archival/unzip.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * Mini unzip implementation for busybox
   4  *
   5  * Copyright (C) 2004 by Ed Clark
   6  *
   7  * Loosely based on original busybox unzip applet by Laurence Anderson.
   8  * All options and features should work in this version.
   9  *
  10  * Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
  11  */
  12
  13 /* For reference see
  14  * http://www.pkware.com/company/standards/appnote/
  15  * http://www.info-zip.org/pub/infozip/doc/appnote-iz-latest.zip
  16  */
  17
  18 /* TODO
  19  * Zip64 + other methods
  20  */
  21
  22 #include "libbb.h"
  23 #include "unarchive.h"
  24
  25 enum {
  26 #if BB_BIG_ENDIAN
  27         ZIP_FILEHEADER_MAGIC = 0x504b0304,
  28         ZIP_CDS_MAGIC        = 0x504b0102,
  29         ZIP_CDE_MAGIC        = 0x504b0506,
  30         ZIP_DD_MAGIC         = 0x504b0708,
  31 #else
  32         ZIP_FILEHEADER_MAGIC = 0x04034b50,
  33         ZIP_CDS_MAGIC        = 0x02014b50,
  34         ZIP_CDE_MAGIC        = 0x06054b50,
  35         ZIP_DD_MAGIC         = 0x08074b50,
  36 #endif
  37 };
  38
  39 #define ZIP_HEADER_LEN 26
  40
  41 typedef union {
  42         uint8_t raw[ZIP_HEADER_LEN];
  43         struct {
  44                 uint16_t version;               /* 0-1 */
  45                 uint16_t flags;                 /* 2-3 */
  46                 uint16_t method;                /* 4-5 */
  47                 uint16_t modtime;               /* 6-7 */
  48                 uint16_t moddate;               /* 8-9 */
  49                 uint32_t crc32 PACKED;          /* 10-13 */
  50                 uint32_t cmpsize PACKED;        /* 14-17 */
  51                 uint32_t ucmpsize PACKED;       /* 18-21 */
  52                 uint16_t filename_len;          /* 22-23 */
  53                 uint16_t extra_len;             /* 24-25 */
  54         } formatted PACKED;
  55 } zip_header_t; /* PACKED - gcc 4.2.1 doesn't like it (spews warning) */
  56
  57 /* Check the offset of the last element, not the length.  This leniency
  58  * allows for poor packing, whereby the overall struct may be too long,
  59  * even though the elements are all in the right place.
  60  */
  61 struct BUG_zip_header_must_be_26_bytes {
  62         char BUG_zip_header_must_be_26_bytes[
  63                 offsetof(zip_header_t, formatted.extra_len) + 2
  64                         == ZIP_HEADER_LEN ? 1 : -1];
  65 };
  66
  67 #define FIX_ENDIANNESS_ZIP(zip_header) do { \
  68         (zip_header).formatted.version      = SWAP_LE16((zip_header).formatted.version     ); \
  69         (zip_header).formatted.flags        = SWAP_LE16((zip_header).formatted.flags       ); \
  70         (zip_header).formatted.method       = SWAP_LE16((zip_header).formatted.method      ); \
  71         (zip_header).formatted.modtime      = SWAP_LE16((zip_header).formatted.modtime     ); \
  72         (zip_header).formatted.moddate      = SWAP_LE16((zip_header).formatted.moddate     ); \
  73         (zip_header).formatted.crc32        = SWAP_LE32((zip_header).formatted.crc32       ); \
  74         (zip_header).formatted.cmpsize      = SWAP_LE32((zip_header).formatted.cmpsize     ); \
  75         (zip_header).formatted.ucmpsize     = SWAP_LE32((zip_header).formatted.ucmpsize    ); \
  76         (zip_header).formatted.filename_len = SWAP_LE16((zip_header).formatted.filename_len); \
  77         (zip_header).formatted.extra_len    = SWAP_LE16((zip_header).formatted.extra_len   ); \
  78 } while (0)
  79
  80 #define CDS_HEADER_LEN 42
  81
  82 typedef union {
  83         uint8_t raw[CDS_HEADER_LEN];
  84         struct {
  85                 /* uint32_t signature; 50 4b 01 02 */
  86                 uint16_t version_made_by;       /* 0-1 */
  87                 uint16_t version_needed;        /* 2-3 */
  88                 uint16_t cds_flags;             /* 4-5 */
  89                 uint16_t method;                /* 6-7 */
  90                 uint16_t mtime;                 /* 8-9 */
  91                 uint16_t mdate;                 /* 10-11 */
  92                 uint32_t crc32;                 /* 12-15 */
  93                 uint32_t cmpsize;               /* 16-19 */
  94                 uint32_t ucmpsize;              /* 20-23 */
  95                 uint16_t file_name_length;      /* 24-25 */
  96                 uint16_t extra_field_length;    /* 26-27 */
  97                 uint16_t file_comment_length;   /* 28-29 */
  98                 uint16_t disk_number_start;     /* 30-31 */
  99                 uint16_t internal_file_attributes; /* 32-33 */
 100                 uint32_t external_file_attributes PACKED; /* 34-37 */
 101                 uint32_t relative_offset_of_local_header PACKED; /* 38-41 */
 102         } formatted PACKED;
 103 } cds_header_t;
 104
 105 struct BUG_cds_header_must_be_42_bytes {
 106         char BUG_cds_header_must_be_42_bytes[
 107                 offsetof(cds_header_t, formatted.relative_offset_of_local_header) + 4
 108                         == CDS_HEADER_LEN ? 1 : -1];
 109 };
 110
 111 #define FIX_ENDIANNESS_CDS(cds_header) do { \
 112         (cds_header).formatted.crc32        = SWAP_LE32((cds_header).formatted.crc32       ); \
 113         (cds_header).formatted.cmpsize      = SWAP_LE32((cds_header).formatted.cmpsize     ); \
 114         (cds_header).formatted.ucmpsize     = SWAP_LE32((cds_header).formatted.ucmpsize    ); \
 115         (cds_header).formatted.file_name_length = SWAP_LE16((cds_header).formatted.file_name_length); \
 116         (cds_header).formatted.extra_field_length = SWAP_LE16((cds_header).formatted.extra_field_length); \
 117         (cds_header).formatted.file_comment_length = SWAP_LE16((cds_header).formatted.file_comment_length); \
 118 } while (0)
 119
 120 #define CDE_HEADER_LEN 16
 121
 122 typedef union {
 123         uint8_t raw[CDE_HEADER_LEN];
 124         struct {
 125                 /* uint32_t signature; 50 4b 05 06 */
 126                 uint16_t this_disk_no;
 127                 uint16_t disk_with_cds_no;
 128                 uint16_t cds_entries_on_this_disk;
 129                 uint16_t cds_entries_total;
 130                 uint32_t cds_size;
 131                 uint32_t cds_offset;
 132                 /* uint16_t file_comment_length; */
 133                 /* .ZIP file comment (variable size) */
 134         } formatted PACKED;
 135 } cde_header_t;
 136
 137 struct BUG_cde_header_must_be_16_bytes {
 138         char BUG_cde_header_must_be_16_bytes[
 139                 sizeof(cde_header_t) == CDE_HEADER_LEN ? 1 : -1];
 140 };
 141
 142 #define FIX_ENDIANNESS_CDE(cde_header) do { \
 143         (cde_header).formatted.cds_offset = SWAP_LE16((cde_header).formatted.cds_offset); \
 144 } while (0)
 145
 146 enum { zip_fd = 3 };
 147
 148
 149 #if ENABLE_DESKTOP
 150 /* NB: does not preserve file position! */
 151 static uint32_t find_cds_offset(void)
 152 {
 153         unsigned char buf[1024];
 154         cde_header_t cde_header;
 155         unsigned char *p;
 156         off_t end;
 157
 158         end = xlseek(zip_fd, 0, SEEK_END);
 159         if (end < 1024)
 160                 end = 1024;
 161         end -= 1024;
 162         xlseek(zip_fd, end, SEEK_SET);
 163         full_read(zip_fd, buf, 1024);
 164
 165         p = buf;
 166         while (p <= buf + 1024 - CDE_HEADER_LEN - 4) {
 167                 if (*p != 'P') {
 168                         p++;
 169                         continue;
 170                 }
 171                 if (*++p != 'K')
 172                         continue;
 173                 if (*++p != 5)
 174                         continue;
 175                 if (*++p != 6)
 176                         continue;
 177                 /* we found CDE! */
 178                 memcpy(cde_header.raw, p + 1, CDE_HEADER_LEN);
 179                 FIX_ENDIANNESS_CDE(cde_header);
 180                 return cde_header.formatted.cds_offset;
 181         }
 182         bb_error_msg_and_die("can't find file table");
 183 };
 184
 185 static uint32_t read_next_cds(int count_m1, uint32_t cds_offset, cds_header_t *cds_ptr)
 186 {
 187         off_t org;
 188
 189         org = xlseek(zip_fd, 0, SEEK_CUR);
 190
 191         if (!cds_offset)
 192                 cds_offset = find_cds_offset();
 193
 194         while (count_m1-- >= 0) {
 195                 xlseek(zip_fd, cds_offset + 4, SEEK_SET);
 196                 xread(zip_fd, cds_ptr->raw, CDS_HEADER_LEN);
 197                 FIX_ENDIANNESS_CDS(*cds_ptr);
 198                 cds_offset += 4 + CDS_HEADER_LEN
 199                         + cds_ptr->formatted.file_name_length
 200                         + cds_ptr->formatted.extra_field_length
 201                         + cds_ptr->formatted.file_comment_length;
 202         }
 203
 204         xlseek(zip_fd, org, SEEK_SET);
 205         return cds_offset;
 206 };
 207 #endif
 208
 209 static void unzip_skip(off_t skip)
 210 {
 211         bb_copyfd_exact_size(zip_fd, -1, skip);
 212 }
 213
 214 static void unzip_create_leading_dirs(const char *fn)
 215 {
 216         /* Create all leading directories */
 217         char *name = xstrdup(fn);
 218         if (bb_make_directory(dirname(name), 0777, FILEUTILS_RECUR)) {
 219                 bb_error_msg_and_die("exiting"); /* bb_make_directory is noisy */
 220         }
 221         free(name);
 222 }
 223
 224 static void unzip_extract(zip_header_t *zip_header, int dst_fd)
 225 {
 226         if (zip_header->formatted.method == 0) {
 227                 /* Method 0 - stored (not compressed) */
 228                 off_t size = zip_header->formatted.ucmpsize;
 229                 if (size)
 230                         bb_copyfd_exact_size(zip_fd, dst_fd, size);
 231         } else {
 232                 /* Method 8 - inflate */
 233                 inflate_unzip_result res;
 234                 if (inflate_unzip(&res, zip_header->formatted.cmpsize, zip_fd, dst_fd) < 0)
 235                         bb_error_msg_and_die("inflate error");
 236                 /* Validate decompression - crc */
 237                 if (zip_header->formatted.crc32 != (res.crc ^ 0xffffffffL)) {
 238                         bb_error_msg_and_die("crc error");
 239                 }
 240                 /* Validate decompression - size */
 241                 if (zip_header->formatted.ucmpsize != res.bytes_out) {
 242                         /* Don't die. Who knows, maybe len calculation
 243                          * was botched somewhere. After all, crc matched! */
 244                         bb_error_msg("bad length");
 245                 }
 246         }
 247 }
 248
 249 int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 250 int unzip_main(int argc, char **argv)
 251 {
 252         enum { O_PROMPT, O_NEVER, O_ALWAYS };
 253
 254         zip_header_t zip_header;
 255         smallint verbose = 1;
 256         smallint listing = 0;
 257         smallint overwrite = O_PROMPT;
 258 #if ENABLE_DESKTOP
 259         uint32_t cds_offset;
 260         unsigned cds_entries;
 261 #endif
 262         unsigned total_size;
 263         unsigned total_entries;
 264         int dst_fd = -1;
 265         char *src_fn = NULL;
 266         char *dst_fn = NULL;
 267         llist_t *zaccept = NULL;
 268         llist_t *zreject = NULL;
 269         char *base_dir = NULL;
 270         int i, opt;
 271         int opt_range = 0;
 272         char key_buf[80];
 273         struct stat stat_buf;
 274
 275         /* '-' makes getopt return 1 for non-options */
 276         while ((opt = getopt(argc, argv, "-d:lnopqx")) != -1) {
 277                 switch (opt_range) {
 278                 case 0: /* Options */
 279                         switch (opt) {
 280                         case 'l': /* List */
 281                                 listing = 1;
 282                                 break;
 283
 284                         case 'n': /* Never overwrite existing files */
 285                                 overwrite = O_NEVER;
 286                                 break;
 287
 288                         case 'o': /* Always overwrite existing files */
 289                                 overwrite = O_ALWAYS;
 290                                 break;
 291
 292                         case 'p': /* Extract files to stdout and fall through to set verbosity */
 293                                 dst_fd = STDOUT_FILENO;
 294
 295                         case 'q': /* Be quiet */
 296                                 verbose = 0;
 297                                 break;
 298
 299                         case 1: /* The zip file */
 300                                 /* +5: space for ".zip" and NUL */
 301                                 src_fn = xmalloc(strlen(optarg) + 5);
 302                                 strcpy(src_fn, optarg);
 303                                 opt_range++;
 304                                 break;
 305
 306                         default:
 307                                 bb_show_usage();
 308
 309                         }
 310                         break;
 311
 312                 case 1: /* Include files */
 313                         if (opt == 1) {
 314                                 llist_add_to(&zaccept, optarg);
 315                                 break;
 316                         }
 317                         if (opt == 'd') {
 318                                 base_dir = optarg;
 319                                 opt_range += 2;
 320                                 break;
 321                         }
 322                         if (opt == 'x') {
 323                                 opt_range++;
 324                                 break;
 325                         }
 326                         bb_show_usage();
 327
 328                 case 2 : /* Exclude files */
 329                         if (opt == 1) {
 330                                 llist_add_to(&zreject, optarg);
 331                                 break;
 332                         }
 333                         if (opt == 'd') { /* Extract to base directory */
 334                                 base_dir = optarg;
 335                                 opt_range++;
 336                                 break;
 337                         }
 338                         /* fall through */
 339
 340                 default:
 341                         bb_show_usage();
 342                 }
 343         }
 344
 345         if (src_fn == NULL) {
 346                 bb_show_usage();
 347         }
 348
 349         /* Open input file */
 350         if (LONE_DASH(src_fn)) {
 351                 xdup2(STDIN_FILENO, zip_fd);
 352                 /* Cannot use prompt mode since zip data is arriving on STDIN */
 353                 if (overwrite == O_PROMPT)
 354                         overwrite = O_NEVER;
 355         } else {
 356                 static const char extn[][5] = {"", ".zip", ".ZIP"};
 357                 int orig_src_fn_len = strlen(src_fn);
 358                 int src_fd = -1;
 359
 360                 for (i = 0; (i < 3) && (src_fd == -1); i++) {
 361                         strcpy(src_fn + orig_src_fn_len, extn[i]);
 362                         src_fd = open(src_fn, O_RDONLY);
 363                 }
 364                 if (src_fd == -1) {
 365                         src_fn[orig_src_fn_len] = '\0';
 366                         bb_error_msg_and_die("can't open %s, %s.zip, %s.ZIP", src_fn, src_fn, src_fn);
 367                 }
 368                 xmove_fd(src_fd, zip_fd);
 369         }
 370
 371         /* Change dir if necessary */
 372         if (base_dir)
 373                 xchdir(base_dir);
 374
 375         if (verbose) {
 376                 printf("Archive:  %s\n", src_fn);
 377                 if (listing){
 378                         puts("  Length     Date   Time    Name\n"
 379                              " --------    ----   ----    ----");
 380                 }
 381         }
 382
 383         total_size = 0;
 384         total_entries = 0;
 385 #if ENABLE_DESKTOP
 386         cds_entries = 0;
 387         cds_offset = 0;
 388 #endif
 389         while (1) {
 390                 uint32_t magic;
 391
 392                 /* Check magic number */
 393                 xread(zip_fd, &magic, 4);
 394                 /* Central directory? It's at the end, so exit */
 395                 if (magic == ZIP_CDS_MAGIC)
 396                         break;
 397 #if ENABLE_DESKTOP
 398                 /* Data descriptor? It was a streaming file, go on */
 399                 if (magic == ZIP_DD_MAGIC) {
 400                         /* skip over duplicate crc32, cmpsize and ucmpsize */
 401                         unzip_skip(3 * 4);
 402                         continue;
 403                 }
 404 #endif
 405                 if (magic != ZIP_FILEHEADER_MAGIC)
 406                         bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
 407
 408                 /* Read the file header */
 409                 xread(zip_fd, zip_header.raw, ZIP_HEADER_LEN);
 410                 FIX_ENDIANNESS_ZIP(zip_header);
 411                 if ((zip_header.formatted.method != 0) && (zip_header.formatted.method != 8)) {
 412                         bb_error_msg_and_die("unsupported method %d", zip_header.formatted.method);
 413                 }
 414 #if !ENABLE_DESKTOP
 415                 if (zip_header.formatted.flags & 0x0009) {
 416                         bb_error_msg_and_die("zip flags 1 and 8 are not supported");
 417                 }
 418 #else
 419                 if (zip_header.formatted.flags & 0x0001) {
 420                         /* 0x0001 - encrypted */
 421                         bb_error_msg_and_die("zip flag 1 (encryption) is not supported");
 422                 }
 423                 if (zip_header.formatted.flags & 0x0008) {
 424                         cds_header_t cds_header;
 425                         /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
 426                          * only from Central Directory. See unzip_doc.txt */
 427                         cds_offset = read_next_cds(total_entries - cds_entries, cds_offset, &cds_header);
 428                         cds_entries = total_entries + 1;
 429                         zip_header.formatted.crc32    = cds_header.formatted.crc32;
 430                         zip_header.formatted.cmpsize  = cds_header.formatted.cmpsize;
 431                         zip_header.formatted.ucmpsize = cds_header.formatted.ucmpsize;
 432                 }
 433 #endif
 434
 435                 /* Read filename */
 436                 free(dst_fn);
 437                 dst_fn = xzalloc(zip_header.formatted.filename_len + 1);
 438                 xread(zip_fd, dst_fn, zip_header.formatted.filename_len);
 439
 440                 /* Skip extra header bytes */
 441                 unzip_skip(zip_header.formatted.extra_len);
 442
 443                 /* Filter zip entries */
 444                 if (find_list_entry(zreject, dst_fn)
 445                  || (zaccept && !find_list_entry(zaccept, dst_fn))
 446                 ) { /* Skip entry */
 447                         i = 'n';
 448
 449                 } else { /* Extract entry */
 450                         if (listing) { /* List entry */
 451                                 if (verbose) {
 452                                         unsigned dostime = zip_header.formatted.modtime | (zip_header.formatted.moddate << 16);
 453                                         printf("%9u  %02u-%02u-%02u %02u:%02u   %s\n",
 454                                            zip_header.formatted.ucmpsize,
 455                                            (dostime & 0x01e00000) >> 21,
 456                                            (dostime & 0x001f0000) >> 16,
 457                                            (((dostime & 0xfe000000) >> 25) + 1980) % 100,
 458                                            (dostime & 0x0000f800) >> 11,
 459                                            (dostime & 0x000007e0) >> 5,
 460                                            dst_fn);
 461                                         total_size += zip_header.formatted.ucmpsize;
 462                                 } else {
 463                                         /* short listing -- filenames only */
 464                                         puts(dst_fn);
 465                                 }
 466                                 i = 'n';
 467                         } else if (dst_fd == STDOUT_FILENO) { /* Extracting to STDOUT */
 468                                 i = -1;
 469                         } else if (last_char_is(dst_fn, '/')) { /* Extract directory */
 470                                 if (stat(dst_fn, &stat_buf) == -1) {
 471                                         if (errno != ENOENT) {
 472                                                 bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 473                                         }
 474                                         if (verbose) {
 475                                                 printf("   creating: %s\n", dst_fn);
 476                                         }
 477                                         unzip_create_leading_dirs(dst_fn);
 478                                         if (bb_make_directory(dst_fn, 0777, 0)) {
 479                                                 bb_error_msg_and_die("exiting");
 480                                         }
 481                                 } else {
 482                                         if (!S_ISDIR(stat_buf.st_mode)) {
 483                                                 bb_error_msg_and_die("'%s' exists but is not directory", dst_fn);
 484                                         }
 485                                 }
 486                                 i = 'n';
 487
 488                         } else {  /* Extract file */
 489  check_file:
 490                                 if (stat(dst_fn, &stat_buf) == -1) { /* File does not exist */
 491                                         if (errno != ENOENT) {
 492                                                 bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 493                                         }
 494                                         i = 'y';
 495                                 } else { /* File already exists */
 496                                         if (overwrite == O_NEVER) {
 497                                                 i = 'n';
 498                                         } else if (S_ISREG(stat_buf.st_mode)) { /* File is regular file */
 499                                                 if (overwrite == O_ALWAYS) {
 500                                                         i = 'y';
 501                                                 } else {
 502                                                         printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
 503                                                         if (!fgets(key_buf, sizeof(key_buf), stdin)) {
 504                                                                 bb_perror_msg_and_die("can't read input");
 505                                                         }
 506                                                         i = key_buf[0];
 507                                                 }
 508                                         } else { /* File is not regular file */
 509                                                 bb_error_msg_and_die("'%s' exists but is not regular file", dst_fn);
 510                                         }
 511                                 }
 512                         }
 513                 }
 514
 515                 switch (i) {
 516                 case 'A':
 517                         overwrite = O_ALWAYS;
 518                 case 'y': /* Open file and fall into unzip */
 519                         unzip_create_leading_dirs(dst_fn);
 520                         dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC);
 521                 case -1: /* Unzip */
 522                         if (verbose) {
 523                                 printf("  inflating: %s\n", dst_fn);
 524                         }
 525                         unzip_extract(&zip_header, dst_fd);
 526                         if (dst_fd != STDOUT_FILENO) {
 527                                 /* closing STDOUT is potentially bad for future business */
 528                                 close(dst_fd);
 529                         }
 530                         break;
 531
 532                 case 'N':
 533                         overwrite = O_NEVER;
 534                 case 'n':
 535                         /* Skip entry data */
 536                         unzip_skip(zip_header.formatted.cmpsize);
 537                         break;
 538
 539                 case 'r':
 540                         /* Prompt for new name */
 541                         printf("new name: ");
 542                         if (!fgets(key_buf, sizeof(key_buf), stdin)) {
 543                                 bb_perror_msg_and_die("can't read input");
 544                         }
 545                         free(dst_fn);
 546                         dst_fn = xstrdup(key_buf);
 547                         chomp(dst_fn);
 548                         goto check_file;
 549
 550                 default:
 551                         printf("error: invalid response [%c]\n",(char)i);
 552                         goto check_file;
 553                 }
 554
 555                 total_entries++;
 556         }
 557
 558         if (listing && verbose) {
 559                 printf(" --------                   -------\n"
 560                        "%9d                   %d files\n",
 561                        total_size, total_entries);
 562         }
 563
 564         return 0;
 565 }