lib/libarchive/archive_read_support_format_mtree.c

   1 /*-
   2  * Copyright (c) 2003-2007 Tim Kientzle
   3  * Copyright (c) 2008 Joerg Sonnenberger
   4  * All rights reserved.
   5  *
   6  * Redistribution and use in source and binary forms, with or without
   7  * modification, are permitted provided that the following conditions
   8  * are met:
   9  * 1. Redistributions of source code must retain the above copyright
  10  *    notice, this list of conditions and the following disclaimer.
  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 "archive_platform.h"
  28 __FBSDID("$FreeBSD$");
  29
  30 #ifdef HAVE_SYS_STAT_H
  31 #include <sys/stat.h>
  32 #endif
  33 #ifdef HAVE_ERRNO_H
  34 #include <errno.h>
  35 #endif
  36 #ifdef HAVE_FCNTL_H
  37 #include <fcntl.h>
  38 #endif
  39 #include <stddef.h>
  40 /* #include <stdint.h> */ /* See archive_platform.h */
  41 #ifdef HAVE_STDLIB_H
  42 #include <stdlib.h>
  43 #endif
  44 #ifdef HAVE_STRING_H
  45 #include <string.h>
  46 #endif
  47
  48 #include "archive.h"
  49 #include "archive_entry.h"
  50 #include "archive_private.h"
  51 #include "archive_read_private.h"
  52 #include "archive_string.h"
  53
  54 #ifndef O_BINARY
  55 #define O_BINARY 0
  56 #endif
  57
  58 #define MTREE_HAS_DEVICE        0x0001
  59 #define MTREE_HAS_FFLAGS        0x0002
  60 #define MTREE_HAS_GID           0x0004
  61 #define MTREE_HAS_GNAME         0x0008
  62 #define MTREE_HAS_MTIME         0x0010
  63 #define MTREE_HAS_NLINK         0x0020
  64 #define MTREE_HAS_PERM          0x0040
  65 #define MTREE_HAS_SIZE          0x0080
  66 #define MTREE_HAS_TYPE          0x0100
  67 #define MTREE_HAS_UID           0x0200
  68 #define MTREE_HAS_UNAME         0x0400
  69
  70 #define MTREE_HAS_OPTIONAL      0x0800
  71
  72 struct mtree_option {
  73         struct mtree_option *next;
  74         char *value;
  75 };
  76
  77 struct mtree_entry {
  78         struct mtree_entry *next;
  79         struct mtree_option *options;
  80         char *name;
  81         char full;
  82         char used;
  83 };
  84
  85 struct mtree {
  86         struct archive_string    line;
  87         size_t                   buffsize;
  88         char                    *buff;
  89         off_t                    offset;
  90         int                      fd;
  91         int                      filetype;
  92         int                      archive_format;
  93         const char              *archive_format_name;
  94         struct mtree_entry      *entries;
  95         struct mtree_entry      *this_entry;
  96         struct archive_string    current_dir;
  97         struct archive_string    contents_name;
  98
  99         struct archive_entry_linkresolver *resolver;
 100
 101         off_t                    cur_size, cur_offset;
 102 };
 103
 104 static int      cleanup(struct archive_read *);
 105 static int      mtree_bid(struct archive_read *);
 106 static int      parse_file(struct archive_read *, struct archive_entry *,
 107                     struct mtree *, struct mtree_entry *, int *);
 108 static void     parse_escapes(char *, struct mtree_entry *);
 109 static int      parse_line(struct archive_read *, struct archive_entry *,
 110                     struct mtree *, struct mtree_entry *, int *);
 111 static int      parse_keyword(struct archive_read *, struct mtree *,
 112                     struct archive_entry *, struct mtree_option *, int *);
 113 static int      read_data(struct archive_read *a,
 114                     const void **buff, size_t *size, off_t *offset);
 115 static ssize_t  readline(struct archive_read *, struct mtree *, char **, ssize_t);
 116 static int      skip(struct archive_read *a);
 117 static int      read_header(struct archive_read *,
 118                     struct archive_entry *);
 119 static int64_t  mtree_atol10(char **);
 120 static int64_t  mtree_atol8(char **);
 121 static int64_t  mtree_atol(char **);
 122
 123 static void
 124 free_options(struct mtree_option *head)
 125 {
 126         struct mtree_option *next;
 127
 128         for (; head != NULL; head = next) {
 129                 next = head->next;
 130                 free(head->value);
 131                 free(head);
 132         }
 133 }
 134
 135 int
 136 archive_read_support_format_mtree(struct archive *_a)
 137 {
 138         struct archive_read *a = (struct archive_read *)_a;
 139         struct mtree *mtree;
 140         int r;
 141
 142         mtree = (struct mtree *)malloc(sizeof(*mtree));
 143         if (mtree == NULL) {
 144                 archive_set_error(&a->archive, ENOMEM,
 145                     "Can't allocate mtree data");
 146                 return (ARCHIVE_FATAL);
 147         }
 148         memset(mtree, 0, sizeof(*mtree));
 149         mtree->fd = -1;
 150
 151         r = __archive_read_register_format(a, mtree,
 152             mtree_bid, read_header, read_data, skip, cleanup);
 153
 154         if (r != ARCHIVE_OK)
 155                 free(mtree);
 156         return (ARCHIVE_OK);
 157 }
 158
 159 static int
 160 cleanup(struct archive_read *a)
 161 {
 162         struct mtree *mtree;
 163         struct mtree_entry *p, *q;
 164
 165         mtree = (struct mtree *)(a->format->data);
 166
 167         p = mtree->entries;
 168         while (p != NULL) {
 169                 q = p->next;
 170                 free(p->name);
 171                 free_options(p->options);
 172                 free(p);
 173                 p = q;
 174         }
 175         archive_string_free(&mtree->line);
 176         archive_string_free(&mtree->current_dir);
 177         archive_string_free(&mtree->contents_name);
 178         archive_entry_linkresolver_free(mtree->resolver);
 179
 180         free(mtree->buff);
 181         free(mtree);
 182         (a->format->data) = NULL;
 183         return (ARCHIVE_OK);
 184 }
 185
 186
 187 static int
 188 mtree_bid(struct archive_read *a)
 189 {
 190         struct mtree *mtree;
 191         ssize_t bytes_read;
 192         const void *h;
 193         const char *signature = "#mtree";
 194         const char *p;
 195         int bid;
 196
 197         mtree = (struct mtree *)(a->format->data);
 198
 199         /* Now let's look at the actual header and see if it matches. */
 200         bytes_read = (a->decompressor->read_ahead)(a, &h, strlen(signature));
 201
 202         if (bytes_read <= 0)
 203                 return (bytes_read);
 204
 205         p = h;
 206         bid = 0;
 207         while (bytes_read > 0 && *signature != '\0') {
 208                 if (*p != *signature)
 209                         return (bid = 0);
 210                 bid += 8;
 211                 p++;
 212                 signature++;
 213                 bytes_read--;
 214         }
 215         return (bid);
 216 }
 217
 218 /*
 219  * The extended mtree format permits multiple lines specifying
 220  * attributes for each file.  For those entries, only the last line
 221  * is actually used.  Practically speaking, that means we have
 222  * to read the entire mtree file into memory up front.
 223  *
 224  * The parsing is done in two steps.  First, it is decided if a line
 225  * changes the global defaults and if it is, processed accordingly.
 226  * Otherwise, the options of the line are merged with the current
 227  * global options.
 228  */
 229 static int
 230 add_option(struct archive_read *a, struct mtree_option **global,
 231     const char *value, size_t len)
 232 {
 233         struct mtree_option *option;
 234
 235         if ((option = malloc(sizeof(*option))) == NULL) {
 236                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 237                 return (ARCHIVE_FATAL);
 238         }
 239         if ((option->value = malloc(len + 1)) == NULL) {
 240                 free(option);
 241                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 242                 return (ARCHIVE_FATAL);
 243         }
 244         memcpy(option->value, value, len);
 245         option->value[len] = '\0';
 246         option->next = *global;
 247         *global = option;
 248         return (ARCHIVE_OK);
 249 }
 250
 251 static void
 252 remove_option(struct mtree_option **global, const char *value, size_t len)
 253 {
 254         struct mtree_option *iter, *last;
 255
 256         last = NULL;
 257         for (iter = *global; iter != NULL; last = iter, iter = iter->next) {
 258                 if (strncmp(iter->value, value, len) == 0 &&
 259                     (iter->value[len] == '\0' ||
 260                      iter->value[len] == '='))
 261                         break;
 262         }
 263         if (iter == NULL)
 264                 return;
 265         if (last == NULL)
 266                 *global = iter->next;
 267         else
 268                 last->next = iter->next;
 269
 270         free(iter->value);
 271         free(iter);
 272 }
 273
 274 static int
 275 process_global_set(struct archive_read *a,
 276     struct mtree_option **global, const char *line)
 277 {
 278         const char *next, *eq;
 279         size_t len;
 280         int r;
 281
 282         line += 4;
 283         for (;;) {
 284                 next = line + strspn(line, " \t\r\n");
 285                 if (*next == '\0')
 286                         return (ARCHIVE_OK);
 287                 line = next;
 288                 next = line + strcspn(line, " \t\r\n");
 289                 eq = strchr(line, '=');
 290                 if (eq > next)
 291                         len = next - line;
 292                 else
 293                         len = eq - line;
 294
 295                 remove_option(global, line, len);
 296                 r = add_option(a, global, line, next - line);
 297                 if (r != ARCHIVE_OK)
 298                         return (r);
 299                 line = next;
 300         }
 301 }
 302
 303 static int
 304 process_global_unset(struct archive_read *a,
 305     struct mtree_option **global, const char *line)
 306 {
 307         const char *next;
 308         size_t len;
 309
 310         line += 6;
 311         if ((next = strchr(line, '=')) != NULL) {
 312                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 313                     "/unset shall not contain `='");
 314                 return ARCHIVE_FATAL;
 315         }
 316
 317         for (;;) {
 318                 next = line + strspn(line, " \t\r\n");
 319                 if (*next == '\0')
 320                         return (ARCHIVE_OK);
 321                 line = next;
 322                 len = strcspn(line, " \t\r\n");
 323
 324                 if (len == 3 && strncmp(line, "all", 3) == 0) {
 325                         free_options(*global);
 326                         *global = NULL;
 327                 } else {
 328                         remove_option(global, line, len);
 329                 }
 330
 331                 line += len;
 332         }
 333 }
 334
 335 static int
 336 process_add_entry(struct archive_read *a, struct mtree *mtree,
 337     struct mtree_option **global, const char *line,
 338     struct mtree_entry **last_entry)
 339 {
 340         struct mtree_entry *entry;
 341         struct mtree_option *iter;
 342         const char *next, *eq;
 343         size_t len;
 344         int r;
 345
 346         if ((entry = malloc(sizeof(*entry))) == NULL) {
 347                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 348                 return (ARCHIVE_FATAL);
 349         }
 350         entry->next = NULL;
 351         entry->options = NULL;
 352         entry->name = NULL;
 353         entry->used = 0;
 354         entry->full = 0;
 355
 356         /* Add this entry to list. */
 357         if (*last_entry == NULL)
 358                 mtree->entries = entry;
 359         else
 360                 (*last_entry)->next = entry;
 361         *last_entry = entry;
 362
 363         len = strcspn(line, " \t\r\n");
 364         if ((entry->name = malloc(len + 1)) == NULL) {
 365                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 366                 return (ARCHIVE_FATAL);
 367         }
 368
 369         memcpy(entry->name, line, len);
 370         entry->name[len] = '\0';
 371         parse_escapes(entry->name, entry);
 372
 373         line += len;
 374         for (iter = *global; iter != NULL; iter = iter->next) {
 375                 r = add_option(a, &entry->options, iter->value,
 376                     strlen(iter->value));
 377                 if (r != ARCHIVE_OK)
 378                         return (r);
 379         }
 380
 381         for (;;) {
 382                 next = line + strspn(line, " \t\r\n");
 383                 if (*next == '\0')
 384                         return (ARCHIVE_OK);
 385                 line = next;
 386                 next = line + strcspn(line, " \t\r\n");
 387                 eq = strchr(line, '=');
 388                 if (eq > next)
 389                         len = next - line;
 390                 else
 391                         len = eq - line;
 392
 393                 remove_option(&entry->options, line, len);
 394                 r = add_option(a, &entry->options, line, next - line);
 395                 if (r != ARCHIVE_OK)
 396                         return (r);
 397                 line = next;
 398         }
 399 }
 400
 401 static int
 402 read_mtree(struct archive_read *a, struct mtree *mtree)
 403 {
 404         ssize_t len;
 405         uintmax_t counter;
 406         char *p;
 407         struct mtree_option *global;
 408         struct mtree_entry *last_entry;
 409         int r;
 410
 411         mtree->archive_format = ARCHIVE_FORMAT_MTREE_V1;
 412         mtree->archive_format_name = "mtree";
 413
 414         global = NULL;
 415         last_entry = NULL;
 416         r = ARCHIVE_OK;
 417
 418         for (counter = 1; ; ++counter) {
 419                 len = readline(a, mtree, &p, 256);
 420                 if (len == 0) {
 421                         mtree->this_entry = mtree->entries;
 422                         return (ARCHIVE_OK);
 423                 }
 424                 if (len < 0)
 425                         return (len);
 426                 /* Leading whitespace is never significant, ignore it. */
 427                 while (*p == ' ' || *p == '\t') {
 428                         ++p;
 429                         --len;
 430                 }
 431                 /* Skip content lines and blank lines. */
 432                 if (*p == '#')
 433                         continue;
 434                 if (*p == '\r' || *p == '\n' || *p == '\0')
 435                         continue;
 436                 if (*p != '/') {
 437                         r = process_add_entry(a, mtree, &global, p,
 438                             &last_entry);
 439                 } else if (strncmp(p, "/set", 4) == 0) {
 440                         if (p[4] != ' ' && p[4] != '\t')
 441                                 break;
 442                         r = process_global_set(a, &global, p);
 443                 } else if (strncmp(p, "/unset", 6) == 0) {
 444                         if (p[6] != ' ' && p[6] != '\t')
 445                                 break;
 446                         r = process_global_unset(a, &global, p);
 447                 } else
 448                         break;
 449
 450                 if (r != ARCHIVE_OK)
 451                         return r;
 452         }
 453
 454         archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
 455             "Can't parse line %ju", counter);
 456         return ARCHIVE_FATAL;
 457 }
 458
 459 /*
 460  * Read in the entire mtree file into memory on the first request.
 461  * Then use the next unused file to satisfy each header request.
 462  */
 463 static int
 464 read_header(struct archive_read *a, struct archive_entry *entry)
 465 {
 466         struct mtree *mtree;
 467         char *p;
 468         int r, use_next;
 469
 470         mtree = (struct mtree *)(a->format->data);
 471
 472         if (mtree->fd >= 0) {
 473                 close(mtree->fd);
 474                 mtree->fd = -1;
 475         }
 476
 477         if (mtree->entries == NULL) {
 478                 mtree->resolver = archive_entry_linkresolver_new();
 479                 if (mtree->resolver == NULL)
 480                         return ARCHIVE_FATAL;
 481                 archive_entry_linkresolver_set_strategy(mtree->resolver,
 482                     ARCHIVE_FORMAT_MTREE);
 483                 r = read_mtree(a, mtree);
 484                 if (r != ARCHIVE_OK)
 485                         return (r);
 486         }
 487
 488         a->archive.archive_format = mtree->archive_format;
 489         a->archive.archive_format_name = mtree->archive_format_name;
 490
 491         for (;;) {
 492                 if (mtree->this_entry == NULL)
 493                         return (ARCHIVE_EOF);
 494                 if (strcmp(mtree->this_entry->name, "..") == 0) {
 495                         mtree->this_entry->used = 1;
 496                         if (archive_strlen(&mtree->current_dir) > 0) {
 497                                 /* Roll back current path. */
 498                                 p = mtree->current_dir.s
 499                                     + mtree->current_dir.length - 1;
 500                                 while (p >= mtree->current_dir.s && *p != '/')
 501                                         --p;
 502                                 if (p >= mtree->current_dir.s)
 503                                         --p;
 504                                 mtree->current_dir.length
 505                                     = p - mtree->current_dir.s + 1;
 506                         }
 507                 }
 508                 if (!mtree->this_entry->used) {
 509                         use_next = 0;
 510                         r = parse_file(a, entry, mtree, mtree->this_entry, &use_next);
 511                         if (use_next == 0)
 512                                 return (r);
 513                 }
 514                 mtree->this_entry = mtree->this_entry->next;
 515         }
 516 }
 517
 518 /*
 519  * A single file can have multiple lines contribute specifications.
 520  * Parse as many lines as necessary, then pull additional information
 521  * from a backing file on disk as necessary.
 522  */
 523 static int
 524 parse_file(struct archive_read *a, struct archive_entry *entry,
 525     struct mtree *mtree, struct mtree_entry *mentry, int *use_next)
 526 {
 527         const char *path;
 528         struct stat st_storage, *st;
 529         struct mtree_entry *mp;
 530         struct archive_entry *sparse_entry;
 531         int r = ARCHIVE_OK, r1, parsed_kws, mismatched_type;
 532
 533         mentry->used = 1;
 534
 535         /* Initialize reasonable defaults. */
 536         mtree->filetype = AE_IFREG;
 537         archive_entry_set_size(entry, 0);
 538
 539         /* Parse options from this line. */
 540         parsed_kws = 0;
 541         r = parse_line(a, entry, mtree, mentry, &parsed_kws);
 542
 543         if (mentry->full) {
 544                 archive_entry_copy_pathname(entry, mentry->name);
 545                 /*
 546                  * "Full" entries are allowed to have multiple lines
 547                  * and those lines aren't required to be adjacent.  We
 548                  * don't support multiple lines for "relative" entries
 549                  * nor do we make any attempt to merge data from
 550                  * separate "relative" and "full" entries.  (Merging
 551                  * "relative" and "full" entries would require dealing
 552                  * with pathname canonicalization, which is a very
 553                  * tricky subject.)
 554                  */
 555                 for (mp = mentry->next; mp != NULL; mp = mp->next) {
 556                         if (mp->full && !mp->used
 557                             && strcmp(mentry->name, mp->name) == 0) {
 558                                 /* Later lines override earlier ones. */
 559                                 mp->used = 1;
 560                                 r1 = parse_line(a, entry, mtree, mp,
 561                                     &parsed_kws);
 562                                 if (r1 < r)
 563                                         r = r1;
 564                         }
 565                 }
 566         } else {
 567                 /*
 568                  * Relative entries require us to construct
 569                  * the full path and possibly update the
 570                  * current directory.
 571                  */
 572                 size_t n = archive_strlen(&mtree->current_dir);
 573                 if (n > 0)
 574                         archive_strcat(&mtree->current_dir, "/");
 575                 archive_strcat(&mtree->current_dir, mentry->name);
 576                 archive_entry_copy_pathname(entry, mtree->current_dir.s);
 577                 if (archive_entry_filetype(entry) != AE_IFDIR)
 578                         mtree->current_dir.length = n;
 579         }
 580
 581         /*
 582          * Try to open and stat the file to get the real size
 583          * and other file info.  It would be nice to avoid
 584          * this here so that getting a listing of an mtree
 585          * wouldn't require opening every referenced contents
 586          * file.  But then we wouldn't know the actual
 587          * contents size, so I don't see a really viable way
 588          * around this.  (Also, we may want to someday pull
 589          * other unspecified info from the contents file on
 590          * disk.)
 591          */
 592         mtree->fd = -1;
 593         if (archive_strlen(&mtree->contents_name) > 0)
 594                 path = mtree->contents_name.s;
 595         else
 596                 path = archive_entry_pathname(entry);
 597
 598         if (archive_entry_filetype(entry) == AE_IFREG ||
 599             archive_entry_filetype(entry) == AE_IFDIR) {
 600                 mtree->fd = open(path,
 601                     O_RDONLY | O_BINARY);
 602                 if (mtree->fd == -1 &&
 603                     (errno != ENOENT ||
 604                      archive_strlen(&mtree->contents_name) > 0)) {
 605                         archive_set_error(&a->archive, errno,
 606                             "Can't open %s", path);
 607                         r = ARCHIVE_WARN;
 608                 }
 609         }
 610
 611         st = &st_storage;
 612         if (mtree->fd >= 0) {
 613                 if (fstat(mtree->fd, st) == -1) {
 614                         archive_set_error(&a->archive, errno,
 615                             "Could not fstat %s", path);
 616                         r = ARCHIVE_WARN;
 617                         /* If we can't stat it, don't keep it open. */
 618                         close(mtree->fd);
 619                         mtree->fd = -1;
 620                         st = NULL;
 621                 }
 622         } else if (lstat(path, st) == -1) {
 623                 st = NULL;
 624         }
 625
 626         /*
 627          * If there is a contents file on disk, use that size;
 628          * otherwise leave it as-is (it might have been set from
 629          * the mtree size= keyword).
 630          */
 631         if (st != NULL) {
 632                 mismatched_type = 0;
 633                 if ((st->st_mode & S_IFMT) == S_IFREG &&
 634                     archive_entry_filetype(entry) != AE_IFREG)
 635                         mismatched_type = 1;
 636                 if ((st->st_mode & S_IFMT) == S_IFLNK &&
 637                     archive_entry_filetype(entry) != AE_IFLNK)
 638                         mismatched_type = 1;
 639                 if ((st->st_mode & S_IFSOCK) == S_IFSOCK &&
 640                     archive_entry_filetype(entry) != AE_IFSOCK)
 641                         mismatched_type = 1;
 642                 if ((st->st_mode & S_IFMT) == S_IFCHR &&
 643                     archive_entry_filetype(entry) != AE_IFCHR)
 644                         mismatched_type = 1;
 645                 if ((st->st_mode & S_IFMT) == S_IFBLK &&
 646                     archive_entry_filetype(entry) != AE_IFBLK)
 647                         mismatched_type = 1;
 648                 if ((st->st_mode & S_IFMT) == S_IFDIR &&
 649                     archive_entry_filetype(entry) != AE_IFDIR)
 650                         mismatched_type = 1;
 651                 if ((st->st_mode & S_IFMT) == S_IFIFO &&
 652                     archive_entry_filetype(entry) != AE_IFIFO)
 653                         mismatched_type = 1;
 654
 655                 if (mismatched_type) {
 656                         if ((parsed_kws & MTREE_HAS_OPTIONAL) == 0) {
 657                                 archive_set_error(&a->archive,
 658                                     ARCHIVE_ERRNO_MISC,
 659                                     "mtree specification has different type for %s",
 660                                     archive_entry_pathname(entry));
 661                                 r = ARCHIVE_WARN;
 662                         } else {
 663                                 *use_next = 1;
 664                         }
 665                         /* Don't hold a non-regular file open. */
 666                         close(mtree->fd);
 667                         mtree->fd = -1;
 668                         st = NULL;
 669                         return r;
 670                 }
 671         }
 672
 673         if (st != NULL) {
 674                 if ((parsed_kws & MTREE_HAS_DEVICE) == 0 &&
 675                     (archive_entry_filetype(entry) == AE_IFCHR ||
 676                      archive_entry_filetype(entry) == AE_IFBLK))
 677                         archive_entry_set_rdev(entry, st->st_rdev);
 678                 if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME)) == 0)
 679                         archive_entry_set_gid(entry, st->st_gid);
 680                 if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME)) == 0)
 681                         archive_entry_set_uid(entry, st->st_uid);
 682                 if ((parsed_kws & MTREE_HAS_MTIME) == 0) {
 683 #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC
 684                         archive_entry_set_mtime(entry, st->st_mtime,
 685                             st->st_mtimespec.tv_nsec);
 686 #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC
 687                         archive_entry_set_mtime(entry, st->st_mtime,
 688                             st->st_mtim.tv_nsec);
 689 #else
 690                         archive_entry_set_mtime(entry, st->st_mtime, 0);
 691 #endif
 692                 }
 693                 if ((parsed_kws & MTREE_HAS_NLINK) == 0)
 694                         archive_entry_set_nlink(entry, st->st_nlink);
 695                 if ((parsed_kws & MTREE_HAS_PERM) == 0)
 696                         archive_entry_set_perm(entry, st->st_mode);
 697                 if ((parsed_kws & MTREE_HAS_SIZE) == 0)
 698                         archive_entry_set_size(entry, st->st_size);
 699                 archive_entry_set_ino(entry, st->st_ino);
 700                 archive_entry_set_dev(entry, st->st_dev);
 701
 702                 archive_entry_linkify(mtree->resolver, &entry, &sparse_entry);
 703         } else if (parsed_kws & MTREE_HAS_OPTIONAL) {
 704                 /*
 705                  * Couldn't open the entry, stat it or the on-disk type
 706                  * didn't match.  If this entry is optional, just ignore it
 707                  * and read the next header entry.
 708                  */
 709                 *use_next = 1;
 710                 return ARCHIVE_OK;
 711         }
 712
 713         mtree->cur_size = archive_entry_size(entry);
 714         mtree->offset = 0;
 715
 716         return r;
 717 }
 718
 719 /*
 720  * Each line contains a sequence of keywords.
 721  */
 722 static int
 723 parse_line(struct archive_read *a, struct archive_entry *entry,
 724     struct mtree *mtree, struct mtree_entry *mp, int *parsed_kws)
 725 {
 726         struct mtree_option *iter;
 727         int r = ARCHIVE_OK, r1;
 728
 729         for (iter = mp->options; iter != NULL; iter = iter->next) {
 730                 r1 = parse_keyword(a, mtree, entry, iter, parsed_kws);
 731                 if (r1 < r)
 732                         r = r1;
 733         }
 734         if ((*parsed_kws & MTREE_HAS_TYPE) == 0) {
 735                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
 736                     "Missing type keyword in mtree specification");
 737                 return (ARCHIVE_WARN);
 738         }
 739         return (r);
 740 }
 741
 742 /*
 743  * Device entries have one of the following forms:
 744  * raw dev_t
 745  * format,major,minor[,subdevice]
 746  *
 747  * Just use major and minor, no translation etc is done
 748  * between formats.
 749  */
 750 static int
 751 parse_device(struct archive *a, struct archive_entry *entry, char *val)
 752 {
 753         char *comma1, *comma2;
 754
 755         comma1 = strchr(val, ',');
 756         if (comma1 == NULL) {
 757                 archive_entry_set_dev(entry, mtree_atol10(&val));
 758                 return (ARCHIVE_OK);
 759         }
 760         ++comma1;
 761         comma2 = strchr(comma1, ',');
 762         if (comma1 == NULL) {
 763                 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
 764                     "Malformed device attribute");
 765                 return (ARCHIVE_WARN);
 766         }
 767         ++comma2;
 768         archive_entry_set_rdevmajor(entry, mtree_atol(&comma1));
 769         archive_entry_set_rdevminor(entry, mtree_atol(&comma2));
 770         return (ARCHIVE_OK);
 771 }
 772
 773 /*
 774  * Parse a single keyword and its value.
 775  */
 776 static int
 777 parse_keyword(struct archive_read *a, struct mtree *mtree,
 778     struct archive_entry *entry, struct mtree_option *option, int *parsed_kws)
 779 {
 780         char *val, *key;
 781
 782         key = option->value;
 783
 784         if (*key == '\0')
 785                 return (ARCHIVE_OK);
 786
 787         if (strcmp(key, "optional") == 0) {
 788                 *parsed_kws |= MTREE_HAS_OPTIONAL;
 789                 return (ARCHIVE_OK);
 790         }
 791         if (strcmp(key, "ignore") == 0) {
 792                 /*
 793                  * The mtree processing is not recursive, so
 794                  * recursion will only happen for explicitly listed
 795                  * entries.
 796                  */
 797                 return (ARCHIVE_OK);
 798         }
 799
 800         val = strchr(key, '=');
 801         if (val == NULL) {
 802                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
 803                     "Malformed attribute \"%s\" (%d)", key, key[0]);
 804                 return (ARCHIVE_WARN);
 805         }
 806
 807         *val = '\0';
 808         ++val;
 809
 810         switch (key[0]) {
 811         case 'c':
 812                 if (strcmp(key, "content") == 0
 813                     || strcmp(key, "contents") == 0) {
 814                         parse_escapes(val, NULL);
 815                         archive_strcpy(&mtree->contents_name, val);
 816                         break;
 817                 }
 818                 if (strcmp(key, "cksum") == 0)
 819                         break;
 820         case 'd':
 821                 if (strcmp(key, "device") == 0) {
 822                         *parsed_kws |= MTREE_HAS_DEVICE;
 823                         return parse_device(&a->archive, entry, val);
 824                 }
 825         case 'f':
 826                 if (strcmp(key, "flags") == 0) {
 827                         *parsed_kws |= MTREE_HAS_FFLAGS;
 828                         archive_entry_copy_fflags_text(entry, val);
 829                         break;
 830                 }
 831         case 'g':
 832                 if (strcmp(key, "gid") == 0) {
 833                         *parsed_kws |= MTREE_HAS_GID;
 834                         archive_entry_set_gid(entry, mtree_atol10(&val));
 835                         break;
 836                 }
 837                 if (strcmp(key, "gname") == 0) {
 838                         *parsed_kws |= MTREE_HAS_GNAME;
 839                         archive_entry_copy_gname(entry, val);
 840                         break;
 841                 }
 842         case 'l':
 843                 if (strcmp(key, "link") == 0) {
 844                         archive_entry_copy_symlink(entry, val);
 845                         break;
 846                 }
 847         case 'm':
 848                 if (strcmp(key, "md5") == 0 || strcmp(key, "md5digest") == 0)
 849                         break;
 850                 if (strcmp(key, "mode") == 0) {
 851                         if (val[0] >= '0' && val[0] <= '9') {
 852                                 *parsed_kws |= MTREE_HAS_PERM;
 853                                 archive_entry_set_perm(entry,
 854                                     mtree_atol8(&val));
 855                         } else {
 856                                 archive_set_error(&a->archive,
 857                                     ARCHIVE_ERRNO_FILE_FORMAT,
 858                                     "Symbolic mode \"%s\" unsupported", val);
 859                                 return ARCHIVE_WARN;
 860                         }
 861                         break;
 862                 }
 863         case 'n':
 864                 if (strcmp(key, "nlink") == 0) {
 865                         *parsed_kws |= MTREE_HAS_NLINK;
 866                         archive_entry_set_nlink(entry, mtree_atol10(&val));
 867                         break;
 868                 }
 869         case 'r':
 870                 if (strcmp(key, "rmd160") == 0 ||
 871                     strcmp(key, "rmd160digest") == 0)
 872                         break;
 873         case 's':
 874                 if (strcmp(key, "sha1") == 0 || strcmp(key, "sha1digest") == 0)
 875                         break;
 876                 if (strcmp(key, "sha256") == 0 ||
 877                     strcmp(key, "sha256digest") == 0)
 878                         break;
 879                 if (strcmp(key, "sha384") == 0 ||
 880                     strcmp(key, "sha384digest") == 0)
 881                         break;
 882                 if (strcmp(key, "sha512") == 0 ||
 883                     strcmp(key, "sha512digest") == 0)
 884                         break;
 885                 if (strcmp(key, "size") == 0) {
 886                         archive_entry_set_size(entry, mtree_atol10(&val));
 887                         break;
 888                 }
 889         case 't':
 890                 if (strcmp(key, "tags") == 0) {
 891                         /*
 892                          * Comma delimited list of tags.
 893                          * Ignore the tags for now, but the interface
 894                          * should be extended to allow inclusion/exclusion.
 895                          */
 896                         break;
 897                 }
 898                 if (strcmp(key, "time") == 0) {
 899                         *parsed_kws |= MTREE_HAS_MTIME;
 900                         archive_entry_set_mtime(entry, mtree_atol10(&val), 0);
 901                         break;
 902                 }
 903                 if (strcmp(key, "type") == 0) {
 904                         *parsed_kws |= MTREE_HAS_TYPE;
 905                         switch (val[0]) {
 906                         case 'b':
 907                                 if (strcmp(val, "block") == 0) {
 908                                         mtree->filetype = AE_IFBLK;
 909                                         break;
 910                                 }
 911                         case 'c':
 912                                 if (strcmp(val, "char") == 0) {
 913                                         mtree->filetype = AE_IFCHR;
 914                                         break;
 915                                 }
 916                         case 'd':
 917                                 if (strcmp(val, "dir") == 0) {
 918                                         mtree->filetype = AE_IFDIR;
 919                                         break;
 920                                 }
 921                         case 'f':
 922                                 if (strcmp(val, "fifo") == 0) {
 923                                         mtree->filetype = AE_IFIFO;
 924                                         break;
 925                                 }
 926                                 if (strcmp(val, "file") == 0) {
 927                                         mtree->filetype = AE_IFREG;
 928                                         break;
 929                                 }
 930                         case 'l':
 931                                 if (strcmp(val, "link") == 0) {
 932                                         mtree->filetype = AE_IFLNK;
 933                                         break;
 934                                 }
 935                         default:
 936                                 archive_set_error(&a->archive,
 937                                     ARCHIVE_ERRNO_FILE_FORMAT,
 938                                     "Unrecognized file type \"%s\"", val);
 939                                 return (ARCHIVE_WARN);
 940                         }
 941                         archive_entry_set_filetype(entry, mtree->filetype);
 942                         break;
 943                 }
 944         case 'u':
 945                 if (strcmp(key, "uid") == 0) {
 946                         *parsed_kws |= MTREE_HAS_UID;
 947                         archive_entry_set_uid(entry, mtree_atol10(&val));
 948                         break;
 949                 }
 950                 if (strcmp(key, "uname") == 0) {
 951                         *parsed_kws |= MTREE_HAS_UNAME;
 952                         archive_entry_copy_uname(entry, val);
 953                         break;
 954                 }
 955         default:
 956                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
 957                     "Unrecognized key %s=%s", key, val);
 958                 return (ARCHIVE_WARN);
 959         }
 960         return (ARCHIVE_OK);
 961 }
 962
 963 static int
 964 read_data(struct archive_read *a, const void **buff, size_t *size, off_t *offset)
 965 {
 966         size_t bytes_to_read;
 967         ssize_t bytes_read;
 968         struct mtree *mtree;
 969
 970         mtree = (struct mtree *)(a->format->data);
 971         if (mtree->fd < 0) {
 972                 *buff = NULL;
 973                 *offset = 0;
 974                 *size = 0;
 975                 return (ARCHIVE_EOF);
 976         }
 977         if (mtree->buff == NULL) {
 978                 mtree->buffsize = 64 * 1024;
 979                 mtree->buff = malloc(mtree->buffsize);
 980                 if (mtree->buff == NULL) {
 981                         archive_set_error(&a->archive, ENOMEM,
 982                             "Can't allocate memory");
 983                 }
 984                 return (ARCHIVE_FATAL);
 985         }
 986
 987         *buff = mtree->buff;
 988         *offset = mtree->offset;
 989         if ((off_t)mtree->buffsize > mtree->cur_size - mtree->offset)
 990                 bytes_to_read = mtree->cur_size - mtree->offset;
 991         else
 992                 bytes_to_read = mtree->buffsize;
 993         bytes_read = read(mtree->fd, mtree->buff, bytes_to_read);
 994         if (bytes_read < 0) {
 995                 archive_set_error(&a->archive, errno, "Can't read");
 996                 return (ARCHIVE_WARN);
 997         }
 998         if (bytes_read == 0) {
 999                 *size = 0;
1000                 return (ARCHIVE_EOF);
1001         }
1002         mtree->offset += bytes_read;
1003         *size = bytes_read;
1004         return (ARCHIVE_OK);
1005 }
1006
1007 /* Skip does nothing except possibly close the contents file. */
1008 static int
1009 skip(struct archive_read *a)
1010 {
1011         struct mtree *mtree;
1012
1013         mtree = (struct mtree *)(a->format->data);
1014         if (mtree->fd >= 0) {
1015                 close(mtree->fd);
1016                 mtree->fd = -1;
1017         }
1018         return (ARCHIVE_OK);
1019 }
1020
1021 /*
1022  * Since parsing octal escapes always makes strings shorter,
1023  * we can always do this conversion in-place.
1024  */
1025 static void
1026 parse_escapes(char *src, struct mtree_entry *mentry)
1027 {
1028         char *dest = src;
1029         char c;
1030
1031         /*
1032          * The current directory is somewhat special, it should be archived
1033          * only once as it will confuse extraction otherwise.
1034          */
1035         if (strcmp(src, ".") == 0)
1036                 mentry->full = 1;
1037
1038         while (*src != '\0') {
1039                 c = *src++;
1040                 if (c == '/' && mentry != NULL)
1041                         mentry->full = 1;
1042                 if (c == '\\') {
1043                         if (src[0] >= '0' && src[0] <= '3'
1044                             && src[1] >= '0' && src[1] <= '7'
1045                             && src[2] >= '0' && src[2] <= '7') {
1046                                 c = (src[0] - '0') << 6;
1047                                 c |= (src[1] - '0') << 3;
1048                                 c |= (src[2] - '0');
1049                                 src += 3;
1050                         }
1051                 }
1052                 *dest++ = c;
1053         }
1054         *dest = '\0';
1055 }
1056
1057 /*
1058  * Note that this implementation does not (and should not!) obey
1059  * locale settings; you cannot simply substitute strtol here, since
1060  * it does obey locale.
1061  */
1062 static int64_t
1063 mtree_atol8(char **p)
1064 {
1065         int64_t l, limit, last_digit_limit;
1066         int digit, base;
1067
1068         base = 8;
1069         limit = INT64_MAX / base;
1070         last_digit_limit = INT64_MAX % base;
1071
1072         l = 0;
1073         digit = **p - '0';
1074         while (digit >= 0 && digit < base) {
1075                 if (l>limit || (l == limit && digit > last_digit_limit)) {
1076                         l = INT64_MAX; /* Truncate on overflow. */
1077                         break;
1078                 }
1079                 l = (l * base) + digit;
1080                 digit = *++(*p) - '0';
1081         }
1082         return (l);
1083 }
1084
1085 /*
1086  * Note that this implementation does not (and should not!) obey
1087  * locale settings; you cannot simply substitute strtol here, since
1088  * it does obey locale.
1089  */
1090 static int64_t
1091 mtree_atol10(char **p)
1092 {
1093         int64_t l, limit, last_digit_limit;
1094         int base, digit, sign;
1095
1096         base = 10;
1097         limit = INT64_MAX / base;
1098         last_digit_limit = INT64_MAX % base;
1099
1100         if (**p == '-') {
1101                 sign = -1;
1102                 ++(*p);
1103         } else
1104                 sign = 1;
1105
1106         l = 0;
1107         digit = **p - '0';
1108         while (digit >= 0 && digit < base) {
1109                 if (l > limit || (l == limit && digit > last_digit_limit)) {
1110                         l = UINT64_MAX; /* Truncate on overflow. */
1111                         break;
1112                 }
1113                 l = (l * base) + digit;
1114                 digit = *++(*p) - '0';
1115         }
1116         return (sign < 0) ? -l : l;
1117 }
1118
1119 /*
1120  * Note that this implementation does not (and should not!) obey
1121  * locale settings; you cannot simply substitute strtol here, since
1122  * it does obey locale.
1123  */
1124 static int64_t
1125 mtree_atol16(char **p)
1126 {
1127         int64_t l, limit, last_digit_limit;
1128         int base, digit, sign;
1129
1130         base = 16;
1131         limit = INT64_MAX / base;
1132         last_digit_limit = INT64_MAX % base;
1133
1134         if (**p == '-') {
1135                 sign = -1;
1136                 ++(*p);
1137         } else
1138                 sign = 1;
1139
1140         l = 0;
1141         if (**p >= '0' && **p <= '9')
1142                 digit = **p - '0';
1143         else if (**p >= 'a' && **p <= 'f')
1144                 digit = **p - 'a' + 10;
1145         else if (**p >= 'A' && **p <= 'F')
1146                 digit = **p - 'A' + 10;
1147         else
1148                 digit = -1;
1149         while (digit >= 0 && digit < base) {
1150                 if (l > limit || (l == limit && digit > last_digit_limit)) {
1151                         l = UINT64_MAX; /* Truncate on overflow. */
1152                         break;
1153                 }
1154                 l = (l * base) + digit;
1155                 if (**p >= '0' && **p <= '9')
1156                         digit = **p - '0';
1157                 else if (**p >= 'a' && **p <= 'f')
1158                         digit = **p - 'a' + 10;
1159                 else if (**p >= 'A' && **p <= 'F')
1160                         digit = **p - 'A' + 10;
1161                 else
1162                         digit = -1;
1163         }
1164         return (sign < 0) ? -l : l;
1165 }
1166
1167 static int64_t
1168 mtree_atol(char **p)
1169 {
1170         if (**p != '0')
1171                 return mtree_atol10(p);
1172         if ((*p)[1] == 'x' || (*p)[1] == 'X') {
1173                 *p += 2;
1174                 return mtree_atol16(p);
1175         }
1176         return mtree_atol8(p);
1177 }
1178
1179 /*
1180  * Returns length of line (including trailing newline)
1181  * or negative on error.  'start' argument is updated to
1182  * point to first character of line.
1183  */
1184 static ssize_t
1185 readline(struct archive_read *a, struct mtree *mtree, char **start, ssize_t limit)
1186 {
1187         ssize_t bytes_read;
1188         ssize_t total_size = 0;
1189         const void *t;
1190         const char *s;
1191         void *p;
1192
1193         /* Accumulate line in a line buffer. */
1194         for (;;) {
1195                 /* Read some more. */
1196                 bytes_read = (a->decompressor->read_ahead)(a, &t, 1);
1197                 if (bytes_read == 0)
1198                         return (0);
1199                 if (bytes_read < 0)
1200                         return (ARCHIVE_FATAL);
1201                 s = t;  /* Start of line? */
1202                 p = memchr(t, '\n', bytes_read);
1203                 /* If we found '\n', trim the read. */
1204                 if (p != NULL) {
1205                         bytes_read = 1 + ((const char *)p) - s;
1206                 }
1207                 if (total_size + bytes_read + 1 > limit) {
1208                         archive_set_error(&a->archive,
1209                             ARCHIVE_ERRNO_FILE_FORMAT,
1210                             "Line too long");
1211                         return (ARCHIVE_FATAL);
1212                 }
1213                 if (archive_string_ensure(&mtree->line,
1214                         total_size + bytes_read + 1) == NULL) {
1215                         archive_set_error(&a->archive, ENOMEM,
1216                             "Can't allocate working buffer");
1217                         return (ARCHIVE_FATAL);
1218                 }
1219                 memcpy(mtree->line.s + total_size, t, bytes_read);
1220                 (a->decompressor->consume)(a, bytes_read);
1221                 total_size += bytes_read;
1222                 /* Null terminate. */
1223                 mtree->line.s[total_size] = '\0';
1224                 /* If we found '\n', clean up and return. */
1225                 if (p != NULL) {
1226                         *start = mtree->line.s;
1227                         return (total_size);
1228                 }
1229         }
1230 }