gzip.c revision 1.81
1 /* $NetBSD: gzip.c,v 1.81 2005/12/13 10:02:04 wiz 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 1.81 2005/12/13 10:02:04 wiz 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 *); 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(int fd, const struct stat *, const char *file); 207 static int check_outfile(const char *outfile); 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 "123456789cdhltV" 283 #else 284 #define OPT_LIST "123456789cdfhlNnqrS:tVv" 285 #endif 286 287 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 288 switch (ch) { 289 case '1': case '2': case '3': 290 case '4': case '5': case '6': 291 case '7': case '8': case '9': 292 numflag = ch - '0'; 293 break; 294 case 'c': 295 cflag = 1; 296 break; 297 case 'd': 298 dflag = 1; 299 break; 300 case 'l': 301 lflag = 1; 302 dflag = 1; 303 break; 304 case 'V': 305 display_version(); 306 /* NOTREACHED */ 307 #ifndef SMALL 308 case 'f': 309 fflag = 1; 310 break; 311 case 'N': 312 nflag = 0; 313 Nflag = 1; 314 break; 315 case 'n': 316 nflag = 1; 317 Nflag = 0; 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 ssize_t w; 601 602 error = deflate(&z, Z_FINISH); 603 if (error != Z_OK && error != Z_STREAM_END) { 604 maybe_warnx("deflate failed"); 605 in_tot = -1; 606 goto out; 607 } 608 609 len = (char *)z.next_out - outbufp; 610 611 w = write(out, outbufp, len); 612 if (w == -1 || (size_t)w != len) { 613 maybe_warn("write"); 614 out_tot = -1; 615 goto out; 616 } 617 out_tot += len; 618 z.next_out = outbufp; 619 z.avail_out = BUFLEN; 620 621 if (error == Z_STREAM_END) 622 break; 623 } 624 625 if (deflateEnd(&z) != Z_OK) { 626 maybe_warnx("deflateEnd failed"); 627 in_tot = -1; 628 goto out; 629 } 630 631 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 632 (int)crc & 0xff, 633 (int)(crc >> 8) & 0xff, 634 (int)(crc >> 16) & 0xff, 635 (int)(crc >> 24) & 0xff, 636 (int)in_tot & 0xff, 637 (int)(in_tot >> 8) & 0xff, 638 (int)(in_tot >> 16) & 0xff, 639 (int)(in_tot >> 24) & 0xff); 640 if (i != 8) 641 maybe_err("snprintf"); 642 #if 0 643 if (in_tot > 0xffffffff) 644 maybe_warn("input file size >= 4GB cannot be saved"); 645 #endif 646 if (write(out, outbufp, i) != i) { 647 maybe_warn("write"); 648 in_tot = -1; 649 } else 650 out_tot += i; 651 652 out: 653 if (inbufp != NULL) 654 free(inbufp); 655 if (outbufp != NULL) 656 free(outbufp); 657 if (gsizep) 658 *gsizep = out_tot; 659 return in_tot; 660 } 661 662 /* 663 * uncompress input to output then close the input. return the 664 * uncompressed size written, and put the compressed sized read 665 * into `*gsizep'. 666 */ 667 static off_t 668 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 669 const char *filename) 670 { 671 z_stream z; 672 char *outbufp, *inbufp; 673 off_t out_tot = -1, in_tot = 0; 674 uint32_t out_sub_tot = 0; 675 enum { 676 GZSTATE_MAGIC0, 677 GZSTATE_MAGIC1, 678 GZSTATE_METHOD, 679 GZSTATE_FLAGS, 680 GZSTATE_SKIPPING, 681 GZSTATE_EXTRA, 682 GZSTATE_EXTRA2, 683 GZSTATE_EXTRA3, 684 GZSTATE_ORIGNAME, 685 GZSTATE_COMMENT, 686 GZSTATE_HEAD_CRC1, 687 GZSTATE_HEAD_CRC2, 688 GZSTATE_INIT, 689 GZSTATE_READ, 690 GZSTATE_CRC, 691 GZSTATE_LEN, 692 } state = GZSTATE_MAGIC0; 693 int flags = 0, skip_count = 0; 694 int error = Z_STREAM_ERROR, done_reading = 0; 695 uLong crc = 0; 696 ssize_t wr; 697 int needmore = 0; 698 699 #define ADVANCE() { z.next_in++; z.avail_in--; } 700 701 if ((outbufp = malloc(BUFLEN)) == NULL) { 702 maybe_err("malloc failed"); 703 goto out2; 704 } 705 if ((inbufp = malloc(BUFLEN)) == NULL) { 706 maybe_err("malloc failed"); 707 goto out1; 708 } 709 710 memset(&z, 0, sizeof z); 711 z.avail_in = prelen; 712 z.next_in = pre; 713 z.avail_out = BUFLEN; 714 z.next_out = outbufp; 715 z.zalloc = NULL; 716 z.zfree = NULL; 717 z.opaque = 0; 718 719 in_tot = prelen; 720 out_tot = 0; 721 722 for (;;) { 723 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 724 ssize_t in_size; 725 726 if (z.avail_in > 0) { 727 memmove(inbufp, z.next_in, z.avail_in); 728 } 729 z.next_in = inbufp; 730 in_size = read(in, z.next_in + z.avail_in, 731 BUFLEN - z.avail_in); 732 733 if (in_size == -1) { 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 free(inbufp); 993 out1: 994 free(outbufp); 995 out2: 996 if (gsizep) 997 *gsizep = in_tot; 998 return (out_tot); 999 } 1000 1001 #ifndef SMALL 1002 /* 1003 * set the owner, mode, flags & utimes using the given file descriptor. 1004 * file is only used in possible warning messages. 1005 */ 1006 static void 1007 copymodes(int fd, const struct stat *sbp, const char *file) 1008 { 1009 struct timeval times[2]; 1010 struct stat sb; 1011 1012 /* 1013 * If we have no info on the input, give this file some 1014 * default values and return.. 1015 */ 1016 if (sbp == NULL) { 1017 mode_t mask = umask(022); 1018 1019 (void)fchmod(fd, DEFFILEMODE & ~mask); 1020 (void)umask(mask); 1021 return; 1022 } 1023 sb = *sbp; 1024 1025 /* if the chown fails, remove set-id bits as-per compress(1) */ 1026 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1027 if (errno != EPERM) 1028 maybe_warn("couldn't fchown: %s", file); 1029 sb.st_mode &= ~(S_ISUID|S_ISGID); 1030 } 1031 1032 /* we only allow set-id and the 9 normal permission bits */ 1033 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1034 if (fchmod(fd, sb.st_mode) < 0) 1035 maybe_warn("couldn't fchmod: %s", file); 1036 1037 /* only try flags if they exist already */ 1038 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1039 maybe_warn("couldn't fchflags: %s", file); 1040 1041 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1042 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1043 if (futimes(fd, times) < 0) 1044 maybe_warn("couldn't utimes: %s", file); 1045 } 1046 #endif 1047 1048 /* what sort of file is this? */ 1049 static enum filetype 1050 file_gettype(u_char *buf) 1051 { 1052 1053 if (buf[0] == GZIP_MAGIC0 && 1054 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1055 return FT_GZIP; 1056 else 1057 #ifndef NO_BZIP2_SUPPORT 1058 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1059 buf[3] >= '0' && buf[3] <= '9') 1060 return FT_BZIP2; 1061 else 1062 #endif 1063 #ifndef NO_COMPRESS_SUPPORT 1064 if (memcmp(buf, Z_MAGIC, 2) == 0) 1065 return FT_Z; 1066 else 1067 #endif 1068 return FT_UNKNOWN; 1069 } 1070 1071 #ifndef SMALL 1072 /* check the outfile is OK. */ 1073 static int 1074 check_outfile(const char *outfile) 1075 { 1076 struct stat sb; 1077 int ok = 1; 1078 1079 if (lflag == 0 && stat(outfile, &sb) == 0) { 1080 if (fflag) 1081 unlink(outfile); 1082 else if (isatty(STDIN_FILENO)) { 1083 char ans[10] = { 'n', '\0' }; /* default */ 1084 1085 fprintf(stderr, "%s already exists -- do you wish to " 1086 "overwrite (y or n)? " , outfile); 1087 (void)fgets(ans, sizeof(ans) - 1, stdin); 1088 if (ans[0] != 'y' && ans[0] != 'Y') { 1089 fprintf(stderr, "\tnot overwritting\n"); 1090 ok = 0; 1091 } else 1092 unlink(outfile); 1093 } else { 1094 maybe_warnx("%s already exists -- skipping", outfile); 1095 ok = 0; 1096 } 1097 } 1098 return ok; 1099 } 1100 1101 static void 1102 unlink_input(const char *file, const struct stat *sb) 1103 { 1104 struct stat nsb; 1105 1106 if (stat(file, &nsb) != 0) 1107 /* Must be gone alrady */ 1108 return; 1109 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1110 /* Definitely a different file */ 1111 return; 1112 unlink(file); 1113 } 1114 #endif 1115 1116 static const suffixes_t * 1117 check_suffix(char *file, int xlate) 1118 { 1119 const suffixes_t *s; 1120 int len = strlen(file); 1121 char *sp; 1122 1123 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1124 /* if it doesn't fit in "a.suf", don't bother */ 1125 if (s->ziplen >= len) 1126 continue; 1127 sp = file + len - s->ziplen; 1128 if (strcmp(s->zipped, sp) != 0) 1129 continue; 1130 if (xlate) 1131 strcpy(sp, s->normal); 1132 return s; 1133 } 1134 return NULL; 1135 } 1136 1137 /* 1138 * compress the given file: create a corresponding .gz file and remove the 1139 * original. 1140 */ 1141 static off_t 1142 file_compress(char *file, char *outfile, size_t outsize) 1143 { 1144 int in; 1145 int out; 1146 off_t size, insize; 1147 #ifndef SMALL 1148 struct stat isb, osb; 1149 const suffixes_t *suff; 1150 #endif 1151 1152 in = open(file, O_RDONLY); 1153 if (in == -1) { 1154 maybe_warn("can't open %s", file); 1155 return -1; 1156 } 1157 1158 if (cflag == 0) { 1159 #ifndef SMALL 1160 if (fstat(in, &isb) == 0) { 1161 if (isb.st_nlink > 1 && fflag == 0) { 1162 maybe_warnx("%s has %d other link%s -- " 1163 "skipping", file, isb.st_nlink - 1, 1164 isb.st_nlink == 1 ? "" : "s"); 1165 close(in); 1166 return -1; 1167 } 1168 } 1169 1170 if (fflag == 0 && (suff = check_suffix(file, 0)) 1171 && suff->zipped[0] != 0) { 1172 maybe_warnx("%s already has %s suffix -- unchanged", 1173 file, suff->zipped); 1174 close(in); 1175 return -1; 1176 } 1177 #endif 1178 1179 /* Add (usually) .gz to filename */ 1180 if ((size_t)snprintf(outfile, outsize, "%s%s", 1181 file, suffixes[0].zipped) >= outsize) 1182 memcpy(outfile - suffixes[0].ziplen - 1, 1183 suffixes[0].zipped, suffixes[0].ziplen + 1); 1184 1185 #ifndef SMALL 1186 if (check_outfile(outfile) == 0) { 1187 close(in); 1188 return -1; 1189 } 1190 #endif 1191 } 1192 1193 if (cflag == 0) { 1194 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1195 if (out == -1) { 1196 maybe_warn("could not create output: %s", outfile); 1197 fclose(stdin); 1198 return -1; 1199 } 1200 } else 1201 out = STDOUT_FILENO; 1202 1203 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1204 1205 (void)close(in); 1206 1207 /* 1208 * If there was an error, insize will be -1. 1209 * If we compressed to stdout, just return the size. 1210 * Otherwise stat the file and check it is the correct size. 1211 * We only blow away the file if we can stat the output and it 1212 * has the expected size. 1213 */ 1214 if (cflag != 0) 1215 return insize == -1 ? -1 : size; 1216 1217 #ifndef SMALL 1218 if (fstat(out, &osb) != 0) { 1219 maybe_warn("couldn't stat: %s", outfile); 1220 goto bad_outfile; 1221 } 1222 1223 if (osb.st_size != size) { 1224 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1225 " != %" PRIdOFF "), deleting", 1226 outfile, osb.st_size, size); 1227 goto bad_outfile; 1228 } 1229 1230 copymodes(out, &isb, outfile); 1231 #endif 1232 if (close(out) == -1) 1233 maybe_warn("couldn't close output"); 1234 1235 /* output is good, ok to delete input */ 1236 unlink_input(file, &isb); 1237 return size; 1238 1239 #ifndef SMALL 1240 bad_outfile: 1241 if (close(out) == -1) 1242 maybe_warn("couldn't close output"); 1243 1244 maybe_warnx("leaving original %s", file); 1245 unlink(outfile); 1246 return size; 1247 #endif 1248 } 1249 1250 /* uncompress the given file and remove the original */ 1251 static off_t 1252 file_uncompress(char *file, char *outfile, size_t outsize) 1253 { 1254 struct stat isb, osb; 1255 off_t size; 1256 ssize_t rbytes; 1257 unsigned char header1[4]; 1258 enum filetype method; 1259 int fd, ofd, zfd = -1; 1260 #ifndef SMALL 1261 time_t timestamp = 0; 1262 unsigned char name[PATH_MAX + 1]; 1263 #endif 1264 1265 /* gather the old name info */ 1266 1267 fd = open(file, O_RDONLY); 1268 if (fd < 0) { 1269 maybe_warn("can't open %s", file); 1270 goto lose; 1271 } 1272 1273 strlcpy(outfile, file, outsize); 1274 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1275 maybe_warnx("%s: unknown suffix -- ignored", file); 1276 goto lose; 1277 } 1278 1279 rbytes = read(fd, header1, sizeof header1); 1280 if (rbytes != sizeof header1) { 1281 /* we don't want to fail here. */ 1282 #ifndef SMALL 1283 if (fflag) 1284 goto lose; 1285 #endif 1286 if (rbytes == -1) 1287 maybe_warn("can't read %s", file); 1288 else 1289 maybe_warnx("%s: unexpected end of file", file); 1290 goto lose; 1291 } 1292 1293 method = file_gettype(header1); 1294 1295 #ifndef SMALL 1296 if (fflag == 0 && method == FT_UNKNOWN) { 1297 maybe_warnx("%s: not in gzip format", file); 1298 goto lose; 1299 } 1300 1301 #endif 1302 1303 #ifndef SMALL 1304 if (method == FT_GZIP && Nflag) { 1305 unsigned char ts[4]; /* timestamp */ 1306 1307 if (pread(fd, ts, sizeof ts, GZIP_TIMESTAMP) != sizeof ts) { 1308 if (!fflag) 1309 maybe_warn("can't read %s", file); 1310 goto lose; 1311 } 1312 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1313 1314 if (header1[3] & ORIG_NAME) { 1315 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1316 if (rbytes < 0) { 1317 maybe_warn("can't read %s", file); 1318 goto lose; 1319 } 1320 if (name[0] != 0) { 1321 /* preserve original directory name */ 1322 char *dp = strrchr(file, '/'); 1323 if (dp == NULL) 1324 dp = file; 1325 else 1326 dp++; 1327 snprintf(outfile, outsize, "%.*s%.*s", 1328 (int) (dp - file), 1329 file, (int) rbytes, name); 1330 } 1331 } 1332 } 1333 #endif 1334 lseek(fd, 0, SEEK_SET); 1335 1336 if (cflag == 0 || lflag) { 1337 if (fstat(fd, &isb) != 0) 1338 goto lose; 1339 #ifndef SMALL 1340 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1341 maybe_warnx("%s has %d other links -- skipping", 1342 file, isb.st_nlink - 1); 1343 goto lose; 1344 } 1345 if (nflag == 0 && timestamp) 1346 isb.st_mtime = timestamp; 1347 if (check_outfile(outfile) == 0) 1348 goto lose; 1349 #endif 1350 } 1351 1352 if (cflag == 0 && lflag == 0) { 1353 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1354 if (zfd == STDOUT_FILENO) { 1355 /* We won't close STDOUT_FILENO later... */ 1356 zfd = dup(zfd); 1357 close(STDOUT_FILENO); 1358 } 1359 if (zfd == -1) { 1360 maybe_warn("can't open %s", outfile); 1361 goto lose; 1362 } 1363 } else 1364 zfd = STDOUT_FILENO; 1365 1366 #ifndef NO_BZIP2_SUPPORT 1367 if (method == FT_BZIP2) { 1368 1369 /* XXX */ 1370 if (lflag) { 1371 maybe_warnx("no -l with bzip2 files"); 1372 goto lose; 1373 } 1374 1375 size = unbzip2(fd, zfd, NULL, 0, NULL); 1376 } else 1377 #endif 1378 1379 #ifndef NO_COMPRESS_SUPPORT 1380 if (method == FT_Z) { 1381 FILE *in, *out; 1382 1383 /* XXX */ 1384 if (lflag) { 1385 maybe_warnx("no -l with Lempel-Ziv files"); 1386 goto lose; 1387 } 1388 1389 if ((in = zdopen(fd)) == NULL) { 1390 maybe_warn("zdopen for read: %s", file); 1391 goto lose; 1392 } 1393 1394 out = fdopen(dup(zfd), "w"); 1395 if (out == NULL) { 1396 maybe_warn("fdopen for write: %s", outfile); 1397 fclose(in); 1398 goto lose; 1399 } 1400 1401 size = zuncompress(in, out, NULL, 0, NULL); 1402 /* need to fclose() if ferror() is true... */ 1403 if (ferror(in) | fclose(in)) { 1404 maybe_warn("failed infile fclose"); 1405 unlink(outfile); 1406 (void)fclose(out); 1407 } 1408 if (fclose(out) != 0) { 1409 maybe_warn("failed outfile fclose"); 1410 unlink(outfile); 1411 goto lose; 1412 } 1413 } else 1414 #endif 1415 1416 #ifndef SMALL 1417 if (method == FT_UNKNOWN) { 1418 if (lflag) { 1419 maybe_warnx("no -l for unknown filetypes"); 1420 goto lose; 1421 } 1422 size = cat_fd(NULL, 0, NULL, fd); 1423 } else 1424 #endif 1425 { 1426 if (lflag) { 1427 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1428 close(fd); 1429 return -1; /* XXX */ 1430 } 1431 1432 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1433 } 1434 1435 if (close(fd) != 0) 1436 maybe_warn("couldn't close input"); 1437 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1438 maybe_warn("couldn't close output"); 1439 1440 if (size == -1) { 1441 if (cflag == 0) 1442 unlink(outfile); 1443 maybe_warnx("%s: uncompress failed", file); 1444 return -1; 1445 } 1446 1447 /* if testing, or we uncompressed to stdout, this is all we need */ 1448 #ifndef SMALL 1449 if (tflag) 1450 return size; 1451 #endif 1452 /* if we are uncompressing to stdin, don't remove the file. */ 1453 if (cflag) 1454 return size; 1455 1456 /* 1457 * if we create a file... 1458 */ 1459 /* 1460 * if we can't stat the file don't remove the file. 1461 */ 1462 1463 ofd = open(outfile, O_RDWR, 0); 1464 if (ofd == -1) { 1465 maybe_warn("couldn't open (leaving original): %s", 1466 outfile); 1467 return -1; 1468 } 1469 if (fstat(ofd, &osb) != 0) { 1470 maybe_warn("couldn't stat (leaving original): %s", 1471 outfile); 1472 close(ofd); 1473 return -1; 1474 } 1475 if (osb.st_size != size) { 1476 maybe_warnx("stat gave different size: %" PRIdOFF 1477 " != %" PRIdOFF " (leaving original)", 1478 size, osb.st_size); 1479 close(ofd); 1480 unlink(outfile); 1481 return -1; 1482 } 1483 unlink_input(file, &isb); 1484 #ifndef SMALL 1485 copymodes(ofd, &isb, outfile); 1486 #endif 1487 close(ofd); 1488 return size; 1489 1490 lose: 1491 if (fd != -1) 1492 close(fd); 1493 if (zfd != -1 && zfd != STDOUT_FILENO) 1494 close(fd); 1495 return -1; 1496 } 1497 1498 #ifndef SMALL 1499 static off_t 1500 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1501 { 1502 char buf[BUFLEN]; 1503 off_t in_tot; 1504 ssize_t w; 1505 1506 in_tot = count; 1507 w = write(STDOUT_FILENO, prepend, count); 1508 if (w == -1 || (size_t)w != count) { 1509 maybe_warn("write to stdout"); 1510 return -1; 1511 } 1512 for (;;) { 1513 ssize_t rv; 1514 1515 rv = read(fd, buf, sizeof buf); 1516 if (rv == 0) 1517 break; 1518 if (rv < 0) { 1519 maybe_warn("read from fd %d", fd); 1520 break; 1521 } 1522 1523 if (write(STDOUT_FILENO, buf, rv) != rv) { 1524 maybe_warn("write to stdout"); 1525 break; 1526 } 1527 in_tot += rv; 1528 } 1529 1530 if (gsizep) 1531 *gsizep = in_tot; 1532 return (in_tot); 1533 } 1534 #endif 1535 1536 static void 1537 handle_stdin(void) 1538 { 1539 unsigned char header1[4]; 1540 off_t usize, gsize; 1541 enum filetype method; 1542 ssize_t bytes_read; 1543 #ifndef NO_COMPRESS_SUPPORT 1544 FILE *in; 1545 #endif 1546 1547 #ifndef SMALL 1548 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1549 maybe_warnx("standard input is a terminal -- ignoring"); 1550 return; 1551 } 1552 #endif 1553 1554 if (lflag) { 1555 struct stat isb; 1556 1557 /* XXX could read the whole file, etc. */ 1558 if (fstat(STDIN_FILENO, &isb) < 0) { 1559 maybe_warn("fstat"); 1560 return; 1561 } 1562 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1563 return; 1564 } 1565 1566 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1567 if (bytes_read == -1) { 1568 maybe_warn("can't read stdin"); 1569 return; 1570 } else if (bytes_read != sizeof(header1)) { 1571 maybe_warnx("unexpected EOF"); 1572 return; 1573 } 1574 1575 method = file_gettype(header1); 1576 switch (method) { 1577 default: 1578 #ifndef SMALL 1579 if (fflag == 0) { 1580 maybe_warnx("unknown compression format"); 1581 return; 1582 } 1583 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1584 break; 1585 #endif 1586 case FT_GZIP: 1587 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1588 header1, sizeof header1, &gsize, "(stdin)"); 1589 break; 1590 #ifndef NO_BZIP2_SUPPORT 1591 case FT_BZIP2: 1592 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1593 header1, sizeof header1, &gsize); 1594 break; 1595 #endif 1596 #ifndef NO_COMPRESS_SUPPORT 1597 case FT_Z: 1598 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1599 maybe_warnx("zopen of stdin"); 1600 return; 1601 } 1602 1603 usize = zuncompress(in, stdout, header1, sizeof header1, &gsize); 1604 fclose(in); 1605 break; 1606 #endif 1607 } 1608 1609 #ifndef SMALL 1610 if (vflag && !tflag && usize != -1 && gsize != -1) 1611 print_verbage(NULL, NULL, usize, gsize); 1612 if (vflag && tflag) 1613 print_test("(stdin)", usize != -1); 1614 #endif 1615 1616 } 1617 1618 static void 1619 handle_stdout(void) 1620 { 1621 off_t gsize, usize; 1622 struct stat sb; 1623 time_t systime; 1624 uint32_t mtime; 1625 int ret; 1626 1627 #ifndef SMALL 1628 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1629 maybe_warnx("standard output is a terminal -- ignoring"); 1630 return; 1631 } 1632 #endif 1633 /* If stdin is a file use it's mtime, otherwise use current time */ 1634 ret = fstat(STDIN_FILENO, &sb); 1635 1636 #ifndef SMALL 1637 if (ret < 0) { 1638 maybe_warn("Can't stat stdin"); 1639 return; 1640 } 1641 #endif 1642 1643 if (S_ISREG(sb.st_mode)) 1644 mtime = (uint32_t)sb.st_mtime; 1645 else { 1646 systime = time(NULL); 1647 #ifndef SMALL 1648 if (systime == -1) { 1649 maybe_warn("time"); 1650 return; 1651 } 1652 #endif 1653 mtime = (uint32_t)systime; 1654 } 1655 1656 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1657 #ifndef SMALL 1658 if (vflag && !tflag && usize != -1 && gsize != -1) 1659 print_verbage(NULL, NULL, usize, gsize); 1660 #endif 1661 } 1662 1663 /* do what is asked for, for the path name */ 1664 static void 1665 handle_pathname(char *path) 1666 { 1667 char *opath = path, *s = NULL; 1668 ssize_t len; 1669 int slen; 1670 struct stat sb; 1671 1672 /* check for stdout/stdin */ 1673 if (path[0] == '-' && path[1] == '\0') { 1674 if (dflag) 1675 handle_stdin(); 1676 else 1677 handle_stdout(); 1678 return; 1679 } 1680 1681 retry: 1682 if (stat(path, &sb) != 0) { 1683 /* lets try <path>.gz if we're decompressing */ 1684 if (dflag && s == NULL && errno == ENOENT) { 1685 len = strlen(path); 1686 slen = suffixes[0].ziplen; 1687 s = malloc(len + slen + 1); 1688 if (s == NULL) 1689 maybe_err("malloc"); 1690 memcpy(s, path, len); 1691 memcpy(s + len, suffixes[0].zipped, slen + 1); 1692 path = s; 1693 goto retry; 1694 } 1695 maybe_warn("can't stat: %s", opath); 1696 goto out; 1697 } 1698 1699 if (S_ISDIR(sb.st_mode)) { 1700 #ifndef SMALL 1701 if (rflag) 1702 handle_dir(path); 1703 else 1704 #endif 1705 maybe_warnx("%s is a directory", path); 1706 goto out; 1707 } 1708 1709 if (S_ISREG(sb.st_mode)) 1710 handle_file(path, &sb); 1711 else 1712 maybe_warnx("%s is not a regular file", path); 1713 1714 out: 1715 if (s) 1716 free(s); 1717 } 1718 1719 /* compress/decompress a file */ 1720 static void 1721 handle_file(char *file, struct stat *sbp) 1722 { 1723 off_t usize, gsize; 1724 char outfile[PATH_MAX]; 1725 1726 infile = file; 1727 if (dflag) { 1728 usize = file_uncompress(file, outfile, sizeof(outfile)); 1729 #ifndef SMALL 1730 if (vflag && tflag) 1731 print_test(file, usize != -1); 1732 #endif 1733 if (usize == -1) 1734 return; 1735 gsize = sbp->st_size; 1736 } else { 1737 gsize = file_compress(file, outfile, sizeof(outfile)); 1738 if (gsize == -1) 1739 return; 1740 usize = sbp->st_size; 1741 } 1742 1743 1744 #ifndef SMALL 1745 if (vflag && !tflag) 1746 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1747 #endif 1748 } 1749 1750 #ifndef SMALL 1751 /* this is used with -r to recursively descend directories */ 1752 static void 1753 handle_dir(char *dir) 1754 { 1755 char *path_argv[2]; 1756 FTS *fts; 1757 FTSENT *entry; 1758 1759 path_argv[0] = dir; 1760 path_argv[1] = 0; 1761 fts = fts_open(path_argv, FTS_PHYSICAL, NULL); 1762 if (fts == NULL) { 1763 warn("couldn't fts_open %s", dir); 1764 return; 1765 } 1766 1767 while ((entry = fts_read(fts))) { 1768 switch(entry->fts_info) { 1769 case FTS_D: 1770 case FTS_DP: 1771 continue; 1772 1773 case FTS_DNR: 1774 case FTS_ERR: 1775 case FTS_NS: 1776 maybe_warn("%s", entry->fts_path); 1777 continue; 1778 case FTS_F: 1779 handle_file(entry->fts_name, entry->fts_statp); 1780 } 1781 } 1782 (void)fts_close(fts); 1783 } 1784 #endif 1785 1786 /* print a ratio - size reduction as a fraction of uncompressed size */ 1787 static void 1788 print_ratio(off_t in, off_t out, FILE *where) 1789 { 1790 int percent10; /* 10 * percent */ 1791 off_t diff; 1792 char buff[8]; 1793 int len; 1794 1795 diff = in - out/2; 1796 if (diff <= 0) 1797 /* 1798 * Output is more than double size of input! print -99.9% 1799 * Quite possibly we've failed to get the original size. 1800 */ 1801 percent10 = -999; 1802 else { 1803 /* 1804 * We only need 12 bits of result from the final division, 1805 * so reduce the values until a 32bit division will suffice. 1806 */ 1807 while (in > 0x100000) { 1808 diff >>= 1; 1809 in >>= 1; 1810 } 1811 if (in != 0) 1812 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1813 else 1814 percent10 = 0; 1815 } 1816 1817 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1818 /* Move the '.' to before the last digit */ 1819 buff[len - 1] = buff[len - 2]; 1820 buff[len - 2] = '.'; 1821 fprintf(where, "%5s%%", buff); 1822 } 1823 1824 #ifndef SMALL 1825 /* print compression statistics, and the new name (if there is one!) */ 1826 static void 1827 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1828 { 1829 if (file) 1830 fprintf(stderr, "%s:%s ", file, 1831 strlen(file) < 7 ? "\t\t" : "\t"); 1832 print_ratio(usize, gsize, stderr); 1833 if (nfile) 1834 fprintf(stderr, " -- replaced with %s", nfile); 1835 fprintf(stderr, "\n"); 1836 fflush(stderr); 1837 } 1838 1839 /* print test results */ 1840 static void 1841 print_test(const char *file, int ok) 1842 { 1843 1844 if (exit_value == 0 && ok == 0) 1845 exit_value = 1; 1846 fprintf(stderr, "%s:%s %s\n", file, 1847 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1848 fflush(stderr); 1849 } 1850 #endif 1851 1852 /* print a file's info ala --list */ 1853 /* eg: 1854 compressed uncompressed ratio uncompressed_name 1855 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1856 */ 1857 static void 1858 print_list(int fd, off_t out, const char *outfile, time_t ts) 1859 { 1860 static int first = 1; 1861 #ifndef SMALL 1862 static off_t in_tot, out_tot; 1863 uint32_t crc = 0; 1864 #endif 1865 off_t in = 0, rv; 1866 1867 if (first) { 1868 #ifndef SMALL 1869 if (vflag) 1870 printf("method crc date time "); 1871 #endif 1872 if (qflag == 0) 1873 printf(" compressed uncompressed " 1874 "ratio uncompressed_name\n"); 1875 } 1876 first = 0; 1877 1878 /* print totals? */ 1879 #ifndef SMALL 1880 if (fd == -1) { 1881 in = in_tot; 1882 out = out_tot; 1883 } else 1884 #endif 1885 { 1886 /* read the last 4 bytes - this is the uncompressed size */ 1887 rv = lseek(fd, (off_t)(-8), SEEK_END); 1888 if (rv != -1) { 1889 unsigned char buf[8]; 1890 uint32_t usize; 1891 1892 if (read(fd, (char *)buf, sizeof(buf)) != sizeof(buf)) 1893 maybe_warn("read of uncompressed size"); 1894 usize = buf[4] | buf[5] << 8 | buf[6] << 16 | buf[7] << 24; 1895 in = (off_t)usize; 1896 #ifndef SMALL 1897 crc = buf[0] | buf[1] << 8 | buf[2] << 16 | buf[3] << 24; 1898 #endif 1899 } 1900 } 1901 1902 #ifndef SMALL 1903 if (vflag && fd == -1) 1904 printf(" "); 1905 else if (vflag) { 1906 char *date = ctime(&ts); 1907 1908 /* skip the day, 1/100th second, and year */ 1909 date += 4; 1910 date[12] = 0; 1911 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1912 } 1913 in_tot += in; 1914 out_tot += out; 1915 #endif 1916 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1917 print_ratio(in, out, stdout); 1918 printf(" %s\n", outfile); 1919 } 1920 1921 /* display the usage of NetBSD gzip */ 1922 static void 1923 usage(void) 1924 { 1925 1926 fprintf(stderr, "%s\n", gzip_version); 1927 fprintf(stderr, 1928 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n" 1929 #ifndef SMALL 1930 " -1 --fast fastest (worst) compression\n" 1931 " -2 .. -8 set compression level\n" 1932 " -9 --best best (slowest) compression\n" 1933 " -c --stdout write to stdout, keep original files\n" 1934 " --to-stdout\n" 1935 " -d --decompress uncompress files\n" 1936 " --uncompress\n" 1937 " -f --force force overwriting & compress links\n" 1938 " -h --help display this help\n" 1939 " -l --list list compressed file contents\n" 1940 " -N --name save or restore original file name and time stamp\n" 1941 " -n --no-name don't save original file name or time stamp\n" 1942 " -q --quiet output no warnings\n" 1943 " -r --recursive recursively compress files in directories\n" 1944 " -S .suf use suffix .suf instead of .gz\n" 1945 " --suffix .suf\n" 1946 " -t --test test compressed file\n" 1947 " -V --version display program version\n" 1948 " -v --verbose print extra statistics\n", 1949 #else 1950 , 1951 #endif 1952 getprogname()); 1953 exit(0); 1954 } 1955 1956 /* display the version of NetBSD gzip */ 1957 static void 1958 display_version(void) 1959 { 1960 1961 fprintf(stderr, "%s\n", gzip_version); 1962 exit(0); 1963 } 1964 1965 #ifndef NO_BZIP2_SUPPORT 1966 #include "unbzip2.c" 1967 #endif 1968 #ifndef NO_COMPRESS_SUPPORT 1969 #include "zuncompress.c" 1970 #endif 1971 1972 static ssize_t 1973 read_retry(int fd, void *buf, size_t sz) 1974 { 1975 char *cp = buf; 1976 size_t left = MIN(sz, (size_t) SSIZE_MAX); 1977 1978 while (left > 0) { 1979 ssize_t ret; 1980 1981 ret = read(fd, cp, left); 1982 if (ret == -1) { 1983 return ret; 1984 } else if (ret == 0) { 1985 break; /* EOF */ 1986 } 1987 cp += ret; 1988 left -= ret; 1989 } 1990 1991 return sz - left; 1992 } 1993
Indexes created Mon Sep 22 05:09:51 GMT 2025