1 /* $NetBSD: gzip.c,v 1.97 2009/10/11 09:17:21 mrg Exp $ */ 2 3 /* 4 * Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 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 1.97 2009/10/11 09:17:21 mrg 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 #ifndef NO_PACK_SUPPORT 82 FT_PACK, 83 #endif 84 FT_LAST, 85 FT_UNKNOWN 86 }; 87 88 #ifndef NO_BZIP2_SUPPORT 89 #include <bzlib.h> 90 91 #define BZ2_SUFFIX ".bz2" 92 #define BZIP2_MAGIC "\102\132\150" 93 #endif 94 95 #ifndef NO_COMPRESS_SUPPORT 96 #define Z_SUFFIX ".Z" 97 #define Z_MAGIC "\037\235" 98 #endif 99 100 #ifndef NO_PACK_SUPPORT 101 #define PACK_MAGIC "\037\036" 102 #endif 103 104 #define GZ_SUFFIX ".gz" 105 106 #define BUFLEN (64 * 1024) 107 108 #define GZIP_MAGIC0 0x1F 109 #define GZIP_MAGIC1 0x8B 110 #define GZIP_OMAGIC1 0x9E 111 112 #define GZIP_TIMESTAMP (off_t)4 113 #define GZIP_ORIGNAME (off_t)10 114 115 #define HEAD_CRC 0x02 116 #define EXTRA_FIELD 0x04 117 #define ORIG_NAME 0x08 118 #define COMMENT 0x10 119 120 #define OS_CODE 3 /* Unix */ 121 122 typedef struct { 123 const char *zipped; 124 int ziplen; 125 const char *normal; /* for unzip - must not be longer than zipped */ 126 } suffixes_t; 127 static suffixes_t suffixes[] = { 128 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 129 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 130 #ifndef SMALL 131 SUFFIX(GZ_SUFFIX, ""), 132 SUFFIX(".z", ""), 133 SUFFIX("-gz", ""), 134 SUFFIX("-z", ""), 135 SUFFIX("_z", ""), 136 SUFFIX(".taz", ".tar"), 137 SUFFIX(".tgz", ".tar"), 138 #ifndef NO_BZIP2_SUPPORT 139 SUFFIX(BZ2_SUFFIX, ""), 140 #endif 141 #ifndef NO_COMPRESS_SUPPORT 142 SUFFIX(Z_SUFFIX, ""), 143 #endif 144 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 145 #endif /* SMALL */ 146 #undef SUFFIX 147 }; 148 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 149 #define SUFFIX_MAXLEN 30 150 151 static const char gzip_version[] = "NetBSD gzip 20091011"; 152 153 static int cflag; /* stdout mode */ 154 static int dflag; /* decompress mode */ 155 static int lflag; /* list mode */ 156 static int numflag = 6; /* gzip -1..-9 value */ 157 158 #ifndef SMALL 159 static int fflag; /* force mode */ 160 static int nflag; /* don't save name/timestamp */ 161 static int Nflag; /* don't restore name/timestamp */ 162 static int qflag; /* quiet mode */ 163 static int rflag; /* recursive mode */ 164 static int tflag; /* test */ 165 static int vflag; /* verbose mode */ 166 #else 167 #define qflag 0 168 #define tflag 0 169 #endif 170 171 static int exit_value = 0; /* exit value */ 172 173 static char *infile; /* name of file coming in */ 174 175 static void maybe_err(const char *fmt, ...) 176 __attribute__((__format__(__printf__, 1, 2))); 177 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) 178 static void maybe_errx(const char *fmt, ...) 179 __attribute__((__format__(__printf__, 1, 2))); 180 #endif 181 static void maybe_warn(const char *fmt, ...) 182 __attribute__((__format__(__printf__, 1, 2))); 183 static void maybe_warnx(const char *fmt, ...) 184 __attribute__((__format__(__printf__, 1, 2))); 185 static enum filetype file_gettype(u_char *); 186 #ifdef SMALL 187 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 188 #endif 189 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 190 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 191 static off_t file_compress(char *, char *, size_t); 192 static off_t file_uncompress(char *, char *, size_t); 193 static void handle_pathname(char *); 194 static void handle_file(char *, struct stat *); 195 static void handle_stdin(void); 196 static void handle_stdout(void); 197 static void print_ratio(off_t, off_t, FILE *); 198 static void print_list(int fd, off_t, const char *, time_t); 199 static void usage(void); 200 static void display_version(void); 201 static const suffixes_t *check_suffix(char *, int); 202 static ssize_t read_retry(int, void *, size_t); 203 204 #ifdef SMALL 205 #define unlink_input(f, sb) unlink(f) 206 #else 207 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 208 static void prepend_gzip(char *, int *, char ***); 209 static void handle_dir(char *); 210 static void print_verbage(const char *, const char *, off_t, off_t); 211 static void print_test(const char *, int); 212 static void copymodes(int fd, const struct stat *, const char *file); 213 static int check_outfile(const char *outfile); 214 #endif 215 216 #ifndef NO_BZIP2_SUPPORT 217 static off_t unbzip2(int, int, char *, size_t, off_t *); 218 #endif 219 220 #ifndef NO_COMPRESS_SUPPORT 221 static FILE *zdopen(int); 222 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 223 #endif 224 225 #ifndef NO_PACK_SUPPORT 226 static off_t unpack(int, int, char *, size_t, off_t *); 227 #endif 228 229 int main(int, char *p[]); 230 231 #ifdef SMALL 232 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 233 #else 234 static const struct option longopts[] = { 235 { "stdout", no_argument, 0, 'c' }, 236 { "to-stdout", no_argument, 0, 'c' }, 237 { "decompress", no_argument, 0, 'd' }, 238 { "uncompress", no_argument, 0, 'd' }, 239 { "force", no_argument, 0, 'f' }, 240 { "help", no_argument, 0, 'h' }, 241 { "list", no_argument, 0, 'l' }, 242 { "no-name", no_argument, 0, 'n' }, 243 { "name", no_argument, 0, 'N' }, 244 { "quiet", no_argument, 0, 'q' }, 245 { "recursive", no_argument, 0, 'r' }, 246 { "suffix", required_argument, 0, 'S' }, 247 { "test", no_argument, 0, 't' }, 248 { "verbose", no_argument, 0, 'v' }, 249 { "version", no_argument, 0, 'V' }, 250 { "fast", no_argument, 0, '1' }, 251 { "best", no_argument, 0, '9' }, 252 #if 0 253 /* 254 * This is what else GNU gzip implements. --ascii isn't useful 255 * on NetBSD, and I don't care to have a --license. 256 */ 257 { "ascii", no_argument, 0, 'a' }, 258 { "license", no_argument, 0, 'L' }, 259 #endif 260 { NULL, no_argument, 0, 0 }, 261 }; 262 #endif 263 264 int 265 main(int argc, char **argv) 266 { 267 const char *progname = getprogname(); 268 #ifndef SMALL 269 char *gzip; 270 int len; 271 #endif 272 int ch; 273 274 /* XXX set up signals */ 275 276 #ifndef SMALL 277 if ((gzip = getenv("GZIP")) != NULL) 278 prepend_gzip(gzip, &argc, &argv); 279 #endif 280 281 /* 282 * XXX 283 * handle being called `gunzip', `zcat' and `gzcat' 284 */ 285 if (strcmp(progname, "gunzip") == 0) 286 dflag = 1; 287 else if (strcmp(progname, "zcat") == 0 || 288 strcmp(progname, "gzcat") == 0) 289 dflag = cflag = 1; 290 291 #ifdef SMALL 292 #define OPT_LIST "123456789cdhltV" 293 #else 294 #define OPT_LIST "123456789cdfhlNnqrS:tVv" 295 #endif 296 297 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 298 switch (ch) { 299 case '1': case '2': case '3': 300 case '4': case '5': case '6': 301 case '7': case '8': case '9': 302 numflag = ch - '0'; 303 break; 304 case 'c': 305 cflag = 1; 306 break; 307 case 'd': 308 dflag = 1; 309 break; 310 case 'l': 311 lflag = 1; 312 dflag = 1; 313 break; 314 case 'V': 315 display_version(); 316 /* NOTREACHED */ 317 #ifndef SMALL 318 case 'f': 319 fflag = 1; 320 break; 321 case 'N': 322 nflag = 0; 323 Nflag = 1; 324 break; 325 case 'n': 326 nflag = 1; 327 Nflag = 0; 328 break; 329 case 'q': 330 qflag = 1; 331 break; 332 case 'r': 333 rflag = 1; 334 break; 335 case 'S': 336 len = strlen(optarg); 337 if (len != 0) { 338 if (len > SUFFIX_MAXLEN) 339 errx(1, "incorrect suffix: '%s'", optarg); 340 suffixes[0].zipped = optarg; 341 suffixes[0].ziplen = len; 342 } else { 343 suffixes[NUM_SUFFIXES - 1].zipped = ""; 344 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 345 } 346 break; 347 case 't': 348 cflag = 1; 349 tflag = 1; 350 dflag = 1; 351 break; 352 case 'v': 353 vflag = 1; 354 break; 355 #endif 356 default: 357 usage(); 358 /* NOTREACHED */ 359 } 360 } 361 argv += optind; 362 argc -= optind; 363 364 if (argc == 0) { 365 if (dflag) /* stdin mode */ 366 handle_stdin(); 367 else /* stdout mode */ 368 handle_stdout(); 369 } else { 370 do { 371 handle_pathname(argv[0]); 372 } while (*++argv); 373 } 374 #ifndef SMALL 375 if (qflag == 0 && lflag && argc > 1) 376 print_list(-1, 0, "(totals)", 0); 377 #endif 378 exit(exit_value); 379 } 380 381 /* maybe print a warning */ 382 void 383 maybe_warn(const char *fmt, ...) 384 { 385 va_list ap; 386 387 if (qflag == 0) { 388 va_start(ap, fmt); 389 vwarn(fmt, ap); 390 va_end(ap); 391 } 392 if (exit_value == 0) 393 exit_value = 1; 394 } 395 396 /* ... without an errno. */ 397 void 398 maybe_warnx(const char *fmt, ...) 399 { 400 va_list ap; 401 402 if (qflag == 0) { 403 va_start(ap, fmt); 404 vwarnx(fmt, ap); 405 va_end(ap); 406 } 407 if (exit_value == 0) 408 exit_value = 1; 409 } 410 411 /* maybe print an error */ 412 void 413 maybe_err(const char *fmt, ...) 414 { 415 va_list ap; 416 417 if (qflag == 0) { 418 va_start(ap, fmt); 419 vwarn(fmt, ap); 420 va_end(ap); 421 } 422 exit(2); 423 } 424 425 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) 426 /* ... without an errno. */ 427 void 428 maybe_errx(const char *fmt, ...) 429 { 430 va_list ap; 431 432 if (qflag == 0) { 433 va_start(ap, fmt); 434 vwarnx(fmt, ap); 435 va_end(ap); 436 } 437 exit(2); 438 } 439 #endif 440 441 #ifndef SMALL 442 /* split up $GZIP and prepend it to the argument list */ 443 static void 444 prepend_gzip(char *gzip, int *argc, char ***argv) 445 { 446 char *s, **nargv, **ac; 447 int nenvarg = 0, i; 448 449 /* scan how many arguments there are */ 450 for (s = gzip;;) { 451 while (*s == ' ' || *s == '\t') 452 s++; 453 if (*s == 0) 454 goto count_done; 455 nenvarg++; 456 while (*s != ' ' && *s != '\t') 457 if (*s++ == 0) 458 goto count_done; 459 } 460 count_done: 461 /* punt early */ 462 if (nenvarg == 0) 463 return; 464 465 *argc += nenvarg; 466 ac = *argv; 467 468 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 469 if (nargv == NULL) 470 maybe_err("malloc"); 471 472 /* stash this away */ 473 *argv = nargv; 474 475 /* copy the program name first */ 476 i = 0; 477 nargv[i++] = *(ac++); 478 479 /* take a copy of $GZIP and add it to the array */ 480 s = strdup(gzip); 481 if (s == NULL) 482 maybe_err("strdup"); 483 for (;;) { 484 /* Skip whitespaces. */ 485 while (*s == ' ' || *s == '\t') 486 s++; 487 if (*s == 0) 488 goto copy_done; 489 nargv[i++] = s; 490 /* Find the end of this argument. */ 491 while (*s != ' ' && *s != '\t') 492 if (*s++ == 0) 493 /* Argument followed by NUL. */ 494 goto copy_done; 495 /* Terminate by overwriting ' ' or '\t' with NUL. */ 496 *s++ = 0; 497 } 498 copy_done: 499 500 /* copy the original arguments and a NULL */ 501 while (*ac) 502 nargv[i++] = *(ac++); 503 nargv[i] = NULL; 504 } 505 #endif 506 507 /* compress input to output. Return bytes read, -1 on error */ 508 static off_t 509 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 510 { 511 z_stream z; 512 char *outbufp, *inbufp; 513 off_t in_tot = 0, out_tot = 0; 514 ssize_t in_size; 515 int i, error; 516 uLong crc; 517 #ifdef SMALL 518 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 519 0, 0, 0, 0, 520 0, OS_CODE }; 521 #endif 522 523 outbufp = malloc(BUFLEN); 524 inbufp = malloc(BUFLEN); 525 if (outbufp == NULL || inbufp == NULL) { 526 maybe_err("malloc failed"); 527 goto out; 528 } 529 530 memset(&z, 0, sizeof z); 531 z.zalloc = Z_NULL; 532 z.zfree = Z_NULL; 533 z.opaque = 0; 534 535 #ifdef SMALL 536 memcpy(outbufp, header, sizeof header); 537 i = sizeof header; 538 #else 539 if (nflag != 0) { 540 mtime = 0; 541 origname = ""; 542 } 543 544 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 545 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 546 *origname ? ORIG_NAME : 0, 547 mtime & 0xff, 548 (mtime >> 8) & 0xff, 549 (mtime >> 16) & 0xff, 550 (mtime >> 24) & 0xff, 551 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 552 OS_CODE, origname); 553 if (i >= BUFLEN) 554 /* this need PATH_MAX > BUFLEN ... */ 555 maybe_err("snprintf"); 556 if (*origname) 557 i++; 558 #endif 559 560 z.next_out = outbufp + i; 561 z.avail_out = BUFLEN - i; 562 563 error = deflateInit2(&z, numflag, Z_DEFLATED, 564 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 565 if (error != Z_OK) { 566 maybe_warnx("deflateInit2 failed"); 567 in_tot = -1; 568 goto out; 569 } 570 571 crc = crc32(0L, Z_NULL, 0); 572 for (;;) { 573 if (z.avail_out == 0) { 574 if (write(out, outbufp, BUFLEN) != BUFLEN) { 575 maybe_warn("write"); 576 out_tot = -1; 577 goto out; 578 } 579 580 out_tot += BUFLEN; 581 z.next_out = outbufp; 582 z.avail_out = BUFLEN; 583 } 584 585 if (z.avail_in == 0) { 586 in_size = read(in, inbufp, BUFLEN); 587 if (in_size < 0) { 588 maybe_warn("read"); 589 in_tot = -1; 590 goto out; 591 } 592 if (in_size == 0) 593 break; 594 595 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 596 in_tot += in_size; 597 z.next_in = inbufp; 598 z.avail_in = in_size; 599 } 600 601 error = deflate(&z, Z_NO_FLUSH); 602 if (error != Z_OK && error != Z_STREAM_END) { 603 maybe_warnx("deflate failed"); 604 in_tot = -1; 605 goto out; 606 } 607 } 608 609 /* clean up */ 610 for (;;) { 611 size_t len; 612 ssize_t w; 613 614 error = deflate(&z, Z_FINISH); 615 if (error != Z_OK && error != Z_STREAM_END) { 616 maybe_warnx("deflate failed"); 617 in_tot = -1; 618 goto out; 619 } 620 621 len = (char *)z.next_out - outbufp; 622 623 w = write(out, outbufp, len); 624 if (w == -1 || (size_t)w != len) { 625 maybe_warn("write"); 626 out_tot = -1; 627 goto out; 628 } 629 out_tot += len; 630 z.next_out = outbufp; 631 z.avail_out = BUFLEN; 632 633 if (error == Z_STREAM_END) 634 break; 635 } 636 637 if (deflateEnd(&z) != Z_OK) { 638 maybe_warnx("deflateEnd failed"); 639 in_tot = -1; 640 goto out; 641 } 642 643 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 644 (int)crc & 0xff, 645 (int)(crc >> 8) & 0xff, 646 (int)(crc >> 16) & 0xff, 647 (int)(crc >> 24) & 0xff, 648 (int)in_tot & 0xff, 649 (int)(in_tot >> 8) & 0xff, 650 (int)(in_tot >> 16) & 0xff, 651 (int)(in_tot >> 24) & 0xff); 652 if (i != 8) 653 maybe_err("snprintf"); 654 #if 0 655 if (in_tot > 0xffffffff) 656 maybe_warn("input file size >= 4GB cannot be saved"); 657 #endif 658 if (write(out, outbufp, i) != i) { 659 maybe_warn("write"); 660 in_tot = -1; 661 } else 662 out_tot += i; 663 664 out: 665 if (inbufp != NULL) 666 free(inbufp); 667 if (outbufp != NULL) 668 free(outbufp); 669 if (gsizep) 670 *gsizep = out_tot; 671 return in_tot; 672 } 673 674 /* 675 * uncompress input to output then close the input. return the 676 * uncompressed size written, and put the compressed sized read 677 * into `*gsizep'. 678 */ 679 static off_t 680 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 681 const char *filename) 682 { 683 z_stream z; 684 char *outbufp, *inbufp; 685 off_t out_tot = -1, in_tot = 0; 686 uint32_t out_sub_tot = 0; 687 enum { 688 GZSTATE_MAGIC0, 689 GZSTATE_MAGIC1, 690 GZSTATE_METHOD, 691 GZSTATE_FLAGS, 692 GZSTATE_SKIPPING, 693 GZSTATE_EXTRA, 694 GZSTATE_EXTRA2, 695 GZSTATE_EXTRA3, 696 GZSTATE_ORIGNAME, 697 GZSTATE_COMMENT, 698 GZSTATE_HEAD_CRC1, 699 GZSTATE_HEAD_CRC2, 700 GZSTATE_INIT, 701 GZSTATE_READ, 702 GZSTATE_CRC, 703 GZSTATE_LEN, 704 } state = GZSTATE_MAGIC0; 705 int flags = 0, skip_count = 0; 706 int error = Z_STREAM_ERROR, done_reading = 0; 707 uLong crc = 0; 708 ssize_t wr; 709 int needmore = 0; 710 711 #define ADVANCE() { z.next_in++; z.avail_in--; } 712 713 if ((outbufp = malloc(BUFLEN)) == NULL) { 714 maybe_err("malloc failed"); 715 goto out2; 716 } 717 if ((inbufp = malloc(BUFLEN)) == NULL) { 718 maybe_err("malloc failed"); 719 goto out1; 720 } 721 722 memset(&z, 0, sizeof z); 723 z.avail_in = prelen; 724 z.next_in = pre; 725 z.avail_out = BUFLEN; 726 z.next_out = outbufp; 727 z.zalloc = NULL; 728 z.zfree = NULL; 729 z.opaque = 0; 730 731 in_tot = prelen; 732 out_tot = 0; 733 734 for (;;) { 735 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 736 ssize_t in_size; 737 738 if (z.avail_in > 0) { 739 memmove(inbufp, z.next_in, z.avail_in); 740 } 741 z.next_in = inbufp; 742 in_size = read(in, z.next_in + z.avail_in, 743 BUFLEN - z.avail_in); 744 745 if (in_size == -1) { 746 maybe_warn("failed to read stdin"); 747 goto stop_and_fail; 748 } else if (in_size == 0) { 749 done_reading = 1; 750 } 751 752 z.avail_in += in_size; 753 needmore = 0; 754 755 in_tot += in_size; 756 } 757 if (z.avail_in == 0) { 758 if (done_reading && state != GZSTATE_MAGIC0) { 759 maybe_warnx("%s: unexpected end of file", 760 filename); 761 goto stop_and_fail; 762 } 763 goto stop; 764 } 765 switch (state) { 766 case GZSTATE_MAGIC0: 767 if (*z.next_in != GZIP_MAGIC0) { 768 if (in_tot > 0) { 769 maybe_warnx("%s: trailing garbage " 770 "ignored", filename); 771 goto stop; 772 } 773 maybe_warnx("input not gziped (MAGIC0)"); 774 goto stop_and_fail; 775 } 776 ADVANCE(); 777 state++; 778 out_sub_tot = 0; 779 crc = crc32(0L, Z_NULL, 0); 780 break; 781 782 case GZSTATE_MAGIC1: 783 if (*z.next_in != GZIP_MAGIC1 && 784 *z.next_in != GZIP_OMAGIC1) { 785 maybe_warnx("input not gziped (MAGIC1)"); 786 goto stop_and_fail; 787 } 788 ADVANCE(); 789 state++; 790 break; 791 792 case GZSTATE_METHOD: 793 if (*z.next_in != Z_DEFLATED) { 794 maybe_warnx("unknown compression method"); 795 goto stop_and_fail; 796 } 797 ADVANCE(); 798 state++; 799 break; 800 801 case GZSTATE_FLAGS: 802 flags = *z.next_in; 803 ADVANCE(); 804 skip_count = 6; 805 state++; 806 break; 807 808 case GZSTATE_SKIPPING: 809 if (skip_count > 0) { 810 skip_count--; 811 ADVANCE(); 812 } else 813 state++; 814 break; 815 816 case GZSTATE_EXTRA: 817 if ((flags & EXTRA_FIELD) == 0) { 818 state = GZSTATE_ORIGNAME; 819 break; 820 } 821 skip_count = *z.next_in; 822 ADVANCE(); 823 state++; 824 break; 825 826 case GZSTATE_EXTRA2: 827 skip_count |= ((*z.next_in) << 8); 828 ADVANCE(); 829 state++; 830 break; 831 832 case GZSTATE_EXTRA3: 833 if (skip_count > 0) { 834 skip_count--; 835 ADVANCE(); 836 } else 837 state++; 838 break; 839 840 case GZSTATE_ORIGNAME: 841 if ((flags & ORIG_NAME) == 0) { 842 state++; 843 break; 844 } 845 if (*z.next_in == 0) 846 state++; 847 ADVANCE(); 848 break; 849 850 case GZSTATE_COMMENT: 851 if ((flags & COMMENT) == 0) { 852 state++; 853 break; 854 } 855 if (*z.next_in == 0) 856 state++; 857 ADVANCE(); 858 break; 859 860 case GZSTATE_HEAD_CRC1: 861 if (flags & HEAD_CRC) 862 skip_count = 2; 863 else 864 skip_count = 0; 865 state++; 866 break; 867 868 case GZSTATE_HEAD_CRC2: 869 if (skip_count > 0) { 870 skip_count--; 871 ADVANCE(); 872 } else 873 state++; 874 break; 875 876 case GZSTATE_INIT: 877 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 878 maybe_warnx("failed to inflateInit"); 879 goto stop_and_fail; 880 } 881 state++; 882 break; 883 884 case GZSTATE_READ: 885 error = inflate(&z, Z_FINISH); 886 switch (error) { 887 /* Z_BUF_ERROR goes with Z_FINISH... */ 888 case Z_BUF_ERROR: 889 case Z_STREAM_END: 890 case Z_OK: 891 break; 892 893 case Z_NEED_DICT: 894 maybe_warnx("Z_NEED_DICT error"); 895 goto stop_and_fail; 896 case Z_DATA_ERROR: 897 maybe_warnx("data stream error"); 898 goto stop_and_fail; 899 case Z_STREAM_ERROR: 900 maybe_warnx("internal stream error"); 901 goto stop_and_fail; 902 case Z_MEM_ERROR: 903 maybe_warnx("memory allocation error"); 904 goto stop_and_fail; 905 906 default: 907 maybe_warn("unknown error from inflate(): %d", 908 error); 909 } 910 wr = BUFLEN - z.avail_out; 911 912 if (wr != 0) { 913 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 914 if ( 915 #ifndef SMALL 916 /* don't write anything with -t */ 917 tflag == 0 && 918 #endif 919 write(out, outbufp, wr) != wr) { 920 maybe_warn("error writing to output"); 921 goto stop_and_fail; 922 } 923 924 out_tot += wr; 925 out_sub_tot += wr; 926 } 927 928 if (error == Z_STREAM_END) { 929 inflateEnd(&z); 930 state++; 931 } 932 933 z.next_out = outbufp; 934 z.avail_out = BUFLEN; 935 936 break; 937 case GZSTATE_CRC: 938 { 939 uLong origcrc; 940 941 if (z.avail_in < 4) { 942 if (!done_reading) { 943 needmore = 1; 944 continue; 945 } 946 maybe_warnx("truncated input"); 947 goto stop_and_fail; 948 } 949 origcrc = ((unsigned)z.next_in[0] & 0xff) | 950 ((unsigned)z.next_in[1] & 0xff) << 8 | 951 ((unsigned)z.next_in[2] & 0xff) << 16 | 952 ((unsigned)z.next_in[3] & 0xff) << 24; 953 if (origcrc != crc) { 954 maybe_warnx("invalid compressed" 955 " data--crc error"); 956 goto stop_and_fail; 957 } 958 } 959 960 z.avail_in -= 4; 961 z.next_in += 4; 962 963 if (!z.avail_in && done_reading) { 964 goto stop; 965 } 966 state++; 967 break; 968 case GZSTATE_LEN: 969 { 970 uLong origlen; 971 972 if (z.avail_in < 4) { 973 if (!done_reading) { 974 needmore = 1; 975 continue; 976 } 977 maybe_warnx("truncated input"); 978 goto stop_and_fail; 979 } 980 origlen = ((unsigned)z.next_in[0] & 0xff) | 981 ((unsigned)z.next_in[1] & 0xff) << 8 | 982 ((unsigned)z.next_in[2] & 0xff) << 16 | 983 ((unsigned)z.next_in[3] & 0xff) << 24; 984 985 if (origlen != out_sub_tot) { 986 maybe_warnx("invalid compressed" 987 " data--length error"); 988 goto stop_and_fail; 989 } 990 } 991 992 z.avail_in -= 4; 993 z.next_in += 4; 994 995 if (error < 0) { 996 maybe_warnx("decompression error"); 997 goto stop_and_fail; 998 } 999 state = GZSTATE_MAGIC0; 1000 break; 1001 } 1002 continue; 1003 stop_and_fail: 1004 out_tot = -1; 1005 stop: 1006 break; 1007 } 1008 if (state > GZSTATE_INIT) 1009 inflateEnd(&z); 1010 1011 free(inbufp); 1012 out1: 1013 free(outbufp); 1014 out2: 1015 if (gsizep) 1016 *gsizep = in_tot; 1017 return (out_tot); 1018 } 1019 1020 #ifndef SMALL 1021 /* 1022 * set the owner, mode, flags & utimes using the given file descriptor. 1023 * file is only used in possible warning messages. 1024 */ 1025 static void 1026 copymodes(int fd, const struct stat *sbp, const char *file) 1027 { 1028 struct timeval times[2]; 1029 struct stat sb; 1030 1031 /* 1032 * If we have no info on the input, give this file some 1033 * default values and return.. 1034 */ 1035 if (sbp == NULL) { 1036 mode_t mask = umask(022); 1037 1038 (void)fchmod(fd, DEFFILEMODE & ~mask); 1039 (void)umask(mask); 1040 return; 1041 } 1042 sb = *sbp; 1043 1044 /* if the chown fails, remove set-id bits as-per compress(1) */ 1045 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1046 if (errno != EPERM) 1047 maybe_warn("couldn't fchown: %s", file); 1048 sb.st_mode &= ~(S_ISUID|S_ISGID); 1049 } 1050 1051 /* we only allow set-id and the 9 normal permission bits */ 1052 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1053 if (fchmod(fd, sb.st_mode) < 0) 1054 maybe_warn("couldn't fchmod: %s", file); 1055 1056 /* only try flags if they exist already */ 1057 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1058 maybe_warn("couldn't fchflags: %s", file); 1059 1060 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1061 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1062 if (futimes(fd, times) < 0) 1063 maybe_warn("couldn't utimes: %s", file); 1064 } 1065 #endif 1066 1067 /* what sort of file is this? */ 1068 static enum filetype 1069 file_gettype(u_char *buf) 1070 { 1071 1072 if (buf[0] == GZIP_MAGIC0 && 1073 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1074 return FT_GZIP; 1075 else 1076 #ifndef NO_BZIP2_SUPPORT 1077 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1078 buf[3] >= '0' && buf[3] <= '9') 1079 return FT_BZIP2; 1080 else 1081 #endif 1082 #ifndef NO_COMPRESS_SUPPORT 1083 if (memcmp(buf, Z_MAGIC, 2) == 0) 1084 return FT_Z; 1085 else 1086 #endif 1087 #ifndef NO_PACK_SUPPORT 1088 if (memcmp(buf, PACK_MAGIC, 2) == 0) 1089 return FT_PACK; 1090 else 1091 #endif 1092 return FT_UNKNOWN; 1093 } 1094 1095 #ifndef SMALL 1096 /* check the outfile is OK. */ 1097 static int 1098 check_outfile(const char *outfile) 1099 { 1100 struct stat sb; 1101 int ok = 1; 1102 1103 if (lflag == 0 && stat(outfile, &sb) == 0) { 1104 if (fflag) 1105 unlink(outfile); 1106 else if (isatty(STDIN_FILENO)) { 1107 char ans[10] = { 'n', '\0' }; /* default */ 1108 1109 fprintf(stderr, "%s already exists -- do you wish to " 1110 "overwrite (y or n)? " , outfile); 1111 (void)fgets(ans, sizeof(ans) - 1, stdin); 1112 if (ans[0] != 'y' && ans[0] != 'Y') { 1113 fprintf(stderr, "\tnot overwriting\n"); 1114 ok = 0; 1115 } else 1116 unlink(outfile); 1117 } else { 1118 maybe_warnx("%s already exists -- skipping", outfile); 1119 ok = 0; 1120 } 1121 } 1122 return ok; 1123 } 1124 1125 static void 1126 unlink_input(const char *file, const struct stat *sb) 1127 { 1128 struct stat nsb; 1129 1130 if (stat(file, &nsb) != 0) 1131 /* Must be gone alrady */ 1132 return; 1133 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1134 /* Definitely a different file */ 1135 return; 1136 unlink(file); 1137 } 1138 #endif 1139 1140 static const suffixes_t * 1141 check_suffix(char *file, int xlate) 1142 { 1143 const suffixes_t *s; 1144 int len = strlen(file); 1145 char *sp; 1146 1147 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1148 /* if it doesn't fit in "a.suf", don't bother */ 1149 if (s->ziplen >= len) 1150 continue; 1151 sp = file + len - s->ziplen; 1152 if (strcmp(s->zipped, sp) != 0) 1153 continue; 1154 if (xlate) 1155 strcpy(sp, s->normal); 1156 return s; 1157 } 1158 return NULL; 1159 } 1160 1161 /* 1162 * compress the given file: create a corresponding .gz file and remove the 1163 * original. 1164 */ 1165 static off_t 1166 file_compress(char *file, char *outfile, size_t outsize) 1167 { 1168 int in; 1169 int out; 1170 off_t size, insize; 1171 #ifndef SMALL 1172 struct stat isb, osb; 1173 const suffixes_t *suff; 1174 #endif 1175 1176 in = open(file, O_RDONLY); 1177 if (in == -1) { 1178 maybe_warn("can't open %s", file); 1179 return -1; 1180 } 1181 1182 if (cflag == 0) { 1183 #ifndef SMALL 1184 if (fstat(in, &isb) == 0) { 1185 if (isb.st_nlink > 1 && fflag == 0) { 1186 maybe_warnx("%s has %d other link%s -- " 1187 "skipping", file, isb.st_nlink - 1, 1188 isb.st_nlink == 1 ? "" : "s"); 1189 close(in); 1190 return -1; 1191 } 1192 } 1193 1194 if (fflag == 0 && (suff = check_suffix(file, 0)) 1195 && suff->zipped[0] != 0) { 1196 maybe_warnx("%s already has %s suffix -- unchanged", 1197 file, suff->zipped); 1198 close(in); 1199 return -1; 1200 } 1201 #endif 1202 1203 /* Add (usually) .gz to filename */ 1204 if ((size_t)snprintf(outfile, outsize, "%s%s", 1205 file, suffixes[0].zipped) >= outsize) 1206 memcpy(outfile + outsize - suffixes[0].ziplen - 1, 1207 suffixes[0].zipped, suffixes[0].ziplen + 1); 1208 1209 #ifndef SMALL 1210 if (check_outfile(outfile) == 0) { 1211 close(in); 1212 return -1; 1213 } 1214 #endif 1215 } 1216 1217 if (cflag == 0) { 1218 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1219 if (out == -1) { 1220 maybe_warn("could not create output: %s", outfile); 1221 fclose(stdin); 1222 return -1; 1223 } 1224 } else 1225 out = STDOUT_FILENO; 1226 1227 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1228 1229 (void)close(in); 1230 1231 /* 1232 * If there was an error, insize will be -1. 1233 * If we compressed to stdout, just return the size. 1234 * Otherwise stat the file and check it is the correct size. 1235 * We only blow away the file if we can stat the output and it 1236 * has the expected size. 1237 */ 1238 if (cflag != 0) 1239 return insize == -1 ? -1 : size; 1240 1241 #ifndef SMALL 1242 if (fstat(out, &osb) != 0) { 1243 maybe_warn("couldn't stat: %s", outfile); 1244 goto bad_outfile; 1245 } 1246 1247 if (osb.st_size != size) { 1248 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1249 " != %" PRIdOFF "), deleting", 1250 outfile, osb.st_size, size); 1251 goto bad_outfile; 1252 } 1253 1254 copymodes(out, &isb, outfile); 1255 #endif 1256 if (close(out) == -1) 1257 maybe_warn("couldn't close output"); 1258 1259 /* output is good, ok to delete input */ 1260 unlink_input(file, &isb); 1261 return size; 1262 1263 #ifndef SMALL 1264 bad_outfile: 1265 if (close(out) == -1) 1266 maybe_warn("couldn't close output"); 1267 1268 maybe_warnx("leaving original %s", file); 1269 unlink(outfile); 1270 return size; 1271 #endif 1272 } 1273 1274 /* uncompress the given file and remove the original */ 1275 static off_t 1276 file_uncompress(char *file, char *outfile, size_t outsize) 1277 { 1278 struct stat isb, osb; 1279 off_t size; 1280 ssize_t rbytes; 1281 unsigned char header1[4]; 1282 enum filetype method; 1283 int fd, ofd, zfd = -1; 1284 #ifndef SMALL 1285 ssize_t rv; 1286 time_t timestamp = 0; 1287 unsigned char name[PATH_MAX + 1]; 1288 #endif 1289 1290 /* gather the old name info */ 1291 1292 fd = open(file, O_RDONLY); 1293 if (fd < 0) { 1294 maybe_warn("can't open %s", file); 1295 goto lose; 1296 } 1297 1298 strlcpy(outfile, file, outsize); 1299 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1300 maybe_warnx("%s: unknown suffix -- ignored", file); 1301 goto lose; 1302 } 1303 1304 rbytes = read(fd, header1, sizeof header1); 1305 if (rbytes != sizeof header1) { 1306 /* we don't want to fail here. */ 1307 #ifndef SMALL 1308 if (fflag) 1309 goto lose; 1310 #endif 1311 if (rbytes == -1) 1312 maybe_warn("can't read %s", file); 1313 else 1314 goto unexpected_EOF; 1315 goto lose; 1316 } 1317 1318 method = file_gettype(header1); 1319 1320 #ifndef SMALL 1321 if (fflag == 0 && method == FT_UNKNOWN) { 1322 maybe_warnx("%s: not in gzip format", file); 1323 goto lose; 1324 } 1325 1326 #endif 1327 1328 #ifndef SMALL 1329 if (method == FT_GZIP && Nflag) { 1330 unsigned char ts[4]; /* timestamp */ 1331 1332 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1333 if (rv >= 0 && rv < (ssize_t)(sizeof ts)) 1334 goto unexpected_EOF; 1335 if (rv == -1) { 1336 if (!fflag) 1337 maybe_warn("can't read %s", file); 1338 goto lose; 1339 } 1340 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1341 1342 if (header1[3] & ORIG_NAME) { 1343 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1344 if (rbytes < 0) { 1345 maybe_warn("can't read %s", file); 1346 goto lose; 1347 } 1348 if (name[0] != 0) { 1349 /* preserve original directory name */ 1350 char *dp = strrchr(file, '/'); 1351 if (dp == NULL) 1352 dp = file; 1353 else 1354 dp++; 1355 snprintf(outfile, outsize, "%.*s%.*s", 1356 (int) (dp - file), 1357 file, (int) rbytes, name); 1358 } 1359 } 1360 } 1361 #endif 1362 lseek(fd, 0, SEEK_SET); 1363 1364 if (cflag == 0 || lflag) { 1365 if (fstat(fd, &isb) != 0) 1366 goto lose; 1367 #ifndef SMALL 1368 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1369 maybe_warnx("%s has %d other links -- skipping", 1370 file, isb.st_nlink - 1); 1371 goto lose; 1372 } 1373 if (nflag == 0 && timestamp) 1374 isb.st_mtime = timestamp; 1375 if (check_outfile(outfile) == 0) 1376 goto lose; 1377 #endif 1378 } 1379 1380 if (cflag == 0 && lflag == 0) { 1381 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1382 if (zfd == STDOUT_FILENO) { 1383 /* We won't close STDOUT_FILENO later... */ 1384 zfd = dup(zfd); 1385 close(STDOUT_FILENO); 1386 } 1387 if (zfd == -1) { 1388 maybe_warn("can't open %s", outfile); 1389 goto lose; 1390 } 1391 } else 1392 zfd = STDOUT_FILENO; 1393 1394 #ifndef NO_BZIP2_SUPPORT 1395 if (method == FT_BZIP2) { 1396 1397 /* XXX */ 1398 if (lflag) { 1399 maybe_warnx("no -l with bzip2 files"); 1400 goto lose; 1401 } 1402 1403 size = unbzip2(fd, zfd, NULL, 0, NULL); 1404 } else 1405 #endif 1406 1407 #ifndef NO_COMPRESS_SUPPORT 1408 if (method == FT_Z) { 1409 FILE *in, *out; 1410 1411 /* XXX */ 1412 if (lflag) { 1413 maybe_warnx("no -l with Lempel-Ziv files"); 1414 goto lose; 1415 } 1416 1417 if ((in = zdopen(fd)) == NULL) { 1418 maybe_warn("zdopen for read: %s", file); 1419 goto lose; 1420 } 1421 1422 out = fdopen(dup(zfd), "w"); 1423 if (out == NULL) { 1424 maybe_warn("fdopen for write: %s", outfile); 1425 fclose(in); 1426 goto lose; 1427 } 1428 1429 size = zuncompress(in, out, NULL, 0, NULL); 1430 /* need to fclose() if ferror() is true... */ 1431 if (ferror(in) | fclose(in)) { 1432 maybe_warn("failed infile fclose"); 1433 unlink(outfile); 1434 (void)fclose(out); 1435 } 1436 if (fclose(out) != 0) { 1437 maybe_warn("failed outfile fclose"); 1438 unlink(outfile); 1439 goto lose; 1440 } 1441 } else 1442 #endif 1443 1444 #ifndef NO_PACK_SUPPORT 1445 if (method == FT_PACK) { 1446 if (lflag) { 1447 maybe_warnx("no -l with packed files"); 1448 goto lose; 1449 } 1450 1451 size = unpack(fd, zfd, NULL, 0, NULL); 1452 } else 1453 #endif 1454 1455 #ifndef SMALL 1456 if (method == FT_UNKNOWN) { 1457 if (lflag) { 1458 maybe_warnx("no -l for unknown filetypes"); 1459 goto lose; 1460 } 1461 size = cat_fd(NULL, 0, NULL, fd); 1462 } else 1463 #endif 1464 { 1465 if (lflag) { 1466 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1467 close(fd); 1468 return -1; /* XXX */ 1469 } 1470 1471 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1472 } 1473 1474 if (close(fd) != 0) 1475 maybe_warn("couldn't close input"); 1476 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1477 maybe_warn("couldn't close output"); 1478 1479 if (size == -1) { 1480 if (cflag == 0) 1481 unlink(outfile); 1482 maybe_warnx("%s: uncompress failed", file); 1483 return -1; 1484 } 1485 1486 /* if testing, or we uncompressed to stdout, this is all we need */ 1487 #ifndef SMALL 1488 if (tflag) 1489 return size; 1490 #endif 1491 /* if we are uncompressing to stdin, don't remove the file. */ 1492 if (cflag) 1493 return size; 1494 1495 /* 1496 * if we create a file... 1497 */ 1498 /* 1499 * if we can't stat the file don't remove the file. 1500 */ 1501 1502 ofd = open(outfile, O_RDWR, 0); 1503 if (ofd == -1) { 1504 maybe_warn("couldn't open (leaving original): %s", 1505 outfile); 1506 return -1; 1507 } 1508 if (fstat(ofd, &osb) != 0) { 1509 maybe_warn("couldn't stat (leaving original): %s", 1510 outfile); 1511 close(ofd); 1512 return -1; 1513 } 1514 if (osb.st_size != size) { 1515 maybe_warnx("stat gave different size: %" PRIdOFF 1516 " != %" PRIdOFF " (leaving original)", 1517 size, osb.st_size); 1518 close(ofd); 1519 unlink(outfile); 1520 return -1; 1521 } 1522 unlink_input(file, &isb); 1523 #ifndef SMALL 1524 copymodes(ofd, &isb, outfile); 1525 #endif 1526 close(ofd); 1527 return size; 1528 1529 unexpected_EOF: 1530 maybe_warnx("%s: unexpected end of file", file); 1531 lose: 1532 if (fd != -1) 1533 close(fd); 1534 if (zfd != -1 && zfd != STDOUT_FILENO) 1535 close(fd); 1536 return -1; 1537 } 1538 1539 #ifndef SMALL 1540 static off_t 1541 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1542 { 1543 char buf[BUFLEN]; 1544 off_t in_tot; 1545 ssize_t w; 1546 1547 in_tot = count; 1548 w = write(STDOUT_FILENO, prepend, count); 1549 if (w == -1 || (size_t)w != count) { 1550 maybe_warn("write to stdout"); 1551 return -1; 1552 } 1553 for (;;) { 1554 ssize_t rv; 1555 1556 rv = read(fd, buf, sizeof buf); 1557 if (rv == 0) 1558 break; 1559 if (rv < 0) { 1560 maybe_warn("read from fd %d", fd); 1561 break; 1562 } 1563 1564 if (write(STDOUT_FILENO, buf, rv) != rv) { 1565 maybe_warn("write to stdout"); 1566 break; 1567 } 1568 in_tot += rv; 1569 } 1570 1571 if (gsizep) 1572 *gsizep = in_tot; 1573 return (in_tot); 1574 } 1575 #endif 1576 1577 static void 1578 handle_stdin(void) 1579 { 1580 unsigned char header1[4]; 1581 off_t usize, gsize; 1582 enum filetype method; 1583 ssize_t bytes_read; 1584 #ifndef NO_COMPRESS_SUPPORT 1585 FILE *in; 1586 #endif 1587 1588 #ifndef SMALL 1589 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1590 maybe_warnx("standard input is a terminal -- ignoring"); 1591 return; 1592 } 1593 #endif 1594 1595 if (lflag) { 1596 struct stat isb; 1597 1598 /* XXX could read the whole file, etc. */ 1599 if (fstat(STDIN_FILENO, &isb) < 0) { 1600 maybe_warn("fstat"); 1601 return; 1602 } 1603 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1604 return; 1605 } 1606 1607 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1608 if (bytes_read == -1) { 1609 maybe_warn("can't read stdin"); 1610 return; 1611 } else if (bytes_read != sizeof(header1)) { 1612 maybe_warnx("(stdin): unexpected end of file"); 1613 return; 1614 } 1615 1616 method = file_gettype(header1); 1617 switch (method) { 1618 default: 1619 #ifndef SMALL 1620 if (fflag == 0) { 1621 maybe_warnx("unknown compression format"); 1622 return; 1623 } 1624 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1625 break; 1626 #endif 1627 case FT_GZIP: 1628 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1629 header1, sizeof header1, &gsize, "(stdin)"); 1630 break; 1631 #ifndef NO_BZIP2_SUPPORT 1632 case FT_BZIP2: 1633 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1634 header1, sizeof header1, &gsize); 1635 break; 1636 #endif 1637 #ifndef NO_COMPRESS_SUPPORT 1638 case FT_Z: 1639 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1640 maybe_warnx("zopen of stdin"); 1641 return; 1642 } 1643 1644 usize = zuncompress(in, stdout, header1, sizeof header1, &gsize); 1645 fclose(in); 1646 break; 1647 #endif 1648 #ifndef NO_PACK_SUPPORT 1649 case FT_PACK: 1650 usize = unpack(STDIN_FILENO, STDOUT_FILENO, 1651 (char *)header1, sizeof header1, &gsize); 1652 break; 1653 #endif 1654 } 1655 1656 #ifndef SMALL 1657 if (vflag && !tflag && usize != -1 && gsize != -1) 1658 print_verbage(NULL, NULL, usize, gsize); 1659 if (vflag && tflag) 1660 print_test("(stdin)", usize != -1); 1661 #endif 1662 1663 } 1664 1665 static void 1666 handle_stdout(void) 1667 { 1668 off_t gsize, usize; 1669 struct stat sb; 1670 time_t systime; 1671 uint32_t mtime; 1672 int ret; 1673 1674 #ifndef SMALL 1675 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1676 maybe_warnx("standard output is a terminal -- ignoring"); 1677 return; 1678 } 1679 #endif 1680 /* If stdin is a file use it's mtime, otherwise use current time */ 1681 ret = fstat(STDIN_FILENO, &sb); 1682 1683 #ifndef SMALL 1684 if (ret < 0) { 1685 maybe_warn("Can't stat stdin"); 1686 return; 1687 } 1688 #endif 1689 1690 if (S_ISREG(sb.st_mode)) 1691 mtime = (uint32_t)sb.st_mtime; 1692 else { 1693 systime = time(NULL); 1694 #ifndef SMALL 1695 if (systime == -1) { 1696 maybe_warn("time"); 1697 return; 1698 } 1699 #endif 1700 mtime = (uint32_t)systime; 1701 } 1702 1703 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1704 #ifndef SMALL 1705 if (vflag && !tflag && usize != -1 && gsize != -1) 1706 print_verbage(NULL, NULL, usize, gsize); 1707 #endif 1708 } 1709 1710 /* do what is asked for, for the path name */ 1711 static void 1712 handle_pathname(char *path) 1713 { 1714 char *opath = path, *s = NULL; 1715 ssize_t len; 1716 int slen; 1717 struct stat sb; 1718 1719 /* check for stdout/stdin */ 1720 if (path[0] == '-' && path[1] == '\0') { 1721 if (dflag) 1722 handle_stdin(); 1723 else 1724 handle_stdout(); 1725 return; 1726 } 1727 1728 retry: 1729 if (stat(path, &sb) != 0) { 1730 /* lets try <path>.gz if we're decompressing */ 1731 if (dflag && s == NULL && errno == ENOENT) { 1732 len = strlen(path); 1733 slen = suffixes[0].ziplen; 1734 s = malloc(len + slen + 1); 1735 if (s == NULL) 1736 maybe_err("malloc"); 1737 memcpy(s, path, len); 1738 memcpy(s + len, suffixes[0].zipped, slen + 1); 1739 path = s; 1740 goto retry; 1741 } 1742 maybe_warn("can't stat: %s", opath); 1743 goto out; 1744 } 1745 1746 if (S_ISDIR(sb.st_mode)) { 1747 #ifndef SMALL 1748 if (rflag) 1749 handle_dir(path); 1750 else 1751 #endif 1752 maybe_warnx("%s is a directory", path); 1753 goto out; 1754 } 1755 1756 if (S_ISREG(sb.st_mode)) 1757 handle_file(path, &sb); 1758 else 1759 maybe_warnx("%s is not a regular file", path); 1760 1761 out: 1762 if (s) 1763 free(s); 1764 } 1765 1766 /* compress/decompress a file */ 1767 static void 1768 handle_file(char *file, struct stat *sbp) 1769 { 1770 off_t usize, gsize; 1771 char outfile[PATH_MAX]; 1772 1773 infile = file; 1774 if (dflag) { 1775 usize = file_uncompress(file, outfile, sizeof(outfile)); 1776 #ifndef SMALL 1777 if (vflag && tflag) 1778 print_test(file, usize != -1); 1779 #endif 1780 if (usize == -1) 1781 return; 1782 gsize = sbp->st_size; 1783 } else { 1784 gsize = file_compress(file, outfile, sizeof(outfile)); 1785 if (gsize == -1) 1786 return; 1787 usize = sbp->st_size; 1788 } 1789 1790 1791 #ifndef SMALL 1792 if (vflag && !tflag) 1793 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1794 #endif 1795 } 1796 1797 #ifndef SMALL 1798 /* this is used with -r to recursively descend directories */ 1799 static void 1800 handle_dir(char *dir) 1801 { 1802 char *path_argv[2]; 1803 FTS *fts; 1804 FTSENT *entry; 1805 1806 path_argv[0] = dir; 1807 path_argv[1] = 0; 1808 fts = fts_open(path_argv, FTS_PHYSICAL, NULL); 1809 if (fts == NULL) { 1810 warn("couldn't fts_open %s", dir); 1811 return; 1812 } 1813 1814 while ((entry = fts_read(fts))) { 1815 switch(entry->fts_info) { 1816 case FTS_D: 1817 case FTS_DP: 1818 continue; 1819 1820 case FTS_DNR: 1821 case FTS_ERR: 1822 case FTS_NS: 1823 maybe_warn("%s", entry->fts_path); 1824 continue; 1825 case FTS_F: 1826 handle_file(entry->fts_name, entry->fts_statp); 1827 } 1828 } 1829 (void)fts_close(fts); 1830 } 1831 #endif 1832 1833 /* print a ratio - size reduction as a fraction of uncompressed size */ 1834 static void 1835 print_ratio(off_t in, off_t out, FILE *where) 1836 { 1837 int percent10; /* 10 * percent */ 1838 off_t diff; 1839 char buff[8]; 1840 int len; 1841 1842 diff = in - out/2; 1843 if (diff <= 0) 1844 /* 1845 * Output is more than double size of input! print -99.9% 1846 * Quite possibly we've failed to get the original size. 1847 */ 1848 percent10 = -999; 1849 else { 1850 /* 1851 * We only need 12 bits of result from the final division, 1852 * so reduce the values until a 32bit division will suffice. 1853 */ 1854 while (in > 0x100000) { 1855 diff >>= 1; 1856 in >>= 1; 1857 } 1858 if (in != 0) 1859 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1860 else 1861 percent10 = 0; 1862 } 1863 1864 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1865 /* Move the '.' to before the last digit */ 1866 buff[len - 1] = buff[len - 2]; 1867 buff[len - 2] = '.'; 1868 fprintf(where, "%5s%%", buff); 1869 } 1870 1871 #ifndef SMALL 1872 /* print compression statistics, and the new name (if there is one!) */ 1873 static void 1874 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1875 { 1876 if (file) 1877 fprintf(stderr, "%s:%s ", file, 1878 strlen(file) < 7 ? "\t\t" : "\t"); 1879 print_ratio(usize, gsize, stderr); 1880 if (nfile) 1881 fprintf(stderr, " -- replaced with %s", nfile); 1882 fprintf(stderr, "\n"); 1883 fflush(stderr); 1884 } 1885 1886 /* print test results */ 1887 static void 1888 print_test(const char *file, int ok) 1889 { 1890 1891 if (exit_value == 0 && ok == 0) 1892 exit_value = 1; 1893 fprintf(stderr, "%s:%s %s\n", file, 1894 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1895 fflush(stderr); 1896 } 1897 #endif 1898 1899 /* print a file's info ala --list */ 1900 /* eg: 1901 compressed uncompressed ratio uncompressed_name 1902 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1903 */ 1904 static void 1905 print_list(int fd, off_t out, const char *outfile, time_t ts) 1906 { 1907 static int first = 1; 1908 #ifndef SMALL 1909 static off_t in_tot, out_tot; 1910 uint32_t crc = 0; 1911 #endif 1912 off_t in = 0, rv; 1913 1914 if (first) { 1915 #ifndef SMALL 1916 if (vflag) 1917 printf("method crc date time "); 1918 #endif 1919 if (qflag == 0) 1920 printf(" compressed uncompressed " 1921 "ratio uncompressed_name\n"); 1922 } 1923 first = 0; 1924 1925 /* print totals? */ 1926 #ifndef SMALL 1927 if (fd == -1) { 1928 in = in_tot; 1929 out = out_tot; 1930 } else 1931 #endif 1932 { 1933 /* read the last 4 bytes - this is the uncompressed size */ 1934 rv = lseek(fd, (off_t)(-8), SEEK_END); 1935 if (rv != -1) { 1936 unsigned char buf[8]; 1937 uint32_t usize; 1938 1939 rv = read(fd, (char *)buf, sizeof(buf)); 1940 if (rv == -1) 1941 maybe_warn("read of uncompressed size"); 1942 else if (rv != sizeof(buf)) 1943 maybe_warnx("read of uncompressed size"); 1944 1945 else { 1946 usize = buf[4] | buf[5] << 8 | 1947 buf[6] << 16 | buf[7] << 24; 1948 in = (off_t)usize; 1949 #ifndef SMALL 1950 crc = buf[0] | buf[1] << 8 | 1951 buf[2] << 16 | buf[3] << 24; 1952 #endif 1953 } 1954 } 1955 } 1956 1957 #ifndef SMALL 1958 if (vflag && fd == -1) 1959 printf(" "); 1960 else if (vflag) { 1961 char *date = ctime(&ts); 1962 1963 /* skip the day, 1/100th second, and year */ 1964 date += 4; 1965 date[12] = 0; 1966 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1967 } 1968 in_tot += in; 1969 out_tot += out; 1970 #endif 1971 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1972 print_ratio(in, out, stdout); 1973 printf(" %s\n", outfile); 1974 } 1975 1976 /* display the usage of NetBSD gzip */ 1977 static void 1978 usage(void) 1979 { 1980 1981 fprintf(stderr, "%s\n", gzip_version); 1982 fprintf(stderr, 1983 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n" 1984 #ifndef SMALL 1985 " -1 --fast fastest (worst) compression\n" 1986 " -2 .. -8 set compression level\n" 1987 " -9 --best best (slowest) compression\n" 1988 " -c --stdout write to stdout, keep original files\n" 1989 " --to-stdout\n" 1990 " -d --decompress uncompress files\n" 1991 " --uncompress\n" 1992 " -f --force force overwriting & compress links\n" 1993 " -h --help display this help\n" 1994 " -l --list list compressed file contents\n" 1995 " -N --name save or restore original file name and time stamp\n" 1996 " -n --no-name don't save original file name or time stamp\n" 1997 " -q --quiet output no warnings\n" 1998 " -r --recursive recursively compress files in directories\n" 1999 " -S .suf use suffix .suf instead of .gz\n" 2000 " --suffix .suf\n" 2001 " -t --test test compressed file\n" 2002 " -V --version display program version\n" 2003 " -v --verbose print extra statistics\n", 2004 #else 2005 , 2006 #endif 2007 getprogname()); 2008 exit(0); 2009 } 2010 2011 /* display the version of NetBSD gzip */ 2012 static void 2013 display_version(void) 2014 { 2015 2016 fprintf(stderr, "%s\n", gzip_version); 2017 exit(0); 2018 } 2019 2020 #ifndef NO_BZIP2_SUPPORT 2021 #include "unbzip2.c" 2022 #endif 2023 #ifndef NO_COMPRESS_SUPPORT 2024 #include "zuncompress.c" 2025 #endif 2026 #ifndef NO_PACK_SUPPORT 2027 #include "unpack.c" 2028 #endif 2029 2030 static ssize_t 2031 read_retry(int fd, void *buf, size_t sz) 2032 { 2033 char *cp = buf; 2034 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2035 2036 while (left > 0) { 2037 ssize_t ret; 2038 2039 ret = read(fd, cp, left); 2040 if (ret == -1) { 2041 return ret; 2042 } else if (ret == 0) { 2043 break; /* EOF */ 2044 } 2045 cp += ret; 2046 left -= ret; 2047 } 2048 2049 return sz - left; 2050 } 2051