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
  38 //applet:IF_UNZIP(APPLET(unzip, BB_DIR_USR_BIN, BB_SUID_DROP))
  39 //kbuild:lib-$(CONFIG_UNZIP) += unzip.o
  40
  41 //usage:#define unzip_trivial_usage
  42 //usage:       "[-lnopq] FILE[.zip] [FILE]... [-x FILE...] [-d DIR]"
  43 //usage:#define unzip_full_usage "\n\n"
  44 //usage:       "Extract FILEs from ZIP archive\n"
  45 //usage:     "\n        -l      List contents (with -q for short form)"
  46 //usage:     "\n        -n      Never overwrite files (default: ask)"
  47 //usage:     "\n        -o      Overwrite"
  48 //usage:     "\n        -p      Print to stdout"
  49 //usage:     "\n        -q      Quiet"
  50 //usage:     "\n        -x FILE Exclude FILEs"
  51 //usage:     "\n        -d DIR  Extract into DIR"
  52
  53 #include "libbb.h"
  54 #include "bb_archive.h"
  55
  56 #if 0
  57 # define dbg(...) bb_error_msg(__VA_ARGS__)
  58 #else
  59 # define dbg(...) ((void)0)
  60 #endif
  61
  62 enum {
  63 #if BB_BIG_ENDIAN
  64         ZIP_FILEHEADER_MAGIC = 0x504b0304,
  65         ZIP_CDF_MAGIC        = 0x504b0102, /* central directory's file header */
  66         ZIP_CDE_MAGIC        = 0x504b0506, /* "end of central directory" record */
  67         ZIP_DD_MAGIC         = 0x504b0708,
  68 #else
  69         ZIP_FILEHEADER_MAGIC = 0x04034b50,
  70         ZIP_CDF_MAGIC        = 0x02014b50,
  71         ZIP_CDE_MAGIC        = 0x06054b50,
  72         ZIP_DD_MAGIC         = 0x08074b50,
  73 #endif
  74 };
  75
  76 #define ZIP_HEADER_LEN 26
  77
  78 typedef union {
  79         uint8_t raw[ZIP_HEADER_LEN];
  80         struct {
  81                 uint16_t version;               /* 0-1 */
  82                 uint16_t zip_flags;             /* 2-3 */
  83                 uint16_t method;                /* 4-5 */
  84                 uint16_t modtime;               /* 6-7 */
  85                 uint16_t moddate;               /* 8-9 */
  86                 uint32_t crc32 PACKED;          /* 10-13 */
  87                 uint32_t cmpsize PACKED;        /* 14-17 */
  88                 uint32_t ucmpsize PACKED;       /* 18-21 */
  89                 uint16_t filename_len;          /* 22-23 */
  90                 uint16_t extra_len;             /* 24-25 */
  91                 /* filename follows (not NUL terminated) */
  92                 /* extra field follows */
  93                 /* data follows */
  94         } formatted PACKED;
  95 } zip_header_t; /* PACKED - gcc 4.2.1 doesn't like it (spews warning) */
  96
  97 #define FIX_ENDIANNESS_ZIP(zip_header) \
  98 do { if (BB_BIG_ENDIAN) { \
  99         (zip_header).formatted.crc32        = SWAP_LE32((zip_header).formatted.crc32       ); \
 100         (zip_header).formatted.cmpsize      = SWAP_LE32((zip_header).formatted.cmpsize     ); \
 101         (zip_header).formatted.ucmpsize     = SWAP_LE32((zip_header).formatted.ucmpsize    ); \
 102         (zip_header).formatted.filename_len = SWAP_LE16((zip_header).formatted.filename_len); \
 103         (zip_header).formatted.extra_len    = SWAP_LE16((zip_header).formatted.extra_len   ); \
 104 }} while (0)
 105
 106 #define CDF_HEADER_LEN 42
 107
 108 typedef union {
 109         uint8_t raw[CDF_HEADER_LEN];
 110         struct {
 111                 /* uint32_t signature; 50 4b 01 02 */
 112                 uint16_t version_made_by;       /* 0-1 */
 113                 uint16_t version_needed;        /* 2-3 */
 114                 uint16_t cdf_flags;             /* 4-5 */
 115                 uint16_t method;                /* 6-7 */
 116                 uint16_t modtime;               /* 8-9 */
 117                 uint16_t moddate;               /* 10-11 */
 118                 uint32_t crc32;                 /* 12-15 */
 119                 uint32_t cmpsize;               /* 16-19 */
 120                 uint32_t ucmpsize;              /* 20-23 */
 121                 uint16_t file_name_length;      /* 24-25 */
 122                 uint16_t extra_field_length;    /* 26-27 */
 123                 uint16_t file_comment_length;   /* 28-29 */
 124                 uint16_t disk_number_start;     /* 30-31 */
 125                 uint16_t internal_file_attributes; /* 32-33 */
 126                 uint32_t external_file_attributes PACKED; /* 34-37 */
 127                 uint32_t relative_offset_of_local_header PACKED; /* 38-41 */
 128                 /* filename follows (not NUL terminated) */
 129                 /* extra field follows */
 130                 /* comment follows */
 131         } formatted PACKED;
 132 } cdf_header_t;
 133
 134 #define FIX_ENDIANNESS_CDF(cdf_header) \
 135 do { if (BB_BIG_ENDIAN) { \
 136         (cdf_header).formatted.version_made_by = SWAP_LE16((cdf_header).formatted.version_made_by); \
 137         (cdf_header).formatted.version_needed = SWAP_LE16((cdf_header).formatted.version_needed); \
 138         (cdf_header).formatted.method       = SWAP_LE16((cdf_header).formatted.method      ); \
 139         (cdf_header).formatted.modtime      = SWAP_LE16((cdf_header).formatted.modtime     ); \
 140         (cdf_header).formatted.moddate      = SWAP_LE16((cdf_header).formatted.moddate     ); \
 141         (cdf_header).formatted.crc32        = SWAP_LE32((cdf_header).formatted.crc32       ); \
 142         (cdf_header).formatted.cmpsize      = SWAP_LE32((cdf_header).formatted.cmpsize     ); \
 143         (cdf_header).formatted.ucmpsize     = SWAP_LE32((cdf_header).formatted.ucmpsize    ); \
 144         (cdf_header).formatted.file_name_length = SWAP_LE16((cdf_header).formatted.file_name_length); \
 145         (cdf_header).formatted.extra_field_length = SWAP_LE16((cdf_header).formatted.extra_field_length); \
 146         (cdf_header).formatted.file_comment_length = SWAP_LE16((cdf_header).formatted.file_comment_length); \
 147         (cdf_header).formatted.external_file_attributes = SWAP_LE32((cdf_header).formatted.external_file_attributes); \
 148 }} while (0)
 149
 150 #define CDE_HEADER_LEN 16
 151
 152 typedef union {
 153         uint8_t raw[CDE_HEADER_LEN];
 154         struct {
 155                 /* uint32_t signature; 50 4b 05 06 */
 156                 uint16_t this_disk_no;
 157                 uint16_t disk_with_cdf_no;
 158                 uint16_t cdf_entries_on_this_disk;
 159                 uint16_t cdf_entries_total;
 160                 uint32_t cdf_size;
 161                 uint32_t cdf_offset;
 162                 /* uint16_t file_comment_length; */
 163                 /* .ZIP file comment (variable size) */
 164         } formatted PACKED;
 165 } cde_header_t;
 166
 167 #define FIX_ENDIANNESS_CDE(cde_header) \
 168 do { if (BB_BIG_ENDIAN) { \
 169         (cde_header).formatted.cdf_offset = SWAP_LE32((cde_header).formatted.cdf_offset); \
 170 }} while (0)
 171
 172 struct BUG {
 173         /* Check the offset of the last element, not the length.  This leniency
 174          * allows for poor packing, whereby the overall struct may be too long,
 175          * even though the elements are all in the right place.
 176          */
 177         char BUG_zip_header_must_be_26_bytes[
 178                 offsetof(zip_header_t, formatted.extra_len) + 2
 179                         == ZIP_HEADER_LEN ? 1 : -1];
 180         char BUG_cdf_header_must_be_42_bytes[
 181                 offsetof(cdf_header_t, formatted.relative_offset_of_local_header) + 4
 182                         == CDF_HEADER_LEN ? 1 : -1];
 183         char BUG_cde_header_must_be_16_bytes[
 184                 sizeof(cde_header_t) == CDE_HEADER_LEN ? 1 : -1];
 185 };
 186
 187
 188 enum { zip_fd = 3 };
 189
 190
 191 /* This value means that we failed to find CDF */
 192 #define BAD_CDF_OFFSET ((uint32_t)0xffffffff)
 193
 194 #if !ENABLE_FEATURE_UNZIP_CDF
 195
 196 # define find_cdf_offset() BAD_CDF_OFFSET
 197
 198 #else
 199 /* Seen in the wild:
 200  * Self-extracting PRO2K3XP_32.exe contains 19078464 byte zip archive,
 201  * where CDE was nearly 48 kbytes before EOF.
 202  * (Surprisingly, it also apparently has *another* CDE structure
 203  * closer to the end, with bogus cdf_offset).
 204  * To make extraction work, bumped PEEK_FROM_END from 16k to 64k.
 205  */
 206 #define PEEK_FROM_END (64*1024)
 207 /* NB: does not preserve file position! */
 208 static uint32_t find_cdf_offset(void)
 209 {
 210         cde_header_t cde_header;
 211         unsigned char *buf;
 212         unsigned char *p;
 213         off_t end;
 214         uint32_t found;
 215
 216         end = lseek(zip_fd, 0, SEEK_END);
 217         if (end == (off_t) -1)
 218                 return BAD_CDF_OFFSET;
 219
 220         end -= PEEK_FROM_END;
 221         if (end < 0)
 222                 end = 0;
 223
 224         dbg("Looking for cdf_offset starting from 0x%"OFF_FMT"x", end);
 225         xlseek(zip_fd, end, SEEK_SET);
 226         buf = xzalloc(PEEK_FROM_END);
 227         full_read(zip_fd, buf, PEEK_FROM_END);
 228
 229         found = BAD_CDF_OFFSET;
 230         p = buf;
 231         while (p <= buf + PEEK_FROM_END - CDE_HEADER_LEN - 4) {
 232                 if (*p != 'P') {
 233                         p++;
 234                         continue;
 235                 }
 236                 if (*++p != 'K')
 237                         continue;
 238                 if (*++p != 5)
 239                         continue;
 240                 if (*++p != 6)
 241                         continue;
 242                 /* we found CDE! */
 243                 memcpy(cde_header.raw, p + 1, CDE_HEADER_LEN);
 244                 FIX_ENDIANNESS_CDE(cde_header);
 245                 /*
 246                  * I've seen .ZIP files with seemingly valid CDEs
 247                  * where cdf_offset points past EOF - ??
 248                  * This check ignores such CDEs:
 249                  */
 250                 if (cde_header.formatted.cdf_offset < end + (p - buf)) {
 251                         found = cde_header.formatted.cdf_offset;
 252                         dbg("Possible cdf_offset:0x%x at 0x%"OFF_FMT"x",
 253                                 (unsigned)found, end + (p-3 - buf));
 254                         dbg("  cdf_offset+cdf_size:0x%x",
 255                                 (unsigned)(found + SWAP_LE32(cde_header.formatted.cdf_size)));
 256                         /*
 257                          * We do not "break" here because only the last CDE is valid.
 258                          * I've seen a .zip archive which contained a .zip file,
 259                          * uncompressed, and taking the first CDE was using
 260                          * the CDE inside that file!
 261                          */
 262                 }
 263         }
 264         free(buf);
 265         dbg("Found cdf_offset:0x%x", (unsigned)found);
 266         return found;
 267 };
 268
 269 static uint32_t read_next_cdf(uint32_t cdf_offset, cdf_header_t *cdf_ptr)
 270 {
 271         uint32_t magic;
 272
 273         if (cdf_offset == BAD_CDF_OFFSET)
 274                 return cdf_offset;
 275
 276         dbg("Reading CDF at 0x%x", (unsigned)cdf_offset);
 277         xlseek(zip_fd, cdf_offset, SEEK_SET);
 278         xread(zip_fd, &magic, 4);
 279         /* Central Directory End? */
 280         if (magic == ZIP_CDE_MAGIC) {
 281                 dbg("got ZIP_CDE_MAGIC");
 282                 return 0; /* EOF */
 283         }
 284         xread(zip_fd, cdf_ptr->raw, CDF_HEADER_LEN);
 285
 286         FIX_ENDIANNESS_CDF(*cdf_ptr);
 287         dbg("  file_name_length:%u extra_field_length:%u file_comment_length:%u",
 288                 (unsigned)cdf_ptr->formatted.file_name_length,
 289                 (unsigned)cdf_ptr->formatted.extra_field_length,
 290                 (unsigned)cdf_ptr->formatted.file_comment_length
 291         );
 292         cdf_offset += 4 + CDF_HEADER_LEN
 293                 + cdf_ptr->formatted.file_name_length
 294                 + cdf_ptr->formatted.extra_field_length
 295                 + cdf_ptr->formatted.file_comment_length;
 296
 297         return cdf_offset;
 298 };
 299 #endif
 300
 301 static void unzip_skip(off_t skip)
 302 {
 303         if (skip != 0)
 304                 if (lseek(zip_fd, skip, SEEK_CUR) == (off_t)-1)
 305                         bb_copyfd_exact_size(zip_fd, -1, skip);
 306 }
 307
 308 static void unzip_create_leading_dirs(const char *fn)
 309 {
 310         /* Create all leading directories */
 311         char *name = xstrdup(fn);
 312         if (bb_make_directory(dirname(name), 0777, FILEUTILS_RECUR)) {
 313                 xfunc_die(); /* bb_make_directory is noisy */
 314         }
 315         free(name);
 316 }
 317
 318 static void unzip_extract(zip_header_t *zip_header, int dst_fd)
 319 {
 320         if (zip_header->formatted.method == 0) {
 321                 /* Method 0 - stored (not compressed) */
 322                 off_t size = zip_header->formatted.ucmpsize;
 323                 if (size)
 324                         bb_copyfd_exact_size(zip_fd, dst_fd, size);
 325         } else {
 326                 /* Method 8 - inflate */
 327                 transformer_state_t xstate;
 328                 init_transformer_state(&xstate);
 329                 xstate.bytes_in = zip_header->formatted.cmpsize;
 330                 xstate.src_fd = zip_fd;
 331                 xstate.dst_fd = dst_fd;
 332                 if (inflate_unzip(&xstate) < 0)
 333                         bb_error_msg_and_die("inflate error");
 334                 /* Validate decompression - crc */
 335                 if (zip_header->formatted.crc32 != (xstate.crc32 ^ 0xffffffffL)) {
 336                         bb_error_msg_and_die("crc error");
 337                 }
 338                 /* Validate decompression - size */
 339                 if (zip_header->formatted.ucmpsize != xstate.bytes_out) {
 340                         /* Don't die. Who knows, maybe len calculation
 341                          * was botched somewhere. After all, crc matched! */
 342                         bb_error_msg("bad length");
 343                 }
 344         }
 345         /* TODO? method 12: bzip2, method 14: LZMA */
 346 }
 347
 348 static void my_fgets80(char *buf80)
 349 {
 350         fflush_all();
 351         if (!fgets(buf80, 80, stdin)) {
 352                 bb_perror_msg_and_die("can't read standard input");
 353         }
 354 }
 355
 356 int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 357 int unzip_main(int argc, char **argv)
 358 {
 359         enum { O_PROMPT, O_NEVER, O_ALWAYS };
 360
 361         smallint quiet = 0;
 362         IF_NOT_FEATURE_UNZIP_CDF(const) smallint verbose = 0;
 363         smallint listing = 0;
 364         smallint overwrite = O_PROMPT;
 365         smallint x_opt_seen;
 366         uint32_t cdf_offset;
 367         unsigned long total_usize;
 368         unsigned long total_size;
 369         unsigned total_entries;
 370         int dst_fd = -1;
 371         char *src_fn = NULL;
 372         char *dst_fn = NULL;
 373         llist_t *zaccept = NULL;
 374         llist_t *zreject = NULL;
 375         char *base_dir = NULL;
 376         int i, opt;
 377         char key_buf[80]; /* must match size used by my_fgets80 */
 378         struct stat stat_buf;
 379
 380 /* -q, -l and -v: UnZip 5.52 of 28 February 2005, by Info-ZIP:
 381  *
 382  * # /usr/bin/unzip -qq -v decompress_unlzma.i.zip
 383  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 384  * # /usr/bin/unzip -q -v decompress_unlzma.i.zip
 385  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 386  * --------  ------  ------- -----   ----   ----   ------    ----
 387  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 388  * --------          -------  ---                            -------
 389  *   204372            35278  83%                            1 file
 390  * # /usr/bin/unzip -v decompress_unlzma.i.zip
 391  * Archive:  decompress_unlzma.i.zip
 392  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 393  * --------  ------  ------- -----   ----   ----   ------    ----
 394  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 395  * --------          -------  ---                            -------
 396  *   204372            35278  83%                            1 file
 397  * # unzip -v decompress_unlzma.i.zip
 398  * Archive:  decompress_unlzma.i.zip
 399  *   Length     Date   Time    Name
 400  *  --------    ----   ----    ----
 401  *    204372  09-06-09 14:23   decompress_unlzma.i
 402  *  --------                   -------
 403  *    204372                   1 files
 404  * # /usr/bin/unzip -l -qq decompress_unlzma.i.zip
 405  *    204372  09-06-09 14:23   decompress_unlzma.i
 406  * # /usr/bin/unzip -l -q decompress_unlzma.i.zip
 407  *   Length     Date   Time    Name
 408  *  --------    ----   ----    ----
 409  *    204372  09-06-09 14:23   decompress_unlzma.i
 410  *  --------                   -------
 411  *    204372                   1 file
 412  * # /usr/bin/unzip -l decompress_unlzma.i.zip
 413  * Archive:  decompress_unlzma.i.zip
 414  *   Length     Date   Time    Name
 415  *  --------    ----   ----    ----
 416  *    204372  09-06-09 14:23   decompress_unlzma.i
 417  *  --------                   -------
 418  *    204372                   1 file
 419  */
 420
 421         x_opt_seen = 0;
 422         /* '-' makes getopt return 1 for non-options */
 423         while ((opt = getopt(argc, argv, "-d:lnopqxv")) != -1) {
 424                 switch (opt) {
 425                 case 'd':  /* Extract to base directory */
 426                         base_dir = optarg;
 427                         break;
 428
 429                 case 'l': /* List */
 430                         listing = 1;
 431                         break;
 432
 433                 case 'n': /* Never overwrite existing files */
 434                         overwrite = O_NEVER;
 435                         break;
 436
 437                 case 'o': /* Always overwrite existing files */
 438                         overwrite = O_ALWAYS;
 439                         break;
 440
 441                 case 'p': /* Extract files to stdout and fall through to set verbosity */
 442                         dst_fd = STDOUT_FILENO;
 443
 444                 case 'q': /* Be quiet */
 445                         quiet++;
 446                         break;
 447
 448                 case 'v': /* Verbose list */
 449                         IF_FEATURE_UNZIP_CDF(verbose++;)
 450                         listing = 1;
 451                         break;
 452
 453                 case 'x':
 454                         x_opt_seen = 1;
 455                         break;
 456
 457                 case 1:
 458                         if (!src_fn) {
 459                                 /* The zip file */
 460                                 /* +5: space for ".zip" and NUL */
 461                                 src_fn = xmalloc(strlen(optarg) + 5);
 462                                 strcpy(src_fn, optarg);
 463                         } else if (!x_opt_seen) {
 464                                 /* Include files */
 465                                 llist_add_to(&zaccept, optarg);
 466                         } else {
 467                                 /* Exclude files */
 468                                 llist_add_to(&zreject, optarg);
 469                         }
 470                         break;
 471
 472                 default:
 473                         bb_show_usage();
 474                 }
 475         }
 476
 477 #ifndef __GLIBC__
 478         /*
 479          * This code is needed for non-GNU getopt
 480          * which doesn't understand "-" in option string.
 481          * The -x option won't work properly in this case:
 482          * "unzip a.zip q -x w e" will be interpreted as
 483          * "unzip a.zip q w e -x" = "unzip a.zip q w e"
 484          */
 485         argv += optind;
 486         if (argv[0]) {
 487                 /* +5: space for ".zip" and NUL */
 488                 src_fn = xmalloc(strlen(argv[0]) + 5);
 489                 strcpy(src_fn, argv[0]);
 490                 while (*++argv)
 491                         llist_add_to(&zaccept, *argv);
 492         }
 493 #endif
 494
 495         if (!src_fn) {
 496                 bb_show_usage();
 497         }
 498
 499         /* Open input file */
 500         if (LONE_DASH(src_fn)) {
 501                 xdup2(STDIN_FILENO, zip_fd);
 502                 /* Cannot use prompt mode since zip data is arriving on STDIN */
 503                 if (overwrite == O_PROMPT)
 504                         overwrite = O_NEVER;
 505         } else {
 506                 static const char extn[][5] ALIGN1 = { ".zip", ".ZIP" };
 507                 char *ext = src_fn + strlen(src_fn);
 508                 int src_fd;
 509
 510                 i = 0;
 511                 for (;;) {
 512                         src_fd = open(src_fn, O_RDONLY);
 513                         if (src_fd >= 0)
 514                                 break;
 515                         if (++i > 2) {
 516                                 *ext = '\0';
 517                                 bb_error_msg_and_die("can't open %s[.zip]", src_fn);
 518                         }
 519                         strcpy(ext, extn[i - 1]);
 520                 }
 521                 xmove_fd(src_fd, zip_fd);
 522         }
 523
 524         /* Change dir if necessary */
 525         if (base_dir)
 526                 xchdir(base_dir);
 527
 528         if (quiet <= 1) { /* not -qq */
 529                 if (quiet == 0)
 530                         printf("Archive:  %s\n", src_fn);
 531                 if (listing) {
 532                         puts(verbose ?
 533                                 " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 534                                 "--------  ------  ------- ---- ---------- ----- --------  ----"
 535                                 :
 536                                 "  Length      Date    Time    Name\n"
 537                                 "---------  ---------- -----   ----"
 538                                 );
 539                 }
 540         }
 541
 542 /* Example of an archive with one 0-byte long file named 'z'
 543  * created by Zip 2.31 on Unix:
 544  * 0000 [50 4b]03 04 0a 00 00 00 00 00 42 1a b8 3c 00 00 |PK........B..<..|
 545  *       sig........ vneed flags compr mtime mdate crc32>
 546  * 0010  00 00 00 00 00 00 00 00 00 00 01 00 15 00 7a 55 |..............zU|
 547  *      >..... csize...... usize...... fnlen exlen fn ex>
 548  * 0020  54 09 00 03 cc d3 f9 4b cc d3 f9 4b 55 78 04 00 |T......K...KUx..|
 549  *      >tra_field......................................
 550  * 0030  00 00 00 00[50 4b]01 02 17 03 0a 00 00 00 00 00 |....PK..........|
 551  *       ........... sig........ vmade vneed flags compr
 552  * 0040  42 1a b8 3c 00 00 00 00 00 00 00 00 00 00 00 00 |B..<............|
 553  *       mtime mdate crc32...... csize...... usize......
 554  * 0050  01 00 0d 00 00 00 00 00 00 00 00 00 a4 81 00 00 |................|
 555  *       fnlen exlen clen. dnum. iattr eattr...... relofs> (eattr = rw-r--r--)
 556  * 0060  00 00 7a 55 54 05 00 03 cc d3 f9 4b 55 78 00 00 |..zUT......KUx..|
 557  *      >..... fn extra_field...........................
 558  * 0070 [50 4b]05 06 00 00 00 00 01 00 01 00 3c 00 00 00 |PK..........<...|
 559  * 0080  34 00 00 00 00 00                               |4.....|
 560  */
 561         total_usize = 0;
 562         total_size = 0;
 563         total_entries = 0;
 564         cdf_offset = find_cdf_offset(); /* try to seek to the end, find CDE and CDF start */
 565         while (1) {
 566                 zip_header_t zip_header;
 567                 mode_t dir_mode = 0777;
 568 #if ENABLE_FEATURE_UNZIP_CDF
 569                 mode_t file_mode = 0666;
 570 #endif
 571
 572                 if (!ENABLE_FEATURE_UNZIP_CDF || cdf_offset == BAD_CDF_OFFSET) {
 573                         /* Normally happens when input is unseekable.
 574                          *
 575                          * Valid ZIP file has Central Directory at the end
 576                          * with central directory file headers (CDFs).
 577                          * After it, there is a Central Directory End structure.
 578                          * CDFs identify what files are in the ZIP and where
 579                          * they are located. This allows ZIP readers to load
 580                          * the list of files without reading the entire ZIP archive.
 581                          * ZIP files may be appended to, only files specified in
 582                          * the CD are valid. Scanning for local file headers is
 583                          * not a correct algorithm.
 584                          *
 585                          * We try to do the above, and resort to "linear" reading
 586                          * of ZIP file only if seek failed or CDE wasn't found.
 587                          */
 588                         uint32_t magic;
 589
 590                         /* Check magic number */
 591                         xread(zip_fd, &magic, 4);
 592                         /* Central directory? It's at the end, so exit */
 593                         if (magic == ZIP_CDF_MAGIC) {
 594                                 dbg("got ZIP_CDF_MAGIC");
 595                                 break;
 596                         }
 597                         /* Data descriptor? It was a streaming file, go on */
 598                         if (magic == ZIP_DD_MAGIC) {
 599                                 dbg("got ZIP_DD_MAGIC");
 600                                 /* skip over duplicate crc32, cmpsize and ucmpsize */
 601                                 unzip_skip(3 * 4);
 602                                 continue;
 603                         }
 604                         if (magic != ZIP_FILEHEADER_MAGIC)
 605                                 bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
 606                         dbg("got ZIP_FILEHEADER_MAGIC");
 607
 608                         xread(zip_fd, zip_header.raw, ZIP_HEADER_LEN);
 609                         FIX_ENDIANNESS_ZIP(zip_header);
 610                         if ((zip_header.formatted.method != 0)
 611                          && (zip_header.formatted.method != 8)
 612                         ) {
 613                                 /* TODO? method 12: bzip2, method 14: LZMA */
 614                                 bb_error_msg_and_die("unsupported method %d", zip_header.formatted.method);
 615                         }
 616                         if (zip_header.formatted.zip_flags & SWAP_LE16(0x0009)) {
 617                                 bb_error_msg_and_die("zip flags 1 and 8 are not supported");
 618                         }
 619                 }
 620 #if ENABLE_FEATURE_UNZIP_CDF
 621                 else {
 622                         /* cdf_offset is valid (and we know the file is seekable) */
 623                         cdf_header_t cdf_header;
 624                         cdf_offset = read_next_cdf(cdf_offset, &cdf_header);
 625                         if (cdf_offset == 0) /* EOF? */
 626                                 break;
 627 # if 0
 628                         xlseek(zip_fd,
 629                                 SWAP_LE32(cdf_header.formatted.relative_offset_of_local_header) + 4,
 630                                 SEEK_SET);
 631                         xread(zip_fd, zip_header.raw, ZIP_HEADER_LEN);
 632                         FIX_ENDIANNESS_ZIP(zip_header);
 633                         if (zip_header.formatted.zip_flags & SWAP_LE16(0x0008)) {
 634                                 /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
 635                                  * only from Central Directory.
 636                                  */
 637                                 zip_header.formatted.crc32    = cdf_header.formatted.crc32;
 638                                 zip_header.formatted.cmpsize  = cdf_header.formatted.cmpsize;
 639                                 zip_header.formatted.ucmpsize = cdf_header.formatted.ucmpsize;
 640                         }
 641 # else
 642                         /* CDF has the same data as local header, no need to read the latter */
 643                         memcpy(&zip_header.formatted.version,
 644                                 &cdf_header.formatted.version_needed, ZIP_HEADER_LEN);
 645                         xlseek(zip_fd,
 646                                 SWAP_LE32(cdf_header.formatted.relative_offset_of_local_header) + 4 + ZIP_HEADER_LEN,
 647                                 SEEK_SET);
 648 # endif
 649                         if ((cdf_header.formatted.version_made_by >> 8) == 3) {
 650                                 /* This archive is created on Unix */
 651                                 dir_mode = file_mode = (cdf_header.formatted.external_file_attributes >> 16);
 652                         }
 653                 }
 654 #endif
 655
 656                 if (zip_header.formatted.zip_flags & SWAP_LE16(0x0001)) {
 657                         /* 0x0001 - encrypted */
 658                         bb_error_msg_and_die("zip flag 1 (encryption) is not supported");
 659                 }
 660                 dbg("File cmpsize:0x%x extra_len:0x%x ucmpsize:0x%x",
 661                         (unsigned)zip_header.formatted.cmpsize,
 662                         (unsigned)zip_header.formatted.extra_len,
 663                         (unsigned)zip_header.formatted.ucmpsize
 664                 );
 665
 666                 /* Read filename */
 667                 free(dst_fn);
 668                 dst_fn = xzalloc(zip_header.formatted.filename_len + 1);
 669                 xread(zip_fd, dst_fn, zip_header.formatted.filename_len);
 670
 671                 /* Skip extra header bytes */
 672                 unzip_skip(zip_header.formatted.extra_len);
 673
 674                 /* Guard against "/abspath", "/../" and similar attacks */
 675                 overlapping_strcpy(dst_fn, strip_unsafe_prefix(dst_fn));
 676
 677                 /* Filter zip entries */
 678                 if (find_list_entry(zreject, dst_fn)
 679                  || (zaccept && !find_list_entry(zaccept, dst_fn))
 680                 ) { /* Skip entry */
 681                         i = 'n';
 682                 } else {
 683                         if (listing) {
 684                                 /* List entry */
 685                                 char dtbuf[sizeof("mm-dd-yyyy hh:mm")];
 686                                 sprintf(dtbuf, "%02u-%02u-%04u %02u:%02u",
 687                                         (zip_header.formatted.moddate >> 5) & 0xf,  // mm: 0x01e0
 688                                         (zip_header.formatted.moddate)      & 0x1f, // dd: 0x001f
 689                                         (zip_header.formatted.moddate >> 9) + 1980, // yy: 0xfe00
 690                                         (zip_header.formatted.modtime >> 11),       // hh: 0xf800
 691                                         (zip_header.formatted.modtime >> 5) & 0x3f  // mm: 0x07e0
 692                                         // seconds/2 are not shown, encoded in ----------- 0x001f
 693                                 );
 694                                 if (!verbose) {
 695                                         //      "  Length      Date    Time    Name\n"
 696                                         //      "---------  ---------- -----   ----"
 697                                         printf(       "%9u  " "%s   "         "%s\n",
 698                                                 (unsigned)zip_header.formatted.ucmpsize,
 699                                                 dtbuf,
 700                                                 dst_fn);
 701                                 } else {
 702                                         unsigned long percents = zip_header.formatted.ucmpsize - zip_header.formatted.cmpsize;
 703                                         if ((int32_t)percents < 0)
 704                                                 percents = 0; /* happens if ucmpsize < cmpsize */
 705                                         percents = percents * 100;
 706                                         if (zip_header.formatted.ucmpsize)
 707                                                 percents /= zip_header.formatted.ucmpsize;
 708                                         //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 709                                         //      "--------  ------  ------- ---- ---------- ----- --------  ----"
 710                                         printf(      "%8u  %s"        "%9u%4u%% " "%s "         "%08x  "  "%s\n",
 711                                                 (unsigned)zip_header.formatted.ucmpsize,
 712                                                 zip_header.formatted.method == 0 ? "Stored" : "Defl:N", /* Defl is method 8 */
 713 /* TODO: show other methods?
 714  *  1 - Shrunk
 715  *  2 - Reduced with compression factor 1
 716  *  3 - Reduced with compression factor 2
 717  *  4 - Reduced with compression factor 3
 718  *  5 - Reduced with compression factor 4
 719  *  6 - Imploded
 720  *  7 - Reserved for Tokenizing compression algorithm
 721  *  9 - Deflate64
 722  * 10 - PKWARE Data Compression Library Imploding
 723  * 11 - Reserved by PKWARE
 724  * 12 - BZIP2
 725  */
 726                                                 (unsigned)zip_header.formatted.cmpsize,
 727                                                 (unsigned)percents,
 728                                                 dtbuf,
 729                                                 zip_header.formatted.crc32,
 730                                                 dst_fn);
 731                                         total_size += zip_header.formatted.cmpsize;
 732                                 }
 733                                 total_usize += zip_header.formatted.ucmpsize;
 734                                 i = 'n';
 735                         } else if (dst_fd == STDOUT_FILENO) {
 736                                 /* Extracting to STDOUT */
 737                                 i = -1;
 738                         } else if (last_char_is(dst_fn, '/')) {
 739                                 /* Extract directory */
 740                                 if (stat(dst_fn, &stat_buf) == -1) {
 741                                         if (errno != ENOENT) {
 742                                                 bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 743                                         }
 744                                         if (!quiet) {
 745                                                 printf("   creating: %s\n", dst_fn);
 746                                         }
 747                                         unzip_create_leading_dirs(dst_fn);
 748                                         if (bb_make_directory(dst_fn, dir_mode, FILEUTILS_IGNORE_CHMOD_ERR)) {
 749                                                 xfunc_die();
 750                                         }
 751                                 } else {
 752                                         if (!S_ISDIR(stat_buf.st_mode)) {
 753                                                 bb_error_msg_and_die("'%s' exists but is not a %s",
 754                                                         dst_fn, "directory");
 755                                         }
 756                                 }
 757                                 i = 'n';
 758                         } else {
 759                                 /* Extract file */
 760  check_file:
 761                                 if (stat(dst_fn, &stat_buf) == -1) {
 762                                         /* File does not exist */
 763                                         if (errno != ENOENT) {
 764                                                 bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 765                                         }
 766                                         i = 'y';
 767                                 } else {
 768                                         /* File already exists */
 769                                         if (overwrite == O_NEVER) {
 770                                                 i = 'n';
 771                                         } else if (S_ISREG(stat_buf.st_mode)) {
 772                                                 /* File is regular file */
 773                                                 if (overwrite == O_ALWAYS) {
 774                                                         i = 'y';
 775                                                 } else {
 776                                                         printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
 777                                                         my_fgets80(key_buf);
 778                                                         i = key_buf[0];
 779                                                 }
 780                                         } else {
 781                                                 /* File is not regular file */
 782                                                 bb_error_msg_and_die("'%s' exists but is not a %s",
 783                                                         dst_fn, "regular file");
 784                                         }
 785                                 }
 786                         }
 787                 }
 788
 789                 switch (i) {
 790                 case 'A':
 791                         overwrite = O_ALWAYS;
 792                 case 'y': /* Open file and fall into unzip */
 793                         unzip_create_leading_dirs(dst_fn);
 794 #if ENABLE_FEATURE_UNZIP_CDF
 795                         dst_fd = xopen3(dst_fn, O_WRONLY | O_CREAT | O_TRUNC, file_mode);
 796 #else
 797                         dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC);
 798 #endif
 799                 case -1: /* Unzip */
 800                         if (!quiet) {
 801                                 printf("  inflating: %s\n", dst_fn);
 802                         }
 803                         unzip_extract(&zip_header, dst_fd);
 804                         if (dst_fd != STDOUT_FILENO) {
 805                                 /* closing STDOUT is potentially bad for future business */
 806                                 close(dst_fd);
 807                         }
 808                         break;
 809
 810                 case 'N':
 811                         overwrite = O_NEVER;
 812                 case 'n':
 813                         /* Skip entry data */
 814                         unzip_skip(zip_header.formatted.cmpsize);
 815                         break;
 816
 817                 case 'r':
 818                         /* Prompt for new name */
 819                         printf("new name: ");
 820                         my_fgets80(key_buf);
 821                         free(dst_fn);
 822                         dst_fn = xstrdup(key_buf);
 823                         chomp(dst_fn);
 824                         goto check_file;
 825
 826                 default:
 827                         printf("error: invalid response [%c]\n", (char)i);
 828                         goto check_file;
 829                 }
 830
 831                 total_entries++;
 832         }
 833
 834         if (listing && quiet <= 1) {
 835                 if (!verbose) {
 836                         //      "  Length      Date    Time    Name\n"
 837                         //      "---------  ---------- -----   ----"
 838                         printf( " --------%21s"               "-------\n"
 839                                      "%9lu%21s"               "%u files\n",
 840                                 "",
 841                                 total_usize, "", total_entries);
 842                 } else {
 843                         unsigned long percents = total_usize - total_size;
 844                         if ((long)percents < 0)
 845                                 percents = 0; /* happens if usize < size */
 846                         percents = percents * 100;
 847                         if (total_usize)
 848                                 percents /= total_usize;
 849                         //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 850                         //      "--------  ------  ------- ---- ---------- ----- --------  ----"
 851                         printf( "--------          ------- ----%28s"                      "----\n"
 852                                 "%8lu"              "%17lu%4u%%%28s"                      "%u files\n",
 853                                 "",
 854                                 total_usize, total_size, (unsigned)percents, "",
 855                                 total_entries);
 856                 }
 857         }
 858
 859         return 0;
 860 }