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 (24 kb)"
  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_symlink(zip_header_t *zip, const char *dst_fn)
 338 {
 339         char *target;
 340
 341         if (zip->fmt.ucmpsize > 0xfff) /* no funny business please */
 342                 bb_error_msg_and_die("bad archive");
 343
 344         if (zip->fmt.method == 0) {
 345                 /* Method 0 - stored (not compressed) */
 346                 target = xzalloc(zip->fmt.ucmpsize + 1);
 347                 xread(zip_fd, target, zip->fmt.ucmpsize);
 348         } else {
 349 #if 1
 350                 bb_error_msg_and_die("compressed symlink is not supported");
 351 #else
 352                 transformer_state_t xstate;
 353                 init_transformer_state(&xstate);
 354                 xstate.mem_output_size_max = zip->fmt.ucmpsize;
 355                 /* ...unpack... */
 356                 if (!xstate.mem_output_buf)
 357                         WTF();
 358                 target = xstate.mem_output_buf;
 359                 target = xrealloc(target, xstate.mem_output_size + 1);
 360                 target[xstate.mem_output_size] = '\0';
 361 #endif
 362         }
 363 //TODO: libbb candidate
 364         if (symlink(target, dst_fn))
 365                 bb_perror_msg_and_die("can't create symlink '%s'", dst_fn);
 366         free(target);
 367 }
 368
 369 static void unzip_extract(zip_header_t *zip, int dst_fd)
 370 {
 371         transformer_state_t xstate;
 372
 373         if (zip->fmt.method == 0) {
 374                 /* Method 0 - stored (not compressed) */
 375                 off_t size = zip->fmt.ucmpsize;
 376                 if (size)
 377                         bb_copyfd_exact_size(zip_fd, dst_fd, size);
 378                 return;
 379         }
 380
 381         init_transformer_state(&xstate);
 382         xstate.bytes_in = zip->fmt.cmpsize;
 383         xstate.src_fd = zip_fd;
 384         xstate.dst_fd = dst_fd;
 385         if (zip->fmt.method == 8) {
 386                 /* Method 8 - inflate */
 387                 if (inflate_unzip(&xstate) < 0)
 388                         bb_error_msg_and_die("inflate error");
 389                 /* Validate decompression - crc */
 390                 if (zip->fmt.crc32 != (xstate.crc32 ^ 0xffffffffL)) {
 391                         bb_error_msg_and_die("crc error");
 392                 }
 393         }
 394 #if ENABLE_FEATURE_UNZIP_BZIP2
 395         else if (zip->fmt.method == 12) {
 396                 /* Tested. Unpacker reads too much, but we use CDF
 397                  * and will seek to the correct beginning of next file.
 398                  */
 399                 xstate.bytes_out = unpack_bz2_stream(&xstate);
 400                 if (xstate.bytes_out < 0)
 401                         bb_error_msg_and_die("inflate error");
 402         }
 403 #endif
 404 #if ENABLE_FEATURE_UNZIP_LZMA
 405         else if (zip->fmt.method == 14) {
 406                 /* Not tested yet */
 407                 xstate.bytes_out = unpack_lzma_stream(&xstate);
 408                 if (xstate.bytes_out < 0)
 409                         bb_error_msg_and_die("inflate error");
 410         }
 411 #endif
 412 #if ENABLE_FEATURE_UNZIP_XZ
 413         else if (zip->fmt.method == 95) {
 414                 /* Not tested yet */
 415                 xstate.bytes_out = unpack_xz_stream(&xstate);
 416                 if (xstate.bytes_out < 0)
 417                         bb_error_msg_and_die("inflate error");
 418         }
 419 #endif
 420         else {
 421                 bb_error_msg_and_die("unsupported method %u", zip->fmt.method);
 422         }
 423
 424         /* Validate decompression - size */
 425         if (zip->fmt.ucmpsize != xstate.bytes_out) {
 426                 /* Don't die. Who knows, maybe len calculation
 427                  * was botched somewhere. After all, crc matched! */
 428                 bb_error_msg("bad length");
 429         }
 430 }
 431
 432 static void my_fgets80(char *buf80)
 433 {
 434         fflush_all();
 435         if (!fgets(buf80, 80, stdin)) {
 436                 bb_perror_msg_and_die("can't read standard input");
 437         }
 438 }
 439
 440 int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 441 int unzip_main(int argc, char **argv)
 442 {
 443         enum { O_PROMPT, O_NEVER, O_ALWAYS };
 444
 445         smallint quiet = 0;
 446         IF_NOT_FEATURE_UNZIP_CDF(const) smallint verbose = 0;
 447         smallint listing = 0;
 448         smallint overwrite = O_PROMPT;
 449         smallint x_opt_seen;
 450         uint32_t cdf_offset;
 451         unsigned long total_usize;
 452         unsigned long total_size;
 453         unsigned total_entries;
 454         int dst_fd = -1;
 455         char *src_fn = NULL;
 456         char *dst_fn = NULL;
 457         llist_t *zaccept = NULL;
 458         llist_t *zreject = NULL;
 459         char *base_dir = NULL;
 460         int i, opt;
 461         char key_buf[80]; /* must match size used by my_fgets80 */
 462         struct stat stat_buf;
 463
 464 /* -q, -l and -v: UnZip 5.52 of 28 February 2005, by Info-ZIP:
 465  *
 466  * # /usr/bin/unzip -qq -v decompress_unlzma.i.zip
 467  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 468  * # /usr/bin/unzip -q -v decompress_unlzma.i.zip
 469  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 470  * --------  ------  ------- -----   ----   ----   ------    ----
 471  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 472  * --------          -------  ---                            -------
 473  *   204372            35278  83%                            1 file
 474  * # /usr/bin/unzip -v decompress_unlzma.i.zip
 475  * Archive:  decompress_unlzma.i.zip
 476  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 477  * --------  ------  ------- -----   ----   ----   ------    ----
 478  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 479  * --------          -------  ---                            -------
 480  *   204372            35278  83%                            1 file
 481  * # unzip -v decompress_unlzma.i.zip
 482  * Archive:  decompress_unlzma.i.zip
 483  *   Length     Date   Time    Name
 484  *  --------    ----   ----    ----
 485  *    204372  09-06-09 14:23   decompress_unlzma.i
 486  *  --------                   -------
 487  *    204372                   1 files
 488  * # /usr/bin/unzip -l -qq decompress_unlzma.i.zip
 489  *    204372  09-06-09 14:23   decompress_unlzma.i
 490  * # /usr/bin/unzip -l -q decompress_unlzma.i.zip
 491  *   Length     Date   Time    Name
 492  *  --------    ----   ----    ----
 493  *    204372  09-06-09 14:23   decompress_unlzma.i
 494  *  --------                   -------
 495  *    204372                   1 file
 496  * # /usr/bin/unzip -l decompress_unlzma.i.zip
 497  * Archive:  decompress_unlzma.i.zip
 498  *   Length     Date   Time    Name
 499  *  --------    ----   ----    ----
 500  *    204372  09-06-09 14:23   decompress_unlzma.i
 501  *  --------                   -------
 502  *    204372                   1 file
 503  */
 504
 505         x_opt_seen = 0;
 506         /* '-' makes getopt return 1 for non-options */
 507         while ((opt = getopt(argc, argv, "-d:lnopqxv")) != -1) {
 508                 switch (opt) {
 509                 case 'd':  /* Extract to base directory */
 510                         base_dir = optarg;
 511                         break;
 512
 513                 case 'l': /* List */
 514                         listing = 1;
 515                         break;
 516
 517                 case 'n': /* Never overwrite existing files */
 518                         overwrite = O_NEVER;
 519                         break;
 520
 521                 case 'o': /* Always overwrite existing files */
 522                         overwrite = O_ALWAYS;
 523                         break;
 524
 525                 case 'p': /* Extract files to stdout and fall through to set verbosity */
 526                         dst_fd = STDOUT_FILENO;
 527
 528                 case 'q': /* Be quiet */
 529                         quiet++;
 530                         break;
 531
 532                 case 'v': /* Verbose list */
 533                         IF_FEATURE_UNZIP_CDF(verbose++;)
 534                         listing = 1;
 535                         break;
 536
 537                 case 'x':
 538                         x_opt_seen = 1;
 539                         break;
 540
 541                 case 1:
 542                         if (!src_fn) {
 543                                 /* The zip file */
 544                                 /* +5: space for ".zip" and NUL */
 545                                 src_fn = xmalloc(strlen(optarg) + 5);
 546                                 strcpy(src_fn, optarg);
 547                         } else if (!x_opt_seen) {
 548                                 /* Include files */
 549                                 llist_add_to(&zaccept, optarg);
 550                         } else {
 551                                 /* Exclude files */
 552                                 llist_add_to(&zreject, optarg);
 553                         }
 554                         break;
 555
 556                 default:
 557                         bb_show_usage();
 558                 }
 559         }
 560
 561 #ifndef __GLIBC__
 562         /*
 563          * This code is needed for non-GNU getopt
 564          * which doesn't understand "-" in option string.
 565          * The -x option won't work properly in this case:
 566          * "unzip a.zip q -x w e" will be interpreted as
 567          * "unzip a.zip q w e -x" = "unzip a.zip q w e"
 568          */
 569         argv += optind;
 570         if (argv[0]) {
 571                 /* +5: space for ".zip" and NUL */
 572                 src_fn = xmalloc(strlen(argv[0]) + 5);
 573                 strcpy(src_fn, argv[0]);
 574                 while (*++argv)
 575                         llist_add_to(&zaccept, *argv);
 576         }
 577 #endif
 578
 579         if (!src_fn) {
 580                 bb_show_usage();
 581         }
 582
 583         /* Open input file */
 584         if (LONE_DASH(src_fn)) {
 585                 xdup2(STDIN_FILENO, zip_fd);
 586                 /* Cannot use prompt mode since zip data is arriving on STDIN */
 587                 if (overwrite == O_PROMPT)
 588                         overwrite = O_NEVER;
 589         } else {
 590                 static const char extn[][5] ALIGN1 = { ".zip", ".ZIP" };
 591                 char *ext = src_fn + strlen(src_fn);
 592                 int src_fd;
 593
 594                 i = 0;
 595                 for (;;) {
 596                         src_fd = open(src_fn, O_RDONLY);
 597                         if (src_fd >= 0)
 598                                 break;
 599                         if (++i > 2) {
 600                                 *ext = '\0';
 601                                 bb_error_msg_and_die("can't open %s[.zip]", src_fn);
 602                         }
 603                         strcpy(ext, extn[i - 1]);
 604                 }
 605                 xmove_fd(src_fd, zip_fd);
 606         }
 607
 608         /* Change dir if necessary */
 609         if (base_dir)
 610                 xchdir(base_dir);
 611
 612         if (quiet <= 1) { /* not -qq */
 613                 if (quiet == 0)
 614                         printf("Archive:  %s\n", src_fn);
 615                 if (listing) {
 616                         puts(verbose ?
 617                                 " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 618                                 "--------  ------  ------- ---- ---------- ----- --------  ----"
 619                                 :
 620                                 "  Length      Date    Time    Name\n"
 621                                 "---------  ---------- -----   ----"
 622                                 );
 623                 }
 624         }
 625
 626 /* Example of an archive with one 0-byte long file named 'z'
 627  * created by Zip 2.31 on Unix:
 628  * 0000 [50 4b]03 04 0a 00 00 00 00 00 42 1a b8 3c 00 00 |PK........B..<..|
 629  *       sig........ vneed flags compr mtime mdate crc32>
 630  * 0010  00 00 00 00 00 00 00 00 00 00 01 00 15 00 7a 55 |..............zU|
 631  *      >..... csize...... usize...... fnlen exlen fn ex>
 632  * 0020  54 09 00 03 cc d3 f9 4b cc d3 f9 4b 55 78 04 00 |T......K...KUx..|
 633  *      >tra_field......................................
 634  * 0030  00 00 00 00[50 4b]01 02 17 03 0a 00 00 00 00 00 |....PK..........|
 635  *       ........... sig........ vmade vneed flags compr
 636  * 0040  42 1a b8 3c 00 00 00 00 00 00 00 00 00 00 00 00 |B..<............|
 637  *       mtime mdate crc32...... csize...... usize......
 638  * 0050  01 00 0d 00 00 00 00 00 00 00 00 00 a4 81 00 00 |................|
 639  *       fnlen exlen clen. dnum. iattr eattr...... relofs> (eattr = rw-r--r--)
 640  * 0060  00 00 7a 55 54 05 00 03 cc d3 f9 4b 55 78 00 00 |..zUT......KUx..|
 641  *      >..... fn extra_field...........................
 642  * 0070 [50 4b]05 06 00 00 00 00 01 00 01 00 3c 00 00 00 |PK..........<...|
 643  * 0080  34 00 00 00 00 00                               |4.....|
 644  */
 645         total_usize = 0;
 646         total_size = 0;
 647         total_entries = 0;
 648         cdf_offset = find_cdf_offset(); /* try to seek to the end, find CDE and CDF start */
 649         while (1) {
 650                 zip_header_t zip;
 651                 mode_t dir_mode = 0777;
 652 #if ENABLE_FEATURE_UNZIP_CDF
 653                 mode_t file_mode = 0666;
 654 #endif
 655
 656                 if (!ENABLE_FEATURE_UNZIP_CDF || cdf_offset == BAD_CDF_OFFSET) {
 657                         /* Normally happens when input is unseekable.
 658                          *
 659                          * Valid ZIP file has Central Directory at the end
 660                          * with central directory file headers (CDFs).
 661                          * After it, there is a Central Directory End structure.
 662                          * CDFs identify what files are in the ZIP and where
 663                          * they are located. This allows ZIP readers to load
 664                          * the list of files without reading the entire ZIP archive.
 665                          * ZIP files may be appended to, only files specified in
 666                          * the CD are valid. Scanning for local file headers is
 667                          * not a correct algorithm.
 668                          *
 669                          * We try to do the above, and resort to "linear" reading
 670                          * of ZIP file only if seek failed or CDE wasn't found.
 671                          */
 672                         uint32_t magic;
 673
 674                         /* Check magic number */
 675                         xread(zip_fd, &magic, 4);
 676                         /* CDF item? Assume there are no more files, exit */
 677                         if (magic == ZIP_CDF_MAGIC) {
 678                                 dbg("got ZIP_CDF_MAGIC");
 679                                 break;
 680                         }
 681                         /* Data descriptor? It was a streaming file, go on */
 682                         if (magic == ZIP_DD_MAGIC) {
 683                                 dbg("got ZIP_DD_MAGIC");
 684                                 /* skip over duplicate crc32, cmpsize and ucmpsize */
 685                                 unzip_skip(3 * 4);
 686                                 continue;
 687                         }
 688                         if (magic != ZIP_FILEHEADER_MAGIC)
 689                                 bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
 690                         dbg("got ZIP_FILEHEADER_MAGIC");
 691
 692                         xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 693                         FIX_ENDIANNESS_ZIP(zip);
 694                         if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 695                                 bb_error_msg_and_die("zip flag %s is not supported",
 696                                         "8 (streaming)");
 697                         }
 698                 }
 699 #if ENABLE_FEATURE_UNZIP_CDF
 700                 else {
 701                         /* cdf_offset is valid (and we know the file is seekable) */
 702                         cdf_header_t cdf;
 703                         cdf_offset = read_next_cdf(cdf_offset, &cdf);
 704                         if (cdf_offset == 0) /* EOF? */
 705                                 break;
 706 # if 1
 707                         xlseek(zip_fd,
 708                                 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4,
 709                                 SEEK_SET);
 710                         xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 711                         FIX_ENDIANNESS_ZIP(zip);
 712                         if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 713                                 /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
 714                                  * only from Central Directory.
 715                                  */
 716                                 zip.fmt.crc32    = cdf.fmt.crc32;
 717                                 zip.fmt.cmpsize  = cdf.fmt.cmpsize;
 718                                 zip.fmt.ucmpsize = cdf.fmt.ucmpsize;
 719                         }
 720 // Seen in some zipfiles: central directory 9 byte extra field contains
 721 // a subfield with ID 0x5455 and 5 data bytes, which is a Unix-style UTC mtime.
 722 // Local header version:
 723 //  u16 0x5455 ("UT")
 724 //  u16 size (1 + 4 * n)
 725 //  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 726 //  u32 mtime
 727 //  u32 atime
 728 //  u32 ctime
 729 // Central header version:
 730 //  u16 0x5455 ("UT")
 731 //  u16 size (5 (or 1?))
 732 //  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 733 //  u32 mtime (CDF does not store atime/ctime)
 734 # else
 735                         /* CDF has the same data as local header, no need to read the latter...
 736                          * ...not really. An archive was seen with cdf.extra_len == 6 but
 737                          * zip.extra_len == 0.
 738                          */
 739                         memcpy(&zip.fmt.version,
 740                                 &cdf.fmt.version_needed, ZIP_HEADER_LEN);
 741                         xlseek(zip_fd,
 742                                 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4 + ZIP_HEADER_LEN,
 743                                 SEEK_SET);
 744 # endif
 745                         if ((cdf.fmt.version_made_by >> 8) == 3) {
 746                                 /* This archive is created on Unix */
 747                                 dir_mode = file_mode = (cdf.fmt.external_attributes >> 16);
 748                         }
 749                 }
 750 #endif
 751
 752                 if (zip.fmt.zip_flags & SWAP_LE16(0x0001)) {
 753                         /* 0x0001 - encrypted */
 754                         bb_error_msg_and_die("zip flag %s is not supported",
 755                                         "1 (encryption)");
 756                 }
 757                 dbg("File cmpsize:0x%x extra_len:0x%x ucmpsize:0x%x",
 758                         (unsigned)zip.fmt.cmpsize,
 759                         (unsigned)zip.fmt.extra_len,
 760                         (unsigned)zip.fmt.ucmpsize
 761                 );
 762
 763                 /* Read filename */
 764                 free(dst_fn);
 765                 dst_fn = xzalloc(zip.fmt.filename_len + 1);
 766                 xread(zip_fd, dst_fn, zip.fmt.filename_len);
 767
 768                 /* Skip extra header bytes */
 769                 unzip_skip(zip.fmt.extra_len);
 770
 771                 /* Guard against "/abspath", "/../" and similar attacks */
 772                 overlapping_strcpy(dst_fn, strip_unsafe_prefix(dst_fn));
 773
 774                 /* Filter zip entries */
 775                 if (find_list_entry(zreject, dst_fn)
 776                  || (zaccept && !find_list_entry(zaccept, dst_fn))
 777                 ) { /* Skip entry */
 778                         goto skip_cmpsize;
 779                 }
 780
 781                 if (listing) {
 782                         /* List entry */
 783                         char dtbuf[sizeof("mm-dd-yyyy hh:mm")];
 784                         sprintf(dtbuf, "%02u-%02u-%04u %02u:%02u",
 785                                 (zip.fmt.moddate >> 5) & 0xf,  // mm: 0x01e0
 786                                 (zip.fmt.moddate)      & 0x1f, // dd: 0x001f
 787                                 (zip.fmt.moddate >> 9) + 1980, // yy: 0xfe00
 788                                 (zip.fmt.modtime >> 11),       // hh: 0xf800
 789                                 (zip.fmt.modtime >> 5) & 0x3f  // mm: 0x07e0
 790                                 // seconds/2 not shown, encoded in -- 0x001f
 791                         );
 792                         if (!verbose) {
 793                                 //      "  Length      Date    Time    Name\n"
 794                                 //      "---------  ---------- -----   ----"
 795                                 printf(       "%9u  " "%s   "         "%s\n",
 796                                         (unsigned)zip.fmt.ucmpsize,
 797                                         dtbuf,
 798                                         dst_fn);
 799                         } else {
 800                                 char method6[7];
 801                                 unsigned long percents;
 802
 803                                 sprintf(method6, "%6u", zip.fmt.method);
 804                                 if (zip.fmt.method == 0) {
 805                                         strcpy(method6, "Stored");
 806                                 }
 807                                 if (zip.fmt.method == 8) {
 808                                         strcpy(method6, "Defl:N");
 809                                         /* normal, maximum, fast, superfast */
 810                                         IF_DESKTOP(method6[5] = "NXFS"[(zip.fmt.zip_flags >> 1) & 3];)
 811                                 }
 812                                 percents = zip.fmt.ucmpsize - zip.fmt.cmpsize;
 813                                 if ((int32_t)percents < 0)
 814                                         percents = 0; /* happens if ucmpsize < cmpsize */
 815                                 percents = percents * 100;
 816                                 if (zip.fmt.ucmpsize)
 817                                         percents /= zip.fmt.ucmpsize;
 818                                 //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 819                                 //      "--------  ------  ------- ---- ---------- ----- --------  ----"
 820                                 printf(      "%8u  %s"        "%9u%4u%% " "%s "         "%08x  "  "%s\n",
 821                                         (unsigned)zip.fmt.ucmpsize,
 822                                         method6,
 823                                         (unsigned)zip.fmt.cmpsize,
 824                                         (unsigned)percents,
 825                                         dtbuf,
 826                                         zip.fmt.crc32,
 827                                         dst_fn);
 828                                 total_size += zip.fmt.cmpsize;
 829                         }
 830                         total_usize += zip.fmt.ucmpsize;
 831                         goto skip_cmpsize;
 832                 }
 833
 834                 if (dst_fd == STDOUT_FILENO) {
 835                         /* Extracting to STDOUT */
 836                         goto do_extract;
 837                 }
 838                 if (last_char_is(dst_fn, '/')) {
 839                         /* Extract directory */
 840                         if (stat(dst_fn, &stat_buf) == -1) {
 841                                 if (errno != ENOENT) {
 842                                         bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 843                                 }
 844                                 if (!quiet) {
 845                                         printf("   creating: %s\n", dst_fn);
 846                                 }
 847                                 unzip_create_leading_dirs(dst_fn);
 848                                 if (bb_make_directory(dst_fn, dir_mode, FILEUTILS_IGNORE_CHMOD_ERR)) {
 849                                         xfunc_die();
 850                                 }
 851                         } else {
 852                                 if (!S_ISDIR(stat_buf.st_mode)) {
 853                                         bb_error_msg_and_die("'%s' exists but is not a %s",
 854                                                 dst_fn, "directory");
 855                                 }
 856                         }
 857                         goto skip_cmpsize;
 858                 }
 859  check_file:
 860                 /* Extract file */
 861                 if (lstat(dst_fn, &stat_buf) == -1) {
 862                         /* File does not exist */
 863                         if (errno != ENOENT) {
 864                                 bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 865                         }
 866                         goto do_open_and_extract;
 867                 }
 868                 /* File already exists */
 869                 if (overwrite == O_NEVER) {
 870                         goto skip_cmpsize;
 871                 }
 872                 if (!S_ISREG(stat_buf.st_mode)) {
 873                         /* File is not regular file */
 874                         bb_error_msg_and_die("'%s' exists but is not a %s",
 875                                 dst_fn, "regular file");
 876                 }
 877                 /* File is regular file */
 878                 if (overwrite == O_ALWAYS)
 879                         goto do_open_and_extract;
 880                 printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
 881                 my_fgets80(key_buf);
 882 //TODO: redo lstat + ISREG check! user input could have taken a long time!
 883
 884                 switch (key_buf[0]) {
 885                 case 'A':
 886                         overwrite = O_ALWAYS;
 887                 case 'y': /* Open file and fall into unzip */
 888  do_open_and_extract:
 889                         unzip_create_leading_dirs(dst_fn);
 890 #if ENABLE_FEATURE_UNZIP_CDF
 891                         if (!S_ISLNK(file_mode))
 892                                 dst_fd = xopen3(dst_fn, O_WRONLY | O_CREAT | O_TRUNC, file_mode);
 893 #else
 894                         dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC);
 895 #endif
 896  do_extract:
 897                         if (!quiet) {
 898                                 printf(/* zip.fmt.method == 0
 899                                         ? " extracting: %s\n"
 900                                         : */ "  inflating: %s\n", dst_fn);
 901                         }
 902 #if ENABLE_FEATURE_UNZIP_CDF
 903                         if (S_ISLNK(file_mode)) {
 904                                 if (dst_fd != STDOUT_FILENO) /* no -p */
 905                                         unzip_extract_symlink(&zip, dst_fn);
 906                         } else
 907 #endif
 908                         {
 909                                 unzip_extract(&zip, dst_fd);
 910                                 if (dst_fd != STDOUT_FILENO) {
 911                                         /* closing STDOUT is potentially bad for future business */
 912                                         close(dst_fd);
 913                                 }
 914                         }
 915                         break;
 916
 917                 case 'N':
 918                         overwrite = O_NEVER;
 919                 case 'n': /* Skip entry data */
 920  skip_cmpsize:
 921                         unzip_skip(zip.fmt.cmpsize);
 922                         break;
 923
 924                 case 'r':
 925                         /* Prompt for new name */
 926                         printf("new name: ");
 927                         my_fgets80(key_buf);
 928                         free(dst_fn);
 929                         dst_fn = xstrdup(key_buf);
 930                         chomp(dst_fn);
 931                         goto check_file;
 932
 933                 default:
 934                         printf("error: invalid response [%c]\n", (char)key_buf[0]);
 935                         goto check_file;
 936                 }
 937
 938                 total_entries++;
 939         }
 940
 941         if (listing && quiet <= 1) {
 942                 if (!verbose) {
 943                         //      "  Length      Date    Time    Name\n"
 944                         //      "---------  ---------- -----   ----"
 945                         printf( " --------%21s"               "-------\n"
 946                                      "%9lu%21s"               "%u files\n",
 947                                 "",
 948                                 total_usize, "", total_entries);
 949                 } else {
 950                         unsigned long percents = total_usize - total_size;
 951                         if ((long)percents < 0)
 952                                 percents = 0; /* happens if usize < size */
 953                         percents = percents * 100;
 954                         if (total_usize)
 955                                 percents /= total_usize;
 956                         //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 957                         //      "--------  ------  ------- ---- ---------- ----- --------  ----"
 958                         printf( "--------          ------- ----%28s"                      "----\n"
 959                                 "%8lu"              "%17lu%4u%%%28s"                      "%u files\n",
 960                                 "",
 961                                 total_usize, total_size, (unsigned)percents, "",
 962                                 total_entries);
 963                 }
 964         }
 965
 966         return 0;
 967 }