1 /* $NetBSD: gzip.c,v 1.94 2009/04/12 10:31:14 lukem Exp $ */ 2 /* $DragonFly: src/usr.bin/gzip/gzip.c,v 1.7 2007/12/06 19:54:52 hasso Exp $ */ 3 4 /* 5 * Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 18 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 19 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 20 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 21 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 22 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 23 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 24 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 25 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 */ 29 30 /* 31 * gzip.c -- GPL free gzip using zlib. 32 * 33 * RFC 1950 covers the zlib format 34 * RFC 1951 covers the deflate format 35 * RFC 1952 covers the gzip format 36 * 37 * TODO: 38 * - use mmap where possible 39 * - handle some signals better (remove outfile?) 40 * - make bzip2/compress -v/-t/-l support work as well as possible 41 */ 42 43 #include <sys/param.h> 44 #include <sys/stat.h> 45 #include <sys/time.h> 46 47 #include <err.h> 48 #include <errno.h> 49 #include <fcntl.h> 50 #include <fts.h> 51 #include <getopt.h> 52 #include <inttypes.h> 53 #include <libgen.h> 54 #include <stdarg.h> 55 #include <stdio.h> 56 #include <stdlib.h> 57 #include <string.h> 58 #include <time.h> 59 #include <unistd.h> 60 #include <zlib.h> 61 62 #ifndef PRIdOFF 63 #define PRIdOFF PRId64 64 #endif 65 66 #ifndef PRId64 67 #define PRId64 "lld" 68 #endif 69 70 /* what type of file are we dealing with */ 71 enum filetype { 72 FT_GZIP, 73 #ifndef NO_BZIP2_SUPPORT 74 FT_BZIP2, 75 #endif 76 #ifndef NO_COMPRESS_SUPPORT 77 FT_Z, 78 #endif 79 #ifndef NO_PACK_SUPPORT 80 FT_PACK, 81 #endif 82 FT_LAST, 83 FT_UNKNOWN 84 }; 85 86 #ifndef NO_BZIP2_SUPPORT 87 #include <bzlib.h> 88 89 #define BZ2_SUFFIX ".bz2" 90 #define BZIP2_MAGIC "\102\132\150" 91 #endif 92 93 #ifndef NO_COMPRESS_SUPPORT 94 #define Z_SUFFIX ".Z" 95 #define Z_MAGIC "\037\235" 96 #endif 97 98 #ifndef NO_PACK_SUPPORT 99 #define PACK_MAGIC "\037\036" 100 #endif 101 102 #define GZ_SUFFIX ".gz" 103 104 #define BUFLEN (64 * 1024) 105 106 #define GZIP_MAGIC0 0x1F 107 #define GZIP_MAGIC1 0x8B 108 #define GZIP_OMAGIC1 0x9E 109 110 #define GZIP_TIMESTAMP (off_t)4 111 #define GZIP_ORIGNAME (off_t)10 112 113 #define HEAD_CRC 0x02 114 #define EXTRA_FIELD 0x04 115 #define ORIG_NAME 0x08 116 #define COMMENT 0x10 117 118 #define OS_CODE 3 /* Unix */ 119 120 typedef struct { 121 const char *zipped; 122 int ziplen; 123 const char *normal; /* for unzip - must not be longer than zipped */ 124 } suffixes_t; 125 static suffixes_t suffixes[] = { 126 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 127 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 128 #ifndef SMALL 129 SUFFIX(GZ_SUFFIX, ""), 130 SUFFIX(".z", ""), 131 SUFFIX("-gz", ""), 132 SUFFIX("-z", ""), 133 SUFFIX("_z", ""), 134 SUFFIX(".taz", ".tar"), 135 SUFFIX(".tgz", ".tar"), 136 #ifndef NO_BZIP2_SUPPORT 137 SUFFIX(BZ2_SUFFIX, ""), 138 #endif 139 #ifndef NO_COMPRESS_SUPPORT 140 SUFFIX(Z_SUFFIX, ""), 141 #endif 142 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 143 #endif /* SMALL */ 144 #undef SUFFIX 145 }; 146 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 147 148 static const char gzip_version[] = "NetBSD gzip 20060927"; 149 150 static int cflag; /* stdout mode */ 151 static int dflag; /* decompress mode */ 152 static int lflag; /* list mode */ 153 static int numflag = 6; /* gzip -1..-9 value */ 154 155 #ifndef SMALL 156 static int fflag; /* force mode */ 157 static int kflag; /* don't delete input files */ 158 static int nflag; /* don't save name/timestamp */ 159 static int Nflag; /* don't restore name/timestamp */ 160 static int qflag; /* quiet mode */ 161 static int rflag; /* recursive mode */ 162 static int tflag; /* test */ 163 static int vflag; /* verbose mode */ 164 #else 165 #define qflag 0 166 #define tflag 0 167 #endif 168 169 static int exit_value = 0; /* exit value */ 170 171 static char *infile; /* name of file coming in */ 172 173 static void maybe_err(const char *fmt, ...) 174 __attribute__((__format__(__printf__, 1, 2))); 175 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) 176 static void maybe_errx(const char *fmt, ...) 177 __attribute__((__format__(__printf__, 1, 2))); 178 #endif 179 static void maybe_warn(const char *fmt, ...) 180 __attribute__((__format__(__printf__, 1, 2))); 181 static void maybe_warnx(const char *fmt, ...) 182 __attribute__((__format__(__printf__, 1, 2))); 183 static enum filetype file_gettype(u_char *); 184 #ifdef SMALL 185 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 186 #endif 187 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 188 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 189 static off_t file_compress(char *, char *, size_t); 190 static off_t file_uncompress(char *, char *, size_t); 191 static void handle_pathname(char *); 192 static void handle_file(char *, struct stat *); 193 static void handle_stdin(void); 194 static void handle_stdout(void); 195 static void print_ratio(off_t, off_t, FILE *); 196 static void print_list(int fd, off_t, const char *, time_t); 197 static void usage(void); 198 static void display_version(void); 199 static const suffixes_t *check_suffix(char *, int); 200 static ssize_t read_retry(int, void *, size_t); 201 202 #ifdef SMALL 203 #define unlink_input(f, sb) unlink(f) 204 #else 205 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 206 static void prepend_gzip(char *, int *, char ***); 207 static void handle_dir(char *); 208 static void print_verbage(const char *, const char *, off_t, off_t); 209 static void print_test(const char *, int); 210 static void copymodes(int fd, const struct stat *, const char *file); 211 static int check_outfile(const char *outfile); 212 #endif 213 214 #ifndef NO_BZIP2_SUPPORT 215 static off_t unbzip2(int, int, char *, size_t, off_t *); 216 #endif 217 218 #ifndef NO_COMPRESS_SUPPORT 219 static FILE *zdopen(int); 220 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 221 #endif 222 223 #ifndef NO_PACK_SUPPORT 224 static off_t unpack(int, int, char *, size_t, off_t *); 225 #endif 226 227 int main(int, char *p[]); 228 229 #ifdef SMALL 230 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 231 #else 232 static const struct option longopts[] = { 233 { "stdout", no_argument, 0, 'c' }, 234 { "to-stdout", no_argument, 0, 'c' }, 235 { "decompress", no_argument, 0, 'd' }, 236 { "uncompress", no_argument, 0, 'd' }, 237 { "force", no_argument, 0, 'f' }, 238 { "help", no_argument, 0, 'h' }, 239 { "keep", no_argument, 0, 'k' }, 240 { "list", no_argument, 0, 'l' }, 241 { "no-name", no_argument, 0, 'n' }, 242 { "name", no_argument, 0, 'N' }, 243 { "quiet", no_argument, 0, 'q' }, 244 { "recursive", no_argument, 0, 'r' }, 245 { "suffix", required_argument, 0, 'S' }, 246 { "test", no_argument, 0, 't' }, 247 { "verbose", no_argument, 0, 'v' }, 248 { "version", no_argument, 0, 'V' }, 249 { "fast", no_argument, 0, '1' }, 250 { "best", no_argument, 0, '9' }, 251 #if 0 252 /* 253 * This is what else GNU gzip implements. --ascii isn't useful 254 * on NetBSD, and I don't care to have a --license. 255 */ 256 { "ascii", no_argument, 0, 'a' }, 257 { "license", no_argument, 0, 'L' }, 258 #endif 259 { NULL, no_argument, 0, 0 }, 260 }; 261 #endif 262 263 int 264 main(int argc, char **argv) 265 { 266 const char *progname = getprogname(); 267 #ifndef SMALL 268 char *gzip; 269 int len; 270 #endif 271 int ch; 272 273 /* XXX set up signals */ 274 275 #ifndef SMALL 276 if ((gzip = getenv("GZIP")) != NULL) 277 prepend_gzip(gzip, &argc, &argv); 278 #endif 279 280 /* 281 * XXX 282 * handle being called `gunzip', `zcat' and `gzcat' 283 */ 284 if (strcmp(progname, "gunzip") == 0) 285 dflag = 1; 286 else if (strcmp(progname, "zcat") == 0 || 287 strcmp(progname, "gzcat") == 0) 288 dflag = cflag = 1; 289 290 #ifdef SMALL 291 #define OPT_LIST "123456789cdhltV" 292 #else 293 #define OPT_LIST "123456789cdfhklNnqrS:tVv" 294 #endif 295 296 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 297 switch (ch) { 298 case '1': case '2': case '3': 299 case '4': case '5': case '6': 300 case '7': case '8': case '9': 301 numflag = ch - '0'; 302 break; 303 case 'c': 304 cflag = 1; 305 break; 306 case 'd': 307 dflag = 1; 308 break; 309 case 'l': 310 lflag = 1; 311 dflag = 1; 312 break; 313 case 'V': 314 display_version(); 315 /* NOTREACHED */ 316 #ifndef SMALL 317 case 'f': 318 fflag = 1; 319 break; 320 case 'k': 321 kflag = 1; 322 break; 323 case 'N': 324 nflag = 0; 325 Nflag = 1; 326 break; 327 case 'n': 328 nflag = 1; 329 Nflag = 0; 330 break; 331 case 'q': 332 qflag = 1; 333 break; 334 case 'r': 335 rflag = 1; 336 break; 337 case 'S': 338 len = strlen(optarg); 339 if (len != 0) { 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 (kflag) 1131 return; 1132 if (stat(file, &nsb) != 0) 1133 /* Must be gone alrady */ 1134 return; 1135 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1136 /* Definitely a different file */ 1137 return; 1138 unlink(file); 1139 } 1140 #endif 1141 1142 static const suffixes_t * 1143 check_suffix(char *file, int xlate) 1144 { 1145 const suffixes_t *s; 1146 int len = strlen(file); 1147 char *sp; 1148 1149 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1150 /* if it doesn't fit in "a.suf", don't bother */ 1151 if (s->ziplen >= len) 1152 continue; 1153 sp = file + len - s->ziplen; 1154 if (strcmp(s->zipped, sp) != 0) 1155 continue; 1156 if (xlate) 1157 strcpy(sp, s->normal); 1158 return s; 1159 } 1160 return NULL; 1161 } 1162 1163 /* 1164 * compress the given file: create a corresponding .gz file and remove the 1165 * original. 1166 */ 1167 static off_t 1168 file_compress(char *file, char *outfile, size_t outsize) 1169 { 1170 int in; 1171 int out; 1172 off_t size, insize; 1173 #ifndef SMALL 1174 struct stat isb, osb; 1175 const suffixes_t *suff; 1176 #endif 1177 1178 in = open(file, O_RDONLY); 1179 if (in == -1) { 1180 maybe_warn("can't open %s", file); 1181 return -1; 1182 } 1183 1184 if (cflag == 0) { 1185 #ifndef SMALL 1186 if (fstat(in, &isb) == 0) { 1187 if (isb.st_nlink > 1 && fflag == 0) { 1188 maybe_warnx("%s has %d other link%s -- " 1189 "skipping", file, isb.st_nlink - 1, 1190 isb.st_nlink == 1 ? "" : "s"); 1191 close(in); 1192 return -1; 1193 } 1194 } 1195 1196 if (fflag == 0 && (suff = check_suffix(file, 0)) 1197 && suff->zipped[0] != 0) { 1198 maybe_warnx("%s already has %s suffix -- unchanged", 1199 file, suff->zipped); 1200 close(in); 1201 return -1; 1202 } 1203 #endif 1204 1205 /* Add (usually) .gz to filename */ 1206 if ((size_t)snprintf(outfile, outsize, "%s%s", 1207 file, suffixes[0].zipped) >= outsize) 1208 memcpy(outfile - suffixes[0].ziplen - 1, 1209 suffixes[0].zipped, suffixes[0].ziplen + 1); 1210 1211 #ifndef SMALL 1212 if (check_outfile(outfile) == 0) { 1213 close(in); 1214 return -1; 1215 } 1216 #endif 1217 } 1218 1219 if (cflag == 0) { 1220 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1221 if (out == -1) { 1222 maybe_warn("could not create output: %s", outfile); 1223 fclose(stdin); 1224 return -1; 1225 } 1226 } else 1227 out = STDOUT_FILENO; 1228 1229 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1230 1231 (void)close(in); 1232 1233 /* 1234 * If there was an error, insize will be -1. 1235 * If we compressed to stdout, just return the size. 1236 * Otherwise stat the file and check it is the correct size. 1237 * We only blow away the file if we can stat the output and it 1238 * has the expected size. 1239 */ 1240 if (cflag != 0) 1241 return insize == -1 ? -1 : size; 1242 1243 #ifndef SMALL 1244 if (fstat(out, &osb) != 0) { 1245 maybe_warn("couldn't stat: %s", outfile); 1246 goto bad_outfile; 1247 } 1248 1249 if (osb.st_size != size) { 1250 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1251 " != %" PRIdOFF "), deleting", 1252 outfile, osb.st_size, size); 1253 goto bad_outfile; 1254 } 1255 1256 copymodes(out, &isb, outfile); 1257 #endif 1258 if (close(out) == -1) 1259 maybe_warn("couldn't close output"); 1260 1261 /* output is good, ok to delete input */ 1262 unlink_input(file, &isb); 1263 return size; 1264 1265 #ifndef SMALL 1266 bad_outfile: 1267 if (close(out) == -1) 1268 maybe_warn("couldn't close output"); 1269 1270 maybe_warnx("leaving original %s", file); 1271 unlink(outfile); 1272 return size; 1273 #endif 1274 } 1275 1276 /* uncompress the given file and remove the original */ 1277 static off_t 1278 file_uncompress(char *file, char *outfile, size_t outsize) 1279 { 1280 struct stat isb, osb; 1281 off_t size; 1282 ssize_t rbytes; 1283 unsigned char header1[4]; 1284 enum filetype method; 1285 int fd, ofd, zfd = -1; 1286 #ifndef SMALL 1287 ssize_t rv; 1288 time_t timestamp = 0; 1289 unsigned char name[PATH_MAX + 1]; 1290 #endif 1291 1292 /* gather the old name info */ 1293 1294 fd = open(file, O_RDONLY); 1295 if (fd < 0) { 1296 maybe_warn("can't open %s", file); 1297 goto lose; 1298 } 1299 1300 strlcpy(outfile, file, outsize); 1301 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1302 maybe_warnx("%s: unknown suffix -- ignored", file); 1303 goto lose; 1304 } 1305 1306 rbytes = read(fd, header1, sizeof header1); 1307 if (rbytes != sizeof header1) { 1308 /* we don't want to fail here. */ 1309 #ifndef SMALL 1310 if (fflag) 1311 goto lose; 1312 #endif 1313 if (rbytes == -1) 1314 maybe_warn("can't read %s", file); 1315 else 1316 goto unexpected_EOF; 1317 goto lose; 1318 } 1319 1320 method = file_gettype(header1); 1321 1322 #ifndef SMALL 1323 if (fflag == 0 && method == FT_UNKNOWN) { 1324 maybe_warnx("%s: not in gzip format", file); 1325 goto lose; 1326 } 1327 1328 #endif 1329 1330 #ifndef SMALL 1331 if (method == FT_GZIP && Nflag) { 1332 unsigned char ts[4]; /* timestamp */ 1333 1334 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1335 if (rv >= 0 && rv < (ssize_t)(sizeof ts)) 1336 goto unexpected_EOF; 1337 if (rv == -1) { 1338 if (!fflag) 1339 maybe_warn("can't read %s", file); 1340 goto lose; 1341 } 1342 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1343 1344 if (header1[3] & ORIG_NAME) { 1345 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1346 if (rbytes < 0) { 1347 maybe_warn("can't read %s", file); 1348 goto lose; 1349 } 1350 if (name[0] != 0) { 1351 /* preserve original directory name */ 1352 char *dp = strrchr(file, '/'); 1353 if (dp == NULL) 1354 dp = file; 1355 else 1356 dp++; 1357 snprintf(outfile, outsize, "%.*s%.*s", 1358 (int) (dp - file), 1359 file, (int) rbytes, name); 1360 } 1361 } 1362 } 1363 #endif 1364 lseek(fd, 0, SEEK_SET); 1365 1366 if (cflag == 0 || lflag) { 1367 if (fstat(fd, &isb) != 0) 1368 goto lose; 1369 #ifndef SMALL 1370 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1371 maybe_warnx("%s has %d other links -- skipping", 1372 file, isb.st_nlink - 1); 1373 goto lose; 1374 } 1375 if (nflag == 0 && timestamp) 1376 isb.st_mtime = timestamp; 1377 if (check_outfile(outfile) == 0) 1378 goto lose; 1379 #endif 1380 } 1381 1382 if (cflag == 0 && lflag == 0) { 1383 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1384 if (zfd == STDOUT_FILENO) { 1385 /* We won't close STDOUT_FILENO later... */ 1386 zfd = dup(zfd); 1387 close(STDOUT_FILENO); 1388 } 1389 if (zfd == -1) { 1390 maybe_warn("can't open %s", outfile); 1391 goto lose; 1392 } 1393 } else 1394 zfd = STDOUT_FILENO; 1395 1396 #ifndef NO_BZIP2_SUPPORT 1397 if (method == FT_BZIP2) { 1398 1399 /* XXX */ 1400 if (lflag) { 1401 maybe_warnx("no -l with bzip2 files"); 1402 goto lose; 1403 } 1404 1405 size = unbzip2(fd, zfd, NULL, 0, NULL); 1406 } else 1407 #endif 1408 1409 #ifndef NO_COMPRESS_SUPPORT 1410 if (method == FT_Z) { 1411 FILE *in, *out; 1412 1413 /* XXX */ 1414 if (lflag) { 1415 maybe_warnx("no -l with Lempel-Ziv files"); 1416 goto lose; 1417 } 1418 1419 if ((in = zdopen(fd)) == NULL) { 1420 maybe_warn("zdopen for read: %s", file); 1421 goto lose; 1422 } 1423 1424 out = fdopen(dup(zfd), "w"); 1425 if (out == NULL) { 1426 maybe_warn("fdopen for write: %s", outfile); 1427 fclose(in); 1428 goto lose; 1429 } 1430 1431 size = zuncompress(in, out, NULL, 0, NULL); 1432 /* need to fclose() if ferror() is true... */ 1433 if (ferror(in) | fclose(in)) { 1434 maybe_warn("failed infile fclose"); 1435 unlink(outfile); 1436 (void)fclose(out); 1437 } 1438 if (fclose(out) != 0) { 1439 maybe_warn("failed outfile fclose"); 1440 unlink(outfile); 1441 goto lose; 1442 } 1443 } else 1444 #endif 1445 1446 #ifndef NO_PACK_SUPPORT 1447 if (method == FT_PACK) { 1448 if (lflag) { 1449 maybe_warnx("no -l with packed files"); 1450 goto lose; 1451 } 1452 1453 size = unpack(fd, zfd, NULL, 0, NULL); 1454 } else 1455 #endif 1456 1457 #ifndef SMALL 1458 if (method == FT_UNKNOWN) { 1459 if (lflag) { 1460 maybe_warnx("no -l for unknown filetypes"); 1461 goto lose; 1462 } 1463 size = cat_fd(NULL, 0, NULL, fd); 1464 } else 1465 #endif 1466 { 1467 if (lflag) { 1468 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1469 close(fd); 1470 return -1; /* XXX */ 1471 } 1472 1473 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1474 } 1475 1476 if (close(fd) != 0) 1477 maybe_warn("couldn't close input"); 1478 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1479 maybe_warn("couldn't close output"); 1480 1481 if (size == -1) { 1482 if (cflag == 0) 1483 unlink(outfile); 1484 maybe_warnx("%s: uncompress failed", file); 1485 return -1; 1486 } 1487 1488 /* if testing, or we uncompressed to stdout, this is all we need */ 1489 #ifndef SMALL 1490 if (tflag) 1491 return size; 1492 #endif 1493 /* if we are uncompressing to stdin, don't remove the file. */ 1494 if (cflag) 1495 return size; 1496 1497 /* 1498 * if we create a file... 1499 */ 1500 /* 1501 * if we can't stat the file don't remove the file. 1502 */ 1503 1504 ofd = open(outfile, O_RDWR, 0); 1505 if (ofd == -1) { 1506 maybe_warn("couldn't open (leaving original): %s", 1507 outfile); 1508 return -1; 1509 } 1510 if (fstat(ofd, &osb) != 0) { 1511 maybe_warn("couldn't stat (leaving original): %s", 1512 outfile); 1513 close(ofd); 1514 return -1; 1515 } 1516 if (osb.st_size != size) { 1517 maybe_warnx("stat gave different size: %" PRIdOFF 1518 " != %" PRIdOFF " (leaving original)", 1519 size, osb.st_size); 1520 close(ofd); 1521 unlink(outfile); 1522 return -1; 1523 } 1524 unlink_input(file, &isb); 1525 #ifndef SMALL 1526 copymodes(ofd, &isb, outfile); 1527 #endif 1528 close(ofd); 1529 return size; 1530 1531 unexpected_EOF: 1532 maybe_warnx("%s: unexpected end of file", file); 1533 lose: 1534 if (fd != -1) 1535 close(fd); 1536 if (zfd != -1 && zfd != STDOUT_FILENO) 1537 close(fd); 1538 return -1; 1539 } 1540 1541 #ifndef SMALL 1542 static off_t 1543 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1544 { 1545 char buf[BUFLEN]; 1546 off_t in_tot; 1547 ssize_t w; 1548 1549 in_tot = count; 1550 w = write(STDOUT_FILENO, prepend, count); 1551 if (w == -1 || (size_t)w != count) { 1552 maybe_warn("write to stdout"); 1553 return -1; 1554 } 1555 for (;;) { 1556 ssize_t rv; 1557 1558 rv = read(fd, buf, sizeof buf); 1559 if (rv == 0) 1560 break; 1561 if (rv < 0) { 1562 maybe_warn("read from fd %d", fd); 1563 break; 1564 } 1565 1566 if (write(STDOUT_FILENO, buf, rv) != rv) { 1567 maybe_warn("write to stdout"); 1568 break; 1569 } 1570 in_tot += rv; 1571 } 1572 1573 if (gsizep) 1574 *gsizep = in_tot; 1575 return (in_tot); 1576 } 1577 #endif 1578 1579 static void 1580 handle_stdin(void) 1581 { 1582 unsigned char header1[4]; 1583 off_t usize, gsize; 1584 enum filetype method; 1585 ssize_t bytes_read; 1586 #ifndef NO_COMPRESS_SUPPORT 1587 FILE *in; 1588 #endif 1589 1590 #ifndef SMALL 1591 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1592 maybe_warnx("standard input is a terminal -- ignoring"); 1593 return; 1594 } 1595 #endif 1596 1597 if (lflag) { 1598 struct stat isb; 1599 1600 /* XXX could read the whole file, etc. */ 1601 if (fstat(STDIN_FILENO, &isb) < 0) { 1602 maybe_warn("fstat"); 1603 return; 1604 } 1605 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1606 return; 1607 } 1608 1609 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1610 if (bytes_read == -1) { 1611 maybe_warn("can't read stdin"); 1612 return; 1613 } else if (bytes_read != sizeof(header1)) { 1614 maybe_warnx("(stdin): unexpected end of file"); 1615 return; 1616 } 1617 1618 method = file_gettype(header1); 1619 switch (method) { 1620 default: 1621 #ifndef SMALL 1622 if (fflag == 0) { 1623 maybe_warnx("unknown compression format"); 1624 return; 1625 } 1626 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1627 break; 1628 #endif 1629 case FT_GZIP: 1630 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1631 header1, sizeof header1, &gsize, "(stdin)"); 1632 break; 1633 #ifndef NO_BZIP2_SUPPORT 1634 case FT_BZIP2: 1635 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1636 header1, sizeof header1, &gsize); 1637 break; 1638 #endif 1639 #ifndef NO_COMPRESS_SUPPORT 1640 case FT_Z: 1641 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1642 maybe_warnx("zopen of stdin"); 1643 return; 1644 } 1645 1646 usize = zuncompress(in, stdout, header1, sizeof header1, &gsize); 1647 fclose(in); 1648 break; 1649 #endif 1650 #ifndef NO_PACK_SUPPORT 1651 case FT_PACK: 1652 usize = unpack(STDIN_FILENO, STDOUT_FILENO, 1653 (char *)header1, sizeof header1, &gsize); 1654 break; 1655 #endif 1656 } 1657 1658 #ifndef SMALL 1659 if (vflag && !tflag && usize != -1 && gsize != -1) 1660 print_verbage(NULL, NULL, usize, gsize); 1661 if (vflag && tflag) 1662 print_test("(stdin)", usize != -1); 1663 #endif 1664 1665 } 1666 1667 static void 1668 handle_stdout(void) 1669 { 1670 off_t gsize, usize; 1671 struct stat sb; 1672 time_t systime; 1673 uint32_t mtime; 1674 int ret; 1675 1676 #ifndef SMALL 1677 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1678 maybe_warnx("standard output is a terminal -- ignoring"); 1679 return; 1680 } 1681 #endif 1682 /* If stdin is a file use it's mtime, otherwise use current time */ 1683 ret = fstat(STDIN_FILENO, &sb); 1684 1685 #ifndef SMALL 1686 if (ret < 0) { 1687 maybe_warn("Can't stat stdin"); 1688 return; 1689 } 1690 #endif 1691 1692 if (S_ISREG(sb.st_mode)) 1693 mtime = (uint32_t)sb.st_mtime; 1694 else { 1695 systime = time(NULL); 1696 #ifndef SMALL 1697 if (systime == -1) { 1698 maybe_warn("time"); 1699 return; 1700 } 1701 #endif 1702 mtime = (uint32_t)systime; 1703 } 1704 1705 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1706 #ifndef SMALL 1707 if (vflag && !tflag && usize != -1 && gsize != -1) 1708 print_verbage(NULL, NULL, usize, gsize); 1709 #endif 1710 } 1711 1712 /* do what is asked for, for the path name */ 1713 static void 1714 handle_pathname(char *path) 1715 { 1716 char *opath = path, *s = NULL; 1717 ssize_t len; 1718 int slen; 1719 struct stat sb; 1720 1721 /* check for stdout/stdin */ 1722 if (path[0] == '-' && path[1] == '\0') { 1723 if (dflag) 1724 handle_stdin(); 1725 else 1726 handle_stdout(); 1727 return; 1728 } 1729 1730 retry: 1731 if (stat(path, &sb) != 0) { 1732 /* lets try <path>.gz if we're decompressing */ 1733 if (dflag && s == NULL && errno == ENOENT) { 1734 len = strlen(path); 1735 slen = suffixes[0].ziplen; 1736 s = malloc(len + slen + 1); 1737 if (s == NULL) 1738 maybe_err("malloc"); 1739 memcpy(s, path, len); 1740 memcpy(s + len, suffixes[0].zipped, slen + 1); 1741 path = s; 1742 goto retry; 1743 } 1744 maybe_warn("can't stat: %s", opath); 1745 goto out; 1746 } 1747 1748 if (S_ISDIR(sb.st_mode)) { 1749 #ifndef SMALL 1750 if (rflag) 1751 handle_dir(path); 1752 else 1753 #endif 1754 maybe_warnx("%s is a directory", path); 1755 goto out; 1756 } 1757 1758 if (S_ISREG(sb.st_mode)) 1759 handle_file(path, &sb); 1760 else 1761 maybe_warnx("%s is not a regular file", path); 1762 1763 out: 1764 if (s) 1765 free(s); 1766 } 1767 1768 /* compress/decompress a file */ 1769 static void 1770 handle_file(char *file, struct stat *sbp) 1771 { 1772 off_t usize, gsize; 1773 char outfile[PATH_MAX]; 1774 1775 infile = file; 1776 if (dflag) { 1777 usize = file_uncompress(file, outfile, sizeof(outfile)); 1778 #ifndef SMALL 1779 if (vflag && tflag) 1780 print_test(file, usize != -1); 1781 #endif 1782 if (usize == -1) 1783 return; 1784 gsize = sbp->st_size; 1785 } else { 1786 gsize = file_compress(file, outfile, sizeof(outfile)); 1787 if (gsize == -1) 1788 return; 1789 usize = sbp->st_size; 1790 } 1791 1792 1793 #ifndef SMALL 1794 if (vflag && !tflag) 1795 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1796 #endif 1797 } 1798 1799 #ifndef SMALL 1800 /* this is used with -r to recursively descend directories */ 1801 static void 1802 handle_dir(char *dir) 1803 { 1804 char *path_argv[2]; 1805 FTS *fts; 1806 FTSENT *entry; 1807 1808 path_argv[0] = dir; 1809 path_argv[1] = 0; 1810 fts = fts_open(path_argv, FTS_PHYSICAL | FTS_NOCHDIR, NULL); 1811 if (fts == NULL) { 1812 warn("couldn't fts_open %s", dir); 1813 return; 1814 } 1815 1816 while ((entry = fts_read(fts))) { 1817 switch(entry->fts_info) { 1818 case FTS_D: 1819 case FTS_DP: 1820 continue; 1821 1822 case FTS_DNR: 1823 case FTS_ERR: 1824 case FTS_NS: 1825 maybe_warn("%s", entry->fts_path); 1826 continue; 1827 case FTS_F: 1828 handle_file(entry->fts_path, entry->fts_statp); 1829 } 1830 } 1831 (void)fts_close(fts); 1832 } 1833 #endif 1834 1835 /* print a ratio - size reduction as a fraction of uncompressed size */ 1836 static void 1837 print_ratio(off_t in, off_t out, FILE *where) 1838 { 1839 int percent10; /* 10 * percent */ 1840 off_t diff; 1841 char buff[8]; 1842 int len; 1843 1844 diff = in - out/2; 1845 if (diff <= 0) 1846 /* 1847 * Output is more than double size of input! print -99.9% 1848 * Quite possibly we've failed to get the original size. 1849 */ 1850 percent10 = -999; 1851 else { 1852 /* 1853 * We only need 12 bits of result from the final division, 1854 * so reduce the values until a 32bit division will suffice. 1855 */ 1856 while (in > 0x100000) { 1857 diff >>= 1; 1858 in >>= 1; 1859 } 1860 if (in != 0) 1861 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1862 else 1863 percent10 = 0; 1864 } 1865 1866 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1867 /* Move the '.' to before the last digit */ 1868 buff[len - 1] = buff[len - 2]; 1869 buff[len - 2] = '.'; 1870 fprintf(where, "%5s%%", buff); 1871 } 1872 1873 #ifndef SMALL 1874 /* print compression statistics, and the new name (if there is one!) */ 1875 static void 1876 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1877 { 1878 if (file) 1879 fprintf(stderr, "%s:%s ", file, 1880 strlen(file) < 7 ? "\t\t" : "\t"); 1881 print_ratio(usize, gsize, stderr); 1882 if (nfile) 1883 fprintf(stderr, " -- replaced with %s", nfile); 1884 fprintf(stderr, "\n"); 1885 fflush(stderr); 1886 } 1887 1888 /* print test results */ 1889 static void 1890 print_test(const char *file, int ok) 1891 { 1892 1893 if (exit_value == 0 && ok == 0) 1894 exit_value = 1; 1895 fprintf(stderr, "%s:%s %s\n", file, 1896 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1897 fflush(stderr); 1898 } 1899 #endif 1900 1901 /* print a file's info ala --list */ 1902 /* eg: 1903 compressed uncompressed ratio uncompressed_name 1904 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1905 */ 1906 static void 1907 print_list(int fd, off_t out, const char *outfile, time_t ts) 1908 { 1909 static int first = 1; 1910 #ifndef SMALL 1911 static off_t in_tot, out_tot; 1912 uint32_t crc = 0; 1913 #endif 1914 off_t in = 0, rv; 1915 1916 if (first) { 1917 #ifndef SMALL 1918 if (vflag) 1919 printf("method crc date time "); 1920 #endif 1921 if (qflag == 0) 1922 printf(" compressed uncompressed " 1923 "ratio uncompressed_name\n"); 1924 } 1925 first = 0; 1926 1927 /* print totals? */ 1928 #ifndef SMALL 1929 if (fd == -1) { 1930 in = in_tot; 1931 out = out_tot; 1932 } else 1933 #endif 1934 { 1935 /* read the last 4 bytes - this is the uncompressed size */ 1936 rv = lseek(fd, (off_t)(-8), SEEK_END); 1937 if (rv != -1) { 1938 unsigned char buf[8]; 1939 uint32_t usize; 1940 1941 rv = read(fd, (char *)buf, sizeof(buf)); 1942 if (rv == -1) 1943 maybe_warn("read of uncompressed size"); 1944 else if (rv != sizeof(buf)) 1945 maybe_warnx("read of uncompressed size"); 1946 1947 else { 1948 usize = buf[4] | buf[5] << 8 | 1949 buf[6] << 16 | buf[7] << 24; 1950 in = (off_t)usize; 1951 #ifndef SMALL 1952 crc = buf[0] | buf[1] << 8 | 1953 buf[2] << 16 | buf[3] << 24; 1954 #endif 1955 } 1956 } 1957 } 1958 1959 #ifndef SMALL 1960 if (vflag && fd == -1) 1961 printf(" "); 1962 else if (vflag) { 1963 char *date = ctime(&ts); 1964 1965 /* skip the day, 1/100th second, and year */ 1966 date += 4; 1967 date[12] = 0; 1968 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1969 } 1970 in_tot += in; 1971 out_tot += out; 1972 #endif 1973 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1974 print_ratio(in, out, stdout); 1975 printf(" %s\n", outfile); 1976 } 1977 1978 /* display the usage of NetBSD gzip */ 1979 static void 1980 usage(void) 1981 { 1982 1983 fprintf(stderr, "%s\n", gzip_version); 1984 fprintf(stderr, 1985 #ifdef SMALL 1986 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n", 1987 #else 1988 "usage: %s [-123456789acdfhklLNnqrtVv] [-S .suffix] [<file> [<file> ...]]\n" 1989 " -1 --fast fastest (worst) compression\n" 1990 " -2 .. -8 set compression level\n" 1991 " -9 --best best (slowest) compression\n" 1992 " -c --stdout write to stdout, keep original files\n" 1993 " --to-stdout\n" 1994 " -d --decompress uncompress files\n" 1995 " --uncompress\n" 1996 " -f --force force overwriting & compress links\n" 1997 " -h --help display this help\n" 1998 " -k --keep don't delete input files during operation\n" 1999 " -l --list list compressed file contents\n" 2000 " -N --name save or restore original file name and time stamp\n" 2001 " -n --no-name don't save original file name or time stamp\n" 2002 " -q --quiet output no warnings\n" 2003 " -r --recursive recursively compress files in directories\n" 2004 " -S .suf use suffix .suf instead of .gz\n" 2005 " --suffix .suf\n" 2006 " -t --test test compressed file\n" 2007 " -V --version display program version\n" 2008 " -v --verbose print extra statistics\n", 2009 #endif 2010 getprogname()); 2011 exit(0); 2012 } 2013 2014 /* display the version of NetBSD gzip */ 2015 static void 2016 display_version(void) 2017 { 2018 2019 fprintf(stderr, "%s\n", gzip_version); 2020 exit(0); 2021 } 2022 2023 #ifndef NO_BZIP2_SUPPORT 2024 #include "unbzip2.c" 2025 #endif 2026 #ifndef NO_COMPRESS_SUPPORT 2027 #include "zuncompress.c" 2028 #endif 2029 #ifndef NO_PACK_SUPPORT 2030 #include "unpack.c" 2031 #endif 2032 2033 static ssize_t 2034 read_retry(int fd, void *buf, size_t sz) 2035 { 2036 char *cp = buf; 2037 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2038 2039 while (left > 0) { 2040 ssize_t ret; 2041 2042 ret = read(fd, cp, left); 2043 if (ret == -1) { 2044 return ret; 2045 } else if (ret == 0) { 2046 break; /* EOF */ 2047 } 2048 cp += ret; 2049 left -= ret; 2050 } 2051 2052 return sz - left; 2053 } 2054