contrib/libarchive/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: src/lib/libarchive/archive_read_support_format_mtree.c,v 1.11 2008/12/06 06:45:15 kientzle Exp $");
  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, "mtree",
 152             mtree_bid, NULL, 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         const char *signature = "#mtree";
 191         const char *p;
 192
 193         /* Now let's look at the actual header and see if it matches. */
 194         p = __archive_read_ahead(a, strlen(signature), NULL);
 195         if (p == NULL)
 196                 return (-1);
 197
 198         if (strncmp(p, signature, strlen(signature)) == 0)
 199                 return (8 * strlen(signature));
 200         return (0);
 201 }
 202
 203 /*
 204  * The extended mtree format permits multiple lines specifying
 205  * attributes for each file.  For those entries, only the last line
 206  * is actually used.  Practically speaking, that means we have
 207  * to read the entire mtree file into memory up front.
 208  *
 209  * The parsing is done in two steps.  First, it is decided if a line
 210  * changes the global defaults and if it is, processed accordingly.
 211  * Otherwise, the options of the line are merged with the current
 212  * global options.
 213  */
 214 static int
 215 add_option(struct archive_read *a, struct mtree_option **global,
 216     const char *value, size_t len)
 217 {
 218         struct mtree_option *option;
 219
 220         if ((option = malloc(sizeof(*option))) == NULL) {
 221                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 222                 return (ARCHIVE_FATAL);
 223         }
 224         if ((option->value = malloc(len + 1)) == NULL) {
 225                 free(option);
 226                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 227                 return (ARCHIVE_FATAL);
 228         }
 229         memcpy(option->value, value, len);
 230         option->value[len] = '\0';
 231         option->next = *global;
 232         *global = option;
 233         return (ARCHIVE_OK);
 234 }
 235
 236 static void
 237 remove_option(struct mtree_option **global, const char *value, size_t len)
 238 {
 239         struct mtree_option *iter, *last;
 240
 241         last = NULL;
 242         for (iter = *global; iter != NULL; last = iter, iter = iter->next) {
 243                 if (strncmp(iter->value, value, len) == 0 &&
 244                     (iter->value[len] == '\0' ||
 245                      iter->value[len] == '='))
 246                         break;
 247         }
 248         if (iter == NULL)
 249                 return;
 250         if (last == NULL)
 251                 *global = iter->next;
 252         else
 253                 last->next = iter->next;
 254
 255         free(iter->value);
 256         free(iter);
 257 }
 258
 259 static int
 260 process_global_set(struct archive_read *a,
 261     struct mtree_option **global, const char *line)
 262 {
 263         const char *next, *eq;
 264         size_t len;
 265         int r;
 266
 267         line += 4;
 268         for (;;) {
 269                 next = line + strspn(line, " \t\r\n");
 270                 if (*next == '\0')
 271                         return (ARCHIVE_OK);
 272                 line = next;
 273                 next = line + strcspn(line, " \t\r\n");
 274                 eq = strchr(line, '=');
 275                 if (eq > next)
 276                         len = next - line;
 277                 else
 278                         len = eq - line;
 279
 280                 remove_option(global, line, len);
 281                 r = add_option(a, global, line, next - line);
 282                 if (r != ARCHIVE_OK)
 283                         return (r);
 284                 line = next;
 285         }
 286 }
 287
 288 static int
 289 process_global_unset(struct archive_read *a,
 290     struct mtree_option **global, const char *line)
 291 {
 292         const char *next;
 293         size_t len;
 294
 295         line += 6;
 296         if (strchr(line, '=') != NULL) {
 297                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 298                     "/unset shall not contain `='");
 299                 return ARCHIVE_FATAL;
 300         }
 301
 302         for (;;) {
 303                 next = line + strspn(line, " \t\r\n");
 304                 if (*next == '\0')
 305                         return (ARCHIVE_OK);
 306                 line = next;
 307                 len = strcspn(line, " \t\r\n");
 308
 309                 if (len == 3 && strncmp(line, "all", 3) == 0) {
 310                         free_options(*global);
 311                         *global = NULL;
 312                 } else {
 313                         remove_option(global, line, len);
 314                 }
 315
 316                 line += len;
 317         }
 318 }
 319
 320 static int
 321 process_add_entry(struct archive_read *a, struct mtree *mtree,
 322     struct mtree_option **global, const char *line,
 323     struct mtree_entry **last_entry)
 324 {
 325         struct mtree_entry *entry;
 326         struct mtree_option *iter;
 327         const char *next, *eq;
 328         size_t len;
 329         int r;
 330
 331         if ((entry = malloc(sizeof(*entry))) == NULL) {
 332                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 333                 return (ARCHIVE_FATAL);
 334         }
 335         entry->next = NULL;
 336         entry->options = NULL;
 337         entry->name = NULL;
 338         entry->used = 0;
 339         entry->full = 0;
 340
 341         /* Add this entry to list. */
 342         if (*last_entry == NULL)
 343                 mtree->entries = entry;
 344         else
 345                 (*last_entry)->next = entry;
 346         *last_entry = entry;
 347
 348         len = strcspn(line, " \t\r\n");
 349         if ((entry->name = malloc(len + 1)) == NULL) {
 350                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 351                 return (ARCHIVE_FATAL);
 352         }
 353
 354         memcpy(entry->name, line, len);
 355         entry->name[len] = '\0';
 356         parse_escapes(entry->name, entry);
 357
 358         line += len;
 359         for (iter = *global; iter != NULL; iter = iter->next) {
 360                 r = add_option(a, &entry->options, iter->value,
 361                     strlen(iter->value));
 362                 if (r != ARCHIVE_OK)
 363                         return (r);
 364         }
 365
 366         for (;;) {
 367                 next = line + strspn(line, " \t\r\n");
 368                 if (*next == '\0')
 369                         return (ARCHIVE_OK);
 370                 line = next;
 371                 next = line + strcspn(line, " \t\r\n");
 372                 eq = strchr(line, '=');
 373                 if (eq > next)
 374                         len = next - line;
 375                 else
 376                         len = eq - line;
 377
 378                 remove_option(&entry->options, line, len);
 379                 r = add_option(a, &entry->options, line, next - line);
 380                 if (r != ARCHIVE_OK)
 381                         return (r);
 382                 line = next;
 383         }
 384 }
 385
 386 static int
 387 read_mtree(struct archive_read *a, struct mtree *mtree)
 388 {
 389         ssize_t len;
 390         uintmax_t counter;
 391         char *p;
 392         struct mtree_option *global;
 393         struct mtree_entry *last_entry;
 394         int r;
 395
 396         mtree->archive_format = ARCHIVE_FORMAT_MTREE;
 397         mtree->archive_format_name = "mtree";
 398
 399         global = NULL;
 400         last_entry = NULL;
 401         r = ARCHIVE_OK;
 402
 403         for (counter = 1; ; ++counter) {
 404                 len = readline(a, mtree, &p, 256);
 405                 if (len == 0) {
 406                         mtree->this_entry = mtree->entries;
 407                         free_options(global);
 408                         return (ARCHIVE_OK);
 409                 }
 410                 if (len < 0) {
 411                         free_options(global);
 412                         return (len);
 413                 }
 414                 /* Leading whitespace is never significant, ignore it. */
 415                 while (*p == ' ' || *p == '\t') {
 416                         ++p;
 417                         --len;
 418                 }
 419                 /* Skip content lines and blank lines. */
 420                 if (*p == '#')
 421                         continue;
 422                 if (*p == '\r' || *p == '\n' || *p == '\0')
 423                         continue;
 424                 if (*p != '/') {
 425                         r = process_add_entry(a, mtree, &global, p,
 426                             &last_entry);
 427                 } else if (strncmp(p, "/set", 4) == 0) {
 428                         if (p[4] != ' ' && p[4] != '\t')
 429                                 break;
 430                         r = process_global_set(a, &global, p);
 431                 } else if (strncmp(p, "/unset", 6) == 0) {
 432                         if (p[6] != ' ' && p[6] != '\t')
 433                                 break;
 434                         r = process_global_unset(a, &global, p);
 435                 } else
 436                         break;
 437
 438                 if (r != ARCHIVE_OK) {
 439                         free_options(global);
 440                         return r;
 441                 }
 442         }
 443
 444         archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
 445             "Can't parse line %ju", counter);
 446         free_options(global);
 447         return (ARCHIVE_FATAL);
 448 }
 449
 450 /*
 451  * Read in the entire mtree file into memory on the first request.
 452  * Then use the next unused file to satisfy each header request.
 453  */
 454 static int
 455 read_header(struct archive_read *a, struct archive_entry *entry)
 456 {
 457         struct mtree *mtree;
 458         char *p;
 459         int r, use_next;
 460
 461         mtree = (struct mtree *)(a->format->data);
 462
 463         if (mtree->fd >= 0) {
 464                 close(mtree->fd);
 465                 mtree->fd = -1;
 466         }
 467
 468         if (mtree->entries == NULL) {
 469                 mtree->resolver = archive_entry_linkresolver_new();
 470                 if (mtree->resolver == NULL)
 471                         return ARCHIVE_FATAL;
 472                 archive_entry_linkresolver_set_strategy(mtree->resolver,
 473                     ARCHIVE_FORMAT_MTREE);
 474                 r = read_mtree(a, mtree);
 475                 if (r != ARCHIVE_OK)
 476                         return (r);
 477         }
 478
 479         a->archive.archive_format = mtree->archive_format;
 480         a->archive.archive_format_name = mtree->archive_format_name;
 481
 482         for (;;) {
 483                 if (mtree->this_entry == NULL)
 484                         return (ARCHIVE_EOF);
 485                 if (strcmp(mtree->this_entry->name, "..") == 0) {
 486                         mtree->this_entry->used = 1;
 487                         if (archive_strlen(&mtree->current_dir) > 0) {
 488                                 /* Roll back current path. */
 489                                 p = mtree->current_dir.s
 490                                     + mtree->current_dir.length - 1;
 491                                 while (p >= mtree->current_dir.s && *p != '/')
 492                                         --p;
 493                                 if (p >= mtree->current_dir.s)
 494                                         --p;
 495                                 mtree->current_dir.length
 496                                     = p - mtree->current_dir.s + 1;
 497                         }
 498                 }
 499                 if (!mtree->this_entry->used) {
 500                         use_next = 0;
 501                         r = parse_file(a, entry, mtree, mtree->this_entry, &use_next);
 502                         if (use_next == 0)
 503                                 return (r);
 504                 }
 505                 mtree->this_entry = mtree->this_entry->next;
 506         }
 507 }
 508
 509 /*
 510  * A single file can have multiple lines contribute specifications.
 511  * Parse as many lines as necessary, then pull additional information
 512  * from a backing file on disk as necessary.
 513  */
 514 static int
 515 parse_file(struct archive_read *a, struct archive_entry *entry,
 516     struct mtree *mtree, struct mtree_entry *mentry, int *use_next)
 517 {
 518         const char *path;
 519         struct stat st_storage, *st;
 520         struct mtree_entry *mp;
 521         struct archive_entry *sparse_entry;
 522         int r = ARCHIVE_OK, r1, parsed_kws, mismatched_type;
 523
 524         mentry->used = 1;
 525
 526         /* Initialize reasonable defaults. */
 527         mtree->filetype = AE_IFREG;
 528         archive_entry_set_size(entry, 0);
 529
 530         /* Parse options from this line. */
 531         parsed_kws = 0;
 532         r = parse_line(a, entry, mtree, mentry, &parsed_kws);
 533
 534         if (mentry->full) {
 535                 archive_entry_copy_pathname(entry, mentry->name);
 536                 /*
 537                  * "Full" entries are allowed to have multiple lines
 538                  * and those lines aren't required to be adjacent.  We
 539                  * don't support multiple lines for "relative" entries
 540                  * nor do we make any attempt to merge data from
 541                  * separate "relative" and "full" entries.  (Merging
 542                  * "relative" and "full" entries would require dealing
 543                  * with pathname canonicalization, which is a very
 544                  * tricky subject.)
 545                  */
 546                 for (mp = mentry->next; mp != NULL; mp = mp->next) {
 547                         if (mp->full && !mp->used
 548                             && strcmp(mentry->name, mp->name) == 0) {
 549                                 /* Later lines override earlier ones. */
 550                                 mp->used = 1;
 551                                 r1 = parse_line(a, entry, mtree, mp,
 552                                     &parsed_kws);
 553                                 if (r1 < r)
 554                                         r = r1;
 555                         }
 556                 }
 557         } else {
 558                 /*
 559                  * Relative entries require us to construct
 560                  * the full path and possibly update the
 561                  * current directory.
 562                  */
 563                 size_t n = archive_strlen(&mtree->current_dir);
 564                 if (n > 0)
 565                         archive_strcat(&mtree->current_dir, "/");
 566                 archive_strcat(&mtree->current_dir, mentry->name);
 567                 archive_entry_copy_pathname(entry, mtree->current_dir.s);
 568                 if (archive_entry_filetype(entry) != AE_IFDIR)
 569                         mtree->current_dir.length = n;
 570         }
 571
 572         /*
 573          * Try to open and stat the file to get the real size
 574          * and other file info.  It would be nice to avoid
 575          * this here so that getting a listing of an mtree
 576          * wouldn't require opening every referenced contents
 577          * file.  But then we wouldn't know the actual
 578          * contents size, so I don't see a really viable way
 579          * around this.  (Also, we may want to someday pull
 580          * other unspecified info from the contents file on
 581          * disk.)
 582          */
 583         mtree->fd = -1;
 584         if (archive_strlen(&mtree->contents_name) > 0)
 585                 path = mtree->contents_name.s;
 586         else
 587                 path = archive_entry_pathname(entry);
 588
 589         if (archive_entry_filetype(entry) == AE_IFREG ||
 590             archive_entry_filetype(entry) == AE_IFDIR) {
 591                 mtree->fd = open(path,
 592                     O_RDONLY | O_BINARY);
 593                 if (mtree->fd == -1 &&
 594                     (errno != ENOENT ||
 595                      archive_strlen(&mtree->contents_name) > 0)) {
 596                         archive_set_error(&a->archive, errno,
 597                             "Can't open %s", path);
 598                         r = ARCHIVE_WARN;
 599                 }
 600         }
 601
 602         st = &st_storage;
 603         if (mtree->fd >= 0) {
 604                 if (fstat(mtree->fd, st) == -1) {
 605                         archive_set_error(&a->archive, errno,
 606                             "Could not fstat %s", path);
 607                         r = ARCHIVE_WARN;
 608                         /* If we can't stat it, don't keep it open. */
 609                         close(mtree->fd);
 610                         mtree->fd = -1;
 611                         st = NULL;
 612                 }
 613         } else if (lstat(path, st) == -1) {
 614                 st = NULL;
 615         }
 616
 617         /*
 618          * If there is a contents file on disk, use that size;
 619          * otherwise leave it as-is (it might have been set from
 620          * the mtree size= keyword).
 621          */
 622         if (st != NULL) {
 623                 mismatched_type = 0;
 624                 if ((st->st_mode & S_IFMT) == S_IFREG &&
 625                     archive_entry_filetype(entry) != AE_IFREG)
 626                         mismatched_type = 1;
 627                 if ((st->st_mode & S_IFMT) == S_IFLNK &&
 628                     archive_entry_filetype(entry) != AE_IFLNK)
 629                         mismatched_type = 1;
 630                 if ((st->st_mode & S_IFSOCK) == S_IFSOCK &&
 631                     archive_entry_filetype(entry) != AE_IFSOCK)
 632                         mismatched_type = 1;
 633                 if ((st->st_mode & S_IFMT) == S_IFCHR &&
 634                     archive_entry_filetype(entry) != AE_IFCHR)
 635                         mismatched_type = 1;
 636                 if ((st->st_mode & S_IFMT) == S_IFBLK &&
 637                     archive_entry_filetype(entry) != AE_IFBLK)
 638                         mismatched_type = 1;
 639                 if ((st->st_mode & S_IFMT) == S_IFDIR &&
 640                     archive_entry_filetype(entry) != AE_IFDIR)
 641                         mismatched_type = 1;
 642                 if ((st->st_mode & S_IFMT) == S_IFIFO &&
 643                     archive_entry_filetype(entry) != AE_IFIFO)
 644                         mismatched_type = 1;
 645
 646                 if (mismatched_type) {
 647                         if ((parsed_kws & MTREE_HAS_OPTIONAL) == 0) {
 648                                 archive_set_error(&a->archive,
 649                                     ARCHIVE_ERRNO_MISC,
 650                                     "mtree specification has different type for %s",
 651                                     archive_entry_pathname(entry));
 652                                 r = ARCHIVE_WARN;
 653                         } else {
 654                                 *use_next = 1;
 655                         }
 656                         /* Don't hold a non-regular file open. */
 657                         if (mtree->fd >= 0)
 658                                 close(mtree->fd);
 659                         mtree->fd = -1;
 660                         st = NULL;
 661                         return r;
 662                 }
 663         }
 664
 665         if (st != NULL) {
 666                 if ((parsed_kws & MTREE_HAS_DEVICE) == 0 &&
 667                     (archive_entry_filetype(entry) == AE_IFCHR ||
 668                      archive_entry_filetype(entry) == AE_IFBLK))
 669                         archive_entry_set_rdev(entry, st->st_rdev);
 670                 if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME)) == 0)
 671                         archive_entry_set_gid(entry, st->st_gid);
 672                 if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME)) == 0)
 673                         archive_entry_set_uid(entry, st->st_uid);
 674                 if ((parsed_kws & MTREE_HAS_MTIME) == 0) {
 675 #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC
 676                         archive_entry_set_mtime(entry, st->st_mtime,
 677                             st->st_mtimespec.tv_nsec);
 678 #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC
 679                         archive_entry_set_mtime(entry, st->st_mtime,
 680                             st->st_mtim.tv_nsec);
 681 #elif HAVE_STRUCT_STAT_ST_MTIME_N
 682                         archive_entry_set_mtime(entry, st->st_mtime,
 683                             st->st_mtime_n);
 684 #elif HAVE_STRUCT_STAT_ST_UMTIME
 685                         archive_entry_set_mtime(entry, st->st_mtime,
 686                             st->st_umtime*1000);
 687 #elif HAVE_STRUCT_STAT_ST_MTIME_USEC
 688                         archive_entry_set_mtime(entry, st->st_mtime,
 689                             st->st_mtime_usec*1000);
 690 #else
 691                         archive_entry_set_mtime(entry, st->st_mtime, 0);
 692 #endif
 693                 }
 694                 if ((parsed_kws & MTREE_HAS_NLINK) == 0)
 695                         archive_entry_set_nlink(entry, st->st_nlink);
 696                 if ((parsed_kws & MTREE_HAS_PERM) == 0)
 697                         archive_entry_set_perm(entry, st->st_mode);
 698                 if ((parsed_kws & MTREE_HAS_SIZE) == 0)
 699                         archive_entry_set_size(entry, st->st_size);
 700                 archive_entry_set_ino(entry, st->st_ino);
 701                 archive_entry_set_dev(entry, st->st_dev);
 702
 703                 archive_entry_linkify(mtree->resolver, &entry, &sparse_entry);
 704         } else if (parsed_kws & MTREE_HAS_OPTIONAL) {
 705                 /*
 706                  * Couldn't open the entry, stat it or the on-disk type
 707                  * didn't match.  If this entry is optional, just ignore it
 708                  * and read the next header entry.
 709                  */
 710                 *use_next = 1;
 711                 return ARCHIVE_OK;
 712         }
 713
 714         mtree->cur_size = archive_entry_size(entry);
 715         mtree->offset = 0;
 716
 717         return r;
 718 }
 719
 720 /*
 721  * Each line contains a sequence of keywords.
 722  */
 723 static int
 724 parse_line(struct archive_read *a, struct archive_entry *entry,
 725     struct mtree *mtree, struct mtree_entry *mp, int *parsed_kws)
 726 {
 727         struct mtree_option *iter;
 728         int r = ARCHIVE_OK, r1;
 729
 730         for (iter = mp->options; iter != NULL; iter = iter->next) {
 731                 r1 = parse_keyword(a, mtree, entry, iter, parsed_kws);
 732                 if (r1 < r)
 733                         r = r1;
 734         }
 735         if ((*parsed_kws & MTREE_HAS_TYPE) == 0) {
 736                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
 737                     "Missing type keyword in mtree specification");
 738                 return (ARCHIVE_WARN);
 739         }
 740         return (r);
 741 }
 742
 743 /*
 744  * Device entries have one of the following forms:
 745  * raw dev_t
 746  * format,major,minor[,subdevice]
 747  *
 748  * Just use major and minor, no translation etc is done
 749  * between formats.
 750  */
 751 static int
 752 parse_device(struct archive *a, struct archive_entry *entry, char *val)
 753 {
 754         char *comma1, *comma2;
 755
 756         comma1 = strchr(val, ',');
 757         if (comma1 == NULL) {
 758                 archive_entry_set_dev(entry, mtree_atol10(&val));
 759                 return (ARCHIVE_OK);
 760         }
 761         ++comma1;
 762         comma2 = strchr(comma1, ',');
 763         if (comma2 == NULL) {
 764                 archive_set_error(a, ARCHIVE_ERRNO_FILE_FORMAT,
 765                     "Malformed device attribute");
 766                 return (ARCHIVE_WARN);
 767         }
 768         ++comma2;
 769         archive_entry_set_rdevmajor(entry, mtree_atol(&comma1));
 770         archive_entry_set_rdevminor(entry, mtree_atol(&comma2));
 771         return (ARCHIVE_OK);
 772 }
 773
 774 /*
 775  * Parse a single keyword and its value.
 776  */
 777 static int
 778 parse_keyword(struct archive_read *a, struct mtree *mtree,
 779     struct archive_entry *entry, struct mtree_option *option, int *parsed_kws)
 780 {
 781         char *val, *key;
 782
 783         key = option->value;
 784
 785         if (*key == '\0')
 786                 return (ARCHIVE_OK);
 787
 788         if (strcmp(key, "optional") == 0) {
 789                 *parsed_kws |= MTREE_HAS_OPTIONAL;
 790                 return (ARCHIVE_OK);
 791         }
 792         if (strcmp(key, "ignore") == 0) {
 793                 /*
 794                  * The mtree processing is not recursive, so
 795                  * recursion will only happen for explicitly listed
 796                  * entries.
 797                  */
 798                 return (ARCHIVE_OK);
 799         }
 800
 801         val = strchr(key, '=');
 802         if (val == NULL) {
 803                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
 804                     "Malformed attribute \"%s\" (%d)", key, key[0]);
 805                 return (ARCHIVE_WARN);
 806         }
 807
 808         *val = '\0';
 809         ++val;
 810
 811         switch (key[0]) {
 812         case 'c':
 813                 if (strcmp(key, "content") == 0
 814                     || strcmp(key, "contents") == 0) {
 815                         parse_escapes(val, NULL);
 816                         archive_strcpy(&mtree->contents_name, val);
 817                         break;
 818                 }
 819                 if (strcmp(key, "cksum") == 0)
 820                         break;
 821         case 'd':
 822                 if (strcmp(key, "device") == 0) {
 823                         *parsed_kws |= MTREE_HAS_DEVICE;
 824                         return parse_device(&a->archive, entry, val);
 825                 }
 826         case 'f':
 827                 if (strcmp(key, "flags") == 0) {
 828                         *parsed_kws |= MTREE_HAS_FFLAGS;
 829                         archive_entry_copy_fflags_text(entry, val);
 830                         break;
 831                 }
 832         case 'g':
 833                 if (strcmp(key, "gid") == 0) {
 834                         *parsed_kws |= MTREE_HAS_GID;
 835                         archive_entry_set_gid(entry, mtree_atol10(&val));
 836                         break;
 837                 }
 838                 if (strcmp(key, "gname") == 0) {
 839                         *parsed_kws |= MTREE_HAS_GNAME;
 840                         archive_entry_copy_gname(entry, val);
 841                         break;
 842                 }
 843         case 'l':
 844                 if (strcmp(key, "link") == 0) {
 845                         archive_entry_copy_symlink(entry, val);
 846                         break;
 847                 }
 848         case 'm':
 849                 if (strcmp(key, "md5") == 0 || strcmp(key, "md5digest") == 0)
 850                         break;
 851                 if (strcmp(key, "mode") == 0) {
 852                         if (val[0] >= '0' && val[0] <= '9') {
 853                                 *parsed_kws |= MTREE_HAS_PERM;
 854                                 archive_entry_set_perm(entry,
 855                                     mtree_atol8(&val));
 856                         } else {
 857                                 archive_set_error(&a->archive,
 858                                     ARCHIVE_ERRNO_FILE_FORMAT,
 859                                     "Symbolic mode \"%s\" unsupported", val);
 860                                 return ARCHIVE_WARN;
 861                         }
 862                         break;
 863                 }
 864         case 'n':
 865                 if (strcmp(key, "nlink") == 0) {
 866                         *parsed_kws |= MTREE_HAS_NLINK;
 867                         archive_entry_set_nlink(entry, mtree_atol10(&val));
 868                         break;
 869                 }
 870         case 'r':
 871                 if (strcmp(key, "rmd160") == 0 ||
 872                     strcmp(key, "rmd160digest") == 0)
 873                         break;
 874         case 's':
 875                 if (strcmp(key, "sha1") == 0 || strcmp(key, "sha1digest") == 0)
 876                         break;
 877                 if (strcmp(key, "sha256") == 0 ||
 878                     strcmp(key, "sha256digest") == 0)
 879                         break;
 880                 if (strcmp(key, "sha384") == 0 ||
 881                     strcmp(key, "sha384digest") == 0)
 882                         break;
 883                 if (strcmp(key, "sha512") == 0 ||
 884                     strcmp(key, "sha512digest") == 0)
 885                         break;
 886                 if (strcmp(key, "size") == 0) {
 887                         archive_entry_set_size(entry, mtree_atol10(&val));
 888                         break;
 889                 }
 890         case 't':
 891                 if (strcmp(key, "tags") == 0) {
 892                         /*
 893                          * Comma delimited list of tags.
 894                          * Ignore the tags for now, but the interface
 895                          * should be extended to allow inclusion/exclusion.
 896                          */
 897                         break;
 898                 }
 899                 if (strcmp(key, "time") == 0) {
 900                         time_t m;
 901                         long ns;
 902
 903                         *parsed_kws |= MTREE_HAS_MTIME;
 904                         m = (time_t)mtree_atol10(&val);
 905                         if (*val == '.') {
 906                                 ++val;
 907                                 ns = (long)mtree_atol10(&val);
 908                         } else
 909                                 ns = 0;
 910                         archive_entry_set_mtime(entry, m, ns);
 911                         break;
 912                 }
 913                 if (strcmp(key, "type") == 0) {
 914                         *parsed_kws |= MTREE_HAS_TYPE;
 915                         switch (val[0]) {
 916                         case 'b':
 917                                 if (strcmp(val, "block") == 0) {
 918                                         mtree->filetype = AE_IFBLK;
 919                                         break;
 920                                 }
 921                         case 'c':
 922                                 if (strcmp(val, "char") == 0) {
 923                                         mtree->filetype = AE_IFCHR;
 924                                         break;
 925                                 }
 926                         case 'd':
 927                                 if (strcmp(val, "dir") == 0) {
 928                                         mtree->filetype = AE_IFDIR;
 929                                         break;
 930                                 }
 931                         case 'f':
 932                                 if (strcmp(val, "fifo") == 0) {
 933                                         mtree->filetype = AE_IFIFO;
 934                                         break;
 935                                 }
 936                                 if (strcmp(val, "file") == 0) {
 937                                         mtree->filetype = AE_IFREG;
 938                                         break;
 939                                 }
 940                         case 'l':
 941                                 if (strcmp(val, "link") == 0) {
 942                                         mtree->filetype = AE_IFLNK;
 943                                         break;
 944                                 }
 945                         default:
 946                                 archive_set_error(&a->archive,
 947                                     ARCHIVE_ERRNO_FILE_FORMAT,
 948                                     "Unrecognized file type \"%s\"", val);
 949                                 return (ARCHIVE_WARN);
 950                         }
 951                         archive_entry_set_filetype(entry, mtree->filetype);
 952                         break;
 953                 }
 954         case 'u':
 955                 if (strcmp(key, "uid") == 0) {
 956                         *parsed_kws |= MTREE_HAS_UID;
 957                         archive_entry_set_uid(entry, mtree_atol10(&val));
 958                         break;
 959                 }
 960                 if (strcmp(key, "uname") == 0) {
 961                         *parsed_kws |= MTREE_HAS_UNAME;
 962                         archive_entry_copy_uname(entry, val);
 963                         break;
 964                 }
 965         default:
 966                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
 967                     "Unrecognized key %s=%s", key, val);
 968                 return (ARCHIVE_WARN);
 969         }
 970         return (ARCHIVE_OK);
 971 }
 972
 973 static int
 974 read_data(struct archive_read *a, const void **buff, size_t *size, off_t *offset)
 975 {
 976         size_t bytes_to_read;
 977         ssize_t bytes_read;
 978         struct mtree *mtree;
 979
 980         mtree = (struct mtree *)(a->format->data);
 981         if (mtree->fd < 0) {
 982                 *buff = NULL;
 983                 *offset = 0;
 984                 *size = 0;
 985                 return (ARCHIVE_EOF);
 986         }
 987         if (mtree->buff == NULL) {
 988                 mtree->buffsize = 64 * 1024;
 989                 mtree->buff = malloc(mtree->buffsize);
 990                 if (mtree->buff == NULL) {
 991                         archive_set_error(&a->archive, ENOMEM,
 992                             "Can't allocate memory");
 993                 }
 994                 return (ARCHIVE_FATAL);
 995         }
 996
 997         *buff = mtree->buff;
 998         *offset = mtree->offset;
 999         if ((off_t)mtree->buffsize > mtree->cur_size - mtree->offset)
1000                 bytes_to_read = mtree->cur_size - mtree->offset;
1001         else
1002                 bytes_to_read = mtree->buffsize;
1003         bytes_read = read(mtree->fd, mtree->buff, bytes_to_read);
1004         if (bytes_read < 0) {
1005                 archive_set_error(&a->archive, errno, "Can't read");
1006                 return (ARCHIVE_WARN);
1007         }
1008         if (bytes_read == 0) {
1009                 *size = 0;
1010                 return (ARCHIVE_EOF);
1011         }
1012         mtree->offset += bytes_read;
1013         *size = bytes_read;
1014         return (ARCHIVE_OK);
1015 }
1016
1017 /* Skip does nothing except possibly close the contents file. */
1018 static int
1019 skip(struct archive_read *a)
1020 {
1021         struct mtree *mtree;
1022
1023         mtree = (struct mtree *)(a->format->data);
1024         if (mtree->fd >= 0) {
1025                 close(mtree->fd);
1026                 mtree->fd = -1;
1027         }
1028         return (ARCHIVE_OK);
1029 }
1030
1031 /*
1032  * Since parsing backslash sequences always makes strings shorter,
1033  * we can always do this conversion in-place.
1034  */
1035 static void
1036 parse_escapes(char *src, struct mtree_entry *mentry)
1037 {
1038         char *dest = src;
1039         char c;
1040
1041         /*
1042          * The current directory is somewhat special, it should be archived
1043          * only once as it will confuse extraction otherwise.
1044          */
1045         if (strcmp(src, ".") == 0)
1046                 mentry->full = 1;
1047
1048         while (*src != '\0') {
1049                 c = *src++;
1050                 if (c == '/' && mentry != NULL)
1051                         mentry->full = 1;
1052                 if (c == '\\') {
1053                         switch (src[0]) {
1054                         case '0':
1055                                 if (src[1] < '0' || src[1] > '7') {
1056                                         c = 0;
1057                                         ++src;
1058                                         break;
1059                                 }
1060                                 /* FALLTHROUGH */
1061                         case '1':
1062                         case '2':
1063                         case '3':
1064                                 if (src[1] >= '0' && src[1] <= '7' &&
1065                                     src[2] >= '0' && src[2] <= '7') {
1066                                         c = (src[0] - '0') << 6;
1067                                         c |= (src[1] - '0') << 3;
1068                                         c |= (src[2] - '0');
1069                                         src += 3;
1070                                 }
1071                                 break;
1072                         case 'a':
1073                                 c = '\a';
1074                                 ++src;
1075                                 break;
1076                         case 'b':
1077                                 c = '\b';
1078                                 ++src;
1079                                 break;
1080                         case 'f':
1081                                 c = '\f';
1082                                 ++src;
1083                                 break;
1084                         case 'n':
1085                                 c = '\n';
1086                                 ++src;
1087                                 break;
1088                         case 'r':
1089                                 c = '\r';
1090                                 ++src;
1091                                 break;
1092                         case 's':
1093                                 c = ' ';
1094                                 ++src;
1095                                 break;
1096                         case 't':
1097                                 c = '\t';
1098                                 ++src;
1099                                 break;
1100                         case 'v':
1101                                 c = '\v';
1102                                 ++src;
1103                                 break;
1104                         }
1105                 }
1106                 *dest++ = c;
1107         }
1108         *dest = '\0';
1109 }
1110
1111 /*
1112  * Note that this implementation does not (and should not!) obey
1113  * locale settings; you cannot simply substitute strtol here, since
1114  * it does obey locale.
1115  */
1116 static int64_t
1117 mtree_atol8(char **p)
1118 {
1119         int64_t l, limit, last_digit_limit;
1120         int digit, base;
1121
1122         base = 8;
1123         limit = INT64_MAX / base;
1124         last_digit_limit = INT64_MAX % base;
1125
1126         l = 0;
1127         digit = **p - '0';
1128         while (digit >= 0 && digit < base) {
1129                 if (l>limit || (l == limit && digit > last_digit_limit)) {
1130                         l = INT64_MAX; /* Truncate on overflow. */
1131                         break;
1132                 }
1133                 l = (l * base) + digit;
1134                 digit = *++(*p) - '0';
1135         }
1136         return (l);
1137 }
1138
1139 /*
1140  * Note that this implementation does not (and should not!) obey
1141  * locale settings; you cannot simply substitute strtol here, since
1142  * it does obey locale.
1143  */
1144 static int64_t
1145 mtree_atol10(char **p)
1146 {
1147         int64_t l, limit, last_digit_limit;
1148         int base, digit, sign;
1149
1150         base = 10;
1151         limit = INT64_MAX / base;
1152         last_digit_limit = INT64_MAX % base;
1153
1154         if (**p == '-') {
1155                 sign = -1;
1156                 ++(*p);
1157         } else
1158                 sign = 1;
1159
1160         l = 0;
1161         digit = **p - '0';
1162         while (digit >= 0 && digit < base) {
1163                 if (l > limit || (l == limit && digit > last_digit_limit)) {
1164                         l = UINT64_MAX; /* Truncate on overflow. */
1165                         break;
1166                 }
1167                 l = (l * base) + digit;
1168                 digit = *++(*p) - '0';
1169         }
1170         return (sign < 0) ? -l : l;
1171 }
1172
1173 /*
1174  * Note that this implementation does not (and should not!) obey
1175  * locale settings; you cannot simply substitute strtol here, since
1176  * it does obey locale.
1177  */
1178 static int64_t
1179 mtree_atol16(char **p)
1180 {
1181         int64_t l, limit, last_digit_limit;
1182         int base, digit, sign;
1183
1184         base = 16;
1185         limit = INT64_MAX / base;
1186         last_digit_limit = INT64_MAX % base;
1187
1188         if (**p == '-') {
1189                 sign = -1;
1190                 ++(*p);
1191         } else
1192                 sign = 1;
1193
1194         l = 0;
1195         if (**p >= '0' && **p <= '9')
1196                 digit = **p - '0';
1197         else if (**p >= 'a' && **p <= 'f')
1198                 digit = **p - 'a' + 10;
1199         else if (**p >= 'A' && **p <= 'F')
1200                 digit = **p - 'A' + 10;
1201         else
1202                 digit = -1;
1203         while (digit >= 0 && digit < base) {
1204                 if (l > limit || (l == limit && digit > last_digit_limit)) {
1205                         l = UINT64_MAX; /* Truncate on overflow. */
1206                         break;
1207                 }
1208                 l = (l * base) + digit;
1209                 if (**p >= '0' && **p <= '9')
1210                         digit = **p - '0';
1211                 else if (**p >= 'a' && **p <= 'f')
1212                         digit = **p - 'a' + 10;
1213                 else if (**p >= 'A' && **p <= 'F')
1214                         digit = **p - 'A' + 10;
1215                 else
1216                         digit = -1;
1217         }
1218         return (sign < 0) ? -l : l;
1219 }
1220
1221 static int64_t
1222 mtree_atol(char **p)
1223 {
1224         if (**p != '0')
1225                 return mtree_atol10(p);
1226         if ((*p)[1] == 'x' || (*p)[1] == 'X') {
1227                 *p += 2;
1228                 return mtree_atol16(p);
1229         }
1230         return mtree_atol8(p);
1231 }
1232
1233 /*
1234  * Returns length of line (including trailing newline)
1235  * or negative on error.  'start' argument is updated to
1236  * point to first character of line.
1237  */
1238 static ssize_t
1239 readline(struct archive_read *a, struct mtree *mtree, char **start, ssize_t limit)
1240 {
1241         ssize_t bytes_read;
1242         ssize_t total_size = 0;
1243         ssize_t find_off = 0;
1244         const void *t;
1245         const char *s;
1246         void *p;
1247         char *u;
1248
1249         /* Accumulate line in a line buffer. */
1250         for (;;) {
1251                 /* Read some more. */
1252                 t = __archive_read_ahead(a, 1, &bytes_read);
1253                 if (t == NULL)
1254                         return (0);
1255                 if (bytes_read < 0)
1256                         return (ARCHIVE_FATAL);
1257                 s = t;  /* Start of line? */
1258                 p = memchr(t, '\n', bytes_read);
1259                 /* If we found '\n', trim the read. */
1260                 if (p != NULL) {
1261                         bytes_read = 1 + ((const char *)p) - s;
1262                 }
1263                 if (total_size + bytes_read + 1 > limit) {
1264                         archive_set_error(&a->archive,
1265                             ARCHIVE_ERRNO_FILE_FORMAT,
1266                             "Line too long");
1267                         return (ARCHIVE_FATAL);
1268                 }
1269                 if (archive_string_ensure(&mtree->line,
1270                         total_size + bytes_read + 1) == NULL) {
1271                         archive_set_error(&a->archive, ENOMEM,
1272                             "Can't allocate working buffer");
1273                         return (ARCHIVE_FATAL);
1274                 }
1275                 memcpy(mtree->line.s + total_size, t, bytes_read);
1276                 __archive_read_consume(a, bytes_read);
1277                 total_size += bytes_read;
1278                 /* Null terminate. */
1279                 mtree->line.s[total_size] = '\0';
1280                 /* If we found an unescaped '\n', clean up and return. */
1281                 for (u = mtree->line.s + find_off; *u; ++u) {
1282                         if (u[0] == '\n') {
1283                                 *start = mtree->line.s;
1284                                 return total_size;
1285                         }
1286                         if (u[0] == '#') {
1287                                 if (p == NULL)
1288                                         break;
1289                                 *start = mtree->line.s;
1290                                 return total_size;
1291                         }
1292                         if (u[0] != '\\')
1293                                 continue;
1294                         if (u[1] == '\\') {
1295                                 ++u;
1296                                 continue;
1297                         }
1298                         if (u[1] == '\n') {
1299                                 memmove(u, u + 1,
1300                                     total_size - (u - mtree->line.s) + 1);
1301                                 --total_size;
1302                                 ++u;
1303                                 break;
1304                         }
1305                         if (u[1] == '\0')
1306                                 break;
1307                 }
1308                 find_off = u - mtree->line.s;
1309         }
1310 }