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        -j      Do not restore paths"
  66 //usage:     "\n        -p      Print to stdout"
  67 //usage:     "\n        -q      Quiet"
  68 //usage:     "\n        -x FILE Exclude FILEs"
  69 //usage:     "\n        -d DIR  Extract into DIR"
  70
  71 #include "libbb.h"
  72 #include "bb_archive.h"
  73
  74 #if 0
  75 # define dbg(...) bb_error_msg(__VA_ARGS__)
  76 #else
  77 # define dbg(...) ((void)0)
  78 #endif
  79
  80 enum {
  81 #if BB_BIG_ENDIAN
  82         ZIP_FILEHEADER_MAGIC = 0x504b0304,
  83         ZIP_CDF_MAGIC        = 0x504b0102, /* CDF item */
  84         ZIP_CDE_MAGIC        = 0x504b0506, /* End of CDF */
  85         ZIP_DD_MAGIC         = 0x504b0708,
  86 #else
  87         ZIP_FILEHEADER_MAGIC = 0x04034b50,
  88         ZIP_CDF_MAGIC        = 0x02014b50,
  89         ZIP_CDE_MAGIC        = 0x06054b50,
  90         ZIP_DD_MAGIC         = 0x08074b50,
  91 #endif
  92 };
  93
  94 #define ZIP_HEADER_LEN 26
  95
  96 typedef union {
  97         uint8_t raw[ZIP_HEADER_LEN];
  98         struct {
  99                 uint16_t version;               /* 0-1 */
 100                 uint16_t zip_flags;             /* 2-3 */
 101                 uint16_t method;                /* 4-5 */
 102                 uint16_t modtime;               /* 6-7 */
 103                 uint16_t moddate;               /* 8-9 */
 104                 uint32_t crc32 PACKED;          /* 10-13 */
 105                 uint32_t cmpsize PACKED;        /* 14-17 */
 106                 uint32_t ucmpsize PACKED;       /* 18-21 */
 107                 uint16_t filename_len;          /* 22-23 */
 108                 uint16_t extra_len;             /* 24-25 */
 109                 /* filename follows (not NUL terminated) */
 110                 /* extra field follows */
 111                 /* data follows */
 112         } fmt PACKED;
 113 } zip_header_t; /* PACKED - gcc 4.2.1 doesn't like it (spews warning) */
 114
 115 #define FIX_ENDIANNESS_ZIP(zip) \
 116 do { if (BB_BIG_ENDIAN) { \
 117         (zip).fmt.crc32         = SWAP_LE32((zip).fmt.crc32       ); \
 118         (zip).fmt.cmpsize       = SWAP_LE32((zip).fmt.cmpsize     ); \
 119         (zip).fmt.ucmpsize      = SWAP_LE32((zip).fmt.ucmpsize    ); \
 120         (zip).fmt.filename_len  = SWAP_LE16((zip).fmt.filename_len); \
 121         (zip).fmt.extra_len     = SWAP_LE16((zip).fmt.extra_len   ); \
 122 }} while (0)
 123
 124 #define CDF_HEADER_LEN 42
 125
 126 typedef union {
 127         uint8_t raw[CDF_HEADER_LEN];
 128         struct {
 129                 /* uint32_t signature; 50 4b 01 02 */
 130                 uint16_t version_made_by;       /* 0-1 */
 131                 uint16_t version_needed;        /* 2-3 */
 132                 uint16_t cdf_flags;             /* 4-5 */
 133                 uint16_t method;                /* 6-7 */
 134                 uint16_t modtime;               /* 8-9 */
 135                 uint16_t moddate;               /* 10-11 */
 136                 uint32_t crc32;                 /* 12-15 */
 137                 uint32_t cmpsize;               /* 16-19 */
 138                 uint32_t ucmpsize;              /* 20-23 */
 139                 uint16_t filename_len;          /* 24-25 */
 140                 uint16_t extra_len;             /* 26-27 */
 141                 uint16_t file_comment_length;   /* 28-29 */
 142                 uint16_t disk_number_start;     /* 30-31 */
 143                 uint16_t internal_attributes;   /* 32-33 */
 144                 uint32_t external_attributes PACKED; /* 34-37 */
 145                 uint32_t relative_offset_of_local_header PACKED; /* 38-41 */
 146                 /* filename follows (not NUL terminated) */
 147                 /* extra field follows */
 148                 /* file comment follows */
 149         } fmt PACKED;
 150 } cdf_header_t;
 151
 152 #define FIX_ENDIANNESS_CDF(cdf) \
 153 do { if (BB_BIG_ENDIAN) { \
 154         (cdf).fmt.version_made_by = SWAP_LE16((cdf).fmt.version_made_by); \
 155         (cdf).fmt.version_needed = SWAP_LE16((cdf).fmt.version_needed); \
 156         (cdf).fmt.method        = SWAP_LE16((cdf).fmt.method      ); \
 157         (cdf).fmt.modtime       = SWAP_LE16((cdf).fmt.modtime     ); \
 158         (cdf).fmt.moddate       = SWAP_LE16((cdf).fmt.moddate     ); \
 159         (cdf).fmt.crc32         = SWAP_LE32((cdf).fmt.crc32       ); \
 160         (cdf).fmt.cmpsize       = SWAP_LE32((cdf).fmt.cmpsize     ); \
 161         (cdf).fmt.ucmpsize      = SWAP_LE32((cdf).fmt.ucmpsize    ); \
 162         (cdf).fmt.filename_len  = SWAP_LE16((cdf).fmt.filename_len); \
 163         (cdf).fmt.extra_len     = SWAP_LE16((cdf).fmt.extra_len   ); \
 164         (cdf).fmt.file_comment_length = SWAP_LE16((cdf).fmt.file_comment_length); \
 165         (cdf).fmt.external_attributes = SWAP_LE32((cdf).fmt.external_attributes); \
 166 }} while (0)
 167
 168 #define CDE_LEN 16
 169
 170 typedef union {
 171         uint8_t raw[CDE_LEN];
 172         struct {
 173                 /* uint32_t signature; 50 4b 05 06 */
 174                 uint16_t this_disk_no;
 175                 uint16_t disk_with_cdf_no;
 176                 uint16_t cdf_entries_on_this_disk;
 177                 uint16_t cdf_entries_total;
 178                 uint32_t cdf_size;
 179                 uint32_t cdf_offset;
 180                 /* uint16_t archive_comment_length; */
 181                 /* archive comment follows */
 182         } fmt PACKED;
 183 } cde_t;
 184
 185 #define FIX_ENDIANNESS_CDE(cde) \
 186 do { if (BB_BIG_ENDIAN) { \
 187         (cde).fmt.cdf_offset = SWAP_LE32((cde).fmt.cdf_offset); \
 188 }} while (0)
 189
 190 struct BUG {
 191         /* Check the offset of the last element, not the length.  This leniency
 192          * allows for poor packing, whereby the overall struct may be too long,
 193          * even though the elements are all in the right place.
 194          */
 195         char BUG_zip_header_must_be_26_bytes[
 196                 offsetof(zip_header_t, fmt.extra_len) + 2
 197                         == ZIP_HEADER_LEN ? 1 : -1];
 198         char BUG_cdf_header_must_be_42_bytes[
 199                 offsetof(cdf_header_t, fmt.relative_offset_of_local_header) + 4
 200                         == CDF_HEADER_LEN ? 1 : -1];
 201         char BUG_cde_must_be_16_bytes[
 202                 sizeof(cde_t) == CDE_LEN ? 1 : -1];
 203 };
 204
 205
 206 enum { zip_fd = 3 };
 207
 208
 209 /* This value means that we failed to find CDF */
 210 #define BAD_CDF_OFFSET ((uint32_t)0xffffffff)
 211
 212 #if !ENABLE_FEATURE_UNZIP_CDF
 213
 214 # define find_cdf_offset() BAD_CDF_OFFSET
 215
 216 #else
 217 /* Seen in the wild:
 218  * Self-extracting PRO2K3XP_32.exe contains 19078464 byte zip archive,
 219  * where CDE was nearly 48 kbytes before EOF.
 220  * (Surprisingly, it also apparently has *another* CDE structure
 221  * closer to the end, with bogus cdf_offset).
 222  * To make extraction work, bumped PEEK_FROM_END from 16k to 64k.
 223  */
 224 #define PEEK_FROM_END (64*1024)
 225 /* NB: does not preserve file position! */
 226 static uint32_t find_cdf_offset(void)
 227 {
 228         cde_t cde;
 229         unsigned char *buf;
 230         unsigned char *p;
 231         off_t end;
 232         uint32_t found;
 233
 234         end = lseek(zip_fd, 0, SEEK_END);
 235         if (end == (off_t) -1)
 236                 return BAD_CDF_OFFSET;
 237
 238         end -= PEEK_FROM_END;
 239         if (end < 0)
 240                 end = 0;
 241
 242         dbg("Looking for cdf_offset starting from 0x%"OFF_FMT"x", end);
 243         xlseek(zip_fd, end, SEEK_SET);
 244         buf = xzalloc(PEEK_FROM_END);
 245         full_read(zip_fd, buf, PEEK_FROM_END);
 246
 247         found = BAD_CDF_OFFSET;
 248         p = buf;
 249         while (p <= buf + PEEK_FROM_END - CDE_LEN - 4) {
 250                 if (*p != 'P') {
 251                         p++;
 252                         continue;
 253                 }
 254                 if (*++p != 'K')
 255                         continue;
 256                 if (*++p != 5)
 257                         continue;
 258                 if (*++p != 6)
 259                         continue;
 260                 /* we found CDE! */
 261                 memcpy(cde.raw, p + 1, CDE_LEN);
 262                 FIX_ENDIANNESS_CDE(cde);
 263                 /*
 264                  * I've seen .ZIP files with seemingly valid CDEs
 265                  * where cdf_offset points past EOF - ??
 266                  * This check ignores such CDEs:
 267                  */
 268                 if (cde.fmt.cdf_offset < end + (p - buf)) {
 269                         found = cde.fmt.cdf_offset;
 270                         dbg("Possible cdf_offset:0x%x at 0x%"OFF_FMT"x",
 271                                 (unsigned)found, end + (p-3 - buf));
 272                         dbg("  cdf_offset+cdf_size:0x%x",
 273                                 (unsigned)(found + SWAP_LE32(cde.fmt.cdf_size)));
 274                         /*
 275                          * We do not "break" here because only the last CDE is valid.
 276                          * I've seen a .zip archive which contained a .zip file,
 277                          * uncompressed, and taking the first CDE was using
 278                          * the CDE inside that file!
 279                          */
 280                 }
 281         }
 282         free(buf);
 283         dbg("Found cdf_offset:0x%x", (unsigned)found);
 284         return found;
 285 };
 286
 287 static uint32_t read_next_cdf(uint32_t cdf_offset, cdf_header_t *cdf)
 288 {
 289         uint32_t magic;
 290
 291         if (cdf_offset == BAD_CDF_OFFSET)
 292                 return cdf_offset;
 293
 294         dbg("Reading CDF at 0x%x", (unsigned)cdf_offset);
 295         xlseek(zip_fd, cdf_offset, SEEK_SET);
 296         xread(zip_fd, &magic, 4);
 297         /* Central Directory End? Assume CDF has ended.
 298          * (more correct method is to use cde.cdf_entries_total counter)
 299          */
 300         if (magic == ZIP_CDE_MAGIC) {
 301                 dbg("got ZIP_CDE_MAGIC");
 302                 return 0; /* EOF */
 303         }
 304         xread(zip_fd, cdf->raw, CDF_HEADER_LEN);
 305
 306         FIX_ENDIANNESS_CDF(*cdf);
 307         dbg("  filename_len:%u extra_len:%u file_comment_length:%u",
 308                 (unsigned)cdf->fmt.filename_len,
 309                 (unsigned)cdf->fmt.extra_len,
 310                 (unsigned)cdf->fmt.file_comment_length
 311         );
 312         cdf_offset += 4 + CDF_HEADER_LEN
 313                 + cdf->fmt.filename_len
 314                 + cdf->fmt.extra_len
 315                 + cdf->fmt.file_comment_length;
 316
 317         return cdf_offset;
 318 };
 319 #endif
 320
 321 static void die_if_bad_fnamesize(unsigned sz)
 322 {
 323         if (sz > 0xfff) /* more than 4k?! no funny business please */
 324                 bb_error_msg_and_die("bad archive");
 325 }
 326
 327 static void unzip_skip(off_t skip)
 328 {
 329         if (skip != 0)
 330                 if (lseek(zip_fd, skip, SEEK_CUR) == (off_t)-1)
 331                         bb_copyfd_exact_size(zip_fd, -1, skip);
 332 }
 333
 334 static void unzip_create_leading_dirs(const char *fn)
 335 {
 336         /* Create all leading directories */
 337         char *name = xstrdup(fn);
 338         if (bb_make_directory(dirname(name), 0777, FILEUTILS_RECUR)) {
 339                 xfunc_die(); /* bb_make_directory is noisy */
 340         }
 341         free(name);
 342 }
 343
 344 #if ENABLE_FEATURE_UNZIP_CDF
 345 static void unzip_extract_symlink(zip_header_t *zip, const char *dst_fn)
 346 {
 347         char *target;
 348
 349         die_if_bad_fnamesize(zip->fmt.ucmpsize);
 350
 351         if (zip->fmt.method == 0) {
 352                 /* Method 0 - stored (not compressed) */
 353                 target = xzalloc(zip->fmt.ucmpsize + 1);
 354                 xread(zip_fd, target, zip->fmt.ucmpsize);
 355         } else {
 356 #if 1
 357                 bb_error_msg_and_die("compressed symlink is not supported");
 358 #else
 359                 transformer_state_t xstate;
 360                 init_transformer_state(&xstate);
 361                 xstate.mem_output_size_max = zip->fmt.ucmpsize;
 362                 /* ...unpack... */
 363                 if (!xstate.mem_output_buf)
 364                         WTF();
 365                 target = xstate.mem_output_buf;
 366                 target = xrealloc(target, xstate.mem_output_size + 1);
 367                 target[xstate.mem_output_size] = '\0';
 368 #endif
 369         }
 370 //TODO: libbb candidate
 371         if (symlink(target, dst_fn))
 372                 bb_perror_msg_and_die("can't create symlink '%s'", dst_fn);
 373         free(target);
 374 }
 375 #endif
 376
 377 static void unzip_extract(zip_header_t *zip, int dst_fd)
 378 {
 379         transformer_state_t xstate;
 380
 381         if (zip->fmt.method == 0) {
 382                 /* Method 0 - stored (not compressed) */
 383                 off_t size = zip->fmt.ucmpsize;
 384                 if (size)
 385                         bb_copyfd_exact_size(zip_fd, dst_fd, size);
 386                 return;
 387         }
 388
 389         init_transformer_state(&xstate);
 390         xstate.bytes_in = zip->fmt.cmpsize;
 391         xstate.src_fd = zip_fd;
 392         xstate.dst_fd = dst_fd;
 393         if (zip->fmt.method == 8) {
 394                 /* Method 8 - inflate */
 395                 if (inflate_unzip(&xstate) < 0)
 396                         bb_error_msg_and_die("inflate error");
 397                 /* Validate decompression - crc */
 398                 if (zip->fmt.crc32 != (xstate.crc32 ^ 0xffffffffL)) {
 399                         bb_error_msg_and_die("crc error");
 400                 }
 401         }
 402 #if ENABLE_FEATURE_UNZIP_BZIP2
 403         else if (zip->fmt.method == 12) {
 404                 /* Tested. Unpacker reads too much, but we use CDF
 405                  * and will seek to the correct beginning of next file.
 406                  */
 407                 xstate.bytes_out = unpack_bz2_stream(&xstate);
 408                 if (xstate.bytes_out < 0)
 409                         bb_error_msg_and_die("inflate error");
 410         }
 411 #endif
 412 #if ENABLE_FEATURE_UNZIP_LZMA
 413         else if (zip->fmt.method == 14) {
 414                 /* Not tested yet */
 415                 xstate.bytes_out = unpack_lzma_stream(&xstate);
 416                 if (xstate.bytes_out < 0)
 417                         bb_error_msg_and_die("inflate error");
 418         }
 419 #endif
 420 #if ENABLE_FEATURE_UNZIP_XZ
 421         else if (zip->fmt.method == 95) {
 422                 /* Not tested yet */
 423                 xstate.bytes_out = unpack_xz_stream(&xstate);
 424                 if (xstate.bytes_out < 0)
 425                         bb_error_msg_and_die("inflate error");
 426         }
 427 #endif
 428         else {
 429                 bb_error_msg_and_die("unsupported method %u", zip->fmt.method);
 430         }
 431
 432         /* Validate decompression - size */
 433         if (zip->fmt.ucmpsize != xstate.bytes_out) {
 434                 /* Don't die. Who knows, maybe len calculation
 435                  * was botched somewhere. After all, crc matched! */
 436                 bb_error_msg("bad length");
 437         }
 438 }
 439
 440 static void my_fgets80(char *buf80)
 441 {
 442         fflush_all();
 443         if (!fgets(buf80, 80, stdin)) {
 444                 bb_perror_msg_and_die("can't read standard input");
 445         }
 446 }
 447
 448 static int get_lstat_mode(const char *dst_fn)
 449 {
 450         struct stat stat_buf;
 451         if (lstat(dst_fn, &stat_buf) == -1) {
 452                 if (errno != ENOENT) {
 453                         bb_perror_msg_and_die("can't stat '%s'", dst_fn);
 454                 }
 455                 /* File does not exist */
 456                 return -1;
 457         }
 458         return stat_buf.st_mode;
 459 }
 460
 461 int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 462 int unzip_main(int argc, char **argv)
 463 {
 464         enum {
 465                 OPT_l = (1 << 0),
 466                 OPT_x = (1 << 1),
 467                 OPT_j = (1 << 2),
 468         };
 469         unsigned opts;
 470         smallint quiet = 0;
 471         IF_NOT_FEATURE_UNZIP_CDF(const) smallint verbose = 0;
 472         enum { O_PROMPT, O_NEVER, O_ALWAYS };
 473         smallint overwrite = O_PROMPT;
 474         uint32_t cdf_offset;
 475         unsigned long total_usize;
 476         unsigned long total_size;
 477         unsigned total_entries;
 478         int dst_fd = -1;
 479         char *src_fn = NULL;
 480         char *dst_fn = NULL;
 481         llist_t *zaccept = NULL;
 482         llist_t *zreject = NULL;
 483         char *base_dir = NULL;
 484         int i;
 485         char key_buf[80]; /* must match size used by my_fgets80 */
 486
 487 /* -q, -l and -v: UnZip 5.52 of 28 February 2005, by Info-ZIP:
 488  *
 489  * # /usr/bin/unzip -qq -v decompress_unlzma.i.zip
 490  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 491  * # /usr/bin/unzip -q -v decompress_unlzma.i.zip
 492  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 493  * --------  ------  ------- -----   ----   ----   ------    ----
 494  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 495  * --------          -------  ---                            -------
 496  *   204372            35278  83%                            1 file
 497  * # /usr/bin/unzip -v decompress_unlzma.i.zip
 498  * Archive:  decompress_unlzma.i.zip
 499  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 500  * --------  ------  ------- -----   ----   ----   ------    ----
 501  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 502  * --------          -------  ---                            -------
 503  *   204372            35278  83%                            1 file
 504  * # unzip -v decompress_unlzma.i.zip
 505  * Archive:  decompress_unlzma.i.zip
 506  *   Length     Date   Time    Name
 507  *  --------    ----   ----    ----
 508  *    204372  09-06-09 14:23   decompress_unlzma.i
 509  *  --------                   -------
 510  *    204372                   1 files
 511  * # /usr/bin/unzip -l -qq decompress_unlzma.i.zip
 512  *    204372  09-06-09 14:23   decompress_unlzma.i
 513  * # /usr/bin/unzip -l -q decompress_unlzma.i.zip
 514  *   Length     Date   Time    Name
 515  *  --------    ----   ----    ----
 516  *    204372  09-06-09 14:23   decompress_unlzma.i
 517  *  --------                   -------
 518  *    204372                   1 file
 519  * # /usr/bin/unzip -l decompress_unlzma.i.zip
 520  * Archive:  decompress_unlzma.i.zip
 521  *   Length     Date   Time    Name
 522  *  --------    ----   ----    ----
 523  *    204372  09-06-09 14:23   decompress_unlzma.i
 524  *  --------                   -------
 525  *    204372                   1 file
 526  */
 527
 528         opts = 0;
 529         /* '-' makes getopt return 1 for non-options */
 530         while ((i = getopt(argc, argv, "-d:lnopqxjv")) != -1) {
 531                 switch (i) {
 532                 case 'd':  /* Extract to base directory */
 533                         base_dir = optarg;
 534                         break;
 535
 536                 case 'l': /* List */
 537                         opts |= OPT_l;
 538                         break;
 539
 540                 case 'n': /* Never overwrite existing files */
 541                         overwrite = O_NEVER;
 542                         break;
 543
 544                 case 'o': /* Always overwrite existing files */
 545                         overwrite = O_ALWAYS;
 546                         break;
 547
 548                 case 'p': /* Extract files to stdout and fall through to set verbosity */
 549                         dst_fd = STDOUT_FILENO;
 550
 551                 case 'q': /* Be quiet */
 552                         quiet++;
 553                         break;
 554
 555                 case 'v': /* Verbose list */
 556                         IF_FEATURE_UNZIP_CDF(verbose++;)
 557                         opts |= OPT_l;
 558                         break;
 559
 560                 case 'x':
 561                         opts |= OPT_x;
 562                         break;
 563
 564                 case 'j':
 565                         opts |= OPT_j;
 566                         break;
 567
 568                 case 1:
 569                         if (!src_fn) {
 570                                 /* The zip file */
 571                                 /* +5: space for ".zip" and NUL */
 572                                 src_fn = xmalloc(strlen(optarg) + 5);
 573                                 strcpy(src_fn, optarg);
 574                         } else if (!(opts & OPT_x)) {
 575                                 /* Include files */
 576                                 llist_add_to(&zaccept, optarg);
 577                         } else {
 578                                 /* Exclude files */
 579                                 llist_add_to(&zreject, optarg);
 580                         }
 581                         break;
 582
 583                 default:
 584                         bb_show_usage();
 585                 }
 586         }
 587
 588 #ifndef __GLIBC__
 589         /*
 590          * This code is needed for non-GNU getopt
 591          * which doesn't understand "-" in option string.
 592          * The -x option won't work properly in this case:
 593          * "unzip a.zip q -x w e" will be interpreted as
 594          * "unzip a.zip q w e -x" = "unzip a.zip q w e"
 595          */
 596         argv += optind;
 597         if (argv[0]) {
 598                 /* +5: space for ".zip" and NUL */
 599                 src_fn = xmalloc(strlen(argv[0]) + 5);
 600                 strcpy(src_fn, argv[0]);
 601                 while (*++argv)
 602                         llist_add_to(&zaccept, *argv);
 603         }
 604 #endif
 605
 606         if (!src_fn) {
 607                 bb_show_usage();
 608         }
 609
 610         /* Open input file */
 611         if (LONE_DASH(src_fn)) {
 612                 xdup2(STDIN_FILENO, zip_fd);
 613                 /* Cannot use prompt mode since zip data is arriving on STDIN */
 614                 if (overwrite == O_PROMPT)
 615                         overwrite = O_NEVER;
 616         } else {
 617                 static const char extn[][5] ALIGN1 = { ".zip", ".ZIP" };
 618                 char *ext = src_fn + strlen(src_fn);
 619                 int src_fd;
 620
 621                 i = 0;
 622                 for (;;) {
 623                         src_fd = open(src_fn, O_RDONLY);
 624                         if (src_fd >= 0)
 625                                 break;
 626                         if (++i > 2) {
 627                                 *ext = '\0';
 628                                 bb_error_msg_and_die("can't open %s[.zip]", src_fn);
 629                         }
 630                         strcpy(ext, extn[i - 1]);
 631                 }
 632                 xmove_fd(src_fd, zip_fd);
 633         }
 634
 635         /* Change dir if necessary */
 636         if (base_dir)
 637                 xchdir(base_dir);
 638
 639         if (quiet <= 1) { /* not -qq */
 640                 if (quiet == 0)
 641                         printf("Archive:  %s\n", src_fn);
 642                 if (opts & OPT_l) {
 643                         puts(verbose ?
 644                                 " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 645                                 "--------  ------  ------- ---- ---------- ----- --------  ----"
 646                                 :
 647                                 "  Length      Date    Time    Name\n"
 648                                 "---------  ---------- -----   ----"
 649                                 );
 650                 }
 651         }
 652
 653 /* Example of an archive with one 0-byte long file named 'z'
 654  * created by Zip 2.31 on Unix:
 655  * 0000 [50 4b]03 04 0a 00 00 00 00 00 42 1a b8 3c 00 00 |PK........B..<..|
 656  *       sig........ vneed flags compr mtime mdate crc32>
 657  * 0010  00 00 00 00 00 00 00 00 00 00 01 00 15 00 7a 55 |..............zU|
 658  *      >..... csize...... usize...... fnlen exlen fn ex>
 659  * 0020  54 09 00 03 cc d3 f9 4b cc d3 f9 4b 55 78 04 00 |T......K...KUx..|
 660  *      >tra_field......................................
 661  * 0030  00 00 00 00[50 4b]01 02 17 03 0a 00 00 00 00 00 |....PK..........|
 662  *       ........... sig........ vmade vneed flags compr
 663  * 0040  42 1a b8 3c 00 00 00 00 00 00 00 00 00 00 00 00 |B..<............|
 664  *       mtime mdate crc32...... csize...... usize......
 665  * 0050  01 00 0d 00 00 00 00 00 00 00 00 00 a4 81 00 00 |................|
 666  *       fnlen exlen clen. dnum. iattr eattr...... relofs> (eattr = rw-r--r--)
 667  * 0060  00 00 7a 55 54 05 00 03 cc d3 f9 4b 55 78 00 00 |..zUT......KUx..|
 668  *      >..... fn extra_field...........................
 669  * 0070 [50 4b]05 06 00 00 00 00 01 00 01 00 3c 00 00 00 |PK..........<...|
 670  * 0080  34 00 00 00 00 00                               |4.....|
 671  */
 672         total_usize = 0;
 673         total_size = 0;
 674         total_entries = 0;
 675         cdf_offset = find_cdf_offset(); /* try to seek to the end, find CDE and CDF start */
 676         while (1) {
 677                 zip_header_t zip;
 678                 mode_t dir_mode = 0777;
 679 #if ENABLE_FEATURE_UNZIP_CDF
 680                 mode_t file_mode = 0666;
 681 #endif
 682
 683                 if (!ENABLE_FEATURE_UNZIP_CDF || cdf_offset == BAD_CDF_OFFSET) {
 684                         /* Normally happens when input is unseekable.
 685                          *
 686                          * Valid ZIP file has Central Directory at the end
 687                          * with central directory file headers (CDFs).
 688                          * After it, there is a Central Directory End structure.
 689                          * CDFs identify what files are in the ZIP and where
 690                          * they are located. This allows ZIP readers to load
 691                          * the list of files without reading the entire ZIP archive.
 692                          * ZIP files may be appended to, only files specified in
 693                          * the CD are valid. Scanning for local file headers is
 694                          * not a correct algorithm.
 695                          *
 696                          * We try to do the above, and resort to "linear" reading
 697                          * of ZIP file only if seek failed or CDE wasn't found.
 698                          */
 699                         uint32_t magic;
 700
 701                         /* Check magic number */
 702                         xread(zip_fd, &magic, 4);
 703                         /* CDF item? Assume there are no more files, exit */
 704                         if (magic == ZIP_CDF_MAGIC) {
 705                                 dbg("got ZIP_CDF_MAGIC");
 706                                 break;
 707                         }
 708                         /* Data descriptor? It was a streaming file, go on */
 709                         if (magic == ZIP_DD_MAGIC) {
 710                                 dbg("got ZIP_DD_MAGIC");
 711                                 /* skip over duplicate crc32, cmpsize and ucmpsize */
 712                                 unzip_skip(3 * 4);
 713                                 continue;
 714                         }
 715                         if (magic != ZIP_FILEHEADER_MAGIC)
 716                                 bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
 717                         dbg("got ZIP_FILEHEADER_MAGIC");
 718
 719                         xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 720                         FIX_ENDIANNESS_ZIP(zip);
 721                         if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 722                                 bb_error_msg_and_die("zip flag %s is not supported",
 723                                         "8 (streaming)");
 724                         }
 725                 }
 726 #if ENABLE_FEATURE_UNZIP_CDF
 727                 else {
 728                         /* cdf_offset is valid (and we know the file is seekable) */
 729                         cdf_header_t cdf;
 730                         cdf_offset = read_next_cdf(cdf_offset, &cdf);
 731                         if (cdf_offset == 0) /* EOF? */
 732                                 break;
 733 # if 1
 734                         xlseek(zip_fd,
 735                                 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4,
 736                                 SEEK_SET);
 737                         xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 738                         FIX_ENDIANNESS_ZIP(zip);
 739                         if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 740                                 /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
 741                                  * only from Central Directory.
 742                                  */
 743                                 zip.fmt.crc32    = cdf.fmt.crc32;
 744                                 zip.fmt.cmpsize  = cdf.fmt.cmpsize;
 745                                 zip.fmt.ucmpsize = cdf.fmt.ucmpsize;
 746                         }
 747 // Seen in some zipfiles: central directory 9 byte extra field contains
 748 // a subfield with ID 0x5455 and 5 data bytes, which is a Unix-style UTC mtime.
 749 // Local header version:
 750 //  u16 0x5455 ("UT")
 751 //  u16 size (1 + 4 * n)
 752 //  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 753 //  u32 mtime
 754 //  u32 atime
 755 //  u32 ctime
 756 // Central header version:
 757 //  u16 0x5455 ("UT")
 758 //  u16 size (5 (or 1?))
 759 //  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 760 //  u32 mtime (CDF does not store atime/ctime)
 761 # else
 762                         /* CDF has the same data as local header, no need to read the latter...
 763                          * ...not really. An archive was seen with cdf.extra_len == 6 but
 764                          * zip.extra_len == 0.
 765                          */
 766                         memcpy(&zip.fmt.version,
 767                                 &cdf.fmt.version_needed, ZIP_HEADER_LEN);
 768                         xlseek(zip_fd,
 769                                 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4 + ZIP_HEADER_LEN,
 770                                 SEEK_SET);
 771 # endif
 772                         if ((cdf.fmt.version_made_by >> 8) == 3) {
 773                                 /* This archive is created on Unix */
 774                                 dir_mode = file_mode = (cdf.fmt.external_attributes >> 16);
 775                         }
 776                 }
 777 #endif
 778
 779                 if (zip.fmt.zip_flags & SWAP_LE16(0x0001)) {
 780                         /* 0x0001 - encrypted */
 781                         bb_error_msg_and_die("zip flag %s is not supported",
 782                                         "1 (encryption)");
 783                 }
 784                 dbg("File cmpsize:0x%x extra_len:0x%x ucmpsize:0x%x",
 785                         (unsigned)zip.fmt.cmpsize,
 786                         (unsigned)zip.fmt.extra_len,
 787                         (unsigned)zip.fmt.ucmpsize
 788                 );
 789
 790                 /* Read filename */
 791                 free(dst_fn);
 792                 die_if_bad_fnamesize(zip.fmt.filename_len);
 793                 dst_fn = xzalloc(zip.fmt.filename_len + 1);
 794                 xread(zip_fd, dst_fn, zip.fmt.filename_len);
 795                 /* Skip extra header bytes */
 796                 unzip_skip(zip.fmt.extra_len);
 797
 798                 /* Guard against "/abspath", "/../" and similar attacks */
 799                 overlapping_strcpy(dst_fn, strip_unsafe_prefix(dst_fn));
 800
 801                 if (opts & OPT_j) /* Strip paths? */
 802                         overlapping_strcpy(dst_fn, bb_basename(dst_fn));
 803
 804                 /* Did this strip everything ("DIR/" case)? Then skip */
 805                 if (!dst_fn[0])
 806                         goto skip_cmpsize;
 807
 808                 /* Filter zip entries */
 809                 if (find_list_entry(zreject, dst_fn)
 810                  || (zaccept && !find_list_entry(zaccept, dst_fn))
 811                 ) { /* Skip entry */
 812                         goto skip_cmpsize;
 813                 }
 814
 815                 if (opts & OPT_l) {
 816                         /* List entry */
 817                         char dtbuf[sizeof("mm-dd-yyyy hh:mm")];
 818                         sprintf(dtbuf, "%02u-%02u-%04u %02u:%02u",
 819                                 (zip.fmt.moddate >> 5) & 0xf,  // mm: 0x01e0
 820                                 (zip.fmt.moddate)      & 0x1f, // dd: 0x001f
 821                                 (zip.fmt.moddate >> 9) + 1980, // yy: 0xfe00
 822                                 (zip.fmt.modtime >> 11),       // hh: 0xf800
 823                                 (zip.fmt.modtime >> 5) & 0x3f  // mm: 0x07e0
 824                                 // seconds/2 not shown, encoded in -- 0x001f
 825                         );
 826                         if (!verbose) {
 827                                 //      "  Length      Date    Time    Name\n"
 828                                 //      "---------  ---------- -----   ----"
 829                                 printf(       "%9u  " "%s   "         "%s\n",
 830                                         (unsigned)zip.fmt.ucmpsize,
 831                                         dtbuf,
 832                                         dst_fn);
 833                         } else {
 834                                 char method6[7];
 835                                 unsigned long percents;
 836
 837                                 sprintf(method6, "%6u", zip.fmt.method);
 838                                 if (zip.fmt.method == 0) {
 839                                         strcpy(method6, "Stored");
 840                                 }
 841                                 if (zip.fmt.method == 8) {
 842                                         strcpy(method6, "Defl:N");
 843                                         /* normal, maximum, fast, superfast */
 844                                         IF_DESKTOP(method6[5] = "NXFS"[(zip.fmt.zip_flags >> 1) & 3];)
 845                                 }
 846                                 percents = zip.fmt.ucmpsize - zip.fmt.cmpsize;
 847                                 if ((int32_t)percents < 0)
 848                                         percents = 0; /* happens if ucmpsize < cmpsize */
 849                                 percents = percents * 100;
 850                                 if (zip.fmt.ucmpsize)
 851                                         percents /= zip.fmt.ucmpsize;
 852                                 //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 853                                 //      "--------  ------  ------- ---- ---------- ----- --------  ----"
 854                                 printf(      "%8u  %s"        "%9u%4u%% " "%s "         "%08x  "  "%s\n",
 855                                         (unsigned)zip.fmt.ucmpsize,
 856                                         method6,
 857                                         (unsigned)zip.fmt.cmpsize,
 858                                         (unsigned)percents,
 859                                         dtbuf,
 860                                         zip.fmt.crc32,
 861                                         dst_fn);
 862                                 total_size += zip.fmt.cmpsize;
 863                         }
 864                         total_usize += zip.fmt.ucmpsize;
 865                         goto skip_cmpsize;
 866                 }
 867
 868                 if (dst_fd == STDOUT_FILENO) {
 869                         /* Extracting to STDOUT */
 870                         goto do_extract;
 871                 }
 872                 if (last_char_is(dst_fn, '/')) {
 873                         int mode;
 874
 875                         /* Extract directory */
 876                         mode = get_lstat_mode(dst_fn);
 877                         if (mode == -1) { /* ENOENT */
 878                                 if (!quiet) {
 879                                         printf("   creating: %s\n", dst_fn);
 880                                 }
 881                                 unzip_create_leading_dirs(dst_fn);
 882                                 if (bb_make_directory(dst_fn, dir_mode, FILEUTILS_IGNORE_CHMOD_ERR)) {
 883                                         xfunc_die();
 884                                 }
 885                         } else {
 886                                 if (!S_ISDIR(mode)) {
 887                                         bb_error_msg_and_die("'%s' exists but is not a %s",
 888                                                 dst_fn, "directory");
 889                                 }
 890                         }
 891                         goto skip_cmpsize;
 892                 }
 893  check_file:
 894                 /* Does target file already exist? */
 895                 {
 896                         int mode = get_lstat_mode(dst_fn);
 897                         if (mode == -1) {
 898                                 /* ENOENT: does not exist */
 899                                 goto do_open_and_extract;
 900                         }
 901                         if (overwrite == O_NEVER) {
 902                                 goto skip_cmpsize;
 903                         }
 904                         if (!S_ISREG(mode)) {
 905  fishy:
 906                                 bb_error_msg_and_die("'%s' exists but is not a %s",
 907                                         dst_fn, "regular file");
 908                         }
 909                         if (overwrite == O_ALWAYS) {
 910                                 goto do_open_and_extract;
 911                         }
 912                         printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
 913                         my_fgets80(key_buf);
 914                         /* User input could take a long time. Is it still a regular file? */
 915                         mode = get_lstat_mode(dst_fn);
 916                         if (!S_ISREG(mode))
 917                                 goto fishy;
 918                 }
 919
 920                 /* Extract (or skip) it */
 921                 switch (key_buf[0]) {
 922                 case 'A':
 923                         overwrite = O_ALWAYS;
 924                 case 'y': /* Open file and fall into unzip */
 925  do_open_and_extract:
 926                         unzip_create_leading_dirs(dst_fn);
 927 #if ENABLE_FEATURE_UNZIP_CDF
 928                         dst_fd = -1;
 929                         if (!S_ISLNK(file_mode)) {
 930                                 dst_fd = xopen3(dst_fn,
 931                                         O_WRONLY | O_CREAT | O_TRUNC | O_NOFOLLOW,
 932                                         file_mode);
 933                         }
 934 #else
 935                         /* O_NOFOLLOW defends against symlink attacks */
 936                         dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC | O_NOFOLLOW);
 937 #endif
 938  do_extract:
 939                         if (!quiet) {
 940                                 printf(/* zip.fmt.method == 0
 941                                         ? " extracting: %s\n"
 942                                         : */ "  inflating: %s\n", dst_fn);
 943                         }
 944 #if ENABLE_FEATURE_UNZIP_CDF
 945                         if (S_ISLNK(file_mode)) {
 946                                 if (dst_fd != STDOUT_FILENO) /* not -p? */
 947                                         unzip_extract_symlink(&zip, dst_fn);
 948                         } else
 949 #endif
 950                         {
 951                                 unzip_extract(&zip, dst_fd);
 952                                 if (dst_fd != STDOUT_FILENO) {
 953                                         /* closing STDOUT is potentially bad for future business */
 954                                         close(dst_fd);
 955                                 }
 956                         }
 957                         break;
 958
 959                 case 'N':
 960                         overwrite = O_NEVER;
 961                 case 'n': /* Skip entry data */
 962  skip_cmpsize:
 963                         unzip_skip(zip.fmt.cmpsize);
 964                         break;
 965
 966                 case 'r':
 967                         /* Prompt for new name */
 968                         printf("new name: ");
 969                         my_fgets80(key_buf);
 970                         free(dst_fn);
 971                         dst_fn = xstrdup(key_buf);
 972                         chomp(dst_fn);
 973                         goto check_file;
 974
 975                 default:
 976                         printf("error: invalid response [%c]\n", (char)key_buf[0]);
 977                         goto check_file;
 978                 }
 979
 980                 total_entries++;
 981         }
 982
 983         if ((opts & OPT_l) && quiet <= 1) {
 984                 if (!verbose) {
 985                         //      "  Length      Date    Time    Name\n"
 986                         //      "---------  ---------- -----   ----"
 987                         printf( " --------%21s"               "-------\n"
 988                                      "%9lu%21s"               "%u files\n",
 989                                 "",
 990                                 total_usize, "", total_entries);
 991                 } else {
 992                         unsigned long percents = total_usize - total_size;
 993                         if ((long)percents < 0)
 994                                 percents = 0; /* happens if usize < size */
 995                         percents = percents * 100;
 996                         if (total_usize)
 997                                 percents /= total_usize;
 998                         //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 999                         //      "--------  ------  ------- ---- ---------- ----- --------  ----"
1000                         printf( "--------          ------- ----%28s"                      "----\n"
1001                                 "%8lu"              "%17lu%4u%%%28s"                      "%u files\n",
1002                                 "",
1003                                 total_usize, total_size, (unsigned)percents, "",
1004                                 total_entries);
1005                 }
1006         }
1007
1008         return 0;
1009 }