usr.bin/gzip/gzip.c

   1 /*      $NetBSD: gzip.c,v 1.94 2009/04/12 10:31:14 lukem Exp $  */
   2 /*      $DragonFly: src/usr.bin/gzip/gzip.c,v 1.7 2007/12/06 19:54:52 hasso Exp $ */
   3
   4 /*
   5  * Copyright (c) 1997, 1998, 2003, 2004, 2006 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 /*
  31  * gzip.c -- GPL free gzip using zlib.
  32  *
  33  * RFC 1950 covers the zlib format
  34  * RFC 1951 covers the deflate format
  35  * RFC 1952 covers the gzip format
  36  *
  37  * TODO:
  38  *      - use mmap where possible
  39  *      - handle some signals better (remove outfile?)
  40  *      - make bzip2/compress -v/-t/-l support work as well as possible
  41  */
  42
  43 #include <sys/param.h>
  44 #include <sys/stat.h>
  45 #include <sys/time.h>
  46
  47 #include <err.h>
  48 #include <errno.h>
  49 #include <fcntl.h>
  50 #include <fts.h>
  51 #include <getopt.h>
  52 #include <inttypes.h>
  53 #include <libgen.h>
  54 #include <stdarg.h>
  55 #include <stdio.h>
  56 #include <stdlib.h>
  57 #include <string.h>
  58 #include <time.h>
  59 #include <unistd.h>
  60 #include <zlib.h>
  61
  62 #ifndef PRIdOFF
  63 #define PRIdOFF PRId64
  64 #endif
  65
  66 #ifndef PRId64
  67 #define PRId64  "lld"
  68 #endif
  69
  70 /* what type of file are we dealing with */
  71 enum filetype {
  72         FT_GZIP,
  73 #ifndef NO_BZIP2_SUPPORT
  74         FT_BZIP2,
  75 #endif
  76 #ifndef NO_COMPRESS_SUPPORT
  77         FT_Z,
  78 #endif
  79 #ifndef NO_PACK_SUPPORT
  80         FT_PACK,
  81 #endif
  82         FT_LAST,
  83         FT_UNKNOWN
  84 };
  85
  86 #ifndef NO_BZIP2_SUPPORT
  87 #include <bzlib.h>
  88
  89 #define BZ2_SUFFIX      ".bz2"
  90 #define BZIP2_MAGIC     "\102\132\150"
  91 #endif
  92
  93 #ifndef NO_COMPRESS_SUPPORT
  94 #define Z_SUFFIX        ".Z"
  95 #define Z_MAGIC         "\037\235"
  96 #endif
  97
  98 #ifndef NO_PACK_SUPPORT
  99 #define PACK_MAGIC      "\037\036"
 100 #endif
 101
 102 #define GZ_SUFFIX       ".gz"
 103
 104 #define BUFLEN          (64 * 1024)
 105
 106 #define GZIP_MAGIC0     0x1F
 107 #define GZIP_MAGIC1     0x8B
 108 #define GZIP_OMAGIC1    0x9E
 109
 110 #define GZIP_TIMESTAMP  (off_t)4
 111 #define GZIP_ORIGNAME   (off_t)10
 112
 113 #define HEAD_CRC        0x02
 114 #define EXTRA_FIELD     0x04
 115 #define ORIG_NAME       0x08
 116 #define COMMENT         0x10
 117
 118 #define OS_CODE         3       /* Unix */
 119
 120 typedef struct {
 121     const char  *zipped;
 122     int         ziplen;
 123     const char  *normal;        /* for unzip - must not be longer than zipped */
 124 } suffixes_t;
 125 static suffixes_t suffixes[] = {
 126 #define SUFFIX(Z, N) {Z, sizeof Z - 1, N}
 127         SUFFIX(GZ_SUFFIX,       ""),    /* Overwritten by -S .xxx */
 128 #ifndef SMALL
 129         SUFFIX(GZ_SUFFIX,       ""),
 130         SUFFIX(".z",            ""),
 131         SUFFIX("-gz",           ""),
 132         SUFFIX("-z",            ""),
 133         SUFFIX("_z",            ""),
 134         SUFFIX(".taz",          ".tar"),
 135         SUFFIX(".tgz",          ".tar"),
 136 #ifndef NO_BZIP2_SUPPORT
 137         SUFFIX(BZ2_SUFFIX,      ""),
 138 #endif
 139 #ifndef NO_COMPRESS_SUPPORT
 140         SUFFIX(Z_SUFFIX,        ""),
 141 #endif
 142         SUFFIX(GZ_SUFFIX,       ""),    /* Overwritten by -S "" */
 143 #endif /* SMALL */
 144 #undef SUFFIX
 145 };
 146 #define NUM_SUFFIXES (sizeof suffixes / sizeof suffixes[0])
 147
 148 #define SUFFIX_MAXLEN  30
 149
 150 static  const char      gzip_version[] = "NetBSD gzip 20060927";
 151
 152 static  int     cflag;                  /* stdout mode */
 153 static  int     dflag;                  /* decompress mode */
 154 static  int     lflag;                  /* list mode */
 155 static  int     numflag = 6;            /* gzip -1..-9 value */
 156
 157 #ifndef SMALL
 158 static  int     fflag;                  /* force mode */
 159 static  int     kflag;                  /* don't delete input files */
 160 static  int     nflag;                  /* don't save name/timestamp */
 161 static  int     Nflag;                  /* don't restore name/timestamp */
 162 static  int     qflag;                  /* quiet mode */
 163 static  int     rflag;                  /* recursive mode */
 164 static  int     tflag;                  /* test */
 165 static  int     vflag;                  /* verbose mode */
 166 #else
 167 #define         qflag   0
 168 #define         tflag   0
 169 #endif
 170
 171 static  int     exit_value = 0;         /* exit value */
 172
 173 static  char    *infile;                /* name of file coming in */
 174
 175 static  void    maybe_err(const char *fmt, ...)
 176     __attribute__((__format__(__printf__, 1, 2)));
 177 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT)
 178 static  void    maybe_errx(const char *fmt, ...)
 179     __attribute__((__format__(__printf__, 1, 2)));
 180 #endif
 181 static  void    maybe_warn(const char *fmt, ...)
 182     __attribute__((__format__(__printf__, 1, 2)));
 183 static  void    maybe_warnx(const char *fmt, ...)
 184     __attribute__((__format__(__printf__, 1, 2)));
 185 static  enum filetype file_gettype(u_char *);
 186 #ifdef SMALL
 187 #define gz_compress(if, of, sz, fn, tm) gz_compress(if, of, sz)
 188 #endif
 189 static  off_t   gz_compress(int, int, off_t *, const char *, uint32_t);
 190 static  off_t   gz_uncompress(int, int, char *, size_t, off_t *, const char *);
 191 static  off_t   file_compress(char *, char *, size_t);
 192 static  off_t   file_uncompress(char *, char *, size_t);
 193 static  void    handle_pathname(char *);
 194 static  void    handle_file(char *, struct stat *);
 195 static  void    handle_stdin(void);
 196 static  void    handle_stdout(void);
 197 static  void    print_ratio(off_t, off_t, FILE *);
 198 static  void    print_list(int fd, off_t, const char *, time_t);
 199 static  void    usage(void);
 200 static  void    display_version(void);
 201 static  const suffixes_t *check_suffix(char *, int);
 202 static  ssize_t read_retry(int, void *, size_t);
 203
 204 #ifdef SMALL
 205 #define unlink_input(f, sb) unlink(f)
 206 #else
 207 static  off_t   cat_fd(unsigned char *, size_t, off_t *, int fd);
 208 static  void    prepend_gzip(char *, int *, char ***);
 209 static  void    handle_dir(char *);
 210 static  void    print_verbage(const char *, const char *, off_t, off_t);
 211 static  void    print_test(const char *, int);
 212 static  void    copymodes(int fd, const struct stat *, const char *file);
 213 static  int     check_outfile(const char *outfile);
 214 #endif
 215
 216 #ifndef NO_BZIP2_SUPPORT
 217 static  off_t   unbzip2(int, int, char *, size_t, off_t *);
 218 #endif
 219
 220 #ifndef NO_COMPRESS_SUPPORT
 221 static  FILE    *zdopen(int);
 222 static  off_t   zuncompress(FILE *, FILE *, char *, size_t, off_t *);
 223 #endif
 224
 225 #ifndef NO_PACK_SUPPORT
 226 static  off_t   unpack(int, int, char *, size_t, off_t *);
 227 #endif
 228
 229 int main(int, char *p[]);
 230
 231 #ifdef SMALL
 232 #define getopt_long(a,b,c,d,e) getopt(a,b,c)
 233 #else
 234 static const struct option longopts[] = {
 235         { "stdout",             no_argument,            0,      'c' },
 236         { "to-stdout",          no_argument,            0,      'c' },
 237         { "decompress",         no_argument,            0,      'd' },
 238         { "uncompress",         no_argument,            0,      'd' },
 239         { "force",              no_argument,            0,      'f' },
 240         { "help",               no_argument,            0,      'h' },
 241         { "keep",               no_argument,            0,      'k' },
 242         { "list",               no_argument,            0,      'l' },
 243         { "no-name",            no_argument,            0,      'n' },
 244         { "name",               no_argument,            0,      'N' },
 245         { "quiet",              no_argument,            0,      'q' },
 246         { "recursive",          no_argument,            0,      'r' },
 247         { "suffix",             required_argument,      0,      'S' },
 248         { "test",               no_argument,            0,      't' },
 249         { "verbose",            no_argument,            0,      'v' },
 250         { "version",            no_argument,            0,      'V' },
 251         { "fast",               no_argument,            0,      '1' },
 252         { "best",               no_argument,            0,      '9' },
 253 #if 0
 254         /*
 255          * This is what else GNU gzip implements.  --ascii isn't useful
 256          * on NetBSD, and I don't care to have a --license.
 257          */
 258         { "ascii",              no_argument,            0,      'a' },
 259         { "license",            no_argument,            0,      'L' },
 260 #endif
 261         { NULL,                 no_argument,            0,      0 },
 262 };
 263 #endif
 264
 265 int
 266 main(int argc, char **argv)
 267 {
 268         const char *progname = getprogname();
 269 #ifndef SMALL
 270         char *gzip;
 271         int len;
 272 #endif
 273         int ch;
 274
 275         /* XXX set up signals */
 276
 277 #ifndef SMALL
 278         if ((gzip = getenv("GZIP")) != NULL)
 279                 prepend_gzip(gzip, &argc, &argv);
 280 #endif
 281
 282         /*
 283          * XXX
 284          * handle being called `gunzip', `zcat' and `gzcat'
 285          */
 286         if (strcmp(progname, "gunzip") == 0)
 287                 dflag = 1;
 288         else if (strcmp(progname, "zcat") == 0 ||
 289                  strcmp(progname, "gzcat") == 0)
 290                 dflag = cflag = 1;
 291
 292 #ifdef SMALL
 293 #define OPT_LIST "123456789cdhltV"
 294 #else
 295 #define OPT_LIST "123456789cdfhklNnqrS:tVv"
 296 #endif
 297
 298         while ((ch = getopt_long(argc, argv, OPT_LIST, longopts, NULL)) != -1) {
 299                 switch (ch) {
 300                 case '1': case '2': case '3':
 301                 case '4': case '5': case '6':
 302                 case '7': case '8': case '9':
 303                         numflag = ch - '0';
 304                         break;
 305                 case 'c':
 306                         cflag = 1;
 307                         break;
 308                 case 'd':
 309                         dflag = 1;
 310                         break;
 311                 case 'l':
 312                         lflag = 1;
 313                         dflag = 1;
 314                         break;
 315                 case 'V':
 316                         display_version();
 317                         /* NOTREACHED */
 318 #ifndef SMALL
 319                 case 'f':
 320                         fflag = 1;
 321                         break;
 322                 case 'k':
 323                         kflag = 1;
 324                         break;
 325                 case 'N':
 326                         nflag = 0;
 327                         Nflag = 1;
 328                         break;
 329                 case 'n':
 330                         nflag = 1;
 331                         Nflag = 0;
 332                         break;
 333                 case 'q':
 334                         qflag = 1;
 335                         break;
 336                 case 'r':
 337                         rflag = 1;
 338                         break;
 339                 case 'S':
 340                         len = strlen(optarg);
 341                         if (len != 0) {
 342                                 if (len >= SUFFIX_MAXLEN)
 343                                         errx(1, "incorrect suffix: '%s'", optarg);
 344                                 suffixes[0].zipped = optarg;
 345                                 suffixes[0].ziplen = len;
 346                         } else {
 347                                 suffixes[NUM_SUFFIXES - 1].zipped = "";
 348                                 suffixes[NUM_SUFFIXES - 1].ziplen = 0;
 349                         }
 350                         break;
 351                 case 't':
 352                         cflag = 1;
 353                         tflag = 1;
 354                         dflag = 1;
 355                         break;
 356                 case 'v':
 357                         vflag = 1;
 358                         break;
 359 #endif
 360                 default:
 361                         usage();
 362                         /* NOTREACHED */
 363                 }
 364         }
 365         argv += optind;
 366         argc -= optind;
 367
 368         if (argc == 0) {
 369                 if (dflag)      /* stdin mode */
 370                         handle_stdin();
 371                 else            /* stdout mode */
 372                         handle_stdout();
 373         } else {
 374                 do {
 375                         handle_pathname(argv[0]);
 376                 } while (*++argv);
 377         }
 378 #ifndef SMALL
 379         if (qflag == 0 && lflag && argc > 1)
 380                 print_list(-1, 0, "(totals)", 0);
 381 #endif
 382         exit(exit_value);
 383 }
 384
 385 /* maybe print a warning */
 386 void
 387 maybe_warn(const char *fmt, ...)
 388 {
 389         va_list ap;
 390
 391         if (qflag == 0) {
 392                 va_start(ap, fmt);
 393                 vwarn(fmt, ap);
 394                 va_end(ap);
 395         }
 396         if (exit_value == 0)
 397                 exit_value = 1;
 398 }
 399
 400 /* ... without an errno. */
 401 void
 402 maybe_warnx(const char *fmt, ...)
 403 {
 404         va_list ap;
 405
 406         if (qflag == 0) {
 407                 va_start(ap, fmt);
 408                 vwarnx(fmt, ap);
 409                 va_end(ap);
 410         }
 411         if (exit_value == 0)
 412                 exit_value = 1;
 413 }
 414
 415 /* maybe print an error */
 416 void
 417 maybe_err(const char *fmt, ...)
 418 {
 419         va_list ap;
 420
 421         if (qflag == 0) {
 422                 va_start(ap, fmt);
 423                 vwarn(fmt, ap);
 424                 va_end(ap);
 425         }
 426         exit(2);
 427 }
 428
 429 #if !defined(NO_BZIP2_SUPPORT) || !defined(NO_PACK_SUPPORT)
 430 /* ... without an errno. */
 431 void
 432 maybe_errx(const char *fmt, ...)
 433 {
 434         va_list ap;
 435
 436         if (qflag == 0) {
 437                 va_start(ap, fmt);
 438                 vwarnx(fmt, ap);
 439                 va_end(ap);
 440         }
 441         exit(2);
 442 }
 443 #endif
 444
 445 #ifndef SMALL
 446 /* split up $GZIP and prepend it to the argument list */
 447 static void
 448 prepend_gzip(char *gzip, int *argc, char ***argv)
 449 {
 450         char *s, **nargv, **ac;
 451         int nenvarg = 0, i;
 452
 453         /* scan how many arguments there are */
 454         for (s = gzip;;) {
 455                 while (*s == ' ' || *s == '\t')
 456                         s++;
 457                 if (*s == 0)
 458                         goto count_done;
 459                 nenvarg++;
 460                 while (*s != ' ' && *s != '\t')
 461                         if (*s++ == 0)
 462                                 goto count_done;
 463         }
 464 count_done:
 465         /* punt early */
 466         if (nenvarg == 0)
 467                 return;
 468
 469         *argc += nenvarg;
 470         ac = *argv;
 471
 472         nargv = (char **)malloc((*argc + 1) * sizeof(char *));
 473         if (nargv == NULL)
 474                 maybe_err("malloc");
 475
 476         /* stash this away */
 477         *argv = nargv;
 478
 479         /* copy the program name first */
 480         i = 0;
 481         nargv[i++] = *(ac++);
 482
 483         /* take a copy of $GZIP and add it to the array */
 484         s = strdup(gzip);
 485         if (s == NULL)
 486                 maybe_err("strdup");
 487         for (;;) {
 488                 /* Skip whitespaces. */
 489                 while (*s == ' ' || *s == '\t')
 490                         s++;
 491                 if (*s == 0)
 492                         goto copy_done;
 493                 nargv[i++] = s;
 494                 /* Find the end of this argument. */
 495                 while (*s != ' ' && *s != '\t')
 496                         if (*s++ == 0)
 497                                 /* Argument followed by NUL. */
 498                                 goto copy_done;
 499                 /* Terminate by overwriting ' ' or '\t' with NUL. */
 500                 *s++ = 0;
 501         }
 502 copy_done:
 503
 504         /* copy the original arguments and a NULL */
 505         while (*ac)
 506                 nargv[i++] = *(ac++);
 507         nargv[i] = NULL;
 508 }
 509 #endif
 510
 511 /* compress input to output. Return bytes read, -1 on error */
 512 static off_t
 513 gz_compress(int in, int out, off_t *gsizep, const char *origname, uint32_t mtime)
 514 {
 515         z_stream z;
 516         char *outbufp, *inbufp;
 517         off_t in_tot = 0, out_tot = 0;
 518         ssize_t in_size;
 519         int i, error;
 520         uLong crc;
 521 #ifdef SMALL
 522         static char header[] = { GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED, 0,
 523                                  0, 0, 0, 0,
 524                                  0, OS_CODE };
 525 #endif
 526
 527         outbufp = malloc(BUFLEN);
 528         inbufp = malloc(BUFLEN);
 529         if (outbufp == NULL || inbufp == NULL) {
 530                 maybe_err("malloc failed");
 531                 goto out;
 532         }
 533
 534         memset(&z, 0, sizeof z);
 535         z.zalloc = Z_NULL;
 536         z.zfree = Z_NULL;
 537         z.opaque = 0;
 538
 539 #ifdef SMALL
 540         memcpy(outbufp, header, sizeof header);
 541         i = sizeof header;
 542 #else
 543         if (nflag != 0) {
 544                 mtime = 0;
 545                 origname = "";
 546         }
 547
 548         i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c%c%c%s",
 549                      GZIP_MAGIC0, GZIP_MAGIC1, Z_DEFLATED,
 550                      *origname ? ORIG_NAME : 0,
 551                      mtime & 0xff,
 552                      (mtime >> 8) & 0xff,
 553                      (mtime >> 16) & 0xff,
 554                      (mtime >> 24) & 0xff,
 555                      numflag == 1 ? 4 : numflag == 9 ? 2 : 0,
 556                      OS_CODE, origname);
 557         if (i >= BUFLEN)
 558                 /* this need PATH_MAX > BUFLEN ... */
 559                 maybe_err("snprintf");
 560         if (*origname)
 561                 i++;
 562 #endif
 563
 564         z.next_out = outbufp + i;
 565         z.avail_out = BUFLEN - i;
 566
 567         error = deflateInit2(&z, numflag, Z_DEFLATED,
 568                              (-MAX_WBITS), 8, Z_DEFAULT_STRATEGY);
 569         if (error != Z_OK) {
 570                 maybe_warnx("deflateInit2 failed");
 571                 in_tot = -1;
 572                 goto out;
 573         }
 574
 575         crc = crc32(0L, Z_NULL, 0);
 576         for (;;) {
 577                 if (z.avail_out == 0) {
 578                         if (write(out, outbufp, BUFLEN) != BUFLEN) {
 579                                 maybe_warn("write");
 580                                 out_tot = -1;
 581                                 goto out;
 582                         }
 583
 584                         out_tot += BUFLEN;
 585                         z.next_out = outbufp;
 586                         z.avail_out = BUFLEN;
 587                 }
 588
 589                 if (z.avail_in == 0) {
 590                         in_size = read(in, inbufp, BUFLEN);
 591                         if (in_size < 0) {
 592                                 maybe_warn("read");
 593                                 in_tot = -1;
 594                                 goto out;
 595                         }
 596                         if (in_size == 0)
 597                                 break;
 598
 599                         crc = crc32(crc, (const Bytef *)inbufp, (unsigned)in_size);
 600                         in_tot += in_size;
 601                         z.next_in = inbufp;
 602                         z.avail_in = in_size;
 603                 }
 604
 605                 error = deflate(&z, Z_NO_FLUSH);
 606                 if (error != Z_OK && error != Z_STREAM_END) {
 607                         maybe_warnx("deflate failed");
 608                         in_tot = -1;
 609                         goto out;
 610                 }
 611         }
 612
 613         /* clean up */
 614         for (;;) {
 615                 size_t len;
 616                 ssize_t w;
 617
 618                 error = deflate(&z, Z_FINISH);
 619                 if (error != Z_OK && error != Z_STREAM_END) {
 620                         maybe_warnx("deflate failed");
 621                         in_tot = -1;
 622                         goto out;
 623                 }
 624
 625                 len = (char *)z.next_out - outbufp;
 626
 627                 w = write(out, outbufp, len);
 628                 if (w == -1 || (size_t)w != len) {
 629                         maybe_warn("write");
 630                         out_tot = -1;
 631                         goto out;
 632                 }
 633                 out_tot += len;
 634                 z.next_out = outbufp;
 635                 z.avail_out = BUFLEN;
 636
 637                 if (error == Z_STREAM_END)
 638                         break;
 639         }
 640
 641         if (deflateEnd(&z) != Z_OK) {
 642                 maybe_warnx("deflateEnd failed");
 643                 in_tot = -1;
 644                 goto out;
 645         }
 646
 647         i = snprintf(outbufp, BUFLEN, "%c%c%c%c%c%c%c%c",
 648                  (int)crc & 0xff,
 649                  (int)(crc >> 8) & 0xff,
 650                  (int)(crc >> 16) & 0xff,
 651                  (int)(crc >> 24) & 0xff,
 652                  (int)in_tot & 0xff,
 653                  (int)(in_tot >> 8) & 0xff,
 654                  (int)(in_tot >> 16) & 0xff,
 655                  (int)(in_tot >> 24) & 0xff);
 656         if (i != 8)
 657                 maybe_err("snprintf");
 658 #if 0
 659         if (in_tot > 0xffffffff)
 660                 maybe_warn("input file size >= 4GB cannot be saved");
 661 #endif
 662         if (write(out, outbufp, i) != i) {
 663                 maybe_warn("write");
 664                 in_tot = -1;
 665         } else
 666                 out_tot += i;
 667
 668 out:
 669         if (inbufp != NULL)
 670                 free(inbufp);
 671         if (outbufp != NULL)
 672                 free(outbufp);
 673         if (gsizep)
 674                 *gsizep = out_tot;
 675         return in_tot;
 676 }
 677
 678 /*
 679  * uncompress input to output then close the input.  return the
 680  * uncompressed size written, and put the compressed sized read
 681  * into `*gsizep'.
 682  */
 683 static off_t
 684 gz_uncompress(int in, int out, char *pre, size_t prelen, off_t *gsizep,
 685               const char *filename)
 686 {
 687         z_stream z;
 688         char *outbufp, *inbufp;
 689         off_t out_tot = -1, in_tot = 0;
 690         uint32_t out_sub_tot = 0;
 691         enum {
 692                 GZSTATE_MAGIC0,
 693                 GZSTATE_MAGIC1,
 694                 GZSTATE_METHOD,
 695                 GZSTATE_FLAGS,
 696                 GZSTATE_SKIPPING,
 697                 GZSTATE_EXTRA,
 698                 GZSTATE_EXTRA2,
 699                 GZSTATE_EXTRA3,
 700                 GZSTATE_ORIGNAME,
 701                 GZSTATE_COMMENT,
 702                 GZSTATE_HEAD_CRC1,
 703                 GZSTATE_HEAD_CRC2,
 704                 GZSTATE_INIT,
 705                 GZSTATE_READ,
 706                 GZSTATE_CRC,
 707                 GZSTATE_LEN,
 708         } state = GZSTATE_MAGIC0;
 709         int flags = 0, skip_count = 0;
 710         int error = Z_STREAM_ERROR, done_reading = 0;
 711         uLong crc = 0;
 712         ssize_t wr;
 713         int needmore = 0;
 714
 715 #define ADVANCE()       { z.next_in++; z.avail_in--; }
 716
 717         if ((outbufp = malloc(BUFLEN)) == NULL) {
 718                 maybe_err("malloc failed");
 719                 goto out2;
 720         }
 721         if ((inbufp = malloc(BUFLEN)) == NULL) {
 722                 maybe_err("malloc failed");
 723                 goto out1;
 724         }
 725
 726         memset(&z, 0, sizeof z);
 727         z.avail_in = prelen;
 728         z.next_in = pre;
 729         z.avail_out = BUFLEN;
 730         z.next_out = outbufp;
 731         z.zalloc = NULL;
 732         z.zfree = NULL;
 733         z.opaque = 0;
 734
 735         in_tot = prelen;
 736         out_tot = 0;
 737
 738         for (;;) {
 739                 if ((z.avail_in == 0 || needmore) && done_reading == 0) {
 740                         ssize_t in_size;
 741
 742                         if (z.avail_in > 0) {
 743                                 memmove(inbufp, z.next_in, z.avail_in);
 744                         }
 745                         z.next_in = inbufp;
 746                         in_size = read(in, z.next_in + z.avail_in,
 747                             BUFLEN - z.avail_in);
 748
 749                         if (in_size == -1) {
 750                                 maybe_warn("failed to read stdin");
 751                                 goto stop_and_fail;
 752                         } else if (in_size == 0) {
 753                                 done_reading = 1;
 754                         }
 755
 756                         z.avail_in += in_size;
 757                         needmore = 0;
 758
 759                         in_tot += in_size;
 760                 }
 761                 if (z.avail_in == 0) {
 762                         if (done_reading && state != GZSTATE_MAGIC0) {
 763                                 maybe_warnx("%s: unexpected end of file",
 764                                             filename);
 765                                 goto stop_and_fail;
 766                         }
 767                         goto stop;
 768                 }
 769                 switch (state) {
 770                 case GZSTATE_MAGIC0:
 771                         if (*z.next_in != GZIP_MAGIC0) {
 772                                 if (in_tot > 0) {
 773                                         maybe_warnx("%s: trailing garbage "
 774                                                     "ignored", filename);
 775                                         goto stop;
 776                                 }
 777                                 maybe_warnx("input not gziped (MAGIC0)");
 778                                 goto stop_and_fail;
 779                         }
 780                         ADVANCE();
 781                         state++;
 782                         out_sub_tot = 0;
 783                         crc = crc32(0L, Z_NULL, 0);
 784                         break;
 785
 786                 case GZSTATE_MAGIC1:
 787                         if (*z.next_in != GZIP_MAGIC1 &&
 788                             *z.next_in != GZIP_OMAGIC1) {
 789                                 maybe_warnx("input not gziped (MAGIC1)");
 790                                 goto stop_and_fail;
 791                         }
 792                         ADVANCE();
 793                         state++;
 794                         break;
 795
 796                 case GZSTATE_METHOD:
 797                         if (*z.next_in != Z_DEFLATED) {
 798                                 maybe_warnx("unknown compression method");
 799                                 goto stop_and_fail;
 800                         }
 801                         ADVANCE();
 802                         state++;
 803                         break;
 804
 805                 case GZSTATE_FLAGS:
 806                         flags = *z.next_in;
 807                         ADVANCE();
 808                         skip_count = 6;
 809                         state++;
 810                         break;
 811
 812                 case GZSTATE_SKIPPING:
 813                         if (skip_count > 0) {
 814                                 skip_count--;
 815                                 ADVANCE();
 816                         } else
 817                                 state++;
 818                         break;
 819
 820                 case GZSTATE_EXTRA:
 821                         if ((flags & EXTRA_FIELD) == 0) {
 822                                 state = GZSTATE_ORIGNAME;
 823                                 break;
 824                         }
 825                         skip_count = *z.next_in;
 826                         ADVANCE();
 827                         state++;
 828                         break;
 829
 830                 case GZSTATE_EXTRA2:
 831                         skip_count |= ((*z.next_in) << 8);
 832                         ADVANCE();
 833                         state++;
 834                         break;
 835
 836                 case GZSTATE_EXTRA3:
 837                         if (skip_count > 0) {
 838                                 skip_count--;
 839                                 ADVANCE();
 840                         } else
 841                                 state++;
 842                         break;
 843
 844                 case GZSTATE_ORIGNAME:
 845                         if ((flags & ORIG_NAME) == 0) {
 846                                 state++;
 847                                 break;
 848                         }
 849                         if (*z.next_in == 0)
 850                                 state++;
 851                         ADVANCE();
 852                         break;
 853
 854                 case GZSTATE_COMMENT:
 855                         if ((flags & COMMENT) == 0) {
 856                                 state++;
 857                                 break;
 858                         }
 859                         if (*z.next_in == 0)
 860                                 state++;
 861                         ADVANCE();
 862                         break;
 863
 864                 case GZSTATE_HEAD_CRC1:
 865                         if (flags & HEAD_CRC)
 866                                 skip_count = 2;
 867                         else
 868                                 skip_count = 0;
 869                         state++;
 870                         break;
 871
 872                 case GZSTATE_HEAD_CRC2:
 873                         if (skip_count > 0) {
 874                                 skip_count--;
 875                                 ADVANCE();
 876                         } else
 877                                 state++;
 878                         break;
 879
 880                 case GZSTATE_INIT:
 881                         if (inflateInit2(&z, -MAX_WBITS) != Z_OK) {
 882                                 maybe_warnx("failed to inflateInit");
 883                                 goto stop_and_fail;
 884                         }
 885                         state++;
 886                         break;
 887
 888                 case GZSTATE_READ:
 889                         error = inflate(&z, Z_FINISH);
 890                         switch (error) {
 891                         /* Z_BUF_ERROR goes with Z_FINISH... */
 892                         case Z_BUF_ERROR:
 893                         case Z_STREAM_END:
 894                         case Z_OK:
 895                                 break;
 896
 897                         case Z_NEED_DICT:
 898                                 maybe_warnx("Z_NEED_DICT error");
 899                                 goto stop_and_fail;
 900                         case Z_DATA_ERROR:
 901                                 maybe_warnx("data stream error");
 902                                 goto stop_and_fail;
 903                         case Z_STREAM_ERROR:
 904                                 maybe_warnx("internal stream error");
 905                                 goto stop_and_fail;
 906                         case Z_MEM_ERROR:
 907                                 maybe_warnx("memory allocation error");
 908                                 goto stop_and_fail;
 909
 910                         default:
 911                                 maybe_warn("unknown error from inflate(): %d",
 912                                     error);
 913                         }
 914                         wr = BUFLEN - z.avail_out;
 915
 916                         if (wr != 0) {
 917                                 crc = crc32(crc, (const Bytef *)outbufp, (unsigned)wr);
 918                                 if (
 919 #ifndef SMALL
 920                                     /* don't write anything with -t */
 921                                     tflag == 0 &&
 922 #endif
 923                                     write(out, outbufp, wr) != wr) {
 924                                         maybe_warn("error writing to output");
 925                                         goto stop_and_fail;
 926                                 }
 927
 928                                 out_tot += wr;
 929                                 out_sub_tot += wr;
 930                         }
 931
 932                         if (error == Z_STREAM_END) {
 933                                 inflateEnd(&z);
 934                                 state++;
 935                         }
 936
 937                         z.next_out = outbufp;
 938                         z.avail_out = BUFLEN;
 939
 940                         break;
 941                 case GZSTATE_CRC:
 942                         {
 943                                 uLong origcrc;
 944
 945                                 if (z.avail_in < 4) {
 946                                         if (!done_reading) {
 947                                                 needmore = 1;
 948                                                 continue;
 949                                         }
 950                                         maybe_warnx("truncated input");
 951                                         goto stop_and_fail;
 952                                 }
 953                                 origcrc = ((unsigned)z.next_in[0] & 0xff) |
 954                                         ((unsigned)z.next_in[1] & 0xff) << 8 |
 955                                         ((unsigned)z.next_in[2] & 0xff) << 16 |
 956                                         ((unsigned)z.next_in[3] & 0xff) << 24;
 957                                 if (origcrc != crc) {
 958                                         maybe_warnx("invalid compressed"
 959                                              " data--crc error");
 960                                         goto stop_and_fail;
 961                                 }
 962                         }
 963
 964                         z.avail_in -= 4;
 965                         z.next_in += 4;
 966
 967                         if (!z.avail_in && done_reading) {
 968                                 goto stop;
 969                         }
 970                         state++;
 971                         break;
 972                 case GZSTATE_LEN:
 973                         {
 974                                 uLong origlen;
 975
 976                                 if (z.avail_in < 4) {
 977                                         if (!done_reading) {
 978                                                 needmore = 1;
 979                                                 continue;
 980                                         }
 981                                         maybe_warnx("truncated input");
 982                                         goto stop_and_fail;
 983                                 }
 984                                 origlen = ((unsigned)z.next_in[0] & 0xff) |
 985                                         ((unsigned)z.next_in[1] & 0xff) << 8 |
 986                                         ((unsigned)z.next_in[2] & 0xff) << 16 |
 987                                         ((unsigned)z.next_in[3] & 0xff) << 24;
 988
 989                                 if (origlen != out_sub_tot) {
 990                                         maybe_warnx("invalid compressed"
 991                                              " data--length error");
 992                                         goto stop_and_fail;
 993                                 }
 994                         }
 995
 996                         z.avail_in -= 4;
 997                         z.next_in += 4;
 998
 999                         if (error < 0) {
1000                                 maybe_warnx("decompression error");
1001                                 goto stop_and_fail;
1002                         }
1003                         state = GZSTATE_MAGIC0;
1004                         break;
1005                 }
1006                 continue;
1007 stop_and_fail:
1008                 out_tot = -1;
1009 stop:
1010                 break;
1011         }
1012         if (state > GZSTATE_INIT)
1013                 inflateEnd(&z);
1014
1015         free(inbufp);
1016 out1:
1017         free(outbufp);
1018 out2:
1019         if (gsizep)
1020                 *gsizep = in_tot;
1021         return (out_tot);
1022 }
1023
1024 #ifndef SMALL
1025 /*
1026  * set the owner, mode, flags & utimes using the given file descriptor.
1027  * file is only used in possible warning messages.
1028  */
1029 static void
1030 copymodes(int fd, const struct stat *sbp, const char *file)
1031 {
1032         struct timeval times[2];
1033         struct stat sb;
1034
1035         /*
1036          * If we have no info on the input, give this file some
1037          * default values and return..
1038          */
1039         if (sbp == NULL) {
1040                 mode_t mask = umask(022);
1041
1042                 (void)fchmod(fd, DEFFILEMODE & ~mask);
1043                 (void)umask(mask);
1044                 return;
1045         }
1046         sb = *sbp;
1047
1048         /* if the chown fails, remove set-id bits as-per compress(1) */
1049         if (fchown(fd, sb.st_uid, sb.st_gid) < 0) {
1050                 if (errno != EPERM)
1051                         maybe_warn("couldn't fchown: %s", file);
1052                 sb.st_mode &= ~(S_ISUID|S_ISGID);
1053         }
1054
1055         /* we only allow set-id and the 9 normal permission bits */
1056         sb.st_mode &= S_ISUID | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO;
1057         if (fchmod(fd, sb.st_mode) < 0)
1058                 maybe_warn("couldn't fchmod: %s", file);
1059
1060         /* only try flags if they exist already */
1061         if (sb.st_flags != 0 && fchflags(fd, sb.st_flags) < 0)
1062                 maybe_warn("couldn't fchflags: %s", file);
1063
1064         TIMESPEC_TO_TIMEVAL(&times[0], &sb.st_atimespec);
1065         TIMESPEC_TO_TIMEVAL(&times[1], &sb.st_mtimespec);
1066         if (futimes(fd, times) < 0)
1067                 maybe_warn("couldn't utimes: %s", file);
1068 }
1069 #endif
1070
1071 /* what sort of file is this? */
1072 static enum filetype
1073 file_gettype(u_char *buf)
1074 {
1075
1076         if (buf[0] == GZIP_MAGIC0 &&
1077             (buf[1] == GZIP_MAGIC1 || buf[1] == GZIP_OMAGIC1))
1078                 return FT_GZIP;
1079         else
1080 #ifndef NO_BZIP2_SUPPORT
1081         if (memcmp(buf, BZIP2_MAGIC, 3) == 0 &&
1082             buf[3] >= '0' && buf[3] <= '9')
1083                 return FT_BZIP2;
1084         else
1085 #endif
1086 #ifndef NO_COMPRESS_SUPPORT
1087         if (memcmp(buf, Z_MAGIC, 2) == 0)
1088                 return FT_Z;
1089         else
1090 #endif
1091 #ifndef NO_PACK_SUPPORT
1092         if (memcmp(buf, PACK_MAGIC, 2) == 0)
1093                 return FT_PACK;
1094         else
1095 #endif
1096                 return FT_UNKNOWN;
1097 }
1098
1099 #ifndef SMALL
1100 /* check the outfile is OK. */
1101 static int
1102 check_outfile(const char *outfile)
1103 {
1104         struct stat sb;
1105         int ok = 1;
1106
1107         if (lflag == 0 && stat(outfile, &sb) == 0) {
1108                 if (fflag)
1109                         unlink(outfile);
1110                 else if (isatty(STDIN_FILENO)) {
1111                         char ans[10] = { 'n', '\0' };   /* default */
1112
1113                         fprintf(stderr, "%s already exists -- do you wish to "
1114                                         "overwrite (y or n)? " , outfile);
1115                         (void)fgets(ans, sizeof(ans) - 1, stdin);
1116                         if (ans[0] != 'y' && ans[0] != 'Y') {
1117                                 fprintf(stderr, "\tnot overwriting\n");
1118                                 ok = 0;
1119                         } else
1120                                 unlink(outfile);
1121                 } else {
1122                         maybe_warnx("%s already exists -- skipping", outfile);
1123                         ok = 0;
1124                 }
1125         }
1126         return ok;
1127 }
1128
1129 static void
1130 unlink_input(const char *file, const struct stat *sb)
1131 {
1132         struct stat nsb;
1133
1134         if (kflag)
1135                 return;
1136         if (stat(file, &nsb) != 0)
1137                 /* Must be gone alrady */
1138                 return;
1139         if (nsb.st_dev != sb->st_dev || nsb.st_ino != sb->st_ino)
1140                 /* Definitely a different file */
1141                 return;
1142         unlink(file);
1143 }
1144 #endif
1145
1146 static const suffixes_t *
1147 check_suffix(char *file, int xlate)
1148 {
1149         const suffixes_t *s;
1150         int len = strlen(file);
1151         char *sp;
1152
1153         for (s = suffixes; s != suffixes + NUM_SUFFIXES; s++) {
1154                 /* if it doesn't fit in "a.suf", don't bother */
1155                 if (s->ziplen >= len)
1156                         continue;
1157                 sp = file + len - s->ziplen;
1158                 if (strcmp(s->zipped, sp) != 0)
1159                         continue;
1160                 if (xlate)
1161                         strcpy(sp, s->normal);
1162                 return s;
1163         }
1164         return NULL;
1165 }
1166
1167 /*
1168  * compress the given file: create a corresponding .gz file and remove the
1169  * original.
1170  */
1171 static off_t
1172 file_compress(char *file, char *outfile, size_t outsize)
1173 {
1174         int in;
1175         int out;
1176         off_t size, insize;
1177 #ifndef SMALL
1178         struct stat isb, osb;
1179         const suffixes_t *suff;
1180 #endif
1181
1182         in = open(file, O_RDONLY);
1183         if (in == -1) {
1184                 maybe_warn("can't open %s", file);
1185                 return -1;
1186         }
1187
1188         if (cflag == 0) {
1189 #ifndef SMALL
1190                 if (fstat(in, &isb) == 0) {
1191                         if (isb.st_nlink > 1 && fflag == 0) {
1192                                 maybe_warnx("%s has %d other link%s -- "
1193                                             "skipping", file, isb.st_nlink - 1,
1194                                             isb.st_nlink == 1 ? "" : "s");
1195                                 close(in);
1196                                 return -1;
1197                         }
1198                 }
1199
1200                 if (fflag == 0 && (suff = check_suffix(file, 0))
1201                     && suff->zipped[0] != 0) {
1202                         maybe_warnx("%s already has %s suffix -- unchanged",
1203                                     file, suff->zipped);
1204                         close(in);
1205                         return -1;
1206                 }
1207 #endif
1208
1209                 /* Add (usually) .gz to filename */
1210                 if ((size_t)snprintf(outfile, outsize, "%s%s",
1211                                      file, suffixes[0].zipped) >= outsize) {
1212                         errx(1, "file path too long: %s", file);
1213                 }
1214 #ifndef SMALL
1215                 if (check_outfile(outfile) == 0) {
1216                         close(in);
1217                         return -1;
1218                 }
1219 #endif
1220         }
1221
1222         if (cflag == 0) {
1223                 out = open(outfile, O_WRONLY | O_CREAT | O_EXCL, 0600);
1224                 if (out == -1) {
1225                         maybe_warn("could not create output: %s", outfile);
1226                         fclose(stdin);
1227                         return -1;
1228                 }
1229         } else
1230                 out = STDOUT_FILENO;
1231
1232         insize = gz_compress(in, out, &size, basename(file), (uint32_t)isb.st_mtime);
1233
1234         (void)close(in);
1235
1236         /*
1237          * If there was an error, insize will be -1.
1238          * If we compressed to stdout, just return the size.
1239          * Otherwise stat the file and check it is the correct size.
1240          * We only blow away the file if we can stat the output and it
1241          * has the expected size.
1242          */
1243         if (cflag != 0)
1244                 return insize == -1 ? -1 : size;
1245
1246 #ifndef SMALL
1247         if (fstat(out, &osb) != 0) {
1248                 maybe_warn("couldn't stat: %s", outfile);
1249                 goto bad_outfile;
1250         }
1251
1252         if (osb.st_size != size) {
1253                 maybe_warnx("output file: %s wrong size (%" PRIdOFF
1254                                 " != %" PRIdOFF "), deleting",
1255                                 outfile, osb.st_size, size);
1256                 goto bad_outfile;
1257         }
1258
1259         copymodes(out, &isb, outfile);
1260 #endif
1261         if (close(out) == -1)
1262                 maybe_warn("couldn't close output");
1263
1264         /* output is good, ok to delete input */
1265         unlink_input(file, &isb);
1266         return size;
1267
1268 #ifndef SMALL
1269     bad_outfile:
1270         if (close(out) == -1)
1271                 maybe_warn("couldn't close output");
1272
1273         maybe_warnx("leaving original %s", file);
1274         unlink(outfile);
1275         return size;
1276 #endif
1277 }
1278
1279 /* uncompress the given file and remove the original */
1280 static off_t
1281 file_uncompress(char *file, char *outfile, size_t outsize)
1282 {
1283         struct stat isb, osb;
1284         off_t size;
1285         ssize_t rbytes;
1286         unsigned char header1[4];
1287         enum filetype method;
1288         int fd, ofd, zfd = -1;
1289 #ifndef SMALL
1290         ssize_t rv;
1291         time_t timestamp = 0;
1292         unsigned char name[PATH_MAX + 1];
1293 #endif
1294
1295         /* gather the old name info */
1296
1297         fd = open(file, O_RDONLY);
1298         if (fd < 0) {
1299                 maybe_warn("can't open %s", file);
1300                 goto lose;
1301         }
1302
1303         if ((size_t)snprintf(outfile, outsize, "%s", file) >= outsize)
1304                 errx(1, "file path too long: %s", file);
1305         if (check_suffix(outfile, 1) == NULL && !(cflag || lflag)) {
1306                 maybe_warnx("%s: unknown suffix -- ignored", file);
1307                 goto lose;
1308         }
1309
1310         rbytes = read(fd, header1, sizeof header1);
1311         if (rbytes != sizeof header1) {
1312                 /* we don't want to fail here. */
1313 #ifndef SMALL
1314                 if (fflag)
1315                         goto lose;
1316 #endif
1317                 if (rbytes == -1)
1318                         maybe_warn("can't read %s", file);
1319                 else
1320                         goto unexpected_EOF;
1321                 goto lose;
1322         }
1323
1324         method = file_gettype(header1);
1325
1326 #ifndef SMALL
1327         if (fflag == 0 && method == FT_UNKNOWN) {
1328                 maybe_warnx("%s: not in gzip format", file);
1329                 goto lose;
1330         }
1331
1332 #endif
1333
1334 #ifndef SMALL
1335         if (method == FT_GZIP && Nflag) {
1336                 unsigned char ts[4];    /* timestamp */
1337
1338                 rv = pread(fd, ts, sizeof ts, GZIP_TIMESTAMP);
1339                 if (rv >= 0 && rv < (ssize_t)(sizeof ts))
1340                         goto unexpected_EOF;
1341                 if (rv == -1) {
1342                         if (!fflag)
1343                                 maybe_warn("can't read %s", file);
1344                         goto lose;
1345                 }
1346                 timestamp = ts[3] << 24 | ts[2] << 16 | ts[1] << 8 | ts[0];
1347
1348                 if (header1[3] & ORIG_NAME) {
1349                         rbytes = pread(fd, name, sizeof name, GZIP_ORIGNAME);
1350                         if (rbytes < 0) {
1351                                 maybe_warn("can't read %s", file);
1352                                 goto lose;
1353                         }
1354                         if (name[0] != 0) {
1355                                 /* preserve original directory name */
1356                                 char *dp = strrchr(file, '/');
1357                                 if (dp == NULL)
1358                                         dp = file;
1359                                 else
1360                                         dp++;
1361                                 snprintf(outfile, outsize, "%.*s%.*s",
1362                                                 (int) (dp - file),
1363                                                 file, (int) rbytes, name);
1364                         }
1365                 }
1366         }
1367 #endif
1368         lseek(fd, 0, SEEK_SET);
1369
1370         if (cflag == 0 || lflag) {
1371                 if (fstat(fd, &isb) != 0)
1372                         goto lose;
1373 #ifndef SMALL
1374                 if (isb.st_nlink > 1 && lflag == 0 && fflag == 0) {
1375                         maybe_warnx("%s has %d other links -- skipping",
1376                             file, isb.st_nlink - 1);
1377                         goto lose;
1378                 }
1379                 if (nflag == 0 && timestamp)
1380                         isb.st_mtime = timestamp;
1381                 if (check_outfile(outfile) == 0)
1382                         goto lose;
1383 #endif
1384         }
1385
1386         if (cflag == 0 && lflag == 0) {
1387                 zfd = open(outfile, O_WRONLY|O_CREAT|O_EXCL, 0600);
1388                 if (zfd == STDOUT_FILENO) {
1389                         /* We won't close STDOUT_FILENO later... */
1390                         zfd = dup(zfd);
1391                         close(STDOUT_FILENO);
1392                 }
1393                 if (zfd == -1) {
1394                         maybe_warn("can't open %s", outfile);
1395                         goto lose;
1396                 }
1397         } else
1398                 zfd = STDOUT_FILENO;
1399
1400 #ifndef NO_BZIP2_SUPPORT
1401         if (method == FT_BZIP2) {
1402
1403                 /* XXX */
1404                 if (lflag) {
1405                         maybe_warnx("no -l with bzip2 files");
1406                         goto lose;
1407                 }
1408
1409                 size = unbzip2(fd, zfd, NULL, 0, NULL);
1410         } else
1411 #endif
1412
1413 #ifndef NO_COMPRESS_SUPPORT
1414         if (method == FT_Z) {
1415                 FILE *in, *out;
1416
1417                 /* XXX */
1418                 if (lflag) {
1419                         maybe_warnx("no -l with Lempel-Ziv files");
1420                         goto lose;
1421                 }
1422
1423                 if ((in = zdopen(fd)) == NULL) {
1424                         maybe_warn("zdopen for read: %s", file);
1425                         goto lose;
1426                 }
1427
1428                 out = fdopen(dup(zfd), "w");
1429                 if (out == NULL) {
1430                         maybe_warn("fdopen for write: %s", outfile);
1431                         fclose(in);
1432                         goto lose;
1433                 }
1434
1435                 size = zuncompress(in, out, NULL, 0, NULL);
1436                 /* need to fclose() if ferror() is true... */
1437                 if (ferror(in) | fclose(in)) {
1438                         maybe_warn("failed infile fclose");
1439                         unlink(outfile);
1440                         (void)fclose(out);
1441                 }
1442                 if (fclose(out) != 0) {
1443                         maybe_warn("failed outfile fclose");
1444                         unlink(outfile);
1445                         goto lose;
1446                 }
1447         } else
1448 #endif
1449
1450 #ifndef NO_PACK_SUPPORT
1451         if (method == FT_PACK) {
1452                 if (lflag) {
1453                         maybe_warnx("no -l with packed files");
1454                         goto lose;
1455                 }
1456
1457                 size = unpack(fd, zfd, NULL, 0, NULL);
1458         } else
1459 #endif
1460
1461 #ifndef SMALL
1462         if (method == FT_UNKNOWN) {
1463                 if (lflag) {
1464                         maybe_warnx("no -l for unknown filetypes");
1465                         goto lose;
1466                 }
1467                 size = cat_fd(NULL, 0, NULL, fd);
1468         } else
1469 #endif
1470         {
1471                 if (lflag) {
1472                         print_list(fd, isb.st_size, outfile, isb.st_mtime);
1473                         close(fd);
1474                         return -1;      /* XXX */
1475                 }
1476
1477                 size = gz_uncompress(fd, zfd, NULL, 0, NULL, file);
1478         }
1479
1480         if (close(fd) != 0)
1481                 maybe_warn("couldn't close input");
1482         if (zfd != STDOUT_FILENO && close(zfd) != 0)
1483                 maybe_warn("couldn't close output");
1484
1485         if (size == -1) {
1486                 if (cflag == 0)
1487                         unlink(outfile);
1488                 maybe_warnx("%s: uncompress failed", file);
1489                 return -1;
1490         }
1491
1492         /* if testing, or we uncompressed to stdout, this is all we need */
1493 #ifndef SMALL
1494         if (tflag)
1495                 return size;
1496 #endif
1497         /* if we are uncompressing to stdin, don't remove the file. */
1498         if (cflag)
1499                 return size;
1500
1501         /*
1502          * if we create a file...
1503          */
1504         /*
1505          * if we can't stat the file don't remove the file.
1506          */
1507
1508         ofd = open(outfile, O_RDWR, 0);
1509         if (ofd == -1) {
1510                 maybe_warn("couldn't open (leaving original): %s",
1511                            outfile);
1512                 return -1;
1513         }
1514         if (fstat(ofd, &osb) != 0) {
1515                 maybe_warn("couldn't stat (leaving original): %s",
1516                            outfile);
1517                 close(ofd);
1518                 return -1;
1519         }
1520         if (osb.st_size != size) {
1521                 maybe_warnx("stat gave different size: %" PRIdOFF
1522                                 " != %" PRIdOFF " (leaving original)",
1523                                 size, osb.st_size);
1524                 close(ofd);
1525                 unlink(outfile);
1526                 return -1;
1527         }
1528         unlink_input(file, &isb);
1529 #ifndef SMALL
1530         copymodes(ofd, &isb, outfile);
1531 #endif
1532         close(ofd);
1533         return size;
1534
1535     unexpected_EOF:
1536         maybe_warnx("%s: unexpected end of file", file);
1537     lose:
1538         if (fd != -1)
1539                 close(fd);
1540         if (zfd != -1 && zfd != STDOUT_FILENO)
1541                 close(fd);
1542         return -1;
1543 }
1544
1545 #ifndef SMALL
1546 static off_t
1547 cat_fd(unsigned char * prepend, size_t count, off_t *gsizep, int fd)
1548 {
1549         char buf[BUFLEN];
1550         off_t in_tot;
1551         ssize_t w;
1552
1553         in_tot = count;
1554         w = write(STDOUT_FILENO, prepend, count);
1555         if (w == -1 || (size_t)w != count) {
1556                 maybe_warn("write to stdout");
1557                 return -1;
1558         }
1559         for (;;) {
1560                 ssize_t rv;
1561
1562                 rv = read(fd, buf, sizeof buf);
1563                 if (rv == 0)
1564                         break;
1565                 if (rv < 0) {
1566                         maybe_warn("read from fd %d", fd);
1567                         break;
1568                 }
1569
1570                 if (write(STDOUT_FILENO, buf, rv) != rv) {
1571                         maybe_warn("write to stdout");
1572                         break;
1573                 }
1574                 in_tot += rv;
1575         }
1576
1577         if (gsizep)
1578                 *gsizep = in_tot;
1579         return (in_tot);
1580 }
1581 #endif
1582
1583 static void
1584 handle_stdin(void)
1585 {
1586         unsigned char header1[4];
1587         off_t usize, gsize;
1588         enum filetype method;
1589         ssize_t bytes_read;
1590 #ifndef NO_COMPRESS_SUPPORT
1591         FILE *in;
1592 #endif
1593
1594 #ifndef SMALL
1595         if (fflag == 0 && lflag == 0 && isatty(STDIN_FILENO)) {
1596                 maybe_warnx("standard input is a terminal -- ignoring");
1597                 return;
1598         }
1599 #endif
1600
1601         if (lflag) {
1602                 struct stat isb;
1603
1604                 /* XXX could read the whole file, etc. */
1605                 if (fstat(STDIN_FILENO, &isb) < 0) {
1606                         maybe_warn("fstat");
1607                         return;
1608                 }
1609                 print_list(STDIN_FILENO, isb.st_size, "stdout", isb.st_mtime);
1610                 return;
1611         }
1612
1613         bytes_read = read_retry(STDIN_FILENO, header1, sizeof header1);
1614         if (bytes_read == -1) {
1615                 maybe_warn("can't read stdin");
1616                 return;
1617         } else if (bytes_read != sizeof(header1)) {
1618                 maybe_warnx("(stdin): unexpected end of file");
1619                 return;
1620         }
1621
1622         method = file_gettype(header1);
1623         switch (method) {
1624         default:
1625 #ifndef SMALL
1626                 if (fflag == 0) {
1627                         maybe_warnx("unknown compression format");
1628                         return;
1629                 }
1630                 usize = cat_fd(header1, sizeof header1, &gsize, STDIN_FILENO);
1631                 break;
1632 #endif
1633         case FT_GZIP:
1634                 usize = gz_uncompress(STDIN_FILENO, STDOUT_FILENO,
1635                               header1, sizeof header1, &gsize, "(stdin)");
1636                 break;
1637 #ifndef NO_BZIP2_SUPPORT
1638         case FT_BZIP2:
1639                 usize = unbzip2(STDIN_FILENO, STDOUT_FILENO,
1640                                 header1, sizeof header1, &gsize);
1641                 break;
1642 #endif
1643 #ifndef NO_COMPRESS_SUPPORT
1644         case FT_Z:
1645                 if ((in = zdopen(STDIN_FILENO)) == NULL) {
1646                         maybe_warnx("zopen of stdin");
1647                         return;
1648                 }
1649
1650                 usize = zuncompress(in, stdout, header1, sizeof header1, &gsize);
1651                 fclose(in);
1652                 break;
1653 #endif
1654 #ifndef NO_PACK_SUPPORT
1655         case FT_PACK:
1656                 usize = unpack(STDIN_FILENO, STDOUT_FILENO,
1657                                (char *)header1, sizeof header1, &gsize);
1658                 break;
1659 #endif
1660         }
1661
1662 #ifndef SMALL
1663         if (vflag && !tflag && usize != -1 && gsize != -1)
1664                 print_verbage(NULL, NULL, usize, gsize);
1665         if (vflag && tflag)
1666                 print_test("(stdin)", usize != -1);
1667 #endif
1668
1669 }
1670
1671 static void
1672 handle_stdout(void)
1673 {
1674         off_t gsize, usize;
1675         struct stat sb;
1676         time_t systime;
1677         uint32_t mtime;
1678         int ret;
1679
1680 #ifndef SMALL
1681         if (fflag == 0 && isatty(STDOUT_FILENO)) {
1682                 maybe_warnx("standard output is a terminal -- ignoring");
1683                 return;
1684         }
1685 #endif
1686         /* If stdin is a file use it's mtime, otherwise use current time */
1687         ret = fstat(STDIN_FILENO, &sb);
1688
1689 #ifndef SMALL
1690         if (ret < 0) {
1691                 maybe_warn("Can't stat stdin");
1692                 return;
1693         }
1694 #endif
1695
1696         if (S_ISREG(sb.st_mode))
1697                 mtime = (uint32_t)sb.st_mtime;
1698         else {
1699                 systime = time(NULL);
1700 #ifndef SMALL
1701                 if (systime == -1) {
1702                         maybe_warn("time");
1703                         return;
1704                 }
1705 #endif
1706                 mtime = (uint32_t)systime;
1707         }
1708
1709         usize = gz_compress(STDIN_FILENO, STDOUT_FILENO, &gsize, "", mtime);
1710 #ifndef SMALL
1711         if (vflag && !tflag && usize != -1 && gsize != -1)
1712                 print_verbage(NULL, NULL, usize, gsize);
1713 #endif
1714 }
1715
1716 /* do what is asked for, for the path name */
1717 static void
1718 handle_pathname(char *path)
1719 {
1720         char *opath = path, *s = NULL;
1721         ssize_t len;
1722         int slen;
1723         struct stat sb;
1724
1725         /* check for stdout/stdin */
1726         if (path[0] == '-' && path[1] == '\0') {
1727                 if (dflag)
1728                         handle_stdin();
1729                 else
1730                         handle_stdout();
1731                 return;
1732         }
1733
1734 retry:
1735         if (stat(path, &sb) != 0) {
1736                 /* lets try <path>.gz if we're decompressing */
1737                 if (dflag && s == NULL && errno == ENOENT) {
1738                         len = strlen(path);
1739                         slen = suffixes[0].ziplen;
1740                         s = malloc(len + slen + 1);
1741                         if (s == NULL)
1742                                 maybe_err("malloc");
1743                         memcpy(s, path, len);
1744                         memcpy(s + len, suffixes[0].zipped, slen + 1);
1745                         path = s;
1746                         goto retry;
1747                 }
1748                 maybe_warn("can't stat: %s", opath);
1749                 goto out;
1750         }
1751
1752         if (S_ISDIR(sb.st_mode)) {
1753 #ifndef SMALL
1754                 if (rflag)
1755                         handle_dir(path);
1756                 else
1757 #endif
1758                         maybe_warnx("%s is a directory", path);
1759                 goto out;
1760         }
1761
1762         if (S_ISREG(sb.st_mode))
1763                 handle_file(path, &sb);
1764         else
1765                 maybe_warnx("%s is not a regular file", path);
1766
1767 out:
1768         if (s)
1769                 free(s);
1770 }
1771
1772 /* compress/decompress a file */
1773 static void
1774 handle_file(char *file, struct stat *sbp)
1775 {
1776         off_t usize, gsize;
1777         char    outfile[PATH_MAX];
1778
1779         infile = file;
1780         if (dflag) {
1781                 usize = file_uncompress(file, outfile, sizeof(outfile));
1782 #ifndef SMALL
1783                 if (vflag && tflag)
1784                         print_test(file, usize != -1);
1785 #endif
1786                 if (usize == -1)
1787                         return;
1788                 gsize = sbp->st_size;
1789         } else {
1790                 gsize = file_compress(file, outfile, sizeof(outfile));
1791                 if (gsize == -1)
1792                         return;
1793                 usize = sbp->st_size;
1794         }
1795
1796
1797 #ifndef SMALL
1798         if (vflag && !tflag)
1799                 print_verbage(file, (cflag) ? NULL : outfile, usize, gsize);
1800 #endif
1801 }
1802
1803 #ifndef SMALL
1804 /* this is used with -r to recursively descend directories */
1805 static void
1806 handle_dir(char *dir)
1807 {
1808         char *path_argv[2];
1809         FTS *fts;
1810         FTSENT *entry;
1811
1812         path_argv[0] = dir;
1813         path_argv[1] = 0;
1814         fts = fts_open(path_argv, FTS_PHYSICAL | FTS_NOCHDIR, NULL);
1815         if (fts == NULL) {
1816                 warn("couldn't fts_open %s", dir);
1817                 return;
1818         }
1819
1820         while ((entry = fts_read(fts))) {
1821                 switch(entry->fts_info) {
1822                 case FTS_D:
1823                 case FTS_DP:
1824                         continue;
1825
1826                 case FTS_DNR:
1827                 case FTS_ERR:
1828                 case FTS_NS:
1829                         maybe_warn("%s", entry->fts_path);
1830                         continue;
1831                 case FTS_F:
1832                         handle_file(entry->fts_path, entry->fts_statp);
1833                 }
1834         }
1835         (void)fts_close(fts);
1836 }
1837 #endif
1838
1839 /* print a ratio - size reduction as a fraction of uncompressed size */
1840 static void
1841 print_ratio(off_t in, off_t out, FILE *where)
1842 {
1843         int percent10;  /* 10 * percent */
1844         off_t diff;
1845         char buff[8];
1846         int len;
1847
1848         diff = in - out/2;
1849         if (diff <= 0)
1850                 /*
1851                  * Output is more than double size of input! print -99.9%
1852                  * Quite possibly we've failed to get the original size.
1853                  */
1854                 percent10 = -999;
1855         else {
1856                 /*
1857                  * We only need 12 bits of result from the final division,
1858                  * so reduce the values until a 32bit division will suffice.
1859                  */
1860                 while (in > 0x100000) {
1861                         diff >>= 1;
1862                         in >>= 1;
1863                 }
1864                 if (in != 0)
1865                         percent10 = ((u_int)diff * 2000) / (u_int)in - 1000;
1866                 else
1867                         percent10 = 0;
1868         }
1869
1870         len = snprintf(buff, sizeof buff, "%2.2d.", percent10);
1871         /* Move the '.' to before the last digit */
1872         buff[len - 1] = buff[len - 2];
1873         buff[len - 2] = '.';
1874         fprintf(where, "%5s%%", buff);
1875 }
1876
1877 #ifndef SMALL
1878 /* print compression statistics, and the new name (if there is one!) */
1879 static void
1880 print_verbage(const char *file, const char *nfile, off_t usize, off_t gsize)
1881 {
1882         if (file)
1883                 fprintf(stderr, "%s:%s  ", file,
1884                     strlen(file) < 7 ? "\t\t" : "\t");
1885         print_ratio(usize, gsize, stderr);
1886         if (nfile)
1887                 fprintf(stderr, " -- replaced with %s", nfile);
1888         fprintf(stderr, "\n");
1889         fflush(stderr);
1890 }
1891
1892 /* print test results */
1893 static void
1894 print_test(const char *file, int ok)
1895 {
1896
1897         if (exit_value == 0 && ok == 0)
1898                 exit_value = 1;
1899         fprintf(stderr, "%s:%s  %s\n", file,
1900             strlen(file) < 7 ? "\t\t" : "\t", ok ? "OK" : "NOT OK");
1901         fflush(stderr);
1902 }
1903 #endif
1904
1905 /* print a file's info ala --list */
1906 /* eg:
1907   compressed uncompressed  ratio uncompressed_name
1908       354841      1679360  78.8% /usr/pkgsrc/distfiles/libglade-2.0.1.tar
1909 */
1910 static void
1911 print_list(int fd, off_t out, const char *outfile, time_t ts)
1912 {
1913         static int first = 1;
1914 #ifndef SMALL
1915         static off_t in_tot, out_tot;
1916         uint32_t crc = 0;
1917 #endif
1918         off_t in = 0, rv;
1919
1920         if (first) {
1921 #ifndef SMALL
1922                 if (vflag)
1923                         printf("method  crc     date  time  ");
1924 #endif
1925                 if (qflag == 0)
1926                         printf("  compressed uncompressed  "
1927                                "ratio uncompressed_name\n");
1928         }
1929         first = 0;
1930
1931         /* print totals? */
1932 #ifndef SMALL
1933         if (fd == -1) {
1934                 in = in_tot;
1935                 out = out_tot;
1936         } else
1937 #endif
1938         {
1939                 /* read the last 4 bytes - this is the uncompressed size */
1940                 rv = lseek(fd, (off_t)(-8), SEEK_END);
1941                 if (rv != -1) {
1942                         unsigned char buf[8];
1943                         uint32_t usize;
1944
1945                         rv = read(fd, (char *)buf, sizeof(buf));
1946                         if (rv == -1)
1947                                 maybe_warn("read of uncompressed size");
1948                         else if (rv != sizeof(buf))
1949                                 maybe_warnx("read of uncompressed size");
1950
1951                         else {
1952                                 usize = buf[4] | buf[5] << 8 |
1953                                         buf[6] << 16 | buf[7] << 24;
1954                                 in = (off_t)usize;
1955 #ifndef SMALL
1956                                 crc = buf[0] | buf[1] << 8 |
1957                                       buf[2] << 16 | buf[3] << 24;
1958 #endif
1959                         }
1960                 }
1961         }
1962
1963 #ifndef SMALL
1964         if (vflag && fd == -1)
1965                 printf("                            ");
1966         else if (vflag) {
1967                 char *date = ctime(&ts);
1968
1969                 /* skip the day, 1/100th second, and year */
1970                 date += 4;
1971                 date[12] = 0;
1972                 printf("%5s %08x %11s ", "defla"/*XXX*/, crc, date);
1973         }
1974         in_tot += in;
1975         out_tot += out;
1976 #endif
1977         printf("%12llu %12llu ", (unsigned long long)out, (unsigned long long)in);
1978         print_ratio(in, out, stdout);
1979         printf(" %s\n", outfile);
1980 }
1981
1982 /* display the usage of NetBSD gzip */
1983 static void
1984 usage(void)
1985 {
1986
1987         fprintf(stderr, "%s\n", gzip_version);
1988         fprintf(stderr,
1989 #ifdef SMALL
1990         "usage: %s [-" OPT_LIST "] [<file> [<file> ...]]\n",
1991 #else
1992         "usage: %s [-123456789acdfhklLNnqrtVv] [-S .suffix] [<file> [<file> ...]]\n"
1993         " -1 --fast            fastest (worst) compression\n"
1994         " -2 .. -8             set compression level\n"
1995         " -9 --best            best (slowest) compression\n"
1996         " -c --stdout          write to stdout, keep original files\n"
1997         "    --to-stdout\n"
1998         " -d --decompress      uncompress files\n"
1999         "    --uncompress\n"
2000         " -f --force           force overwriting & compress links\n"
2001         " -h --help            display this help\n"
2002         " -k --keep            don't delete input files during operation\n"
2003         " -l --list            list compressed file contents\n"
2004         " -N --name            save or restore original file name and time stamp\n"
2005         " -n --no-name         don't save original file name or time stamp\n"
2006         " -q --quiet           output no warnings\n"
2007         " -r --recursive       recursively compress files in directories\n"
2008         " -S .suf              use suffix .suf instead of .gz\n"
2009         "    --suffix .suf\n"
2010         " -t --test            test compressed file\n"
2011         " -V --version         display program version\n"
2012         " -v --verbose         print extra statistics\n",
2013 #endif
2014         getprogname());
2015         exit(0);
2016 }
2017
2018 /* display the version of NetBSD gzip */
2019 static void
2020 display_version(void)
2021 {
2022
2023         fprintf(stderr, "%s\n", gzip_version);
2024         exit(0);
2025 }
2026
2027 #ifndef NO_BZIP2_SUPPORT
2028 #include "unbzip2.c"
2029 #endif
2030 #ifndef NO_COMPRESS_SUPPORT
2031 #include "zuncompress.c"
2032 #endif
2033 #ifndef NO_PACK_SUPPORT
2034 #include "unpack.c"
2035 #endif
2036
2037 static ssize_t
2038 read_retry(int fd, void *buf, size_t sz)
2039 {
2040         char *cp = buf;
2041         size_t left = MIN(sz, (size_t) SSIZE_MAX);
2042
2043         while (left > 0) {
2044                 ssize_t ret;
2045
2046                 ret = read(fd, cp, left);
2047                 if (ret == -1) {
2048                         return ret;
2049                 } else if (ret == 0) {
2050                         break; /* EOF */
2051                 }
2052                 cp += ret;
2053                 left -= ret;
2054         }
2055
2056         return sz - left;
2057 }