1 /* $NetBSD: gzip.c,v 1.116 2018/10/27 11:39:12 skrll Exp $ */ 2 3 /*- 4 * SPDX-License-Identifier: BSD-2-Clause-NetBSD 5 * 6 * Copyright (c) 1997, 1998, 2003, 2004, 2006, 2008, 2009, 2010, 2011, 2015, 2017 7 * Matthew R. Green 8 * All rights reserved. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 20 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 21 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 22 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 23 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 24 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 25 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 26 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 27 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 * 31 */ 32 33 #include <sys/cdefs.h> 34 #ifndef lint 35 __COPYRIGHT("@(#) Copyright (c) 1997, 1998, 2003, 2004, 2006, 2008,\ 36 2009, 2010, 2011, 2015, 2017 Matthew R. Green. All rights reserved."); 37 __FBSDID("$FreeBSD$"); 38 #endif /* not lint */ 39 40 /* 41 * gzip.c -- GPL free gzip using zlib. 42 * 43 * RFC 1950 covers the zlib format 44 * RFC 1951 covers the deflate format 45 * RFC 1952 covers the gzip format 46 * 47 * TODO: 48 * - use mmap where possible 49 * - make bzip2/compress -v/-t/-l support work as well as possible 50 */ 51 52 #include <sys/endian.h> 53 #include <sys/param.h> 54 #include <sys/stat.h> 55 #include <sys/time.h> 56 57 #include <inttypes.h> 58 #include <unistd.h> 59 #include <stdio.h> 60 #include <string.h> 61 #include <stdlib.h> 62 #include <err.h> 63 #include <errno.h> 64 #include <fcntl.h> 65 #include <zlib.h> 66 #include <fts.h> 67 #include <libgen.h> 68 #include <stdarg.h> 69 #include <getopt.h> 70 #include <time.h> 71 72 /* what type of file are we dealing with */ 73 enum filetype { 74 FT_GZIP, 75 #ifndef NO_BZIP2_SUPPORT 76 FT_BZIP2, 77 #endif 78 #ifndef NO_COMPRESS_SUPPORT 79 FT_Z, 80 #endif 81 #ifndef NO_PACK_SUPPORT 82 FT_PACK, 83 #endif 84 #ifndef NO_XZ_SUPPORT 85 FT_XZ, 86 #endif 87 #ifndef NO_LZ_SUPPORT 88 FT_LZ, 89 #endif 90 FT_LAST, 91 FT_UNKNOWN 92 }; 93 94 #ifndef NO_BZIP2_SUPPORT 95 #include <bzlib.h> 96 97 #define BZ2_SUFFIX ".bz2" 98 #define BZIP2_MAGIC "BZh" 99 #endif 100 101 #ifndef NO_COMPRESS_SUPPORT 102 #define Z_SUFFIX ".Z" 103 #define Z_MAGIC "\037\235" 104 #endif 105 106 #ifndef NO_PACK_SUPPORT 107 #define PACK_MAGIC "\037\036" 108 #endif 109 110 #ifndef NO_XZ_SUPPORT 111 #include <lzma.h> 112 #define XZ_SUFFIX ".xz" 113 #define XZ_MAGIC "\3757zXZ" 114 #endif 115 116 #ifndef NO_LZ_SUPPORT 117 #define LZ_SUFFIX ".lz" 118 #define LZ_MAGIC "LZIP" 119 #endif 120 121 #define GZ_SUFFIX ".gz" 122 123 #define BUFLEN (64 * 1024) 124 125 #define GZIP_MAGIC0 0x1F 126 #define GZIP_MAGIC1 0x8B 127 #define GZIP_OMAGIC1 0x9E 128 129 #define GZIP_TIMESTAMP (off_t)4 130 #define GZIP_ORIGNAME (off_t)10 131 132 #define HEAD_CRC 0x02 133 #define EXTRA_FIELD 0x04 134 #define ORIG_NAME 0x08 135 #define COMMENT 0x10 136 137 #define OS_CODE 3 /* Unix */ 138 139 typedef struct { 140 const char *zipped; 141 int ziplen; 142 const char *normal; /* for unzip - must not be longer than zipped */ 143 } suffixes_t; 144 static suffixes_t suffixes[] = { 145 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N} 146 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S .xxx */ 147 SUFFIX(GZ_SUFFIX, ""), 148 SUFFIX(".z", ""), 149 SUFFIX("-gz", ""), 150 SUFFIX("-z", ""), 151 SUFFIX("_z", ""), 152 SUFFIX(".taz", ".tar"), 153 SUFFIX(".tgz", ".tar"), 154 #ifndef NO_BZIP2_SUPPORT 155 SUFFIX(BZ2_SUFFIX, ""), 156 SUFFIX(".tbz", ".tar"), 157 SUFFIX(".tbz2", ".tar"), 158 #endif 159 #ifndef NO_COMPRESS_SUPPORT 160 SUFFIX(Z_SUFFIX, ""), 161 #endif 162 #ifndef NO_XZ_SUPPORT 163 SUFFIX(XZ_SUFFIX, ""), 164 #endif 165 #ifndef NO_LZ_SUPPORT 166 SUFFIX(LZ_SUFFIX, ""), 167 #endif 168 SUFFIX(GZ_SUFFIX, ""), /* Overwritten by -S "" */ 169 #undef SUFFIX 170 }; 171 #define NUM_SUFFIXES (nitems(suffixes)) 172 #define SUFFIX_MAXLEN 30 173 174 static const char gzip_version[] = "FreeBSD gzip 20190107"; 175 176 static const char gzip_copyright[] = \ 177 " Copyright (c) 1997, 1998, 2003, 2004, 2006 Matthew R. Green\n" 178 " All rights reserved.\n" 179 "\n" 180 " Redistribution and use in source and binary forms, with or without\n" 181 " modification, are permitted provided that the following conditions\n" 182 " are met:\n" 183 " 1. Redistributions of source code must retain the above copyright\n" 184 " notice, this list of conditions and the following disclaimer.\n" 185 " 2. Redistributions in binary form must reproduce the above copyright\n" 186 " notice, this list of conditions and the following disclaimer in the\n" 187 " documentation and/or other materials provided with the distribution.\n" 188 "\n" 189 " THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR\n" 190 " IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES\n" 191 " OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.\n" 192 " IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,\n" 193 " INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,\n" 194 " BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;\n" 195 " LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED\n" 196 " AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,\n" 197 " OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY\n" 198 " OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF\n" 199 " SUCH DAMAGE."; 200 201 static int cflag; /* stdout mode */ 202 static int dflag; /* decompress mode */ 203 static int lflag; /* list mode */ 204 static int numflag = 6; /* gzip -1..-9 value */ 205 206 static const char *remove_file = NULL; /* file to be removed upon SIGINT */ 207 208 static int fflag; /* force mode */ 209 static int kflag; /* don't delete input files */ 210 static int nflag; /* don't save name/timestamp */ 211 static int Nflag; /* don't restore name/timestamp */ 212 static int qflag; /* quiet mode */ 213 static int rflag; /* recursive mode */ 214 static int tflag; /* test */ 215 static int vflag; /* verbose mode */ 216 static sig_atomic_t print_info = 0; 217 218 static int exit_value = 0; /* exit value */ 219 220 static const char *infile; /* name of file coming in */ 221 222 static void maybe_err(const char *fmt, ...) __printflike(1, 2) __dead2; 223 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \ 224 !defined(NO_XZ_SUPPORT) 225 static void maybe_errx(const char *fmt, ...) __printflike(1, 2) __dead2; 226 #endif 227 static void maybe_warn(const char *fmt, ...) __printflike(1, 2); 228 static void maybe_warnx(const char *fmt, ...) __printflike(1, 2); 229 static enum filetype file_gettype(u_char *); 230 static off_t gz_compress(int, int, off_t *, const char *, uint32_t); 231 static off_t gz_uncompress(int, int, char *, size_t, off_t *, const char *); 232 static off_t file_compress(char *, char *, size_t); 233 static off_t file_uncompress(char *, char *, size_t); 234 static void handle_pathname(char *); 235 static void handle_file(char *, struct stat *); 236 static void handle_stdin(void); 237 static void handle_stdout(void); 238 static void print_ratio(off_t, off_t, FILE *); 239 static void print_list(int fd, off_t, const char *, time_t); 240 static void usage(void) __dead2; 241 static void display_version(void) __dead2; 242 static void display_license(void); 243 static const suffixes_t *check_suffix(char *, int); 244 static ssize_t read_retry(int, void *, size_t); 245 static ssize_t write_retry(int, const void *, size_t); 246 static void print_list_out(off_t, off_t, const char*); 247 248 static void infile_set(const char *newinfile, off_t total); 249 250 static off_t infile_total; /* total expected to read/write */ 251 static off_t infile_current; /* current read/write */ 252 253 static void check_siginfo(void); 254 static off_t cat_fd(unsigned char *, size_t, off_t *, int fd); 255 static void prepend_gzip(char *, int *, char ***); 256 static void handle_dir(char *); 257 static void print_verbage(const char *, const char *, off_t, off_t); 258 static void print_test(const char *, int); 259 static void copymodes(int fd, const struct stat *, const char *file); 260 static int check_outfile(const char *outfile); 261 static void setup_signals(void); 262 static void infile_newdata(size_t newdata); 263 static void infile_clear(void); 264 265 #ifndef NO_BZIP2_SUPPORT 266 static off_t unbzip2(int, int, char *, size_t, off_t *); 267 #endif 268 269 #ifndef NO_COMPRESS_SUPPORT 270 static FILE *zdopen(int); 271 static off_t zuncompress(FILE *, FILE *, char *, size_t, off_t *); 272 #endif 273 274 #ifndef NO_PACK_SUPPORT 275 static off_t unpack(int, int, char *, size_t, off_t *); 276 #endif 277 278 #ifndef NO_XZ_SUPPORT 279 static off_t unxz(int, int, char *, size_t, off_t *); 280 static off_t unxz_len(int); 281 #endif 282 283 #ifndef NO_LZ_SUPPORT 284 static off_t unlz(int, int, char *, size_t, off_t *); 285 #endif 286 287 static const struct option longopts[] = { 288 { "stdout", no_argument, 0, 'c' }, 289 { "to-stdout", no_argument, 0, 'c' }, 290 { "decompress", no_argument, 0, 'd' }, 291 { "uncompress", no_argument, 0, 'd' }, 292 { "force", no_argument, 0, 'f' }, 293 { "help", no_argument, 0, 'h' }, 294 { "keep", no_argument, 0, 'k' }, 295 { "list", no_argument, 0, 'l' }, 296 { "no-name", no_argument, 0, 'n' }, 297 { "name", no_argument, 0, 'N' }, 298 { "quiet", no_argument, 0, 'q' }, 299 { "recursive", no_argument, 0, 'r' }, 300 { "suffix", required_argument, 0, 'S' }, 301 { "test", no_argument, 0, 't' }, 302 { "verbose", no_argument, 0, 'v' }, 303 { "version", no_argument, 0, 'V' }, 304 { "fast", no_argument, 0, '1' }, 305 { "best", no_argument, 0, '9' }, 306 { "ascii", no_argument, 0, 'a' }, 307 { "license", no_argument, 0, 'L' }, 308 { NULL, no_argument, 0, 0 }, 309 }; 310 311 int 312 main(int argc, char **argv) 313 { 314 const char *progname = getprogname(); 315 char *gzip; 316 int len; 317 int ch; 318 319 setup_signals(); 320 321 if ((gzip = getenv("GZIP")) != NULL) 322 prepend_gzip(gzip, &argc, &argv); 323 324 /* 325 * XXX 326 * handle being called `gunzip', `zcat' and `gzcat' 327 */ 328 if (strcmp(progname, "gunzip") == 0) 329 dflag = 1; 330 else if (strcmp(progname, "zcat") == 0 || 331 strcmp(progname, "gzcat") == 0) 332 dflag = cflag = 1; 333 334 #define OPT_LIST "123456789acdfhklLNnqrS:tVv" 335 336 while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) { 337 switch (ch) { 338 case '1': case '2': case '3': 339 case '4': case '5': case '6': 340 case '7': case '8': case '9': 341 numflag = ch - '0'; 342 break; 343 case 'c': 344 cflag = 1; 345 break; 346 case 'd': 347 dflag = 1; 348 break; 349 case 'l': 350 lflag = 1; 351 dflag = 1; 352 break; 353 case 'V': 354 display_version(); 355 /* NOTREACHED */ 356 case 'a': 357 fprintf(stderr, "%s: option --ascii ignored on this system\n", progname); 358 break; 359 case 'f': 360 fflag = 1; 361 break; 362 case 'k': 363 kflag = 1; 364 break; 365 case 'L': 366 display_license(); 367 /* NOT REACHED */ 368 case 'N': 369 nflag = 0; 370 Nflag = 1; 371 break; 372 case 'n': 373 nflag = 1; 374 Nflag = 0; 375 break; 376 case 'q': 377 qflag = 1; 378 break; 379 case 'r': 380 rflag = 1; 381 break; 382 case 'S': 383 len = strlen(optarg); 384 if (len != 0) { 385 if (len > SUFFIX_MAXLEN) 386 errx(1, "incorrect suffix: '%s': too long", optarg); 387 suffixes[0].zipped = optarg; 388 suffixes[0].ziplen = len; 389 } else { 390 suffixes[NUM_SUFFIXES - 1].zipped = ""; 391 suffixes[NUM_SUFFIXES - 1].ziplen = 0; 392 } 393 break; 394 case 't': 395 cflag = 1; 396 tflag = 1; 397 dflag = 1; 398 break; 399 case 'v': 400 vflag = 1; 401 break; 402 default: 403 usage(); 404 /* NOTREACHED */ 405 } 406 } 407 argv += optind; 408 argc -= optind; 409 410 if (argc == 0) { 411 if (dflag) /* stdin mode */ 412 handle_stdin(); 413 else /* stdout mode */ 414 handle_stdout(); 415 } else { 416 do { 417 handle_pathname(argv[0]); 418 } while (*++argv); 419 } 420 if (qflag == 0 && lflag && argc > 1) 421 print_list(-1, 0, "(totals)", 0); 422 exit(exit_value); 423 } 424 425 /* maybe print a warning */ 426 void 427 maybe_warn(const char *fmt, ...) 428 { 429 va_list ap; 430 431 if (qflag == 0) { 432 va_start(ap, fmt); 433 vwarn(fmt, ap); 434 va_end(ap); 435 } 436 if (exit_value == 0) 437 exit_value = 1; 438 } 439 440 /* ... without an errno. */ 441 void 442 maybe_warnx(const char *fmt, ...) 443 { 444 va_list ap; 445 446 if (qflag == 0) { 447 va_start(ap, fmt); 448 vwarnx(fmt, ap); 449 va_end(ap); 450 } 451 if (exit_value == 0) 452 exit_value = 1; 453 } 454 455 /* maybe print an error */ 456 void 457 maybe_err(const char *fmt, ...) 458 { 459 va_list ap; 460 461 if (qflag == 0) { 462 va_start(ap, fmt); 463 vwarn(fmt, ap); 464 va_end(ap); 465 } 466 exit(2); 467 } 468 469 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT) || \ 470 !defined(NO_XZ_SUPPORT) 471 /* ... without an errno. */ 472 void 473 maybe_errx(const char *fmt, ...) 474 { 475 va_list ap; 476 477 if (qflag == 0) { 478 va_start(ap, fmt); 479 vwarnx(fmt, ap); 480 va_end(ap); 481 } 482 exit(2); 483 } 484 #endif 485 486 /* split up $GZIP and prepend it to the argument list */ 487 static void 488 prepend_gzip(char *gzip, int *argc, char ***argv) 489 { 490 char *s, **nargv, **ac; 491 int nenvarg = 0, i; 492 493 /* scan how many arguments there are */ 494 for (s = gzip;;) { 495 while (*s == ' ' || *s == '\t') 496 s++; 497 if (*s == 0) 498 goto count_done; 499 nenvarg++; 500 while (*s != ' ' && *s != '\t') 501 if (*s++ == 0) 502 goto count_done; 503 } 504 count_done: 505 /* punt early */ 506 if (nenvarg == 0) 507 return; 508 509 *argc += nenvarg; 510 ac = *argv; 511 512 nargv = (char **)malloc((*argc + 1) * sizeof(char *)); 513 if (nargv == NULL) 514 maybe_err("malloc"); 515 516 /* stash this away */ 517 *argv = nargv; 518 519 /* copy the program name first */ 520 i = 0; 521 nargv[i++] = *(ac++); 522 523 /* take a copy of $GZIP and add it to the array */ 524 s = strdup(gzip); 525 if (s == NULL) 526 maybe_err("strdup"); 527 for (;;) { 528 /* Skip whitespaces. */ 529 while (*s == ' ' || *s == '\t') 530 s++; 531 if (*s == 0) 532 goto copy_done; 533 nargv[i++] = s; 534 /* Find the end of this argument. */ 535 while (*s != ' ' && *s != '\t') 536 if (*s++ == 0) 537 /* Argument followed by NUL. */ 538 goto copy_done; 539 /* Terminate by overwriting ' ' or '\t' with NUL. */ 540 *s++ = 0; 541 } 542 copy_done: 543 544 /* copy the original arguments and a NULL */ 545 while (*ac) 546 nargv[i++] = *(ac++); 547 nargv[i] = NULL; 548 } 549 550 /* compress input to output. Return bytes read, -1 on error */ 551 static off_t 552 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime) 553 { 554 z_stream z; 555 char *outbufp, *inbufp; 556 off_t in_tot = 0, out_tot = 0; 557 ssize_t in_size; 558 int i, error; 559 uLong crc; 560 561 outbufp = malloc(BUFLEN); 562 inbufp = malloc(BUFLEN); 563 if (outbufp == NULL || inbufp == NULL) { 564 maybe_err("malloc failed"); 565 goto out; 566 } 567 568 memset(&z, 0, sizeof z); 569 z.zalloc = Z_NULL; 570 z.zfree = Z_NULL; 571 z.opaque = 0; 572 573 if (nflag != 0) { 574 mtime = 0; 575 origname = ""; 576 } 577 578 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s", 579 GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 580 *origname ? ORIG_NAME : 0, 581 mtime & 0xff, 582 (mtime >> 8) & 0xff, 583 (mtime >> 16) & 0xff, 584 (mtime >> 24) & 0xff, 585 numflag == 1 ? 4 : numflag == 9 ? 2 : 0, 586 OS_CODE, origname); 587 if (i >= BUFLEN) 588 /* this need PATH_MAX > BUFLEN ... */ 589 maybe_err("snprintf"); 590 if (*origname) 591 i++; 592 593 z.next_out = (unsigned char *)outbufp + i; 594 z.avail_out = BUFLEN - i; 595 596 error = deflateInit2(&z, numflag, Z_DEFLATED, 597 (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY); 598 if (error != Z_OK) { 599 maybe_warnx("deflateInit2 failed"); 600 in_tot = -1; 601 goto out; 602 } 603 604 crc = crc32(0L, Z_NULL, 0); 605 for (;;) { 606 if (z.avail_out == 0) { 607 if (write_retry(out, outbufp, BUFLEN) != BUFLEN) { 608 maybe_warn("write"); 609 out_tot = -1; 610 goto out; 611 } 612 613 out_tot += BUFLEN; 614 z.next_out = (unsigned char *)outbufp; 615 z.avail_out = BUFLEN; 616 } 617 618 if (z.avail_in == 0) { 619 in_size = read(in, inbufp, BUFLEN); 620 if (in_size < 0) { 621 maybe_warn("read"); 622 in_tot = -1; 623 goto out; 624 } 625 if (in_size == 0) 626 break; 627 infile_newdata(in_size); 628 629 crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size); 630 in_tot += in_size; 631 z.next_in = (unsigned char *)inbufp; 632 z.avail_in = in_size; 633 } 634 635 error = deflate(&z, Z_NO_FLUSH); 636 if (error != Z_OK && error != Z_STREAM_END) { 637 maybe_warnx("deflate failed"); 638 in_tot = -1; 639 goto out; 640 } 641 } 642 643 /* clean up */ 644 for (;;) { 645 size_t len; 646 ssize_t w; 647 648 error = deflate(&z, Z_FINISH); 649 if (error != Z_OK && error != Z_STREAM_END) { 650 maybe_warnx("deflate failed"); 651 in_tot = -1; 652 goto out; 653 } 654 655 len = (char *)z.next_out - outbufp; 656 657 w = write_retry(out, outbufp, len); 658 if (w == -1 || (size_t)w != len) { 659 maybe_warn("write"); 660 out_tot = -1; 661 goto out; 662 } 663 out_tot += len; 664 z.next_out = (unsigned char *)outbufp; 665 z.avail_out = BUFLEN; 666 667 if (error == Z_STREAM_END) 668 break; 669 } 670 671 if (deflateEnd(&z) != Z_OK) { 672 maybe_warnx("deflateEnd failed"); 673 in_tot = -1; 674 goto out; 675 } 676 677 i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c", 678 (int)crc & 0xff, 679 (int)(crc >> 8) & 0xff, 680 (int)(crc >> 16) & 0xff, 681 (int)(crc >> 24) & 0xff, 682 (int)in_tot & 0xff, 683 (int)(in_tot >> 8) & 0xff, 684 (int)(in_tot >> 16) & 0xff, 685 (int)(in_tot >> 24) & 0xff); 686 if (i != 8) 687 maybe_err("snprintf"); 688 if (write_retry(out, outbufp, i) != i) { 689 maybe_warn("write"); 690 in_tot = -1; 691 } else 692 out_tot += i; 693 694 out: 695 if (inbufp != NULL) 696 free(inbufp); 697 if (outbufp != NULL) 698 free(outbufp); 699 if (gsizep) 700 *gsizep = out_tot; 701 return in_tot; 702 } 703 704 /* 705 * uncompress input to output then close the input. return the 706 * uncompressed size written, and put the compressed sized read 707 * into `*gsizep'. 708 */ 709 static off_t 710 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep, 711 const char *filename) 712 { 713 z_stream z; 714 char *outbufp, *inbufp; 715 off_t out_tot = -1, in_tot = 0; 716 uint32_t out_sub_tot = 0; 717 enum { 718 GZSTATE_MAGIC0, 719 GZSTATE_MAGIC1, 720 GZSTATE_METHOD, 721 GZSTATE_FLAGS, 722 GZSTATE_SKIPPING, 723 GZSTATE_EXTRA, 724 GZSTATE_EXTRA2, 725 GZSTATE_EXTRA3, 726 GZSTATE_ORIGNAME, 727 GZSTATE_COMMENT, 728 GZSTATE_HEAD_CRC1, 729 GZSTATE_HEAD_CRC2, 730 GZSTATE_INIT, 731 GZSTATE_READ, 732 GZSTATE_CRC, 733 GZSTATE_LEN, 734 } state = GZSTATE_MAGIC0; 735 int flags = 0, skip_count = 0; 736 int error = Z_STREAM_ERROR, done_reading = 0; 737 uLong crc = 0; 738 ssize_t wr; 739 int needmore = 0; 740 741 #define ADVANCE() { z.next_in++; z.avail_in--; } 742 743 if ((outbufp = malloc(BUFLEN)) == NULL) { 744 maybe_err("malloc failed"); 745 goto out2; 746 } 747 if ((inbufp = malloc(BUFLEN)) == NULL) { 748 maybe_err("malloc failed"); 749 goto out1; 750 } 751 752 memset(&z, 0, sizeof z); 753 z.avail_in = prelen; 754 z.next_in = (unsigned char *)pre; 755 z.avail_out = BUFLEN; 756 z.next_out = (unsigned char *)outbufp; 757 z.zalloc = NULL; 758 z.zfree = NULL; 759 z.opaque = 0; 760 761 in_tot = prelen; 762 out_tot = 0; 763 764 for (;;) { 765 check_siginfo(); 766 if ((z.avail_in == 0 || needmore) && done_reading == 0) { 767 ssize_t in_size; 768 769 if (z.avail_in > 0) { 770 memmove(inbufp, z.next_in, z.avail_in); 771 } 772 z.next_in = (unsigned char *)inbufp; 773 in_size = read(in, z.next_in + z.avail_in, 774 BUFLEN - z.avail_in); 775 776 if (in_size == -1) { 777 maybe_warn("failed to read stdin"); 778 goto stop_and_fail; 779 } else if (in_size == 0) { 780 done_reading = 1; 781 } 782 infile_newdata(in_size); 783 784 z.avail_in += in_size; 785 needmore = 0; 786 787 in_tot += in_size; 788 } 789 if (z.avail_in == 0) { 790 if (done_reading && state != GZSTATE_MAGIC0) { 791 maybe_warnx("%s: unexpected end of file", 792 filename); 793 goto stop_and_fail; 794 } 795 goto stop; 796 } 797 switch (state) { 798 case GZSTATE_MAGIC0: 799 if (*z.next_in != GZIP_MAGIC0) { 800 if (in_tot > 0) { 801 maybe_warnx("%s: trailing garbage " 802 "ignored", filename); 803 exit_value = 2; 804 goto stop; 805 } 806 maybe_warnx("input not gziped (MAGIC0)"); 807 goto stop_and_fail; 808 } 809 ADVANCE(); 810 state++; 811 out_sub_tot = 0; 812 crc = crc32(0L, Z_NULL, 0); 813 break; 814 815 case GZSTATE_MAGIC1: 816 if (*z.next_in != GZIP_MAGIC1 && 817 *z.next_in != GZIP_OMAGIC1) { 818 maybe_warnx("input not gziped (MAGIC1)"); 819 goto stop_and_fail; 820 } 821 ADVANCE(); 822 state++; 823 break; 824 825 case GZSTATE_METHOD: 826 if (*z.next_in != Z_DEFLATED) { 827 maybe_warnx("unknown compression method"); 828 goto stop_and_fail; 829 } 830 ADVANCE(); 831 state++; 832 break; 833 834 case GZSTATE_FLAGS: 835 flags = *z.next_in; 836 ADVANCE(); 837 skip_count = 6; 838 state++; 839 break; 840 841 case GZSTATE_SKIPPING: 842 if (skip_count > 0) { 843 skip_count--; 844 ADVANCE(); 845 } else 846 state++; 847 break; 848 849 case GZSTATE_EXTRA: 850 if ((flags & EXTRA_FIELD) == 0) { 851 state = GZSTATE_ORIGNAME; 852 break; 853 } 854 skip_count = *z.next_in; 855 ADVANCE(); 856 state++; 857 break; 858 859 case GZSTATE_EXTRA2: 860 skip_count |= ((*z.next_in) << 8); 861 ADVANCE(); 862 state++; 863 break; 864 865 case GZSTATE_EXTRA3: 866 if (skip_count > 0) { 867 skip_count--; 868 ADVANCE(); 869 } else 870 state++; 871 break; 872 873 case GZSTATE_ORIGNAME: 874 if ((flags & ORIG_NAME) == 0) { 875 state++; 876 break; 877 } 878 if (*z.next_in == 0) 879 state++; 880 ADVANCE(); 881 break; 882 883 case GZSTATE_COMMENT: 884 if ((flags & COMMENT) == 0) { 885 state++; 886 break; 887 } 888 if (*z.next_in == 0) 889 state++; 890 ADVANCE(); 891 break; 892 893 case GZSTATE_HEAD_CRC1: 894 if (flags & HEAD_CRC) 895 skip_count = 2; 896 else 897 skip_count = 0; 898 state++; 899 break; 900 901 case GZSTATE_HEAD_CRC2: 902 if (skip_count > 0) { 903 skip_count--; 904 ADVANCE(); 905 } else 906 state++; 907 break; 908 909 case GZSTATE_INIT: 910 if (inflateInit2(&z, -MAX_WBITS) != Z_OK) { 911 maybe_warnx("failed to inflateInit"); 912 goto stop_and_fail; 913 } 914 state++; 915 break; 916 917 case GZSTATE_READ: 918 error = inflate(&z, Z_FINISH); 919 switch (error) { 920 /* Z_BUF_ERROR goes with Z_FINISH... */ 921 case Z_BUF_ERROR: 922 if (z.avail_out > 0 && !done_reading) 923 continue; 924 925 case Z_STREAM_END: 926 case Z_OK: 927 break; 928 929 case Z_NEED_DICT: 930 maybe_warnx("Z_NEED_DICT error"); 931 goto stop_and_fail; 932 case Z_DATA_ERROR: 933 maybe_warnx("data stream error"); 934 goto stop_and_fail; 935 case Z_STREAM_ERROR: 936 maybe_warnx("internal stream error"); 937 goto stop_and_fail; 938 case Z_MEM_ERROR: 939 maybe_warnx("memory allocation error"); 940 goto stop_and_fail; 941 942 default: 943 maybe_warn("unknown error from inflate(): %d", 944 error); 945 } 946 wr = BUFLEN - z.avail_out; 947 948 if (wr != 0) { 949 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr); 950 if ( 951 /* don't write anything with -t */ 952 tflag == 0 && 953 write_retry(out, outbufp, wr) != wr) { 954 maybe_warn("error writing to output"); 955 goto stop_and_fail; 956 } 957 958 out_tot += wr; 959 out_sub_tot += wr; 960 } 961 962 if (error == Z_STREAM_END) { 963 inflateEnd(&z); 964 state++; 965 } 966 967 z.next_out = (unsigned char *)outbufp; 968 z.avail_out = BUFLEN; 969 970 break; 971 case GZSTATE_CRC: 972 { 973 uLong origcrc; 974 975 if (z.avail_in < 4) { 976 if (!done_reading) { 977 needmore = 1; 978 continue; 979 } 980 maybe_warnx("truncated input"); 981 goto stop_and_fail; 982 } 983 origcrc = le32dec(&z.next_in[0]); 984 if (origcrc != crc) { 985 maybe_warnx("invalid compressed" 986 " data--crc error"); 987 goto stop_and_fail; 988 } 989 } 990 991 z.avail_in -= 4; 992 z.next_in += 4; 993 994 if (!z.avail_in && done_reading) { 995 goto stop; 996 } 997 state++; 998 break; 999 case GZSTATE_LEN: 1000 { 1001 uLong origlen; 1002 1003 if (z.avail_in < 4) { 1004 if (!done_reading) { 1005 needmore = 1; 1006 continue; 1007 } 1008 maybe_warnx("truncated input"); 1009 goto stop_and_fail; 1010 } 1011 origlen = le32dec(&z.next_in[0]); 1012 1013 if (origlen != out_sub_tot) { 1014 maybe_warnx("invalid compressed" 1015 " data--length error"); 1016 goto stop_and_fail; 1017 } 1018 } 1019 1020 z.avail_in -= 4; 1021 z.next_in += 4; 1022 1023 if (error < 0) { 1024 maybe_warnx("decompression error"); 1025 goto stop_and_fail; 1026 } 1027 state = GZSTATE_MAGIC0; 1028 break; 1029 } 1030 continue; 1031 stop_and_fail: 1032 out_tot = -1; 1033 stop: 1034 break; 1035 } 1036 if (state > GZSTATE_INIT) 1037 inflateEnd(&z); 1038 1039 free(inbufp); 1040 out1: 1041 free(outbufp); 1042 out2: 1043 if (gsizep) 1044 *gsizep = in_tot; 1045 return (out_tot); 1046 } 1047 1048 /* 1049 * set the owner, mode, flags & utimes using the given file descriptor. 1050 * file is only used in possible warning messages. 1051 */ 1052 static void 1053 copymodes(int fd, const struct stat *sbp, const char *file) 1054 { 1055 struct timespec times[2]; 1056 struct stat sb; 1057 1058 /* 1059 * If we have no info on the input, give this file some 1060 * default values and return.. 1061 */ 1062 if (sbp == NULL) { 1063 mode_t mask = umask(022); 1064 1065 (void)fchmod(fd, DEFFILEMODE & ~mask); 1066 (void)umask(mask); 1067 return; 1068 } 1069 sb = *sbp; 1070 1071 /* if the chown fails, remove set-id bits as-per compress(1) */ 1072 if (fchown(fd, sb.st_uid, sb.st_gid) < 0) { 1073 if (errno != EPERM) 1074 maybe_warn("couldn't fchown: %s", file); 1075 sb.st_mode &= ~(S_ISUID|S_ISGID); 1076 } 1077 1078 /* we only allow set-id and the 9 normal permission bits */ 1079 sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO; 1080 if (fchmod(fd, sb.st_mode) < 0) 1081 maybe_warn("couldn't fchmod: %s", file); 1082 1083 times[0] = sb.st_atim; 1084 times[1] = sb.st_mtim; 1085 if (futimens(fd, times) < 0) 1086 maybe_warn("couldn't futimens: %s", file); 1087 1088 /* only try flags if they exist already */ 1089 if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0) 1090 maybe_warn("couldn't fchflags: %s", file); 1091 } 1092 1093 /* what sort of file is this? */ 1094 static enum filetype 1095 file_gettype(u_char *buf) 1096 { 1097 1098 if (buf[0] == GZIP_MAGIC0 && 1099 (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1)) 1100 return FT_GZIP; 1101 else 1102 #ifndef NO_BZIP2_SUPPORT 1103 if (memcmp(buf, BZIP2_MAGIC, 3) == 0 && 1104 buf[3] >= '0' && buf[3] <= '9') 1105 return FT_BZIP2; 1106 else 1107 #endif 1108 #ifndef NO_COMPRESS_SUPPORT 1109 if (memcmp(buf, Z_MAGIC, 2) == 0) 1110 return FT_Z; 1111 else 1112 #endif 1113 #ifndef NO_PACK_SUPPORT 1114 if (memcmp(buf, PACK_MAGIC, 2) == 0) 1115 return FT_PACK; 1116 else 1117 #endif 1118 #ifndef NO_XZ_SUPPORT 1119 if (memcmp(buf, XZ_MAGIC, 4) == 0) /* XXX: We only have 4 bytes */ 1120 return FT_XZ; 1121 else 1122 #endif 1123 #ifndef NO_LZ_SUPPORT 1124 if (memcmp(buf, LZ_MAGIC, 4) == 0) 1125 return FT_LZ; 1126 else 1127 #endif 1128 return FT_UNKNOWN; 1129 } 1130 1131 /* check the outfile is OK. */ 1132 static int 1133 check_outfile(const char *outfile) 1134 { 1135 struct stat sb; 1136 int ok = 1; 1137 1138 if (lflag == 0 && stat(outfile, &sb) == 0) { 1139 if (fflag) 1140 unlink(outfile); 1141 else if (isatty(STDIN_FILENO)) { 1142 char ans[10] = { 'n', '\0' }; /* default */ 1143 1144 fprintf(stderr, "%s already exists -- do you wish to " 1145 "overwrite (y or n)? " , outfile); 1146 (void)fgets(ans, sizeof(ans) - 1, stdin); 1147 if (ans[0] != 'y' && ans[0] != 'Y') { 1148 fprintf(stderr, "\tnot overwriting\n"); 1149 ok = 0; 1150 } else 1151 unlink(outfile); 1152 } else { 1153 maybe_warnx("%s already exists -- skipping", outfile); 1154 ok = 0; 1155 } 1156 } 1157 return ok; 1158 } 1159 1160 static void 1161 unlink_input(const char *file, const struct stat *sb) 1162 { 1163 struct stat nsb; 1164 1165 if (kflag) 1166 return; 1167 if (stat(file, &nsb) != 0) 1168 /* Must be gone already */ 1169 return; 1170 if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino) 1171 /* Definitely a different file */ 1172 return; 1173 unlink(file); 1174 } 1175 1176 static void 1177 got_sigint(int signo __unused) 1178 { 1179 1180 if (remove_file != NULL) 1181 unlink(remove_file); 1182 _exit(2); 1183 } 1184 1185 static void 1186 got_siginfo(int signo __unused) 1187 { 1188 1189 print_info = 1; 1190 } 1191 1192 static void 1193 setup_signals(void) 1194 { 1195 1196 signal(SIGINFO, got_siginfo); 1197 signal(SIGINT, got_sigint); 1198 } 1199 1200 static void 1201 infile_newdata(size_t newdata) 1202 { 1203 1204 infile_current += newdata; 1205 } 1206 1207 static void 1208 infile_set(const char *newinfile, off_t total) 1209 { 1210 1211 if (newinfile) 1212 infile = newinfile; 1213 infile_total = total; 1214 } 1215 1216 static void 1217 infile_clear(void) 1218 { 1219 1220 infile = NULL; 1221 infile_total = infile_current = 0; 1222 } 1223 1224 static const suffixes_t * 1225 check_suffix(char *file, int xlate) 1226 { 1227 const suffixes_t *s; 1228 int len = strlen(file); 1229 char *sp; 1230 1231 for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) { 1232 /* if it doesn't fit in "a.suf", don't bother */ 1233 if (s->ziplen >= len) 1234 continue; 1235 sp = file + len - s->ziplen; 1236 if (strcmp(s->zipped, sp) != 0) 1237 continue; 1238 if (xlate) 1239 strcpy(sp, s->normal); 1240 return s; 1241 } 1242 return NULL; 1243 } 1244 1245 /* 1246 * compress the given file: create a corresponding .gz file and remove the 1247 * original. 1248 */ 1249 static off_t 1250 file_compress(char *file, char *outfile, size_t outsize) 1251 { 1252 int in; 1253 int out; 1254 off_t size, in_size; 1255 struct stat isb, osb; 1256 const suffixes_t *suff; 1257 1258 in = open(file, O_RDONLY); 1259 if (in == -1) { 1260 maybe_warn("can't open %s", file); 1261 return (-1); 1262 } 1263 1264 if (fstat(in, &isb) != 0) { 1265 maybe_warn("couldn't stat: %s", file); 1266 close(in); 1267 return (-1); 1268 } 1269 1270 if (fstat(in, &isb) != 0) { 1271 close(in); 1272 maybe_warn("can't stat %s", file); 1273 return -1; 1274 } 1275 infile_set(file, isb.st_size); 1276 1277 if (cflag == 0) { 1278 if (isb.st_nlink > 1 && fflag == 0) { 1279 maybe_warnx("%s has %ju other link%s -- " 1280 "skipping", file, 1281 (uintmax_t)isb.st_nlink - 1, 1282 isb.st_nlink == 1 ? "" : "s"); 1283 close(in); 1284 return -1; 1285 } 1286 1287 if (fflag == 0 && (suff = check_suffix(file, 0)) && 1288 suff->zipped[0] != 0) { 1289 maybe_warnx("%s already has %s suffix -- unchanged", 1290 file, suff->zipped); 1291 close(in); 1292 return (-1); 1293 } 1294 1295 /* Add (usually) .gz to filename */ 1296 if ((size_t)snprintf(outfile, outsize, "%s%s", 1297 file, suffixes[0].zipped) >= outsize) 1298 memcpy(outfile + outsize - suffixes[0].ziplen - 1, 1299 suffixes[0].zipped, suffixes[0].ziplen + 1); 1300 1301 if (check_outfile(outfile) == 0) { 1302 close(in); 1303 return (-1); 1304 } 1305 } 1306 1307 if (cflag == 0) { 1308 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600); 1309 if (out == -1) { 1310 maybe_warn("could not create output: %s", outfile); 1311 fclose(stdin); 1312 return (-1); 1313 } 1314 remove_file = outfile; 1315 } else 1316 out = STDOUT_FILENO; 1317 1318 in_size = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime); 1319 1320 (void)close(in); 1321 1322 /* 1323 * If there was an error, in_size will be -1. 1324 * If we compressed to stdout, just return the size. 1325 * Otherwise stat the file and check it is the correct size. 1326 * We only blow away the file if we can stat the output and it 1327 * has the expected size. 1328 */ 1329 if (cflag != 0) 1330 return in_size == -1 ? -1 : size; 1331 1332 if (fstat(out, &osb) != 0) { 1333 maybe_warn("couldn't stat: %s", outfile); 1334 goto bad_outfile; 1335 } 1336 1337 if (osb.st_size != size) { 1338 maybe_warnx("output file: %s wrong size (%ju != %ju), deleting", 1339 outfile, (uintmax_t)osb.st_size, (uintmax_t)size); 1340 goto bad_outfile; 1341 } 1342 1343 copymodes(out, &isb, outfile); 1344 remove_file = NULL; 1345 if (close(out) == -1) 1346 maybe_warn("couldn't close output"); 1347 1348 /* output is good, ok to delete input */ 1349 unlink_input(file, &isb); 1350 return (size); 1351 1352 bad_outfile: 1353 if (close(out) == -1) 1354 maybe_warn("couldn't close output"); 1355 1356 maybe_warnx("leaving original %s", file); 1357 unlink(outfile); 1358 return (size); 1359 } 1360 1361 /* uncompress the given file and remove the original */ 1362 static off_t 1363 file_uncompress(char *file, char *outfile, size_t outsize) 1364 { 1365 struct stat isb, osb; 1366 off_t size; 1367 ssize_t rbytes; 1368 unsigned char fourbytes[4]; 1369 enum filetype method; 1370 int fd, ofd, zfd = -1; 1371 int error; 1372 size_t in_size; 1373 ssize_t rv; 1374 time_t timestamp = 0; 1375 char name[PATH_MAX + 1]; 1376 1377 /* gather the old name info */ 1378 1379 fd = open(file, O_RDONLY); 1380 if (fd < 0) { 1381 maybe_warn("can't open %s", file); 1382 goto lose; 1383 } 1384 if (fstat(fd, &isb) != 0) { 1385 maybe_warn("can't stat %s", file); 1386 goto lose; 1387 } 1388 if (S_ISREG(isb.st_mode)) 1389 in_size = isb.st_size; 1390 else 1391 in_size = 0; 1392 infile_set(file, in_size); 1393 1394 strlcpy(outfile, file, outsize); 1395 if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) { 1396 maybe_warnx("%s: unknown suffix -- ignored", file); 1397 goto lose; 1398 } 1399 1400 rbytes = read(fd, fourbytes, sizeof fourbytes); 1401 if (rbytes != sizeof fourbytes) { 1402 /* we don't want to fail here. */ 1403 if (fflag) 1404 goto lose; 1405 if (rbytes == -1) 1406 maybe_warn("can't read %s", file); 1407 else 1408 goto unexpected_EOF; 1409 goto lose; 1410 } 1411 infile_newdata(rbytes); 1412 1413 method = file_gettype(fourbytes); 1414 if (fflag == 0 && method == FT_UNKNOWN) { 1415 maybe_warnx("%s: not in gzip format", file); 1416 goto lose; 1417 } 1418 1419 1420 if (method == FT_GZIP && Nflag) { 1421 unsigned char ts[4]; /* timestamp */ 1422 1423 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP); 1424 if (rv >= 0 && rv < (ssize_t)(sizeof ts)) 1425 goto unexpected_EOF; 1426 if (rv == -1) { 1427 if (!fflag) 1428 maybe_warn("can't read %s", file); 1429 goto lose; 1430 } 1431 infile_newdata(rv); 1432 timestamp = le32dec(&ts[0]); 1433 1434 if (fourbytes[3] & ORIG_NAME) { 1435 rbytes = pread(fd, name, sizeof(name) - 1, GZIP_ORIGNAME); 1436 if (rbytes < 0) { 1437 maybe_warn("can't read %s", file); 1438 goto lose; 1439 } 1440 if (name[0] != '\0') { 1441 char *dp, *nf; 1442 1443 /* Make sure that name is NUL-terminated */ 1444 name[rbytes] = '\0'; 1445 1446 /* strip saved directory name */ 1447 nf = strrchr(name, '/'); 1448 if (nf == NULL) 1449 nf = name; 1450 else 1451 nf++; 1452 1453 /* preserve original directory name */ 1454 dp = strrchr(file, '/'); 1455 if (dp == NULL) 1456 dp = file; 1457 else 1458 dp++; 1459 snprintf(outfile, outsize, "%.*s%.*s", 1460 (int) (dp - file), 1461 file, (int) rbytes, nf); 1462 } 1463 } 1464 } 1465 lseek(fd, 0, SEEK_SET); 1466 1467 if (cflag == 0 || lflag) { 1468 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) { 1469 maybe_warnx("%s has %ju other links -- skipping", 1470 file, (uintmax_t)isb.st_nlink - 1); 1471 goto lose; 1472 } 1473 if (nflag == 0 && timestamp) 1474 isb.st_mtime = timestamp; 1475 if (check_outfile(outfile) == 0) 1476 goto lose; 1477 } 1478 1479 if (cflag) 1480 zfd = STDOUT_FILENO; 1481 else if (lflag) 1482 zfd = -1; 1483 else { 1484 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600); 1485 if (zfd == STDOUT_FILENO) { 1486 /* We won't close STDOUT_FILENO later... */ 1487 zfd = dup(zfd); 1488 close(STDOUT_FILENO); 1489 } 1490 if (zfd == -1) { 1491 maybe_warn("can't open %s", outfile); 1492 goto lose; 1493 } 1494 remove_file = outfile; 1495 } 1496 1497 switch (method) { 1498 #ifndef NO_BZIP2_SUPPORT 1499 case FT_BZIP2: 1500 /* XXX */ 1501 if (lflag) { 1502 maybe_warnx("no -l with bzip2 files"); 1503 goto lose; 1504 } 1505 1506 size = unbzip2(fd, zfd, NULL, 0, NULL); 1507 break; 1508 #endif 1509 1510 #ifndef NO_COMPRESS_SUPPORT 1511 case FT_Z: { 1512 FILE *in, *out; 1513 1514 /* XXX */ 1515 if (lflag) { 1516 maybe_warnx("no -l with Lempel-Ziv files"); 1517 goto lose; 1518 } 1519 1520 if ((in = zdopen(fd)) == NULL) { 1521 maybe_warn("zdopen for read: %s", file); 1522 goto lose; 1523 } 1524 1525 out = fdopen(dup(zfd), "w"); 1526 if (out == NULL) { 1527 maybe_warn("fdopen for write: %s", outfile); 1528 fclose(in); 1529 goto lose; 1530 } 1531 1532 size = zuncompress(in, out, NULL, 0, NULL); 1533 /* need to fclose() if ferror() is true... */ 1534 error = ferror(in); 1535 if (error | fclose(in)) { 1536 if (error) 1537 maybe_warn("failed infile"); 1538 else 1539 maybe_warn("failed infile fclose"); 1540 if (cflag == 0) 1541 unlink(outfile); 1542 (void)fclose(out); 1543 goto lose; 1544 } 1545 if (fclose(out) != 0) { 1546 maybe_warn("failed outfile fclose"); 1547 if (cflag == 0) 1548 unlink(outfile); 1549 goto lose; 1550 } 1551 break; 1552 } 1553 #endif 1554 1555 #ifndef NO_PACK_SUPPORT 1556 case FT_PACK: 1557 if (lflag) { 1558 maybe_warnx("no -l with packed files"); 1559 goto lose; 1560 } 1561 1562 size = unpack(fd, zfd, NULL, 0, NULL); 1563 break; 1564 #endif 1565 1566 #ifndef NO_XZ_SUPPORT 1567 case FT_XZ: 1568 if (lflag) { 1569 size = unxz_len(fd); 1570 if (!tflag) { 1571 print_list_out(in_size, size, file); 1572 close(fd); 1573 return -1; 1574 } 1575 } else 1576 size = unxz(fd, zfd, NULL, 0, NULL); 1577 break; 1578 #endif 1579 1580 #ifndef NO_LZ_SUPPORT 1581 case FT_LZ: 1582 if (lflag) { 1583 maybe_warnx("no -l with lzip files"); 1584 goto lose; 1585 } 1586 size = unlz(fd, zfd, NULL, 0, NULL); 1587 break; 1588 #endif 1589 case FT_UNKNOWN: 1590 if (lflag) { 1591 maybe_warnx("no -l for unknown filetypes"); 1592 goto lose; 1593 } 1594 size = cat_fd(NULL, 0, NULL, fd); 1595 break; 1596 default: 1597 if (lflag) { 1598 print_list(fd, in_size, outfile, isb.st_mtime); 1599 if (!tflag) { 1600 close(fd); 1601 return -1; /* XXX */ 1602 } 1603 } 1604 1605 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file); 1606 break; 1607 } 1608 1609 if (close(fd) != 0) 1610 maybe_warn("couldn't close input"); 1611 if (zfd != STDOUT_FILENO && close(zfd) != 0) 1612 maybe_warn("couldn't close output"); 1613 1614 if (size == -1) { 1615 if (cflag == 0) 1616 unlink(outfile); 1617 maybe_warnx("%s: uncompress failed", file); 1618 return -1; 1619 } 1620 1621 /* if testing, or we uncompressed to stdout, this is all we need */ 1622 if (tflag) 1623 return size; 1624 /* if we are uncompressing to stdin, don't remove the file. */ 1625 if (cflag) 1626 return size; 1627 1628 /* 1629 * if we create a file... 1630 */ 1631 /* 1632 * if we can't stat the file don't remove the file. 1633 */ 1634 1635 ofd = open(outfile, O_RDWR, 0); 1636 if (ofd == -1) { 1637 maybe_warn("couldn't open (leaving original): %s", 1638 outfile); 1639 return -1; 1640 } 1641 if (fstat(ofd, &osb) != 0) { 1642 maybe_warn("couldn't stat (leaving original): %s", 1643 outfile); 1644 close(ofd); 1645 return -1; 1646 } 1647 if (osb.st_size != size) { 1648 maybe_warnx("stat gave different size: %ju != %ju (leaving original)", 1649 (uintmax_t)size, (uintmax_t)osb.st_size); 1650 close(ofd); 1651 unlink(outfile); 1652 return -1; 1653 } 1654 copymodes(ofd, &isb, outfile); 1655 remove_file = NULL; 1656 close(ofd); 1657 unlink_input(file, &isb); 1658 return size; 1659 1660 unexpected_EOF: 1661 maybe_warnx("%s: unexpected end of file", file); 1662 lose: 1663 if (fd != -1) 1664 close(fd); 1665 if (zfd != -1 && zfd != STDOUT_FILENO) 1666 close(zfd); 1667 return -1; 1668 } 1669 1670 static void 1671 check_siginfo(void) 1672 { 1673 if (print_info == 0) 1674 return; 1675 if (infile) { 1676 if (infile_total) { 1677 int pcent = (int)((100.0 * infile_current) / infile_total); 1678 1679 fprintf(stderr, "%s: done %llu/%llu bytes %d%%\n", 1680 infile, (unsigned long long)infile_current, 1681 (unsigned long long)infile_total, pcent); 1682 } else 1683 fprintf(stderr, "%s: done %llu bytes\n", 1684 infile, (unsigned long long)infile_current); 1685 } 1686 print_info = 0; 1687 } 1688 1689 static off_t 1690 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd) 1691 { 1692 char buf[BUFLEN]; 1693 off_t in_tot; 1694 ssize_t w; 1695 1696 in_tot = count; 1697 w = write_retry(STDOUT_FILENO, prepend, count); 1698 if (w == -1 || (size_t)w != count) { 1699 maybe_warn("write to stdout"); 1700 return -1; 1701 } 1702 for (;;) { 1703 ssize_t rv; 1704 1705 rv = read(fd, buf, sizeof buf); 1706 if (rv == 0) 1707 break; 1708 if (rv < 0) { 1709 maybe_warn("read from fd %d", fd); 1710 break; 1711 } 1712 infile_newdata(rv); 1713 1714 if (write_retry(STDOUT_FILENO, buf, rv) != rv) { 1715 maybe_warn("write to stdout"); 1716 break; 1717 } 1718 in_tot += rv; 1719 } 1720 1721 if (gsizep) 1722 *gsizep = in_tot; 1723 return (in_tot); 1724 } 1725 1726 static void 1727 handle_stdin(void) 1728 { 1729 struct stat isb; 1730 unsigned char fourbytes[4]; 1731 size_t in_size; 1732 off_t usize, gsize; 1733 enum filetype method; 1734 ssize_t bytes_read; 1735 #ifndef NO_COMPRESS_SUPPORT 1736 FILE *in; 1737 #endif 1738 1739 if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) { 1740 maybe_warnx("standard input is a terminal -- ignoring"); 1741 goto out; 1742 } 1743 1744 if (fstat(STDIN_FILENO, &isb) < 0) { 1745 maybe_warn("fstat"); 1746 goto out; 1747 } 1748 if (S_ISREG(isb.st_mode)) 1749 in_size = isb.st_size; 1750 else 1751 in_size = 0; 1752 infile_set("(stdin)", in_size); 1753 1754 if (lflag) { 1755 print_list(STDIN_FILENO, in_size, infile, isb.st_mtime); 1756 goto out; 1757 } 1758 1759 bytes_read = read_retry(STDIN_FILENO, fourbytes, sizeof fourbytes); 1760 if (bytes_read == -1) { 1761 maybe_warn("can't read stdin"); 1762 goto out; 1763 } else if (bytes_read != sizeof(fourbytes)) { 1764 maybe_warnx("(stdin): unexpected end of file"); 1765 goto out; 1766 } 1767 1768 method = file_gettype(fourbytes); 1769 switch (method) { 1770 default: 1771 if (fflag == 0) { 1772 maybe_warnx("unknown compression format"); 1773 goto out; 1774 } 1775 usize = cat_fd(fourbytes, sizeof fourbytes, &gsize, STDIN_FILENO); 1776 break; 1777 case FT_GZIP: 1778 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO, 1779 (char *)fourbytes, sizeof fourbytes, &gsize, "(stdin)"); 1780 break; 1781 #ifndef NO_BZIP2_SUPPORT 1782 case FT_BZIP2: 1783 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO, 1784 (char *)fourbytes, sizeof fourbytes, &gsize); 1785 break; 1786 #endif 1787 #ifndef NO_COMPRESS_SUPPORT 1788 case FT_Z: 1789 if ((in = zdopen(STDIN_FILENO)) == NULL) { 1790 maybe_warnx("zopen of stdin"); 1791 goto out; 1792 } 1793 1794 usize = zuncompress(in, stdout, (char *)fourbytes, 1795 sizeof fourbytes, &gsize); 1796 fclose(in); 1797 break; 1798 #endif 1799 #ifndef NO_PACK_SUPPORT 1800 case FT_PACK: 1801 usize = unpack(STDIN_FILENO, STDOUT_FILENO, 1802 (char *)fourbytes, sizeof fourbytes, &gsize); 1803 break; 1804 #endif 1805 #ifndef NO_XZ_SUPPORT 1806 case FT_XZ: 1807 usize = unxz(STDIN_FILENO, STDOUT_FILENO, 1808 (char *)fourbytes, sizeof fourbytes, &gsize); 1809 break; 1810 #endif 1811 #ifndef NO_LZ_SUPPORT 1812 case FT_LZ: 1813 usize = unlz(STDIN_FILENO, STDOUT_FILENO, 1814 (char *)fourbytes, sizeof fourbytes, &gsize); 1815 break; 1816 #endif 1817 } 1818 1819 if (vflag && !tflag && usize != -1 && gsize != -1) 1820 print_verbage(NULL, NULL, usize, gsize); 1821 if (vflag && tflag) 1822 print_test("(stdin)", usize != -1); 1823 1824 out: 1825 infile_clear(); 1826 } 1827 1828 static void 1829 handle_stdout(void) 1830 { 1831 off_t gsize; 1832 off_t usize; 1833 struct stat sb; 1834 time_t systime; 1835 uint32_t mtime; 1836 int ret; 1837 1838 infile_set("(stdout)", 0); 1839 1840 if (fflag == 0 && isatty(STDOUT_FILENO)) { 1841 maybe_warnx("standard output is a terminal -- ignoring"); 1842 return; 1843 } 1844 1845 /* If stdin is a file use its mtime, otherwise use current time */ 1846 ret = fstat(STDIN_FILENO, &sb); 1847 if (ret < 0) { 1848 maybe_warn("Can't stat stdin"); 1849 return; 1850 } 1851 1852 if (S_ISREG(sb.st_mode)) { 1853 infile_set("(stdout)", sb.st_size); 1854 mtime = (uint32_t)sb.st_mtime; 1855 } else { 1856 systime = time(NULL); 1857 if (systime == -1) { 1858 maybe_warn("time"); 1859 return; 1860 } 1861 mtime = (uint32_t)systime; 1862 } 1863 1864 usize = 1865 gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime); 1866 if (vflag && !tflag && usize != -1 && gsize != -1) 1867 print_verbage(NULL, NULL, usize, gsize); 1868 } 1869 1870 /* do what is asked for, for the path name */ 1871 static void 1872 handle_pathname(char *path) 1873 { 1874 char *opath = path, *s = NULL; 1875 ssize_t len; 1876 int slen; 1877 struct stat sb; 1878 1879 /* check for stdout/stdin */ 1880 if (path[0] == '-' && path[1] == '\0') { 1881 if (dflag) 1882 handle_stdin(); 1883 else 1884 handle_stdout(); 1885 return; 1886 } 1887 1888 retry: 1889 if (stat(path, &sb) != 0 || (fflag == 0 && cflag == 0 && 1890 lstat(path, &sb) != 0)) { 1891 /* lets try <path>.gz if we're decompressing */ 1892 if (dflag && s == NULL && errno == ENOENT) { 1893 len = strlen(path); 1894 slen = suffixes[0].ziplen; 1895 s = malloc(len + slen + 1); 1896 if (s == NULL) 1897 maybe_err("malloc"); 1898 memcpy(s, path, len); 1899 memcpy(s + len, suffixes[0].zipped, slen + 1); 1900 path = s; 1901 goto retry; 1902 } 1903 maybe_warn("can't stat: %s", opath); 1904 goto out; 1905 } 1906 1907 if (S_ISDIR(sb.st_mode)) { 1908 if (rflag) 1909 handle_dir(path); 1910 else 1911 maybe_warnx("%s is a directory", path); 1912 goto out; 1913 } 1914 1915 if (S_ISREG(sb.st_mode)) 1916 handle_file(path, &sb); 1917 else 1918 maybe_warnx("%s is not a regular file", path); 1919 1920 out: 1921 if (s) 1922 free(s); 1923 } 1924 1925 /* compress/decompress a file */ 1926 static void 1927 handle_file(char *file, struct stat *sbp) 1928 { 1929 off_t usize, gsize; 1930 char outfile[PATH_MAX]; 1931 1932 infile_set(file, sbp->st_size); 1933 if (dflag) { 1934 usize = file_uncompress(file, outfile, sizeof(outfile)); 1935 if (vflag && tflag) 1936 print_test(file, usize != -1); 1937 if (usize == -1) 1938 return; 1939 gsize = sbp->st_size; 1940 } else { 1941 gsize = file_compress(file, outfile, sizeof(outfile)); 1942 if (gsize == -1) 1943 return; 1944 usize = sbp->st_size; 1945 } 1946 infile_clear(); 1947 1948 if (vflag && !tflag) 1949 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize); 1950 } 1951 1952 /* this is used with -r to recursively descend directories */ 1953 static void 1954 handle_dir(char *dir) 1955 { 1956 char *path_argv[2]; 1957 FTS *fts; 1958 FTSENT *entry; 1959 1960 path_argv[0] = dir; 1961 path_argv[1] = 0; 1962 fts = fts_open(path_argv, FTS_PHYSICAL | FTS_NOCHDIR, NULL); 1963 if (fts == NULL) { 1964 warn("couldn't fts_open %s", dir); 1965 return; 1966 } 1967 1968 while (errno = 0, (entry = fts_read(fts))) { 1969 switch(entry->fts_info) { 1970 case FTS_D: 1971 case FTS_DP: 1972 continue; 1973 1974 case FTS_DNR: 1975 case FTS_ERR: 1976 case FTS_NS: 1977 maybe_warn("%s", entry->fts_path); 1978 continue; 1979 case FTS_F: 1980 handle_file(entry->fts_path, entry->fts_statp); 1981 } 1982 } 1983 if (errno != 0) 1984 warn("error with fts_read %s", dir); 1985 (void)fts_close(fts); 1986 } 1987 1988 /* print a ratio - size reduction as a fraction of uncompressed size */ 1989 static void 1990 print_ratio(off_t in, off_t out, FILE *where) 1991 { 1992 int percent10; /* 10 * percent */ 1993 off_t diff; 1994 char buff[8]; 1995 int len; 1996 1997 diff = in - out/2; 1998 if (in == 0 && out == 0) 1999 percent10 = 0; 2000 else if (diff < 0) 2001 /* 2002 * Output is more than double size of input! print -99.9% 2003 * Quite possibly we've failed to get the original size. 2004 */ 2005 percent10 = -999; 2006 else { 2007 /* 2008 * We only need 12 bits of result from the final division, 2009 * so reduce the values until a 32bit division will suffice. 2010 */ 2011 while (in > 0x100000) { 2012 diff >>= 1; 2013 in >>= 1; 2014 } 2015 if (in != 0) 2016 percent10 = ((u_int)diff * 2000) / (u_int)in - 1000; 2017 else 2018 percent10 = 0; 2019 } 2020 2021 len = snprintf(buff, sizeof buff, "%2.2d.", percent10); 2022 /* Move the '.' to before the last digit */ 2023 buff[len - 1] = buff[len - 2]; 2024 buff[len - 2] = '.'; 2025 fprintf(where, "%5s%%", buff); 2026 } 2027 2028 /* print compression statistics, and the new name (if there is one!) */ 2029 static void 2030 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize) 2031 { 2032 if (file) 2033 fprintf(stderr, "%s:%s ", file, 2034 strlen(file) < 7 ? "\t\t" : "\t"); 2035 print_ratio(usize, gsize, stderr); 2036 if (nfile) 2037 fprintf(stderr, " -- replaced with %s", nfile); 2038 fprintf(stderr, "\n"); 2039 fflush(stderr); 2040 } 2041 2042 /* print test results */ 2043 static void 2044 print_test(const char *file, int ok) 2045 { 2046 2047 if (exit_value == 0 && ok == 0) 2048 exit_value = 1; 2049 fprintf(stderr, "%s:%s %s\n", file, 2050 strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK"); 2051 fflush(stderr); 2052 } 2053 2054 /* print a file's info ala --list */ 2055 /* eg: 2056 compressed uncompressed ratio uncompressed_name 2057 354841 1679360 78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar 2058 */ 2059 static void 2060 print_list(int fd, off_t out, const char *outfile, time_t ts) 2061 { 2062 static int first = 1; 2063 static off_t in_tot, out_tot; 2064 uint32_t crc = 0; 2065 off_t in = 0, rv; 2066 2067 if (first) { 2068 if (vflag) 2069 printf("method crc date time "); 2070 if (qflag == 0) 2071 printf(" compressed uncompressed " 2072 "ratio uncompressed_name\n"); 2073 } 2074 first = 0; 2075 2076 /* print totals? */ 2077 if (fd == -1) { 2078 in = in_tot; 2079 out = out_tot; 2080 } else 2081 { 2082 /* read the last 4 bytes - this is the uncompressed size */ 2083 rv = lseek(fd, (off_t)(-8), SEEK_END); 2084 if (rv != -1) { 2085 unsigned char buf[8]; 2086 uint32_t usize; 2087 2088 rv = read(fd, (char *)buf, sizeof(buf)); 2089 if (rv == -1) 2090 maybe_warn("read of uncompressed size"); 2091 else if (rv != sizeof(buf)) 2092 maybe_warnx("read of uncompressed size"); 2093 2094 else { 2095 usize = le32dec(&buf[4]); 2096 in = (off_t)usize; 2097 crc = le32dec(&buf[0]); 2098 } 2099 } 2100 } 2101 2102 if (vflag && fd == -1) 2103 printf(" "); 2104 else if (vflag) { 2105 char *date = ctime(&ts); 2106 2107 /* skip the day, 1/100th second, and year */ 2108 date += 4; 2109 date[12] = 0; 2110 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date); 2111 } 2112 in_tot += in; 2113 out_tot += out; 2114 print_list_out(out, in, outfile); 2115 } 2116 2117 static void 2118 print_list_out(off_t out, off_t in, const char *outfile) 2119 { 2120 printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in); 2121 print_ratio(in, out, stdout); 2122 printf(" %s\n", outfile); 2123 } 2124 2125 /* display the usage of NetBSD gzip */ 2126 static void 2127 usage(void) 2128 { 2129 2130 fprintf(stderr, "%s\n", gzip_version); 2131 fprintf(stderr, 2132 "usage: %s [-123456789acdfhklLNnqrtVv] [-S .suffix] [<file> [<file> ...]]\n" 2133 " -1 --fast fastest (worst) compression\n" 2134 " -2 .. -8 set compression level\n" 2135 " -9 --best best (slowest) compression\n" 2136 " -c --stdout write to stdout, keep original files\n" 2137 " --to-stdout\n" 2138 " -d --decompress uncompress files\n" 2139 " --uncompress\n" 2140 " -f --force force overwriting & compress links\n" 2141 " -h --help display this help\n" 2142 " -k --keep don't delete input files during operation\n" 2143 " -l --list list compressed file contents\n" 2144 " -N --name save or restore original file name and time stamp\n" 2145 " -n --no-name don't save original file name or time stamp\n" 2146 " -q --quiet output no warnings\n" 2147 " -r --recursive recursively compress files in directories\n" 2148 " -S .suf use suffix .suf instead of .gz\n" 2149 " --suffix .suf\n" 2150 " -t --test test compressed file\n" 2151 " -V --version display program version\n" 2152 " -v --verbose print extra statistics\n", 2153 getprogname()); 2154 exit(0); 2155 } 2156 2157 /* display the license information of FreeBSD gzip */ 2158 static void 2159 display_license(void) 2160 { 2161 2162 fprintf(stderr, "%s (based on NetBSD gzip 20150113)\n", gzip_version); 2163 fprintf(stderr, "%s\n", gzip_copyright); 2164 exit(0); 2165 } 2166 2167 /* display the version of NetBSD gzip */ 2168 static void 2169 display_version(void) 2170 { 2171 2172 fprintf(stderr, "%s\n", gzip_version); 2173 exit(0); 2174 } 2175 2176 #ifndef NO_BZIP2_SUPPORT 2177 #include "unbzip2.c" 2178 #endif 2179 #ifndef NO_COMPRESS_SUPPORT 2180 #include "zuncompress.c" 2181 #endif 2182 #ifndef NO_PACK_SUPPORT 2183 #include "unpack.c" 2184 #endif 2185 #ifndef NO_XZ_SUPPORT 2186 #include "unxz.c" 2187 #endif 2188 #ifndef NO_LZ_SUPPORT 2189 #include "unlz.c" 2190 #endif 2191 2192 static ssize_t 2193 read_retry(int fd, void *buf, size_t sz) 2194 { 2195 char *cp = buf; 2196 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2197 2198 while (left > 0) { 2199 ssize_t ret; 2200 2201 ret = read(fd, cp, left); 2202 if (ret == -1) { 2203 return ret; 2204 } else if (ret == 0) { 2205 break; /* EOF */ 2206 } 2207 cp += ret; 2208 left -= ret; 2209 } 2210 2211 return sz - left; 2212 } 2213 2214 static ssize_t 2215 write_retry(int fd, const void *buf, size_t sz) 2216 { 2217 const char *cp = buf; 2218 size_t left = MIN(sz, (size_t) SSIZE_MAX); 2219 2220 while (left > 0) { 2221 ssize_t ret; 2222 2223 ret = write(fd, cp, left); 2224 if (ret == -1) { 2225 return ret; 2226 } else if (ret == 0) { 2227 abort(); /* Can't happen */ 2228 } 2229 cp += ret; 2230 left -= ret; 2231 } 2232 2233 return sz - left; 2234 } 2235