1 /* $NetBSD: gzip.c,v 1.73 2005/08/28 10:17:50 mrg Exp $ */ 2 3 /* 4 * Copyright (c) 1997, 1998, 2003, 2004 Matthew R. Green 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. The name of the author may not be used to endorse or promote products 16 * derived from this software without specific prior written permission. 17 * 18 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 19 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 20 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 21 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 22 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 23 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 24 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 25 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 26 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 28 * SUCH DAMAGE. 29 */ 30 31 #include <sys/cdefs.h> 32 #ifndef lint 33 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004 Matthew R. Green\n\ 34 All rights reserved.\n"); 35 __RCSID("$NetBSD: gzip.c,v 1.73 2005/08/28 10:17:50 mrg Exp $"); 36 #endif /* not lint */ 37 38 /* 39 * gzip.c -- GPL free gzip using zlib. 40 * 41 * RFC 1950 covers the zlib format 42 * RFC 1951 covers the deflate format 43 * RFC 1952 covers the gzip format 44 * 45 * TODO: 46 * - use mmap where possible 47 * - handle some signals better (remove outfile?) 48 * - make bzip2/compress -v/-t/-l support work as well as possible 49 */ 50 51 #include <sys/param.h> 52 #include <sys/stat.h> 53 #include <sys/time.h> 54 55 #include <inttypes.h> 56 #include <unistd.h> 57 #include <stdio.h> 58 #include <string.h> 59 #include <stdlib.h> 60 #include <err.h> 61 #include <errno.h> 62 #include <fcntl.h> 63 #include <zlib.h> 64 #include <fts.h> 65 #include <libgen.h> 66 #include <stdarg.h> 67 #include <getopt.h> 68 #include <time.h> 69 70 #ifndef PRIdOFF 71 #define PRIdOFF PRId64 72 #endif 73 74 /* what type of file are we dealing with */ 75 enum filetype { 76 FT_GZIP, 77 #ifndef NO_BZIP2_SUPPORT 78 FT_BZIP2, 79 #endif 80 #ifndef NO_COMPRESS_SUPPORT 81 FT_Z, 82 #endif 83 FT_LAST, 84 FT_UNKNOWN 85 }; 86 87 #ifndef NO_BZIP2_SUPPORT 88 #include <bzlib.h> 89 90 #define BZ2_SUFFIX ".bz2" 91 #define BZIP2_MAGIC "\102\132\150" 92 #endif 93 94 #ifndef NO_COMPRESS_SUPPORT 95 #define Z_SUFFIX ".Z" 96 #define Z_MAGIC "\037\235" 97 #endif 98 99 #define GZ_SUFFIX ".gz" 100 101 #define BUFLEN (64 * 1024) 102 103 #define GZIP_MAGIC0 0x1F 104 #define GZIP_MAGIC1 0x8B 105 #define GZIP_OMAGIC1 0x9E 106 107 #define GZIP_TIMESTAMP (off_t)4 108 #define GZIP_ORIGNAME (off_t)10 109 110 #define HEAD_CRC 0x02 111 #define EXTRA_FIELD 0x04 112 #define ORIG_NAME 0x08 113 #define COMMENT 0x10 114 115 #define OS_CODE 3 /* Unix */ 116 117 typedef struct { 118 const char *zipped; 119 int ziplen; 120 const char *normal; /* for unzip - must not be longer than zipped */ 121 } suffixes_t; 122 static suffixes_t suffixes[] = { 123 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 124 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 125 #ifndef SMALL 126 SUFFIX(GZ_SUFFIX, ""), 127 SUFFIX(".z", ""), 128 SUFFIX("-gz", ""), 129 SUFFIX("-z", ""), 130 SUFFIX("_z", ""), 131 SUFFIX(".taz", ".tar"), 132 SUFFIX(".tgz", ".tar"), 133 #ifndef NO_BZIP2_SUPPORT 134 SUFFIX(BZ2_SUFFIX, ""), 135 #endif 136 #ifndef NO_COMPRESS_SUPPORT 137 SUFFIX(Z_SUFFIX, ""), 138 #endif 139 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 140 #endif /* SMALL */ 141 #undef SUFFIX 142 }; 143 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 144 145 static const char gzip_version[] = "NetBSD gzip 20040830"; 146 147 static int cflag; /* stdout mode */ 148 static int dflag; /* decompress mode */ 149 static int lflag; /* list mode */ 150 static int numflag = 6; /* gzip -1..-9 value */ 151 152 #ifndef SMALL 153 static int fflag; /* force mode */ 154 static int nflag; /* don't save name/timestamp */ 155 static int Nflag; /* don't restore name/timestamp */ 156 static int qflag; /* quiet mode */ 157 static int rflag; /* recursive mode */ 158 static int tflag; /* test */ 159 static int vflag; /* verbose mode */ 160 #else 161 #define qflag 0 162 #define tflag 0 163 #endif 164 165 static int exit_value = 0; /* exit value */ 166 167 static char *infile; /* name of file coming in */ 168 169 static void maybe_err(const char *fmt, ...) 170 __attribute__((__format__(__printf__, 1, 2))); 171 #ifndef NO_BZIP2_SUPPORT 172 static void maybe_errx(const char *fmt, ...) 173 __attribute__((__format__(__printf__, 1, 2))); 174 #endif 175 static void maybe_warn(const char *fmt, ...) 176 __attribute__((__format__(__printf__, 1, 2))); 177 static void maybe_warnx(const char *fmt, ...) 178 __attribute__((__format__(__printf__, 1, 2))); 179 static enum filetype file_gettype(u_char *); 180 #ifdef SMALL 181 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 182 #endif 183 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 184 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 185 static off_t file_compress(char *, char *, size_t); 186 static off_t file_uncompress(char *, char *, size_t); 187 static void handle_pathname(char *); 188 static void handle_file(char *, struct stat *); 189 static void handle_stdin(void); 190 static void handle_stdout(void); 191 static void print_ratio(off_t, off_t, FILE *); 192 static void print_list(int fd, off_t, const char *, time_t); 193 static void usage(void); 194 static void display_version(void); 195 static const suffixes_t *check_suffix(char *, int); 196 static ssize_t read_retry(int, void *, size_t); 197 198 #ifdef SMALL 199 #define unlink_input(f, sb) unlink(f) 200 #else 201 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 202 static void prepend_gzip(char *, int *, char ***); 203 static void handle_dir(char *, struct stat *); 204 static void print_verbage(const char *, const char *, off_t, off_t); 205 static void print_test(const char *, int); 206 static void copymodes(const char *, struct stat *); 207 static int check_outfile(const char *outfile, struct stat *sb); 208 #endif 209 210 #ifndef NO_BZIP2_SUPPORT 211 static off_t unbzip2(int, int, char *, size_t, off_t *); 212 #endif 213 214 #ifndef NO_COMPRESS_SUPPORT 215 static FILE *zdopen(int); 216 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 217 #endif 218 219 int main(int, char *p[]); 220 221 #ifdef SMALL 222 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 223 #else 224 static const struct option longopts[] = { 225 { "stdout", no_argument, 0, 'c' }, 226 { "to-stdout", no_argument, 0, 'c' }, 227 { "decompress", no_argument, 0, 'd' }, 228 { "uncompress", no_argument, 0, 'd' }, 229 { "force", no_argument, 0, 'f' }, 230 { "help", no_argument, 0, 'h' }, 231 { "list", no_argument, 0, 'l' }, 232 { "no-name", no_argument, 0, 'n' }, 233 { "name", no_argument, 0, 'N' }, 234 { "quiet", no_argument, 0, 'q' }, 235 { "recursive", no_argument, 0, 'r' }, 236 { "suffix", required_argument, 0, 'S' }, 237 { "test", no_argument, 0, 't' }, 238 { "verbose", no_argument, 0, 'v' }, 239 { "version", no_argument, 0, 'V' }, 240 { "fast", no_argument, 0, '1' }, 241 { "best", no_argument, 0, '9' }, 242 #if 0 243 /* 244 * This is what else GNU gzip implements. --ascii isn't useful 245 * on NetBSD, and I don't care to have a --license. 246 */ 247 { "ascii", no_argument, 0, 'a' }, 248 { "license", no_argument, 0, 'L' }, 249 #endif 250 { NULL, no_argument, 0, 0 }, 251 }; 252 #endif 253 254 int 255 main(int argc, char **argv) 256 { 257 const char *progname = getprogname(); 258 #ifndef SMALL 259 char *gzip; 260 int len; 261 #endif 262 int ch; 263 264 /* XXX set up signals */ 265 266 #ifndef SMALL 267 if ((gzip = getenv("GZIP")) != NULL) 268 prepend_gzip(gzip, &argc, &argv); 269 #endif 270 271 /* 272 * XXX 273 * handle being called `gunzip', `zcat' and `gzcat' 274 */ 275 if (strcmp(progname, "gunzip") == 0) 276 dflag = 1; 277 else if (strcmp(progname, "zcat") == 0 || 278 strcmp(progname, "gzcat") == 0) 279 dflag = cflag = 1; 280 281 #ifdef SMALL 282 #define OPT_LIST "cdhHltV123456789" 283 #else 284 #define OPT_LIST "cdfhHlnNqrS:tvV123456789" 285 #endif 286 287 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 288 switch (ch) { 289 case 'c': 290 cflag = 1; 291 break; 292 case 'd': 293 dflag = 1; 294 break; 295 case 'l': 296 lflag = 1; 297 dflag = 1; 298 break; 299 case 'V': 300 display_version(); 301 /* NOTREACHED */ 302 case '1': case '2': case '3': 303 case '4': case '5': case '6': 304 case '7': case '8': case '9': 305 numflag = ch - '0'; 306 break; 307 #ifndef SMALL 308 case 'f': 309 fflag = 1; 310 break; 311 case 'n': 312 nflag = 1; 313 Nflag = 0; 314 break; 315 case 'N': 316 nflag = 0; 317 Nflag = 1; 318 break; 319 case 'q': 320 qflag = 1; 321 break; 322 case 'r': 323 rflag = 1; 324 break; 325 case 'S': 326 len = strlen(optarg); 327 if (len != 0) { 328 suffixes[0].zipped = optarg; 329 suffixes[0].ziplen = len; 330 } else { 331 suffixes[NUM_SUFFIXES - 1].zipped = ""; 332 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 333 } 334 break; 335 case 't': 336 cflag = 1; 337 tflag = 1; 338 dflag = 1; 339 break; 340 case 'v': 341 vflag = 1; 342 break; 343 #endif 344 default: 345 usage(); 346 /* NOTREACHED */ 347 } 348 } 349 argv += optind; 350 argc -= optind; 351 352 if (argc == 0) { 353 if (dflag) /* stdin mode */ 354 handle_stdin(); 355 else /* stdout mode */ 356 handle_stdout(); 357 } else { 358 do { 359 handle_pathname(argv[0]); 360 } while (*++argv); 361 } 362 #ifndef SMALL 363 if (qflag == 0 && lflag && argc > 1) 364 print_list(-1, 0, "(totals)", 0); 365 #endif 366 exit(exit_value); 367 } 368 369 /* maybe print a warning */ 370 void 371 maybe_warn(const char *fmt, ...) 372 { 373 va_list ap; 374 375 if (qflag == 0) { 376 va_start(ap, fmt); 377 vwarn(fmt, ap); 378 va_end(ap); 379 } 380 if (exit_value == 0) 381 exit_value = 1; 382 } 383 384 /* ... without an errno. */ 385 void 386 maybe_warnx(const char *fmt, ...) 387 { 388 va_list ap; 389 390 if (qflag == 0) { 391 va_start(ap, fmt); 392 vwarnx(fmt, ap); 393 va_end(ap); 394 } 395 if (exit_value == 0) 396 exit_value = 1; 397 } 398 399 /* maybe print an error */ 400 void 401 maybe_err(const char *fmt, ...) 402 { 403 va_list ap; 404 405 if (qflag == 0) { 406 va_start(ap, fmt); 407 vwarn(fmt, ap); 408 va_end(ap); 409 } 410 exit(2); 411 } 412 413 #ifndef NO_BZIP2_SUPPORT 414 /* ... without an errno. */ 415 void 416 maybe_errx(const char *fmt, ...) 417 { 418 va_list ap; 419 420 if (qflag == 0) { 421 va_start(ap, fmt); 422 vwarnx(fmt, ap); 423 va_end(ap); 424 } 425 exit(2); 426 } 427 #endif 428 429 #ifndef SMALL 430 /* split up $GZIP and prepend it to the argument list */ 431 static void 432 prepend_gzip(char *gzip, int *argc, char ***argv) 433 { 434 char *s, **nargv, **ac; 435 int nenvarg = 0, i; 436 437 /* scan how many arguments there are */ 438 for (s = gzip;;) { 439 while (*s == ' ' || *s == '\t') 440 s++; 441 if (*s == 0) 442 goto count_done; 443 nenvarg++; 444 while (*s != ' ' && *s != '\t') 445 if (*s++ == 0) 446 goto count_done; 447 } 448 count_done: 449 /* punt early */ 450 if (nenvarg == 0) 451 return; 452 453 *argc += nenvarg; 454 ac = *argv; 455 456 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 457 if (nargv == NULL) 458 maybe_err("malloc"); 459 460 /* stash this away */ 461 *argv = nargv; 462 463 /* copy the program name first */ 464 i = 0; 465 nargv[i++] = *(ac++); 466 467 /* take a copy of $GZIP and add it to the array */ 468 s = strdup(gzip); 469 if (s == NULL) 470 maybe_err("strdup"); 471 for (;;) { 472 /* Skip whitespaces. */ 473 while (*s == ' ' || *s == '\t') 474 s++; 475 if (*s == 0) 476 goto copy_done; 477 nargv[i++] = s; 478 /* Find the end of this argument. */ 479 while (*s != ' ' && *s != '\t') 480 if (*s++ == 0) 481 /* Argument followed by NUL. */ 482 goto copy_done; 483 /* Terminate by overwriting ' ' or '\t' with NUL. */ 484 *s++ = 0; 485 } 486 copy_done: 487 488 /* copy the original arguments and a NULL */ 489 while (*ac) 490 nargv[i++] = *(ac++); 491 nargv[i] = NULL; 492 } 493 #endif 494 495 /* compress input to output. Return bytes read, -1 on error */ 496 static off_t 497 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 498 { 499 z_stream z; 500 char *outbufp, *inbufp; 501 off_t in_tot = 0, out_tot = 0; 502 ssize_t in_size; 503 int i, error; 504 uLong crc; 505 #ifdef SMALL 506 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 507 0, 0, 0, 0, 508 0, OS_CODE }; 509 #endif 510 511 outbufp = malloc(BUFLEN); 512 inbufp = malloc(BUFLEN); 513 if (outbufp == NULL || inbufp == NULL) { 514 maybe_err("malloc failed"); 515 goto out; 516 } 517 518 memset(&z, 0, sizeof z); 519 z.zalloc = Z_NULL; 520 z.zfree = Z_NULL; 521 z.opaque = 0; 522 523 #ifdef SMALL 524 memcpy(outbufp, header, sizeof header); 525 i = sizeof header; 526 #else 527 if (nflag != 0) { 528 mtime = 0; 529 origname = ""; 530 } 531 532 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 533 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 534 *origname ? ORIG_NAME : 0, 535 mtime & 0xff, 536 (mtime >> 8) & 0xff, 537 (mtime >> 16) & 0xff, 538 (mtime >> 24) & 0xff, 539 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 540 OS_CODE, origname); 541 if (i >= BUFLEN) 542 /* this need PATH_MAX > BUFLEN ... */ 543 maybe_err("snprintf"); 544 if (*origname) 545 i++; 546 #endif 547 548 z.next_out = outbufp + i; 549 z.avail_out = BUFLEN - i; 550 551 error = deflateInit2(&z, numflag, Z_DEFLATED, 552 -MAX_WBITS, 8, Z_DEFAULT_STRATEGY); 553 if (error != Z_OK) { 554 maybe_warnx("deflateInit2 failed"); 555 in_tot = -1; 556 goto out; 557 } 558 559 crc = crc32(0L, Z_NULL, 0); 560 for (;;) { 561 if (z.avail_out == 0) { 562 if (write(out, outbufp, BUFLEN) != BUFLEN) { 563 maybe_warn("write"); 564 in_tot = -1; 565 goto out; 566 } 567 568 out_tot += BUFLEN; 569 z.next_out = outbufp; 570 z.avail_out = BUFLEN; 571 } 572 573 if (z.avail_in == 0) { 574 in_size = read(in, inbufp, BUFLEN); 575 if (in_size < 0) { 576 maybe_warn("read"); 577 in_tot = -1; 578 goto out; 579 } 580 if (in_size == 0) 581 break; 582 583 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 584 in_tot += in_size; 585 z.next_in = inbufp; 586 z.avail_in = in_size; 587 } 588 589 error = deflate(&z, Z_NO_FLUSH); 590 if (error != Z_OK && error != Z_STREAM_END) { 591 maybe_warnx("deflate failed"); 592 in_tot = -1; 593 goto out; 594 } 595 } 596 597 /* clean up */ 598 for (;;) { 599 size_t len; 600 601 error = deflate(&z, Z_FINISH); 602 if (error != Z_OK && error != Z_STREAM_END) { 603 maybe_warnx("deflate failed"); 604 in_tot = -1; 605 goto out; 606 } 607 608 len = (char *)z.next_out - outbufp; 609 610 if (write(out, outbufp, len) != len) { 611 maybe_warn("write"); 612 out_tot = -1; 613 goto out; 614 } 615 out_tot += len; 616 z.next_out = outbufp; 617 z.avail_out = BUFLEN; 618 619 if (error == Z_STREAM_END) 620 break; 621 } 622 623 if (deflateEnd(&z) != Z_OK) { 624 maybe_warnx("deflateEnd failed"); 625 in_tot = -1; 626 goto out; 627 } 628 629 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 630 (int)crc & 0xff, 631 (int)(crc >> 8) & 0xff, 632 (int)(crc >> 16) & 0xff, 633 (int)(crc >> 24) & 0xff, 634 (int)in_tot & 0xff, 635 (int)(in_tot >> 8) & 0xff, 636 (int)(in_tot >> 16) & 0xff, 637 (int)(in_tot >> 24) & 0xff); 638 if (i != 8) 639 maybe_err("snprintf"); 640 if (in_tot > 0xffffffff) 641 maybe_warn("input file size >= 4GB cannot be saved"); 642 if (write(out, outbufp, i) != i) { 643 maybe_warn("write"); 644 in_tot = -1; 645 } else 646 out_tot += i; 647 648 out: 649 if (inbufp != NULL) 650 free(inbufp); 651 if (outbufp != NULL) 652 free(outbufp); 653 if (gsizep) 654 *gsizep = out_tot; 655 return in_tot; 656 } 657 658 /* 659 * uncompress input to output then close the input. return the 660 * uncompressed size written, and put the compressed sized read 661 * into `*gsizep'. 662 */ 663 static off_t 664 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 665 const char *filename) 666 { 667 z_stream z; 668 char *outbufp, *inbufp; 669 off_t out_tot = -1, in_tot = 0; 670 uint32_t out_sub_tot = 0; 671 enum { 672 GZSTATE_MAGIC0, 673 GZSTATE_MAGIC1, 674 GZSTATE_METHOD, 675 GZSTATE_FLAGS, 676 GZSTATE_SKIPPING, 677 GZSTATE_EXTRA, 678 GZSTATE_EXTRA2, 679 GZSTATE_EXTRA3, 680 GZSTATE_ORIGNAME, 681 GZSTATE_COMMENT, 682 GZSTATE_HEAD_CRC1, 683 GZSTATE_HEAD_CRC2, 684 GZSTATE_INIT, 685 GZSTATE_READ, 686 GZSTATE_CRC, 687 GZSTATE_LEN, 688 } state = GZSTATE_MAGIC0; 689 int flags = 0, skip_count = 0; 690 int error = Z_STREAM_ERROR, done_reading = 0; 691 uLong crc = 0; 692 ssize_t wr; 693 int needmore = 0; 694 695 #define ADVANCE() { z.next_in++; z.avail_in--; } 696 697 if ((outbufp = malloc(BUFLEN)) == NULL) { 698 maybe_err("malloc failed"); 699 goto out2; 700 } 701 if ((inbufp = malloc(BUFLEN)) == NULL) { 702 maybe_err("malloc failed"); 703 goto out1; 704 } 705 706 memset(&z, 0, sizeof z); 707 z.avail_in = prelen; 708 z.next_in = pre; 709 z.avail_out = BUFLEN; 710 z.next_out = outbufp; 711 z.zalloc = NULL; 712 z.zfree = NULL; 713 z.opaque = 0; 714 715 in_tot = prelen; 716 out_tot = 0; 717 718 for (;;) { 719 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 720 size_t in_size; 721 722 if (z.avail_in > 0) { 723 memmove(inbufp, z.next_in, z.avail_in); 724 } 725 z.next_in = inbufp; 726 in_size = read(in, z.next_in + z.avail_in, 727 BUFLEN - z.avail_in); 728 729 if (in_size == -1) { 730 #ifndef SMALL 731 if (tflag && vflag) 732 print_test(filename, 0); 733 #endif 734 maybe_warn("failed to read stdin"); 735 goto stop_and_fail; 736 } else if (in_size == 0) { 737 done_reading = 1; 738 } 739 740 z.avail_in += in_size; 741 needmore = 0; 742 743 in_tot += in_size; 744 } 745 if (z.avail_in == 0) { 746 if (done_reading && state != GZSTATE_MAGIC0) 747 maybe_warnx("%s: unexpected end of file", 748 filename); 749 goto stop; 750 } 751 switch (state) { 752 case GZSTATE_MAGIC0: 753 if (*z.next_in != GZIP_MAGIC0) { 754 maybe_warnx("input not gziped (MAGIC0)"); 755 goto stop_and_fail; 756 } 757 ADVANCE(); 758 state++; 759 out_sub_tot = 0; 760 crc = crc32(0L, Z_NULL, 0); 761 break; 762 763 case GZSTATE_MAGIC1: 764 if (*z.next_in != GZIP_MAGIC1 && 765 *z.next_in != GZIP_OMAGIC1) { 766 maybe_warnx("input not gziped (MAGIC1)"); 767 goto stop_and_fail; 768 } 769 ADVANCE(); 770 state++; 771 break; 772 773 case GZSTATE_METHOD: 774 if (*z.next_in != Z_DEFLATED) { 775 maybe_warnx("unknown compression method"); 776 goto stop_and_fail; 777 } 778 ADVANCE(); 779 state++; 780 break; 781 782 case GZSTATE_FLAGS: 783 flags = *z.next_in; 784 ADVANCE(); 785 skip_count = 6; 786 state++; 787 break; 788 789 case GZSTATE_SKIPPING: 790 if (skip_count > 0) { 791 skip_count--; 792 ADVANCE(); 793 } else 794 state++; 795 break; 796 797 case GZSTATE_EXTRA: 798 if ((flags & EXTRA_FIELD) == 0) { 799 state = GZSTATE_ORIGNAME; 800 break; 801 } 802 skip_count = *z.next_in; 803 ADVANCE(); 804 state++; 805 break; 806 807 case GZSTATE_EXTRA2: 808 skip_count |= ((*z.next_in) << 8); 809 ADVANCE(); 810 state++; 811 break; 812 813 case GZSTATE_EXTRA3: 814 if (skip_count > 0) { 815 skip_count--; 816 ADVANCE(); 817 } else 818 state++; 819 break; 820 821 case GZSTATE_ORIGNAME: 822 if ((flags & ORIG_NAME) == 0) { 823 state++; 824 break; 825 } 826 if (*z.next_in == 0) 827 state++; 828 ADVANCE(); 829 break; 830 831 case GZSTATE_COMMENT: 832 if ((flags & COMMENT) == 0) { 833 state++; 834 break; 835 } 836 if (*z.next_in == 0) 837 state++; 838 ADVANCE(); 839 break; 840 841 case GZSTATE_HEAD_CRC1: 842 if (flags & HEAD_CRC) 843 skip_count = 2; 844 else 845 skip_count = 0; 846 state++; 847 break; 848 849 case GZSTATE_HEAD_CRC2: 850 if (skip_count > 0) { 851 skip_count--; 852 ADVANCE(); 853 } else 854 state++; 855 break; 856 857 case GZSTATE_INIT: 858 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 859 maybe_warnx("failed to inflateInit"); 860 goto stop_and_fail; 861 } 862 state++; 863 break; 864 865 case GZSTATE_READ: 866 error = inflate(&z, Z_FINISH); 867 switch (error) { 868 /* Z_BUF_ERROR goes with Z_FINISH... */ 869 case Z_BUF_ERROR: 870 case Z_STREAM_END: 871 case Z_OK: 872 break; 873 874 case Z_NEED_DICT: 875 maybe_warnx("Z_NEED_DICT error"); 876 goto stop_and_fail; 877 case Z_DATA_ERROR: 878 maybe_warnx("data stream error"); 879 goto stop_and_fail; 880 case Z_STREAM_ERROR: 881 maybe_warnx("internal stream error"); 882 goto stop_and_fail; 883 case Z_MEM_ERROR: 884 maybe_warnx("memory allocation error"); 885 goto stop_and_fail; 886 887 default: 888 maybe_warn("unknown error from inflate(): %d", 889 error); 890 } 891 wr = BUFLEN - z.avail_out; 892 893 if (wr != 0) { 894 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 895 if ( 896 #ifndef SMALL 897 /* don't write anything with -t */ 898 tflag == 0 && 899 #endif 900 write(out, outbufp, wr) != wr) { 901 maybe_warn("error writing to output"); 902 goto stop_and_fail; 903 } 904 905 out_tot += wr; 906 out_sub_tot += wr; 907 } 908 909 if (error == Z_STREAM_END) { 910 inflateEnd(&z); 911 state++; 912 } 913 914 z.next_out = outbufp; 915 z.avail_out = BUFLEN; 916 917 break; 918 case GZSTATE_CRC: 919 { 920 uLong origcrc; 921 922 if (z.avail_in < 4) { 923 if (!done_reading) { 924 needmore = 1; 925 continue; 926 } 927 maybe_warnx("truncated input"); 928 goto stop_and_fail; 929 } 930 origcrc = ((unsigned)z.next_in[0] & 0xff) | 931 ((unsigned)z.next_in[1] & 0xff) << 8 | 932 ((unsigned)z.next_in[2] & 0xff) << 16 | 933 ((unsigned)z.next_in[3] & 0xff) << 24; 934 if (origcrc != crc) { 935 maybe_warnx("invalid compressed" 936 " data--crc error"); 937 goto stop_and_fail; 938 } 939 } 940 941 z.avail_in -= 4; 942 z.next_in += 4; 943 944 if (!z.avail_in && done_reading) { 945 goto stop; 946 } 947 state++; 948 break; 949 case GZSTATE_LEN: 950 { 951 uLong origlen; 952 953 if (z.avail_in < 4) { 954 if (!done_reading) { 955 needmore = 1; 956 continue; 957 } 958 maybe_warnx("truncated input"); 959 goto stop_and_fail; 960 } 961 origlen = ((unsigned)z.next_in[0] & 0xff) | 962 ((unsigned)z.next_in[1] & 0xff) << 8 | 963 ((unsigned)z.next_in[2] & 0xff) << 16 | 964 ((unsigned)z.next_in[3] & 0xff) << 24; 965 966 if (origlen != out_sub_tot) { 967 maybe_warnx("invalid compressed" 968 " data--length error"); 969 goto stop_and_fail; 970 } 971 } 972 973 z.avail_in -= 4; 974 z.next_in += 4; 975 976 if (error < 0) { 977 maybe_warnx("decompression error"); 978 goto stop_and_fail; 979 } 980 state = GZSTATE_MAGIC0; 981 break; 982 } 983 continue; 984 stop_and_fail: 985 out_tot = 1; 986 stop: 987 break; 988 } 989 if (state > GZSTATE_INIT) 990 inflateEnd(&z); 991 992 #ifndef SMALL 993 if (tflag && vflag) 994 print_test(filename, out_tot != -1); 995 #endif 996 997 free(inbufp); 998 out1: 999 free(outbufp); 1000 out2: 1001 if (gsizep) 1002 *gsizep = in_tot; 1003 return (out_tot); 1004 } 1005 1006 #ifndef SMALL 1007 /* 1008 * set the owner, mode, flags & utimes for a file 1009 */ 1010 static void 1011 copymodes(const char *file, struct stat *sbp) 1012 { 1013 struct timeval times[2]; 1014 1015 /* 1016 * If we have no info on the input, give this file some 1017 * default values and return.. 1018 */ 1019 if (sbp == NULL) { 1020 mode_t mask = umask(022); 1021 1022 (void)chmod(file, DEFFILEMODE & ~mask); 1023 (void)umask(mask); 1024 return; 1025 } 1026 1027 /* if the chown fails, remove set-id bits as-per compress(1) */ 1028 if (chown(file, sbp->st_uid, sbp->st_gid) < 0) { 1029 if (errno != EPERM) 1030 maybe_warn("couldn't chown: %s", file); 1031 sbp->st_mode &= ~(S_ISUID|S_ISGID); 1032 } 1033 1034 /* we only allow set-id and the 9 normal permission bits */ 1035 sbp->st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1036 if (chmod(file, sbp->st_mode) < 0) 1037 maybe_warn("couldn't chmod: %s", file); 1038 1039 /* only try flags if they exist already */ 1040 if (sbp->st_flags != 0 && chflags(file, sbp->st_flags) < 0) 1041 maybe_warn("couldn't chflags: %s", file); 1042 1043 TIMESPEC_TO_TIMEVAL(×[0], &sbp->st_atimespec); 1044 TIMESPEC_TO_TIMEVAL(×[1], &sbp->st_mtimespec); 1045 if (utimes(file, times) < 0) 1046 maybe_warn("couldn't utimes: %s", file); 1047 } 1048 #endif 1049 1050 /* what sort of file is this? */ 1051 static enum filetype 1052 file_gettype(u_char *buf) 1053 { 1054 1055 if (buf[0] == GZIP_MAGIC0 && 1056 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1057 return FT_GZIP; 1058 else 1059 #ifndef NO_BZIP2_SUPPORT 1060 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1061 buf[3] >= '0' && buf[3] <= '9') 1062 return FT_BZIP2; 1063 else 1064 #endif 1065 #ifndef NO_COMPRESS_SUPPORT 1066 if (memcmp(buf, Z_MAGIC, 2) == 0) 1067 return FT_Z; 1068 else 1069 #endif 1070 return FT_UNKNOWN; 1071 } 1072 1073 #ifndef SMALL 1074 /* check the outfile is OK. */ 1075 static int 1076 check_outfile(const char *outfile, struct stat *sb) 1077 { 1078 int ok = 1; 1079 1080 if (lflag == 0 && stat(outfile, sb) == 0) { 1081 if (fflag) 1082 unlink(outfile); 1083 else if (isatty(STDIN_FILENO)) { 1084 char ans[10] = { 'n', '\0' }; /* default */ 1085 1086 fprintf(stderr, "%s already exists -- do you wish to " 1087 "overwrite (y or n)? " , outfile); 1088 (void)fgets(ans, sizeof(ans) - 1, stdin); 1089 if (ans[0] != 'y' && ans[0] != 'Y') { 1090 fprintf(stderr, "\tnot overwritting\n"); 1091 ok = 0; 1092 } else 1093 unlink(outfile); 1094 } else { 1095 maybe_warnx("%s already exists -- skipping", outfile); 1096 ok = 0; 1097 } 1098 } 1099 return ok; 1100 } 1101 1102 static void 1103 unlink_input(const char *file, struct stat *sb) 1104 { 1105 struct stat nsb; 1106 1107 if (stat(file, &nsb) != 0) 1108 /* Must be gone alrady */ 1109 return; 1110 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1111 /* Definitely a different file */ 1112 return; 1113 unlink(file); 1114 } 1115 #endif 1116 1117 static const suffixes_t * 1118 check_suffix(char *file, int xlate) 1119 { 1120 const suffixes_t *s; 1121 int len = strlen(file); 1122 char *sp; 1123 1124 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1125 /* if it doesn't fit in "a.suf", don't bother */ 1126 if (s->ziplen >= len) 1127 continue; 1128 sp = file + len - s->ziplen; 1129 if (strcmp(s->zipped, sp) != 0) 1130 continue; 1131 if (xlate) 1132 strcpy(sp, s->normal); 1133 return s; 1134 } 1135 return NULL; 1136 } 1137 1138 /* 1139 * compress the given file: create a corresponding .gz file and remove the 1140 * original. 1141 */ 1142 static off_t 1143 file_compress(char *file, char *outfile, size_t outsize) 1144 { 1145 int in; 1146 int out; 1147 off_t size, insize; 1148 #ifndef SMALL 1149 struct stat isb, osb; 1150 const suffixes_t *suff; 1151 #endif 1152 1153 in = open(file, O_RDONLY); 1154 if (in == -1) { 1155 maybe_warn("can't open %s", file); 1156 return -1; 1157 } 1158 1159 if (cflag == 0) { 1160 #ifndef SMALL 1161 if (stat(file, &isb) == 0) { 1162 if (isb.st_nlink > 1 && fflag == 0) { 1163 maybe_warnx("%s has %d other link%s -- " 1164 "skipping", file, isb.st_nlink - 1, 1165 isb.st_nlink == 1 ? "" : "s"); 1166 close(in); 1167 return -1; 1168 } 1169 } 1170 1171 if (fflag == 0 && (suff = check_suffix(file, 0)) 1172 && suff->zipped[0] != 0) { 1173 maybe_warnx("%s already has %s suffix -- unchanged", 1174 file, suff->zipped); 1175 close(in); 1176 return -1; 1177 } 1178 #endif 1179 1180 /* Add (usually) .gz to filename */ 1181 if (snprintf(outfile, outsize, "%s%s", 1182 file, suffixes[0].zipped) >= outsize) 1183 memcpy(outfile - suffixes[0].ziplen - 1, 1184 suffixes[0].zipped, suffixes[0].ziplen + 1); 1185 1186 #ifndef SMALL 1187 if (check_outfile(outfile, &osb) == 0) { 1188 close(in); 1189 return -1; 1190 } 1191 #endif 1192 } 1193 1194 if (cflag == 0) { 1195 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1196 if (out == -1) { 1197 maybe_warn("could not create output: %s", outfile); 1198 fclose(stdin); 1199 return -1; 1200 } 1201 } else 1202 out = STDOUT_FILENO; 1203 1204 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1205 1206 (void)close(in); 1207 1208 /* 1209 * If there was an error, insize will be -1. 1210 * If we compressed to stdout, just return the size. 1211 * Otherwise stat the file and check it is the correct size. 1212 * We only blow away the file if we can stat the output and it 1213 * has the expected size. 1214 */ 1215 if (cflag != 0) 1216 return insize == -1 ? -1 : size; 1217 1218 if (close(out) == -1) 1219 maybe_warn("couldn't close ouput"); 1220 1221 #ifndef SMALL 1222 if (stat(outfile, &osb) != 0) { 1223 maybe_warn("couldn't stat: %s", outfile); 1224 goto bad_outfile; 1225 } 1226 1227 if (osb.st_size != size) { 1228 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1229 " != %" PRIdOFF "), deleting", 1230 outfile, osb.st_size, size); 1231 goto bad_outfile; 1232 } 1233 1234 copymodes(outfile, &isb); 1235 #endif 1236 1237 /* output is good, ok to delete input */ 1238 unlink_input(file, &isb); 1239 return size; 1240 1241 #ifndef SMALL 1242 bad_outfile: 1243 maybe_warnx("leaving original %s", file); 1244 unlink(outfile); 1245 return size; 1246 #endif 1247 } 1248 1249 /* uncompress the given file and remove the original */ 1250 static off_t 1251 file_uncompress(char *file, char *outfile, size_t outsize) 1252 { 1253 struct stat isb, osb; 1254 off_t size; 1255 ssize_t rbytes; 1256 unsigned char header1[4]; 1257 enum filetype method; 1258 int fd, zfd = -1; 1259 #ifndef SMALL 1260 time_t timestamp = 0; 1261 unsigned char name[PATH_MAX + 1]; 1262 #endif 1263 1264 /* gather the old name info */ 1265 1266 fd = open(file, O_RDONLY); 1267 if (fd < 0) { 1268 maybe_warn("can't open %s", file); 1269 goto lose; 1270 } 1271 1272 strlcpy(outfile, file, outsize); 1273 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1274 maybe_warnx("%s: unknown suffix -- ignored", file); 1275 goto lose; 1276 } 1277 1278 rbytes = read(fd, header1, sizeof header1); 1279 if (rbytes != sizeof header1) { 1280 /* we don't want to fail here. */ 1281 #ifndef SMALL 1282 if (fflag) 1283 goto lose; 1284 #endif 1285 if (rbytes == -1) 1286 maybe_warn("can't read %s", file); 1287 else 1288 maybe_warnx("%s: unexpected end of file", file); 1289 goto lose; 1290 } 1291 1292 method = file_gettype(header1); 1293 1294 #ifndef SMALL 1295 if (fflag == 0 && method == FT_UNKNOWN) { 1296 maybe_warnx("%s: not in gzip format", file); 1297 goto lose; 1298 } 1299 1300 #endif 1301 1302 #ifndef SMALL 1303 if (method == FT_GZIP && Nflag) { 1304 unsigned char ts[4]; /* timestamp */ 1305 1306 if (pread(fd, ts, sizeof ts, GZIP_TIMESTAMP) != sizeof ts) { 1307 if (!fflag) 1308 maybe_warn("can't read %s", file); 1309 goto lose; 1310 } 1311 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1312 1313 if (header1[3] & ORIG_NAME) { 1314 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1315 if (rbytes < 0) { 1316 maybe_warn("can't read %s", file); 1317 goto lose; 1318 } 1319 if (name[0] != 0) { 1320 /* preserve original directory name */ 1321 char *dp = strrchr(file, '/'); 1322 if (dp == NULL) 1323 dp = file; 1324 else 1325 dp++; 1326 snprintf(outfile, outsize, "%.*s%.*s", 1327 (int) (dp - file), 1328 file, (int) rbytes, name); 1329 } 1330 } 1331 } 1332 #endif 1333 lseek(fd, 0, SEEK_SET); 1334 1335 if (cflag == 0 || lflag) { 1336 if (fstat(fd, &isb) != 0) 1337 goto lose; 1338 #ifndef SMALL 1339 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1340 maybe_warnx("%s has %d other links -- skipping", 1341 file, isb.st_nlink - 1); 1342 goto lose; 1343 } 1344 if (nflag == 0 && timestamp) 1345 isb.st_mtime = timestamp; 1346 if (check_outfile(outfile, &osb) == 0) 1347 goto lose; 1348 #endif 1349 } 1350 1351 if (cflag == 0 && lflag == 0) { 1352 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1353 if (zfd == STDOUT_FILENO) { 1354 /* We won't close STDOUT_FILENO later... */ 1355 zfd = dup(zfd); 1356 close(STDOUT_FILENO); 1357 } 1358 if (zfd == -1) { 1359 maybe_warn("can't open %s", outfile); 1360 goto lose; 1361 } 1362 } else 1363 zfd = STDOUT_FILENO; 1364 1365 #ifndef NO_BZIP2_SUPPORT 1366 if (method == FT_BZIP2) { 1367 1368 /* XXX */ 1369 if (lflag) { 1370 maybe_warnx("no -l with bzip2 files"); 1371 goto lose; 1372 } 1373 1374 size = unbzip2(fd, zfd, NULL, 0, NULL); 1375 } else 1376 #endif 1377 1378 #ifndef NO_COMPRESS_SUPPORT 1379 if (method == FT_Z) { 1380 FILE *in, *out; 1381 1382 /* XXX */ 1383 if (lflag) { 1384 maybe_warnx("no -l with Lempel-Ziv files"); 1385 goto lose; 1386 } 1387 1388 if ((in = zdopen(fd)) == NULL) { 1389 maybe_warn("zdopen for read: %s", file); 1390 goto lose; 1391 } 1392 1393 out = fdopen(dup(zfd), "w"); 1394 if (out == NULL) { 1395 maybe_warn("fdopen for write: %s", outfile); 1396 fclose(in); 1397 goto lose; 1398 } 1399 1400 size = zuncompress(in, out, NULL, 0, NULL); 1401 /* need to fclose() if ferror() is true... */ 1402 if (ferror(in) | fclose(in)) { 1403 maybe_warn("failed infile fclose"); 1404 unlink(outfile); 1405 (void)fclose(out); 1406 } 1407 if (fclose(out) != 0) { 1408 maybe_warn("failed outfile fclose"); 1409 unlink(outfile); 1410 goto lose; 1411 } 1412 } else 1413 #endif 1414 1415 #ifndef SMALL 1416 if (method == FT_UNKNOWN) { 1417 if (lflag) { 1418 maybe_warnx("no -l for unknown filetypes"); 1419 goto lose; 1420 } 1421 size = cat_fd(NULL, 0, NULL, fd); 1422 } else 1423 #endif 1424 { 1425 if (lflag) { 1426 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1427 close(fd); 1428 return -1; /* XXX */ 1429 } 1430 1431 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1432 } 1433 1434 if (close(fd) != 0) 1435 maybe_warn("couldn't close input"); 1436 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1437 maybe_warn("couldn't close output"); 1438 1439 if (size == -1) { 1440 if (cflag == 0) 1441 unlink(outfile); 1442 maybe_warnx("%s: uncompress failed", file); 1443 return -1; 1444 } 1445 1446 /* if testing, or we uncompressed to stdout, this is all we need */ 1447 #ifndef SMALL 1448 if (tflag) 1449 return size; 1450 #endif 1451 /* if we are uncompressing to stdin, don't remove the file. */ 1452 if (cflag) 1453 return size; 1454 1455 /* 1456 * if we create a file... 1457 */ 1458 /* 1459 * if we can't stat the file don't remove the file. 1460 */ 1461 if (stat(outfile, &osb) != 0) { 1462 maybe_warn("couldn't stat (leaving original): %s", 1463 outfile); 1464 return -1; 1465 } 1466 if (osb.st_size != size) { 1467 maybe_warnx("stat gave different size: %" PRIdOFF 1468 " != %" PRIdOFF " (leaving original)", 1469 size, osb.st_size); 1470 unlink(outfile); 1471 return -1; 1472 } 1473 unlink_input(file, &isb); 1474 #ifndef SMALL 1475 copymodes(outfile, &isb); 1476 #endif 1477 return size; 1478 1479 lose: 1480 if (fd != -1) 1481 close(fd); 1482 if (zfd != -1 && zfd != STDOUT_FILENO) 1483 close(fd); 1484 return -1; 1485 } 1486 1487 #ifndef SMALL 1488 static off_t 1489 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1490 { 1491 char buf[BUFLEN]; 1492 size_t rv; 1493 off_t in_tot; 1494 1495 in_tot = count; 1496 if (write(STDOUT_FILENO, prepend, count) != count) { 1497 maybe_warn("write to stdout"); 1498 return -1; 1499 } 1500 for (;;) { 1501 rv = read(fd, buf, sizeof buf); 1502 if (rv == 0) 1503 break; 1504 if (rv < 0) { 1505 maybe_warn("read from fd %d", fd); 1506 break; 1507 } 1508 1509 if (write(STDOUT_FILENO, buf, rv) != rv) { 1510 maybe_warn("write to stdout"); 1511 break; 1512 } 1513 in_tot += rv; 1514 } 1515 1516 if (gsizep) 1517 *gsizep = in_tot; 1518 return (in_tot); 1519 } 1520 #endif 1521 1522 static void 1523 handle_stdin(void) 1524 { 1525 unsigned char header1[4]; 1526 off_t usize, gsize; 1527 enum filetype method; 1528 ssize_t bytes_read; 1529 #ifndef NO_COMPRESS_SUPPORT 1530 FILE *in; 1531 #endif 1532 1533 #ifndef SMALL 1534 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1535 maybe_warnx("standard input is a terminal -- ignoring"); 1536 return; 1537 } 1538 #endif 1539 1540 if (lflag) { 1541 struct stat isb; 1542 1543 /* XXX could read the whole file, etc. */ 1544 if (fstat(STDIN_FILENO, &isb) < 0) { 1545 maybe_warn("fstat"); 1546 return; 1547 } 1548 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1549 return; 1550 } 1551 1552 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1553 if (bytes_read == -1) { 1554 maybe_warn("can't read stdin"); 1555 return; 1556 } else if (bytes_read != sizeof(header1)) { 1557 maybe_warnx("unexpected EOF"); 1558 return; 1559 } 1560 1561 method = file_gettype(header1); 1562 switch (method) { 1563 default: 1564 #ifndef SMALL 1565 if (fflag == 0) { 1566 maybe_warnx("unknown compression format"); 1567 return; 1568 } 1569 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1570 break; 1571 #endif 1572 case FT_GZIP: 1573 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1574 header1, sizeof header1, &gsize, "(stdin)"); 1575 break; 1576 #ifndef NO_BZIP2_SUPPORT 1577 case FT_BZIP2: 1578 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1579 header1, sizeof header1, &gsize); 1580 break; 1581 #endif 1582 #ifndef NO_COMPRESS_SUPPORT 1583 case FT_Z: 1584 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1585 maybe_warnx("zopen of stdin"); 1586 return; 1587 } 1588 1589 usize = zuncompress(in, stdout, header1, sizeof header1, &gsize); 1590 fclose(in); 1591 break; 1592 #endif 1593 } 1594 1595 #ifndef SMALL 1596 if (vflag && !tflag && usize != -1 && gsize != -1) 1597 print_verbage(NULL, NULL, usize, gsize); 1598 #endif 1599 1600 } 1601 1602 static void 1603 handle_stdout(void) 1604 { 1605 off_t gsize, usize; 1606 struct stat sb; 1607 time_t systime; 1608 uint32_t mtime; 1609 int ret; 1610 1611 #ifndef SMALL 1612 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1613 maybe_warnx("standard output is a terminal -- ignoring"); 1614 return; 1615 } 1616 #endif 1617 /* If stdin is a file use it's mtime, otherwise use current time */ 1618 ret = fstat(STDIN_FILENO, &sb); 1619 1620 #ifndef SMALL 1621 if (ret < 0) { 1622 maybe_warn("Can't stat stdin"); 1623 return; 1624 } 1625 #endif 1626 1627 if (S_ISREG(sb.st_mode)) 1628 mtime = (uint32_t)sb.st_mtime; 1629 else { 1630 systime = time(NULL); 1631 #ifndef SMALL 1632 if (systime == -1) { 1633 maybe_warn("time"); 1634 return; 1635 } 1636 #endif 1637 mtime = (uint32_t)systime; 1638 } 1639 1640 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1641 #ifndef SMALL 1642 if (vflag && !tflag && usize != -1 && gsize != -1) 1643 print_verbage(NULL, NULL, usize, gsize); 1644 #endif 1645 } 1646 1647 /* do what is asked for, for the path name */ 1648 static void 1649 handle_pathname(char *path) 1650 { 1651 char *opath = path, *s = NULL; 1652 ssize_t len; 1653 int slen; 1654 struct stat sb; 1655 1656 /* check for stdout/stdin */ 1657 if (path[0] == '-' && path[1] == '\0') { 1658 if (dflag) 1659 handle_stdin(); 1660 else 1661 handle_stdout(); 1662 return; 1663 } 1664 1665 retry: 1666 if (stat(path, &sb) != 0) { 1667 /* lets try <path>.gz if we're decompressing */ 1668 if (dflag && s == NULL && errno == ENOENT) { 1669 len = strlen(path); 1670 slen = suffixes[0].ziplen; 1671 s = malloc(len + slen + 1); 1672 if (s == NULL) 1673 maybe_err("malloc"); 1674 memcpy(s, path, len); 1675 memcpy(s + len, suffixes[0].zipped, slen + 1); 1676 path = s; 1677 goto retry; 1678 } 1679 maybe_warn("can't stat: %s", opath); 1680 goto out; 1681 } 1682 1683 if (S_ISDIR(sb.st_mode)) { 1684 #ifndef SMALL 1685 if (rflag) 1686 handle_dir(path, &sb); 1687 else 1688 #endif 1689 maybe_warnx("%s is a directory", path); 1690 goto out; 1691 } 1692 1693 if (S_ISREG(sb.st_mode)) 1694 handle_file(path, &sb); 1695 else 1696 maybe_warnx("%s is not a regular file", path); 1697 1698 out: 1699 if (s) 1700 free(s); 1701 } 1702 1703 /* compress/decompress a file */ 1704 static void 1705 handle_file(char *file, struct stat *sbp) 1706 { 1707 off_t usize, gsize; 1708 char outfile[PATH_MAX]; 1709 1710 infile = file; 1711 if (dflag) { 1712 usize = file_uncompress(file, outfile, sizeof(outfile)); 1713 if (usize == -1) 1714 return; 1715 gsize = sbp->st_size; 1716 } else { 1717 gsize = file_compress(file, outfile, sizeof(outfile)); 1718 if (gsize == -1) 1719 return; 1720 usize = sbp->st_size; 1721 } 1722 1723 1724 #ifndef SMALL 1725 if (vflag && !tflag) 1726 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1727 #endif 1728 } 1729 1730 #ifndef SMALL 1731 /* this is used with -r to recursively descend directories */ 1732 static void 1733 handle_dir(char *dir, struct stat *sbp) 1734 { 1735 char *path_argv[2]; 1736 FTS *fts; 1737 FTSENT *entry; 1738 1739 path_argv[0] = dir; 1740 path_argv[1] = 0; 1741 fts = fts_open(path_argv, FTS_PHYSICAL, NULL); 1742 if (fts == NULL) { 1743 warn("couldn't fts_open %s", dir); 1744 return; 1745 } 1746 1747 while ((entry = fts_read(fts))) { 1748 switch(entry->fts_info) { 1749 case FTS_D: 1750 case FTS_DP: 1751 continue; 1752 1753 case FTS_DNR: 1754 case FTS_ERR: 1755 case FTS_NS: 1756 maybe_warn("%s", entry->fts_path); 1757 continue; 1758 case FTS_F: 1759 handle_file(entry->fts_name, entry->fts_statp); 1760 } 1761 } 1762 (void)fts_close(fts); 1763 } 1764 #endif 1765 1766 /* print a ratio - size reduction as a fraction of uncompressed size */ 1767 static void 1768 print_ratio(off_t in, off_t out, FILE *where) 1769 { 1770 int percent10; /* 10 * percent */ 1771 off_t diff; 1772 char buff[8]; 1773 int len; 1774 1775 diff = in - out/2; 1776 if (diff <= 0) 1777 /* 1778 * Output is more than double size of input! print -99.9% 1779 * Quite possibly we've failed to get the original size. 1780 */ 1781 percent10 = -999; 1782 else { 1783 /* 1784 * We only need 12 bits of result from the final division, 1785 * so reduce the values until a 32bit division will suffice. 1786 */ 1787 while (in > 0x100000) { 1788 diff >>= 1; 1789 in >>= 1; 1790 } 1791 if (in != 0) 1792 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1793 else 1794 percent10 = 0; 1795 } 1796 1797 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1798 /* Move the '.' to before the last digit */ 1799 buff[len - 1] = buff[len - 2]; 1800 buff[len - 2] = '.'; 1801 fprintf(where, "%5s%%", buff); 1802 } 1803 1804 #ifndef SMALL 1805 /* print compression statistics, and the new name (if there is one!) */ 1806 static void 1807 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1808 { 1809 if (file) 1810 fprintf(stderr, "%s:%s ", file, 1811 strlen(file) < 7 ? "\t\t" : "\t"); 1812 print_ratio(usize, gsize, stderr); 1813 if (nfile) 1814 fprintf(stderr, " -- replaced with %s", nfile); 1815 fprintf(stderr, "\n"); 1816 fflush(stderr); 1817 } 1818 1819 /* print test results */ 1820 static void 1821 print_test(const char *file, int ok) 1822 { 1823 1824 if (exit_value == 0 && ok == 0) 1825 exit_value = 1; 1826 fprintf(stderr, "%s:%s %s\n", file, 1827 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1828 fflush(stderr); 1829 } 1830 #endif 1831 1832 /* print a file's info ala --list */ 1833 /* eg: 1834 compressed uncompressed ratio uncompressed_name 1835 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1836 */ 1837 static void 1838 print_list(int fd, off_t out, const char *outfile, time_t ts) 1839 { 1840 static int first = 1; 1841 #ifndef SMALL 1842 static off_t in_tot, out_tot; 1843 uint32_t crc = 0; 1844 #endif 1845 off_t in = 0; 1846 int rv; 1847 1848 if (first) { 1849 #ifndef SMALL 1850 if (vflag) 1851 printf("method crc date time "); 1852 #endif 1853 if (qflag == 0) 1854 printf(" compressed uncompressed " 1855 "ratio uncompressed_name\n"); 1856 } 1857 first = 0; 1858 1859 /* print totals? */ 1860 #ifndef SMALL 1861 if (fd == -1) { 1862 in = in_tot; 1863 out = out_tot; 1864 } else 1865 #endif 1866 { 1867 /* read the last 4 bytes - this is the uncompressed size */ 1868 rv = lseek(fd, (off_t)(-8), SEEK_END); 1869 if (rv != -1) { 1870 unsigned char buf[8]; 1871 uint32_t usize; 1872 1873 if (read(fd, (char *)buf, sizeof(buf)) != sizeof(buf)) 1874 maybe_warn("read of uncompressed size"); 1875 usize = buf[4] | buf[5] << 8 | buf[6] << 16 | buf[7] << 24; 1876 in = (off_t)usize; 1877 #ifndef SMALL 1878 crc = buf[0] | buf[1] << 8 | buf[2] << 16 | buf[3] << 24; 1879 #endif 1880 } 1881 } 1882 1883 #ifndef SMALL 1884 if (vflag && fd == -1) 1885 printf(" "); 1886 else if (vflag) { 1887 char *date = ctime(&ts); 1888 1889 /* skip the day, 1/100th second, and year */ 1890 date += 4; 1891 date[12] = 0; 1892 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1893 } 1894 in_tot += in; 1895 out_tot += out; 1896 #endif 1897 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1898 print_ratio(in, out, stdout); 1899 printf(" %s\n", outfile); 1900 } 1901 1902 /* display the usage of NetBSD gzip */ 1903 static void 1904 usage(void) 1905 { 1906 1907 fprintf(stderr, "%s\n", gzip_version); 1908 fprintf(stderr, 1909 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n" 1910 #ifndef SMALL 1911 " -c --stdout write to stdout, keep original files\n" 1912 " --to-stdout\n" 1913 " -d --decompress uncompress files\n" 1914 " --uncompress\n" 1915 " -f --force force overwriting & compress links\n" 1916 " -h --help display this help\n" 1917 " -n --no-name don't save original file name or time stamp\n" 1918 " -N --name save or restore original file name and time stamp\n" 1919 " -q --quiet output no warnings\n" 1920 " -r --recursive recursively compress files in directories\n" 1921 " -S .suf use suffix .suf instead of .gz\n" 1922 " --suffix .suf\n" 1923 " -t --test test compressed file\n" 1924 " -v --verbose print extra statistics\n" 1925 " -V --version display program version\n" 1926 " -1 --fast fastest (worst) compression\n" 1927 " -2 .. -8 set compression level\n" 1928 " -9 --best best (slowest) compression\n", 1929 #else 1930 , 1931 #endif 1932 getprogname()); 1933 exit(0); 1934 } 1935 1936 /* display the version of NetBSD gzip */ 1937 static void 1938 display_version(void) 1939 { 1940 1941 fprintf(stderr, "%s\n", gzip_version); 1942 exit(0); 1943 } 1944 1945 #ifndef NO_BZIP2_SUPPORT 1946 #include "unbzip2.c" 1947 #endif 1948 #ifndef NO_COMPRESS_SUPPORT 1949 #include "zuncompress.c" 1950 #endif 1951 1952 static ssize_t 1953 read_retry(int fd, void *buf, size_t sz) 1954 { 1955 char *cp = buf; 1956 ssize_t left = sz; 1957 1958 while (left > 0) { 1959 ssize_t ret; 1960 1961 ret = read(fd, cp, sz); 1962 if (ret == -1) { 1963 return ret; 1964 } else if (ret == 0) { 1965 break; /* EOF */ 1966 } 1967 cp += ret; 1968 left -= ret; 1969 } 1970 1971 return sz - left; 1972 } 1973