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 GPLv2 or later, see file LICENSE in this source tree.
  11  */
  12 /* For reference see
  13  * http://www.pkware.com/company/standards/appnote/
  14  * http://www.info-zip.org/pub/infozip/doc/appnote-iz-latest.zip
  15  *
  16  * TODO
  17  * Zip64 + other methods
  18  */
  19 //config:config UNZIP
  20 //config:       bool "unzip"
  21 //config:       default y
  22 //config:       help
  23 //config:         unzip will list or extract files from a ZIP archive,
  24 //config:         commonly found on DOS/WIN systems. The default behavior
  25 //config:         (with no options) is to extract the archive into the
  26 //config:         current directory.
  27 //config:
  28 //config:config FEATURE_UNZIP_CDF
  29 //config:       bool "Read and use Central Directory data"
  30 //config:       default y
  31 //config:       depends on UNZIP
  32 //config:       help
  33 //config:         If you know that you only need to deal with simple
  34 //config:         ZIP files without deleted/updated files, SFX archives etc,
  35 //config:         you can reduce code size by unselecting this option.
  36 //config:         To support less trivial ZIPs, say Y.
  37 //config:
  38 //config:config FEATURE_UNZIP_BZIP2
  39 //config:       bool "Support compression method 12 (bzip2)"
  40 //config:       default y
  41 //config:       depends on FEATURE_UNZIP_CDF && DESKTOP
  42 // FEATURE_UNZIP_CDF is needed, otherwise we can't find start of next file
  43 // DESKTOP is needed to get back uncompressed length
  44 //config:
  45 //config:config FEATURE_UNZIP_LZMA
  46 //config:       bool "Support compression method 14 (lzma)"
  47 //config:       default y
  48 //config:       depends on FEATURE_UNZIP_CDF && DESKTOP
  49 //config:
  50 //config:config FEATURE_UNZIP_XZ
  51 //config:       bool "Support compression method 95 (xz)"
  52 //config:       default y
  53 //config:       depends on FEATURE_UNZIP_CDF && DESKTOP
  54
  55 //applet:IF_UNZIP(APPLET(unzip, BB_DIR_USR_BIN, BB_SUID_DROP))
  56 //kbuild:lib-$(CONFIG_UNZIP) += unzip.o
  57
  58 //usage:#define unzip_trivial_usage
  59 //usage:       "[-lnopq] FILE[.zip] [FILE]... [-x FILE...] [-d DIR]"
  60 //usage:#define unzip_full_usage "\n\n"
  61 //usage:       "Extract FILEs from ZIP archive\n"
  62 //usage:     "\n        -l      List contents (with -q for short form)"
  63 //usage:     "\n        -n      Never overwrite files (default: ask)"
  64 //usage:     "\n        -o      Overwrite"
  65 //usage:     "\n        -p      Print to stdout"
  66 //usage:     "\n        -q      Quiet"
  67 //usage:     "\n        -x FILE Exclude FILEs"
  68 //usage:     "\n        -d DIR  Extract into DIR"
  69
  70 #include "libbb.h"
  71 #include "bb_archive.h"
  72
  73 #if 0
  74 # define dbg(...) bb_error_msg(__VA_ARGS__)
  75 #else
  76 # define dbg(...) ((void)0)
  77 #endif
  78
  79 enum {
  80 #if BB_BIG_ENDIAN
  81         ZIP_FILEHEADER_MAGIC = 0x504b0304,
  82         ZIP_CDF_MAGIC        = 0x504b0102, /* CDF item */
  83         ZIP_CDE_MAGIC        = 0x504b0506, /* End of CDF */
  84         ZIP_DD_MAGIC         = 0x504b0708,
  85 #else
  86         ZIP_FILEHEADER_MAGIC = 0x04034b50,
  87         ZIP_CDF_MAGIC        = 0x02014b50,
  88         ZIP_CDE_MAGIC        = 0x06054b50,
  89         ZIP_DD_MAGIC         = 0x08074b50,
  90 #endif
  91 };
  92
  93 #define ZIP_HEADER_LEN 26
  94
  95 typedef union {
  96         uint8_t raw[ZIP_HEADER_LEN];
  97         struct {
  98                 uint16_t version;               /* 0-1 */
  99                 uint16_t zip_flags;             /* 2-3 */
 100                 uint16_t method;                /* 4-5 */
 101                 uint16_t modtime;               /* 6-7 */
 102                 uint16_t moddate;               /* 8-9 */
 103                 uint32_t crc32 PACKED;          /* 10-13 */
 104                 uint32_t cmpsize PACKED;        /* 14-17 */
 105                 uint32_t ucmpsize PACKED;       /* 18-21 */
 106                 uint16_t filename_len;          /* 22-23 */
 107                 uint16_t extra_len;             /* 24-25 */
 108                 /* filename follows (not NUL terminated) */
 109                 /* extra field follows */
 110                 /* data follows */
 111         } fmt PACKED;
 112 } zip_header_t; /* PACKED - gcc 4.2.1 doesn't like it (spews warning) */
 113
 114 #define FIX_ENDIANNESS_ZIP(zip) \
 115 do { if (BB_BIG_ENDIAN) { \
 116         (zip).fmt.crc32         = SWAP_LE32((zip).fmt.crc32       ); \
 117         (zip).fmt.cmpsize       = SWAP_LE32((zip).fmt.cmpsize     ); \
 118         (zip).fmt.ucmpsize      = SWAP_LE32((zip).fmt.ucmpsize    ); \
 119         (zip).fmt.filename_len  = SWAP_LE16((zip).fmt.filename_len); \
 120         (zip).fmt.extra_len     = SWAP_LE16((zip).fmt.extra_len   ); \
 121 }} while (0)
 122
 123 #define CDF_HEADER_LEN 42
 124
 125 typedef union {
 126         uint8_t raw[CDF_HEADER_LEN];
 127         struct {
 128                 /* uint32_t signature; 50 4b 01 02 */
 129                 uint16_t version_made_by;       /* 0-1 */
 130                 uint16_t version_needed;        /* 2-3 */
 131                 uint16_t cdf_flags;             /* 4-5 */
 132                 uint16_t method;                /* 6-7 */
 133                 uint16_t modtime;               /* 8-9 */
 134                 uint16_t moddate;               /* 10-11 */
 135                 uint32_t crc32;                 /* 12-15 */
 136                 uint32_t cmpsize;               /* 16-19 */
 137                 uint32_t ucmpsize;              /* 20-23 */
 138                 uint16_t filename_len;          /* 24-25 */
 139                 uint16_t extra_len;             /* 26-27 */
 140                 uint16_t file_comment_length;   /* 28-29 */
 141                 uint16_t disk_number_start;     /* 30-31 */
 142                 uint16_t internal_attributes;   /* 32-33 */
 143                 uint32_t external_attributes PACKED; /* 34-37 */
 144                 uint32_t relative_offset_of_local_header PACKED; /* 38-41 */
 145                 /* filename follows (not NUL terminated) */
 146                 /* extra field follows */
 147                 /* file comment follows */
 148         } fmt PACKED;
 149 } cdf_header_t;
 150
 151 #define FIX_ENDIANNESS_CDF(cdf) \
 152 do { if (BB_BIG_ENDIAN) { \
 153         (cdf).fmt.version_made_by = SWAP_LE16((cdf).fmt.version_made_by); \
 154         (cdf).fmt.version_needed = SWAP_LE16((cdf).fmt.version_needed); \
 155         (cdf).fmt.method        = SWAP_LE16((cdf).fmt.method      ); \
 156         (cdf).fmt.modtime       = SWAP_LE16((cdf).fmt.modtime     ); \
 157         (cdf).fmt.moddate       = SWAP_LE16((cdf).fmt.moddate     ); \
 158         (cdf).fmt.crc32         = SWAP_LE32((cdf).fmt.crc32       ); \
 159         (cdf).fmt.cmpsize       = SWAP_LE32((cdf).fmt.cmpsize     ); \
 160         (cdf).fmt.ucmpsize      = SWAP_LE32((cdf).fmt.ucmpsize    ); \
 161         (cdf).fmt.filename_len  = SWAP_LE16((cdf).fmt.filename_len); \
 162         (cdf).fmt.extra_len     = SWAP_LE16((cdf).fmt.extra_len   ); \
 163         (cdf).fmt.file_comment_length = SWAP_LE16((cdf).fmt.file_comment_length); \
 164         (cdf).fmt.external_attributes = SWAP_LE32((cdf).fmt.external_attributes); \
 165 }} while (0)
 166
 167 #define CDE_LEN 16
 168
 169 typedef union {
 170         uint8_t raw[CDE_LEN];
 171         struct {
 172                 /* uint32_t signature; 50 4b 05 06 */
 173                 uint16_t this_disk_no;
 174                 uint16_t disk_with_cdf_no;
 175                 uint16_t cdf_entries_on_this_disk;
 176                 uint16_t cdf_entries_total;
 177                 uint32_t cdf_size;
 178                 uint32_t cdf_offset;
 179                 /* uint16_t archive_comment_length; */
 180                 /* archive comment follows */
 181         } fmt PACKED;
 182 } cde_t;
 183
 184 #define FIX_ENDIANNESS_CDE(cde) \
 185 do { if (BB_BIG_ENDIAN) { \
 186         (cde).fmt.cdf_offset = SWAP_LE32((cde).fmt.cdf_offset); \
 187 }} while (0)
 188
 189 struct BUG {
 190         /* Check the offset of the last element, not the length.  This leniency
 191          * allows for poor packing, whereby the overall struct may be too long,
 192          * even though the elements are all in the right place.
 193          */
 194         char BUG_zip_header_must_be_26_bytes[
 195                 offsetof(zip_header_t, fmt.extra_len) + 2
 196                         == ZIP_HEADER_LEN ? 1 : -1];
 197         char BUG_cdf_header_must_be_42_bytes[
 198                 offsetof(cdf_header_t, fmt.relative_offset_of_local_header) + 4
 199                         == CDF_HEADER_LEN ? 1 : -1];
 200         char BUG_cde_must_be_16_bytes[
 201                 sizeof(cde_t) == CDE_LEN ? 1 : -1];
 202 };
 203
 204
 205 enum { zip_fd = 3 };
 206
 207
 208 /* This value means that we failed to find CDF */
 209 #define BAD_CDF_OFFSET ((uint32_t)0xffffffff)
 210
 211 #if !ENABLE_FEATURE_UNZIP_CDF
 212
 213 # define find_cdf_offset() BAD_CDF_OFFSET
 214
 215 #else
 216 /* Seen in the wild:
 217  * Self-extracting PRO2K3XP_32.exe contains 19078464 byte zip archive,
 218  * where CDE was nearly 48 kbytes before EOF.
 219  * (Surprisingly, it also apparently has *another* CDE structure
 220  * closer to the end, with bogus cdf_offset).
 221  * To make extraction work, bumped PEEK_FROM_END from 16k to 64k.
 222  */
 223 #define PEEK_FROM_END (64*1024)
 224 /* NB: does not preserve file position! */
 225 static uint32_t find_cdf_offset(void)
 226 {
 227         cde_t cde;
 228         unsigned char *buf;
 229         unsigned char *p;
 230         off_t end;
 231         uint32_t found;
 232
 233         end = lseek(zip_fd, 0, SEEK_END);
 234         if (end == (off_t) -1)
 235                 return BAD_CDF_OFFSET;
 236
 237         end -= PEEK_FROM_END;
 238         if (end < 0)
 239                 end = 0;
 240
 241         dbg("Looking for cdf_offset starting from 0x%"OFF_FMT"x", end);
 242         xlseek(zip_fd, end, SEEK_SET);
 243         buf = xzalloc(PEEK_FROM_END);
 244         full_read(zip_fd, buf, PEEK_FROM_END);
 245
 246         found = BAD_CDF_OFFSET;
 247         p = buf;
 248         while (p <= buf + PEEK_FROM_END - CDE_LEN - 4) {
 249                 if (*p != 'P') {
 250                         p++;
 251                         continue;
 252                 }
 253                 if (*++p != 'K')
 254                         continue;
 255                 if (*++p != 5)
 256                         continue;
 257                 if (*++p != 6)
 258                         continue;
 259                 /* we found CDE! */
 260                 memcpy(cde.raw, p + 1, CDE_LEN);
 261                 FIX_ENDIANNESS_CDE(cde);
 262                 /*
 263                  * I've seen .ZIP files with seemingly valid CDEs
 264                  * where cdf_offset points past EOF - ??
 265                  * This check ignores such CDEs:
 266                  */
 267                 if (cde.fmt.cdf_offset < end + (p - buf)) {
 268                         found = cde.fmt.cdf_offset;
 269                         dbg("Possible cdf_offset:0x%x at 0x%"OFF_FMT"x",
 270                                 (unsigned)found, end + (p-3 - buf));
 271                         dbg("  cdf_offset+cdf_size:0x%x",
 272                                 (unsigned)(found + SWAP_LE32(cde.fmt.cdf_size)));
 273                         /*
 274                          * We do not "break" here because only the last CDE is valid.
 275                          * I've seen a .zip archive which contained a .zip file,
 276                          * uncompressed, and taking the first CDE was using
 277                          * the CDE inside that file!
 278                          */
 279                 }
 280         }
 281         free(buf);
 282         dbg("Found cdf_offset:0x%x", (unsigned)found);
 283         return found;
 284 };
 285
 286 static uint32_t read_next_cdf(uint32_t cdf_offset, cdf_header_t *cdf)
 287 {
 288         uint32_t magic;
 289
 290         if (cdf_offset == BAD_CDF_OFFSET)
 291                 return cdf_offset;
 292
 293         dbg("Reading CDF at 0x%x", (unsigned)cdf_offset);
 294         xlseek(zip_fd, cdf_offset, SEEK_SET);
 295         xread(zip_fd, &magic, 4);
 296         /* Central Directory End? Assume CDF has ended.
 297          * (more correct method is to use cde.cdf_entries_total counter)
 298          */
 299         if (magic == ZIP_CDE_MAGIC) {
 300                 dbg("got ZIP_CDE_MAGIC");
 301                 return 0; /* EOF */
 302         }
 303         xread(zip_fd, cdf->raw, CDF_HEADER_LEN);
 304
 305         FIX_ENDIANNESS_CDF(*cdf);
 306         dbg("  filename_len:%u extra_len:%u file_comment_length:%u",
 307                 (unsigned)cdf->fmt.filename_len,
 308                 (unsigned)cdf->fmt.extra_len,
 309                 (unsigned)cdf->fmt.file_comment_length
 310         );
 311         cdf_offset += 4 + CDF_HEADER_LEN
 312                 + cdf->fmt.filename_len
 313                 + cdf->fmt.extra_len
 314                 + cdf->fmt.file_comment_length;
 315
 316         return cdf_offset;
 317 };
 318 #endif
 319
 320 static void unzip_skip(off_t skip)
 321 {
 322         if (skip != 0)
 323                 if (lseek(zip_fd, skip, SEEK_CUR) == (off_t)-1)
 324                         bb_copyfd_exact_size(zip_fd, -1, skip);
 325 }
 326
 327 static void unzip_create_leading_dirs(const char *fn)
 328 {
 329         /* Create all leading directories */
 330         char *name = xstrdup(fn);
 331         if (bb_make_directory(dirname(name), 0777, FILEUTILS_RECUR)) {
 332                 xfunc_die(); /* bb_make_directory is noisy */
 333         }
 334         free(name);
 335 }
 336
 337 static void unzip_extract(zip_header_t *zip, int dst_fd)
 338 {
 339         transformer_state_t xstate;
 340
 341         if (zip->fmt.method == 0) {
 342                 /* Method 0 - stored (not compressed) */
 343                 off_t size = zip->fmt.ucmpsize;
 344                 if (size)
 345                         bb_copyfd_exact_size(zip_fd, dst_fd, size);
 346                 return;
 347         }
 348
 349 // NB: to support symlinks, need to extract symlink target. A-la:
 350 // xstate.mem_output_size_max = zip->fmt.ucmpsize;
 351 // ...unpack...
 352 // if (xstate.mem_output_buf) { success, xstate.mem_output_size is the size }
 353 // Although archives I've seen have fmt.method == 0 for symlinks.
 354
 355         init_transformer_state(&xstate);
 356         xstate.bytes_in = zip->fmt.cmpsize;
 357         xstate.src_fd = zip_fd;
 358         xstate.dst_fd = dst_fd;
 359         if (zip->fmt.method == 8) {
 360                 /* Method 8 - inflate */
 361                 if (inflate_unzip(&xstate) < 0)
 362                         bb_error_msg_and_die("inflate error");
 363                 /* Validate decompression - crc */
 364                 if (zip->fmt.crc32 != (xstate.crc32 ^ 0xffffffffL)) {
 365                         bb_error_msg_and_die("crc error");
 366                 }
 367         }
 368 #if ENABLE_FEATURE_UNZIP_BZIP2
 369         else if (zip->fmt.method == 12) {
 370                 /* Tested. Unpacker reads too much, but we use CDF
 371                  * and will seek to the correct beginning of next file.
 372                  */
 373                 xstate.bytes_out = unpack_bz2_stream(&xstate);
 374                 if (xstate.bytes_out < 0)
 375                         bb_error_msg_and_die("inflate error");
 376         }
 377 #endif
 378 #if ENABLE_FEATURE_UNZIP_LZMA
 379         else if (zip->fmt.method == 14) {
 380                 /* Not tested yet */
 381                 xstate.bytes_out = unpack_lzma_stream(&xstate);
 382                 if (xstate.bytes_out < 0)
 383                         bb_error_msg_and_die("inflate error");
 384         }
 385 #endif
 386 #if ENABLE_FEATURE_UNZIP_XZ
 387         else if (zip->fmt.method == 95) {
 388                 /* Not tested yet */
 389                 xstate.bytes_out = unpack_xz_stream(&xstate);
 390                 if (xstate.bytes_out < 0)
 391                         bb_error_msg_and_die("inflate error");
 392         }
 393 #endif
 394         else {
 395                 bb_error_msg_and_die("unsupported method %u", zip->fmt.method);
 396         }
 397
 398         /* Validate decompression - size */
 399         if (zip->fmt.ucmpsize != xstate.bytes_out) {
 400                 /* Don't die. Who knows, maybe len calculation
 401                  * was botched somewhere. After all, crc matched! */
 402                 bb_error_msg("bad length");
 403         }
 404 }
 405
 406 static void my_fgets80(char *buf80)
 407 {
 408         fflush_all();
 409         if (!fgets(buf80, 80, stdin)) {
 410                 bb_perror_msg_and_die("can't read standard input");
 411         }
 412 }
 413
 414 int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 415 int unzip_main(int argc, char **argv)
 416 {
 417         enum { O_PROMPT, O_NEVER, O_ALWAYS };
 418
 419         smallint quiet = 0;
 420         IF_NOT_FEATURE_UNZIP_CDF(const) smallint verbose = 0;
 421         smallint listing = 0;
 422         smallint overwrite = O_PROMPT;
 423         smallint x_opt_seen;
 424         uint32_t cdf_offset;
 425         unsigned long total_usize;
 426         unsigned long total_size;
 427         unsigned total_entries;
 428         int dst_fd = -1;
 429         char *src_fn = NULL;
 430         char *dst_fn = NULL;
 431         llist_t *zaccept = NULL;
 432         llist_t *zreject = NULL;
 433         char *base_dir = NULL;
 434         int i, opt;
 435         char key_buf[80]; /* must match size used by my_fgets80 */
 436         struct stat stat_buf;
 437
 438 /* -q, -l and -v: UnZip 5.52 of 28 February 2005, by Info-ZIP:
 439  *
 440  * # /usr/bin/unzip -qq -v decompress_unlzma.i.zip
 441  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 442  * # /usr/bin/unzip -q -v decompress_unlzma.i.zip
 443  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 444  * --------  ------  ------- -----   ----   ----   ------    ----
 445  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 446  * --------          -------  ---                            -------
 447  *   204372            35278  83%                            1 file
 448  * # /usr/bin/unzip -v decompress_unlzma.i.zip
 449  * Archive:  decompress_unlzma.i.zip
 450  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 451  * --------  ------  ------- -----   ----   ----   ------    ----
 452  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 453  * --------          -------  ---                            -------
 454  *   204372            35278  83%                            1 file
 455  * # unzip -v decompress_unlzma.i.zip
 456  * Archive:  decompress_unlzma.i.zip
 457  *   Length     Date   Time    Name
 458  *  --------    ----   ----    ----
 459  *    204372  09-06-09 14:23   decompress_unlzma.i
 460  *  --------                   -------
 461  *    204372                   1 files
 462  * # /usr/bin/unzip -l -qq decompress_unlzma.i.zip
 463  *    204372  09-06-09 14:23   decompress_unlzma.i
 464  * # /usr/bin/unzip -l -q decompress_unlzma.i.zip
 465  *   Length     Date   Time    Name
 466  *  --------    ----   ----    ----
 467  *    204372  09-06-09 14:23   decompress_unlzma.i
 468  *  --------                   -------
 469  *    204372                   1 file
 470  * # /usr/bin/unzip -l decompress_unlzma.i.zip
 471  * Archive:  decompress_unlzma.i.zip
 472  *   Length     Date   Time    Name
 473  *  --------    ----   ----    ----
 474  *    204372  09-06-09 14:23   decompress_unlzma.i
 475  *  --------                   -------
 476  *    204372                   1 file
 477  */
 478
 479         x_opt_seen = 0;
 480         /* '-' makes getopt return 1 for non-options */
 481         while ((opt = getopt(argc, argv, "-d:lnopqxv")) != -1) {
 482                 switch (opt) {
 483                 case 'd':  /* Extract to base directory */
 484                         base_dir = optarg;
 485                         break;
 486
 487                 case 'l': /* List */
 488                         listing = 1;
 489                         break;
 490
 491                 case 'n': /* Never overwrite existing files */
 492                         overwrite = O_NEVER;
 493                         break;
 494
 495                 case 'o': /* Always overwrite existing files */
 496                         overwrite = O_ALWAYS;
 497                         break;
 498
 499                 case 'p': /* Extract files to stdout and fall through to set verbosity */
 500                         dst_fd = STDOUT_FILENO;
 501
 502                 case 'q': /* Be quiet */
 503                         quiet++;
 504                         break;
 505
 506                 case 'v': /* Verbose list */
 507                         IF_FEATURE_UNZIP_CDF(verbose++;)
 508                         listing = 1;
 509                         break;
 510
 511                 case 'x':
 512                         x_opt_seen = 1;
 513                         break;
 514
 515                 case 1:
 516                         if (!src_fn) {
 517                                 /* The zip file */
 518                                 /* +5: space for ".zip" and NUL */
 519                                 src_fn = xmalloc(strlen(optarg) + 5);
 520                                 strcpy(src_fn, optarg);
 521                         } else if (!x_opt_seen) {
 522                                 /* Include files */
 523                                 llist_add_to(&zaccept, optarg);
 524                         } else {
 525                                 /* Exclude files */
 526                                 llist_add_to(&zreject, optarg);
 527                         }
 528                         break;
 529
 530                 default:
 531                         bb_show_usage();
 532                 }
 533         }
 534
 535 #ifndef __GLIBC__
 536         /*
 537          * This code is needed for non-GNU getopt
 538          * which doesn't understand "-" in option string.
 539          * The -x option won't work properly in this case:
 540          * "unzip a.zip q -x w e" will be interpreted as
 541          * "unzip a.zip q w e -x" = "unzip a.zip q w e"
 542          */
 543         argv += optind;
 544         if (argv[0]) {
 545                 /* +5: space for ".zip" and NUL */
 546                 src_fn = xmalloc(strlen(argv[0]) + 5);
 547                 strcpy(src_fn, argv[0]);
 548                 while (*++argv)
 549                         llist_add_to(&zaccept, *argv);
 550         }
 551 #endif
 552
 553         if (!src_fn) {
 554                 bb_show_usage();
 555         }
 556
 557         /* Open input file */
 558         if (LONE_DASH(src_fn)) {
 559                 xdup2(STDIN_FILENO, zip_fd);
 560                 /* Cannot use prompt mode since zip data is arriving on STDIN */
 561                 if (overwrite == O_PROMPT)
 562                         overwrite = O_NEVER;
 563         } else {
 564                 static const char extn[][5] ALIGN1 = { ".zip", ".ZIP" };
 565                 char *ext = src_fn + strlen(src_fn);
 566                 int src_fd;
 567
 568                 i = 0;
 569                 for (;;) {
 570                         src_fd = open(src_fn, O_RDONLY);
 571                         if (src_fd >= 0)
 572                                 break;
 573                         if (++i > 2) {
 574                                 *ext = '\0';
 575                                 bb_error_msg_and_die("can't open %s[.zip]", src_fn);
 576                         }
 577                         strcpy(ext, extn[i - 1]);
 578                 }
 579                 xmove_fd(src_fd, zip_fd);
 580         }
 581
 582         /* Change dir if necessary */
 583         if (base_dir)
 584                 xchdir(base_dir);
 585
 586         if (quiet <= 1) { /* not -qq */
 587                 if (quiet == 0)
 588                         printf("Archive:  %s\n", src_fn);
 589                 if (listing) {
 590                         puts(verbose ?
 591                                 " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 592                                 "--------  ------  ------- ---- ---------- ----- --------  ----"
 593                                 :
 594                                 "  Length      Date    Time    Name\n"
 595                                 "---------  ---------- -----   ----"
 596                                 );
 597                 }
 598         }
 599
 600 /* Example of an archive with one 0-byte long file named 'z'
 601  * created by Zip 2.31 on Unix:
 602  * 0000 [50 4b]03 04 0a 00 00 00 00 00 42 1a b8 3c 00 00 |PK........B..<..|
 603  *       sig........ vneed flags compr mtime mdate crc32>
 604  * 0010  00 00 00 00 00 00 00 00 00 00 01 00 15 00 7a 55 |..............zU|
 605  *      >..... csize...... usize...... fnlen exlen fn ex>
 606  * 0020  54 09 00 03 cc d3 f9 4b cc d3 f9 4b 55 78 04 00 |T......K...KUx..|
 607  *      >tra_field......................................
 608  * 0030  00 00 00 00[50 4b]01 02 17 03 0a 00 00 00 00 00 |....PK..........|
 609  *       ........... sig........ vmade vneed flags compr
 610  * 0040  42 1a b8 3c 00 00 00 00 00 00 00 00 00 00 00 00 |B..<............|
 611  *       mtime mdate crc32...... csize...... usize......
 612  * 0050  01 00 0d 00 00 00 00 00 00 00 00 00 a4 81 00 00 |................|
 613  *       fnlen exlen clen. dnum. iattr eattr...... relofs> (eattr = rw-r--r--)
 614  * 0060  00 00 7a 55 54 05 00 03 cc d3 f9 4b 55 78 00 00 |..zUT......KUx..|
 615  *      >..... fn extra_field...........................
 616  * 0070 [50 4b]05 06 00 00 00 00 01 00 01 00 3c 00 00 00 |PK..........<...|
 617  * 0080  34 00 00 00 00 00                               |4.....|
 618  */
 619         total_usize = 0;
 620         total_size = 0;
 621         total_entries = 0;
 622         cdf_offset = find_cdf_offset(); /* try to seek to the end, find CDE and CDF start */
 623         while (1) {
 624                 zip_header_t zip;
 625                 mode_t dir_mode = 0777;
 626 #if ENABLE_FEATURE_UNZIP_CDF
 627                 mode_t file_mode = 0666;
 628 #endif
 629
 630                 if (!ENABLE_FEATURE_UNZIP_CDF || cdf_offset == BAD_CDF_OFFSET) {
 631                         /* Normally happens when input is unseekable.
 632                          *
 633                          * Valid ZIP file has Central Directory at the end
 634                          * with central directory file headers (CDFs).
 635                          * After it, there is a Central Directory End structure.
 636                          * CDFs identify what files are in the ZIP and where
 637                          * they are located. This allows ZIP readers to load
 638                          * the list of files without reading the entire ZIP archive.
 639                          * ZIP files may be appended to, only files specified in
 640                          * the CD are valid. Scanning for local file headers is
 641                          * not a correct algorithm.
 642                          *
 643                          * We try to do the above, and resort to "linear" reading
 644                          * of ZIP file only if seek failed or CDE wasn't found.
 645                          */
 646                         uint32_t magic;
 647
 648                         /* Check magic number */
 649                         xread(zip_fd, &magic, 4);
 650                         /* CDF item? Assume there are no more files, exit */
 651                         if (magic == ZIP_CDF_MAGIC) {
 652                                 dbg("got ZIP_CDF_MAGIC");
 653                                 break;
 654                         }
 655                         /* Data descriptor? It was a streaming file, go on */
 656                         if (magic == ZIP_DD_MAGIC) {
 657                                 dbg("got ZIP_DD_MAGIC");
 658                                 /* skip over duplicate crc32, cmpsize and ucmpsize */
 659                                 unzip_skip(3 * 4);
 660                                 continue;
 661                         }
 662                         if (magic != ZIP_FILEHEADER_MAGIC)
 663                                 bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
 664                         dbg("got ZIP_FILEHEADER_MAGIC");
 665
 666                         xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 667                         FIX_ENDIANNESS_ZIP(zip);
 668                         if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 669                                 bb_error_msg_and_die("zip flag %s is not supported",
 670                                         "8 (streaming)");
 671                         }
 672                 }
 673 #if ENABLE_FEATURE_UNZIP_CDF
 674                 else {
 675                         /* cdf_offset is valid (and we know the file is seekable) */
 676                         cdf_header_t cdf;
 677                         cdf_offset = read_next_cdf(cdf_offset, &cdf);
 678                         if (cdf_offset == 0) /* EOF? */
 679                                 break;
 680 # if 1
 681                         xlseek(zip_fd,
 682                                 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4,
 683                                 SEEK_SET);
 684                         xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 685                         FIX_ENDIANNESS_ZIP(zip);
 686                         if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 687                                 /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
 688                                  * only from Central Directory.
 689                                  */
 690                                 zip.fmt.crc32    = cdf.fmt.crc32;
 691                                 zip.fmt.cmpsize  = cdf.fmt.cmpsize;
 692                                 zip.fmt.ucmpsize = cdf.fmt.ucmpsize;
 693                         }
 694 // Seen in some zipfiles: central directory 9 byte extra field contains
 695 // a subfield with ID 0x5455 and 5 data bytes, which is a Unix-style UTC mtime.
 696 // Local header version:
 697 //  u16 0x5455 ("UT")
 698 //  u16 size (1 + 4 * n)
 699 //  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 700 //  u32 mtime
 701 //  u32 atime
 702 //  u32 ctime
 703 // Central header version:
 704 //  u16 0x5455 ("UT")
 705 //  u16 size (5 (or 1?))
 706 //  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 707 //  u32 mtime (CDF does not store atime/ctime)
 708 # else
 709                         /* CDF has the same data as local header, no need to read the latter...
 710                          * ...not really. An archive was seen with cdf.extra_len == 6 but
 711                          * zip.extra_len == 0.
 712                          */
 713                         memcpy(&zip.fmt.version,
 714                                 &cdf.fmt.version_needed, ZIP_HEADER_LEN);
 715                         xlseek(zip_fd,
 716                                 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4 + ZIP_HEADER_LEN,
 717                                 SEEK_SET);
 718 # endif
 719                         if ((cdf.fmt.version_made_by >> 8) == 3) {
 720                                 /* This archive is created on Unix */
 721                                 dir_mode = file_mode = (cdf.fmt.external_attributes >> 16);
 722 //TODO: if (S_ISLNK(file_mode)) this is a symlink
 723                         }
 724                 }
 725 #endif
 726
 727                 if (zip.fmt.zip_flags & SWAP_LE16(0x0001)) {
 728                         /* 0x0001 - encrypted */
 729                         bb_error_msg_and_die("zip flag %s is not supported",
 730                                         "1 (encryption)");
 731                 }
 732                 dbg("File cmpsize:0x%x extra_len:0x%x ucmpsize:0x%x",
 733                         (unsigned)zip.fmt.cmpsize,
 734                         (unsigned)zip.fmt.extra_len,
 735                         (unsigned)zip.fmt.ucmpsize
 736                 );
 737
 738                 /* Read filename */
 739                 free(dst_fn);
 740                 dst_fn = xzalloc(zip.fmt.filename_len + 1);
 741                 xread(zip_fd, dst_fn, zip.fmt.filename_len);
 742
 743                 /* Skip extra header bytes */
 744                 unzip_skip(zip.fmt.extra_len);
 745
 746                 /* Guard against "/abspath", "/../" and similar attacks */
 747                 overlapping_strcpy(dst_fn, strip_unsafe_prefix(dst_fn));
 748
 749                 /* Filter zip entries */
 750                 if (find_list_entry(zreject, dst_fn)
 751                  || (zaccept && !find_list_entry(zaccept, dst_fn))
 752                 ) { /* Skip entry */
 753                         goto skip_cmpsize;
 754                 }
 755
 756                 if (listing) {
 757                         /* List entry */
 758                         char dtbuf[sizeof("mm-dd-yyyy hh:mm")];
 759                         sprintf(dtbuf, "%02u-%02u-%04u %02u:%02u",
 760                                 (zip.fmt.moddate >> 5) & 0xf,  // mm: 0x01e0
 761                                 (zip.fmt.moddate)      & 0x1f, // dd: 0x001f
 762                                 (zip.fmt.moddate >> 9) + 1980, // yy: 0xfe00
 763                                 (zip.fmt.modtime >> 11),       // hh: 0xf800
 764                                 (zip.fmt.modtime >> 5) & 0x3f  // mm: 0x07e0
 765                                 // seconds/2 not shown, encoded in -- 0x001f
 766                         );
 767                         if (!verbose) {
 768                                 //      "  Length      Date    Time    Name\n"
 769                                 //      "---------  ---------- -----   ----"
 770                                 printf(       "%9u  " "%s   "         "%s\n",
 771                                         (unsigned)zip.fmt.ucmpsize,
 772                                         dtbuf,
 773                                         dst_fn);
 774                         } else {
 775                                 char method6[7];
 776                                 unsigned long percents;
 777
 778                                 sprintf(method6, "%6u", zip.fmt.method);
 779                                 if (zip.fmt.method == 0) {
 780                                         strcpy(method6, "Stored");
 781                                 }
 782                                 if (zip.fmt.method == 8) {
 783                                         strcpy(method6, "Defl:N");
 784                                         /* normal, maximum, fast, superfast */
 785                                         IF_DESKTOP(method6[5] = "NXFS"[(zip.fmt.zip_flags >> 1) & 3];)
 786                                 }
 787                                 percents = zip.fmt.ucmpsize - zip.fmt.cmpsize;
 788                                 if ((int32_t)percents < 0)
 789                                         percents = 0; /* happens if ucmpsize < cmpsize */
 790                                 percents = percents * 100;
 791                                 if (zip.fmt.ucmpsize)
 792                                         percents /= zip.fmt.ucmpsize;
 793                                 //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 794                                 //      "--------  ------  ------- ---- ---------- ----- --------  ----"
 795                                 printf(      "%8u  %s"        "%9u%4u%% " "%s "         "%08x  "  "%s\n",
 796                                         (unsigned)zip.fmt.ucmpsize,
 797                                         method6,
 798                                         (unsigned)zip.fmt.cmpsize,
 799                                         (unsigned)percents,
 800                                         dtbuf,
 801                                         zip.fmt.crc32,
 802                                         dst_fn);
 803                                 total_size += zip.fmt.cmpsize;
 804                         }
 805                         total_usize += zip.fmt.ucmpsize;
 806                         goto skip_cmpsize;
 807                 }
 808
 809                 if (dst_fd == STDOUT_FILENO) {
 810                         /* Extracting to STDOUT */
 811                         goto do_extract;
 812                 }
 813                 if (last_char_is(dst_fn, '/')) {
 814                         /* Extract directory */
 815                         if (stat(dst_fn, &stat_buf) == -1) {
 816                                 if (errno != ENOENT) {
 817                                         bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 818                                 }
 819                                 if (!quiet) {
 820                                         printf("   creating: %s\n", dst_fn);
 821                                 }
 822                                 unzip_create_leading_dirs(dst_fn);
 823                                 if (bb_make_directory(dst_fn, dir_mode, FILEUTILS_IGNORE_CHMOD_ERR)) {
 824                                         xfunc_die();
 825                                 }
 826                         } else {
 827                                 if (!S_ISDIR(stat_buf.st_mode)) {
 828                                         bb_error_msg_and_die("'%s' exists but is not a %s",
 829                                                 dst_fn, "directory");
 830                                 }
 831                         }
 832                         goto skip_cmpsize;
 833                 }
 834  check_file:
 835                 /* Extract file */
 836                 if (stat(dst_fn, &stat_buf) == -1) {
 837                         /* File does not exist */
 838                         if (errno != ENOENT) {
 839                                 bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 840                         }
 841                         goto do_open_and_extract;
 842                 }
 843                 /* File already exists */
 844                 if (overwrite == O_NEVER) {
 845                         goto skip_cmpsize;
 846                 }
 847                 if (!S_ISREG(stat_buf.st_mode)) {
 848                         /* File is not regular file */
 849                         bb_error_msg_and_die("'%s' exists but is not a %s",
 850                                 dst_fn, "regular file");
 851                 }
 852                 /* File is regular file */
 853                 if (overwrite == O_ALWAYS)
 854                         goto do_open_and_extract;
 855                 printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
 856                 my_fgets80(key_buf);
 857
 858                 switch (key_buf[0]) {
 859                 case 'A':
 860                         overwrite = O_ALWAYS;
 861                 case 'y': /* Open file and fall into unzip */
 862  do_open_and_extract:
 863                         unzip_create_leading_dirs(dst_fn);
 864 #if ENABLE_FEATURE_UNZIP_CDF
 865                         dst_fd = xopen3(dst_fn, O_WRONLY | O_CREAT | O_TRUNC, file_mode);
 866 #else
 867                         dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC);
 868 #endif
 869  do_extract:
 870                         if (!quiet) {
 871                                 printf(/* zip.fmt.method == 0
 872                                         ? " extracting: %s\n"
 873                                         : */ "  inflating: %s\n", dst_fn);
 874                         }
 875                         unzip_extract(&zip, dst_fd);
 876                         if (dst_fd != STDOUT_FILENO) {
 877                                 /* closing STDOUT is potentially bad for future business */
 878                                 close(dst_fd);
 879                         }
 880                         break;
 881
 882                 case 'N':
 883                         overwrite = O_NEVER;
 884                 case 'n': /* Skip entry data */
 885  skip_cmpsize:
 886                         unzip_skip(zip.fmt.cmpsize);
 887                         break;
 888
 889                 case 'r':
 890                         /* Prompt for new name */
 891                         printf("new name: ");
 892                         my_fgets80(key_buf);
 893                         free(dst_fn);
 894                         dst_fn = xstrdup(key_buf);
 895                         chomp(dst_fn);
 896                         goto check_file;
 897
 898                 default:
 899                         printf("error: invalid response [%c]\n", (char)key_buf[0]);
 900                         goto check_file;
 901                 }
 902
 903                 total_entries++;
 904         }
 905
 906         if (listing && quiet <= 1) {
 907                 if (!verbose) {
 908                         //      "  Length      Date    Time    Name\n"
 909                         //      "---------  ---------- -----   ----"
 910                         printf( " --------%21s"               "-------\n"
 911                                      "%9lu%21s"               "%u files\n",
 912                                 "",
 913                                 total_usize, "", total_entries);
 914                 } else {
 915                         unsigned long percents = total_usize - total_size;
 916                         if ((long)percents < 0)
 917                                 percents = 0; /* happens if usize < size */
 918                         percents = percents * 100;
 919                         if (total_usize)
 920                                 percents /= total_usize;
 921                         //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 922                         //      "--------  ------  ------- ---- ---------- ----- --------  ----"
 923                         printf( "--------          ------- ----%28s"                      "----\n"
 924                                 "%8lu"              "%17lu%4u%%%28s"                      "%u files\n",
 925                                 "",
 926                                 total_usize, total_size, (unsigned)percents, "",
 927                                 total_entries);
 928                 }
 929         }
 930
 931         return 0;
 932 }