1 /* $NetBSD: gzip.c,v 1.118 2022/01/22 14:00:45 christos Exp $ */ 2 3 /* 4 * Copyright (c) 1997, 1998, 2003, 2004, 2006, 2008, 2009, 2010, 2011, 2015, 2017 5 * Matthew R. Green 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 18 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 19 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 20 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 21 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 22 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 23 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 24 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 25 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 */ 29 30 #include <sys/cdefs.h> 31 #ifndef lint 32 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006, 2008,\ 33 2009, 2010, 2011, 2015, 2017 Matthew R. Green. All rights reserved."); 34 __RCSID("$NetBSD: gzip.c,v 1.118 2022/01/22 14:00:45 christos Exp $"); 35 #endif /* not lint */ 36 37 /* 38 * gzip.c -- GPL free gzip using zlib. 39 * 40 * RFC 1950 covers the zlib format 41 * RFC 1951 covers the deflate format 42 * RFC 1952 covers the gzip format 43 * 44 * TODO: 45 * - use mmap where possible 46 * - handle some signals better (remove outfile?) 47 * - make bzip2/compress -v/-t/-l support work as well as possible 48 */ 49 50 #include <sys/param.h> 51 #include <sys/stat.h> 52 #include <sys/time.h> 53 54 #include <inttypes.h> 55 #include <unistd.h> 56 #include <stdio.h> 57 #include <string.h> 58 #include <stdlib.h> 59 #include <err.h> 60 #include <errno.h> 61 #include <fcntl.h> 62 #include <zlib.h> 63 #include <fts.h> 64 #include <libgen.h> 65 #include <stdarg.h> 66 #include <getopt.h> 67 #include <time.h> 68 #include <paths.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 #ifndef NO_PACK_SUPPORT 84 FT_PACK, 85 #endif 86 #ifndef NO_XZ_SUPPORT 87 FT_XZ, 88 #endif 89 #ifndef NO_LZ_SUPPORT 90 FT_LZ, 91 #endif 92 FT_LAST, 93 FT_UNKNOWN 94 }; 95 96 #ifndef NO_BZIP2_SUPPORT 97 #include <bzlib.h> 98 99 #define BZ2_SUFFIX ".bz2" 100 #define BZIP2_MAGIC "\102\132\150" 101 #endif 102 103 #ifndef NO_COMPRESS_SUPPORT 104 #define Z_SUFFIX ".Z" 105 #define Z_MAGIC "\037\235" 106 #endif 107 108 #ifndef NO_PACK_SUPPORT 109 #define PACK_MAGIC "\037\036" 110 #endif 111 112 #ifndef NO_XZ_SUPPORT 113 #include <lzma.h> 114 #define XZ_SUFFIX ".xz" 115 #define XZ_MAGIC "\3757zXZ" 116 #endif 117 118 #ifndef NO_LZ_SUPPORT 119 #define LZ_SUFFIX ".lz" 120 #define LZ_MAGIC "LZIP" 121 #endif 122 123 #define GZ_SUFFIX ".gz" 124 125 #define BUFLEN (64 * 1024) 126 127 #define GZIP_MAGIC0 0x1F 128 #define GZIP_MAGIC1 0x8B 129 #define GZIP_OMAGIC1 0x9E 130 131 #define GZIP_TIMESTAMP (off_t)4 132 #define GZIP_ORIGNAME (off_t)10 133 134 #define HEAD_CRC 0x02 135 #define EXTRA_FIELD 0x04 136 #define ORIG_NAME 0x08 137 #define COMMENT 0x10 138 139 #define OS_CODE 3 /* Unix */ 140 141 typedef struct { 142 const char *zipped; 143 int ziplen; 144 const char *normal; /* for unzip - must not be longer than zipped */ 145 } suffixes_t; 146 static suffixes_t suffixes[] = { 147 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 148 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 149 #ifndef SMALL 150 SUFFIX(GZ_SUFFIX, ""), 151 SUFFIX(".z", ""), 152 SUFFIX("-gz", ""), 153 SUFFIX("-z", ""), 154 SUFFIX("_z", ""), 155 SUFFIX(".taz", ".tar"), 156 SUFFIX(".tgz", ".tar"), 157 #ifndef NO_BZIP2_SUPPORT 158 SUFFIX(BZ2_SUFFIX, ""), 159 #endif 160 #ifndef NO_COMPRESS_SUPPORT 161 SUFFIX(Z_SUFFIX, ""), 162 #endif 163 #ifndef NO_XZ_SUPPORT 164 SUFFIX(XZ_SUFFIX, ""), 165 #endif 166 #ifndef NO_LZ_SUPPORT 167 SUFFIX(LZ_SUFFIX, ""), 168 #endif 169 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 170 #endif /* SMALL */ 171 #undef SUFFIX 172 }; 173 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0]) 174 #define SUFFIX_MAXLEN 30 175 176 static const char gzip_version[] = "NetBSD gzip 20170803"; 177 178 static int cflag; /* stdout mode */ 179 static int dflag; /* decompress mode */ 180 static int lflag; /* list mode */ 181 static int numflag = 6; /* gzip -1..-9 value */ 182 183 #ifndef SMALL 184 static int fflag; /* force mode */ 185 static int kflag; /* don't delete input files */ 186 static int nflag; /* don't save name/timestamp */ 187 static int Nflag; /* don't restore name/timestamp */ 188 static int qflag; /* quiet mode */ 189 static int rflag; /* recursive mode */ 190 static int tflag; /* test */ 191 static int vflag; /* verbose mode */ 192 static sig_atomic_t print_info = 0; 193 #else 194 #define qflag 0 195 #define tflag 0 196 #endif 197 198 static int exit_value = 0; /* exit value */ 199 200 static const char *infile; /* name of file coming in */ 201 202 static void maybe_err(const char *fmt, ...) __printflike(1, 2) __dead; 203 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \ 204 !defined(NO_XZ_SUPPORT) 205 static void maybe_errx(const char *fmt, ...) __printflike(1, 2) __dead; 206 #endif 207 static void maybe_warn(const char *fmt, ...) __printflike(1, 2); 208 static void maybe_warnx(const char *fmt, ...) __printflike(1, 2); 209 static enum filetype file_gettype(u_char *); 210 #ifdef SMALL 211 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz) 212 #endif 213 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 214 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 215 static off_t file_compress(char *, char *, size_t); 216 static off_t file_uncompress(char *, char *, size_t); 217 static void handle_pathname(char *); 218 static void handle_file(char *, struct stat *); 219 static void handle_stdin(void); 220 static void handle_stdout(void); 221 static void print_ratio(off_t, off_t, FILE *); 222 static void print_list(int fd, off_t, const char *, time_t); 223 __dead static void usage(void); 224 __dead static void display_version(void); 225 static const suffixes_t *check_suffix(char *, int); 226 static ssize_t read_retry(int, void *, size_t); 227 static ssize_t write_retry(int, const void *, size_t); 228 static void print_list_out(off_t, off_t, const char*); 229 230 #ifdef SMALL 231 #define infile_set(f,t) infile_set(f) 232 #endif 233 static void infile_set(const char *newinfile, off_t total); 234 235 #ifdef SMALL 236 #define unlink_input(f, sb) unlink(f) 237 #define check_siginfo() /* nothing */ 238 #define setup_signals() /* nothing */ 239 #define infile_newdata(t) /* nothing */ 240 #else 241 static off_t infile_total; /* total expected to read/write */ 242 static off_t infile_current; /* current read/write */ 243 244 static void check_siginfo(void); 245 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 246 static void prepend_gzip(char *, int *, char ***); 247 static void handle_dir(char *); 248 static void print_verbage(const char *, const char *, off_t, off_t); 249 static void print_test(const char *, int); 250 static void copymodes(int fd, const struct stat *, const char *file); 251 static int check_outfile(const char *outfile); 252 static void setup_signals(void); 253 static void infile_newdata(size_t newdata); 254 static void infile_clear(void); 255 #endif 256 257 #ifndef NO_BZIP2_SUPPORT 258 static off_t unbzip2(int, int, char *, size_t, off_t *); 259 #endif 260 261 #ifndef NO_COMPRESS_SUPPORT 262 static FILE *zdopen(int); 263 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 264 #endif 265 266 #ifndef NO_PACK_SUPPORT 267 static off_t unpack(int, int, char *, size_t, off_t *); 268 #endif 269 270 #ifndef NO_XZ_SUPPORT 271 static off_t unxz(int, int, char *, size_t, off_t *); 272 static off_t unxz_len(int); 273 #endif 274 275 #ifndef NO_LZ_SUPPORT 276 static off_t unlz(int, int, char *, size_t, off_t *); 277 #endif 278 279 #ifdef SMALL 280 #define getopt_long(a,b,c,d,e) getopt(a,b,c) 281 #else 282 static const struct option longopts[] = { 283 { "stdout", no_argument, 0, 'c' }, 284 { "to-stdout", no_argument, 0, 'c' }, 285 { "decompress", no_argument, 0, 'd' }, 286 { "uncompress", no_argument, 0, 'd' }, 287 { "force", no_argument, 0, 'f' }, 288 { "help", no_argument, 0, 'h' }, 289 { "keep", no_argument, 0, 'k' }, 290 { "list", no_argument, 0, 'l' }, 291 { "no-name", no_argument, 0, 'n' }, 292 { "name", no_argument, 0, 'N' }, 293 { "quiet", no_argument, 0, 'q' }, 294 { "recursive", no_argument, 0, 'r' }, 295 { "suffix", required_argument, 0, 'S' }, 296 { "test", no_argument, 0, 't' }, 297 { "verbose", no_argument, 0, 'v' }, 298 { "version", no_argument, 0, 'V' }, 299 { "fast", no_argument, 0, '1' }, 300 { "best", no_argument, 0, '9' }, 301 #if 0 302 /* 303 * This is what else GNU gzip implements. --ascii isn't useful 304 * on NetBSD, and I don't care to have a --license. 305 */ 306 { "ascii", no_argument, 0, 'a' }, 307 { "license", no_argument, 0, 'L' }, 308 #endif 309 { NULL, no_argument, 0, 0 }, 310 }; 311 #endif 312 313 int 314 main(int argc, char **argv) 315 { 316 const char *progname = getprogname(); 317 #ifndef SMALL 318 char *gzip; 319 int len; 320 #endif 321 int ch; 322 323 setup_signals(); 324 325 #ifndef SMALL 326 if ((gzip = getenv("GZIP")) != NULL) 327 prepend_gzip(gzip, &argc, &argv); 328 #endif 329 330 /* 331 * XXX 332 * handle being called `gunzip', `zcat' and `gzcat' 333 */ 334 if (strcmp(progname, "gunzip") == 0) 335 dflag = 1; 336 else if (strcmp(progname, "zcat") == 0 || 337 strcmp(progname, "gzcat") == 0) 338 dflag = cflag = 1; 339 340 #ifdef SMALL 341 #define OPT_LIST "123456789cdhlV" 342 #else 343 #define OPT_LIST "123456789cdfhklNnqrS:tVv" 344 #endif 345 346 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 347 switch (ch) { 348 case '1': case '2': case '3': 349 case '4': case '5': case '6': 350 case '7': case '8': case '9': 351 numflag = ch - '0'; 352 break; 353 case 'c': 354 cflag = 1; 355 break; 356 case 'd': 357 dflag = 1; 358 break; 359 case 'l': 360 lflag = 1; 361 dflag = 1; 362 break; 363 case 'V': 364 display_version(); 365 /* NOTREACHED */ 366 #ifndef SMALL 367 case 'f': 368 fflag = 1; 369 break; 370 case 'k': 371 kflag = 1; 372 break; 373 case 'N': 374 nflag = 0; 375 Nflag = 1; 376 break; 377 case 'n': 378 nflag = 1; 379 Nflag = 0; 380 break; 381 case 'q': 382 qflag = 1; 383 break; 384 case 'r': 385 rflag = 1; 386 break; 387 case 'S': 388 len = strlen(optarg); 389 if (len != 0) { 390 if (len > SUFFIX_MAXLEN) 391 errx(1, "incorrect suffix: '%s'", optarg); 392 suffixes[0].zipped = optarg; 393 suffixes[0].ziplen = len; 394 } else { 395 suffixes[NUM_SUFFIXES - 1].zipped = ""; 396 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 397 } 398 break; 399 case 't': 400 cflag = 1; 401 tflag = 1; 402 dflag = 1; 403 break; 404 case 'v': 405 vflag = 1; 406 break; 407 #endif 408 default: 409 usage(); 410 /* NOTREACHED */ 411 } 412 } 413 argv += optind; 414 argc -= optind; 415 416 if (argc == 0) { 417 if (dflag) /* stdin mode */ 418 handle_stdin(); 419 else /* stdout mode */ 420 handle_stdout(); 421 } else { 422 do { 423 handle_pathname(argv[0]); 424 } while (*++argv); 425 } 426 #ifndef SMALL 427 if (qflag == 0 && lflag && argc > 1) 428 print_list(-1, 0, "(totals)", 0); 429 #endif 430 exit(exit_value); 431 } 432 433 /* maybe print a warning */ 434 void 435 maybe_warn(const char *fmt, ...) 436 { 437 va_list ap; 438 439 if (qflag == 0) { 440 va_start(ap, fmt); 441 vwarn(fmt, ap); 442 va_end(ap); 443 } 444 if (exit_value == 0) 445 exit_value = 1; 446 } 447 448 /* ... without an errno. */ 449 void 450 maybe_warnx(const char *fmt, ...) 451 { 452 va_list ap; 453 454 if (qflag == 0) { 455 va_start(ap, fmt); 456 vwarnx(fmt, ap); 457 va_end(ap); 458 } 459 if (exit_value == 0) 460 exit_value = 1; 461 } 462 463 /* maybe print an error */ 464 void 465 maybe_err(const char *fmt, ...) 466 { 467 va_list ap; 468 469 if (qflag == 0) { 470 va_start(ap, fmt); 471 vwarn(fmt, ap); 472 va_end(ap); 473 } 474 exit(2); 475 } 476 477 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \ 478 !defined(NO_XZ_SUPPORT) 479 /* ... without an errno. */ 480 void 481 maybe_errx(const char *fmt, ...) 482 { 483 va_list ap; 484 485 if (qflag == 0) { 486 va_start(ap, fmt); 487 vwarnx(fmt, ap); 488 va_end(ap); 489 } 490 exit(2); 491 } 492 #endif 493 494 #ifndef SMALL 495 /* split up $GZIP and prepend it to the argument list */ 496 static void 497 prepend_gzip(char *gzip, int *argc, char ***argv) 498 { 499 char *s, **nargv, **ac; 500 int nenvarg = 0, i; 501 502 /* scan how many arguments there are */ 503 for (s = gzip;;) { 504 while (*s == ' ' || *s == '\t') 505 s++; 506 if (*s == 0) 507 goto count_done; 508 nenvarg++; 509 while (*s != ' ' && *s != '\t') 510 if (*s++ == 0) 511 goto count_done; 512 } 513 count_done: 514 /* punt early */ 515 if (nenvarg == 0) 516 return; 517 518 *argc += nenvarg; 519 ac = *argv; 520 521 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 522 if (nargv == NULL) 523 maybe_err("malloc"); 524 525 /* stash this away */ 526 *argv = nargv; 527 528 /* copy the program name first */ 529 i = 0; 530 nargv[i++] = *(ac++); 531 532 /* take a copy of $GZIP and add it to the array */ 533 s = strdup(gzip); 534 if (s == NULL) 535 maybe_err("strdup"); 536 for (;;) { 537 /* Skip whitespaces. */ 538 while (*s == ' ' || *s == '\t') 539 s++; 540 if (*s == 0) 541 goto copy_done; 542 nargv[i++] = s; 543 /* Find the end of this argument. */ 544 while (*s != ' ' && *s != '\t') 545 if (*s++ == 0) 546 /* Argument followed by NUL. */ 547 goto copy_done; 548 /* Terminate by overwriting ' ' or '\t' with NUL. */ 549 *s++ = 0; 550 } 551 copy_done: 552 553 /* copy the original arguments and a NULL */ 554 while (*ac) 555 nargv[i++] = *(ac++); 556 nargv[i] = NULL; 557 } 558 #endif 559 560 /* compress input to output. Return bytes read, -1 on error */ 561 static off_t 562 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 563 { 564 z_stream z; 565 char *outbufp, *inbufp; 566 off_t in_tot = 0, out_tot = 0; 567 ssize_t in_size; 568 int i, error; 569 uLong crc; 570 #ifdef SMALL 571 static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0, 572 0, 0, 0, 0, 573 0, OS_CODE }; 574 #endif 575 576 outbufp = malloc(BUFLEN); 577 inbufp = malloc(BUFLEN); 578 if (outbufp == NULL || inbufp == NULL) { 579 maybe_err("malloc failed"); 580 goto out; 581 } 582 583 memset(&z, 0, sizeof z); 584 z.zalloc = Z_NULL; 585 z.zfree = Z_NULL; 586 z.opaque = 0; 587 588 #ifdef SMALL 589 memcpy(outbufp, header, sizeof header); 590 i = sizeof header; 591 #else 592 if (nflag != 0) { 593 mtime = 0; 594 origname = ""; 595 } 596 597 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 598 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 599 *origname ? ORIG_NAME : 0, 600 mtime & 0xff, 601 (mtime >> 8) & 0xff, 602 (mtime >> 16) & 0xff, 603 (mtime >> 24) & 0xff, 604 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 605 OS_CODE, origname); 606 if (i >= BUFLEN) 607 /* this need PATH_MAX > BUFLEN ... */ 608 maybe_err("snprintf"); 609 if (*origname) 610 i++; 611 #endif 612 613 z.next_out = (unsigned char *)outbufp + i; 614 z.avail_out = BUFLEN - i; 615 616 error = deflateInit2(&z, numflag, Z_DEFLATED, 617 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 618 if (error != Z_OK) { 619 maybe_warnx("deflateInit2 failed"); 620 in_tot = -1; 621 goto out; 622 } 623 624 crc = crc32(0L, Z_NULL, 0); 625 for (;;) { 626 check_siginfo(); 627 if (z.avail_out == 0) { 628 if (write_retry(out, outbufp, BUFLEN) != BUFLEN) { 629 maybe_warn("write"); 630 out_tot = -1; 631 goto out; 632 } 633 634 out_tot += BUFLEN; 635 z.next_out = (unsigned char *)outbufp; 636 z.avail_out = BUFLEN; 637 } 638 639 if (z.avail_in == 0) { 640 in_size = read(in, inbufp, BUFLEN); 641 if (in_size < 0) { 642 maybe_warn("read"); 643 in_tot = -1; 644 goto out; 645 } 646 if (in_size == 0) 647 break; 648 infile_newdata(in_size); 649 650 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 651 in_tot += in_size; 652 z.next_in = (unsigned char *)inbufp; 653 z.avail_in = in_size; 654 } 655 656 error = deflate(&z, Z_NO_FLUSH); 657 if (error != Z_OK && error != Z_STREAM_END) { 658 maybe_warnx("deflate failed"); 659 in_tot = -1; 660 goto out; 661 } 662 } 663 664 /* clean up */ 665 for (;;) { 666 size_t len; 667 ssize_t w; 668 669 error = deflate(&z, Z_FINISH); 670 if (error != Z_OK && error != Z_STREAM_END) { 671 maybe_warnx("deflate failed"); 672 in_tot = -1; 673 goto out; 674 } 675 676 len = (char *)z.next_out - outbufp; 677 678 w = write_retry(out, outbufp, len); 679 if (w == -1 || (size_t)w != len) { 680 maybe_warn("write"); 681 out_tot = -1; 682 goto out; 683 } 684 out_tot += len; 685 z.next_out = (unsigned char *)outbufp; 686 z.avail_out = BUFLEN; 687 688 if (error == Z_STREAM_END) 689 break; 690 } 691 692 if (deflateEnd(&z) != Z_OK) { 693 maybe_warnx("deflateEnd failed"); 694 in_tot = -1; 695 goto out; 696 } 697 698 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 699 (int)crc & 0xff, 700 (int)(crc >> 8) & 0xff, 701 (int)(crc >> 16) & 0xff, 702 (int)(crc >> 24) & 0xff, 703 (int)in_tot & 0xff, 704 (int)(in_tot >> 8) & 0xff, 705 (int)(in_tot >> 16) & 0xff, 706 (int)(in_tot >> 24) & 0xff); 707 if (i != 8) 708 maybe_err("snprintf"); 709 #if 0 710 if (in_tot > 0xffffffff) 711 maybe_warn("input file size >= 4GB cannot be saved"); 712 #endif 713 if (write_retry(out, outbufp, i) != i) { 714 maybe_warn("write"); 715 in_tot = -1; 716 } else 717 out_tot += i; 718 719 out: 720 if (inbufp != NULL) 721 free(inbufp); 722 if (outbufp != NULL) 723 free(outbufp); 724 if (gsizep) 725 *gsizep = out_tot; 726 return in_tot; 727 } 728 729 /* 730 * uncompress input to output then close the input. return the 731 * uncompressed size written, and put the compressed sized read 732 * into `*gsizep'. 733 */ 734 static off_t 735 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 736 const char *filename) 737 { 738 z_stream z; 739 char *outbufp, *inbufp; 740 off_t out_tot = -1, in_tot = 0; 741 uint32_t out_sub_tot = 0; 742 enum { 743 GZSTATE_MAGIC0, 744 GZSTATE_MAGIC1, 745 GZSTATE_METHOD, 746 GZSTATE_FLAGS, 747 GZSTATE_SKIPPING, 748 GZSTATE_EXTRA, 749 GZSTATE_EXTRA2, 750 GZSTATE_EXTRA3, 751 GZSTATE_ORIGNAME, 752 GZSTATE_COMMENT, 753 GZSTATE_HEAD_CRC1, 754 GZSTATE_HEAD_CRC2, 755 GZSTATE_INIT, 756 GZSTATE_READ, 757 GZSTATE_CRC, 758 GZSTATE_LEN, 759 } state = GZSTATE_MAGIC0; 760 int flags = 0, skip_count = 0; 761 int error = Z_STREAM_ERROR, done_reading = 0; 762 uLong crc = 0; 763 ssize_t wr; 764 int needmore = 0; 765 766 #define ADVANCE() { z.next_in++; z.avail_in--; } 767 768 if ((outbufp = malloc(BUFLEN)) == NULL) { 769 maybe_err("malloc failed"); 770 goto out2; 771 } 772 if ((inbufp = malloc(BUFLEN)) == NULL) { 773 maybe_err("malloc failed"); 774 goto out1; 775 } 776 777 memset(&z, 0, sizeof z); 778 z.avail_in = prelen; 779 z.next_in = (unsigned char *)pre; 780 z.avail_out = BUFLEN; 781 z.next_out = (unsigned char *)outbufp; 782 z.zalloc = NULL; 783 z.zfree = NULL; 784 z.opaque = 0; 785 786 in_tot = prelen; 787 out_tot = 0; 788 789 for (;;) { 790 check_siginfo(); 791 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 792 ssize_t in_size; 793 794 if (z.avail_in > 0) { 795 memmove(inbufp, z.next_in, z.avail_in); 796 } 797 z.next_in = (unsigned char *)inbufp; 798 in_size = read(in, z.next_in + z.avail_in, 799 BUFLEN - z.avail_in); 800 801 if (in_size == -1) { 802 maybe_warn("failed to read stdin"); 803 goto stop_and_fail; 804 } else if (in_size == 0) { 805 done_reading = 1; 806 } 807 infile_newdata(in_size); 808 809 z.avail_in += in_size; 810 needmore = 0; 811 812 in_tot += in_size; 813 } 814 if (z.avail_in == 0) { 815 if (done_reading && state != GZSTATE_MAGIC0) { 816 maybe_warnx("%s: unexpected end of file", 817 filename); 818 goto stop_and_fail; 819 } 820 goto stop; 821 } 822 switch (state) { 823 case GZSTATE_MAGIC0: 824 if (*z.next_in != GZIP_MAGIC0) { 825 if (in_tot > 0) { 826 maybe_warnx("%s: trailing garbage " 827 "ignored", filename); 828 goto stop; 829 } 830 maybe_warnx("input not gziped (MAGIC0)"); 831 exit_value = 2; 832 goto stop_and_fail; 833 } 834 ADVANCE(); 835 state++; 836 out_sub_tot = 0; 837 crc = crc32(0L, Z_NULL, 0); 838 break; 839 840 case GZSTATE_MAGIC1: 841 if (*z.next_in != GZIP_MAGIC1 && 842 *z.next_in != GZIP_OMAGIC1) { 843 maybe_warnx("input not gziped (MAGIC1)"); 844 goto stop_and_fail; 845 } 846 ADVANCE(); 847 state++; 848 break; 849 850 case GZSTATE_METHOD: 851 if (*z.next_in != Z_DEFLATED) { 852 maybe_warnx("unknown compression method"); 853 goto stop_and_fail; 854 } 855 ADVANCE(); 856 state++; 857 break; 858 859 case GZSTATE_FLAGS: 860 flags = *z.next_in; 861 ADVANCE(); 862 skip_count = 6; 863 state++; 864 break; 865 866 case GZSTATE_SKIPPING: 867 if (skip_count > 0) { 868 skip_count--; 869 ADVANCE(); 870 } else 871 state++; 872 break; 873 874 case GZSTATE_EXTRA: 875 if ((flags & EXTRA_FIELD) == 0) { 876 state = GZSTATE_ORIGNAME; 877 break; 878 } 879 skip_count = *z.next_in; 880 ADVANCE(); 881 state++; 882 break; 883 884 case GZSTATE_EXTRA2: 885 skip_count |= ((*z.next_in) << 8); 886 ADVANCE(); 887 state++; 888 break; 889 890 case GZSTATE_EXTRA3: 891 if (skip_count > 0) { 892 skip_count--; 893 ADVANCE(); 894 } else 895 state++; 896 break; 897 898 case GZSTATE_ORIGNAME: 899 if ((flags & ORIG_NAME) == 0) { 900 state++; 901 break; 902 } 903 if (*z.next_in == 0) 904 state++; 905 ADVANCE(); 906 break; 907 908 case GZSTATE_COMMENT: 909 if ((flags & COMMENT) == 0) { 910 state++; 911 break; 912 } 913 if (*z.next_in == 0) 914 state++; 915 ADVANCE(); 916 break; 917 918 case GZSTATE_HEAD_CRC1: 919 if (flags & HEAD_CRC) 920 skip_count = 2; 921 else 922 skip_count = 0; 923 state++; 924 break; 925 926 case GZSTATE_HEAD_CRC2: 927 if (skip_count > 0) { 928 skip_count--; 929 ADVANCE(); 930 } else 931 state++; 932 break; 933 934 case GZSTATE_INIT: 935 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 936 maybe_warnx("failed to inflateInit"); 937 goto stop_and_fail; 938 } 939 state++; 940 break; 941 942 case GZSTATE_READ: 943 error = inflate(&z, Z_FINISH); 944 switch (error) { 945 /* Z_BUF_ERROR goes with Z_FINISH... */ 946 case Z_BUF_ERROR: 947 if (z.avail_out > 0 && !done_reading) 948 continue; 949 950 case Z_STREAM_END: 951 case Z_OK: 952 break; 953 954 case Z_NEED_DICT: 955 maybe_warnx("Z_NEED_DICT error"); 956 goto stop_and_fail; 957 case Z_DATA_ERROR: 958 maybe_warnx("data stream error"); 959 goto stop_and_fail; 960 case Z_STREAM_ERROR: 961 maybe_warnx("internal stream error"); 962 goto stop_and_fail; 963 case Z_MEM_ERROR: 964 maybe_warnx("memory allocation error"); 965 goto stop_and_fail; 966 967 default: 968 maybe_warn("unknown error from inflate(): %d", 969 error); 970 } 971 wr = BUFLEN - z.avail_out; 972 973 if (wr != 0) { 974 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 975 if ( 976 #ifndef SMALL 977 /* don't write anything with -t */ 978 tflag == 0 && 979 #endif 980 write_retry(out, outbufp, wr) != wr) { 981 maybe_warn("error writing to output"); 982 goto stop_and_fail; 983 } 984 985 out_tot += wr; 986 out_sub_tot += wr; 987 } 988 989 if (error == Z_STREAM_END) { 990 inflateEnd(&z); 991 state++; 992 } 993 994 z.next_out = (unsigned char *)outbufp; 995 z.avail_out = BUFLEN; 996 997 break; 998 case GZSTATE_CRC: 999 { 1000 uLong origcrc; 1001 1002 if (z.avail_in < 4) { 1003 if (!done_reading) { 1004 needmore = 1; 1005 continue; 1006 } 1007 maybe_warnx("truncated input"); 1008 goto stop_and_fail; 1009 } 1010 origcrc = ((unsigned)z.next_in[0] & 0xff) | 1011 ((unsigned)z.next_in[1] & 0xff) << 8 | 1012 ((unsigned)z.next_in[2] & 0xff) << 16 | 1013 ((unsigned)z.next_in[3] & 0xff) << 24; 1014 if (origcrc != crc) { 1015 maybe_warnx("invalid compressed" 1016 " data--crc error"); 1017 goto stop_and_fail; 1018 } 1019 } 1020 1021 z.avail_in -= 4; 1022 z.next_in += 4; 1023 1024 if (!z.avail_in && done_reading) { 1025 goto stop; 1026 } 1027 state++; 1028 break; 1029 case GZSTATE_LEN: 1030 { 1031 uLong origlen; 1032 1033 if (z.avail_in < 4) { 1034 if (!done_reading) { 1035 needmore = 1; 1036 continue; 1037 } 1038 maybe_warnx("truncated input"); 1039 goto stop_and_fail; 1040 } 1041 origlen = ((unsigned)z.next_in[0] & 0xff) | 1042 ((unsigned)z.next_in[1] & 0xff) << 8 | 1043 ((unsigned)z.next_in[2] & 0xff) << 16 | 1044 ((unsigned)z.next_in[3] & 0xff) << 24; 1045 1046 if (origlen != out_sub_tot) { 1047 maybe_warnx("invalid compressed" 1048 " data--length error"); 1049 goto stop_and_fail; 1050 } 1051 } 1052 1053 z.avail_in -= 4; 1054 z.next_in += 4; 1055 1056 if (error < 0) { 1057 maybe_warnx("decompression error"); 1058 goto stop_and_fail; 1059 } 1060 state = GZSTATE_MAGIC0; 1061 break; 1062 } 1063 continue; 1064 stop_and_fail: 1065 out_tot = -1; 1066 stop: 1067 break; 1068 } 1069 if (state > GZSTATE_INIT) 1070 inflateEnd(&z); 1071 1072 free(inbufp); 1073 out1: 1074 free(outbufp); 1075 out2: 1076 if (gsizep) 1077 *gsizep = in_tot; 1078 return (out_tot); 1079 } 1080 1081 #ifndef SMALL 1082 /* 1083 * set the owner, mode, flags & utimes using the given file descriptor. 1084 * file is only used in possible warning messages. 1085 */ 1086 static void 1087 copymodes(int fd, const struct stat *sbp, const char *file) 1088 { 1089 struct timeval times[2]; 1090 struct stat sb; 1091 1092 /* 1093 * If we have no info on the input, give this file some 1094 * default values and return.. 1095 */ 1096 if (sbp == NULL) { 1097 mode_t mask = umask(022); 1098 1099 (void)fchmod(fd, DEFFILEMODE & ~mask); 1100 (void)umask(mask); 1101 return; 1102 } 1103 sb = *sbp; 1104 1105 /* if the chown fails, remove set-id bits as-per compress(1) */ 1106 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1107 if (errno != EPERM) 1108 maybe_warn("couldn't fchown: %s", file); 1109 sb.st_mode &= ~(S_ISUID|S_ISGID); 1110 } 1111 1112 /* we only allow set-id and the 9 normal permission bits */ 1113 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1114 if (fchmod(fd, sb.st_mode) < 0) 1115 maybe_warn("couldn't fchmod: %s", file); 1116 1117 /* only try flags if they exist already */ 1118 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1119 maybe_warn("couldn't fchflags: %s", file); 1120 1121 TIMESPEC_TO_TIMEVAL(×[0], &sb.st_atimespec); 1122 TIMESPEC_TO_TIMEVAL(×[1], &sb.st_mtimespec); 1123 if (futimes(fd, times) < 0) 1124 maybe_warn("couldn't utimes: %s", file); 1125 } 1126 #endif 1127 1128 /* what sort of file is this? */ 1129 static enum filetype 1130 file_gettype(u_char *buf) 1131 { 1132 1133 if (buf[0] == GZIP_MAGIC0 && 1134 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1135 return FT_GZIP; 1136 else 1137 #ifndef NO_BZIP2_SUPPORT 1138 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1139 buf[3] >= '0' && buf[3] <= '9') 1140 return FT_BZIP2; 1141 else 1142 #endif 1143 #ifndef NO_COMPRESS_SUPPORT 1144 if (memcmp(buf, Z_MAGIC, 2) == 0) 1145 return FT_Z; 1146 else 1147 #endif 1148 #ifndef NO_PACK_SUPPORT 1149 if (memcmp(buf, PACK_MAGIC, 2) == 0) 1150 return FT_PACK; 1151 else 1152 #endif 1153 #ifndef NO_XZ_SUPPORT 1154 if (memcmp(buf, XZ_MAGIC, 4) == 0) /* XXX: We only have 4 bytes */ 1155 return FT_XZ; 1156 else 1157 #endif 1158 #ifndef NO_LZ_SUPPORT 1159 if (memcmp(buf, LZ_MAGIC, 4) == 0) 1160 return FT_LZ; 1161 else 1162 #endif 1163 return FT_UNKNOWN; 1164 } 1165 1166 #ifndef SMALL 1167 /* check the outfile is OK. */ 1168 static int 1169 check_outfile(const char *outfile) 1170 { 1171 struct stat sb; 1172 int ok = 1; 1173 1174 if (lflag == 0 && stat(outfile, &sb) == 0) { 1175 if (fflag) 1176 unlink(outfile); 1177 else if (isatty(STDIN_FILENO)) { 1178 char ans[10] = { 'n', '\0' }; /* default */ 1179 1180 fprintf(stderr, "%s already exists -- do you wish to " 1181 "overwrite (y or n)? " , outfile); 1182 (void)fgets(ans, sizeof(ans) - 1, stdin); 1183 if (ans[0] != 'y' && ans[0] != 'Y') { 1184 fprintf(stderr, "\tnot overwriting\n"); 1185 ok = 0; 1186 } else 1187 unlink(outfile); 1188 } else { 1189 maybe_warnx("%s already exists -- skipping", outfile); 1190 ok = 0; 1191 } 1192 } 1193 return ok; 1194 } 1195 1196 static void 1197 unlink_input(const char *file, const struct stat *sb) 1198 { 1199 struct stat nsb; 1200 1201 if (kflag) 1202 return; 1203 if (stat(file, &nsb) != 0) 1204 /* Must be gone already */ 1205 return; 1206 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1207 /* Definitely a different file */ 1208 return; 1209 unlink(file); 1210 } 1211 1212 static void 1213 got_siginfo(int signo) 1214 { 1215 1216 print_info = 1; 1217 } 1218 1219 static void 1220 setup_signals(void) 1221 { 1222 1223 signal(SIGINFO, got_siginfo); 1224 } 1225 1226 static void 1227 infile_newdata(size_t newdata) 1228 { 1229 1230 infile_current += newdata; 1231 } 1232 #endif 1233 1234 static void 1235 infile_set(const char *newinfile, off_t total) 1236 { 1237 1238 if (newinfile) 1239 infile = newinfile; 1240 #ifndef SMALL 1241 infile_total = total; 1242 #endif 1243 } 1244 1245 static void 1246 infile_clear(void) 1247 { 1248 1249 infile = NULL; 1250 #ifndef SMALL 1251 infile_total = infile_current = 0; 1252 #endif 1253 } 1254 1255 static const suffixes_t * 1256 check_suffix(char *file, int xlate) 1257 { 1258 const suffixes_t *s; 1259 int len = strlen(file); 1260 char *sp; 1261 1262 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1263 /* if it doesn't fit in "a.suf", don't bother */ 1264 if (s->ziplen >= len) 1265 continue; 1266 sp = file + len - s->ziplen; 1267 if (strcmp(s->zipped, sp) != 0) 1268 continue; 1269 if (xlate) 1270 strcpy(sp, s->normal); 1271 return s; 1272 } 1273 return NULL; 1274 } 1275 1276 /* 1277 * compress the given file: create a corresponding .gz file and remove the 1278 * original. 1279 */ 1280 static off_t 1281 file_compress(char *file, char *outfile, size_t outsize) 1282 { 1283 int in; 1284 int out; 1285 off_t size, in_size; 1286 #ifndef SMALL 1287 struct stat isb, osb; 1288 const suffixes_t *suff; 1289 #endif 1290 1291 in = open(file, O_RDONLY); 1292 if (in == -1) { 1293 maybe_warn("can't open %s", file); 1294 return -1; 1295 } 1296 1297 #ifndef SMALL 1298 if (fstat(in, &isb) != 0) { 1299 close(in); 1300 maybe_warn("can't stat %s", file); 1301 return -1; 1302 } 1303 infile_set(file, isb.st_size); 1304 #endif 1305 1306 if (cflag == 0) { 1307 #ifndef SMALL 1308 if (isb.st_nlink > 1 && fflag == 0) { 1309 maybe_warnx("%s has %d other link%s -- " 1310 "skipping", file, isb.st_nlink - 1, 1311 isb.st_nlink == 1 ? "" : "s"); 1312 close(in); 1313 return -1; 1314 } 1315 1316 if (fflag == 0 && (suff = check_suffix(file, 0)) 1317 && suff->zipped[0] != 0) { 1318 maybe_warnx("%s already has %s suffix -- unchanged", 1319 file, suff->zipped); 1320 close(in); 1321 return -1; 1322 } 1323 #endif 1324 1325 /* Add (usually) .gz to filename */ 1326 if ((size_t)snprintf(outfile, outsize, "%s%s", 1327 file, suffixes[0].zipped) >= outsize) 1328 memcpy(outfile + outsize - suffixes[0].ziplen - 1, 1329 suffixes[0].zipped, suffixes[0].ziplen + 1); 1330 1331 #ifndef SMALL 1332 if (check_outfile(outfile) == 0) { 1333 close(in); 1334 return -1; 1335 } 1336 #endif 1337 } 1338 1339 if (cflag == 0) { 1340 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1341 if (out == -1) { 1342 maybe_warn("could not create output: %s", outfile); 1343 fclose(stdin); 1344 return -1; 1345 } 1346 } else 1347 out = STDOUT_FILENO; 1348 1349 in_size = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1350 1351 (void)close(in); 1352 1353 /* 1354 * If there was an error, in_size will be -1. 1355 * If we compressed to stdout, just return the size. 1356 * Otherwise stat the file and check it is the correct size. 1357 * We only blow away the file if we can stat the output and it 1358 * has the expected size. 1359 */ 1360 if (cflag != 0) 1361 return in_size == -1 ? -1 : size; 1362 1363 #ifndef SMALL 1364 if (fstat(out, &osb) != 0) { 1365 maybe_warn("couldn't stat: %s", outfile); 1366 goto bad_outfile; 1367 } 1368 1369 if (osb.st_size != size) { 1370 maybe_warnx("output file: %s wrong size (%" PRIdOFF 1371 " != %" PRIdOFF "), deleting", 1372 outfile, osb.st_size, size); 1373 goto bad_outfile; 1374 } 1375 1376 copymodes(out, &isb, outfile); 1377 #endif 1378 if (close(out) == -1) 1379 maybe_warn("couldn't close output"); 1380 1381 /* output is good, ok to delete input */ 1382 unlink_input(file, &isb); 1383 return size; 1384 1385 #ifndef SMALL 1386 bad_outfile: 1387 if (close(out) == -1) 1388 maybe_warn("couldn't close output"); 1389 1390 maybe_warnx("leaving original %s", file); 1391 unlink(outfile); 1392 return size; 1393 #endif 1394 } 1395 1396 /* uncompress the given file and remove the original */ 1397 static off_t 1398 file_uncompress(char *file, char *outfile, size_t outsize) 1399 { 1400 struct stat isb, osb; 1401 off_t size; 1402 ssize_t rbytes; 1403 unsigned char header1[4]; 1404 enum filetype method; 1405 int fd, ofd, zfd = -1; 1406 size_t in_size; 1407 #ifndef SMALL 1408 ssize_t rv; 1409 time_t timestamp = 0; 1410 char name[PATH_MAX + 1]; 1411 #endif 1412 1413 /* gather the old name info */ 1414 1415 fd = open(file, O_RDONLY); 1416 if (fd < 0) { 1417 maybe_warn("can't open %s", file); 1418 goto lose; 1419 } 1420 if (fstat(fd, &isb) != 0) { 1421 close(fd); 1422 maybe_warn("can't stat %s", file); 1423 goto lose; 1424 } 1425 if (S_ISREG(isb.st_mode)) 1426 in_size = isb.st_size; 1427 else 1428 in_size = 0; 1429 infile_set(file, in_size); 1430 1431 strlcpy(outfile, file, outsize); 1432 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1433 maybe_warnx("%s: unknown suffix -- ignored", file); 1434 goto lose; 1435 } 1436 1437 rbytes = read(fd, header1, sizeof header1); 1438 if (rbytes != sizeof header1) { 1439 /* we don't want to fail here. */ 1440 #ifndef SMALL 1441 if (fflag) 1442 goto lose; 1443 #endif 1444 if (rbytes == -1) 1445 maybe_warn("can't read %s", file); 1446 else 1447 goto unexpected_EOF; 1448 goto lose; 1449 } 1450 infile_newdata(rbytes); 1451 1452 method = file_gettype(header1); 1453 #ifndef SMALL 1454 if (fflag == 0 && method == FT_UNKNOWN) { 1455 maybe_warnx("%s: not in gzip format", file); 1456 goto lose; 1457 } 1458 1459 #endif 1460 1461 #ifndef SMALL 1462 if (method == FT_GZIP && Nflag) { 1463 unsigned char ts[4]; /* timestamp */ 1464 1465 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1466 if (rv >= 0 && rv < (ssize_t)(sizeof ts)) 1467 goto unexpected_EOF; 1468 if (rv == -1) { 1469 if (!fflag) 1470 maybe_warn("can't read %s", file); 1471 goto lose; 1472 } 1473 infile_newdata(rv); 1474 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0]; 1475 1476 if (header1[3] & ORIG_NAME) { 1477 rbytes = pread(fd, name, sizeof(name) - 1, GZIP_ORIGNAME); 1478 if (rbytes < 0) { 1479 maybe_warn("can't read %s", file); 1480 goto lose; 1481 } 1482 if (name[0] != '\0') { 1483 char *dp, *nf; 1484 1485 /* Make sure that name is NUL-terminated */ 1486 name[rbytes] = '\0'; 1487 1488 /* strip saved directory name */ 1489 nf = strrchr(name, '/'); 1490 if (nf == NULL) 1491 nf = name; 1492 else 1493 nf++; 1494 1495 /* preserve original directory name */ 1496 dp = strrchr(file, '/'); 1497 if (dp == NULL) 1498 dp = file; 1499 else 1500 dp++; 1501 snprintf(outfile, outsize, "%.*s%.*s", 1502 (int) (dp - file), 1503 file, (int) rbytes, nf); 1504 } 1505 } 1506 } 1507 #endif 1508 lseek(fd, 0, SEEK_SET); 1509 1510 if (cflag == 0 || lflag) { 1511 #ifndef SMALL 1512 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1513 maybe_warnx("%s has %d other links -- skipping", 1514 file, isb.st_nlink - 1); 1515 goto lose; 1516 } 1517 if (nflag == 0 && timestamp) 1518 isb.st_mtime = timestamp; 1519 if (check_outfile(outfile) == 0) 1520 goto lose; 1521 #endif 1522 } 1523 1524 if (cflag) 1525 zfd = STDOUT_FILENO; 1526 else if (lflag) 1527 zfd = -1; 1528 else { 1529 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1530 if (zfd == STDOUT_FILENO) { 1531 /* We won't close STDOUT_FILENO later... */ 1532 zfd = dup(zfd); 1533 close(STDOUT_FILENO); 1534 } 1535 if (zfd == -1) { 1536 maybe_warn("can't open %s", outfile); 1537 goto lose; 1538 } 1539 } 1540 1541 switch (method) { 1542 #ifndef NO_BZIP2_SUPPORT 1543 case FT_BZIP2: 1544 /* XXX */ 1545 if (lflag) { 1546 maybe_warnx("no -l with bzip2 files"); 1547 goto lose; 1548 } 1549 1550 size = unbzip2(fd, zfd, NULL, 0, NULL); 1551 break; 1552 #endif 1553 1554 #ifndef NO_COMPRESS_SUPPORT 1555 case FT_Z: { 1556 FILE *in, *out; 1557 1558 /* XXX */ 1559 if (lflag) { 1560 maybe_warnx("no -l with Lempel-Ziv files"); 1561 goto lose; 1562 } 1563 1564 if ((in = zdopen(fd)) == NULL) { 1565 maybe_warn("zdopen for read: %s", file); 1566 goto lose; 1567 } 1568 1569 out = fdopen(dup(zfd), "w"); 1570 if (out == NULL) { 1571 maybe_warn("fdopen for write: %s", outfile); 1572 fclose(in); 1573 goto lose; 1574 } 1575 1576 size = zuncompress(in, out, NULL, 0, NULL); 1577 /* need to fclose() if ferror() is true... */ 1578 if (ferror(in) | fclose(in)) { 1579 maybe_warn("failed infile fclose"); 1580 unlink(outfile); 1581 (void)fclose(out); 1582 } 1583 if (fclose(out) != 0) { 1584 maybe_warn("failed outfile fclose"); 1585 unlink(outfile); 1586 goto lose; 1587 } 1588 break; 1589 } 1590 #endif 1591 1592 #ifndef NO_PACK_SUPPORT 1593 case FT_PACK: 1594 if (lflag) { 1595 maybe_warnx("no -l with packed files"); 1596 goto lose; 1597 } 1598 1599 size = unpack(fd, zfd, NULL, 0, NULL); 1600 break; 1601 #endif 1602 1603 #ifndef NO_XZ_SUPPORT 1604 case FT_XZ: 1605 if (lflag) { 1606 size = unxz_len(fd); 1607 print_list_out(in_size, size, file); 1608 return -1; 1609 } 1610 size = unxz(fd, zfd, NULL, 0, NULL); 1611 break; 1612 #endif 1613 1614 #ifndef NO_LZ_SUPPORT 1615 case FT_LZ: 1616 if (lflag) { 1617 maybe_warnx("no -l with lzip files"); 1618 goto lose; 1619 } 1620 size = unlz(fd, zfd, NULL, 0, NULL); 1621 break; 1622 #endif 1623 #ifndef SMALL 1624 case FT_UNKNOWN: 1625 if (lflag) { 1626 maybe_warnx("no -l for unknown filetypes"); 1627 goto lose; 1628 } 1629 size = cat_fd(NULL, 0, NULL, fd); 1630 break; 1631 #endif 1632 default: 1633 if (lflag) { 1634 print_list(fd, in_size, outfile, isb.st_mtime); 1635 close(fd); 1636 return -1; /* XXX */ 1637 } 1638 1639 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1640 break; 1641 } 1642 1643 if (close(fd) != 0) 1644 maybe_warn("couldn't close input"); 1645 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1646 maybe_warn("couldn't close output"); 1647 1648 if (size == -1) { 1649 if (cflag == 0) 1650 unlink(outfile); 1651 maybe_warnx("%s: uncompress failed", file); 1652 return -1; 1653 } 1654 1655 /* if testing, or we uncompressed to stdout, this is all we need */ 1656 #ifndef SMALL 1657 if (tflag) 1658 return size; 1659 #endif 1660 /* if we are uncompressing to stdin, don't remove the file. */ 1661 if (cflag) 1662 return size; 1663 1664 /* 1665 * if we create a file... 1666 */ 1667 /* 1668 * if we can't stat the file don't remove the file. 1669 */ 1670 1671 ofd = open(outfile, O_RDWR, 0); 1672 if (ofd == -1) { 1673 maybe_warn("couldn't open (leaving original): %s", 1674 outfile); 1675 return -1; 1676 } 1677 if (fstat(ofd, &osb) != 0) { 1678 maybe_warn("couldn't stat (leaving original): %s", 1679 outfile); 1680 close(ofd); 1681 return -1; 1682 } 1683 if (osb.st_size != size) { 1684 maybe_warnx("stat gave different size: %" PRIdOFF 1685 " != %" PRIdOFF " (leaving original)", 1686 size, osb.st_size); 1687 close(ofd); 1688 unlink(outfile); 1689 return -1; 1690 } 1691 unlink_input(file, &isb); 1692 #ifndef SMALL 1693 copymodes(ofd, &isb, outfile); 1694 #endif 1695 close(ofd); 1696 return size; 1697 1698 unexpected_EOF: 1699 maybe_warnx("%s: unexpected end of file", file); 1700 lose: 1701 if (fd != -1) 1702 close(fd); 1703 if (zfd != -1 && zfd != STDOUT_FILENO) 1704 close(fd); 1705 return -1; 1706 } 1707 1708 #ifndef SMALL 1709 static void 1710 check_siginfo(void) 1711 { 1712 static int ttyfd = -2; 1713 char buf[2048]; 1714 int n; 1715 1716 if (print_info == 0) 1717 return; 1718 1719 if (!infile) 1720 goto out; 1721 1722 if (ttyfd == -2) 1723 ttyfd = open(_PATH_TTY, O_RDWR | O_CLOEXEC); 1724 1725 if (ttyfd == -1) 1726 goto out; 1727 1728 if (infile_total) { 1729 const double pcent = (100.0 * infile_current) / infile_total; 1730 1731 n = snprintf(buf, sizeof(buf), 1732 "%s: %s: done %ju/%ju bytes (%3.2f%%)\n", 1733 getprogname(), infile, (uintmax_t)infile_current, 1734 (uintmax_t)infile_total, pcent); 1735 } else { 1736 n = snprintf(buf, sizeof(buf), "%s: %s: done %ju bytes\n", 1737 getprogname(), infile, (uintmax_t)infile_current); 1738 } 1739 1740 if (n <= 0) 1741 goto out; 1742 1743 write(ttyfd, buf, (size_t)n); 1744 out: 1745 print_info = 0; 1746 } 1747 1748 static off_t 1749 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1750 { 1751 char buf[BUFLEN]; 1752 off_t in_tot; 1753 ssize_t w; 1754 1755 in_tot = count; 1756 w = write_retry(STDOUT_FILENO, prepend, count); 1757 if (w == -1 || (size_t)w != count) { 1758 maybe_warn("write to stdout"); 1759 return -1; 1760 } 1761 for (;;) { 1762 ssize_t rv; 1763 1764 rv = read(fd, buf, sizeof buf); 1765 if (rv == 0) 1766 break; 1767 if (rv < 0) { 1768 maybe_warn("read from fd %d", fd); 1769 break; 1770 } 1771 infile_newdata(rv); 1772 1773 if (write_retry(STDOUT_FILENO, buf, rv) != rv) { 1774 maybe_warn("write to stdout"); 1775 break; 1776 } 1777 in_tot += rv; 1778 } 1779 1780 if (gsizep) 1781 *gsizep = in_tot; 1782 return (in_tot); 1783 } 1784 #endif 1785 1786 static void 1787 handle_stdin(void) 1788 { 1789 struct stat isb; 1790 unsigned char header1[4]; 1791 size_t in_size; 1792 off_t usize, gsize; 1793 enum filetype method; 1794 ssize_t bytes_read; 1795 #ifndef NO_COMPRESS_SUPPORT 1796 FILE *in; 1797 #endif 1798 1799 #ifndef SMALL 1800 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1801 maybe_warnx("standard input is a terminal -- ignoring"); 1802 goto out; 1803 } 1804 #endif 1805 1806 if (fstat(STDIN_FILENO, &isb) < 0) { 1807 maybe_warn("fstat"); 1808 goto out; 1809 } 1810 if (S_ISREG(isb.st_mode)) 1811 in_size = isb.st_size; 1812 else 1813 in_size = 0; 1814 infile_set("(stdin)", in_size); 1815 1816 if (lflag) { 1817 print_list(STDIN_FILENO, in_size, infile, isb.st_mtime); 1818 goto out; 1819 } 1820 1821 bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1); 1822 if (bytes_read == -1) { 1823 maybe_warn("can't read stdin"); 1824 goto out; 1825 } else if (bytes_read != sizeof(header1)) { 1826 maybe_warnx("(stdin): unexpected end of file"); 1827 goto out; 1828 } 1829 1830 method = file_gettype(header1); 1831 switch (method) { 1832 default: 1833 #ifndef SMALL 1834 if (fflag == 0) { 1835 maybe_warnx("unknown compression format"); 1836 goto out; 1837 } 1838 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO); 1839 break; 1840 #endif 1841 case FT_GZIP: 1842 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1843 (char *)header1, sizeof header1, &gsize, "(stdin)"); 1844 break; 1845 #ifndef NO_BZIP2_SUPPORT 1846 case FT_BZIP2: 1847 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1848 (char *)header1, sizeof header1, &gsize); 1849 break; 1850 #endif 1851 #ifndef NO_COMPRESS_SUPPORT 1852 case FT_Z: 1853 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1854 maybe_warnx("zopen of stdin"); 1855 goto out; 1856 } 1857 1858 usize = zuncompress(in, stdout, (char *)header1, 1859 sizeof header1, &gsize); 1860 fclose(in); 1861 break; 1862 #endif 1863 #ifndef NO_PACK_SUPPORT 1864 case FT_PACK: 1865 usize = unpack(STDIN_FILENO, STDOUT_FILENO, 1866 (char *)header1, sizeof header1, &gsize); 1867 break; 1868 #endif 1869 #ifndef NO_XZ_SUPPORT 1870 case FT_XZ: 1871 usize = unxz(STDIN_FILENO, STDOUT_FILENO, 1872 (char *)header1, sizeof header1, &gsize); 1873 break; 1874 #endif 1875 #ifndef NO_LZ_SUPPORT 1876 case FT_LZ: 1877 usize = unlz(STDIN_FILENO, STDOUT_FILENO, 1878 (char *)header1, sizeof header1, &gsize); 1879 break; 1880 #endif 1881 } 1882 1883 #ifndef SMALL 1884 if (vflag && !tflag && usize != -1 && gsize != -1) 1885 print_verbage(NULL, NULL, usize, gsize); 1886 if (vflag && tflag) 1887 print_test("(stdin)", usize != -1); 1888 #else 1889 (void)&usize; 1890 #endif 1891 1892 out: 1893 infile_clear(); 1894 } 1895 1896 static void 1897 handle_stdout(void) 1898 { 1899 off_t gsize; 1900 #ifndef SMALL 1901 off_t usize; 1902 struct stat sb; 1903 time_t systime; 1904 uint32_t mtime; 1905 int ret; 1906 1907 infile_set("<stdout>", 0); 1908 1909 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1910 maybe_warnx("standard output is a terminal -- ignoring"); 1911 return; 1912 } 1913 1914 /* If stdin is a file use its mtime, otherwise use current time */ 1915 ret = fstat(STDIN_FILENO, &sb); 1916 if (ret < 0) { 1917 maybe_warn("Can't stat stdin"); 1918 return; 1919 } 1920 1921 if (S_ISREG(sb.st_mode)) { 1922 infile_set("<stdout>", sb.st_size); 1923 mtime = (uint32_t)sb.st_mtime; 1924 } else { 1925 systime = time(NULL); 1926 if (systime == -1) { 1927 maybe_warn("time"); 1928 return; 1929 } 1930 mtime = (uint32_t)systime; 1931 } 1932 1933 usize = 1934 #endif 1935 gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1936 #ifndef SMALL 1937 if (vflag && !tflag && usize != -1 && gsize != -1) 1938 print_verbage(NULL, NULL, usize, gsize); 1939 #endif 1940 } 1941 1942 /* do what is asked for, for the path name */ 1943 static void 1944 handle_pathname(char *path) 1945 { 1946 char *opath = path, *s = NULL; 1947 ssize_t len; 1948 int slen; 1949 struct stat sb; 1950 1951 /* check for stdout/stdin */ 1952 if (path[0] == '-' && path[1] == '\0') { 1953 if (dflag) 1954 handle_stdin(); 1955 else 1956 handle_stdout(); 1957 return; 1958 } 1959 1960 retry: 1961 if (stat(path, &sb) != 0) { 1962 /* lets try <path>.gz if we're decompressing */ 1963 if (dflag && s == NULL && errno == ENOENT) { 1964 len = strlen(path); 1965 slen = suffixes[0].ziplen; 1966 s = malloc(len + slen + 1); 1967 if (s == NULL) 1968 maybe_err("malloc"); 1969 memcpy(s, path, len); 1970 memcpy(s + len, suffixes[0].zipped, slen + 1); 1971 path = s; 1972 goto retry; 1973 } 1974 maybe_warn("can't stat: %s", opath); 1975 goto out; 1976 } 1977 1978 if (S_ISDIR(sb.st_mode)) { 1979 #ifndef SMALL 1980 if (rflag) 1981 handle_dir(path); 1982 else 1983 #endif 1984 maybe_warnx("%s is a directory", path); 1985 goto out; 1986 } 1987 1988 if (S_ISREG(sb.st_mode)) 1989 handle_file(path, &sb); 1990 else 1991 maybe_warnx("%s is not a regular file", path); 1992 1993 out: 1994 if (s) 1995 free(s); 1996 } 1997 1998 /* compress/decompress a file */ 1999 static void 2000 handle_file(char *file, struct stat *sbp) 2001 { 2002 off_t usize, gsize; 2003 char outfile[PATH_MAX]; 2004 2005 infile_set(file, sbp->st_size); 2006 if (dflag) { 2007 usize = file_uncompress(file, outfile, sizeof(outfile)); 2008 #ifndef SMALL 2009 if (vflag && tflag) 2010 print_test(file, usize != -1); 2011 #endif 2012 if (usize == -1) 2013 return; 2014 gsize = sbp->st_size; 2015 } else { 2016 gsize = file_compress(file, outfile, sizeof(outfile)); 2017 if (gsize == -1) 2018 return; 2019 usize = sbp->st_size; 2020 } 2021 infile_clear(); 2022 2023 #ifndef SMALL 2024 if (vflag && !tflag) 2025 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 2026 #endif 2027 } 2028 2029 #ifndef SMALL 2030 /* this is used with -r to recursively descend directories */ 2031 static void 2032 handle_dir(char *dir) 2033 { 2034 char *path_argv[2]; 2035 FTS *fts; 2036 FTSENT *entry; 2037 2038 path_argv[0] = dir; 2039 path_argv[1] = 0; 2040 fts = fts_open(path_argv, FTS_PHYSICAL, NULL); 2041 if (fts == NULL) { 2042 warn("couldn't fts_open %s", dir); 2043 return; 2044 } 2045 2046 while ((entry = fts_read(fts))) { 2047 switch(entry->fts_info) { 2048 case FTS_D: 2049 case FTS_DP: 2050 continue; 2051 2052 case FTS_DNR: 2053 case FTS_ERR: 2054 case FTS_NS: 2055 maybe_warn("%s", entry->fts_path); 2056 continue; 2057 case FTS_F: 2058 handle_file(entry->fts_name, entry->fts_statp); 2059 } 2060 } 2061 (void)fts_close(fts); 2062 } 2063 #endif 2064 2065 /* print a ratio - size reduction as a fraction of uncompressed size */ 2066 static void 2067 print_ratio(off_t in, off_t out, FILE *where) 2068 { 2069 int percent10; /* 10 * percent */ 2070 off_t diff; 2071 char buff[8]; 2072 int len; 2073 2074 diff = in - out/2; 2075 if (in == 0 && out == 0) 2076 percent10 = 0; 2077 else if (diff < 0) 2078 /* 2079 * Output is more than double size of input! print -99.9% 2080 * Quite possibly we've failed to get the original size. 2081 */ 2082 percent10 = -999; 2083 else { 2084 /* 2085 * We only need 12 bits of result from the final division, 2086 * so reduce the values until a 32bit division will suffice. 2087 */ 2088 while (in > 0x100000) { 2089 diff >>= 1; 2090 in >>= 1; 2091 } 2092 if (in != 0) 2093 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 2094 else 2095 percent10 = 0; 2096 } 2097 2098 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 2099 /* Move the '.' to before the last digit */ 2100 buff[len - 1] = buff[len - 2]; 2101 buff[len - 2] = '.'; 2102 fprintf(where, "%5s%%", buff); 2103 } 2104 2105 #ifndef SMALL 2106 /* print compression statistics, and the new name (if there is one!) */ 2107 static void 2108 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 2109 { 2110 if (file) 2111 fprintf(stderr, "%s:%s ", file, 2112 strlen(file) < 7 ? "\t\t" : "\t"); 2113 print_ratio(usize, gsize, stderr); 2114 if (nfile) 2115 fprintf(stderr, " -- replaced with %s", nfile); 2116 fprintf(stderr, "\n"); 2117 fflush(stderr); 2118 } 2119 2120 /* print test results */ 2121 static void 2122 print_test(const char *file, int ok) 2123 { 2124 2125 if (exit_value == 0 && ok == 0) 2126 exit_value = 1; 2127 fprintf(stderr, "%s:%s %s\n", file, 2128 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 2129 fflush(stderr); 2130 } 2131 #endif 2132 2133 /* print a file's info ala --list */ 2134 /* eg: 2135 compressed uncompressed ratio uncompressed_name 2136 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 2137 */ 2138 static void 2139 print_list(int fd, off_t out, const char *outfile, time_t ts) 2140 { 2141 static int first = 1; 2142 #ifndef SMALL 2143 static off_t in_tot, out_tot; 2144 uint32_t crc = 0; 2145 #endif 2146 off_t in = 0, rv; 2147 2148 if (first) { 2149 #ifndef SMALL 2150 if (vflag) 2151 printf("method crc date time "); 2152 #endif 2153 if (qflag == 0) 2154 printf(" compressed uncompressed " 2155 "ratio uncompressed_name\n"); 2156 } 2157 first = 0; 2158 2159 /* print totals? */ 2160 #ifndef SMALL 2161 if (fd == -1) { 2162 in = in_tot; 2163 out = out_tot; 2164 } else 2165 #endif 2166 { 2167 /* read the last 4 bytes - this is the uncompressed size */ 2168 rv = lseek(fd, (off_t)(-8), SEEK_END); 2169 if (rv != -1) { 2170 unsigned char buf[8]; 2171 uint32_t usize; 2172 2173 rv = read(fd, (char *)buf, sizeof(buf)); 2174 if (rv == -1) 2175 maybe_warn("read of uncompressed size"); 2176 else if (rv != sizeof(buf)) 2177 maybe_warnx("read of uncompressed size"); 2178 2179 else { 2180 usize = buf[4]; 2181 usize |= (unsigned int)buf[5] << 8; 2182 usize |= (unsigned int)buf[6] << 16; 2183 usize |= (unsigned int)buf[7] << 24; 2184 in = (off_t)usize; 2185 #ifndef SMALL 2186 crc = buf[0]; 2187 crc |= (unsigned int)buf[1] << 8; 2188 crc |= (unsigned int)buf[2] << 16; 2189 crc |= (unsigned int)buf[3] << 24; 2190 #endif 2191 } 2192 } 2193 } 2194 2195 #ifndef SMALL 2196 if (vflag && fd == -1) 2197 printf(" "); 2198 else if (vflag) { 2199 char *date = ctime(&ts); 2200 2201 /* skip the day, 1/100th second, and year */ 2202 date += 4; 2203 date[12] = 0; 2204 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 2205 } 2206 in_tot += in; 2207 out_tot += out; 2208 #endif 2209 print_list_out(out, in, outfile); 2210 } 2211 2212 static void 2213 print_list_out(off_t out, off_t in, const char *outfile) 2214 { 2215 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 2216 print_ratio(in, out, stdout); 2217 printf(" %s\n", outfile); 2218 } 2219 2220 /* display the usage of NetBSD gzip */ 2221 static void 2222 usage(void) 2223 { 2224 2225 fprintf(stderr, "%s\n", gzip_version); 2226 fprintf(stderr, 2227 "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n" 2228 #ifndef SMALL 2229 " -1 --fast fastest (worst) compression\n" 2230 " -2 .. -8 set compression level\n" 2231 " -9 --best best (slowest) compression\n" 2232 " -c --stdout write to stdout, keep original files\n" 2233 " --to-stdout\n" 2234 " -d --decompress uncompress files\n" 2235 " --uncompress\n" 2236 " -f --force force overwriting & compress links\n" 2237 " -h --help display this help\n" 2238 " -k --keep don't delete input files during operation\n" 2239 " -l --list list compressed file contents\n" 2240 " -N --name save or restore original file name and time stamp\n" 2241 " -n --no-name don't save original file name or time stamp\n" 2242 " -q --quiet output no warnings\n" 2243 " -r --recursive recursively compress files in directories\n" 2244 " -S .suf use suffix .suf instead of .gz\n" 2245 " --suffix .suf\n" 2246 " -t --test test compressed file\n" 2247 " -V --version display program version\n" 2248 " -v --verbose print extra statistics\n", 2249 #else 2250 , 2251 #endif 2252 getprogname()); 2253 exit(0); 2254 } 2255 2256 /* display the version of NetBSD gzip */ 2257 static void 2258 display_version(void) 2259 { 2260 2261 fprintf(stderr, "%s\n", gzip_version); 2262 exit(0); 2263 } 2264 2265 #ifndef NO_BZIP2_SUPPORT 2266 #include "unbzip2.c" 2267 #endif 2268 #ifndef NO_COMPRESS_SUPPORT 2269 #include "zuncompress.c" 2270 #endif 2271 #ifndef NO_PACK_SUPPORT 2272 #include "unpack.c" 2273 #endif 2274 #ifndef NO_XZ_SUPPORT 2275 #include "unxz.c" 2276 #endif 2277 #ifndef NO_LZ_SUPPORT 2278 #include "unlz.c" 2279 #endif 2280 2281 static ssize_t 2282 read_retry(int fd, void *buf, size_t sz) 2283 { 2284 char *cp = buf; 2285 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2286 2287 while (left > 0) { 2288 ssize_t ret; 2289 2290 ret = read(fd, cp, left); 2291 if (ret == -1) { 2292 return ret; 2293 } else if (ret == 0) { 2294 break; /* EOF */ 2295 } 2296 cp += ret; 2297 left -= ret; 2298 } 2299 2300 return sz - left; 2301 } 2302 2303 static ssize_t 2304 write_retry(int fd, const void *buf, size_t sz) 2305 { 2306 const char *cp = buf; 2307 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2308 2309 while (left > 0) { 2310 ssize_t ret; 2311 2312 ret = write(fd, cp, left); 2313 if (ret == -1) { 2314 return ret; 2315 } else if (ret == 0) { 2316 abort(); /* Can't happen */ 2317 } 2318 cp += ret; 2319 left -= ret; 2320 } 2321 2322 return sz - left; 2323 } 2324