1 /* $NetBSD: gzip.c,v 1.98 2010/11/06 21:42:32 mrg Exp $ */ 2 3 /* 4 * Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 21 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 22 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 23 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 24 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 26 * SUCH DAMAGE. 27 */ 28 29 #include <sys/cdefs.h> 30 #ifndef lint 31 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006\ 32 Matthew R. Green. All rights reserved."); 33 __RCSID("$NetBSD: gzip.c,v 1.98 2010/11/06 21:42:32 mrg Exp $"); 34 #endif /* not lint */ 35 36 /* 37 * gzip.c -- GPL free gzip using zlib. 38 * 39 * RFC 1950 covers the zlib format 40 * RFC 1951 covers the deflate format 41 * RFC 1952 covers the gzip format 42 * 43 * TODO: 44 * - use mmap where possible 45 * - handle some signals better (remove outfile?) 46 * - make bzip2/compress -v/-t/-l support work as well as possible 47 */ 48 49 #include <sys/param.h> 50 #include <sys/stat.h> 51 #include <sys/time.h> 52 53 #include <inttypes.h> 54 #include <unistd.h> 55 #include <stdio.h> 56 #include <string.h> 57 #include <stdlib.h> 58 #include <err.h> 59 #include <errno.h> 60 #include <fcntl.h> 61 #include <zlib.h> 62 #include <fts.h> 63 #include <libgen.h> 64 #include <stdarg.h> 65 #include <getopt.h> 66 #include <time.h> 67 68 #ifndef PRIdOFF 69 #define PRIdOFF PRId64 70 #endif 71 72 /* what type of file are we dealing with */ 73 enum filetype { 74 FT_GZIP, 75 #ifndef NO_BZIP2_SUPPORT 76 FT_BZIP2, 77 #endif 78 #ifndef NO_COMPRESS_SUPPORT 79 FT_Z, 80 #endif 81 #ifndef NO_PACK_SUPPORT 82 FT_PACK, 83 #endif 84 FT_LAST, 85 FT_UNKNOWN 86 }; 87 88 #ifndef NO_BZIP2_SUPPORT 89 #include <bzlib.h> 90 91 #define BZ2_SUFFIX ".bz2" 92 #define BZIP2_MAGIC "\102\132\150" 93 #endif 94 95 #ifndef NO_COMPRESS_SUPPORT 96 #define Z_SUFFIX ".Z" 97 #define Z_MAGIC "\037\235" 98 #endif 99 100 #ifndef NO_PACK_SUPPORT 101 #define PACK_MAGIC "\037\036" 102 #endif 103 104 #define GZ_SUFFIX ".gz" 105 106 #define BUFLEN (64 * 1024) 107 108 #define GZIP_MAGIC0 0x1F 109 #define GZIP_MAGIC1 0x8B 110 #define GZIP_OMAGIC1 0x9E 111 112 #define GZIP_TIMESTAMP (off_t)4 113 #define GZIP_ORIGNAME (off_t)10 114 115 #define HEAD_CRC 0x02 116 #define EXTRA_FIELD 0x04 117 #define ORIG_NAME 0x08 118 #define COMMENT 0x10 119 120 #define OS_CODE 3 /* Unix */ 121 122 typedef struct { 123 const char *zipped; 124 int ziplen; 125 const char *normal; /* for unzip - must not be longer than zipped */ 126 } suffixes_t; 127 static suffixes_t suffixes[] = { 128 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 129 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 130 #ifndef SMALL 131 SUFFIX(GZ_SUFFIX, ""), 132 SUFFIX(".z", ""), 133 SUFFIX("-gz", ""), 134 SUFFIX("-z", ""), 135 SUFFIX("_z", ""), 136 SUFFIX(".taz", ".tar"), 137 SUFFIX(".tgz", ".tar"), 138 #ifndef NO_BZIP2_SUPPORT 139 SUFFIX(BZ2_SUFFIX, ""), 140 #endif 141 #ifndef NO_COMPRESS_SUPPORT 142 SUFFIX(Z_SUFFIX, ""), 143 #endif 144 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 145 #endif /* SMALL */ 146 #undef SUFFIX 147 }; 148 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 149 #define SUFFIX_MAXLEN 30 150 151 static const char gzip_version[] = "NetBSD gzip 20101018"; 152 153 static int cflag; /* stdout mode */ 154 static int dflag; /* decompress mode */ 155 static int lflag; /* list mode */ 156 static int numflag = 6; /* gzip -1..-9 value */ 157 158 #ifndef SMALL 159 static int fflag; /* force mode */ 160 static int kflag; /* don't delete input files */ 161 static int nflag; /* don't save name/timestamp */ 162 static int Nflag; /* don't restore name/timestamp */ 163 static int qflag; /* quiet mode */ 164 static int rflag; /* recursive mode */ 165 static int tflag; /* test */ 166 static int vflag; /* verbose mode */ 167 #else 168 #define qflag 0 169 #define tflag 0 170 #endif 171 172 static int exit_value = 0; /* exit value */ 173 174 static char *infile; /* name of file coming in */ 175 176 static void maybe_err(const char *fmt, ...) 177 __attribute__((__format__(__printf__, 1, 2))); 178 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) 179 static void maybe_errx(const char *fmt, ...) 180 __attribute__((__format__(__printf__, 1, 2))); 181 #endif 182 static void maybe_warn(const char *fmt, ...) 183 __attribute__((__format__(__printf__, 1, 2))); 184 static void maybe_warnx(const char *fmt, ...) 185 __attribute__((__format__(__printf__, 1, 2))); 186 static enum filetype file_gettype(u_char *); 187 #ifdef SMALL 188 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 189 #endif 190 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 191 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 192 static off_t file_compress(char *, char *, size_t); 193 static off_t file_uncompress(char *, char *, size_t); 194 static void handle_pathname(char *); 195 static void handle_file(char *, struct stat *); 196 static void handle_stdin(void); 197 static void handle_stdout(void); 198 static void print_ratio(off_t, off_t, FILE *); 199 static void print_list(int fd, off_t, const char *, time_t); 200 static void usage(void); 201 static void display_version(void); 202 static const suffixes_t *check_suffix(char *, int); 203 static ssize_t read_retry(int, void *, size_t); 204 205 #ifdef SMALL 206 #define unlink_input(f, sb) unlink(f) 207 #else 208 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 209 static void prepend_gzip(char *, int *, char ***); 210 static void handle_dir(char *); 211 static void print_verbage(const char *, const char *, off_t, off_t); 212 static void print_test(const char *, int); 213 static void copymodes(int fd, const struct stat *, const char *file); 214 static int check_outfile(const char *outfile); 215 #endif 216 217 #ifndef NO_BZIP2_SUPPORT 218 static off_t unbzip2(int, int, char *, size_t, off_t *); 219 #endif 220 221 #ifndef NO_COMPRESS_SUPPORT 222 static FILE *zdopen(int); 223 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 224 #endif 225 226 #ifndef NO_PACK_SUPPORT 227 static off_t unpack(int, int, char *, size_t, off_t *); 228 #endif 229 230 int main(int, char *p[]); 231 232 #ifdef SMALL 233 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 234 #else 235 static const struct option longopts[] = { 236 { "stdout", no_argument, 0, 'c' }, 237 { "to-stdout", no_argument, 0, 'c' }, 238 { "decompress", no_argument, 0, 'd' }, 239 { "uncompress", no_argument, 0, 'd' }, 240 { "force", no_argument, 0, 'f' }, 241 { "help", no_argument, 0, 'h' }, 242 { "keep", no_argument, 0, 'k' }, 243 { "list", no_argument, 0, 'l' }, 244 { "no-name", no_argument, 0, 'n' }, 245 { "name", no_argument, 0, 'N' }, 246 { "quiet", no_argument, 0, 'q' }, 247 { "recursive", no_argument, 0, 'r' }, 248 { "suffix", required_argument, 0, 'S' }, 249 { "test", no_argument, 0, 't' }, 250 { "verbose", no_argument, 0, 'v' }, 251 { "version", no_argument, 0, 'V' }, 252 { "fast", no_argument, 0, '1' }, 253 { "best", no_argument, 0, '9' }, 254 #if 0 255 /* 256 * This is what else GNU gzip implements. --ascii isn't useful 257 * on NetBSD, and I don't care to have a --license. 258 */ 259 { "ascii", no_argument, 0, 'a' }, 260 { "license", no_argument, 0, 'L' }, 261 #endif 262 { NULL, no_argument, 0, 0 }, 263 }; 264 #endif 265 266 int 267 main(int argc, char **argv) 268 { 269 const char *progname = getprogname(); 270 #ifndef SMALL 271 char *gzip; 272 int len; 273 #endif 274 int ch; 275 276 /* XXX set up signals */ 277 278 #ifndef SMALL 279 if ((gzip = getenv("GZIP")) != NULL) 280 prepend_gzip(gzip, &argc, &argv); 281 #endif 282 283 /* 284 * XXX 285 * handle being called `gunzip', `zcat' and `gzcat' 286 */ 287 if (strcmp(progname, "gunzip") == 0) 288 dflag = 1; 289 else if (strcmp(progname, "zcat") == 0 || 290 strcmp(progname, "gzcat") == 0) 291 dflag = cflag = 1; 292 293 #ifdef SMALL 294 #define OPT_LIST "123456789cdhltV" 295 #else 296 #define OPT_LIST "123456789cdfhklNnqrS:tVv" 297 #endif 298 299 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 300 switch (ch) { 301 case '1': case '2': case '3': 302 case '4': case '5': case '6': 303 case '7': case '8': case '9': 304 numflag = ch - '0'; 305 break; 306 case 'c': 307 cflag = 1; 308 break; 309 case 'd': 310 dflag = 1; 311 break; 312 case 'l': 313 lflag = 1; 314 dflag = 1; 315 break; 316 case 'V': 317 display_version(); 318 /* NOTREACHED */ 319 #ifndef SMALL 320 case 'f': 321 fflag = 1; 322 break; 323 case 'k': 324 kflag = 1; 325 break; 326 case 'N': 327 nflag = 0; 328 Nflag = 1; 329 break; 330 case 'n': 331 nflag = 1; 332 Nflag = 0; 333 break; 334 case 'q': 335 qflag = 1; 336 break; 337 case 'r': 338 rflag = 1; 339 break; 340 case 'S': 341 len = strlen(optarg); 342 if (len != 0) { 343 if (len > SUFFIX_MAXLEN) 344 errx(1, "incorrect suffix: '%s'", optarg); 345 suffixes[0].zipped = optarg; 346 suffixes[0].ziplen = len; 347 } else { 348 suffixes[NUM_SUFFIXES - 1].zipped = ""; 349 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 350 } 351 break; 352 case 't': 353 cflag = 1; 354 tflag = 1; 355 dflag = 1; 356 break; 357 case 'v': 358 vflag = 1; 359 break; 360 #endif 361 default: 362 usage(); 363 /* NOTREACHED */ 364 } 365 } 366 argv += optind; 367 argc -= optind; 368 369 if (argc == 0) { 370 if (dflag) /* stdin mode */ 371 handle_stdin(); 372 else /* stdout mode */ 373 handle_stdout(); 374 } else { 375 do { 376 handle_pathname(argv[0]); 377 } while (*++argv); 378 } 379 #ifndef SMALL 380 if (qflag == 0 && lflag && argc > 1) 381 print_list(-1, 0, "(totals)", 0); 382 #endif 383 exit(exit_value); 384 } 385 386 /* maybe print a warning */ 387 void 388 maybe_warn(const char *fmt, ...) 389 { 390 va_list ap; 391 392 if (qflag == 0) { 393 va_start(ap, fmt); 394 vwarn(fmt, ap); 395 va_end(ap); 396 } 397 if (exit_value == 0) 398 exit_value = 1; 399 } 400 401 /* ... without an errno. */ 402 void 403 maybe_warnx(const char *fmt, ...) 404 { 405 va_list ap; 406 407 if (qflag == 0) { 408 va_start(ap, fmt); 409 vwarnx(fmt, ap); 410 va_end(ap); 411 } 412 if (exit_value == 0) 413 exit_value = 1; 414 } 415 416 /* maybe print an error */ 417 void 418 maybe_err(const char *fmt, ...) 419 { 420 va_list ap; 421 422 if (qflag == 0) { 423 va_start(ap, fmt); 424 vwarn(fmt, ap); 425 va_end(ap); 426 } 427 exit(2); 428 } 429 430 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) 431 /* ... without an errno. */ 432 void 433 maybe_errx(const char *fmt, ...) 434 { 435 va_list ap; 436 437 if (qflag == 0) { 438 va_start(ap, fmt); 439 vwarnx(fmt, ap); 440 va_end(ap); 441 } 442 exit(2); 443 } 444 #endif 445 446 #ifndef SMALL 447 /* split up $GZIP and prepend it to the argument list */ 448 static void 449 prepend_gzip(char *gzip, int *argc, char ***argv) 450 { 451 char *s, **nargv, **ac; 452 int nenvarg = 0, i; 453 454 /* scan how many arguments there are */ 455 for (s = gzip;;) { 456 while (*s == ' ' || *s == '\t') 457 s++; 458 if (*s == 0) 459 goto count_done; 460 nenvarg++; 461 while (*s != ' ' && *s != '\t') 462 if (*s++ == 0) 463 goto count_done; 464 } 465 count_done: 466 /* punt early */ 467 if (nenvarg == 0) 468 return; 469 470 *argc += nenvarg; 471 ac = *argv; 472 473 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 474 if (nargv == NULL) 475 maybe_err("malloc"); 476 477 /* stash this away */ 478 *argv = nargv; 479 480 /* copy the program name first */ 481 i = 0; 482 nargv[i++] = *(ac++); 483 484 /* take a copy of $GZIP and add it to the array */ 485 s = strdup(gzip); 486 if (s == NULL) 487 maybe_err("strdup"); 488 for (;;) { 489 /* Skip whitespaces. */ 490 while (*s == ' ' || *s == '\t') 491 s++; 492 if (*s == 0) 493 goto copy_done; 494 nargv[i++] = s; 495 /* Find the end of this argument. */ 496 while (*s != ' ' && *s != '\t') 497 if (*s++ == 0) 498 /* Argument followed by NUL. */ 499 goto copy_done; 500 /* Terminate by overwriting ' ' or '\t' with NUL. */ 501 *s++ = 0; 502 } 503 copy_done: 504 505 /* copy the original arguments and a NULL */ 506 while (*ac) 507 nargv[i++] = *(ac++); 508 nargv[i] = NULL; 509 } 510 #endif 511 512 /* compress input to output. Return bytes read, -1 on error */ 513 static off_t 514 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 515 { 516 z_stream z; 517 char *outbufp, *inbufp; 518 off_t in_tot = 0, out_tot = 0; 519 ssize_t in_size; 520 int i, error; 521 uLong crc; 522 #ifdef SMALL 523 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 524 0, 0, 0, 0, 525 0, OS_CODE }; 526 #endif 527 528 outbufp = malloc(BUFLEN); 529 inbufp = malloc(BUFLEN); 530 if (outbufp == NULL || inbufp == NULL) { 531 maybe_err("malloc failed"); 532 goto out; 533 } 534 535 memset(&z, 0, sizeof z); 536 z.zalloc = Z_NULL; 537 z.zfree = Z_NULL; 538 z.opaque = 0; 539 540 #ifdef SMALL 541 memcpy(outbufp, header, sizeof header); 542 i = sizeof header; 543 #else 544 if (nflag != 0) { 545 mtime = 0; 546 origname = ""; 547 } 548 549 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 550 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 551 *origname ? ORIG_NAME : 0, 552 mtime & 0xff, 553 (mtime >> 8) & 0xff, 554 (mtime >> 16) & 0xff, 555 (mtime >> 24) & 0xff, 556 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 557 OS_CODE, origname); 558 if (i >= BUFLEN) 559 /* this need PATH_MAX > BUFLEN ... */ 560 maybe_err("snprintf"); 561 if (*origname) 562 i++; 563 #endif 564 565 z.next_out = outbufp + i; 566 z.avail_out = BUFLEN - i; 567 568 error = deflateInit2(&z, numflag, Z_DEFLATED, 569 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 570 if (error != Z_OK) { 571 maybe_warnx("deflateInit2 failed"); 572 in_tot = -1; 573 goto out; 574 } 575 576 crc = crc32(0L, Z_NULL, 0); 577 for (;;) { 578 if (z.avail_out == 0) { 579 if (write(out, outbufp, BUFLEN) != BUFLEN) { 580 maybe_warn("write"); 581 out_tot = -1; 582 goto out; 583 } 584 585 out_tot += BUFLEN; 586 z.next_out = outbufp; 587 z.avail_out = BUFLEN; 588 } 589 590 if (z.avail_in == 0) { 591 in_size = read(in, inbufp, BUFLEN); 592 if (in_size < 0) { 593 maybe_warn("read"); 594 in_tot = -1; 595 goto out; 596 } 597 if (in_size == 0) 598 break; 599 600 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 601 in_tot += in_size; 602 z.next_in = inbufp; 603 z.avail_in = in_size; 604 } 605 606 error = deflate(&z, Z_NO_FLUSH); 607 if (error != Z_OK && error != Z_STREAM_END) { 608 maybe_warnx("deflate failed"); 609 in_tot = -1; 610 goto out; 611 } 612 } 613 614 /* clean up */ 615 for (;;) { 616 size_t len; 617 ssize_t w; 618 619 error = deflate(&z, Z_FINISH); 620 if (error != Z_OK && error != Z_STREAM_END) { 621 maybe_warnx("deflate failed"); 622 in_tot = -1; 623 goto out; 624 } 625 626 len = (char *)z.next_out - outbufp; 627 628 w = write(out, outbufp, len); 629 if (w == -1 || (size_t)w != len) { 630 maybe_warn("write"); 631 out_tot = -1; 632 goto out; 633 } 634 out_tot += len; 635 z.next_out = outbufp; 636 z.avail_out = BUFLEN; 637 638 if (error == Z_STREAM_END) 639 break; 640 } 641 642 if (deflateEnd(&z) != Z_OK) { 643 maybe_warnx("deflateEnd failed"); 644 in_tot = -1; 645 goto out; 646 } 647 648 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 649 (int)crc & 0xff, 650 (int)(crc >> 8) & 0xff, 651 (int)(crc >> 16) & 0xff, 652 (int)(crc >> 24) & 0xff, 653 (int)in_tot & 0xff, 654 (int)(in_tot >> 8) & 0xff, 655 (int)(in_tot >> 16) & 0xff, 656 (int)(in_tot >> 24) & 0xff); 657 if (i != 8) 658 maybe_err("snprintf"); 659 #if 0 660 if (in_tot > 0xffffffff) 661 maybe_warn("input file size >= 4GB cannot be saved"); 662 #endif 663 if (write(out, outbufp, i) != i) { 664 maybe_warn("write"); 665 in_tot = -1; 666 } else 667 out_tot += i; 668 669 out: 670 if (inbufp != NULL) 671 free(inbufp); 672 if (outbufp != NULL) 673 free(outbufp); 674 if (gsizep) 675 *gsizep = out_tot; 676 return in_tot; 677 } 678 679 /* 680 * uncompress input to output then close the input. return the 681 * uncompressed size written, and put the compressed sized read 682 * into `*gsizep'. 683 */ 684 static off_t 685 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 686 const char *filename) 687 { 688 z_stream z; 689 char *outbufp, *inbufp; 690 off_t out_tot = -1, in_tot = 0; 691 uint32_t out_sub_tot = 0; 692 enum { 693 GZSTATE_MAGIC0, 694 GZSTATE_MAGIC1, 695 GZSTATE_METHOD, 696 GZSTATE_FLAGS, 697 GZSTATE_SKIPPING, 698 GZSTATE_EXTRA, 699 GZSTATE_EXTRA2, 700 GZSTATE_EXTRA3, 701 GZSTATE_ORIGNAME, 702 GZSTATE_COMMENT, 703 GZSTATE_HEAD_CRC1, 704 GZSTATE_HEAD_CRC2, 705 GZSTATE_INIT, 706 GZSTATE_READ, 707 GZSTATE_CRC, 708 GZSTATE_LEN, 709 } state = GZSTATE_MAGIC0; 710 int flags = 0, skip_count = 0; 711 int error = Z_STREAM_ERROR, done_reading = 0; 712 uLong crc = 0; 713 ssize_t wr; 714 int needmore = 0; 715 716 #define ADVANCE() { z.next_in++; z.avail_in--; } 717 718 if ((outbufp = malloc(BUFLEN)) == NULL) { 719 maybe_err("malloc failed"); 720 goto out2; 721 } 722 if ((inbufp = malloc(BUFLEN)) == NULL) { 723 maybe_err("malloc failed"); 724 goto out1; 725 } 726 727 memset(&z, 0, sizeof z); 728 z.avail_in = prelen; 729 z.next_in = pre; 730 z.avail_out = BUFLEN; 731 z.next_out = outbufp; 732 z.zalloc = NULL; 733 z.zfree = NULL; 734 z.opaque = 0; 735 736 in_tot = prelen; 737 out_tot = 0; 738 739 for (;;) { 740 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 741 ssize_t in_size; 742 743 if (z.avail_in > 0) { 744 memmove(inbufp, z.next_in, z.avail_in); 745 } 746 z.next_in = inbufp; 747 in_size = read(in, z.next_in + z.avail_in, 748 BUFLEN - z.avail_in); 749 750 if (in_size == -1) { 751 maybe_warn("failed to read stdin"); 752 goto stop_and_fail; 753 } else if (in_size == 0) { 754 done_reading = 1; 755 } 756 757 z.avail_in += in_size; 758 needmore = 0; 759 760 in_tot += in_size; 761 } 762 if (z.avail_in == 0) { 763 if (done_reading && state != GZSTATE_MAGIC0) { 764 maybe_warnx("%s: unexpected end of file", 765 filename); 766 goto stop_and_fail; 767 } 768 goto stop; 769 } 770 switch (state) { 771 case GZSTATE_MAGIC0: 772 if (*z.next_in != GZIP_MAGIC0) { 773 if (in_tot > 0) { 774 maybe_warnx("%s: trailing garbage " 775 "ignored", filename); 776 goto stop; 777 } 778 maybe_warnx("input not gziped (MAGIC0)"); 779 goto stop_and_fail; 780 } 781 ADVANCE(); 782 state++; 783 out_sub_tot = 0; 784 crc = crc32(0L, Z_NULL, 0); 785 break; 786 787 case GZSTATE_MAGIC1: 788 if (*z.next_in != GZIP_MAGIC1 && 789 *z.next_in != GZIP_OMAGIC1) { 790 maybe_warnx("input not gziped (MAGIC1)"); 791 goto stop_and_fail; 792 } 793 ADVANCE(); 794 state++; 795 break; 796 797 case GZSTATE_METHOD: 798 if (*z.next_in != Z_DEFLATED) { 799 maybe_warnx("unknown compression method"); 800 goto stop_and_fail; 801 } 802 ADVANCE(); 803 state++; 804 break; 805 806 case GZSTATE_FLAGS: 807 flags = *z.next_in; 808 ADVANCE(); 809 skip_count = 6; 810 state++; 811 break; 812 813 case GZSTATE_SKIPPING: 814 if (skip_count > 0) { 815 skip_count--; 816 ADVANCE(); 817 } else 818 state++; 819 break; 820 821 case GZSTATE_EXTRA: 822 if ((flags & EXTRA_FIELD) == 0) { 823 state = GZSTATE_ORIGNAME; 824 break; 825 } 826 skip_count = *z.next_in; 827 ADVANCE(); 828 state++; 829 break; 830 831 case GZSTATE_EXTRA2: 832 skip_count |= ((*z.next_in) << 8); 833 ADVANCE(); 834 state++; 835 break; 836 837 case GZSTATE_EXTRA3: 838 if (skip_count > 0) { 839 skip_count--; 840 ADVANCE(); 841 } else 842 state++; 843 break; 844 845 case GZSTATE_ORIGNAME: 846 if ((flags & ORIG_NAME) == 0) { 847 state++; 848 break; 849 } 850 if (*z.next_in == 0) 851 state++; 852 ADVANCE(); 853 break; 854 855 case GZSTATE_COMMENT: 856 if ((flags & COMMENT) == 0) { 857 state++; 858 break; 859 } 860 if (*z.next_in == 0) 861 state++; 862 ADVANCE(); 863 break; 864 865 case GZSTATE_HEAD_CRC1: 866 if (flags & HEAD_CRC) 867 skip_count = 2; 868 else 869 skip_count = 0; 870 state++; 871 break; 872 873 case GZSTATE_HEAD_CRC2: 874 if (skip_count > 0) { 875 skip_count--; 876 ADVANCE(); 877 } else 878 state++; 879 break; 880 881 case GZSTATE_INIT: 882 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 883 maybe_warnx("failed to inflateInit"); 884 goto stop_and_fail; 885 } 886 state++; 887 break; 888 889 case GZSTATE_READ: 890 error = inflate(&z, Z_FINISH); 891 switch (error) { 892 /* Z_BUF_ERROR goes with Z_FINISH... */ 893 case Z_BUF_ERROR: 894 if (z.avail_out > 0 && !done_reading) 895 continue; 896 897 case Z_STREAM_END: 898 case Z_OK: 899 break; 900 901 case Z_NEED_DICT: 902 maybe_warnx("Z_NEED_DICT error"); 903 goto stop_and_fail; 904 case Z_DATA_ERROR: 905 maybe_warnx("data stream error"); 906 goto stop_and_fail; 907 case Z_STREAM_ERROR: 908 maybe_warnx("internal stream error"); 909 goto stop_and_fail; 910 case Z_MEM_ERROR: 911 maybe_warnx("memory allocation error"); 912 goto stop_and_fail; 913 914 default: 915 maybe_warn("unknown error from inflate(): %d", 916 error); 917 } 918 wr = BUFLEN - z.avail_out; 919 920 if (wr != 0) { 921 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 922 if ( 923 #ifndef SMALL 924 /* don't write anything with -t */ 925 tflag == 0 && 926 #endif 927 write(out, outbufp, wr) != wr) { 928 maybe_warn("error writing to output"); 929 goto stop_and_fail; 930 } 931 932 out_tot += wr; 933 out_sub_tot += wr; 934 } 935 936 if (error == Z_STREAM_END) { 937 inflateEnd(&z); 938 state++; 939 } 940 941 z.next_out = outbufp; 942 z.avail_out = BUFLEN; 943 944 break; 945 case GZSTATE_CRC: 946 { 947 uLong origcrc; 948 949 if (z.avail_in < 4) { 950 if (!done_reading) { 951 needmore = 1; 952 continue; 953 } 954 maybe_warnx("truncated input"); 955 goto stop_and_fail; 956 } 957 origcrc = ((unsigned)z.next_in[0] & 0xff) | 958 ((unsigned)z.next_in[1] & 0xff) << 8 | 959 ((unsigned)z.next_in[2] & 0xff) << 16 | 960 ((unsigned)z.next_in[3] & 0xff) << 24; 961 if (origcrc != crc) { 962 maybe_warnx("invalid compressed" 963 " data--crc error"); 964 goto stop_and_fail; 965 } 966 } 967 968 z.avail_in -= 4; 969 z.next_in += 4; 970 971 if (!z.avail_in && done_reading) { 972 goto stop; 973 } 974 state++; 975 break; 976 case GZSTATE_LEN: 977 { 978 uLong origlen; 979 980 if (z.avail_in < 4) { 981 if (!done_reading) { 982 needmore = 1; 983 continue; 984 } 985 maybe_warnx("truncated input"); 986 goto stop_and_fail; 987 } 988 origlen = ((unsigned)z.next_in[0] & 0xff) | 989 ((unsigned)z.next_in[1] & 0xff) << 8 | 990 ((unsigned)z.next_in[2] & 0xff) << 16 | 991 ((unsigned)z.next_in[3] & 0xff) << 24; 992 993 if (origlen != out_sub_tot) { 994 maybe_warnx("invalid compressed" 995 " data--length error"); 996 goto stop_and_fail; 997 } 998 } 999 1000 z.avail_in -= 4; 1001 z.next_in += 4; 1002 1003 if (error < 0) { 1004 maybe_warnx("decompression error"); 1005 goto stop_and_fail; 1006 } 1007 state = GZSTATE_MAGIC0; 1008 break; 1009 } 1010 continue; 1011 stop_and_fail: 1012 out_tot = -1; 1013 stop: 1014 break; 1015 } 1016 if (state > GZSTATE_INIT) 1017 inflateEnd(&z); 1018 1019 free(inbufp); 1020 out1: 1021 free(outbufp); 1022 out2: 1023 if (gsizep) 1024 *gsizep = in_tot; 1025 return (out_tot); 1026 } 1027 1028 #ifndef SMALL 1029 /* 1030 * set the owner, mode, flags & utimes using the given file descriptor. 1031 * file is only used in possible warning messages. 1032 */ 1033 static void 1034 copymodes(int fd, const struct stat *sbp, const char *file) 1035 { 1036 struct timeval times[2]; 1037 struct stat sb; 1038 1039 /* 1040 * If we have no info on the input, give this file some 1041 * default values and return.. 1042 */ 1043 if (sbp == NULL) { 1044 mode_t mask = umask(022); 1045 1046 (void)fchmod(fd, DEFFILEMODE & ~mask); 1047 (void)umask(mask); 1048 return; 1049 } 1050 sb = *sbp; 1051 1052 /* if the chown fails, remove set-id bits as-per compress(1) */ 1053 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1054 if (errno != EPERM) 1055 maybe_warn("couldn't fchown: %s", file); 1056 sb.st_mode &= ~(S_ISUID|S_ISGID); 1057 } 1058 1059 /* we only allow set-id and the 9 normal permission bits */ 1060 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1061 if (fchmod(fd, sb.st_mode) < 0) 1062 maybe_warn("couldn't fchmod: %s", file); 1063 1064 /* only try flags if they exist already */ 1065 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1066 maybe_warn("couldn't fchflags: %s", file); 1067 1068 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1069 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1070 if (futimes(fd, times) < 0) 1071 maybe_warn("couldn't utimes: %s", file); 1072 } 1073 #endif 1074 1075 /* what sort of file is this? */ 1076 static enum filetype 1077 file_gettype(u_char *buf) 1078 { 1079 1080 if (buf[0] == GZIP_MAGIC0 && 1081 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1082 return FT_GZIP; 1083 else 1084 #ifndef NO_BZIP2_SUPPORT 1085 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1086 buf[3] >= '0' && buf[3] <= '9') 1087 return FT_BZIP2; 1088 else 1089 #endif 1090 #ifndef NO_COMPRESS_SUPPORT 1091 if (memcmp(buf, Z_MAGIC, 2) == 0) 1092 return FT_Z; 1093 else 1094 #endif 1095 #ifndef NO_PACK_SUPPORT 1096 if (memcmp(buf, PACK_MAGIC, 2) == 0) 1097 return FT_PACK; 1098 else 1099 #endif 1100 return FT_UNKNOWN; 1101 } 1102 1103 #ifndef SMALL 1104 /* check the outfile is OK. */ 1105 static int 1106 check_outfile(const char *outfile) 1107 { 1108 struct stat sb; 1109 int ok = 1; 1110 1111 if (lflag == 0 && stat(outfile, &sb) == 0) { 1112 if (fflag) 1113 unlink(outfile); 1114 else if (isatty(STDIN_FILENO)) { 1115 char ans[10] = { 'n', '\0' }; /* default */ 1116 1117 fprintf(stderr, "%s already exists -- do you wish to " 1118 "overwrite (y or n)? " , outfile); 1119 (void)fgets(ans, sizeof(ans) - 1, stdin); 1120 if (ans[0] != 'y' && ans[0] != 'Y') { 1121 fprintf(stderr, "\tnot overwriting\n"); 1122 ok = 0; 1123 } else 1124 unlink(outfile); 1125 } else { 1126 maybe_warnx("%s already exists -- skipping", outfile); 1127 ok = 0; 1128 } 1129 } 1130 return ok; 1131 } 1132 1133 static void 1134 unlink_input(const char *file, const struct stat *sb) 1135 { 1136 struct stat nsb; 1137 1138 if (kflag) 1139 return; 1140 if (stat(file, &nsb) != 0) 1141 /* Must be gone already */ 1142 return; 1143 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1144 /* Definitely a different file */ 1145 return; 1146 unlink(file); 1147 } 1148 #endif 1149 1150 static const suffixes_t * 1151 check_suffix(char *file, int xlate) 1152 { 1153 const suffixes_t *s; 1154 int len = strlen(file); 1155 char *sp; 1156 1157 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1158 /* if it doesn't fit in "a.suf", don't bother */ 1159 if (s->ziplen >= len) 1160 continue; 1161 sp = file + len - s->ziplen; 1162 if (strcmp(s->zipped, sp) != 0) 1163 continue; 1164 if (xlate) 1165 strcpy(sp, s->normal); 1166 return s; 1167 } 1168 return NULL; 1169 } 1170 1171 /* 1172 * compress the given file: create a corresponding .gz file and remove the 1173 * original. 1174 */ 1175 static off_t 1176 file_compress(char *file, char *outfile, size_t outsize) 1177 { 1178 int in; 1179 int out; 1180 off_t size, insize; 1181 #ifndef SMALL 1182 struct stat isb, osb; 1183 const suffixes_t *suff; 1184 #endif 1185 1186 in = open(file, O_RDONLY); 1187 if (in == -1) { 1188 maybe_warn("can't open %s", file); 1189 return -1; 1190 } 1191 1192 if (cflag == 0) { 1193 #ifndef SMALL 1194 if (fstat(in, &isb) == 0) { 1195 if (isb.st_nlink > 1 && fflag == 0) { 1196 maybe_warnx("%s has %d other link%s -- " 1197 "skipping", file, isb.st_nlink - 1, 1198 isb.st_nlink == 1 ? "" : "s"); 1199 close(in); 1200 return -1; 1201 } 1202 } 1203 1204 if (fflag == 0 && (suff = check_suffix(file, 0)) 1205 && suff->zipped[0] != 0) { 1206 maybe_warnx("%s already has %s suffix -- unchanged", 1207 file, suff->zipped); 1208 close(in); 1209 return -1; 1210 } 1211 #endif 1212 1213 /* Add (usually) .gz to filename */ 1214 if ((size_t)snprintf(outfile, outsize, "%s%s", 1215 file, suffixes[0].zipped) >= outsize) 1216 memcpy(outfile + outsize - suffixes[0].ziplen - 1, 1217 suffixes[0].zipped, suffixes[0].ziplen + 1); 1218 1219 #ifndef SMALL 1220 if (check_outfile(outfile) == 0) { 1221 close(in); 1222 return -1; 1223 } 1224 #endif 1225 } 1226 1227 if (cflag == 0) { 1228 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1229 if (out == -1) { 1230 maybe_warn("could not create output: %s", outfile); 1231 fclose(stdin); 1232 return -1; 1233 } 1234 } else 1235 out = STDOUT_FILENO; 1236 1237 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1238 1239 (void)close(in); 1240 1241 /* 1242 * If there was an error, insize will be -1. 1243 * If we compressed to stdout, just return the size. 1244 * Otherwise stat the file and check it is the correct size. 1245 * We only blow away the file if we can stat the output and it 1246 * has the expected size. 1247 */ 1248 if (cflag != 0) 1249 return insize == -1 ? -1 : size; 1250 1251 #ifndef SMALL 1252 if (fstat(out, &osb) != 0) { 1253 maybe_warn("couldn't stat: %s", outfile); 1254 goto bad_outfile; 1255 } 1256 1257 if (osb.st_size != size) { 1258 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1259 " != %" PRIdOFF "), deleting", 1260 outfile, osb.st_size, size); 1261 goto bad_outfile; 1262 } 1263 1264 copymodes(out, &isb, outfile); 1265 #endif 1266 if (close(out) == -1) 1267 maybe_warn("couldn't close output"); 1268 1269 /* output is good, ok to delete input */ 1270 unlink_input(file, &isb); 1271 return size; 1272 1273 #ifndef SMALL 1274 bad_outfile: 1275 if (close(out) == -1) 1276 maybe_warn("couldn't close output"); 1277 1278 maybe_warnx("leaving original %s", file); 1279 unlink(outfile); 1280 return size; 1281 #endif 1282 } 1283 1284 /* uncompress the given file and remove the original */ 1285 static off_t 1286 file_uncompress(char *file, char *outfile, size_t outsize) 1287 { 1288 struct stat isb, osb; 1289 off_t size; 1290 ssize_t rbytes; 1291 unsigned char header1[4]; 1292 enum filetype method; 1293 int fd, ofd, zfd = -1; 1294 #ifndef SMALL 1295 ssize_t rv; 1296 time_t timestamp = 0; 1297 unsigned char name[PATH_MAX + 1]; 1298 #endif 1299 1300 /* gather the old name info */ 1301 1302 fd = open(file, O_RDONLY); 1303 if (fd < 0) { 1304 maybe_warn("can't open %s", file); 1305 goto lose; 1306 } 1307 1308 strlcpy(outfile, file, outsize); 1309 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1310 maybe_warnx("%s: unknown suffix -- ignored", file); 1311 goto lose; 1312 } 1313 1314 rbytes = read(fd, header1, sizeof header1); 1315 if (rbytes != sizeof header1) { 1316 /* we don't want to fail here. */ 1317 #ifndef SMALL 1318 if (fflag) 1319 goto lose; 1320 #endif 1321 if (rbytes == -1) 1322 maybe_warn("can't read %s", file); 1323 else 1324 goto unexpected_EOF; 1325 goto lose; 1326 } 1327 1328 method = file_gettype(header1); 1329 1330 #ifndef SMALL 1331 if (fflag == 0 && method == FT_UNKNOWN) { 1332 maybe_warnx("%s: not in gzip format", file); 1333 goto lose; 1334 } 1335 1336 #endif 1337 1338 #ifndef SMALL 1339 if (method == FT_GZIP && Nflag) { 1340 unsigned char ts[4]; /* timestamp */ 1341 1342 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1343 if (rv >= 0 && rv < (ssize_t)(sizeof ts)) 1344 goto unexpected_EOF; 1345 if (rv == -1) { 1346 if (!fflag) 1347 maybe_warn("can't read %s", file); 1348 goto lose; 1349 } 1350 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1351 1352 if (header1[3] & ORIG_NAME) { 1353 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1354 if (rbytes < 0) { 1355 maybe_warn("can't read %s", file); 1356 goto lose; 1357 } 1358 if (name[0] != 0) { 1359 /* preserve original directory name */ 1360 char *dp = strrchr(file, '/'); 1361 if (dp == NULL) 1362 dp = file; 1363 else 1364 dp++; 1365 snprintf(outfile, outsize, "%.*s%.*s", 1366 (int) (dp - file), 1367 file, (int) rbytes, name); 1368 } 1369 } 1370 } 1371 #endif 1372 lseek(fd, 0, SEEK_SET); 1373 1374 if (cflag == 0 || lflag) { 1375 if (fstat(fd, &isb) != 0) 1376 goto lose; 1377 #ifndef SMALL 1378 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1379 maybe_warnx("%s has %d other links -- skipping", 1380 file, isb.st_nlink - 1); 1381 goto lose; 1382 } 1383 if (nflag == 0 && timestamp) 1384 isb.st_mtime = timestamp; 1385 if (check_outfile(outfile) == 0) 1386 goto lose; 1387 #endif 1388 } 1389 1390 if (cflag == 0 && lflag == 0) { 1391 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1392 if (zfd == STDOUT_FILENO) { 1393 /* We won't close STDOUT_FILENO later... */ 1394 zfd = dup(zfd); 1395 close(STDOUT_FILENO); 1396 } 1397 if (zfd == -1) { 1398 maybe_warn("can't open %s", outfile); 1399 goto lose; 1400 } 1401 } else 1402 zfd = STDOUT_FILENO; 1403 1404 #ifndef NO_BZIP2_SUPPORT 1405 if (method == FT_BZIP2) { 1406 1407 /* XXX */ 1408 if (lflag) { 1409 maybe_warnx("no -l with bzip2 files"); 1410 goto lose; 1411 } 1412 1413 size = unbzip2(fd, zfd, NULL, 0, NULL); 1414 } else 1415 #endif 1416 1417 #ifndef NO_COMPRESS_SUPPORT 1418 if (method == FT_Z) { 1419 FILE *in, *out; 1420 1421 /* XXX */ 1422 if (lflag) { 1423 maybe_warnx("no -l with Lempel-Ziv files"); 1424 goto lose; 1425 } 1426 1427 if ((in = zdopen(fd)) == NULL) { 1428 maybe_warn("zdopen for read: %s", file); 1429 goto lose; 1430 } 1431 1432 out = fdopen(dup(zfd), "w"); 1433 if (out == NULL) { 1434 maybe_warn("fdopen for write: %s", outfile); 1435 fclose(in); 1436 goto lose; 1437 } 1438 1439 size = zuncompress(in, out, NULL, 0, NULL); 1440 /* need to fclose() if ferror() is true... */ 1441 if (ferror(in) | fclose(in)) { 1442 maybe_warn("failed infile fclose"); 1443 unlink(outfile); 1444 (void)fclose(out); 1445 } 1446 if (fclose(out) != 0) { 1447 maybe_warn("failed outfile fclose"); 1448 unlink(outfile); 1449 goto lose; 1450 } 1451 } else 1452 #endif 1453 1454 #ifndef NO_PACK_SUPPORT 1455 if (method == FT_PACK) { 1456 if (lflag) { 1457 maybe_warnx("no -l with packed files"); 1458 goto lose; 1459 } 1460 1461 size = unpack(fd, zfd, NULL, 0, NULL); 1462 } else 1463 #endif 1464 1465 #ifndef SMALL 1466 if (method == FT_UNKNOWN) { 1467 if (lflag) { 1468 maybe_warnx("no -l for unknown filetypes"); 1469 goto lose; 1470 } 1471 size = cat_fd(NULL, 0, NULL, fd); 1472 } else 1473 #endif 1474 { 1475 if (lflag) { 1476 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1477 close(fd); 1478 return -1; /* XXX */ 1479 } 1480 1481 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1482 } 1483 1484 if (close(fd) != 0) 1485 maybe_warn("couldn't close input"); 1486 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1487 maybe_warn("couldn't close output"); 1488 1489 if (size == -1) { 1490 if (cflag == 0) 1491 unlink(outfile); 1492 maybe_warnx("%s: uncompress failed", file); 1493 return -1; 1494 } 1495 1496 /* if testing, or we uncompressed to stdout, this is all we need */ 1497 #ifndef SMALL 1498 if (tflag) 1499 return size; 1500 #endif 1501 /* if we are uncompressing to stdin, don't remove the file. */ 1502 if (cflag) 1503 return size; 1504 1505 /* 1506 * if we create a file... 1507 */ 1508 /* 1509 * if we can't stat the file don't remove the file. 1510 */ 1511 1512 ofd = open(outfile, O_RDWR, 0); 1513 if (ofd == -1) { 1514 maybe_warn("couldn't open (leaving original): %s", 1515 outfile); 1516 return -1; 1517 } 1518 if (fstat(ofd, &osb) != 0) { 1519 maybe_warn("couldn't stat (leaving original): %s", 1520 outfile); 1521 close(ofd); 1522 return -1; 1523 } 1524 if (osb.st_size != size) { 1525 maybe_warnx("stat gave different size: %" PRIdOFF 1526 " != %" PRIdOFF " (leaving original)", 1527 size, osb.st_size); 1528 close(ofd); 1529 unlink(outfile); 1530 return -1; 1531 } 1532 unlink_input(file, &isb); 1533 #ifndef SMALL 1534 copymodes(ofd, &isb, outfile); 1535 #endif 1536 close(ofd); 1537 return size; 1538 1539 unexpected_EOF: 1540 maybe_warnx("%s: unexpected end of file", file); 1541 lose: 1542 if (fd != -1) 1543 close(fd); 1544 if (zfd != -1 && zfd != STDOUT_FILENO) 1545 close(fd); 1546 return -1; 1547 } 1548 1549 #ifndef SMALL 1550 static off_t 1551 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1552 { 1553 char buf[BUFLEN]; 1554 off_t in_tot; 1555 ssize_t w; 1556 1557 in_tot = count; 1558 w = write(STDOUT_FILENO, prepend, count); 1559 if (w == -1 || (size_t)w != count) { 1560 maybe_warn("write to stdout"); 1561 return -1; 1562 } 1563 for (;;) { 1564 ssize_t rv; 1565 1566 rv = read(fd, buf, sizeof buf); 1567 if (rv == 0) 1568 break; 1569 if (rv < 0) { 1570 maybe_warn("read from fd %d", fd); 1571 break; 1572 } 1573 1574 if (write(STDOUT_FILENO, buf, rv) != rv) { 1575 maybe_warn("write to stdout"); 1576 break; 1577 } 1578 in_tot += rv; 1579 } 1580 1581 if (gsizep) 1582 *gsizep = in_tot; 1583 return (in_tot); 1584 } 1585 #endif 1586 1587 static void 1588 handle_stdin(void) 1589 { 1590 unsigned char header1[4]; 1591 off_t usize, gsize; 1592 enum filetype method; 1593 ssize_t bytes_read; 1594 #ifndef NO_COMPRESS_SUPPORT 1595 FILE *in; 1596 #endif 1597 1598 #ifndef SMALL 1599 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1600 maybe_warnx("standard input is a terminal -- ignoring"); 1601 return; 1602 } 1603 #endif 1604 1605 if (lflag) { 1606 struct stat isb; 1607 1608 /* XXX could read the whole file, etc. */ 1609 if (fstat(STDIN_FILENO, &isb) < 0) { 1610 maybe_warn("fstat"); 1611 return; 1612 } 1613 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1614 return; 1615 } 1616 1617 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1618 if (bytes_read == -1) { 1619 maybe_warn("can't read stdin"); 1620 return; 1621 } else if (bytes_read != sizeof(header1)) { 1622 maybe_warnx("(stdin): unexpected end of file"); 1623 return; 1624 } 1625 1626 method = file_gettype(header1); 1627 switch (method) { 1628 default: 1629 #ifndef SMALL 1630 if (fflag == 0) { 1631 maybe_warnx("unknown compression format"); 1632 return; 1633 } 1634 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1635 break; 1636 #endif 1637 case FT_GZIP: 1638 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1639 header1, sizeof header1, &gsize, "(stdin)"); 1640 break; 1641 #ifndef NO_BZIP2_SUPPORT 1642 case FT_BZIP2: 1643 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1644 header1, sizeof header1, &gsize); 1645 break; 1646 #endif 1647 #ifndef NO_COMPRESS_SUPPORT 1648 case FT_Z: 1649 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1650 maybe_warnx("zopen of stdin"); 1651 return; 1652 } 1653 1654 usize = zuncompress(in, stdout, header1, sizeof header1, &gsize); 1655 fclose(in); 1656 break; 1657 #endif 1658 #ifndef NO_PACK_SUPPORT 1659 case FT_PACK: 1660 usize = unpack(STDIN_FILENO, STDOUT_FILENO, 1661 (char *)header1, sizeof header1, &gsize); 1662 break; 1663 #endif 1664 } 1665 1666 #ifndef SMALL 1667 if (vflag && !tflag && usize != -1 && gsize != -1) 1668 print_verbage(NULL, NULL, usize, gsize); 1669 if (vflag && tflag) 1670 print_test("(stdin)", usize != -1); 1671 #endif 1672 1673 } 1674 1675 static void 1676 handle_stdout(void) 1677 { 1678 off_t gsize, usize; 1679 struct stat sb; 1680 time_t systime; 1681 uint32_t mtime; 1682 int ret; 1683 1684 #ifndef SMALL 1685 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1686 maybe_warnx("standard output is a terminal -- ignoring"); 1687 return; 1688 } 1689 #endif 1690 /* If stdin is a file use it's mtime, otherwise use current time */ 1691 ret = fstat(STDIN_FILENO, &sb); 1692 1693 #ifndef SMALL 1694 if (ret < 0) { 1695 maybe_warn("Can't stat stdin"); 1696 return; 1697 } 1698 #endif 1699 1700 if (S_ISREG(sb.st_mode)) 1701 mtime = (uint32_t)sb.st_mtime; 1702 else { 1703 systime = time(NULL); 1704 #ifndef SMALL 1705 if (systime == -1) { 1706 maybe_warn("time"); 1707 return; 1708 } 1709 #endif 1710 mtime = (uint32_t)systime; 1711 } 1712 1713 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1714 #ifndef SMALL 1715 if (vflag && !tflag && usize != -1 && gsize != -1) 1716 print_verbage(NULL, NULL, usize, gsize); 1717 #endif 1718 } 1719 1720 /* do what is asked for, for the path name */ 1721 static void 1722 handle_pathname(char *path) 1723 { 1724 char *opath = path, *s = NULL; 1725 ssize_t len; 1726 int slen; 1727 struct stat sb; 1728 1729 /* check for stdout/stdin */ 1730 if (path[0] == '-' && path[1] == '\0') { 1731 if (dflag) 1732 handle_stdin(); 1733 else 1734 handle_stdout(); 1735 return; 1736 } 1737 1738 retry: 1739 if (stat(path, &sb) != 0) { 1740 /* lets try <path>.gz if we're decompressing */ 1741 if (dflag && s == NULL && errno == ENOENT) { 1742 len = strlen(path); 1743 slen = suffixes[0].ziplen; 1744 s = malloc(len + slen + 1); 1745 if (s == NULL) 1746 maybe_err("malloc"); 1747 memcpy(s, path, len); 1748 memcpy(s + len, suffixes[0].zipped, slen + 1); 1749 path = s; 1750 goto retry; 1751 } 1752 maybe_warn("can't stat: %s", opath); 1753 goto out; 1754 } 1755 1756 if (S_ISDIR(sb.st_mode)) { 1757 #ifndef SMALL 1758 if (rflag) 1759 handle_dir(path); 1760 else 1761 #endif 1762 maybe_warnx("%s is a directory", path); 1763 goto out; 1764 } 1765 1766 if (S_ISREG(sb.st_mode)) 1767 handle_file(path, &sb); 1768 else 1769 maybe_warnx("%s is not a regular file", path); 1770 1771 out: 1772 if (s) 1773 free(s); 1774 } 1775 1776 /* compress/decompress a file */ 1777 static void 1778 handle_file(char *file, struct stat *sbp) 1779 { 1780 off_t usize, gsize; 1781 char outfile[PATH_MAX]; 1782 1783 infile = file; 1784 if (dflag) { 1785 usize = file_uncompress(file, outfile, sizeof(outfile)); 1786 #ifndef SMALL 1787 if (vflag && tflag) 1788 print_test(file, usize != -1); 1789 #endif 1790 if (usize == -1) 1791 return; 1792 gsize = sbp->st_size; 1793 } else { 1794 gsize = file_compress(file, outfile, sizeof(outfile)); 1795 if (gsize == -1) 1796 return; 1797 usize = sbp->st_size; 1798 } 1799 1800 1801 #ifndef SMALL 1802 if (vflag && !tflag) 1803 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1804 #endif 1805 } 1806 1807 #ifndef SMALL 1808 /* this is used with -r to recursively descend directories */ 1809 static void 1810 handle_dir(char *dir) 1811 { 1812 char *path_argv[2]; 1813 FTS *fts; 1814 FTSENT *entry; 1815 1816 path_argv[0] = dir; 1817 path_argv[1] = 0; 1818 fts = fts_open(path_argv, FTS_PHYSICAL, NULL); 1819 if (fts == NULL) { 1820 warn("couldn't fts_open %s", dir); 1821 return; 1822 } 1823 1824 while ((entry = fts_read(fts))) { 1825 switch(entry->fts_info) { 1826 case FTS_D: 1827 case FTS_DP: 1828 continue; 1829 1830 case FTS_DNR: 1831 case FTS_ERR: 1832 case FTS_NS: 1833 maybe_warn("%s", entry->fts_path); 1834 continue; 1835 case FTS_F: 1836 handle_file(entry->fts_name, entry->fts_statp); 1837 } 1838 } 1839 (void)fts_close(fts); 1840 } 1841 #endif 1842 1843 /* print a ratio - size reduction as a fraction of uncompressed size */ 1844 static void 1845 print_ratio(off_t in, off_t out, FILE *where) 1846 { 1847 int percent10; /* 10 * percent */ 1848 off_t diff; 1849 char buff[8]; 1850 int len; 1851 1852 diff = in - out/2; 1853 if (diff <= 0) 1854 /* 1855 * Output is more than double size of input! print -99.9% 1856 * Quite possibly we've failed to get the original size. 1857 */ 1858 percent10 = -999; 1859 else { 1860 /* 1861 * We only need 12 bits of result from the final division, 1862 * so reduce the values until a 32bit division will suffice. 1863 */ 1864 while (in > 0x100000) { 1865 diff >>= 1; 1866 in >>= 1; 1867 } 1868 if (in != 0) 1869 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1870 else 1871 percent10 = 0; 1872 } 1873 1874 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1875 /* Move the '.' to before the last digit */ 1876 buff[len - 1] = buff[len - 2]; 1877 buff[len - 2] = '.'; 1878 fprintf(where, "%5s%%", buff); 1879 } 1880 1881 #ifndef SMALL 1882 /* print compression statistics, and the new name (if there is one!) */ 1883 static void 1884 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1885 { 1886 if (file) 1887 fprintf(stderr, "%s:%s ", file, 1888 strlen(file) < 7 ? "\t\t" : "\t"); 1889 print_ratio(usize, gsize, stderr); 1890 if (nfile) 1891 fprintf(stderr, " -- replaced with %s", nfile); 1892 fprintf(stderr, "\n"); 1893 fflush(stderr); 1894 } 1895 1896 /* print test results */ 1897 static void 1898 print_test(const char *file, int ok) 1899 { 1900 1901 if (exit_value == 0 && ok == 0) 1902 exit_value = 1; 1903 fprintf(stderr, "%s:%s %s\n", file, 1904 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1905 fflush(stderr); 1906 } 1907 #endif 1908 1909 /* print a file's info ala --list */ 1910 /* eg: 1911 compressed uncompressed ratio uncompressed_name 1912 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1913 */ 1914 static void 1915 print_list(int fd, off_t out, const char *outfile, time_t ts) 1916 { 1917 static int first = 1; 1918 #ifndef SMALL 1919 static off_t in_tot, out_tot; 1920 uint32_t crc = 0; 1921 #endif 1922 off_t in = 0, rv; 1923 1924 if (first) { 1925 #ifndef SMALL 1926 if (vflag) 1927 printf("method crc date time "); 1928 #endif 1929 if (qflag == 0) 1930 printf(" compressed uncompressed " 1931 "ratio uncompressed_name\n"); 1932 } 1933 first = 0; 1934 1935 /* print totals? */ 1936 #ifndef SMALL 1937 if (fd == -1) { 1938 in = in_tot; 1939 out = out_tot; 1940 } else 1941 #endif 1942 { 1943 /* read the last 4 bytes - this is the uncompressed size */ 1944 rv = lseek(fd, (off_t)(-8), SEEK_END); 1945 if (rv != -1) { 1946 unsigned char buf[8]; 1947 uint32_t usize; 1948 1949 rv = read(fd, (char *)buf, sizeof(buf)); 1950 if (rv == -1) 1951 maybe_warn("read of uncompressed size"); 1952 else if (rv != sizeof(buf)) 1953 maybe_warnx("read of uncompressed size"); 1954 1955 else { 1956 usize = buf[4] | buf[5] << 8 | 1957 buf[6] << 16 | buf[7] << 24; 1958 in = (off_t)usize; 1959 #ifndef SMALL 1960 crc = buf[0] | buf[1] << 8 | 1961 buf[2] << 16 | buf[3] << 24; 1962 #endif 1963 } 1964 } 1965 } 1966 1967 #ifndef SMALL 1968 if (vflag && fd == -1) 1969 printf(" "); 1970 else if (vflag) { 1971 char *date = ctime(&ts); 1972 1973 /* skip the day, 1/100th second, and year */ 1974 date += 4; 1975 date[12] = 0; 1976 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1977 } 1978 in_tot += in; 1979 out_tot += out; 1980 #endif 1981 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1982 print_ratio(in, out, stdout); 1983 printf(" %s\n", outfile); 1984 } 1985 1986 /* display the usage of NetBSD gzip */ 1987 static void 1988 usage(void) 1989 { 1990 1991 fprintf(stderr, "%s\n", gzip_version); 1992 fprintf(stderr, 1993 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n" 1994 #ifndef SMALL 1995 " -1 --fast fastest (worst) compression\n" 1996 " -2 .. -8 set compression level\n" 1997 " -9 --best best (slowest) compression\n" 1998 " -c --stdout write to stdout, keep original files\n" 1999 " --to-stdout\n" 2000 " -d --decompress uncompress files\n" 2001 " --uncompress\n" 2002 " -f --force force overwriting & compress links\n" 2003 " -h --help display this help\n" 2004 " -k --keep don't delete input files during operation\n" 2005 " -l --list list compressed file contents\n" 2006 " -N --name save or restore original file name and time stamp\n" 2007 " -n --no-name don't save original file name or time stamp\n" 2008 " -q --quiet output no warnings\n" 2009 " -r --recursive recursively compress files in directories\n" 2010 " -S .suf use suffix .suf instead of .gz\n" 2011 " --suffix .suf\n" 2012 " -t --test test compressed file\n" 2013 " -V --version display program version\n" 2014 " -v --verbose print extra statistics\n", 2015 #else 2016 , 2017 #endif 2018 getprogname()); 2019 exit(0); 2020 } 2021 2022 /* display the version of NetBSD gzip */ 2023 static void 2024 display_version(void) 2025 { 2026 2027 fprintf(stderr, "%s\n", gzip_version); 2028 exit(0); 2029 } 2030 2031 #ifndef NO_BZIP2_SUPPORT 2032 #include "unbzip2.c" 2033 #endif 2034 #ifndef NO_COMPRESS_SUPPORT 2035 #include "zuncompress.c" 2036 #endif 2037 #ifndef NO_PACK_SUPPORT 2038 #include "unpack.c" 2039 #endif 2040 2041 static ssize_t 2042 read_retry(int fd, void *buf, size_t sz) 2043 { 2044 char *cp = buf; 2045 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2046 2047 while (left > 0) { 2048 ssize_t ret; 2049 2050 ret = read(fd, cp, left); 2051 if (ret == -1) { 2052 return ret; 2053 } else if (ret == 0) { 2054 break; /* EOF */ 2055 } 2056 cp += ret; 2057 left -= ret; 2058 } 2059 2060 return sz - left; 2061 } 2062