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 1.9 2008/06/21 19:06:37 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,
 152             mtree_bid, read_header, read_data, skip, cleanup);
 153
 154         if (r != ARCHIVE_OK)
 155                 free(mtree);
 156         return (ARCHIVE_OK);
 157 }
 158
 159 static int
 160 cleanup(struct archive_read *a)
 161 {
 162         struct mtree *mtree;
 163         struct mtree_entry *p, *q;
 164
 165         mtree = (struct mtree *)(a->format->data);
 166
 167         p = mtree->entries;
 168         while (p != NULL) {
 169                 q = p->next;
 170                 free(p->name);
 171                 free_options(p->options);
 172                 free(p);
 173                 p = q;
 174         }
 175         archive_string_free(&mtree->line);
 176         archive_string_free(&mtree->current_dir);
 177         archive_string_free(&mtree->contents_name);
 178         archive_entry_linkresolver_free(mtree->resolver);
 179
 180         free(mtree->buff);
 181         free(mtree);
 182         (a->format->data) = NULL;
 183         return (ARCHIVE_OK);
 184 }
 185
 186
 187 static int
 188 mtree_bid(struct archive_read *a)
 189 {
 190         struct mtree *mtree;
 191         ssize_t bytes_read;
 192         const void *h;
 193         const char *signature = "#mtree";
 194         const char *p;
 195         int bid;
 196
 197         mtree = (struct mtree *)(a->format->data);
 198
 199         /* Now let's look at the actual header and see if it matches. */
 200         bytes_read = (a->decompressor->read_ahead)(a, &h, strlen(signature));
 201
 202         if (bytes_read <= 0)
 203                 return (bytes_read);
 204
 205         p = h;
 206         bid = 0;
 207         while (bytes_read > 0 && *signature != '\0') {
 208                 if (*p != *signature)
 209                         return (bid = 0);
 210                 bid += 8;
 211                 p++;
 212                 signature++;
 213                 bytes_read--;
 214         }
 215         return (bid);
 216 }
 217
 218 /*
 219  * The extended mtree format permits multiple lines specifying
 220  * attributes for each file.  For those entries, only the last line
 221  * is actually used.  Practically speaking, that means we have
 222  * to read the entire mtree file into memory up front.
 223  *
 224  * The parsing is done in two steps.  First, it is decided if a line
 225  * changes the global defaults and if it is, processed accordingly.
 226  * Otherwise, the options of the line are merged with the current
 227  * global options.
 228  */
 229 static int
 230 add_option(struct archive_read *a, struct mtree_option **global,
 231     const char *value, size_t len)
 232 {
 233         struct mtree_option *option;
 234
 235         if ((option = malloc(sizeof(*option))) == NULL) {
 236                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 237                 return (ARCHIVE_FATAL);
 238         }
 239         if ((option->value = malloc(len + 1)) == NULL) {
 240                 free(option);
 241                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 242                 return (ARCHIVE_FATAL);
 243         }
 244         memcpy(option->value, value, len);
 245         option->value[len] = '\0';
 246         option->next = *global;
 247         *global = option;
 248         return (ARCHIVE_OK);
 249 }
 250
 251 static void
 252 remove_option(struct mtree_option **global, const char *value, size_t len)
 253 {
 254         struct mtree_option *iter, *last;
 255
 256         last = NULL;
 257         for (iter = *global; iter != NULL; last = iter, iter = iter->next) {
 258                 if (strncmp(iter->value, value, len) == 0 &&
 259                     (iter->value[len] == '\0' ||
 260                      iter->value[len] == '='))
 261                         break;
 262         }
 263         if (iter == NULL)
 264                 return;
 265         if (last == NULL)
 266                 *global = iter->next;
 267         else
 268                 last->next = iter->next;
 269
 270         free(iter->value);
 271         free(iter);
 272 }
 273
 274 static int
 275 process_global_set(struct archive_read *a,
 276     struct mtree_option **global, const char *line)
 277 {
 278         const char *next, *eq;
 279         size_t len;
 280         int r;
 281
 282         line += 4;
 283         for (;;) {
 284                 next = line + strspn(line, " \t\r\n");
 285                 if (*next == '\0')
 286                         return (ARCHIVE_OK);
 287                 line = next;
 288                 next = line + strcspn(line, " \t\r\n");
 289                 eq = strchr(line, '=');
 290                 if (eq > next)
 291                         len = next - line;
 292                 else
 293                         len = eq - line;
 294
 295                 remove_option(global, line, len);
 296                 r = add_option(a, global, line, next - line);
 297                 if (r != ARCHIVE_OK)
 298                         return (r);
 299                 line = next;
 300         }
 301 }
 302
 303 static int
 304 process_global_unset(struct archive_read *a,
 305     struct mtree_option **global, const char *line)
 306 {
 307         const char *next;
 308         size_t len;
 309
 310         line += 6;
 311         if (strchr(line, '=') != NULL) {
 312                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 313                     "/unset shall not contain `='");
 314                 return ARCHIVE_FATAL;
 315         }
 316
 317         for (;;) {
 318                 next = line + strspn(line, " \t\r\n");
 319                 if (*next == '\0')
 320                         return (ARCHIVE_OK);
 321                 line = next;
 322                 len = strcspn(line, " \t\r\n");
 323
 324                 if (len == 3 && strncmp(line, "all", 3) == 0) {
 325                         free_options(*global);
 326                         *global = NULL;
 327                 } else {
 328                         remove_option(global, line, len);
 329                 }
 330
 331                 line += len;
 332         }
 333 }
 334
 335 static int
 336 process_add_entry(struct archive_read *a, struct mtree *mtree,
 337     struct mtree_option **global, const char *line,
 338     struct mtree_entry **last_entry)
 339 {
 340         struct mtree_entry *entry;
 341         struct mtree_option *iter;
 342         const char *next, *eq;
 343         size_t len;
 344         int r;
 345
 346         if ((entry = malloc(sizeof(*entry))) == NULL) {
 347                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 348                 return (ARCHIVE_FATAL);
 349         }
 350         entry->next = NULL;
 351         entry->options = NULL;
 352         entry->name = NULL;
 353         entry->used = 0;
 354         entry->full = 0;
 355
 356         /* Add this entry to list. */
 357         if (*last_entry == NULL)
 358                 mtree->entries = entry;
 359         else
 360                 (*last_entry)->next = entry;
 361         *last_entry = entry;
 362
 363         len = strcspn(line, " \t\r\n");
 364         if ((entry->name = malloc(len + 1)) == NULL) {
 365                 archive_set_error(&a->archive, errno, "Can't allocate memory");
 366                 return (ARCHIVE_FATAL);
 367         }
 368
 369         memcpy(entry->name, line, len);
 370         entry->name[len] = '\0';
 371         parse_escapes(entry->name, entry);
 372
 373         line += len;
 374         for (iter = *global; iter != NULL; iter = iter->next) {
 375                 r = add_option(a, &entry->options, iter->value,
 376                     strlen(iter->value));
 377                 if (r != ARCHIVE_OK)
 378                         return (r);
 379         }
 380
 381         for (;;) {
 382                 next = line + strspn(line, " \t\r\n");
 383                 if (*next == '\0')
 384                         return (ARCHIVE_OK);
 385                 line = next;
 386                 next = line + strcspn(line, " \t\r\n");
 387                 eq = strchr(line, '=');
 388                 if (eq > next)
 389                         len = next - line;
 390                 else
 391                         len = eq - line;
 392
 393                 remove_option(&entry->options, line, len);
 394                 r = add_option(a, &entry->options, line, next - line);
 395                 if (r != ARCHIVE_OK)
 396                         return (r);
 397                 line = next;
 398         }
 399 }
 400
 401 static int
 402 read_mtree(struct archive_read *a, struct mtree *mtree)
 403 {
 404         ssize_t len;
 405         uintmax_t counter;
 406         char *p;
 407         struct mtree_option *global;
 408         struct mtree_entry *last_entry;
 409         int r;
 410
 411         mtree->archive_format = ARCHIVE_FORMAT_MTREE_V1;
 412         mtree->archive_format_name = "mtree";
 413
 414         global = NULL;
 415         last_entry = NULL;
 416         r = ARCHIVE_OK;
 417
 418         for (counter = 1; ; ++counter) {
 419                 len = readline(a, mtree, &p, 256);
 420                 if (len == 0) {
 421                         mtree->this_entry = mtree->entries;
 422                         return (ARCHIVE_OK);
 423                 }
 424                 if (len < 0)
 425                         return (len);
 426                 /* Leading whitespace is never significant, ignore it. */
 427                 while (*p == ' ' || *p == '\t') {
 428                         ++p;
 429                         --len;
 430                 }
 431                 /* Skip content lines and blank lines. */
 432                 if (*p == '#')
 433                         continue;
 434                 if (*p == '\r' || *p == '\n' || *p == '\0')
 435                         continue;
 436                 if (*p != '/') {
 437                         r = process_add_entry(a, mtree, &global, p,
 438                             &last_entry);
 439                 } else if (strncmp(p, "/set", 4) == 0) {
 440                         if (p[4] != ' ' && p[4] != '\t')
 441                                 break;
 442                         r = process_global_set(a, &global, p);
 443                 } else if (strncmp(p, "/unset", 6) == 0) {
 444                         if (p[6] != ' ' && p[6] != '\t')
 445                                 break;
 446                         r = process_global_unset(a, &global, p);
 447                 } else
 448                         break;
 449
 450                 if (r != ARCHIVE_OK)
 451                         return r;
 452         }
 453
 454         archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
 455             "Can't parse line %ju", counter);
 456         return ARCHIVE_FATAL;
 457 }
 458
 459 /*
 460  * Read in the entire mtree file into memory on the first request.
 461  * Then use the next unused file to satisfy each header request.
 462  */
 463 static int
 464 read_header(struct archive_read *a, struct archive_entry *entry)
 465 {
 466         struct mtree *mtree;
 467         char *p;
 468         int r, use_next;
 469
 470         mtree = (struct mtree *)(a->format->data);
 471
 472         if (mtree->fd >= 0) {
 473                 close(mtree->fd);
 474                 mtree->fd = -1;
 475         }
 476
 477         if (mtree->entries == NULL) {
 478                 mtree->resolver = archive_entry_linkresolver_new();
 479                 if (mtree->resolver == NULL)
 480                         return ARCHIVE_FATAL;
 481                 archive_entry_linkresolver_set_strategy(mtree->resolver,
 482                     ARCHIVE_FORMAT_MTREE);
 483                 r = read_mtree(a, mtree);
 484                 if (r != ARCHIVE_OK)
 485                         return (r);
 486         }
 487
 488         a->archive.archive_format = mtree->archive_format;
 489         a->archive.archive_format_name = mtree->archive_format_name;
 490
 491         for (;;) {
 492                 if (mtree->this_entry == NULL)
 493                         return (ARCHIVE_EOF);
 494                 if (strcmp(mtree->this_entry->name, "..") == 0) {
 495                         mtree->this_entry->used = 1;
 496                         if (archive_strlen(&mtree->current_dir) > 0) {
 497                                 /* Roll back current path. */
 498                                 p = mtree->current_dir.s
 499                                     + mtree->current_dir.length - 1;
 500                                 while (p >= mtree->current_dir.s && *p != '/')
 501                                         --p;
 502                                 if (p >= mtree->current_dir.s)
 503                                         --p;
 504                                 mtree->current_dir.length
 505                                     = p - mtree->current_dir.s + 1;
 506                         }
 507                 }
 508                 if (!mtree->this_entry->used) {
 509                         use_next = 0;
 510                         r = parse_file(a, entry, mtree, mtree->this_entry, &use_next);
 511                         if (use_next == 0)
 512                                 return (r);
 513                 }
 514                 mtree->this_entry = mtree->this_entry->next;
 515         }
 516 }
 517
 518 /*
 519  * A single file can have multiple lines contribute specifications.
 520  * Parse as many lines as necessary, then pull additional information
 521  * from a backing file on disk as necessary.
 522  */
 523 static int
 524 parse_file(struct archive_read *a, struct archive_entry *entry,
 525     struct mtree *mtree, struct mtree_entry *mentry, int *use_next)
 526 {
 527         const char *path;
 528         struct stat st_storage, *st;
 529         struct mtree_entry *mp;
 530         struct archive_entry *sparse_entry;
 531         int r = ARCHIVE_OK, r1, parsed_kws, mismatched_type;
 532
 533         mentry->used = 1;
 534
 535         /* Initialize reasonable defaults. */
 536         mtree->filetype = AE_IFREG;
 537         archive_entry_set_size(entry, 0);
 538
 539         /* Parse options from this line. */
 540         parsed_kws = 0;
 541         r = parse_line(a, entry, mtree, mentry, &parsed_kws);
 542
 543         if (mentry->full) {
 544                 archive_entry_copy_pathname(entry, mentry->name);
 545                 /*
 546                  * "Full" entries are allowed to have multiple lines
 547                  * and those lines aren't required to be adjacent.  We
 548                  * don't support multiple lines for "relative" entries
 549                  * nor do we make any attempt to merge data from
 550                  * separate "relative" and "full" entries.  (Merging
 551                  * "relative" and "full" entries would require dealing
 552                  * with pathname canonicalization, which is a very
 553                  * tricky subject.)
 554                  */
 555                 for (mp = mentry->next; mp != NULL; mp = mp->next) {
 556                         if (mp->full && !mp->used
 557                             && strcmp(mentry->name, mp->name) == 0) {
 558                                 /* Later lines override earlier ones. */
 559                                 mp->used = 1;
 560                                 r1 = parse_line(a, entry, mtree, mp,
 561                                     &parsed_kws);
 562                                 if (r1 < r)
 563                                         r = r1;
 564                         }
 565                 }
 566         } else {
 567                 /*
 568                  * Relative entries require us to construct
 569                  * the full path and possibly update the
 570                  * current directory.
 571                  */
 572                 size_t n = archive_strlen(&mtree->current_dir);
 573                 if (n > 0)
 574                         archive_strcat(&mtree->current_dir, "/");
 575                 archive_strcat(&mtree->current_dir, mentry->name);
 576                 archive_entry_copy_pathname(entry, mtree->current_dir.s);
 577                 if (archive_entry_filetype(entry) != AE_IFDIR)
 578                         mtree->current_dir.length = n;
 579         }
 580
 581         /*
 582          * Try to open and stat the file to get the real size
 583          * and other file info.  It would be nice to avoid
 584          * this here so that getting a listing of an mtree
 585          * wouldn't require opening every referenced contents
 586          * file.  But then we wouldn't know the actual
 587          * contents size, so I don't see a really viable way
 588          * around this.  (Also, we may want to someday pull
 589          * other unspecified info from the contents file on
 590          * disk.)
 591          */
 592         mtree->fd = -1;
 593         if (archive_strlen(&mtree->contents_name) > 0)
 594                 path = mtree->contents_name.s;
 595         else
 596                 path = archive_entry_pathname(entry);
 597
 598         if (archive_entry_filetype(entry) == AE_IFREG ||
 599             archive_entry_filetype(entry) == AE_IFDIR) {
 600                 mtree->fd = open(path,
 601                     O_RDONLY | O_BINARY);
 602                 if (mtree->fd == -1 &&
 603                     (errno != ENOENT ||
 604                      archive_strlen(&mtree->contents_name) > 0)) {
 605                         archive_set_error(&a->archive, errno,
 606                             "Can't open %s", path);
 607                         r = ARCHIVE_WARN;
 608                 }
 609         }
 610
 611         st = &st_storage;
 612         if (mtree->fd >= 0) {
 613                 if (fstat(mtree->fd, st) == -1) {
 614                         archive_set_error(&a->archive, errno,
 615                             "Could not fstat %s", path);
 616                         r = ARCHIVE_WARN;
 617                         /* If we can't stat it, don't keep it open. */
 618                         close(mtree->fd);
 619                         mtree->fd = -1;
 620                         st = NULL;
 621                 }
 622         } else if (lstat(path, st) == -1) {
 623                 st = NULL;
 624         }
 625
 626         /*
 627          * If there is a contents file on disk, use that size;
 628          * otherwise leave it as-is (it might have been set from
 629          * the mtree size= keyword).
 630          */
 631         if (st != NULL) {
 632                 mismatched_type = 0;
 633                 if ((st->st_mode & S_IFMT) == S_IFREG &&
 634                     archive_entry_filetype(entry) != AE_IFREG)
 635                         mismatched_type = 1;
 636                 if ((st->st_mode & S_IFMT) == S_IFLNK &&
 637                     archive_entry_filetype(entry) != AE_IFLNK)
 638                         mismatched_type = 1;
 639                 if ((st->st_mode & S_IFSOCK) == S_IFSOCK &&
 640                     archive_entry_filetype(entry) != AE_IFSOCK)
 641                         mismatched_type = 1;
 642                 if ((st->st_mode & S_IFMT) == S_IFCHR &&
 643                     archive_entry_filetype(entry) != AE_IFCHR)
 644                         mismatched_type = 1;
 645                 if ((st->st_mode & S_IFMT) == S_IFBLK &&
 646                     archive_entry_filetype(entry) != AE_IFBLK)
 647                         mismatched_type = 1;
 648                 if ((st->st_mode & S_IFMT) == S_IFDIR &&
 649                     archive_entry_filetype(entry) != AE_IFDIR)
 650                         mismatched_type = 1;
 651                 if ((st->st_mode & S_IFMT) == S_IFIFO &&
 652                     archive_entry_filetype(entry) != AE_IFIFO)
 653                         mismatched_type = 1;
 654
 655                 if (mismatched_type) {
 656                         if ((parsed_kws & MTREE_HAS_OPTIONAL) == 0) {
 657                                 archive_set_error(&a->archive,
 658                                     ARCHIVE_ERRNO_MISC,
 659                                     "mtree specification has different type for %s",
 660                                     archive_entry_pathname(entry));
 661                                 r = ARCHIVE_WARN;
 662                         } else {
 663                                 *use_next = 1;
 664                         }
 665                         /* Don't hold a non-regular file open. */
 666                         if (mtree->fd >= 0)
 667                                 close(mtree->fd);
 668                         mtree->fd = -1;
 669                         st = NULL;
 670                         return r;
 671                 }
 672         }
 673
 674         if (st != NULL) {
 675                 if ((parsed_kws & MTREE_HAS_DEVICE) == 0 &&
 676                     (archive_entry_filetype(entry) == AE_IFCHR ||
 677                      archive_entry_filetype(entry) == AE_IFBLK))
 678                         archive_entry_set_rdev(entry, st->st_rdev);
 679                 if ((parsed_kws & (MTREE_HAS_GID | MTREE_HAS_GNAME)) == 0)
 680                         archive_entry_set_gid(entry, st->st_gid);
 681                 if ((parsed_kws & (MTREE_HAS_UID | MTREE_HAS_UNAME)) == 0)
 682                         archive_entry_set_uid(entry, st->st_uid);
 683                 if ((parsed_kws & MTREE_HAS_MTIME) == 0) {
 684 #if HAVE_STRUCT_STAT_ST_MTIMESPEC_TV_NSEC
 685                         archive_entry_set_mtime(entry, st->st_mtime,
 686                             st->st_mtimespec.tv_nsec);
 687 #elif HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC
 688                         archive_entry_set_mtime(entry, st->st_mtime,
 689                             st->st_mtim.tv_nsec);
 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                         *parsed_kws |= MTREE_HAS_MTIME;
 901                         archive_entry_set_mtime(entry, mtree_atol10(&val), 0);
 902                         break;
 903                 }
 904                 if (strcmp(key, "type") == 0) {
 905                         *parsed_kws |= MTREE_HAS_TYPE;
 906                         switch (val[0]) {
 907                         case 'b':
 908                                 if (strcmp(val, "block") == 0) {
 909                                         mtree->filetype = AE_IFBLK;
 910                                         break;
 911                                 }
 912                         case 'c':
 913                                 if (strcmp(val, "char") == 0) {
 914                                         mtree->filetype = AE_IFCHR;
 915                                         break;
 916                                 }
 917                         case 'd':
 918                                 if (strcmp(val, "dir") == 0) {
 919                                         mtree->filetype = AE_IFDIR;
 920                                         break;
 921                                 }
 922                         case 'f':
 923                                 if (strcmp(val, "fifo") == 0) {
 924                                         mtree->filetype = AE_IFIFO;
 925                                         break;
 926                                 }
 927                                 if (strcmp(val, "file") == 0) {
 928                                         mtree->filetype = AE_IFREG;
 929                                         break;
 930                                 }
 931                         case 'l':
 932                                 if (strcmp(val, "link") == 0) {
 933                                         mtree->filetype = AE_IFLNK;
 934                                         break;
 935                                 }
 936                         default:
 937                                 archive_set_error(&a->archive,
 938                                     ARCHIVE_ERRNO_FILE_FORMAT,
 939                                     "Unrecognized file type \"%s\"", val);
 940                                 return (ARCHIVE_WARN);
 941                         }
 942                         archive_entry_set_filetype(entry, mtree->filetype);
 943                         break;
 944                 }
 945         case 'u':
 946                 if (strcmp(key, "uid") == 0) {
 947                         *parsed_kws |= MTREE_HAS_UID;
 948                         archive_entry_set_uid(entry, mtree_atol10(&val));
 949                         break;
 950                 }
 951                 if (strcmp(key, "uname") == 0) {
 952                         *parsed_kws |= MTREE_HAS_UNAME;
 953                         archive_entry_copy_uname(entry, val);
 954                         break;
 955                 }
 956         default:
 957                 archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT,
 958                     "Unrecognized key %s=%s", key, val);
 959                 return (ARCHIVE_WARN);
 960         }
 961         return (ARCHIVE_OK);
 962 }
 963
 964 static int
 965 read_data(struct archive_read *a, const void **buff, size_t *size, off_t *offset)
 966 {
 967         size_t bytes_to_read;
 968         ssize_t bytes_read;
 969         struct mtree *mtree;
 970
 971         mtree = (struct mtree *)(a->format->data);
 972         if (mtree->fd < 0) {
 973                 *buff = NULL;
 974                 *offset = 0;
 975                 *size = 0;
 976                 return (ARCHIVE_EOF);
 977         }
 978         if (mtree->buff == NULL) {
 979                 mtree->buffsize = 64 * 1024;
 980                 mtree->buff = malloc(mtree->buffsize);
 981                 if (mtree->buff == NULL) {
 982                         archive_set_error(&a->archive, ENOMEM,
 983                             "Can't allocate memory");
 984                 }
 985                 return (ARCHIVE_FATAL);
 986         }
 987
 988         *buff = mtree->buff;
 989         *offset = mtree->offset;
 990         if ((off_t)mtree->buffsize > mtree->cur_size - mtree->offset)
 991                 bytes_to_read = mtree->cur_size - mtree->offset;
 992         else
 993                 bytes_to_read = mtree->buffsize;
 994         bytes_read = read(mtree->fd, mtree->buff, bytes_to_read);
 995         if (bytes_read < 0) {
 996                 archive_set_error(&a->archive, errno, "Can't read");
 997                 return (ARCHIVE_WARN);
 998         }
 999         if (bytes_read == 0) {
1000                 *size = 0;
1001                 return (ARCHIVE_EOF);
1002         }
1003         mtree->offset += bytes_read;
1004         *size = bytes_read;
1005         return (ARCHIVE_OK);
1006 }
1007
1008 /* Skip does nothing except possibly close the contents file. */
1009 static int
1010 skip(struct archive_read *a)
1011 {
1012         struct mtree *mtree;
1013
1014         mtree = (struct mtree *)(a->format->data);
1015         if (mtree->fd >= 0) {
1016                 close(mtree->fd);
1017                 mtree->fd = -1;
1018         }
1019         return (ARCHIVE_OK);
1020 }
1021
1022 /*
1023  * Since parsing backslash sequences always makes strings shorter,
1024  * we can always do this conversion in-place.
1025  */
1026 static void
1027 parse_escapes(char *src, struct mtree_entry *mentry)
1028 {
1029         char *dest = src;
1030         char c;
1031
1032         /*
1033          * The current directory is somewhat special, it should be archived
1034          * only once as it will confuse extraction otherwise.
1035          */
1036         if (strcmp(src, ".") == 0)
1037                 mentry->full = 1;
1038
1039         while (*src != '\0') {
1040                 c = *src++;
1041                 if (c == '/' && mentry != NULL)
1042                         mentry->full = 1;
1043                 if (c == '\\') {
1044                         switch (src[0]) {
1045                         case '0':
1046                                 if (src[1] < '0' || src[1] > '7') {
1047                                         c = 0;
1048                                         ++src;
1049                                         break;
1050                                 }
1051                                 /* FALLTHROUGH */
1052                         case '1':
1053                         case '2':
1054                         case '3':
1055                                 if (src[1] >= '0' && src[1] <= '7' &&
1056                                     src[2] >= '0' && src[2] <= '7') {
1057                                         c = (src[0] - '0') << 6;
1058                                         c |= (src[1] - '0') << 3;
1059                                         c |= (src[2] - '0');
1060                                         src += 3;
1061                                 }
1062                                 break;
1063                         case 'a':
1064                                 c = '\a';
1065                                 ++src;
1066                                 break;
1067                         case 'b':
1068                                 c = '\b';
1069                                 ++src;
1070                                 break;
1071                         case 'f':
1072                                 c = '\f';
1073                                 ++src;
1074                                 break;
1075                         case 'n':
1076                                 c = '\n';
1077                                 ++src;
1078                                 break;
1079                         case 'r':
1080                                 c = '\r';
1081                                 ++src;
1082                                 break;
1083                         case 's':
1084                                 c = ' ';
1085                                 ++src;
1086                                 break;
1087                         case 't':
1088                                 c = '\t';
1089                                 ++src;
1090                                 break;
1091                         case 'v':
1092                                 c = '\v';
1093                                 ++src;
1094                                 break;
1095                         }
1096                 }
1097                 *dest++ = c;
1098         }
1099         *dest = '\0';
1100 }
1101
1102 /*
1103  * Note that this implementation does not (and should not!) obey
1104  * locale settings; you cannot simply substitute strtol here, since
1105  * it does obey locale.
1106  */
1107 static int64_t
1108 mtree_atol8(char **p)
1109 {
1110         int64_t l, limit, last_digit_limit;
1111         int digit, base;
1112
1113         base = 8;
1114         limit = INT64_MAX / base;
1115         last_digit_limit = INT64_MAX % base;
1116
1117         l = 0;
1118         digit = **p - '0';
1119         while (digit >= 0 && digit < base) {
1120                 if (l>limit || (l == limit && digit > last_digit_limit)) {
1121                         l = INT64_MAX; /* Truncate on overflow. */
1122                         break;
1123                 }
1124                 l = (l * base) + digit;
1125                 digit = *++(*p) - '0';
1126         }
1127         return (l);
1128 }
1129
1130 /*
1131  * Note that this implementation does not (and should not!) obey
1132  * locale settings; you cannot simply substitute strtol here, since
1133  * it does obey locale.
1134  */
1135 static int64_t
1136 mtree_atol10(char **p)
1137 {
1138         int64_t l, limit, last_digit_limit;
1139         int base, digit, sign;
1140
1141         base = 10;
1142         limit = INT64_MAX / base;
1143         last_digit_limit = INT64_MAX % base;
1144
1145         if (**p == '-') {
1146                 sign = -1;
1147                 ++(*p);
1148         } else
1149                 sign = 1;
1150
1151         l = 0;
1152         digit = **p - '0';
1153         while (digit >= 0 && digit < base) {
1154                 if (l > limit || (l == limit && digit > last_digit_limit)) {
1155                         l = UINT64_MAX; /* Truncate on overflow. */
1156                         break;
1157                 }
1158                 l = (l * base) + digit;
1159                 digit = *++(*p) - '0';
1160         }
1161         return (sign < 0) ? -l : l;
1162 }
1163
1164 /*
1165  * Note that this implementation does not (and should not!) obey
1166  * locale settings; you cannot simply substitute strtol here, since
1167  * it does obey locale.
1168  */
1169 static int64_t
1170 mtree_atol16(char **p)
1171 {
1172         int64_t l, limit, last_digit_limit;
1173         int base, digit, sign;
1174
1175         base = 16;
1176         limit = INT64_MAX / base;
1177         last_digit_limit = INT64_MAX % base;
1178
1179         if (**p == '-') {
1180                 sign = -1;
1181                 ++(*p);
1182         } else
1183                 sign = 1;
1184
1185         l = 0;
1186         if (**p >= '0' && **p <= '9')
1187                 digit = **p - '0';
1188         else if (**p >= 'a' && **p <= 'f')
1189                 digit = **p - 'a' + 10;
1190         else if (**p >= 'A' && **p <= 'F')
1191                 digit = **p - 'A' + 10;
1192         else
1193                 digit = -1;
1194         while (digit >= 0 && digit < base) {
1195                 if (l > limit || (l == limit && digit > last_digit_limit)) {
1196                         l = UINT64_MAX; /* Truncate on overflow. */
1197                         break;
1198                 }
1199                 l = (l * base) + digit;
1200                 if (**p >= '0' && **p <= '9')
1201                         digit = **p - '0';
1202                 else if (**p >= 'a' && **p <= 'f')
1203                         digit = **p - 'a' + 10;
1204                 else if (**p >= 'A' && **p <= 'F')
1205                         digit = **p - 'A' + 10;
1206                 else
1207                         digit = -1;
1208         }
1209         return (sign < 0) ? -l : l;
1210 }
1211
1212 static int64_t
1213 mtree_atol(char **p)
1214 {
1215         if (**p != '0')
1216                 return mtree_atol10(p);
1217         if ((*p)[1] == 'x' || (*p)[1] == 'X') {
1218                 *p += 2;
1219                 return mtree_atol16(p);
1220         }
1221         return mtree_atol8(p);
1222 }
1223
1224 /*
1225  * Returns length of line (including trailing newline)
1226  * or negative on error.  'start' argument is updated to
1227  * point to first character of line.
1228  */
1229 static ssize_t
1230 readline(struct archive_read *a, struct mtree *mtree, char **start, ssize_t limit)
1231 {
1232         ssize_t bytes_read;
1233         ssize_t total_size = 0;
1234         const void *t;
1235         const char *s;
1236         void *p;
1237         char *u;
1238
1239         /* Accumulate line in a line buffer. */
1240         for (;;) {
1241                 /* Read some more. */
1242                 bytes_read = (a->decompressor->read_ahead)(a, &t, 1);
1243                 if (bytes_read == 0)
1244                         return (0);
1245                 if (bytes_read < 0)
1246                         return (ARCHIVE_FATAL);
1247                 s = t;  /* Start of line? */
1248                 p = memchr(t, '\n', bytes_read);
1249                 /* If we found '\n', trim the read. */
1250                 if (p != NULL) {
1251                         bytes_read = 1 + ((const char *)p) - s;
1252                 }
1253                 if (total_size + bytes_read + 1 > limit) {
1254                         archive_set_error(&a->archive,
1255                             ARCHIVE_ERRNO_FILE_FORMAT,
1256                             "Line too long");
1257                         return (ARCHIVE_FATAL);
1258                 }
1259                 if (archive_string_ensure(&mtree->line,
1260                         total_size + bytes_read + 1) == NULL) {
1261                         archive_set_error(&a->archive, ENOMEM,
1262                             "Can't allocate working buffer");
1263                         return (ARCHIVE_FATAL);
1264                 }
1265                 memcpy(mtree->line.s + total_size, t, bytes_read);
1266                 (a->decompressor->consume)(a, bytes_read);
1267                 total_size += bytes_read;
1268                 /* Null terminate. */
1269                 mtree->line.s[total_size] = '\0';
1270                 /* If we found an unescaped '\n', clean up and return. */
1271                 if (p == NULL)
1272                         continue;
1273                 for (u = mtree->line.s; *u; ++u) {
1274                         if (u[0] == '\n') {
1275                                 *start = mtree->line.s;
1276                                 return total_size;
1277                         }
1278                         if (u[0] == '#') {
1279                                 if (p == NULL)
1280                                         break;
1281                                 *start = mtree->line.s;
1282                                 return total_size;
1283                         }
1284                         if (u[0] != '\\')
1285                                 continue;
1286                         if (u[1] == '\\') {
1287                                 ++u;
1288                                 continue;
1289                         }
1290                         if (u[1] == '\n') {
1291                                 memmove(u, u + 1,
1292                                     total_size - (u - mtree->line.s) + 1);
1293                                 --total_size;
1294                                 continue;
1295                         }
1296                 }
1297         }
1298 }