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