1 /* $NetBSD: gzip.c,v 1.90 2007/01/31 08:23:22 martin 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 * 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, 2006 Matthew R. Green\n\ 34 All rights reserved.\n"); 35 __RCSID("$NetBSD: gzip.c,v 1.90 2007/01/31 08:23:22 martin 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 20060927"; 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 out_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_and_fail; 750 } 751 goto stop; 752 } 753 switch (state) { 754 case GZSTATE_MAGIC0: 755 if (*z.next_in != GZIP_MAGIC0) { 756 if (in_tot > 0) { 757 maybe_warnx("%s: trailing garbage " 758 "ignored", filename); 759 goto stop; 760 } 761 maybe_warnx("input not gziped (MAGIC0)"); 762 goto stop_and_fail; 763 } 764 ADVANCE(); 765 state++; 766 out_sub_tot = 0; 767 crc = crc32(0L, Z_NULL, 0); 768 break; 769 770 case GZSTATE_MAGIC1: 771 if (*z.next_in != GZIP_MAGIC1 && 772 *z.next_in != GZIP_OMAGIC1) { 773 maybe_warnx("input not gziped (MAGIC1)"); 774 goto stop_and_fail; 775 } 776 ADVANCE(); 777 state++; 778 break; 779 780 case GZSTATE_METHOD: 781 if (*z.next_in != Z_DEFLATED) { 782 maybe_warnx("unknown compression method"); 783 goto stop_and_fail; 784 } 785 ADVANCE(); 786 state++; 787 break; 788 789 case GZSTATE_FLAGS: 790 flags = *z.next_in; 791 ADVANCE(); 792 skip_count = 6; 793 state++; 794 break; 795 796 case GZSTATE_SKIPPING: 797 if (skip_count > 0) { 798 skip_count--; 799 ADVANCE(); 800 } else 801 state++; 802 break; 803 804 case GZSTATE_EXTRA: 805 if ((flags & EXTRA_FIELD) == 0) { 806 state = GZSTATE_ORIGNAME; 807 break; 808 } 809 skip_count = *z.next_in; 810 ADVANCE(); 811 state++; 812 break; 813 814 case GZSTATE_EXTRA2: 815 skip_count |= ((*z.next_in) << 8); 816 ADVANCE(); 817 state++; 818 break; 819 820 case GZSTATE_EXTRA3: 821 if (skip_count > 0) { 822 skip_count--; 823 ADVANCE(); 824 } else 825 state++; 826 break; 827 828 case GZSTATE_ORIGNAME: 829 if ((flags & ORIG_NAME) == 0) { 830 state++; 831 break; 832 } 833 if (*z.next_in == 0) 834 state++; 835 ADVANCE(); 836 break; 837 838 case GZSTATE_COMMENT: 839 if ((flags & COMMENT) == 0) { 840 state++; 841 break; 842 } 843 if (*z.next_in == 0) 844 state++; 845 ADVANCE(); 846 break; 847 848 case GZSTATE_HEAD_CRC1: 849 if (flags & HEAD_CRC) 850 skip_count = 2; 851 else 852 skip_count = 0; 853 state++; 854 break; 855 856 case GZSTATE_HEAD_CRC2: 857 if (skip_count > 0) { 858 skip_count--; 859 ADVANCE(); 860 } else 861 state++; 862 break; 863 864 case GZSTATE_INIT: 865 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 866 maybe_warnx("failed to inflateInit"); 867 goto stop_and_fail; 868 } 869 state++; 870 break; 871 872 case GZSTATE_READ: 873 error = inflate(&z, Z_FINISH); 874 switch (error) { 875 /* Z_BUF_ERROR goes with Z_FINISH... */ 876 case Z_BUF_ERROR: 877 case Z_STREAM_END: 878 case Z_OK: 879 break; 880 881 case Z_NEED_DICT: 882 maybe_warnx("Z_NEED_DICT error"); 883 goto stop_and_fail; 884 case Z_DATA_ERROR: 885 maybe_warnx("data stream error"); 886 goto stop_and_fail; 887 case Z_STREAM_ERROR: 888 maybe_warnx("internal stream error"); 889 goto stop_and_fail; 890 case Z_MEM_ERROR: 891 maybe_warnx("memory allocation error"); 892 goto stop_and_fail; 893 894 default: 895 maybe_warn("unknown error from inflate(): %d", 896 error); 897 } 898 wr = BUFLEN - z.avail_out; 899 900 if (wr != 0) { 901 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 902 if ( 903 #ifndef SMALL 904 /* don't write anything with -t */ 905 tflag == 0 && 906 #endif 907 write(out, outbufp, wr) != wr) { 908 maybe_warn("error writing to output"); 909 goto stop_and_fail; 910 } 911 912 out_tot += wr; 913 out_sub_tot += wr; 914 } 915 916 if (error == Z_STREAM_END) { 917 inflateEnd(&z); 918 state++; 919 } 920 921 z.next_out = outbufp; 922 z.avail_out = BUFLEN; 923 924 break; 925 case GZSTATE_CRC: 926 { 927 uLong origcrc; 928 929 if (z.avail_in < 4) { 930 if (!done_reading) { 931 needmore = 1; 932 continue; 933 } 934 maybe_warnx("truncated input"); 935 goto stop_and_fail; 936 } 937 origcrc = ((unsigned)z.next_in[0] & 0xff) | 938 ((unsigned)z.next_in[1] & 0xff) << 8 | 939 ((unsigned)z.next_in[2] & 0xff) << 16 | 940 ((unsigned)z.next_in[3] & 0xff) << 24; 941 if (origcrc != crc) { 942 maybe_warnx("invalid compressed" 943 " data--crc error"); 944 goto stop_and_fail; 945 } 946 } 947 948 z.avail_in -= 4; 949 z.next_in += 4; 950 951 if (!z.avail_in && done_reading) { 952 goto stop; 953 } 954 state++; 955 break; 956 case GZSTATE_LEN: 957 { 958 uLong origlen; 959 960 if (z.avail_in < 4) { 961 if (!done_reading) { 962 needmore = 1; 963 continue; 964 } 965 maybe_warnx("truncated input"); 966 goto stop_and_fail; 967 } 968 origlen = ((unsigned)z.next_in[0] & 0xff) | 969 ((unsigned)z.next_in[1] & 0xff) << 8 | 970 ((unsigned)z.next_in[2] & 0xff) << 16 | 971 ((unsigned)z.next_in[3] & 0xff) << 24; 972 973 if (origlen != out_sub_tot) { 974 maybe_warnx("invalid compressed" 975 " data--length error"); 976 goto stop_and_fail; 977 } 978 } 979 980 z.avail_in -= 4; 981 z.next_in += 4; 982 983 if (error < 0) { 984 maybe_warnx("decompression error"); 985 goto stop_and_fail; 986 } 987 state = GZSTATE_MAGIC0; 988 break; 989 } 990 continue; 991 stop_and_fail: 992 out_tot = -1; 993 stop: 994 break; 995 } 996 if (state > GZSTATE_INIT) 997 inflateEnd(&z); 998 999 free(inbufp); 1000 out1: 1001 free(outbufp); 1002 out2: 1003 if (gsizep) 1004 *gsizep = in_tot; 1005 return (out_tot); 1006 } 1007 1008 #ifndef SMALL 1009 /* 1010 * set the owner, mode, flags & utimes using the given file descriptor. 1011 * file is only used in possible warning messages. 1012 */ 1013 static void 1014 copymodes(int fd, const struct stat *sbp, const char *file) 1015 { 1016 struct timeval times[2]; 1017 struct stat sb; 1018 1019 /* 1020 * If we have no info on the input, give this file some 1021 * default values and return.. 1022 */ 1023 if (sbp == NULL) { 1024 mode_t mask = umask(022); 1025 1026 (void)fchmod(fd, DEFFILEMODE & ~mask); 1027 (void)umask(mask); 1028 return; 1029 } 1030 sb = *sbp; 1031 1032 /* if the chown fails, remove set-id bits as-per compress(1) */ 1033 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1034 if (errno != EPERM) 1035 maybe_warn("couldn't fchown: %s", file); 1036 sb.st_mode &= ~(S_ISUID|S_ISGID); 1037 } 1038 1039 /* we only allow set-id and the 9 normal permission bits */ 1040 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1041 if (fchmod(fd, sb.st_mode) < 0) 1042 maybe_warn("couldn't fchmod: %s", file); 1043 1044 /* only try flags if they exist already */ 1045 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1046 maybe_warn("couldn't fchflags: %s", file); 1047 1048 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1049 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1050 if (futimes(fd, times) < 0) 1051 maybe_warn("couldn't utimes: %s", file); 1052 } 1053 #endif 1054 1055 /* what sort of file is this? */ 1056 static enum filetype 1057 file_gettype(u_char *buf) 1058 { 1059 1060 if (buf[0] == GZIP_MAGIC0 && 1061 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1062 return FT_GZIP; 1063 else 1064 #ifndef NO_BZIP2_SUPPORT 1065 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1066 buf[3] >= '0' && buf[3] <= '9') 1067 return FT_BZIP2; 1068 else 1069 #endif 1070 #ifndef NO_COMPRESS_SUPPORT 1071 if (memcmp(buf, Z_MAGIC, 2) == 0) 1072 return FT_Z; 1073 else 1074 #endif 1075 return FT_UNKNOWN; 1076 } 1077 1078 #ifndef SMALL 1079 /* check the outfile is OK. */ 1080 static int 1081 check_outfile(const char *outfile) 1082 { 1083 struct stat sb; 1084 int ok = 1; 1085 1086 if (lflag == 0 && stat(outfile, &sb) == 0) { 1087 if (fflag) 1088 unlink(outfile); 1089 else if (isatty(STDIN_FILENO)) { 1090 char ans[10] = { 'n', '\0' }; /* default */ 1091 1092 fprintf(stderr, "%s already exists -- do you wish to " 1093 "overwrite (y or n)? " , outfile); 1094 (void)fgets(ans, sizeof(ans) - 1, stdin); 1095 if (ans[0] != 'y' && ans[0] != 'Y') { 1096 fprintf(stderr, "\tnot overwriting\n"); 1097 ok = 0; 1098 } else 1099 unlink(outfile); 1100 } else { 1101 maybe_warnx("%s already exists -- skipping", outfile); 1102 ok = 0; 1103 } 1104 } 1105 return ok; 1106 } 1107 1108 static void 1109 unlink_input(const char *file, const struct stat *sb) 1110 { 1111 struct stat nsb; 1112 1113 if (stat(file, &nsb) != 0) 1114 /* Must be gone alrady */ 1115 return; 1116 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1117 /* Definitely a different file */ 1118 return; 1119 unlink(file); 1120 } 1121 #endif 1122 1123 static const suffixes_t * 1124 check_suffix(char *file, int xlate) 1125 { 1126 const suffixes_t *s; 1127 int len = strlen(file); 1128 char *sp; 1129 1130 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1131 /* if it doesn't fit in "a.suf", don't bother */ 1132 if (s->ziplen >= len) 1133 continue; 1134 sp = file + len - s->ziplen; 1135 if (strcmp(s->zipped, sp) != 0) 1136 continue; 1137 if (xlate) 1138 strcpy(sp, s->normal); 1139 return s; 1140 } 1141 return NULL; 1142 } 1143 1144 /* 1145 * compress the given file: create a corresponding .gz file and remove the 1146 * original. 1147 */ 1148 static off_t 1149 file_compress(char *file, char *outfile, size_t outsize) 1150 { 1151 int in; 1152 int out; 1153 off_t size, insize; 1154 #ifndef SMALL 1155 struct stat isb, osb; 1156 const suffixes_t *suff; 1157 #endif 1158 1159 in = open(file, O_RDONLY); 1160 if (in == -1) { 1161 maybe_warn("can't open %s", file); 1162 return -1; 1163 } 1164 1165 if (cflag == 0) { 1166 #ifndef SMALL 1167 if (fstat(in, &isb) == 0) { 1168 if (isb.st_nlink > 1 && fflag == 0) { 1169 maybe_warnx("%s has %d other link%s -- " 1170 "skipping", file, isb.st_nlink - 1, 1171 isb.st_nlink == 1 ? "" : "s"); 1172 close(in); 1173 return -1; 1174 } 1175 } 1176 1177 if (fflag == 0 && (suff = check_suffix(file, 0)) 1178 && suff->zipped[0] != 0) { 1179 maybe_warnx("%s already has %s suffix -- unchanged", 1180 file, suff->zipped); 1181 close(in); 1182 return -1; 1183 } 1184 #endif 1185 1186 /* Add (usually) .gz to filename */ 1187 if ((size_t)snprintf(outfile, outsize, "%s%s", 1188 file, suffixes[0].zipped) >= outsize) 1189 memcpy(outfile - suffixes[0].ziplen - 1, 1190 suffixes[0].zipped, suffixes[0].ziplen + 1); 1191 1192 #ifndef SMALL 1193 if (check_outfile(outfile) == 0) { 1194 close(in); 1195 return -1; 1196 } 1197 #endif 1198 } 1199 1200 if (cflag == 0) { 1201 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1202 if (out == -1) { 1203 maybe_warn("could not create output: %s", outfile); 1204 fclose(stdin); 1205 return -1; 1206 } 1207 } else 1208 out = STDOUT_FILENO; 1209 1210 insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1211 1212 (void)close(in); 1213 1214 /* 1215 * If there was an error, insize will be -1. 1216 * If we compressed to stdout, just return the size. 1217 * Otherwise stat the file and check it is the correct size. 1218 * We only blow away the file if we can stat the output and it 1219 * has the expected size. 1220 */ 1221 if (cflag != 0) 1222 return insize == -1 ? -1 : size; 1223 1224 #ifndef SMALL 1225 if (fstat(out, &osb) != 0) { 1226 maybe_warn("couldn't stat: %s", outfile); 1227 goto bad_outfile; 1228 } 1229 1230 if (osb.st_size != size) { 1231 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1232 " != %" PRIdOFF "), deleting", 1233 outfile, osb.st_size, size); 1234 goto bad_outfile; 1235 } 1236 1237 copymodes(out, &isb, outfile); 1238 #endif 1239 if (close(out) == -1) 1240 maybe_warn("couldn't close output"); 1241 1242 /* output is good, ok to delete input */ 1243 unlink_input(file, &isb); 1244 return size; 1245 1246 #ifndef SMALL 1247 bad_outfile: 1248 if (close(out) == -1) 1249 maybe_warn("couldn't close output"); 1250 1251 maybe_warnx("leaving original %s", file); 1252 unlink(outfile); 1253 return size; 1254 #endif 1255 } 1256 1257 /* uncompress the given file and remove the original */ 1258 static off_t 1259 file_uncompress(char *file, char *outfile, size_t outsize) 1260 { 1261 struct stat isb, osb; 1262 off_t size; 1263 ssize_t rbytes; 1264 unsigned char header1[4]; 1265 enum filetype method; 1266 int rv, fd, ofd, zfd = -1; 1267 #ifndef SMALL 1268 time_t timestamp = 0; 1269 unsigned char name[PATH_MAX + 1]; 1270 #endif 1271 1272 /* gather the old name info */ 1273 1274 fd = open(file, O_RDONLY); 1275 if (fd < 0) { 1276 maybe_warn("can't open %s", file); 1277 goto lose; 1278 } 1279 1280 strlcpy(outfile, file, outsize); 1281 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1282 maybe_warnx("%s: unknown suffix -- ignored", file); 1283 goto lose; 1284 } 1285 1286 rbytes = read(fd, header1, sizeof header1); 1287 if (rbytes != sizeof header1) { 1288 /* we don't want to fail here. */ 1289 #ifndef SMALL 1290 if (fflag) 1291 goto lose; 1292 #endif 1293 if (rbytes == -1) 1294 maybe_warn("can't read %s", file); 1295 else 1296 goto unexpected_EOF; 1297 goto lose; 1298 } 1299 1300 method = file_gettype(header1); 1301 1302 #ifndef SMALL 1303 if (fflag == 0 && method == FT_UNKNOWN) { 1304 maybe_warnx("%s: not in gzip format", file); 1305 goto lose; 1306 } 1307 1308 #endif 1309 1310 #ifndef SMALL 1311 if (method == FT_GZIP && Nflag) { 1312 unsigned char ts[4]; /* timestamp */ 1313 1314 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1315 if (rv >= 0 && rv < sizeof ts) 1316 goto unexpected_EOF; 1317 if (rv == -1) { 1318 if (!fflag) 1319 maybe_warn("can't read %s", file); 1320 goto lose; 1321 } 1322 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1323 1324 if (header1[3] & ORIG_NAME) { 1325 rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME); 1326 if (rbytes < 0) { 1327 maybe_warn("can't read %s", file); 1328 goto lose; 1329 } 1330 if (name[0] != 0) { 1331 /* preserve original directory name */ 1332 char *dp = strrchr(file, '/'); 1333 if (dp == NULL) 1334 dp = file; 1335 else 1336 dp++; 1337 snprintf(outfile, outsize, "%.*s%.*s", 1338 (int) (dp - file), 1339 file, (int) rbytes, name); 1340 } 1341 } 1342 } 1343 #endif 1344 lseek(fd, 0, SEEK_SET); 1345 1346 if (cflag == 0 || lflag) { 1347 if (fstat(fd, &isb) != 0) 1348 goto lose; 1349 #ifndef SMALL 1350 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1351 maybe_warnx("%s has %d other links -- skipping", 1352 file, isb.st_nlink - 1); 1353 goto lose; 1354 } 1355 if (nflag == 0 && timestamp) 1356 isb.st_mtime = timestamp; 1357 if (check_outfile(outfile) == 0) 1358 goto lose; 1359 #endif 1360 } 1361 1362 if (cflag == 0 && lflag == 0) { 1363 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1364 if (zfd == STDOUT_FILENO) { 1365 /* We won't close STDOUT_FILENO later... */ 1366 zfd = dup(zfd); 1367 close(STDOUT_FILENO); 1368 } 1369 if (zfd == -1) { 1370 maybe_warn("can't open %s", outfile); 1371 goto lose; 1372 } 1373 } else 1374 zfd = STDOUT_FILENO; 1375 1376 #ifndef NO_BZIP2_SUPPORT 1377 if (method == FT_BZIP2) { 1378 1379 /* XXX */ 1380 if (lflag) { 1381 maybe_warnx("no -l with bzip2 files"); 1382 goto lose; 1383 } 1384 1385 size = unbzip2(fd, zfd, NULL, 0, NULL); 1386 } else 1387 #endif 1388 1389 #ifndef NO_COMPRESS_SUPPORT 1390 if (method == FT_Z) { 1391 FILE *in, *out; 1392 1393 /* XXX */ 1394 if (lflag) { 1395 maybe_warnx("no -l with Lempel-Ziv files"); 1396 goto lose; 1397 } 1398 1399 if ((in = zdopen(fd)) == NULL) { 1400 maybe_warn("zdopen for read: %s", file); 1401 goto lose; 1402 } 1403 1404 out = fdopen(dup(zfd), "w"); 1405 if (out == NULL) { 1406 maybe_warn("fdopen for write: %s", outfile); 1407 fclose(in); 1408 goto lose; 1409 } 1410 1411 size = zuncompress(in, out, NULL, 0, NULL); 1412 /* need to fclose() if ferror() is true... */ 1413 if (ferror(in) | fclose(in)) { 1414 maybe_warn("failed infile fclose"); 1415 unlink(outfile); 1416 (void)fclose(out); 1417 } 1418 if (fclose(out) != 0) { 1419 maybe_warn("failed outfile fclose"); 1420 unlink(outfile); 1421 goto lose; 1422 } 1423 } else 1424 #endif 1425 1426 #ifndef SMALL 1427 if (method == FT_UNKNOWN) { 1428 if (lflag) { 1429 maybe_warnx("no -l for unknown filetypes"); 1430 goto lose; 1431 } 1432 size = cat_fd(NULL, 0, NULL, fd); 1433 } else 1434 #endif 1435 { 1436 if (lflag) { 1437 print_list(fd, isb.st_size, outfile, isb.st_mtime); 1438 close(fd); 1439 return -1; /* XXX */ 1440 } 1441 1442 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1443 } 1444 1445 if (close(fd) != 0) 1446 maybe_warn("couldn't close input"); 1447 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1448 maybe_warn("couldn't close output"); 1449 1450 if (size == -1) { 1451 if (cflag == 0) 1452 unlink(outfile); 1453 maybe_warnx("%s: uncompress failed", file); 1454 return -1; 1455 } 1456 1457 /* if testing, or we uncompressed to stdout, this is all we need */ 1458 #ifndef SMALL 1459 if (tflag) 1460 return size; 1461 #endif 1462 /* if we are uncompressing to stdin, don't remove the file. */ 1463 if (cflag) 1464 return size; 1465 1466 /* 1467 * if we create a file... 1468 */ 1469 /* 1470 * if we can't stat the file don't remove the file. 1471 */ 1472 1473 ofd = open(outfile, O_RDWR, 0); 1474 if (ofd == -1) { 1475 maybe_warn("couldn't open (leaving original): %s", 1476 outfile); 1477 return -1; 1478 } 1479 if (fstat(ofd, &osb) != 0) { 1480 maybe_warn("couldn't stat (leaving original): %s", 1481 outfile); 1482 close(ofd); 1483 return -1; 1484 } 1485 if (osb.st_size != size) { 1486 maybe_warnx("stat gave different size: %" PRIdOFF 1487 " != %" PRIdOFF " (leaving original)", 1488 size, osb.st_size); 1489 close(ofd); 1490 unlink(outfile); 1491 return -1; 1492 } 1493 unlink_input(file, &isb); 1494 #ifndef SMALL 1495 copymodes(ofd, &isb, outfile); 1496 #endif 1497 close(ofd); 1498 return size; 1499 1500 unexpected_EOF: 1501 maybe_warnx("%s: unexpected end of file", file); 1502 lose: 1503 if (fd != -1) 1504 close(fd); 1505 if (zfd != -1 && zfd != STDOUT_FILENO) 1506 close(fd); 1507 return -1; 1508 } 1509 1510 #ifndef SMALL 1511 static off_t 1512 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1513 { 1514 char buf[BUFLEN]; 1515 off_t in_tot; 1516 ssize_t w; 1517 1518 in_tot = count; 1519 w = write(STDOUT_FILENO, prepend, count); 1520 if (w == -1 || (size_t)w != count) { 1521 maybe_warn("write to stdout"); 1522 return -1; 1523 } 1524 for (;;) { 1525 ssize_t rv; 1526 1527 rv = read(fd, buf, sizeof buf); 1528 if (rv == 0) 1529 break; 1530 if (rv < 0) { 1531 maybe_warn("read from fd %d", fd); 1532 break; 1533 } 1534 1535 if (write(STDOUT_FILENO, buf, rv) != rv) { 1536 maybe_warn("write to stdout"); 1537 break; 1538 } 1539 in_tot += rv; 1540 } 1541 1542 if (gsizep) 1543 *gsizep = in_tot; 1544 return (in_tot); 1545 } 1546 #endif 1547 1548 static void 1549 handle_stdin(void) 1550 { 1551 unsigned char header1[4]; 1552 off_t usize, gsize; 1553 enum filetype method; 1554 ssize_t bytes_read; 1555 #ifndef NO_COMPRESS_SUPPORT 1556 FILE *in; 1557 #endif 1558 1559 #ifndef SMALL 1560 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1561 maybe_warnx("standard input is a terminal -- ignoring"); 1562 return; 1563 } 1564 #endif 1565 1566 if (lflag) { 1567 struct stat isb; 1568 1569 /* XXX could read the whole file, etc. */ 1570 if (fstat(STDIN_FILENO, &isb) < 0) { 1571 maybe_warn("fstat"); 1572 return; 1573 } 1574 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime); 1575 return; 1576 } 1577 1578 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1579 if (bytes_read == -1) { 1580 maybe_warn("can't read stdin"); 1581 return; 1582 } else if (bytes_read != sizeof(header1)) { 1583 maybe_warnx("(stdin): unexpected end of file"); 1584 return; 1585 } 1586 1587 method = file_gettype(header1); 1588 switch (method) { 1589 default: 1590 #ifndef SMALL 1591 if (fflag == 0) { 1592 maybe_warnx("unknown compression format"); 1593 return; 1594 } 1595 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1596 break; 1597 #endif 1598 case FT_GZIP: 1599 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1600 header1, sizeof header1, &gsize, "(stdin)"); 1601 break; 1602 #ifndef NO_BZIP2_SUPPORT 1603 case FT_BZIP2: 1604 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1605 header1, sizeof header1, &gsize); 1606 break; 1607 #endif 1608 #ifndef NO_COMPRESS_SUPPORT 1609 case FT_Z: 1610 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1611 maybe_warnx("zopen of stdin"); 1612 return; 1613 } 1614 1615 usize = zuncompress(in, stdout, header1, sizeof header1, &gsize); 1616 fclose(in); 1617 break; 1618 #endif 1619 } 1620 1621 #ifndef SMALL 1622 if (vflag && !tflag && usize != -1 && gsize != -1) 1623 print_verbage(NULL, NULL, usize, gsize); 1624 if (vflag && tflag) 1625 print_test("(stdin)", usize != -1); 1626 #endif 1627 1628 } 1629 1630 static void 1631 handle_stdout(void) 1632 { 1633 off_t gsize, usize; 1634 struct stat sb; 1635 time_t systime; 1636 uint32_t mtime; 1637 int ret; 1638 1639 #ifndef SMALL 1640 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1641 maybe_warnx("standard output is a terminal -- ignoring"); 1642 return; 1643 } 1644 #endif 1645 /* If stdin is a file use it's mtime, otherwise use current time */ 1646 ret = fstat(STDIN_FILENO, &sb); 1647 1648 #ifndef SMALL 1649 if (ret < 0) { 1650 maybe_warn("Can't stat stdin"); 1651 return; 1652 } 1653 #endif 1654 1655 if (S_ISREG(sb.st_mode)) 1656 mtime = (uint32_t)sb.st_mtime; 1657 else { 1658 systime = time(NULL); 1659 #ifndef SMALL 1660 if (systime == -1) { 1661 maybe_warn("time"); 1662 return; 1663 } 1664 #endif 1665 mtime = (uint32_t)systime; 1666 } 1667 1668 usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1669 #ifndef SMALL 1670 if (vflag && !tflag && usize != -1 && gsize != -1) 1671 print_verbage(NULL, NULL, usize, gsize); 1672 #endif 1673 } 1674 1675 /* do what is asked for, for the path name */ 1676 static void 1677 handle_pathname(char *path) 1678 { 1679 char *opath = path, *s = NULL; 1680 ssize_t len; 1681 int slen; 1682 struct stat sb; 1683 1684 /* check for stdout/stdin */ 1685 if (path[0] == '-' && path[1] == '\0') { 1686 if (dflag) 1687 handle_stdin(); 1688 else 1689 handle_stdout(); 1690 return; 1691 } 1692 1693 retry: 1694 if (stat(path, &sb) != 0) { 1695 /* lets try <path>.gz if we're decompressing */ 1696 if (dflag && s == NULL && errno == ENOENT) { 1697 len = strlen(path); 1698 slen = suffixes[0].ziplen; 1699 s = malloc(len + slen + 1); 1700 if (s == NULL) 1701 maybe_err("malloc"); 1702 memcpy(s, path, len); 1703 memcpy(s + len, suffixes[0].zipped, slen + 1); 1704 path = s; 1705 goto retry; 1706 } 1707 maybe_warn("can't stat: %s", opath); 1708 goto out; 1709 } 1710 1711 if (S_ISDIR(sb.st_mode)) { 1712 #ifndef SMALL 1713 if (rflag) 1714 handle_dir(path); 1715 else 1716 #endif 1717 maybe_warnx("%s is a directory", path); 1718 goto out; 1719 } 1720 1721 if (S_ISREG(sb.st_mode)) 1722 handle_file(path, &sb); 1723 else 1724 maybe_warnx("%s is not a regular file", path); 1725 1726 out: 1727 if (s) 1728 free(s); 1729 } 1730 1731 /* compress/decompress a file */ 1732 static void 1733 handle_file(char *file, struct stat *sbp) 1734 { 1735 off_t usize, gsize; 1736 char outfile[PATH_MAX]; 1737 1738 infile = file; 1739 if (dflag) { 1740 usize = file_uncompress(file, outfile, sizeof(outfile)); 1741 #ifndef SMALL 1742 if (vflag && tflag) 1743 print_test(file, usize != -1); 1744 #endif 1745 if (usize == -1) 1746 return; 1747 gsize = sbp->st_size; 1748 } else { 1749 gsize = file_compress(file, outfile, sizeof(outfile)); 1750 if (gsize == -1) 1751 return; 1752 usize = sbp->st_size; 1753 } 1754 1755 1756 #ifndef SMALL 1757 if (vflag && !tflag) 1758 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1759 #endif 1760 } 1761 1762 #ifndef SMALL 1763 /* this is used with -r to recursively descend directories */ 1764 static void 1765 handle_dir(char *dir) 1766 { 1767 char *path_argv[2]; 1768 FTS *fts; 1769 FTSENT *entry; 1770 1771 path_argv[0] = dir; 1772 path_argv[1] = 0; 1773 fts = fts_open(path_argv, FTS_PHYSICAL, NULL); 1774 if (fts == NULL) { 1775 warn("couldn't fts_open %s", dir); 1776 return; 1777 } 1778 1779 while ((entry = fts_read(fts))) { 1780 switch(entry->fts_info) { 1781 case FTS_D: 1782 case FTS_DP: 1783 continue; 1784 1785 case FTS_DNR: 1786 case FTS_ERR: 1787 case FTS_NS: 1788 maybe_warn("%s", entry->fts_path); 1789 continue; 1790 case FTS_F: 1791 handle_file(entry->fts_name, entry->fts_statp); 1792 } 1793 } 1794 (void)fts_close(fts); 1795 } 1796 #endif 1797 1798 /* print a ratio - size reduction as a fraction of uncompressed size */ 1799 static void 1800 print_ratio(off_t in, off_t out, FILE *where) 1801 { 1802 int percent10; /* 10 * percent */ 1803 off_t diff; 1804 char buff[8]; 1805 int len; 1806 1807 diff = in - out/2; 1808 if (diff <= 0) 1809 /* 1810 * Output is more than double size of input! print -99.9% 1811 * Quite possibly we've failed to get the original size. 1812 */ 1813 percent10 = -999; 1814 else { 1815 /* 1816 * We only need 12 bits of result from the final division, 1817 * so reduce the values until a 32bit division will suffice. 1818 */ 1819 while (in > 0x100000) { 1820 diff >>= 1; 1821 in >>= 1; 1822 } 1823 if (in != 0) 1824 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 1825 else 1826 percent10 = 0; 1827 } 1828 1829 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 1830 /* Move the '.' to before the last digit */ 1831 buff[len - 1] = buff[len - 2]; 1832 buff[len - 2] = '.'; 1833 fprintf(where, "%5s%%", buff); 1834 } 1835 1836 #ifndef SMALL 1837 /* print compression statistics, and the new name (if there is one!) */ 1838 static void 1839 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 1840 { 1841 if (file) 1842 fprintf(stderr, "%s:%s ", file, 1843 strlen(file) < 7 ? "\t\t" : "\t"); 1844 print_ratio(usize, gsize, stderr); 1845 if (nfile) 1846 fprintf(stderr, " -- replaced with %s", nfile); 1847 fprintf(stderr, "\n"); 1848 fflush(stderr); 1849 } 1850 1851 /* print test results */ 1852 static void 1853 print_test(const char *file, int ok) 1854 { 1855 1856 if (exit_value == 0 && ok == 0) 1857 exit_value = 1; 1858 fprintf(stderr, "%s:%s %s\n", file, 1859 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 1860 fflush(stderr); 1861 } 1862 #endif 1863 1864 /* print a file's info ala --list */ 1865 /* eg: 1866 compressed uncompressed ratio uncompressed_name 1867 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 1868 */ 1869 static void 1870 print_list(int fd, off_t out, const char *outfile, time_t ts) 1871 { 1872 static int first = 1; 1873 #ifndef SMALL 1874 static off_t in_tot, out_tot; 1875 uint32_t crc = 0; 1876 #endif 1877 off_t in = 0, rv; 1878 1879 if (first) { 1880 #ifndef SMALL 1881 if (vflag) 1882 printf("method crc date time "); 1883 #endif 1884 if (qflag == 0) 1885 printf(" compressed uncompressed " 1886 "ratio uncompressed_name\n"); 1887 } 1888 first = 0; 1889 1890 /* print totals? */ 1891 #ifndef SMALL 1892 if (fd == -1) { 1893 in = in_tot; 1894 out = out_tot; 1895 } else 1896 #endif 1897 { 1898 /* read the last 4 bytes - this is the uncompressed size */ 1899 rv = lseek(fd, (off_t)(-8), SEEK_END); 1900 if (rv != -1) { 1901 unsigned char buf[8]; 1902 uint32_t usize; 1903 1904 rv = read(fd, (char *)buf, sizeof(buf)); 1905 if (rv == -1) 1906 maybe_warn("read of uncompressed size"); 1907 else if (rv != sizeof(buf)) 1908 maybe_warnx("read of uncompressed size"); 1909 1910 else { 1911 usize = buf[4] | buf[5] << 8 | 1912 buf[6] << 16 | buf[7] << 24; 1913 in = (off_t)usize; 1914 #ifndef SMALL 1915 crc = buf[0] | buf[1] << 8 | 1916 buf[2] << 16 | buf[3] << 24; 1917 #endif 1918 } 1919 } 1920 } 1921 1922 #ifndef SMALL 1923 if (vflag && fd == -1) 1924 printf(" "); 1925 else if (vflag) { 1926 char *date = ctime(&ts); 1927 1928 /* skip the day, 1/100th second, and year */ 1929 date += 4; 1930 date[12] = 0; 1931 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 1932 } 1933 in_tot += in; 1934 out_tot += out; 1935 #endif 1936 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 1937 print_ratio(in, out, stdout); 1938 printf(" %s\n", outfile); 1939 } 1940 1941 /* display the usage of NetBSD gzip */ 1942 static void 1943 usage(void) 1944 { 1945 1946 fprintf(stderr, "%s\n", gzip_version); 1947 fprintf(stderr, 1948 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n" 1949 #ifndef SMALL 1950 " -1 --fast fastest (worst) compression\n" 1951 " -2 .. -8 set compression level\n" 1952 " -9 --best best (slowest) compression\n" 1953 " -c --stdout write to stdout, keep original files\n" 1954 " --to-stdout\n" 1955 " -d --decompress uncompress files\n" 1956 " --uncompress\n" 1957 " -f --force force overwriting & compress links\n" 1958 " -h --help display this help\n" 1959 " -l --list list compressed file contents\n" 1960 " -N --name save or restore original file name and time stamp\n" 1961 " -n --no-name don't save original file name or time stamp\n" 1962 " -q --quiet output no warnings\n" 1963 " -r --recursive recursively compress files in directories\n" 1964 " -S .suf use suffix .suf instead of .gz\n" 1965 " --suffix .suf\n" 1966 " -t --test test compressed file\n" 1967 " -V --version display program version\n" 1968 " -v --verbose print extra statistics\n", 1969 #else 1970 , 1971 #endif 1972 getprogname()); 1973 exit(0); 1974 } 1975 1976 /* display the version of NetBSD gzip */ 1977 static void 1978 display_version(void) 1979 { 1980 1981 fprintf(stderr, "%s\n", gzip_version); 1982 exit(0); 1983 } 1984 1985 #ifndef NO_BZIP2_SUPPORT 1986 #include "unbzip2.c" 1987 #endif 1988 #ifndef NO_COMPRESS_SUPPORT 1989 #include "zuncompress.c" 1990 #endif 1991 1992 static ssize_t 1993 read_retry(int fd, void *buf, size_t sz) 1994 { 1995 char *cp = buf; 1996 size_t left = MIN(sz, (size_t) SSIZE_MAX); 1997 1998 while (left > 0) { 1999 ssize_t ret; 2000 2001 ret = read(fd, cp, left); 2002 if (ret == -1) { 2003 return ret; 2004 } else if (ret == 0) { 2005 break; /* EOF */ 2006 } 2007 cp += ret; 2008 left -= ret; 2009 } 2010 2011 return sz - left; 2012 } 2013