1 /* $NetBSD: gzip.c,v 1.94 2009/04/12 10:31:14 lukem Exp $ */ 2 3 /* 4 * Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 21 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 22 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 23 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 24 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 26 * SUCH DAMAGE. 27 */ 28 29 /* 30 * gzip.c -- GPL free gzip using zlib. 31 * 32 * RFC 1950 covers the zlib format 33 * RFC 1951 covers the deflate format 34 * RFC 1952 covers the gzip format 35 * 36 * TODO: 37 * - use mmap where possible 38 * - handle some signals better (remove outfile?) 39 * - make bzip2/compress -v/-t/-l support work as well as possible 40 */ 41 42 #include <sys/param.h> 43 #include <sys/stat.h> 44 #include <sys/time.h> 45 46 #include <err.h> 47 #include <errno.h> 48 #include <fcntl.h> 49 #include <fts.h> 50 #include <getopt.h> 51 #include <inttypes.h> 52 #include <libgen.h> 53 #include <stdarg.h> 54 #include <stdio.h> 55 #include <stdlib.h> 56 #include <string.h> 57 #include <time.h> 58 #include <unistd.h> 59 #include <zlib.h> 60 61 #ifndef PRIdOFF 62 #define PRIdOFF PRId64 63 #endif 64 65 #ifndef PRId64 66 #define PRId64 "lld" 67 #endif 68 69 /* what type of file are we dealing with */ 70 enum filetype { 71 FT_GZIP, 72 #ifndef NO_BZIP2_SUPPORT 73 FT_BZIP2, 74 #endif 75 #ifndef NO_COMPRESS_SUPPORT 76 FT_Z, 77 #endif 78 #ifndef NO_PACK_SUPPORT 79 FT_PACK, 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 #ifndef NO_PACK_SUPPORT 98 #define PACK_MAGIC "\037\036" 99 #endif 100 101 #define GZ_SUFFIX ".gz" 102 103 #define BUFLEN (64 * 1024) 104 105 #define GZIP_MAGIC0 0x1F 106 #define GZIP_MAGIC1 0x8B 107 #define GZIP_OMAGIC1 0x9E 108 109 #define GZIP_TIMESTAMP (off_t)4 110 #define GZIP_ORIGNAME (off_t)10 111 112 #define HEAD_CRC 0x02 113 #define EXTRA_FIELD 0x04 114 #define ORIG_NAME 0x08 115 #define COMMENT 0x10 116 117 #define OS_CODE 3 /* Unix */ 118 119 typedef struct { 120 const char *zipped; 121 int ziplen; 122 const char *normal; /* for unzip - must not be longer than zipped */ 123 } suffixes_t; 124 static suffixes_t suffixes[] = { 125 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 126 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 127 #ifndef SMALL 128 SUFFIX(GZ_SUFFIX, ""), 129 SUFFIX(".z", ""), 130 SUFFIX("-gz", ""), 131 SUFFIX("-z", ""), 132 SUFFIX("_z", ""), 133 SUFFIX(".taz", ".tar"), 134 SUFFIX(".tgz", ".tar"), 135 #ifndef NO_BZIP2_SUPPORT 136 SUFFIX(BZ2_SUFFIX, ""), 137 #endif 138 #ifndef NO_COMPRESS_SUPPORT 139 SUFFIX(Z_SUFFIX, ""), 140 #endif 141 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 142 #endif /* SMALL */ 143 #undef SUFFIX 144 }; 145 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 146 147 #define SUFFIX_MAXLEN 30 148 149 static const char gzip_version[] = "NetBSD gzip 20060927"; 150 151 static int cflag; /* stdout mode */ 152 static int dflag; /* decompress mode */ 153 static int lflag; /* list mode */ 154 static int numflag = 6; /* gzip -1..-9 value */ 155 156 #ifndef SMALL 157 static int fflag; /* force mode */ 158 static int kflag; /* don't delete input files */ 159 static int nflag; /* don't save name/timestamp */ 160 static int Nflag; /* don't restore name/timestamp */ 161 static int qflag; /* quiet mode */ 162 static int rflag; /* recursive mode */ 163 static int tflag; /* test */ 164 static int vflag; /* verbose mode */ 165 #else 166 #define qflag 0 167 #define tflag 0 168 #endif 169 170 static int exit_value = 0; /* exit value */ 171 172 static char *infile; /* name of file coming in */ 173 174 static void maybe_err(const char *fmt, ...) 175 __attribute__((__format__(__printf__, 1, 2))); 176 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) 177 static void maybe_errx(const char *fmt, ...) 178 __attribute__((__format__(__printf__, 1, 2))); 179 #endif 180 static void maybe_warn(const char *fmt, ...) 181 __attribute__((__format__(__printf__, 1, 2))); 182 static void maybe_warnx(const char *fmt, ...) 183 __attribute__((__format__(__printf__, 1, 2))); 184 static enum filetype file_gettype(u_char *); 185 #ifdef SMALL 186 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 187 #endif 188 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 189 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 190 static off_t file_compress(char *, char *, size_t); 191 static off_t file_uncompress(char *, char *, size_t); 192 static void handle_pathname(char *); 193 static void handle_file(char *, struct stat *); 194 static void handle_stdin(void); 195 static void handle_stdout(void); 196 static void print_ratio(off_t, off_t, FILE *); 197 static void print_list(int fd, off_t, const char *, time_t); 198 static void usage(void); 199 static void display_version(void); 200 static const suffixes_t *check_suffix(char *, int); 201 static ssize_t read_retry(int, void *, size_t); 202 203 #ifdef SMALL 204 #define unlink_input(f, sb) unlink(f) 205 #else 206 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 207 static void prepend_gzip(char *, int *, char ***); 208 static void handle_dir(char *); 209 static void print_verbage(const char *, const char *, off_t, off_t); 210 static void print_test(const char *, int); 211 static void copymodes(int fd, const struct stat *, const char *file); 212 static int check_outfile(const char *outfile); 213 #endif 214 215 #ifndef NO_BZIP2_SUPPORT 216 static off_t unbzip2(int, int, char *, size_t, off_t *); 217 #endif 218 219 #ifndef NO_COMPRESS_SUPPORT 220 static FILE *zdopen(int); 221 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 222 #endif 223 224 #ifndef NO_PACK_SUPPORT 225 static off_t unpack(int, int, char *, size_t, off_t *); 226 #endif 227 228 #ifdef SMALL 229 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 230 #else 231 static const struct option longopts[] = { 232 { "stdout", no_argument, 0, 'c' }, 233 { "to-stdout", no_argument, 0, 'c' }, 234 { "decompress", no_argument, 0, 'd' }, 235 { "uncompress", no_argument, 0, 'd' }, 236 { "force", no_argument, 0, 'f' }, 237 { "help", no_argument, 0, 'h' }, 238 { "keep", no_argument, 0, 'k' }, 239 { "list", no_argument, 0, 'l' }, 240 { "no-name", no_argument, 0, 'n' }, 241 { "name", no_argument, 0, 'N' }, 242 { "quiet", no_argument, 0, 'q' }, 243 { "recursive", no_argument, 0, 'r' }, 244 { "suffix", required_argument, 0, 'S' }, 245 { "test", no_argument, 0, 't' }, 246 { "verbose", no_argument, 0, 'v' }, 247 { "version", no_argument, 0, 'V' }, 248 { "fast", no_argument, 0, '1' }, 249 { "best", no_argument, 0, '9' }, 250 #if 0 251 /* 252 * This is what else GNU gzip implements. --ascii isn't useful 253 * on NetBSD, and I don't care to have a --license. 254 */ 255 { "ascii", no_argument, 0, 'a' }, 256 { "license", no_argument, 0, 'L' }, 257 #endif 258 { NULL, no_argument, 0, 0 }, 259 }; 260 #endif 261 262 int 263 main(int argc, char **argv) 264 { 265 const char *progname = getprogname(); 266 #ifndef SMALL 267 char *gzip; 268 int len; 269 #endif 270 int ch; 271 272 /* XXX set up signals */ 273 274 #ifndef SMALL 275 if ((gzip = getenv("GZIP")) != NULL) 276 prepend_gzip(gzip, &argc, &argv); 277 #endif 278 279 /* 280 * XXX 281 * handle being called `gunzip', `zcat' and `gzcat' 282 */ 283 if (strcmp(progname, "gunzip") == 0) 284 dflag = 1; 285 else if (strcmp(progname, "zcat") == 0 || 286 strcmp(progname, "gzcat") == 0) 287 dflag = cflag = 1; 288 289 #ifdef SMALL 290 #define OPT_LIST "123456789cdhltV" 291 #else 292 #define OPT_LIST "123456789cdfhklNnqrS:tVv" 293 #endif 294 295 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 296 switch (ch) { 297 case '1': case '2': case '3': 298 case '4': case '5': case '6': 299 case '7': case '8': case '9': 300 numflag = ch - '0'; 301 break; 302 case 'c': 303 cflag = 1; 304 break; 305 case 'd': 306 dflag = 1; 307 break; 308 case 'l': 309 lflag = 1; 310 dflag = 1; 311 break; 312 case 'V': 313 display_version(); 314 /* NOTREACHED */ 315 #ifndef SMALL 316 case 'f': 317 fflag = 1; 318 break; 319 case 'k': 320 kflag = 1; 321 break; 322 case 'N': 323 nflag = 0; 324 Nflag = 1; 325 break; 326 case 'n': 327 nflag = 1; 328 Nflag = 0; 329 break; 330 case 'q': 331 qflag = 1; 332 break; 333 case 'r': 334 rflag = 1; 335 break; 336 case 'S': 337 len = strlen(optarg); 338 if (len != 0) { 339 if (len >= SUFFIX_MAXLEN) 340 errx(1, "incorrect suffix: '%s'", optarg); 341 suffixes[0].zipped = optarg; 342 suffixes[0].ziplen = len; 343 } else { 344 suffixes[NUM_SUFFIXES - 1].zipped = ""; 345 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 346 } 347 break; 348 case 't': 349 cflag = 1; 350 tflag = 1; 351 dflag = 1; 352 break; 353 case 'v': 354 vflag = 1; 355 break; 356 #endif 357 default: 358 usage(); 359 /* NOTREACHED */ 360 } 361 } 362 argv += optind; 363 argc -= optind; 364 365 if (argc == 0) { 366 if (dflag) /* stdin mode */ 367 handle_stdin(); 368 else /* stdout mode */ 369 handle_stdout(); 370 } else { 371 do { 372 handle_pathname(argv[0]); 373 } while (*++argv); 374 } 375 #ifndef SMALL 376 if (qflag == 0 && lflag && argc > 1) 377 print_list(-1, 0, "(totals)", 0); 378 #endif 379 exit(exit_value); 380 } 381 382 /* maybe print a warning */ 383 void 384 maybe_warn(const char *fmt, ...) 385 { 386 va_list ap; 387 388 if (qflag == 0) { 389 va_start(ap, fmt); 390 vwarn(fmt, ap); 391 va_end(ap); 392 } 393 if (exit_value == 0) 394 exit_value = 1; 395 } 396 397 /* ... without an errno. */ 398 void 399 maybe_warnx(const char *fmt, ...) 400 { 401 va_list ap; 402 403 if (qflag == 0) { 404 va_start(ap, fmt); 405 vwarnx(fmt, ap); 406 va_end(ap); 407 } 408 if (exit_value == 0) 409 exit_value = 1; 410 } 411 412 /* maybe print an error */ 413 void 414 maybe_err(const char *fmt, ...) 415 { 416 va_list ap; 417 418 if (qflag == 0) { 419 va_start(ap, fmt); 420 vwarn(fmt, ap); 421 va_end(ap); 422 } 423 exit(2); 424 } 425 426 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) 427 /* ... without an errno. */ 428 void 429 maybe_errx(const char *fmt, ...) 430 { 431 va_list ap; 432 433 if (qflag == 0) { 434 va_start(ap, fmt); 435 vwarnx(fmt, ap); 436 va_end(ap); 437 } 438 exit(2); 439 } 440 #endif 441 442 #ifndef SMALL 443 /* split up $GZIP and prepend it to the argument list */ 444 static void 445 prepend_gzip(char *gzip, int *argc, char ***argv) 446 { 447 char *s, **nargv, **ac; 448 int nenvarg = 0, i; 449 450 /* scan how many arguments there are */ 451 for (s = gzip;;) { 452 while (*s == ' ' || *s == '\t') 453 s++; 454 if (*s == 0) 455 goto count_done; 456 nenvarg++; 457 while (*s != ' ' && *s != '\t') 458 if (*s++ == 0) 459 goto count_done; 460 } 461 count_done: 462 /* punt early */ 463 if (nenvarg == 0) 464 return; 465 466 *argc += nenvarg; 467 ac = *argv; 468 469 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 470 if (nargv == NULL) 471 maybe_err("malloc"); 472 473 /* stash this away */ 474 *argv = nargv; 475 476 /* copy the program name first */ 477 i = 0; 478 nargv[i++] = *(ac++); 479 480 /* take a copy of $GZIP and add it to the array */ 481 s = strdup(gzip); 482 if (s == NULL) 483 maybe_err("strdup"); 484 for (;;) { 485 /* Skip whitespaces. */ 486 while (*s == ' ' || *s == '\t') 487 s++; 488 if (*s == 0) 489 goto copy_done; 490 nargv[i++] = s; 491 /* Find the end of this argument. */ 492 while (*s != ' ' && *s != '\t') 493 if (*s++ == 0) 494 /* Argument followed by NUL. */ 495 goto copy_done; 496 /* Terminate by overwriting ' ' or '\t' with NUL. */ 497 *s++ = 0; 498 } 499 copy_done: 500 501 /* copy the original arguments and a NULL */ 502 while (*ac) 503 nargv[i++] = *(ac++); 504 nargv[i] = NULL; 505 } 506 #endif 507 508 /* compress input to output. Return bytes read, -1 on error */ 509 static off_t 510 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 511 { 512 z_stream z; 513 char *outbufp, *inbufp; 514 off_t in_tot = 0, out_tot = 0; 515 ssize_t in_size; 516 int i, error; 517 uLong crc; 518 #ifdef SMALL 519 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 520 0, 0, 0, 0, 521 0, OS_CODE }; 522 #endif 523 524 outbufp = malloc(BUFLEN); 525 inbufp = malloc(BUFLEN); 526 if (outbufp == NULL || inbufp == NULL) { 527 maybe_err("malloc failed"); 528 goto out; 529 } 530 531 memset(&z, 0, sizeof z); 532 z.zalloc = Z_NULL; 533 z.zfree = Z_NULL; 534 z.opaque = 0; 535 536 #ifdef SMALL 537 memcpy(outbufp, header, sizeof header); 538 i = sizeof header; 539 #else 540 if (nflag != 0) { 541 mtime = 0; 542 origname = ""; 543 } 544 545 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 546 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 547 *origname ? ORIG_NAME : 0, 548 mtime & 0xff, 549 (mtime >> 8) & 0xff, 550 (mtime >> 16) & 0xff, 551 (mtime >> 24) & 0xff, 552 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 553 OS_CODE, origname); 554 if (i >= BUFLEN) 555 /* this need PATH_MAX > BUFLEN ... */ 556 maybe_err("snprintf"); 557 if (*origname) 558 i++; 559 #endif 560 561 z.next_out = outbufp + i; 562 z.avail_out = BUFLEN - i; 563 564 error = deflateInit2(&z, numflag, Z_DEFLATED, 565 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 566 if (error != Z_OK) { 567 maybe_warnx("deflateInit2 failed"); 568 in_tot = -1; 569 goto out; 570 } 571 572 crc = crc32(0L, Z_NULL, 0); 573 for (;;) { 574 if (z.avail_out == 0) { 575 if (write(out, outbufp, BUFLEN) != BUFLEN) { 576 maybe_warn("write"); 577 out_tot = -1; 578 goto out; 579 } 580 581 out_tot += BUFLEN; 582 z.next_out = outbufp; 583 z.avail_out = BUFLEN; 584 } 585 586 if (z.avail_in == 0) { 587 in_size = read(in, inbufp, BUFLEN); 588 if (in_size < 0) { 589 maybe_warn("read"); 590 in_tot = -1; 591 goto out; 592 } 593 if (in_size == 0) 594 break; 595 596 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 597 in_tot += in_size; 598 z.next_in = inbufp; 599 z.avail_in = in_size; 600 } 601 602 error = deflate(&z, Z_NO_FLUSH); 603 if (error != Z_OK && error != Z_STREAM_END) { 604 maybe_warnx("deflate failed"); 605 in_tot = -1; 606 goto out; 607 } 608 } 609 610 /* clean up */ 611 for (;;) { 612 size_t len; 613 ssize_t w; 614 615 error = deflate(&z, Z_FINISH); 616 if (error != Z_OK && error != Z_STREAM_END) { 617 maybe_warnx("deflate failed"); 618 in_tot = -1; 619 goto out; 620 } 621 622 len = (char *)z.next_out - outbufp; 623 624 w = write(out, outbufp, len); 625 if (w == -1 || (size_t)w != len) { 626 maybe_warn("write"); 627 out_tot = -1; 628 goto out; 629 } 630 out_tot += len; 631 z.next_out = outbufp; 632 z.avail_out = BUFLEN; 633 634 if (error == Z_STREAM_END) 635 break; 636 } 637 638 if (deflateEnd(&z) != Z_OK) { 639 maybe_warnx("deflateEnd failed"); 640 in_tot = -1; 641 goto out; 642 } 643 644 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 645 (int)crc & 0xff, 646 (int)(crc >> 8) & 0xff, 647 (int)(crc >> 16) & 0xff, 648 (int)(crc >> 24) & 0xff, 649 (int)in_tot & 0xff, 650 (int)(in_tot >> 8) & 0xff, 651 (int)(in_tot >> 16) & 0xff, 652 (int)(in_tot >> 24) & 0xff); 653 if (i != 8) 654 maybe_err("snprintf"); 655 #if 0 656 if (in_tot > 0xffffffff) 657 maybe_warn("input file size >= 4GB cannot be saved"); 658 #endif 659 if (write(out, outbufp, i) != i) { 660 maybe_warn("write"); 661 in_tot = -1; 662 } else 663 out_tot += i; 664 665 out: 666 if (inbufp != NULL) 667 free(inbufp); 668 if (outbufp != NULL) 669 free(outbufp); 670 if (gsizep) 671 *gsizep = out_tot; 672 return in_tot; 673 } 674 675 /* 676 * uncompress input to output then close the input. return the 677 * uncompressed size written, and put the compressed sized read 678 * into `*gsizep'. 679 */ 680 static off_t 681 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 682 const char *filename) 683 { 684 z_stream z; 685 char *outbufp, *inbufp; 686 off_t out_tot = -1, in_tot = 0; 687 uint32_t out_sub_tot = 0; 688 enum { 689 GZSTATE_MAGIC0, 690 GZSTATE_MAGIC1, 691 GZSTATE_METHOD, 692 GZSTATE_FLAGS, 693 GZSTATE_SKIPPING, 694 GZSTATE_EXTRA, 695 GZSTATE_EXTRA2, 696 GZSTATE_EXTRA3, 697 GZSTATE_ORIGNAME, 698 GZSTATE_COMMENT, 699 GZSTATE_HEAD_CRC1, 700 GZSTATE_HEAD_CRC2, 701 GZSTATE_INIT, 702 GZSTATE_READ, 703 GZSTATE_CRC, 704 GZSTATE_LEN, 705 } state = GZSTATE_MAGIC0; 706 int flags = 0, skip_count = 0; 707 int error = Z_STREAM_ERROR, done_reading = 0; 708 uLong crc = 0; 709 ssize_t wr; 710 int needmore = 0; 711 712 #define ADVANCE() { z.next_in++; z.avail_in--; } 713 714 if ((outbufp = malloc(BUFLEN)) == NULL) { 715 maybe_err("malloc failed"); 716 goto out2; 717 } 718 if ((inbufp = malloc(BUFLEN)) == NULL) { 719 maybe_err("malloc failed"); 720 goto out1; 721 } 722 723 memset(&z, 0, sizeof z); 724 z.avail_in = prelen; 725 z.next_in = pre; 726 z.avail_out = BUFLEN; 727 z.next_out = outbufp; 728 z.zalloc = NULL; 729 z.zfree = NULL; 730 z.opaque = 0; 731 732 in_tot = prelen; 733 out_tot = 0; 734 735 for (;;) { 736 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 737 ssize_t in_size; 738 739 if (z.avail_in > 0) { 740 memmove(inbufp, z.next_in, z.avail_in); 741 } 742 z.next_in = inbufp; 743 in_size = read(in, z.next_in + z.avail_in, 744 BUFLEN - z.avail_in); 745 746 if (in_size == -1) { 747 maybe_warn("failed to read stdin"); 748 goto stop_and_fail; 749 } else if (in_size == 0) { 750 done_reading = 1; 751 } 752 753 z.avail_in += in_size; 754 needmore = 0; 755 756 in_tot += in_size; 757 } 758 if (z.avail_in == 0) { 759 if (done_reading && state != GZSTATE_MAGIC0) { 760 maybe_warnx("%s: unexpected end of file", 761 filename); 762 goto stop_and_fail; 763 } 764 goto stop; 765 } 766 switch (state) { 767 case GZSTATE_MAGIC0: 768 if (*z.next_in != GZIP_MAGIC0) { 769 if (in_tot > 0) { 770 maybe_warnx("%s: trailing garbage " 771 "ignored", filename); 772 goto stop; 773 } 774 maybe_warnx("input not gziped (MAGIC0)"); 775 goto stop_and_fail; 776 } 777 ADVANCE(); 778 state++; 779 out_sub_tot = 0; 780 crc = crc32(0L, Z_NULL, 0); 781 break; 782 783 case GZSTATE_MAGIC1: 784 if (*z.next_in != GZIP_MAGIC1 && 785 *z.next_in != GZIP_OMAGIC1) { 786 maybe_warnx("input not gziped (MAGIC1)"); 787 goto stop_and_fail; 788 } 789 ADVANCE(); 790 state++; 791 break; 792 793 case GZSTATE_METHOD: 794 if (*z.next_in != Z_DEFLATED) { 795 maybe_warnx("unknown compression method"); 796 goto stop_and_fail; 797 } 798 ADVANCE(); 799 state++; 800 break; 801 802 case GZSTATE_FLAGS: 803 flags = *z.next_in; 804 ADVANCE(); 805 skip_count = 6; 806 state++; 807 break; 808 809 case GZSTATE_SKIPPING: 810 if (skip_count > 0) { 811 skip_count--; 812 ADVANCE(); 813 } else 814 state++; 815 break; 816 817 case GZSTATE_EXTRA: 818 if ((flags & EXTRA_FIELD) == 0) { 819 state = GZSTATE_ORIGNAME; 820 break; 821 } 822 skip_count = *z.next_in; 823 ADVANCE(); 824 state++; 825 break; 826 827 case GZSTATE_EXTRA2: 828 skip_count |= ((*z.next_in) << 8); 829 ADVANCE(); 830 state++; 831 break; 832 833 case GZSTATE_EXTRA3: 834 if (skip_count > 0) { 835 skip_count--; 836 ADVANCE(); 837 } else 838 state++; 839 break; 840 841 case GZSTATE_ORIGNAME: 842 if ((flags & ORIG_NAME) == 0) { 843 state++; 844 break; 845 } 846 if (*z.next_in == 0) 847 state++; 848 ADVANCE(); 849 break; 850 851 case GZSTATE_COMMENT: 852 if ((flags & COMMENT) == 0) { 853 state++; 854 break; 855 } 856 if (*z.next_in == 0) 857 state++; 858 ADVANCE(); 859 break; 860 861 case GZSTATE_HEAD_CRC1: 862 if (flags & HEAD_CRC) 863 skip_count = 2; 864 else 865 skip_count = 0; 866 state++; 867 break; 868 869 case GZSTATE_HEAD_CRC2: 870 if (skip_count > 0) { 871 skip_count--; 872 ADVANCE(); 873 } else 874 state++; 875 break; 876 877 case GZSTATE_INIT: 878 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 879 maybe_warnx("failed to inflateInit"); 880 goto stop_and_fail; 881 } 882 state++; 883 break; 884 885 case GZSTATE_READ: 886 error = inflate(&z, Z_FINISH); 887 switch (error) { 888 /* Z_BUF_ERROR goes with Z_FINISH... */ 889 case Z_BUF_ERROR: 890 if (z.avail_out > 0 && !done_reading) 891 continue; 892 case Z_STREAM_END: 893 case Z_OK: 894 break; 895 896 case Z_NEED_DICT: 897 maybe_warnx("Z_NEED_DICT error"); 898 goto stop_and_fail; 899 case Z_DATA_ERROR: 900 maybe_warnx("data stream error"); 901 goto stop_and_fail; 902 case Z_STREAM_ERROR: 903 maybe_warnx("internal stream error"); 904 goto stop_and_fail; 905 case Z_MEM_ERROR: 906 maybe_warnx("memory allocation error"); 907 goto stop_and_fail; 908 909 default: 910 maybe_warn("unknown error from inflate(): %d", 911 error); 912 } 913 wr = BUFLEN - z.avail_out; 914 915 if (wr != 0) { 916 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 917 if ( 918 #ifndef SMALL 919 /* don't write anything with -t */ 920 tflag == 0 && 921 #endif 922 write(out, outbufp, wr) != wr) { 923 maybe_warn("error writing to output"); 924 goto stop_and_fail; 925 } 926 927 out_tot += wr; 928 out_sub_tot += wr; 929 } 930 931 if (error == Z_STREAM_END) { 932 inflateEnd(&z); 933 state++; 934 } 935 936 z.next_out = outbufp; 937 z.avail_out = BUFLEN; 938 939 break; 940 case GZSTATE_CRC: 941 { 942 uLong origcrc; 943 944 if (z.avail_in < 4) { 945 if (!done_reading) { 946 needmore = 1; 947 continue; 948 } 949 maybe_warnx("truncated input"); 950 goto stop_and_fail; 951 } 952 origcrc = ((unsigned)z.next_in[0] & 0xff) | 953 ((unsigned)z.next_in[1] & 0xff) << 8 | 954 ((unsigned)z.next_in[2] & 0xff) << 16 | 955 ((unsigned)z.next_in[3] & 0xff) << 24; 956 if (origcrc != crc) { 957 maybe_warnx("invalid compressed" 958 " data--crc error"); 959 goto stop_and_fail; 960 } 961 } 962 963 z.avail_in -= 4; 964 z.next_in += 4; 965 966 if (!z.avail_in && done_reading) { 967 goto stop; 968 } 969 state++; 970 break; 971 case GZSTATE_LEN: 972 { 973 uLong origlen; 974 975 if (z.avail_in < 4) { 976 if (!done_reading) { 977 needmore = 1; 978 continue; 979 } 980 maybe_warnx("truncated input"); 981 goto stop_and_fail; 982 } 983 origlen = ((unsigned)z.next_in[0] & 0xff) | 984 ((unsigned)z.next_in[1] & 0xff) << 8 | 985 ((unsigned)z.next_in[2] & 0xff) << 16 | 986 ((unsigned)z.next_in[3] & 0xff) << 24; 987 988 if (origlen != out_sub_tot) { 989 maybe_warnx("invalid compressed" 990 " data--length error"); 991 goto stop_and_fail; 992 } 993 } 994 995 z.avail_in -= 4; 996 z.next_in += 4; 997 998 if (error < 0) { 999 maybe_warnx("decompression error"); 1000 goto stop_and_fail; 1001 } 1002 state = GZSTATE_MAGIC0; 1003 break; 1004 } 1005 continue; 1006 stop_and_fail: 1007 out_tot = -1; 1008 stop: 1009 break; 1010 } 1011 if (state > GZSTATE_INIT) 1012 inflateEnd(&z); 1013 1014 free(inbufp); 1015 out1: 1016 free(outbufp); 1017 out2: 1018 if (gsizep) 1019 *gsizep = in_tot; 1020 return (out_tot); 1021 } 1022 1023 #ifndef SMALL 1024 /* 1025 * set the owner, mode, flags & utimes using the given file descriptor. 1026 * file is only used in possible warning messages. 1027 */ 1028 static void 1029 copymodes(int fd, const struct stat *sbp, const char *file) 1030 { 1031 struct timeval times[2]; 1032 struct stat sb; 1033 1034 /* 1035 * If we have no info on the input, give this file some 1036 * default values and return.. 1037 */ 1038 if (sbp == NULL) { 1039 mode_t mask = umask(022); 1040 1041 (void)fchmod(fd, DEFFILEMODE & ~mask); 1042 (void)umask(mask); 1043 return; 1044 } 1045 sb = *sbp; 1046 1047 /* if the chown fails, remove set-id bits as-per compress(1) */ 1048 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1049 if (errno != EPERM) 1050 maybe_warn("couldn't fchown: %s", file); 1051 sb.st_mode &= ~(S_ISUID|S_ISGID); 1052 } 1053 1054 /* we only allow set-id and the 9 normal permission bits */ 1055 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1056 if (fchmod(fd, sb.st_mode) < 0) 1057 maybe_warn("couldn't fchmod: %s", file); 1058 1059 /* only try flags if they exist already */ 1060 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1061 maybe_warn("couldn't fchflags: %s", file); 1062 1063 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1064 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1065 if (futimes(fd, times) < 0) 1066 maybe_warn("couldn't utimes: %s", file); 1067 } 1068 #endif 1069 1070 /* what sort of file is this? */ 1071 static enum filetype 1072 file_gettype(u_char *buf) 1073 { 1074 1075 if (buf[0] == GZIP_MAGIC0 && 1076 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1077 return FT_GZIP; 1078 else 1079 #ifndef NO_BZIP2_SUPPORT 1080 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1081 buf[3] >= '0' && buf[3] <= '9') 1082 return FT_BZIP2; 1083 else 1084 #endif 1085 #ifndef NO_COMPRESS_SUPPORT 1086 if (memcmp(buf, Z_MAGIC, 2) == 0) 1087 return FT_Z; 1088 else 1089 #endif 1090 #ifndef NO_PACK_SUPPORT 1091 if (memcmp(buf, PACK_MAGIC, 2) == 0) 1092 return FT_PACK; 1093 else 1094 #endif 1095 return FT_UNKNOWN; 1096 } 1097 1098 #ifndef SMALL 1099 /* check the outfile is OK. */ 1100 static int 1101 check_outfile(const char *outfile) 1102 { 1103 struct stat sb; 1104 int ok = 1; 1105 1106 if (lflag == 0 && stat(outfile, &sb) == 0) { 1107 if (fflag) 1108 unlink(outfile); 1109 else if (isatty(STDIN_FILENO)) { 1110 char ans[10] = { 'n', '\0' }; /* default */ 1111 1112 fprintf(stderr, "%s already exists -- do you wish to " 1113 "overwrite (y or n)? " , outfile); 1114 (void)fgets(ans, sizeof(ans) - 1, stdin); 1115 if (ans[0] != 'y' && ans[0] != 'Y') { 1116 fprintf(stderr, "\tnot overwriting\n"); 1117 ok = 0; 1118 } else 1119 unlink(outfile); 1120 } else { 1121 maybe_warnx("%s already exists -- skipping", outfile); 1122 ok = 0; 1123 } 1124 } 1125 return ok; 1126 } 1127 1128 static void 1129 unlink_input(const char *file, const struct stat *sb) 1130 { 1131 struct stat nsb; 1132 1133 if (kflag) 1134 return; 1135 if (stat(file, &nsb) != 0) 1136 /* Must be gone alrady */ 1137 return; 1138 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1139 /* Definitely a different file */ 1140 return; 1141 unlink(file); 1142 } 1143 #endif 1144 1145 static const suffixes_t * 1146 check_suffix(char *file, int xlate) 1147 { 1148 const suffixes_t *s; 1149 int len = strlen(file); 1150 char *sp; 1151 1152 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1153 /* if it doesn't fit in "a.suf", don't bother */ 1154 if (s->ziplen >= len) 1155 continue; 1156 sp = file + len - s->ziplen; 1157 if (strcmp(s->zipped, sp) != 0) 1158 continue; 1159 if (xlate) 1160 strcpy(sp, s->normal); 1161 return s; 1162 } 1163 return NULL; 1164 } 1165 1166 /* 1167 * compress the given file: create a corresponding .gz file and remove the 1168 * original. 1169 */ 1170 static off_t 1171 file_compress(char *file, char *outfile, size_t outsize) 1172 { 1173 int in; 1174 int out; 1175 off_t size, insize; 1176 #ifndef SMALL 1177 struct stat isb, osb; 1178 const suffixes_t *suff; 1179 #endif 1180 1181 in = open(file, O_RDONLY); 1182 if (in == -1) { 1183 maybe_warn("can't open %s", file); 1184 return -1; 1185 } 1186 1187 if (cflag == 0) { 1188 #ifndef SMALL 1189 if (fstat(in, &isb) == 0) { 1190 if (isb.st_nlink > 1 && fflag == 0) { 1191 maybe_warnx("%s has %d other link%s -- " 1192 "skipping", file, isb.st_nlink - 1, 1193 isb.st_nlink == 1 ? "" : "s"); 1194 close(in); 1195 return -1; 1196 } 1197 } 1198 1199 if (fflag == 0 && (suff = check_suffix(file, 0)) 1200 && suff->zipped[0] != 0) { 1201 maybe_warnx("%s already has %s suffix -- unchanged", 1202 file, suff->zipped); 1203 close(in); 1204 return -1; 1205 } 1206 #endif 1207 1208 /* Add (usually) .gz to filename */ 1209 if ((size_t)snprintf(outfile, outsize, "%s%s", 1210 file, suffixes[0].zipped) >= outsize) { 1211 errx(1, "file path too long: %s", file); 1212 } 1213 #ifndef SMALL 1214 if (check_outfile(outfile) == 0) { 1215 close(in); 1216 return -1; 1217 } 1218 #endif 1219 } 1220 1221 if (cflag == 0) { 1222 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1223 if (out == -1) { 1224 maybe_warn("could not create output: %s", outfile); 1225 fclose(stdin); 1226 return -1; 1227 } 1228 } else 1229 out = STDOUT_FILENO; 1230 1231 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1232 1233 (void)close(in); 1234 1235 /* 1236 * If there was an error, insize will be -1. 1237 * If we compressed to stdout, just return the size. 1238 * Otherwise stat the file and check it is the correct size. 1239 * We only blow away the file if we can stat the output and it 1240 * has the expected size. 1241 */ 1242 if (cflag != 0) 1243 return insize == -1 ? -1 : size; 1244 1245 #ifndef SMALL 1246 if (fstat(out, &osb) != 0) { 1247 maybe_warn("couldn't stat: %s", outfile); 1248 goto bad_outfile; 1249 } 1250 1251 if (osb.st_size != size) { 1252 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1253 " != %" PRIdOFF "), deleting", 1254 outfile, osb.st_size, size); 1255 goto bad_outfile; 1256 } 1257 1258 copymodes(out, &isb, outfile); 1259 #endif 1260 if (close(out) == -1) 1261 maybe_warn("couldn't close output"); 1262 1263 /* output is good, ok to delete input */ 1264 unlink_input(file, &isb); 1265 return size; 1266 1267 #ifndef SMALL 1268 bad_outfile: 1269 if (close(out) == -1) 1270 maybe_warn("couldn't close output"); 1271 1272 maybe_warnx("leaving original %s", file); 1273 unlink(outfile); 1274 return size; 1275 #endif 1276 } 1277 1278 /* uncompress the given file and remove the original */ 1279 static off_t 1280 file_uncompress(char *file, char *outfile, size_t outsize) 1281 { 1282 struct stat isb, osb; 1283 off_t size; 1284 ssize_t rbytes; 1285 unsigned char header1[4]; 1286 enum filetype method; 1287 int fd, ofd, zfd = -1; 1288 #ifndef SMALL 1289 ssize_t rv; 1290 time_t timestamp = 0; 1291 unsigned char name[PATH_MAX + 1]; 1292 #endif 1293 1294 /* gather the old name info */ 1295 1296 fd = open(file, O_RDONLY); 1297 if (fd < 0) { 1298 maybe_warn("can't open %s", file); 1299 goto lose; 1300 } 1301 1302 if ((size_t)snprintf(outfile, outsize, "%s", file) >= outsize) 1303 errx(1, "file path too long: %s", file); 1304 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1305 maybe_warnx("%s: unknown suffix -- ignored", file); 1306 goto lose; 1307 } 1308 1309 rbytes = read(fd, header1, sizeof header1); 1310 if (rbytes != sizeof header1) { 1311 /* we don't want to fail here. */ 1312 #ifndef SMALL 1313 if (fflag) 1314 goto lose; 1315 #endif 1316 if (rbytes == -1) 1317 maybe_warn("can't read %s", file); 1318 else 1319 goto unexpected_EOF; 1320 goto lose; 1321 } 1322 1323 method = file_gettype(header1); 1324 1325 #ifndef SMALL 1326 if (fflag == 0 && method == FT_UNKNOWN) { 1327 maybe_warnx("%s: not in gzip format", file); 1328 goto lose; 1329 } 1330 1331 #endif 1332 1333 #ifndef SMALL 1334 if (method == FT_GZIP && Nflag) { 1335 unsigned char ts[4]; /* timestamp */ 1336 1337 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1338 if (rv >= 0 && rv < (ssize_t)(sizeof ts)) 1339 goto unexpected_EOF; 1340 if (rv == -1) { 1341 if (!fflag) 1342 maybe_warn("can't read %s", file); 1343 goto lose; 1344 } 1345 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1346 1347 if (header1[3] & ORIG_NAME) { 1348 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1349 if (rbytes < 0) { 1350 maybe_warn("can't read %s", file); 1351 goto lose; 1352 } 1353 if (name[0] != 0) { 1354 /* preserve original directory name */ 1355 char *dp = strrchr(file, '/'); 1356 if (dp == NULL) 1357 dp = file; 1358 else 1359 dp++; 1360 snprintf(outfile, outsize, "%.*s%.*s", 1361 (int) (dp - file), 1362 file, (int) rbytes, name); 1363 } 1364 } 1365 } 1366 #endif 1367 lseek(fd, 0, SEEK_SET); 1368 1369 if (cflag == 0 || lflag) { 1370 if (fstat(fd, &isb) != 0) 1371 goto lose; 1372 #ifndef SMALL 1373 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1374 maybe_warnx("%s has %d other links -- skipping", 1375 file, isb.st_nlink - 1); 1376 goto lose; 1377 } 1378 if (nflag == 0 && timestamp) 1379 isb.st_mtime = timestamp; 1380 if (check_outfile(outfile) == 0) 1381 goto lose; 1382 #endif 1383 } 1384 1385 if (cflag == 0 && lflag == 0) { 1386 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1387 if (zfd == STDOUT_FILENO) { 1388 /* We won't close STDOUT_FILENO later... */ 1389 zfd = dup(zfd); 1390 close(STDOUT_FILENO); 1391 } 1392 if (zfd == -1) { 1393 maybe_warn("can't open %s", outfile); 1394 goto lose; 1395 } 1396 } else 1397 zfd = STDOUT_FILENO; 1398 1399 #ifndef NO_BZIP2_SUPPORT 1400 if (method == FT_BZIP2) { 1401 1402 /* XXX */ 1403 if (lflag) { 1404 maybe_warnx("no -l with bzip2 files"); 1405 goto lose; 1406 } 1407 1408 size = unbzip2(fd, zfd, NULL, 0, NULL); 1409 } else 1410 #endif 1411 1412 #ifndef NO_COMPRESS_SUPPORT 1413 if (method == FT_Z) { 1414 FILE *in, *out; 1415 1416 /* XXX */ 1417 if (lflag) { 1418 maybe_warnx("no -l with Lempel-Ziv files"); 1419 goto lose; 1420 } 1421 1422 if ((in = zdopen(fd)) == NULL) { 1423 maybe_warn("zdopen for read: %s", file); 1424 goto lose; 1425 } 1426 1427 out = fdopen(dup(zfd), "w"); 1428 if (out == NULL) { 1429 maybe_warn("fdopen for write: %s", outfile); 1430 fclose(in); 1431 goto lose; 1432 } 1433 1434 size = zuncompress(in, out, NULL, 0, NULL); 1435 /* need to fclose() if ferror() is true... */ 1436 if (ferror(in) | fclose(in)) { 1437 maybe_warn("failed infile fclose"); 1438 unlink(outfile); 1439 (void)fclose(out); 1440 } 1441 if (fclose(out) != 0) { 1442 maybe_warn("failed outfile fclose"); 1443 unlink(outfile); 1444 goto lose; 1445 } 1446 } else 1447 #endif 1448 1449 #ifndef NO_PACK_SUPPORT 1450 if (method == FT_PACK) { 1451 if (lflag) { 1452 maybe_warnx("no -l with packed files"); 1453 goto lose; 1454 } 1455 1456 size = unpack(fd, zfd, NULL, 0, NULL); 1457 } else 1458 #endif 1459 1460 #ifndef SMALL 1461 if (method == FT_UNKNOWN) { 1462 if (lflag) { 1463 maybe_warnx("no -l for unknown filetypes"); 1464 goto lose; 1465 } 1466 size = cat_fd(NULL, 0, NULL, fd); 1467 } else 1468 #endif 1469 { 1470 if (lflag) { 1471 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1472 close(fd); 1473 return -1; /* XXX */ 1474 } 1475 1476 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1477 } 1478 1479 if (close(fd) != 0) 1480 maybe_warn("couldn't close input"); 1481 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1482 maybe_warn("couldn't close output"); 1483 1484 if (size == -1) { 1485 if (cflag == 0) 1486 unlink(outfile); 1487 maybe_warnx("%s: uncompress failed", file); 1488 return -1; 1489 } 1490 1491 /* if testing, or we uncompressed to stdout, this is all we need */ 1492 #ifndef SMALL 1493 if (tflag) 1494 return size; 1495 #endif 1496 /* if we are uncompressing to stdin, don't remove the file. */ 1497 if (cflag) 1498 return size; 1499 1500 /* 1501 * if we create a file... 1502 */ 1503 /* 1504 * if we can't stat the file don't remove the file. 1505 */ 1506 1507 ofd = open(outfile, O_RDWR, 0); 1508 if (ofd == -1) { 1509 maybe_warn("couldn't open (leaving original): %s", 1510 outfile); 1511 return -1; 1512 } 1513 if (fstat(ofd, &osb) != 0) { 1514 maybe_warn("couldn't stat (leaving original): %s", 1515 outfile); 1516 close(ofd); 1517 return -1; 1518 } 1519 if (osb.st_size != size) { 1520 maybe_warnx("stat gave different size: %" PRIdOFF 1521 " != %" PRIdOFF " (leaving original)", 1522 size, osb.st_size); 1523 close(ofd); 1524 unlink(outfile); 1525 return -1; 1526 } 1527 unlink_input(file, &isb); 1528 #ifndef SMALL 1529 copymodes(ofd, &isb, outfile); 1530 #endif 1531 close(ofd); 1532 return size; 1533 1534 unexpected_EOF: 1535 maybe_warnx("%s: unexpected end of file", file); 1536 lose: 1537 if (fd != -1) 1538 close(fd); 1539 if (zfd != -1 && zfd != STDOUT_FILENO) 1540 close(fd); 1541 return -1; 1542 } 1543 1544 #ifndef SMALL 1545 static off_t 1546 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1547 { 1548 char buf[BUFLEN]; 1549 off_t in_tot; 1550 ssize_t w; 1551 1552 in_tot = count; 1553 w = write(STDOUT_FILENO, prepend, count); 1554 if (w == -1 || (size_t)w != count) { 1555 maybe_warn("write to stdout"); 1556 return -1; 1557 } 1558 for (;;) { 1559 ssize_t rv; 1560 1561 rv = read(fd, buf, sizeof buf); 1562 if (rv == 0) 1563 break; 1564 if (rv < 0) { 1565 maybe_warn("read from fd %d", fd); 1566 break; 1567 } 1568 1569 if (write(STDOUT_FILENO, buf, rv) != rv) { 1570 maybe_warn("write to stdout"); 1571 break; 1572 } 1573 in_tot += rv; 1574 } 1575 1576 if (gsizep) 1577 *gsizep = in_tot; 1578 return (in_tot); 1579 } 1580 #endif 1581 1582 static void 1583 handle_stdin(void) 1584 { 1585 unsigned char header1[4]; 1586 off_t usize, gsize; 1587 enum filetype method; 1588 ssize_t bytes_read; 1589 #ifndef NO_COMPRESS_SUPPORT 1590 FILE *in; 1591 #endif 1592 1593 #ifndef SMALL 1594 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1595 maybe_warnx("standard input is a terminal -- ignoring"); 1596 return; 1597 } 1598 #endif 1599 1600 if (lflag) { 1601 struct stat isb; 1602 1603 /* XXX could read the whole file, etc. */ 1604 if (fstat(STDIN_FILENO, &isb) < 0) { 1605 maybe_warn("fstat"); 1606 return; 1607 } 1608 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1609 return; 1610 } 1611 1612 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1613 if (bytes_read == -1) { 1614 maybe_warn("can't read stdin"); 1615 return; 1616 } else if (bytes_read != sizeof(header1)) { 1617 maybe_warnx("(stdin): unexpected end of file"); 1618 return; 1619 } 1620 1621 method = file_gettype(header1); 1622 switch (method) { 1623 default: 1624 #ifndef SMALL 1625 if (fflag == 0) { 1626 maybe_warnx("unknown compression format"); 1627 return; 1628 } 1629 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1630 break; 1631 #endif 1632 case FT_GZIP: 1633 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1634 header1, sizeof header1, &gsize, "(stdin)"); 1635 break; 1636 #ifndef NO_BZIP2_SUPPORT 1637 case FT_BZIP2: 1638 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1639 header1, sizeof header1, &gsize); 1640 break; 1641 #endif 1642 #ifndef NO_COMPRESS_SUPPORT 1643 case FT_Z: 1644 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1645 maybe_warnx("zopen of stdin"); 1646 return; 1647 } 1648 1649 usize = zuncompress(in, stdout, header1, sizeof header1, &gsize); 1650 fclose(in); 1651 break; 1652 #endif 1653 #ifndef NO_PACK_SUPPORT 1654 case FT_PACK: 1655 usize = unpack(STDIN_FILENO, STDOUT_FILENO, 1656 (char *)header1, sizeof header1, &gsize); 1657 break; 1658 #endif 1659 } 1660 1661 #ifndef SMALL 1662 if (vflag && !tflag && usize != -1 && gsize != -1) 1663 print_verbage(NULL, NULL, usize, gsize); 1664 if (vflag && tflag) 1665 print_test("(stdin)", usize != -1); 1666 #endif 1667 1668 } 1669 1670 static void 1671 handle_stdout(void) 1672 { 1673 off_t gsize, usize; 1674 struct stat sb; 1675 time_t systime; 1676 uint32_t mtime; 1677 int ret; 1678 1679 #ifndef SMALL 1680 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1681 maybe_warnx("standard output is a terminal -- ignoring"); 1682 return; 1683 } 1684 #endif 1685 /* If stdin is a file use it's mtime, otherwise use current time */ 1686 ret = fstat(STDIN_FILENO, &sb); 1687 1688 #ifndef SMALL 1689 if (ret < 0) { 1690 maybe_warn("Can't stat stdin"); 1691 return; 1692 } 1693 #endif 1694 1695 if (S_ISREG(sb.st_mode)) 1696 mtime = (uint32_t)sb.st_mtime; 1697 else { 1698 systime = time(NULL); 1699 #ifndef SMALL 1700 if (systime == -1) { 1701 maybe_warn("time"); 1702 return; 1703 } 1704 #endif 1705 mtime = (uint32_t)systime; 1706 } 1707 1708 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1709 #ifndef SMALL 1710 if (vflag && !tflag && usize != -1 && gsize != -1) 1711 print_verbage(NULL, NULL, usize, gsize); 1712 #endif 1713 } 1714 1715 /* do what is asked for, for the path name */ 1716 static void 1717 handle_pathname(char *path) 1718 { 1719 char *opath = path, *s = NULL; 1720 ssize_t len; 1721 int slen; 1722 struct stat sb; 1723 1724 /* check for stdout/stdin */ 1725 if (path[0] == '-' && path[1] == '\0') { 1726 if (dflag) 1727 handle_stdin(); 1728 else 1729 handle_stdout(); 1730 return; 1731 } 1732 1733 retry: 1734 if (stat(path, &sb) != 0) { 1735 /* lets try <path>.gz if we're decompressing */ 1736 if (dflag && s == NULL && errno == ENOENT) { 1737 len = strlen(path); 1738 slen = suffixes[0].ziplen; 1739 s = malloc(len + slen + 1); 1740 if (s == NULL) 1741 maybe_err("malloc"); 1742 memcpy(s, path, len); 1743 memcpy(s + len, suffixes[0].zipped, slen + 1); 1744 path = s; 1745 goto retry; 1746 } 1747 maybe_warn("can't stat: %s", opath); 1748 goto out; 1749 } 1750 1751 if (S_ISDIR(sb.st_mode)) { 1752 #ifndef SMALL 1753 if (rflag) 1754 handle_dir(path); 1755 else 1756 #endif 1757 maybe_warnx("%s is a directory", path); 1758 goto out; 1759 } 1760 1761 if (S_ISREG(sb.st_mode)) 1762 handle_file(path, &sb); 1763 else 1764 maybe_warnx("%s is not a regular file", path); 1765 1766 out: 1767 if (s) 1768 free(s); 1769 } 1770 1771 /* compress/decompress a file */ 1772 static void 1773 handle_file(char *file, struct stat *sbp) 1774 { 1775 off_t usize, gsize; 1776 char outfile[PATH_MAX]; 1777 1778 infile = file; 1779 if (dflag) { 1780 usize = file_uncompress(file, outfile, sizeof(outfile)); 1781 #ifndef SMALL 1782 if (vflag && tflag) 1783 print_test(file, usize != -1); 1784 #endif 1785 if (usize == -1) 1786 return; 1787 gsize = sbp->st_size; 1788 } else { 1789 gsize = file_compress(file, outfile, sizeof(outfile)); 1790 if (gsize == -1) 1791 return; 1792 usize = sbp->st_size; 1793 } 1794 1795 1796 #ifndef SMALL 1797 if (vflag && !tflag) 1798 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1799 #endif 1800 } 1801 1802 #ifndef SMALL 1803 /* this is used with -r to recursively descend directories */ 1804 static void 1805 handle_dir(char *dir) 1806 { 1807 char *path_argv[2]; 1808 FTS *fts; 1809 FTSENT *entry; 1810 1811 path_argv[0] = dir; 1812 path_argv[1] = NULL; 1813 fts = fts_open(path_argv, FTS_PHYSICAL | FTS_NOCHDIR, NULL); 1814 if (fts == NULL) { 1815 warn("couldn't fts_open %s", dir); 1816 return; 1817 } 1818 1819 while ((entry = fts_read(fts))) { 1820 switch(entry->fts_info) { 1821 case FTS_D: 1822 case FTS_DP: 1823 continue; 1824 1825 case FTS_DNR: 1826 case FTS_ERR: 1827 case FTS_NS: 1828 maybe_warn("%s", entry->fts_path); 1829 continue; 1830 case FTS_F: 1831 handle_file(entry->fts_path, entry->fts_statp); 1832 } 1833 } 1834 (void)fts_close(fts); 1835 } 1836 #endif 1837 1838 /* print a ratio - size reduction as a fraction of uncompressed size */ 1839 static void 1840 print_ratio(off_t in, off_t out, FILE *where) 1841 { 1842 int percent10; /* 10 * percent */ 1843 off_t diff; 1844 char buff[8]; 1845 int len; 1846 1847 diff = in - out/2; 1848 if (diff <= 0) 1849 /* 1850 * Output is more than double size of input! print -99.9% 1851 * Quite possibly we've failed to get the original size. 1852 */ 1853 percent10 = -999; 1854 else { 1855 /* 1856 * We only need 12 bits of result from the final division, 1857 * so reduce the values until a 32bit division will suffice. 1858 */ 1859 while (in > 0x100000) { 1860 diff >>= 1; 1861 in >>= 1; 1862 } 1863 if (in != 0) 1864 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1865 else 1866 percent10 = 0; 1867 } 1868 1869 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1870 /* Move the '.' to before the last digit */ 1871 buff[len - 1] = buff[len - 2]; 1872 buff[len - 2] = '.'; 1873 fprintf(where, "%5s%%", buff); 1874 } 1875 1876 #ifndef SMALL 1877 /* print compression statistics, and the new name (if there is one!) */ 1878 static void 1879 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1880 { 1881 if (file) 1882 fprintf(stderr, "%s:%s ", file, 1883 strlen(file) < 7 ? "\t\t" : "\t"); 1884 print_ratio(usize, gsize, stderr); 1885 if (nfile) 1886 fprintf(stderr, " -- replaced with %s", nfile); 1887 fprintf(stderr, "\n"); 1888 fflush(stderr); 1889 } 1890 1891 /* print test results */ 1892 static void 1893 print_test(const char *file, int ok) 1894 { 1895 1896 if (exit_value == 0 && ok == 0) 1897 exit_value = 1; 1898 fprintf(stderr, "%s:%s %s\n", file, 1899 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1900 fflush(stderr); 1901 } 1902 #endif 1903 1904 /* print a file's info ala --list */ 1905 /* eg: 1906 compressed uncompressed ratio uncompressed_name 1907 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1908 */ 1909 static void 1910 print_list(int fd, off_t out, const char *outfile, time_t ts) 1911 { 1912 static int first = 1; 1913 #ifndef SMALL 1914 static off_t in_tot, out_tot; 1915 uint32_t crc = 0; 1916 #endif 1917 off_t in = 0, rv; 1918 1919 if (first) { 1920 #ifndef SMALL 1921 if (vflag) 1922 printf("method crc date time "); 1923 #endif 1924 if (qflag == 0) 1925 printf(" compressed uncompressed " 1926 "ratio uncompressed_name\n"); 1927 } 1928 first = 0; 1929 1930 /* print totals? */ 1931 #ifndef SMALL 1932 if (fd == -1) { 1933 in = in_tot; 1934 out = out_tot; 1935 } else 1936 #endif 1937 { 1938 /* read the last 4 bytes - this is the uncompressed size */ 1939 rv = lseek(fd, (off_t)(-8), SEEK_END); 1940 if (rv != -1) { 1941 unsigned char buf[8]; 1942 uint32_t usize; 1943 1944 rv = read(fd, (char *)buf, sizeof(buf)); 1945 if (rv == -1) 1946 maybe_warn("read of uncompressed size"); 1947 else if (rv != sizeof(buf)) 1948 maybe_warnx("read of uncompressed size"); 1949 1950 else { 1951 usize = buf[4] | buf[5] << 8 | 1952 buf[6] << 16 | buf[7] << 24; 1953 in = (off_t)usize; 1954 #ifndef SMALL 1955 crc = buf[0] | buf[1] << 8 | 1956 buf[2] << 16 | buf[3] << 24; 1957 #endif 1958 } 1959 } 1960 } 1961 1962 #ifndef SMALL 1963 if (vflag && fd == -1) 1964 printf(" "); 1965 else if (vflag) { 1966 char *date = ctime(&ts); 1967 1968 /* skip the day, 1/100th second, and year */ 1969 date += 4; 1970 date[12] = 0; 1971 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1972 } 1973 in_tot += in; 1974 out_tot += out; 1975 #endif 1976 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1977 print_ratio(in, out, stdout); 1978 printf(" %s\n", outfile); 1979 } 1980 1981 /* display the usage of NetBSD gzip */ 1982 static void 1983 usage(void) 1984 { 1985 1986 fprintf(stderr, "%s\n", gzip_version); 1987 fprintf(stderr, 1988 #ifdef SMALL 1989 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n", 1990 #else 1991 "usage: %s [-123456789acdfhklLNnqrtVv] [-S .suffix] [<file> [<file> ...]]\n" 1992 " -1 --fast fastest (worst) compression\n" 1993 " -2 .. -8 set compression level\n" 1994 " -9 --best best (slowest) compression\n" 1995 " -c --stdout write to stdout, keep original files\n" 1996 " --to-stdout\n" 1997 " -d --decompress uncompress files\n" 1998 " --uncompress\n" 1999 " -f --force force overwriting & compress links\n" 2000 " -h --help display this help\n" 2001 " -k --keep don't delete input files during operation\n" 2002 " -l --list list compressed file contents\n" 2003 " -N --name save or restore original file name and time stamp\n" 2004 " -n --no-name don't save original file name or time stamp\n" 2005 " -q --quiet output no warnings\n" 2006 " -r --recursive recursively compress files in directories\n" 2007 " -S .suf use suffix .suf instead of .gz\n" 2008 " --suffix .suf\n" 2009 " -t --test test compressed file\n" 2010 " -V --version display program version\n" 2011 " -v --verbose print extra statistics\n", 2012 #endif 2013 getprogname()); 2014 exit(0); 2015 } 2016 2017 /* display the version of NetBSD gzip */ 2018 static void 2019 display_version(void) 2020 { 2021 2022 fprintf(stderr, "%s\n", gzip_version); 2023 exit(0); 2024 } 2025 2026 #ifndef NO_BZIP2_SUPPORT 2027 #include "unbzip2.c" 2028 #endif 2029 #ifndef NO_COMPRESS_SUPPORT 2030 #include "zuncompress.c" 2031 #endif 2032 #ifndef NO_PACK_SUPPORT 2033 #include "unpack.c" 2034 #endif 2035 2036 static ssize_t 2037 read_retry(int fd, void *buf, size_t sz) 2038 { 2039 char *cp = buf; 2040 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2041 2042 while (left > 0) { 2043 ssize_t ret; 2044 2045 ret = read(fd, cp, left); 2046 if (ret == -1) { 2047 return ret; 2048 } else if (ret == 0) { 2049 break; /* EOF */ 2050 } 2051 cp += ret; 2052 left -= ret; 2053 } 2054 2055 return sz - left; 2056 } 2057