contrib/bsdtar/read.c

   1 /*-
   2  * Copyright (c) 2003-2004 Tim Kientzle
   3  * All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer
  10  *    in this position and unchanged.
  11  * 2. Redistributions in binary form must reproduce the above copyright
  12  *    notice, this list of conditions and the following disclaimer in the
  13  *    documentation and/or other materials provided with the distribution.
  14  *
  15  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
  16  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
  17  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
  18  * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
  19  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
  20  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  21  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  22  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  23  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
  24  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  25  */
  26
  27 #include "bsdtar_platform.h"
  28 __FBSDID("$FreeBSD: src/usr.bin/tar/read.c,v 1.23 2005/04/17 17:20:54 kientzle Exp $");
  29
  30 #include <sys/param.h>
  31 #include <sys/types.h>
  32 #include <sys/stat.h>
  33
  34 #include <archive.h>
  35 #include <archive_entry.h>
  36 #include <errno.h>
  37 #include <grp.h>
  38 #include <limits.h>
  39 #include <pwd.h>
  40 #include <stdio.h>
  41 #include <stdlib.h>
  42 #include <string.h>
  43 #include <time.h>
  44 #include <unistd.h>
  45
  46 #include "bsdtar.h"
  47
  48 static void     cleanup_security(struct bsdtar *);
  49 static void     list_item_verbose(struct bsdtar *, FILE *,
  50                     struct archive_entry *);
  51 static void     read_archive(struct bsdtar *bsdtar, char mode);
  52 static int      security_problem(struct bsdtar *, struct archive_entry *);
  53
  54 void
  55 tar_mode_t(struct bsdtar *bsdtar)
  56 {
  57         read_archive(bsdtar, 't');
  58 }
  59
  60 void
  61 tar_mode_x(struct bsdtar *bsdtar)
  62 {
  63         read_archive(bsdtar, 'x');
  64 }
  65
  66 /*
  67  * Handle 'x' and 't' modes.
  68  */
  69 static void
  70 read_archive(struct bsdtar *bsdtar, char mode)
  71 {
  72         FILE                     *out;
  73         struct archive           *a;
  74         struct archive_entry     *entry;
  75         const struct stat        *st;
  76         int                       r;
  77
  78         while (*bsdtar->argv) {
  79                 include(bsdtar, *bsdtar->argv);
  80                 bsdtar->argv++;
  81         }
  82
  83         if (bsdtar->names_from_file != NULL)
  84                 include_from_file(bsdtar, bsdtar->names_from_file);
  85
  86         a = archive_read_new();
  87         archive_read_support_compression_all(a);
  88         archive_read_support_format_all(a);
  89         if (archive_read_open_file(a, bsdtar->filename,
  90             bsdtar->bytes_per_block != 0 ? bsdtar->bytes_per_block :
  91             DEFAULT_BYTES_PER_BLOCK))
  92                 bsdtar_errc(bsdtar, 1, 0, "Error opening archive: %s",
  93                     archive_error_string(a));
  94
  95         do_chdir(bsdtar);
  96         for (;;) {
  97                 /* Support --fast-read option */
  98                 if (bsdtar->option_fast_read &&
  99                     unmatched_inclusions(bsdtar) == 0)
 100                         break;
 101
 102                 r = archive_read_next_header(a, &entry);
 103                 if (r == ARCHIVE_EOF)
 104                         break;
 105                 if (r == ARCHIVE_WARN)
 106                         bsdtar_warnc(bsdtar, 0, "%s", archive_error_string(a));
 107                 if (r == ARCHIVE_FATAL) {
 108                         bsdtar->return_value = 1;
 109                         bsdtar_warnc(bsdtar, 0, "%s", archive_error_string(a));
 110                         break;
 111                 }
 112                 if (r == ARCHIVE_RETRY) {
 113                         /* Retryable error: try again */
 114                         bsdtar_warnc(bsdtar, 0, "%s", archive_error_string(a));
 115                         bsdtar_warnc(bsdtar, 0, "Retrying...");
 116                         continue;
 117                 }
 118
 119                 /*
 120                  * Exclude entries that are too old.
 121                  */
 122                 st = archive_entry_stat(entry);
 123                 if (bsdtar->newer_ctime_sec > 0) {
 124                         if (st->st_ctime < bsdtar->newer_ctime_sec)
 125                                 continue; /* Too old, skip it. */
 126                         if (st->st_ctime == bsdtar->newer_ctime_sec
 127                             && ARCHIVE_STAT_CTIME_NANOS(st)
 128                             <= bsdtar->newer_ctime_nsec)
 129                                 continue; /* Too old, skip it. */
 130                 }
 131                 if (bsdtar->newer_mtime_sec > 0) {
 132                         if (st->st_mtime < bsdtar->newer_mtime_sec)
 133                                 continue; /* Too old, skip it. */
 134                         if (st->st_mtime == bsdtar->newer_mtime_sec
 135                             && ARCHIVE_STAT_MTIME_NANOS(st)
 136                             <= bsdtar->newer_mtime_nsec)
 137                                 continue; /* Too old, skip it. */
 138                 }
 139
 140                 /*
 141                  * Note that pattern exclusions are checked before
 142                  * pathname rewrites are handled.  This gives more
 143                  * control over exclusions, since rewrites always lose
 144                  * information.  (For example, consider a rewrite
 145                  * s/foo[0-9]/foo/.  If we check exclusions after the
 146                  * rewrite, there would be no way to exclude foo1/bar
 147                  * while allowing foo2/bar.)
 148                  */
 149                 if (excluded(bsdtar, archive_entry_pathname(entry)))
 150                         continue; /* Excluded by a pattern test. */
 151
 152                 /*
 153                  * Modify the pathname as requested by the user.  We
 154                  * do this for -t as well to give users a way to
 155                  * preview the effects of their rewrites.  We also do
 156                  * this before extraction security checks (including
 157                  * leading '/' removal).  Note that some rewrite
 158                  * failures prevent extraction.
 159                  */
 160                 if (edit_pathname(bsdtar, entry))
 161                         continue; /* Excluded by a rewrite failure. */
 162
 163                 if (mode == 't') {
 164                         /* Perversely, gtar uses -O to mean "send to stderr"
 165                          * when used with -t. */
 166                         out = bsdtar->option_stdout ? stderr : stdout;
 167
 168                         if (bsdtar->verbose < 2)
 169                                 safe_fprintf(out, "%s",
 170                                     archive_entry_pathname(entry));
 171                         else
 172                                 list_item_verbose(bsdtar, out, entry);
 173                         fflush(out);
 174                         r = archive_read_data_skip(a);
 175                         if (r == ARCHIVE_WARN) {
 176                                 fprintf(out, "\n");
 177                                 bsdtar_warnc(bsdtar, 0, "%s",
 178                                     archive_error_string(a));
 179                         }
 180                         if (r == ARCHIVE_RETRY) {
 181                                 fprintf(out, "\n");
 182                                 bsdtar_warnc(bsdtar, 0, "%s",
 183                                     archive_error_string(a));
 184                         }
 185                         if (r == ARCHIVE_FATAL) {
 186                                 fprintf(out, "\n");
 187                                 bsdtar_warnc(bsdtar, 0, "%s",
 188                                     archive_error_string(a));
 189                                 break;
 190                         }
 191                         fprintf(out, "\n");
 192                 } else {
 193                         /*
 194                          * Skip security problems before prompting.
 195                          * Otherwise, the user may be confused that a
 196                          * file they wanted to extract was
 197                          * subsequently skipped.
 198                          */
 199                         if (security_problem(bsdtar, entry))
 200                                 continue;
 201
 202                         if (bsdtar->option_interactive &&
 203                             !yes("extract '%s'", archive_entry_pathname(entry)))
 204                                 continue;
 205
 206                         /*
 207                          * Format here is from SUSv2, including the
 208                          * deferred '\n'.
 209                          */
 210                         if (bsdtar->verbose) {
 211                                 safe_fprintf(stderr, "x %s",
 212                                     archive_entry_pathname(entry));
 213                                 fflush(stderr);
 214                         }
 215                         if (bsdtar->option_stdout) {
 216                                 /* TODO: Catch/recover any errors here. */
 217                                 archive_read_data_into_fd(a, 1);
 218                         } else if (archive_read_extract(a, entry,
 219                                        bsdtar->extract_flags)) {
 220                                 if (!bsdtar->verbose)
 221                                         safe_fprintf(stderr, "%s",
 222                                             archive_entry_pathname(entry));
 223                                 safe_fprintf(stderr, ": %s",
 224                                     archive_error_string(a));
 225                                 if (!bsdtar->verbose)
 226                                         fprintf(stderr, "\n");
 227                                 /*
 228                                  * TODO: Decide how to handle
 229                                  * extraction error... <sigh>
 230                                  */
 231                                 bsdtar->return_value = 1;
 232                         }
 233                         if (bsdtar->verbose)
 234                                 fprintf(stderr, "\n");
 235                 }
 236         }
 237
 238         if (bsdtar->verbose > 2)
 239                 fprintf(stdout, "Archive Format: %s,  Compression: %s\n",
 240                     archive_format_name(a), archive_compression_name(a));
 241
 242         archive_read_finish(a);
 243         cleanup_security(bsdtar);
 244 }
 245
 246
 247 /*
 248  * Display information about the current file.
 249  *
 250  * The format here roughly duplicates the output of 'ls -l'.
 251  * This is based on SUSv2, where 'tar tv' is documented as
 252  * listing additional information in an "unspecified format,"
 253  * and 'pax -l' is documented as using the same format as 'ls -l'.
 254  */
 255 static void
 256 list_item_verbose(struct bsdtar *bsdtar, FILE *out, struct archive_entry *entry)
 257 {
 258         const struct stat       *st;
 259         char                     tmp[100];
 260         size_t                   w;
 261         const char              *p;
 262         const char              *fmt;
 263         time_t                   tim;
 264         static time_t            now;
 265
 266         st = archive_entry_stat(entry);
 267
 268         /*
 269          * We avoid collecting the entire list in memory at once by
 270          * listing things as we see them.  However, that also means we can't
 271          * just pre-compute the field widths.  Instead, we start with guesses
 272          * and just widen them as necessary.  These numbers are completely
 273          * arbitrary.
 274          */
 275         if (!bsdtar->u_width) {
 276                 bsdtar->u_width = 6;
 277                 bsdtar->gs_width = 13;
 278         }
 279         if (!now)
 280                 time(&now);
 281         bsdtar_strmode(entry, tmp);
 282         fprintf(out, "%s %d ", tmp, st->st_nlink);
 283
 284         /* Use uname if it's present, else uid. */
 285         p = archive_entry_uname(entry);
 286         if ((p == NULL) || (*p == '\0')) {
 287                 sprintf(tmp, "%d ", st->st_uid);
 288                 p = tmp;
 289         }
 290         w = strlen(p);
 291         if (w > bsdtar->u_width)
 292                 bsdtar->u_width = w;
 293         fprintf(out, "%-*s ", (int)bsdtar->u_width, p);
 294
 295         /* Use gname if it's present, else gid. */
 296         p = archive_entry_gname(entry);
 297         if (p != NULL && p[0] != '\0') {
 298                 fprintf(out, "%s", p);
 299                 w = strlen(p);
 300         } else {
 301                 sprintf(tmp, "%d", st->st_gid);
 302                 w = strlen(tmp);
 303                 fprintf(out, "%s", tmp);
 304         }
 305
 306         /*
 307          * Print device number or file size, right-aligned so as to make
 308          * total width of group and devnum/filesize fields be gs_width.
 309          * If gs_width is too small, grow it.
 310          */
 311         if (S_ISCHR(st->st_mode) || S_ISBLK(st->st_mode)) {
 312                 sprintf(tmp, "%d,%u",
 313                     major(st->st_rdev),
 314                     (unsigned)minor(st->st_rdev)); /* ls(1) also casts here. */
 315         } else {
 316                 /*
 317                  * Note the use of platform-dependent macros to format
 318                  * the filesize here.  We need the format string and the
 319                  * corresponding type for the cast.
 320                  */
 321                 sprintf(tmp, BSDTAR_FILESIZE_PRINTF,
 322                     (BSDTAR_FILESIZE_TYPE)st->st_size);
 323         }
 324         if (w + strlen(tmp) >= bsdtar->gs_width)
 325                 bsdtar->gs_width = w+strlen(tmp)+1;
 326         fprintf(out, "%*s", (int)(bsdtar->gs_width - w), tmp);
 327
 328         /* Format the time using 'ls -l' conventions. */
 329         tim = (time_t)st->st_mtime;
 330         if (abs(tim - now) > (365/2)*86400)
 331                 fmt = bsdtar->day_first ? "%e %b  %Y" : "%b %e  %Y";
 332         else
 333                 fmt = bsdtar->day_first ? "%e %b %R" : "%b %e %R";
 334         strftime(tmp, sizeof(tmp), fmt, localtime(&tim));
 335         fprintf(out, " %s ", tmp);
 336         safe_fprintf(out, "%s", archive_entry_pathname(entry));
 337
 338         /* Extra information for links. */
 339         if (archive_entry_hardlink(entry)) /* Hard link */
 340                 safe_fprintf(out, " link to %s",
 341                     archive_entry_hardlink(entry));
 342         else if (S_ISLNK(st->st_mode)) /* Symbolic link */
 343                 safe_fprintf(out, " -> %s", archive_entry_symlink(entry));
 344 }
 345
 346 /*
 347  * Structure for storing path of last successful security check.
 348  */
 349 struct security {
 350         char    *path;
 351         size_t   path_size;
 352 };
 353
 354 /*
 355  * Check for a variety of security issues.  Fix what we can here,
 356  * generate warnings as appropriate, return non-zero to prevent
 357  * this entry from being extracted.
 358  */
 359 static int
 360 security_problem(struct bsdtar *bsdtar, struct archive_entry *entry)
 361 {
 362         struct stat st;
 363         const char *name, *pn;
 364         char *p;
 365         int r;
 366
 367         /* -P option forces us to just accept all pathnames as-is. */
 368         if (bsdtar->option_absolute_paths)
 369                 return (0);
 370
 371         name = archive_entry_pathname(entry);
 372
 373         /* Reject any archive entry with '..' as a path element. */
 374         pn = name;
 375         while (pn != NULL && pn[0] != '\0') {
 376                 if (pn[0] == '.' && pn[1] == '.' &&
 377                     (pn[2] == '\0' || pn[2] == '/')) {
 378                         bsdtar_warnc(bsdtar, 0,
 379                             "Skipping pathname containing ..");
 380                         bsdtar->return_value = 1;
 381                         return (1);
 382                 }
 383                 pn = strchr(pn, '/');
 384                 if (pn != NULL)
 385                         pn++;
 386         }
 387
 388         /*
 389          * Gaurd against symlink tricks.  Reject any archive entry whose
 390          * destination would be altered by a symlink.
 391          */
 392         /* XXX TODO: Make this faster by comparing current path to
 393          * prefix of last successful check to avoid duplicate lstat()
 394          * calls. XXX */
 395         pn = name;
 396         if (bsdtar->security == NULL) {
 397                 bsdtar->security = malloc(sizeof(*bsdtar->security));
 398                 if (bsdtar->security == NULL)
 399                         bsdtar_errc(bsdtar, 1, errno, "No Memory");
 400                 bsdtar->security->path_size = MAXPATHLEN + 1;
 401                 bsdtar->security->path = malloc(bsdtar->security->path_size);
 402                 if (bsdtar->security->path == NULL)
 403                         bsdtar_errc(bsdtar, 1, errno, "No Memory");
 404         }
 405         if (strlen(name) >= bsdtar->security->path_size) {
 406                 free(bsdtar->security->path);
 407                 while (strlen(name) >= bsdtar->security->path_size)
 408                         bsdtar->security->path_size *= 2;
 409                 bsdtar->security->path = malloc(bsdtar->security->path_size);
 410                 if (bsdtar->security->path == NULL)
 411                         bsdtar_errc(bsdtar, 1, errno, "No Memory");
 412         }
 413         p = bsdtar->security->path;
 414         while (pn != NULL && pn[0] != '\0') {
 415                 *p++ = *pn++;
 416                 while (*pn != '\0' && *pn != '/')
 417                         *p++ = *pn++;
 418                 p[0] = '\0';
 419                 r = lstat(bsdtar->security->path, &st);
 420                 if (r != 0) {
 421                         if (errno == ENOENT)
 422                                 break;
 423                 } else if (S_ISLNK(st.st_mode)) {
 424                         if (pn[0] == '\0') {
 425                                 /*
 426                                  * Last element is symlink; remove it
 427                                  * so we can overwrite it with the
 428                                  * item being extracted.
 429                                  */
 430                                 if (!S_ISLNK(archive_entry_mode(entry))) {
 431                                         /*
 432                                          * Warn only if the symlink is being
 433                                          * replaced with a non-symlink.
 434                                          */
 435                                         bsdtar_warnc(bsdtar, 0,
 436                                             "Removing symlink %s",
 437                                             bsdtar->security->path);
 438                                 }
 439                                 if (unlink(bsdtar->security->path))
 440                                         bsdtar_errc(bsdtar, 1, errno,
 441                                             "Unlink failed");
 442                                 /* Symlink gone.  No more problem! */
 443                                 return (0);
 444                         } else if (bsdtar->option_unlink_first) {
 445                                 /* User asked us to remove problems. */
 446                                 if (unlink(bsdtar->security->path))
 447                                         bsdtar_errc(bsdtar, 1, errno,
 448                                             "Unlink failed");
 449                         } else {
 450                                 bsdtar_warnc(bsdtar, 0,
 451                                     "Cannot extract %s through symlink %s",
 452                                     name, bsdtar->security->path);
 453                                 bsdtar->return_value = 1;
 454                                 return (1);
 455                         }
 456                 }
 457         }
 458
 459         return (0);
 460 }
 461
 462 static void
 463 cleanup_security(struct bsdtar *bsdtar)
 464 {
 465         if (bsdtar->security != NULL) {
 466                 free(bsdtar->security->path);
 467                 free(bsdtar->security);
 468         }
 469 }