FIELD) == 0) { 804 state = GZSTATE_ORIGNAME; 805 break; 806 } 807 skip_count = *z.next_in; 808 ADVANCE(); 809 state++; 810 break; 811 812 case GZSTATE_EXTRA2: 813 skip_count |= ((*z.next_in) << 8); 814 ADVANCE(); 815 state++; 816 break; 817 818 case GZSTATE_EXTRA3: 819 if (skip_count > 0) { 820 skip_count--; 821 ADVANCE(); 822 } else 823 state++; 824 break; 825 826 case GZSTATE_ORIGNAME: 827 if ((flags & ORIG_NAME) == 0) { 828 state++; 829 break; 830 } 831 if (*z.next_in == 0) 832 state++; 833 ADVANCE(); 834 break; 835 836 case GZSTATE_COMMENT: 837 if ((flags & COMMENT) == 0) { 838 state++; 839 break; 840 } 841 if (*z.next_in == 0) 842 state++; 843 ADVANCE(); 844 break; 845 846 case GZSTATE_HEAD_CRC1: 847 if (flags & HEAD_CRC) 848 skip_count = 2; 849 else 850 skip_count = 0; 851 state++; 852 break; 853 854 case GZSTATE_HEAD_CRC2: 855 if (skip_count > 0) { 856 skip_count--; 857 ADVANCE(); 858 } else 859 state++; 860 break; 861 862 case GZSTATE_INIT: 863 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 864 maybe_warnx("failed to inflateInit"); 865 goto stop_and_fail; 866 } 867 state++; 868 break; 869 870 case GZSTATE_READ: 871 error = inflate(&z, Z_FINISH); 872 switch (error) { 873 /* Z_BUF_ERROR goes with Z_FINISH... */ 874 case Z_BUF_ERROR: 875 case Z_STREAM_END: 876 case Z_OK: 877 break; 878 879 case Z_NEED_DICT: 880 maybe_warnx("Z_NEED_DICT error"); 881 goto stop_and_fail; 882 case Z_DATA_ERROR: 883 maybe_warnx("data stream error"); 884 goto stop_and_fail; 885 case Z_STREAM_ERROR: 886 maybe_warnx("internal stream error"); 887 goto stop_and_fail; 888 case Z_MEM_ERROR: 889 maybe_warnx("memory allocation error"); 890 goto stop_and_fail; 891 892 default: 893 maybe_warn("unknown error from inflate(): %d", 894 error); 895 } 896 wr = BUFLEN - z.avail_out; 897 898 if (wr != 0) { 899 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 900 if ( 901 #ifndef SMALL 902 /* don't write anything with -t */ 903 tflag == 0 && 904 #endif 905 write(out, outbufp, wr) != wr) { 906 maybe_warn("error writing to output"); 907 goto stop_and_fail; 908 } 909 910 out_tot += wr; 911 out_sub_tot += wr; 912 } 913 914 if (error == Z_STREAM_END) { 915 inflateEnd(&z); 916 state++; 917 } 918 919 z.next_out = outbufp; 920 z.avail_out = BUFLEN; 921 922 break; 923 case GZSTATE_CRC: 924 { 925 uLong origcrc; 926 927 if (z.avail_in < 4) { 928 if (!done_reading) { 929 needmore = 1; 930 continue; 931 } 932 maybe_warnx("truncated input"); 933 goto stop_and_fail; 934 } 935 origcrc = ((unsigned)z.next_in[0] & 0xff) | 936 ((unsigned)z.next_in[1] & 0xff) << 8 | 937 ((unsigned)z.next_in[2] & 0xff) << 16 | 938 ((unsigned)z.next_in[3] & 0xff) << 24; 939 if (origcrc != crc) { 940 maybe_warnx("invalid compressed" 941 " data--crc error"); 942 goto stop_and_fail; 943 } 944 } 945 946 z.avail_in -= 4; 947 z.next_in += 4; 948 949 if (!z.avail_in && done_reading) { 950 goto stop; 951 } 952 state++; 953 break; 954 case GZSTATE_LEN: 955 { 956 uLong origlen; 957 958 if (z.avail_in < 4) { 959 if (!done_reading) { 960 needmore = 1; 961 continue; 962 } 963 maybe_warnx("truncated input"); 964 goto stop_and_fail; 965 } 966 origlen = ((unsigned)z.next_in[0] & 0xff) | 967 ((unsigned)z.next_in[1] & 0xff) << 8 | 968 ((unsigned)z.next_in[2] & 0xff) << 16 | 969 ((unsigned)z.next_in[3] & 0xff) << 24; 970 971 if (origlen != out_sub_tot) { 972 maybe_warnx("invalid compressed" 973 " data--length error"); 974 goto stop_and_fail; 975 } 976 } 977 978 z.avail_in -= 4; 979 z.next_in += 4; 980 981 if (error < 0) { 982 maybe_warnx("decompression error"); 983 goto stop_and_fail; 984 } 985 state = GZSTATE_MAGIC0; 986 break; 987 } 988 continue; 989 stop_and_fail: 990 out_tot = -1; 991 stop: 992 break; 993 } 994 if (state > GZSTATE_INIT) 995 inflateEnd(&z); 996 997 free(inbufp); 998 out1: 999 free(outbufp); 1000 out2: 1001 if (gsizep) 1002 *gsizep = in_tot; 1003 return (out_tot); 1004 } 1005 1006 #ifndef SMALL 1007 /* 1008 * set the owner, mode, flags & utimes using the given file descriptor. 1009 * file is only used in possible warning messages. 1010 */ 1011 static void 1012 copymodes(int fd, const struct stat *sbp, const char *file) 1013 { 1014 struct timeval times[2]; 1015 struct stat sb; 1016 1017 /* 1018 * If we have no info on the input, give this file some 1019 * default values and return.. 1020 */ 1021 if (sbp == NULL) { 1022 mode_t mask = umask(022); 1023 1024 (void)fchmod(fd, DEFFILEMODE & ~mask); 1025 (void)umask(mask); 1026 return; 1027 } 1028 sb = *sbp; 1029 1030 /* if the chown fails, remove set-id bits as-per compress(1) */ 1031 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1032 if (errno != EPERM) 1033 maybe_warn("couldn't fchown: %s", file); 1034 sb.st_mode &= ~(S_ISUID|S_ISGID); 1035 } 1036 1037 /* we only allow set-id and the 9 normal permission bits */ 1038 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1039 if (fchmod(fd, sb.st_mode) < 0) 1040 maybe_warn("couldn't fchmod: %s", file); 1041 1042 /* only try flags if they exist already */ 1043 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1044 maybe_warn("couldn't fchflags: %s", file); 1045 1046 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1047 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1048 if (futimes(fd, times) < 0) 1049 maybe_warn("couldn't utimes: %s", file); 1050 } 1051 #endif 1052 1053 /* what sort of file is this? */ 1054 static enum filetype 1055 file_gettype(u_char *buf) 1056 { 1057 1058 if (buf[0] == GZIP_MAGIC0 && 1059 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1060 return FT_GZIP; 1061 else 1062 #ifndef NO_BZIP2_SUPPORT 1063 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1064 buf[3] >= '0' && buf[3] <= '9') 1065 return FT_BZIP2; 1066 else 1067 #endif 1068 #ifndef NO_COMPRESS_SUPPORT 1069 if (memcmp(buf, Z_MAGIC, 2) == 0) 1070 return FT_Z; 1071 else 1072 #endif 1073 return FT_UNKNOWN; 1074 } 1075 1076 #ifndef SMALL 1077 /* check the outfile is OK. */ 1078 static int 1079 check_outfile(const char *outfile) 1080 { 1081 struct stat sb; 1082 int ok = 1; 1083 1084 if (lflag == 0 && stat(outfile, &sb) == 0) { 1085 if (fflag) 1086 unlink(outfile); 1087 else if (isatty(STDIN_FILENO)) { 1088 char ans[10] = { 'n', '\0' }; /* default */ 1089 1090 fprintf(stderr, "%s already exists -- do you wish to " 1091 "overwrite (y or n)? " , outfile); 1092 (void)fgets(ans, sizeof(ans) - 1, stdin); 1093 if (ans[0] != 'y' && ans[0] != 'Y') { 1094 fprintf(stderr, "\tnot overwriting\n"); 1095 ok = 0; 1096 } else 1097 unlink(outfile); 1098 } else { 1099 maybe_warnx("%s already exists -- skipping", outfile); 1100 ok = 0; 1101 } 1102 } 1103 return ok; 1104 } 1105 1106 static void 1107 unlink_input(const char *file, const struct stat *sb) 1108 { 1109 struct stat nsb; 1110 1111 if (stat(file, &nsb) != 0) 1112 /* Must be gone alrady */ 1113 return; 1114 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1115 /* Definitely a different file */ 1116 return; 1117 unlink(file); 1118 } 1119 #endif 1120 1121 static const suffixes_t * 1122 check_suffix(char *file, int xlate) 1123 { 1124 const suffixes_t *s; 1125 int len = strlen(file); 1126 char *sp; 1127 1128 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1129 /* if it doesn't fit in "a.suf", don't bother */ 1130 if (s->ziplen >= len) 1131 continue; 1132 sp = file + len - s->ziplen; 1133 if (strcmp(s->zipped, sp) != 0) 1134 continue; 1135 if (xlate) 1136 strcpy(sp, s->normal); 1137 return s; 1138 } 1139 return NULL; 1140 } 1141 1142 /* 1143 * compress the given file: create a corresponding .gz file and remove the 1144 * original. 1145 */ 1146 static off_t 1147 file_compress(char *file, char *outfile, size_t outsize) 1148 { 1149 int in; 1150 int out; 1151 off_t size, insize; 1152 #ifndef SMALL 1153 struct stat isb, osb; 1154 const suffixes_t *suff; 1155 #endif 1156 1157 in = open(file, O_RDONLY); 1158 if (in == -1) { 1159 maybe_warn("can't open %s", file); 1160 return -1; 1161 } 1162 1163 if (cflag == 0) { 1164 #ifndef SMALL 1165 if (fstat(in, &isb) == 0) { 1166 if (isb.st_nlink > 1 && fflag == 0) { 1167 maybe_warnx("%s has %d other link%s -- " 1168 "skipping", file, isb.st_nlink - 1, 1169 isb.st_nlink == 1 ? "" : "s"); 1170 close(in); 1171 return -1; 1172 } 1173 } 1174 1175 if (fflag == 0 && (suff = check_suffix(file, 0)) 1176 && suff->zipped[0] != 0) { 1177 maybe_warnx("%s already has %s suffix -- unchanged", 1178 file, suff->zipped); 1179 close(in); 1180 return -1; 1181 } 1182 #endif 1183 1184 /* Add (usually) .gz to filename */ 1185 if ((size_t)snprintf(outfile, outsize, "%s%s", 1186 file, suffixes[0].zipped) >= outsize) 1187 memcpy(outfile - suffixes[0].ziplen - 1, 1188 suffixes[0].zipped, suffixes[0].ziplen + 1); 1189 1190 #ifndef SMALL 1191 if (check_outfile(outfile) == 0) { 1192 close(in); 1193 return -1; 1194 } 1195 #endif 1196 } 1197 1198 if (cflag == 0) { 1199 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1200 if (out == -1) { 1201 maybe_warn("could not create output: %s", outfile); 1202 fclose(stdin); 1203 return -1; 1204 } 1205 } else 1206 out = STDOUT_FILENO; 1207 1208 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1209 1210 (void)close(in); 1211 1212 /* 1213 * If there was an error, insize will be -1. 1214 * If we compressed to stdout, just return the size. 1215 * Otherwise stat the file and check it is the correct size. 1216 * We only blow away the file if we can stat the output and it 1217 * has the expected size. 1218 */ 1219 if (cflag != 0) 1220 return insize == -1 ? -1 : size; 1221 1222 #ifndef SMALL 1223 if (fstat(out, &osb) != 0) { 1224 maybe_warn("couldn't stat: %s", outfile); 1225 goto bad_outfile; 1226 } 1227 1228 if (osb.st_size != size) { 1229 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1230 " != %" PRIdOFF "), deleting", 1231 outfile, osb.st_size, size); 1232 goto bad_outfile; 1233 } 1234 1235 copymodes(out, &isb, outfile); 1236 #endif 1237 if (close(out) == -1) 1238 maybe_warn("couldn't close output"); 1239 1240 /* output is good, ok to delete input */ 1241 unlink_input(file, &isb); 1242 return size; 1243 1244 #ifndef SMALL 1245 bad_outfile: 1246 if (close(out) == -1) 1247 maybe_warn("couldn't close output"); 1248 1249 maybe_warnx("leaving original %s", file); 1250 unlink(outfile); 1251 return size; 1252 #endif 1253 } 1254 1255 /* uncompress the given file and remove the original */ 1256 static off_t 1257 file_uncompress(char *file, char *outfile, size_t outsize) 1258 { 1259 struct stat isb, osb; 1260 off_t size; 1261 ssize_t rbytes; 1262 unsigned char header1[4]; 1263 enum filetype method; 1264 int fd, ofd, zfd = -1; 1265 #ifndef SMALL 1266 int rv; 1267 time_t timestamp = 0; 1268 unsigned char name[PATH_MAX + 1]; 1269 #endif 1270 1271 /* gather the old name info */ 1272 1273 fd = open(file, O_RDONLY); 1274 if (fd < 0) { 1275 maybe_warn("can't open %s", file); 1276 goto lose; 1277 } 1278 1279 strlcpy(outfile, file, outsize); 1280 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1281 maybe_warnx("%s: unknown suffix -- ignored", file); 1282 goto lose; 1283 } 1284 1285 rbytes = read(fd, header1, sizeof header1); 1286 if (rbytes != sizeof header1) { 1287 /* we don't want to fail here. */ 1288 #ifndef SMALL 1289 if (fflag) 1290 goto lose; 1291 #endif 1292 if (rbytes == -1) 1293 maybe_warn("can't read %s", file); 1294 else 1295 goto unexpected_EOF; 1296 goto lose; 1297 } 1298 1299 method = file_gettype(header1); 1300 1301 #ifndef SMALL 1302 if (fflag == 0 && method == FT_UNKNOWN) { 1303 maybe_warnx("%s: not in gzip format", file); 1304 goto lose; 1305 } 1306 1307 #endif 1308 1309 #ifndef SMALL 1310 if (method == FT_GZIP && Nflag) { 1311 unsigned char ts[4]; /* timestamp */ 1312 1313 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1314 if (rv >= 0 && rv < sizeof ts) 1315 goto unexpected_EOF; 1316 if (rv == -1) { 1317 if (!fflag) 1318 maybe_warn("can't read %s", file); 1319 goto lose; 1320 } 1321 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1322 1323 if (header1[3] & ORIG_NAME) { 1324 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1325 if (rbytes < 0) { 1326 maybe_warn("can't read %s", file); 1327 goto lose; 1328 } 1329 if (name[0] != 0) { 1330 /* preserve original directory name */ 1331 char *dp = strrchr(file, '/'); 1332 if (dp == NULL) 1333 dp = file; 1334 else 1335 dp++; 1336 snprintf(outfile, outsize, "%.*s%.*s", 1337 (int) (dp - file), 1338 file, (int) rbytes, name); 1339 } 1340 } 1341 } 1342 #endif 1343 lseek(fd, 0, SEEK_SET); 1344 1345 if (cflag == 0 || lflag) { 1346 if (fstat(fd, &isb) != 0) 1347 goto lose; 1348 #ifndef SMALL 1349 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1350 maybe_warnx("%s has %d other links -- skipping", 1351 file, isb.st_nlink - 1); 1352 goto lose; 1353 } 1354 if (nflag == 0 && timestamp) 1355 isb.st_mtime = timestamp; 1356 if (check_outfile(outfile) == 0) 1357 goto lose; 1358 #endif 1359 } 1360 1361 if (cflag == 0 && lflag == 0) { 1362 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1363 if (zfd == STDOUT_FILENO) { 1364 /* We won't close STDOUT_FILENO later... */ 1365 zfd = dup(zfd); 1366 close(STDOUT_FILENO); 1367 } 1368 if (zfd == -1) { 1369 maybe_warn("can't open %s", outfile); 1370 goto lose; 1371 } 1372 } else 1373 zfd = STDOUT_FILENO; 1374 1375 #ifndef NO_BZIP2_SUPPORT 1376 if (method == FT_BZIP2) { 1377 1378 /* XXX */ 1379 if (lflag) { 1380 maybe_warnx("no -l with bzip2 files"); 1381 goto lose; 1382 } 1383 1384 size = unbzip2(fd, zfd, NULL, 0, NULL); 1385 } else 1386 #endif 1387 1388 #ifndef NO_COMPRESS_SUPPORT 1389 if (method == FT_Z) { 1390 FILE *in, *out; 1391 1392 /* XXX */ 1393 if (lflag) { 1394 maybe_warnx("no -l with Lempel-Ziv files"); 1395 goto lose; 1396 } 1397 1398 if ((in = zdopen(fd)) == NULL) { 1399 maybe_warn("zdopen for read: %s", file); 1400 goto lose; 1401 } 1402 1403 out = fdopen(dup(zfd), "w"); 1404 if (out == NULL) { 1405 maybe_warn("fdopen for write: %s", outfile); 1406 fclose(in); 1407 goto lose; 1408 } 1409 1410 size = zuncompress(in, out, NULL, 0, NULL); 1411 /* need to fclose() if ferror() is true... */ 1412 if (ferror(in) | fclose(in)) { 1413 maybe_warn("failed infile fclose"); 1414 unlink(outfile); 1415 (void)fclose(out); 1416 } 1417 if (fclose(out) != 0) { 1418 maybe_warn("failed outfile fclose"); 1419 unlink(outfile); 1420 goto lose; 1421 } 1422 } else 1423 #endif 1424 1425 #ifndef SMALL 1426 if (method == FT_UNKNOWN) { 1427 if (lflag) { 1428 maybe_warnx("no -l for unknown filetypes"); 1429 goto lose; 1430 } 1431 size = cat_fd(NULL, 0, NULL, fd); 1432 } else 1433 #endif 1434 { 1435 if (lflag) { 1436 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1437 close(fd); 1438 return -1; /* XXX */ 1439 } 1440 1441 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1442 } 1443 1444 if (close(fd) != 0) 1445 maybe_warn("couldn't close input"); 1446 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1447 maybe_warn("couldn't close output"); 1448 1449 if (size == -1) { 1450 if (cflag == 0) 1451 unlink(outfile); 1452 maybe_warnx("%s: uncompress failed", file); 1453 return -1; 1454 } 1455 1456 /* if testing, or we uncompressed to stdout, this is all we need */ 1457 #ifndef SMALL 1458 if (tflag) 1459 return size; 1460 #endif 1461 /* if we are uncompressing to stdin, don't remove the file. */ 1462 if (cflag) 1463 return size; 1464 1465 /* 1466 * if we create a file... 1467 */ 1468 /* 1469 * if we can't stat the file don't remove the file. 1470 */ 1471 1472 ofd = open(outfile, O_RDWR, 0); 1473 if (ofd == -1) { 1474 maybe_warn("couldn't open (leaving original): %s", 1475 outfile); 1476 return -1; 1477 } 1478 if (fstat(ofd, &osb) != 0) { 1479 maybe_warn("couldn't stat (leaving original): %s", 1480 outfile); 1481 close(ofd); 1482 return -1; 1483 } 1484 if (osb.st_size != size) { 1485 maybe_warnx("stat gave different size: %" PRIdOFF 1486 " != %" PRIdOFF " (leaving original)", 1487 size, osb.st_size); 1488 close(ofd); 1489 unlink(outfile); 1490 return -1; 1491 } 1492 unlink_input(file, &isb); 1493 #ifndef SMALL 1494 copymodes(ofd, &isb, outfile); 1495 #endif 1496 close(ofd); 1497 return size; 1498 1499 unexpected_EOF: 1500 maybe_warnx("%s: unexpected end of file", file); 1501 lose: 1502 if (fd != -1) 1503 close(fd); 1504 if (zfd != -1 && zfd != STDOUT_FILENO) 1505 close(fd); 1506 return -1; 1507 } 1508 1509 #ifndef SMALL 1510 static off_t 1511 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1512 { 1513 char buf[BUFLEN]; 1514 off_t in_tot; 1515 ssize_t w; 1516 1517 in_tot = count; 1518 w = write(STDOUT_FILENO, prepend, count); 1519 if (w == -1 || (size_t)w != count) { 1520 maybe_warn("write to stdout"); 1521 return -1; 1522 } 1523 for (;;) { 1524 ssize_t rv; 1525 1526 rv = read(fd, buf, sizeof buf); 1527 if (rv == 0) 1528 break; 1529 if (rv < 0) { 1530 maybe_warn("read from fd %d", fd); 1531 break; 1532 } 1533 1534 if (write(STDOUT_FILENO, buf, rv) != rv) { 1535 maybe_warn("write to stdout"); 1536 break; 1537 } 1538 in_tot += rv; 1539 } 1540 1541 if (gsizep) 1542 *gsizep = in_tot; 1543 return (in_tot); 1544 } 1545 #endif 1546 1547 static void 1548 handle_stdin(void) 1549 { 1550 unsigned char header1[4]; 1551 off_t usize, gsize; 1552 enum filetype method; 1553 ssize_t bytes_read; 1554 #ifndef NO_COMPRESS_SUPPORT 1555 FILE *in; 1556 #endif 1557 1558 #ifndef SMALL 1559 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1560 maybe_warnx("standard input is a terminal -- ignoring"); 1561 return; 1562 } 1563 #endif 1564 1565 if (lflag) { 1566 struct stat isb; 1567 1568 /* XXX could read the whole file, etc. */ 1569 if (fstat(STDIN_FILENO, &isb) < 0) { 1570 maybe_warn("fstat"); 1571 return; 1572 } 1573 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1574 return; 1575 } 1576 1577 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1578 if (bytes_read == -1) { 1579 maybe_warn("can't read stdin"); 1580 return; 1581 } else if (bytes_read != sizeof(header1)) { 1582 maybe_warnx("(stdin): unexpected end of file"); 1583 return; 1584 } 1585 1586 method = file_gettype(header1); 1587 switch (method) { 1588 default: 1589 #ifndef SMALL 1590 if (fflag == 0) { 1591 maybe_warnx("unknown compression format"); 1592 return; 1593 } 1594 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1595 break; 1596 #endif 1597 case FT_GZIP: 1598 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1599 header1, sizeof header1, &gsize, "(stdin)"); 1600 break; 1601 #ifndef NO_BZIP2_SUPPORT 1602 case FT_BZIP2: 1603 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1604 header1, sizeof header1, &gsize); 1605 break; 1606 #endif 1607 #ifndef NO_COMPRESS_SUPPORT 1608 case FT_Z: 1609 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1610 maybe_warnx("zopen of stdin"); 1611 return; 1612 } 1613 1614 usize = zuncompress(in, stdout, header1, sizeof header1, &gsize); 1615 fclose(in); 1616 break; 1617 #endif 1618 } 1619 1620 #ifndef SMALL 1621 if (vflag && !tflag && usize != -1 && gsize != -1) 1622 print_verbage(NULL, NULL, usize, gsize); 1623 if (vflag && tflag) 1624 print_test("(stdin)", usize != -1); 1625 #endif 1626 1627 } 1628 1629 static void 1630 handle_stdout(void) 1631 { 1632 off_t gsize, usize; 1633 struct stat sb; 1634 time_t systime; 1635 uint32_t mtime; 1636 int ret; 1637 1638 #ifndef SMALL 1639 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1640 maybe_warnx("standard output is a terminal -- ignoring"); 1641 return; 1642 } 1643 #endif 1644 /* If stdin is a file use it's mtime, otherwise use current time */ 1645 ret = fstat(STDIN_FILENO, &sb); 1646 1647 #ifndef SMALL 1648 if (ret < 0) { 1649 maybe_warn("Can't stat stdin"); 1650 return; 1651 } 1652 #endif 1653 1654 if (S_ISREG(sb.st_mode)) 1655 mtime = (uint32_t)sb.st_mtime; 1656 else { 1657 systime = time(NULL); 1658 #ifndef SMALL 1659 if (systime == -1) { 1660 maybe_warn("time"); 1661 return; 1662 } 1663 #endif 1664 mtime = (uint32_t)systime; 1665 } 1666 1667 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1668 #ifndef SMALL 1669 if (vflag && !tflag && usize != -1 && gsize != -1) 1670 print_verbage(NULL, NULL, usize, gsize); 1671 #endif 1672 } 1673 1674 /* do what is asked for, for the path name */ 1675 static void 1676 handle_pathname(char *path) 1677 { 1678 char *opath = path, *s = NULL; 1679 ssize_t len; 1680 int slen; 1681 struct stat sb; 1682 1683 /* check for stdout/stdin */ 1684 if (path[0] == '-' && path[1] == '\0') { 1685 if (dflag) 1686 handle_stdin(); 1687 else 1688 handle_stdout(); 1689 return; 1690 } 1691 1692 retry: 1693 if (stat(path, &sb) != 0) { 1694 /* lets try <path>.gz if we're decompressing */ 1695 if (dflag && s == NULL && errno == ENOENT) { 1696 len = strlen(path); 1697 slen = suffixes[0].ziplen; 1698 s = malloc(len + slen + 1); 1699 if (s == NULL) 1700 maybe_err("malloc"); 1701 memcpy(s, path, len); 1702 memcpy(s + len, suffixes[0].zipped, slen + 1); 1703 path = s; 1704 goto retry; 1705 } 1706 maybe_warn("can't stat: %s", opath); 1707 goto out; 1708 } 1709 1710 if (S_ISDIR(sb.st_mode)) { 1711 #ifndef SMALL 1712 if (rflag) 1713 handle_dir(path); 1714 else 1715 #endif 1716 maybe_warnx("%s is a directory", path); 1717 goto out; 1718 } 1719 1720 if (S_ISREG(sb.st_mode)) 1721 handle_file(path, &sb); 1722 else 1723 maybe_warnx("%s is not a regular file", path); 1724 1725 out: 1726 if (s) 1727 free(s); 1728 } 1729 1730 /* compress/decompress a file */ 1731 static void 1732 handle_file(char *file, struct stat *sbp) 1733 { 1734 off_t usize, gsize; 1735 char outfile[PATH_MAX]; 1736 1737 infile = file; 1738 if (dflag) { 1739 usize = file_uncompress(file, outfile, sizeof(outfile)); 1740 #ifndef SMALL 1741 if (vflag && tflag) 1742 print_test(file, usize != -1); 1743 #endif 1744 if (usize == -1) 1745 return; 1746 gsize = sbp->st_size; 1747 } else { 1748 gsize = file_compress(file, outfile, sizeof(outfile)); 1749 if (gsize == -1) 1750 return; 1751 usize = sbp->st_size; 1752 } 1753 1754 1755 #ifndef SMALL 1756 if (vflag && !tflag) 1757 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1758 #endif 1759 } 1760 1761 #ifndef SMALL 1762 /* this is used with -r to recursively descend directories */ 1763 static void 1764 handle_dir(char *dir) 1765 { 1766 char *path_argv[2]; 1767 FTS *fts; 1768 FTSENT *entry; 1769 1770 path_argv[0] = dir; 1771 path_argv[1] = 0; 1772 fts = fts_open(path_argv, FTS_PHYSICAL, NULL); 1773 if (fts == NULL) { 1774 warn("couldn't fts_open %s", dir); 1775 return; 1776 } 1777 1778 while ((entry = fts_read(fts))) { 1779 switch(entry->fts_info) { 1780 case FTS_D: 1781 case FTS_DP: 1782 continue; 1783 1784 case FTS_DNR: 1785 case FTS_ERR: 1786 case FTS_NS: 1787 maybe_warn("%s", entry->fts_path); 1788 continue; 1789 case FTS_F: 1790 handle_file(entry->fts_name, entry->fts_statp); 1791 } 1792 } 1793 (void)fts_close(fts); 1794 } 1795 #endif 1796 1797 /* print a ratio - size reduction as a fraction of uncompressed size */ 1798 static void 1799 print_ratio(off_t in, off_t out, FILE *where) 1800 { 1801 int percent10; /* 10 * percent */ 1802 off_t diff; 1803 char buff[8]; 1804 int len; 1805 1806 diff = in - out/2; 1807 if (diff <= 0) 1808 /* 1809 * Output is more than double size of input! print -99.9% 1810 * Quite possibly we've failed to get the original size. 1811 */ 1812 percent10 = -999; 1813 else { 1814 /* 1815 * We only need 12 bits of result from the final division, 1816 * so reduce the values until a 32bit division will suffice. 1817 */ 1818 while (in > 0x100000) { 1819 diff >>= 1; 1820 in >>= 1; 1821 } 1822 if (in != 0) 1823 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1824 else 1825 percent10 = 0; 1826 } 1827 1828 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1829 /* Move the '.' to before the last digit */ 1830 buff[len - 1] = buff[len - 2]; 1831 buff[len - 2] = '.'; 1832 fprintf(where, "%5s%%", buff); 1833 } 1834 1835 #ifndef SMALL 1836 /* print compression statistics, and the new name (if there is one!) */ 1837 static void 1838 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1839 { 1840 if (file) 1841 fprintf(stderr, "%s:%s ", file, 1842 strlen(file) < 7 ? "\t\t" : "\t"); 1843 print_ratio(usize, gsize, stderr); 1844 if (nfile) 1845 fprintf(stderr, " -- replaced with %s", nfile); 1846 fprintf(stderr, "\n"); 1847 fflush(stderr); 1848 } 1849 1850 /* print test results */ 1851 static void 1852 print_test(const char *file, int ok) 1853 { 1854 1855 if (exit_value == 0 && ok == 0) 1856 exit_value = 1; 1857 fprintf(stderr, "%s:%s %s\n", file, 1858 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1859 fflush(stderr); 1860 } 1861 #endif 1862 1863 /* print a file's info ala --list */ 1864 /* eg: 1865 compressed uncompressed ratio uncompressed_name 1866 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1867 */ 1868 static void 1869 print_list(int fd, off_t out, const char *outfile, time_t ts) 1870 { 1871 static int first = 1; 1872 #ifndef SMALL 1873 static off_t in_tot, out_tot; 1874 uint32_t crc = 0; 1875 #endif 1876 off_t in = 0, rv; 1877 1878 if (first) { 1879 #ifndef SMALL 1880 if (vflag) 1881 printf("method crc date time "); 1882 #endif 1883 if (qflag == 0) 1884 printf(" compressed uncompressed " 1885 "ratio uncompressed_name\n"); 1886 } 1887 first = 0; 1888 1889 /* print totals? */ 1890 #ifndef SMALL 1891 if (fd == -1) { 1892 in = in_tot; 1893 out = out_tot; 1894 } else 1895 #endif 1896 { 1897 /* read the last 4 bytes - this is the uncompressed size */ 1898 rv = lseek(fd, (off_t)(-8), SEEK_END); 1899 if (rv != -1) { 1900 unsigned char buf[8]; 1901 uint32_t usize; 1902 1903 rv = read(fd, (char *)buf, sizeof(buf)); 1904 if (rv == -1) 1905 maybe_warn("read of uncompressed size"); 1906 else if (rv != sizeof(buf)) 1907 maybe_warnx("read of uncompressed size"); 1908 1909 else { 1910 usize = buf[4] | buf[5] << 8 | 1911 buf[6] << 16 | buf[7] << 24; 1912 in = (off_t)usize; 1913 #ifndef SMALL 1914 crc = buf[0] | buf[1] << 8 | 1915 buf[2] << 16 | buf[3] << 24; 1916 #endif 1917 } 1918 } 1919 } 1920 1921 #ifndef SMALL 1922 if (vflag && fd == -1) 1923 printf(" "); 1924 else if (vflag) { 1925 char *date = ctime(&ts); 1926 1927 /* skip the day, 1/100th second, and year */ 1928 date += 4; 1929 date[12] = 0; 1930 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1931 } 1932 in_tot += in; 1933 out_tot += out; 1934 #endif 1935 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1936 print_ratio(in, out, stdout); 1937 printf(" %s\n", outfile); 1938 } 1939 1940 /* display the usage of NetBSD gzip */ 1941 static void 1942 usage(void) 1943 { 1944 1945 fprintf(stderr, "%s\n", gzip_version); 1946 fprintf(stderr, 1947 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n" 1948 #ifndef SMALL 1949 " -1 --fast fastest (worst) compression\n" 1950 " -2 .. -8 set compression level\n" 1951 " -9 --best best (slowest) compression\n" 1952 " -c --stdout write to stdout, keep original files\n" 1953 " --to-stdout\n" 1954 " -d --decompress uncompress files\n" 1955 " --uncompress\n" 1956 " -f --force force overwriting & compress links\n" 1957 " -h --help display this help\n" 1958 " -l --list list compressed file contents\n" 1959 " -N --name save or restore original file name and time stamp\n" 1960 " -n --no-name don't save original file name or time stamp\n" 1961 " -q --quiet output no warnings\n" 1962 " -r --recursive recursively compress files in directories\n" 1963 " -S .suf use suffix .suf instead of .gz\n" 1964 " --suffix .suf\n" 1965 " -t --test test compressed file\n" 1966 " -V --version display program version\n" 1967 " -v --verbose print extra statistics\n", 1968 #else 1969 , 1970 #endif 1971 getprogname()); 1972 exit(0); 1973 } 1974 1975 /* display the version of NetBSD gzip */ 1976 static void 1977 display_version(void) 1978 { 1979 1980 fprintf(stderr, "%s\n", gzip_version); 1981 exit(0); 1982 } 1983 1984 #ifndef NO_BZIP2_SUPPORT 1985 #include "unbzip2.c" 1986 #endif 1987 #ifndef NO_COMPRESS_SUPPORT 1988 #include "zuncompress.c" 1989 #endif 1990 1991 static ssize_t 1992 read_retry(int fd, void *buf, size_t sz) 1993 { 1994 char *cp = buf; 1995 size_t left = MIN(sz, (size_t) SSIZE_MAX); 1996 1997 while (left > 0) { 1998 ssize_t ret; 1999 2000 ret = read(fd, cp, left); 2001 if (ret == -1) { 2002 return ret; 2003 } else if (ret == 0) { 2004 break; /* EOF */ 2005 } 2006 cp += ret; 2007 left -= ret; 2008 } 2009 2010 return sz - left; 2011 } 2012
Indexes created Mon Sep 22 13:09:51 GMT 2025