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 (26 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:       "[-lnojpq] 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.method        = SWAP_LE16((zip).fmt.method      ); \
 118         (zip).fmt.crc32         = SWAP_LE32((zip).fmt.crc32       ); \
 119         (zip).fmt.cmpsize       = SWAP_LE32((zip).fmt.cmpsize     ); \
 120         (zip).fmt.ucmpsize      = SWAP_LE32((zip).fmt.ucmpsize    ); \
 121         (zip).fmt.filename_len  = SWAP_LE16((zip).fmt.filename_len); \
 122         (zip).fmt.extra_len     = SWAP_LE16((zip).fmt.extra_len   ); \
 123 }} while (0)
 124
 125 #define CDF_HEADER_LEN 42
 126
 127 typedef union {
 128         uint8_t raw[CDF_HEADER_LEN];
 129         struct {
 130                 /* uint32_t signature; 50 4b 01 02 */
 131                 uint16_t version_made_by;       /* 0-1 */
 132                 uint16_t version_needed;        /* 2-3 */
 133                 uint16_t cdf_flags;             /* 4-5 */
 134                 uint16_t method;                /* 6-7 */
 135                 uint16_t modtime;               /* 8-9 */
 136                 uint16_t moddate;               /* 10-11 */
 137                 uint32_t crc32;                 /* 12-15 */
 138                 uint32_t cmpsize;               /* 16-19 */
 139                 uint32_t ucmpsize;              /* 20-23 */
 140                 uint16_t filename_len;          /* 24-25 */
 141                 uint16_t extra_len;             /* 26-27 */
 142                 uint16_t file_comment_length;   /* 28-29 */
 143                 uint16_t disk_number_start;     /* 30-31 */
 144                 uint16_t internal_attributes;   /* 32-33 */
 145                 uint32_t external_attributes PACKED; /* 34-37 */
 146                 uint32_t relative_offset_of_local_header PACKED; /* 38-41 */
 147                 /* filename follows (not NUL terminated) */
 148                 /* extra field follows */
 149                 /* file comment follows */
 150         } fmt PACKED;
 151 } cdf_header_t;
 152
 153 #define FIX_ENDIANNESS_CDF(cdf) \
 154 do { if (BB_BIG_ENDIAN) { \
 155         (cdf).fmt.version_made_by = SWAP_LE16((cdf).fmt.version_made_by); \
 156         (cdf).fmt.version_needed  = SWAP_LE16((cdf).fmt.version_needed ); \
 157         (cdf).fmt.method          = SWAP_LE16((cdf).fmt.method         ); \
 158         (cdf).fmt.modtime         = SWAP_LE16((cdf).fmt.modtime        ); \
 159         (cdf).fmt.moddate         = SWAP_LE16((cdf).fmt.moddate        ); \
 160         (cdf).fmt.crc32           = SWAP_LE32((cdf).fmt.crc32          ); \
 161         (cdf).fmt.cmpsize         = SWAP_LE32((cdf).fmt.cmpsize        ); \
 162         (cdf).fmt.ucmpsize        = SWAP_LE32((cdf).fmt.ucmpsize       ); \
 163         (cdf).fmt.filename_len    = SWAP_LE16((cdf).fmt.filename_len   ); \
 164         (cdf).fmt.extra_len       = SWAP_LE16((cdf).fmt.extra_len      ); \
 165         (cdf).fmt.file_comment_length = SWAP_LE16((cdf).fmt.file_comment_length); \
 166         (cdf).fmt.external_attributes = SWAP_LE32((cdf).fmt.external_attributes); \
 167 }} while (0)
 168
 169 #define CDE_LEN 16
 170
 171 typedef union {
 172         uint8_t raw[CDE_LEN];
 173         struct {
 174                 /* uint32_t signature; 50 4b 05 06 */
 175                 uint16_t this_disk_no;
 176                 uint16_t disk_with_cdf_no;
 177                 uint16_t cdf_entries_on_this_disk;
 178                 uint16_t cdf_entries_total;
 179                 uint32_t cdf_size;
 180                 uint32_t cdf_offset;
 181                 /* uint16_t archive_comment_length; */
 182                 /* archive comment follows */
 183         } fmt PACKED;
 184 } cde_t;
 185
 186 #define FIX_ENDIANNESS_CDE(cde) \
 187 do { if (BB_BIG_ENDIAN) { \
 188         (cde).fmt.cdf_offset = SWAP_LE32((cde).fmt.cdf_offset); \
 189 }} while (0)
 190
 191 struct BUG {
 192         /* Check the offset of the last element, not the length.  This leniency
 193          * allows for poor packing, whereby the overall struct may be too long,
 194          * even though the elements are all in the right place.
 195          */
 196         char BUG_zip_header_must_be_26_bytes[
 197                 offsetof(zip_header_t, fmt.extra_len) + 2
 198                         == ZIP_HEADER_LEN ? 1 : -1];
 199         char BUG_cdf_header_must_be_42_bytes[
 200                 offsetof(cdf_header_t, fmt.relative_offset_of_local_header) + 4
 201                         == CDF_HEADER_LEN ? 1 : -1];
 202         char BUG_cde_must_be_16_bytes[
 203                 sizeof(cde_t) == CDE_LEN ? 1 : -1];
 204 };
 205
 206
 207 enum { zip_fd = 3 };
 208
 209
 210 /* This value means that we failed to find CDF */
 211 #define BAD_CDF_OFFSET ((uint32_t)0xffffffff)
 212
 213 #if !ENABLE_FEATURE_UNZIP_CDF
 214
 215 # define find_cdf_offset() BAD_CDF_OFFSET
 216
 217 #else
 218 /* Seen in the wild:
 219  * Self-extracting PRO2K3XP_32.exe contains 19078464 byte zip archive,
 220  * where CDE was nearly 48 kbytes before EOF.
 221  * (Surprisingly, it also apparently has *another* CDE structure
 222  * closer to the end, with bogus cdf_offset).
 223  * To make extraction work, bumped PEEK_FROM_END from 16k to 64k.
 224  */
 225 #define PEEK_FROM_END (64*1024)
 226 /* NB: does not preserve file position! */
 227 static uint32_t find_cdf_offset(void)
 228 {
 229         cde_t cde;
 230         unsigned char *buf;
 231         unsigned char *p;
 232         off_t end;
 233         uint32_t found;
 234
 235         end = lseek(zip_fd, 0, SEEK_END);
 236         if (end == (off_t) -1)
 237                 return BAD_CDF_OFFSET;
 238
 239         end -= PEEK_FROM_END;
 240         if (end < 0)
 241                 end = 0;
 242
 243         dbg("Looking for cdf_offset starting from 0x%"OFF_FMT"x", end);
 244         xlseek(zip_fd, end, SEEK_SET);
 245         buf = xzalloc(PEEK_FROM_END);
 246         full_read(zip_fd, buf, PEEK_FROM_END);
 247
 248         found = BAD_CDF_OFFSET;
 249         p = buf;
 250         while (p <= buf + PEEK_FROM_END - CDE_LEN - 4) {
 251                 if (*p != 'P') {
 252                         p++;
 253                         continue;
 254                 }
 255                 if (*++p != 'K')
 256                         continue;
 257                 if (*++p != 5)
 258                         continue;
 259                 if (*++p != 6)
 260                         continue;
 261                 /* we found CDE! */
 262                 memcpy(cde.raw, p + 1, CDE_LEN);
 263                 FIX_ENDIANNESS_CDE(cde);
 264                 /*
 265                  * I've seen .ZIP files with seemingly valid CDEs
 266                  * where cdf_offset points past EOF - ??
 267                  * This check ignores such CDEs:
 268                  */
 269                 if (cde.fmt.cdf_offset < end + (p - buf)) {
 270                         found = cde.fmt.cdf_offset;
 271                         dbg("Possible cdf_offset:0x%x at 0x%"OFF_FMT"x",
 272                                 (unsigned)found, end + (p-3 - buf));
 273                         dbg("  cdf_offset+cdf_size:0x%x",
 274                                 (unsigned)(found + SWAP_LE32(cde.fmt.cdf_size)));
 275                         /*
 276                          * We do not "break" here because only the last CDE is valid.
 277                          * I've seen a .zip archive which contained a .zip file,
 278                          * uncompressed, and taking the first CDE was using
 279                          * the CDE inside that file!
 280                          */
 281                 }
 282         }
 283         free(buf);
 284         dbg("Found cdf_offset:0x%x", (unsigned)found);
 285         return found;
 286 };
 287
 288 static uint32_t read_next_cdf(uint32_t cdf_offset, cdf_header_t *cdf)
 289 {
 290         uint32_t magic;
 291
 292         if (cdf_offset == BAD_CDF_OFFSET)
 293                 return cdf_offset;
 294
 295         dbg("Reading CDF at 0x%x", (unsigned)cdf_offset);
 296         xlseek(zip_fd, cdf_offset, SEEK_SET);
 297         xread(zip_fd, &magic, 4);
 298         /* Central Directory End? Assume CDF has ended.
 299          * (more correct method is to use cde.cdf_entries_total counter)
 300          */
 301         if (magic == ZIP_CDE_MAGIC) {
 302                 dbg("got ZIP_CDE_MAGIC");
 303                 return 0; /* EOF */
 304         }
 305         xread(zip_fd, cdf->raw, CDF_HEADER_LEN);
 306
 307         FIX_ENDIANNESS_CDF(*cdf);
 308         dbg("  filename_len:%u extra_len:%u file_comment_length:%u",
 309                 (unsigned)cdf->fmt.filename_len,
 310                 (unsigned)cdf->fmt.extra_len,
 311                 (unsigned)cdf->fmt.file_comment_length
 312         );
 313         cdf_offset += 4 + CDF_HEADER_LEN
 314                 + cdf->fmt.filename_len
 315                 + cdf->fmt.extra_len
 316                 + cdf->fmt.file_comment_length;
 317
 318         return cdf_offset;
 319 };
 320 #endif
 321
 322 static void die_if_bad_fnamesize(unsigned sz)
 323 {
 324         if (sz > 0xfff) /* more than 4k?! no funny business please */
 325                 bb_error_msg_and_die("bad archive");
 326 }
 327
 328 static void unzip_skip(off_t skip)
 329 {
 330         if (skip != 0)
 331                 if (lseek(zip_fd, skip, SEEK_CUR) == (off_t)-1)
 332                         bb_copyfd_exact_size(zip_fd, -1, skip);
 333 }
 334
 335 static void unzip_create_leading_dirs(const char *fn)
 336 {
 337         /* Create all leading directories */
 338         char *name = xstrdup(fn);
 339
 340         /* mode of -1: set mode according to umask */
 341         if (bb_make_directory(dirname(name), -1, FILEUTILS_RECUR)) {
 342                 xfunc_die(); /* bb_make_directory is noisy */
 343         }
 344         free(name);
 345 }
 346
 347 #if ENABLE_FEATURE_UNZIP_CDF
 348 static void unzip_extract_symlink(llist_t **symlink_placeholders,
 349                 zip_header_t *zip,
 350                 const char *dst_fn)
 351 {
 352         char *target;
 353
 354         die_if_bad_fnamesize(zip->fmt.ucmpsize);
 355
 356         if (zip->fmt.method == 0) {
 357                 /* Method 0 - stored (not compressed) */
 358                 target = xzalloc(zip->fmt.ucmpsize + 1);
 359                 xread(zip_fd, target, zip->fmt.ucmpsize);
 360         } else {
 361 #if 1
 362                 bb_error_msg_and_die("compressed symlink is not supported");
 363 #else
 364                 transformer_state_t xstate;
 365                 init_transformer_state(&xstate);
 366                 xstate.mem_output_size_max = zip->fmt.ucmpsize;
 367                 /* ...unpack... */
 368                 if (!xstate.mem_output_buf)
 369                         WTF();
 370                 target = xstate.mem_output_buf;
 371                 target = xrealloc(target, xstate.mem_output_size + 1);
 372                 target[xstate.mem_output_size] = '\0';
 373 #endif
 374         }
 375         create_or_remember_link(symlink_placeholders,
 376                         target,
 377                         dst_fn,
 378                         0);
 379         free(target);
 380 }
 381 #endif
 382
 383 static void unzip_extract(zip_header_t *zip, int dst_fd)
 384 {
 385         transformer_state_t xstate;
 386
 387         if (zip->fmt.method == 0) {
 388                 /* Method 0 - stored (not compressed) */
 389                 off_t size = zip->fmt.ucmpsize;
 390                 if (size)
 391                         bb_copyfd_exact_size(zip_fd, dst_fd, size);
 392                 return;
 393         }
 394
 395         init_transformer_state(&xstate);
 396         xstate.bytes_in = zip->fmt.cmpsize;
 397         xstate.src_fd = zip_fd;
 398         xstate.dst_fd = dst_fd;
 399         if (zip->fmt.method == 8) {
 400                 /* Method 8 - inflate */
 401                 if (inflate_unzip(&xstate) < 0)
 402                         bb_error_msg_and_die("inflate error");
 403                 /* Validate decompression - crc */
 404                 if (zip->fmt.crc32 != (xstate.crc32 ^ 0xffffffffL)) {
 405                         bb_error_msg_and_die("crc error");
 406                 }
 407         }
 408 #if ENABLE_FEATURE_UNZIP_BZIP2
 409         else if (zip->fmt.method == 12) {
 410                 /* Tested. Unpacker reads too much, but we use CDF
 411                  * and will seek to the correct beginning of next file.
 412                  */
 413                 xstate.bytes_out = unpack_bz2_stream(&xstate);
 414                 if (xstate.bytes_out < 0)
 415                         bb_error_msg_and_die("inflate error");
 416         }
 417 #endif
 418 #if ENABLE_FEATURE_UNZIP_LZMA
 419         else if (zip->fmt.method == 14) {
 420                 /* Not tested yet */
 421                 xstate.bytes_out = unpack_lzma_stream(&xstate);
 422                 if (xstate.bytes_out < 0)
 423                         bb_error_msg_and_die("inflate error");
 424         }
 425 #endif
 426 #if ENABLE_FEATURE_UNZIP_XZ
 427         else if (zip->fmt.method == 95) {
 428                 /* Not tested yet */
 429                 xstate.bytes_out = unpack_xz_stream(&xstate);
 430                 if (xstate.bytes_out < 0)
 431                         bb_error_msg_and_die("inflate error");
 432         }
 433 #endif
 434         else {
 435                 bb_error_msg_and_die("unsupported method %u", zip->fmt.method);
 436         }
 437
 438         /* Validate decompression - size */
 439         if (zip->fmt.ucmpsize != xstate.bytes_out) {
 440                 /* Don't die. Who knows, maybe len calculation
 441                  * was botched somewhere. After all, crc matched! */
 442                 bb_error_msg("bad length");
 443         }
 444 }
 445
 446 static void my_fgets80(char *buf80)
 447 {
 448         fflush_all();
 449         if (!fgets(buf80, 80, stdin)) {
 450                 bb_perror_msg_and_die("can't read standard input");
 451         }
 452 }
 453
 454 static int get_lstat_mode(const char *dst_fn)
 455 {
 456         struct stat stat_buf;
 457         if (lstat(dst_fn, &stat_buf) == -1) {
 458                 if (errno != ENOENT) {
 459                         bb_perror_msg_and_die("can't stat '%s'",
 460                                 dst_fn
 461                         );
 462                 }
 463                 /* File does not exist */
 464                 return -1;
 465         }
 466         return stat_buf.st_mode;
 467 }
 468
 469 int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 470 int unzip_main(int argc, char **argv)
 471 {
 472         enum {
 473                 OPT_l = (1 << 0),
 474                 OPT_x = (1 << 1),
 475                 OPT_j = (1 << 2),
 476         };
 477         unsigned opts;
 478         smallint quiet = 0;
 479         IF_NOT_FEATURE_UNZIP_CDF(const) smallint verbose = 0;
 480         enum { O_PROMPT, O_NEVER, O_ALWAYS };
 481         smallint overwrite = O_PROMPT;
 482         uint32_t cdf_offset;
 483         unsigned long total_usize;
 484         unsigned long total_size;
 485         unsigned total_entries;
 486         int dst_fd = -1;
 487         char *src_fn = NULL;
 488         char *dst_fn = NULL;
 489         llist_t *zaccept = NULL;
 490         llist_t *zreject = NULL;
 491         char *base_dir = NULL;
 492 #if ENABLE_FEATURE_UNZIP_CDF
 493         llist_t *symlink_placeholders = NULL;
 494 #endif
 495         int i;
 496         char key_buf[80]; /* must match size used by my_fgets80 */
 497
 498 /* -q, -l and -v: UnZip 5.52 of 28 February 2005, by Info-ZIP:
 499  *
 500  * # /usr/bin/unzip -qq -v decompress_unlzma.i.zip
 501  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 502  * # /usr/bin/unzip -q -v decompress_unlzma.i.zip
 503  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 504  * --------  ------  ------- -----   ----   ----   ------    ----
 505  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 506  * --------          -------  ---                            -------
 507  *   204372            35278  83%                            1 file
 508  * # /usr/bin/unzip -v decompress_unlzma.i.zip
 509  * Archive:  decompress_unlzma.i.zip
 510  *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 511  * --------  ------  ------- -----   ----   ----   ------    ----
 512  *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 513  * --------          -------  ---                            -------
 514  *   204372            35278  83%                            1 file
 515  * # unzip -v decompress_unlzma.i.zip
 516  * Archive:  decompress_unlzma.i.zip
 517  *   Length     Date   Time    Name
 518  *  --------    ----   ----    ----
 519  *    204372  09-06-09 14:23   decompress_unlzma.i
 520  *  --------                   -------
 521  *    204372                   1 files
 522  * # /usr/bin/unzip -l -qq decompress_unlzma.i.zip
 523  *    204372  09-06-09 14:23   decompress_unlzma.i
 524  * # /usr/bin/unzip -l -q decompress_unlzma.i.zip
 525  *   Length     Date   Time    Name
 526  *  --------    ----   ----    ----
 527  *    204372  09-06-09 14:23   decompress_unlzma.i
 528  *  --------                   -------
 529  *    204372                   1 file
 530  * # /usr/bin/unzip -l decompress_unlzma.i.zip
 531  * Archive:  decompress_unlzma.i.zip
 532  *   Length     Date   Time    Name
 533  *  --------    ----   ----    ----
 534  *    204372  09-06-09 14:23   decompress_unlzma.i
 535  *  --------                   -------
 536  *    204372                   1 file
 537  */
 538
 539         opts = 0;
 540         /* '-' makes getopt return 1 for non-options */
 541         while ((i = getopt(argc, argv, "-d:lnopqxjv")) != -1) {
 542                 switch (i) {
 543                 case 'd':  /* Extract to base directory */
 544                         base_dir = optarg;
 545                         break;
 546
 547                 case 'l': /* List */
 548                         opts |= OPT_l;
 549                         break;
 550
 551                 case 'n': /* Never overwrite existing files */
 552                         overwrite = O_NEVER;
 553                         break;
 554
 555                 case 'o': /* Always overwrite existing files */
 556                         overwrite = O_ALWAYS;
 557                         break;
 558
 559                 case 'p': /* Extract files to stdout and fall through to set verbosity */
 560                         dst_fd = STDOUT_FILENO;
 561
 562                 case 'q': /* Be quiet */
 563                         quiet++;
 564                         break;
 565
 566                 case 'v': /* Verbose list */
 567                         IF_FEATURE_UNZIP_CDF(verbose++;)
 568                         opts |= OPT_l;
 569                         break;
 570
 571                 case 'x':
 572                         opts |= OPT_x;
 573                         break;
 574
 575                 case 'j':
 576                         opts |= OPT_j;
 577                         break;
 578
 579                 case 1:
 580                         if (!src_fn) {
 581                                 /* The zip file */
 582                                 /* +5: space for ".zip" and NUL */
 583                                 src_fn = xmalloc(strlen(optarg) + 5);
 584                                 strcpy(src_fn, optarg);
 585                         } else if (!(opts & OPT_x)) {
 586                                 /* Include files */
 587                                 llist_add_to(&zaccept, optarg);
 588                         } else {
 589                                 /* Exclude files */
 590                                 llist_add_to(&zreject, optarg);
 591                         }
 592                         break;
 593
 594                 default:
 595                         bb_show_usage();
 596                 }
 597         }
 598
 599 #ifndef __GLIBC__
 600         /*
 601          * This code is needed for non-GNU getopt
 602          * which doesn't understand "-" in option string.
 603          * The -x option won't work properly in this case:
 604          * "unzip a.zip q -x w e" will be interpreted as
 605          * "unzip a.zip q w e -x" = "unzip a.zip q w e"
 606          */
 607         argv += optind;
 608         if (argv[0]) {
 609                 /* +5: space for ".zip" and NUL */
 610                 src_fn = xmalloc(strlen(argv[0]) + 5);
 611                 strcpy(src_fn, argv[0]);
 612                 while (*++argv)
 613                         llist_add_to(&zaccept, *argv);
 614         }
 615 #endif
 616
 617         if (!src_fn) {
 618                 bb_show_usage();
 619         }
 620
 621         /* Open input file */
 622         if (LONE_DASH(src_fn)) {
 623                 xdup2(STDIN_FILENO, zip_fd);
 624                 /* Cannot use prompt mode since zip data is arriving on STDIN */
 625                 if (overwrite == O_PROMPT)
 626                         overwrite = O_NEVER;
 627         } else {
 628                 static const char extn[][5] ALIGN1 = { ".zip", ".ZIP" };
 629                 char *ext = src_fn + strlen(src_fn);
 630                 int src_fd;
 631
 632                 i = 0;
 633                 for (;;) {
 634                         src_fd = open(src_fn, O_RDONLY);
 635                         if (src_fd >= 0)
 636                                 break;
 637                         if (++i > 2) {
 638                                 *ext = '\0';
 639                                 bb_error_msg_and_die("can't open %s[.zip]",
 640                                         src_fn
 641                                 );
 642                         }
 643                         strcpy(ext, extn[i - 1]);
 644                 }
 645                 xmove_fd(src_fd, zip_fd);
 646         }
 647
 648         /* Change dir if necessary */
 649         if (base_dir)
 650                 xchdir(base_dir);
 651
 652         if (quiet <= 1) { /* not -qq */
 653                 if (quiet == 0) {
 654                         printf("Archive:  %s\n",
 655                                 printable_string(src_fn)
 656                         );
 657                 }
 658                 if (opts & OPT_l) {
 659                         puts(verbose ?
 660                                 " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 661                                 "--------  ------  ------- ---- ---------- ----- --------  ----"
 662                                 :
 663                                 "  Length      Date    Time    Name\n"
 664                                 "---------  ---------- -----   ----"
 665                                 );
 666                 }
 667         }
 668
 669 /* Example of an archive with one 0-byte long file named 'z'
 670  * created by Zip 2.31 on Unix:
 671  * 0000 [50 4b]03 04 0a 00 00 00 00 00 42 1a b8 3c 00 00 |PK........B..<..|
 672  *       sig........ vneed flags compr mtime mdate crc32>
 673  * 0010  00 00 00 00 00 00 00 00 00 00 01 00 15 00 7a 55 |..............zU|
 674  *      >..... csize...... usize...... fnlen exlen fn ex>
 675  * 0020  54 09 00 03 cc d3 f9 4b cc d3 f9 4b 55 78 04 00 |T......K...KUx..|
 676  *      >tra_field......................................
 677  * 0030  00 00 00 00[50 4b]01 02 17 03 0a 00 00 00 00 00 |....PK..........|
 678  *       ........... sig........ vmade vneed flags compr
 679  * 0040  42 1a b8 3c 00 00 00 00 00 00 00 00 00 00 00 00 |B..<............|
 680  *       mtime mdate crc32...... csize...... usize......
 681  * 0050  01 00 0d 00 00 00 00 00 00 00 00 00 a4 81 00 00 |................|
 682  *       fnlen exlen clen. dnum. iattr eattr...... relofs> (eattr = rw-r--r--)
 683  * 0060  00 00 7a 55 54 05 00 03 cc d3 f9 4b 55 78 00 00 |..zUT......KUx..|
 684  *      >..... fn extra_field...........................
 685  * 0070 [50 4b]05 06 00 00 00 00 01 00 01 00 3c 00 00 00 |PK..........<...|
 686  * 0080  34 00 00 00 00 00                               |4.....|
 687  */
 688         total_usize = 0;
 689         total_size = 0;
 690         total_entries = 0;
 691         cdf_offset = find_cdf_offset(); /* try to seek to the end, find CDE and CDF start */
 692         while (1) {
 693                 zip_header_t zip;
 694                 mode_t dir_mode = 0777;
 695 #if ENABLE_FEATURE_UNZIP_CDF
 696                 mode_t file_mode = 0666;
 697 #endif
 698
 699                 if (!ENABLE_FEATURE_UNZIP_CDF || cdf_offset == BAD_CDF_OFFSET) {
 700                         /* Normally happens when input is unseekable.
 701                          *
 702                          * Valid ZIP file has Central Directory at the end
 703                          * with central directory file headers (CDFs).
 704                          * After it, there is a Central Directory End structure.
 705                          * CDFs identify what files are in the ZIP and where
 706                          * they are located. This allows ZIP readers to load
 707                          * the list of files without reading the entire ZIP archive.
 708                          * ZIP files may be appended to, only files specified in
 709                          * the CD are valid. Scanning for local file headers is
 710                          * not a correct algorithm.
 711                          *
 712                          * We try to do the above, and resort to "linear" reading
 713                          * of ZIP file only if seek failed or CDE wasn't found.
 714                          */
 715                         uint32_t magic;
 716
 717                         /* Check magic number */
 718                         xread(zip_fd, &magic, 4);
 719                         /* CDF item? Assume there are no more files, exit */
 720                         if (magic == ZIP_CDF_MAGIC) {
 721                                 dbg("got ZIP_CDF_MAGIC");
 722                                 break;
 723                         }
 724                         /* Data descriptor? It was a streaming file, go on */
 725                         if (magic == ZIP_DD_MAGIC) {
 726                                 dbg("got ZIP_DD_MAGIC");
 727                                 /* skip over duplicate crc32, cmpsize and ucmpsize */
 728                                 unzip_skip(3 * 4);
 729                                 continue;
 730                         }
 731                         if (magic != ZIP_FILEHEADER_MAGIC)
 732                                 bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
 733                         dbg("got ZIP_FILEHEADER_MAGIC");
 734
 735                         xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 736                         FIX_ENDIANNESS_ZIP(zip);
 737                         if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 738                                 bb_error_msg_and_die("zip flag %s is not supported",
 739                                         "8 (streaming)");
 740                         }
 741                 }
 742 #if ENABLE_FEATURE_UNZIP_CDF
 743                 else {
 744                         /* cdf_offset is valid (and we know the file is seekable) */
 745                         cdf_header_t cdf;
 746                         cdf_offset = read_next_cdf(cdf_offset, &cdf);
 747                         if (cdf_offset == 0) /* EOF? */
 748                                 break;
 749 # if 1
 750                         xlseek(zip_fd,
 751                                 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4,
 752                                 SEEK_SET);
 753                         xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 754                         FIX_ENDIANNESS_ZIP(zip);
 755                         if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 756                                 /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
 757                                  * only from Central Directory.
 758                                  */
 759                                 zip.fmt.crc32    = cdf.fmt.crc32;
 760                                 zip.fmt.cmpsize  = cdf.fmt.cmpsize;
 761                                 zip.fmt.ucmpsize = cdf.fmt.ucmpsize;
 762                         }
 763 // Seen in some zipfiles: central directory 9 byte extra field contains
 764 // a subfield with ID 0x5455 and 5 data bytes, which is a Unix-style UTC mtime.
 765 // Local header version:
 766 //  u16 0x5455 ("UT")
 767 //  u16 size (1 + 4 * n)
 768 //  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 769 //  u32 mtime
 770 //  u32 atime
 771 //  u32 ctime
 772 // Central header version:
 773 //  u16 0x5455 ("UT")
 774 //  u16 size (5 (or 1?))
 775 //  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 776 //  u32 mtime (CDF does not store atime/ctime)
 777 # else
 778                         /* CDF has the same data as local header, no need to read the latter...
 779                          * ...not really. An archive was seen with cdf.extra_len == 6 but
 780                          * zip.extra_len == 0.
 781                          */
 782                         memcpy(&zip.fmt.version,
 783                                 &cdf.fmt.version_needed, ZIP_HEADER_LEN);
 784                         xlseek(zip_fd,
 785                                 SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4 + ZIP_HEADER_LEN,
 786                                 SEEK_SET);
 787 # endif
 788                         if ((cdf.fmt.version_made_by >> 8) == 3) {
 789                                 /* This archive is created on Unix */
 790                                 dir_mode = file_mode = (cdf.fmt.external_attributes >> 16);
 791                         }
 792                 }
 793 #endif
 794
 795                 if (zip.fmt.zip_flags & SWAP_LE16(0x0001)) {
 796                         /* 0x0001 - encrypted */
 797                         bb_error_msg_and_die("zip flag %s is not supported",
 798                                         "1 (encryption)");
 799                 }
 800                 dbg("File cmpsize:0x%x extra_len:0x%x ucmpsize:0x%x",
 801                         (unsigned)zip.fmt.cmpsize,
 802                         (unsigned)zip.fmt.extra_len,
 803                         (unsigned)zip.fmt.ucmpsize
 804                 );
 805
 806                 /* Read filename */
 807                 free(dst_fn);
 808                 die_if_bad_fnamesize(zip.fmt.filename_len);
 809                 dst_fn = xzalloc(zip.fmt.filename_len + 1);
 810                 xread(zip_fd, dst_fn, zip.fmt.filename_len);
 811                 /* Skip extra header bytes */
 812                 unzip_skip(zip.fmt.extra_len);
 813
 814                 /* Guard against "/abspath", "/../" and similar attacks */
 815                 overlapping_strcpy(dst_fn, strip_unsafe_prefix(dst_fn));
 816
 817                 /* Filter zip entries */
 818                 if (find_list_entry(zreject, dst_fn)
 819                  || (zaccept && !find_list_entry(zaccept, dst_fn))
 820                 ) { /* Skip entry */
 821                         goto skip_cmpsize;
 822                 }
 823
 824                 if (opts & OPT_l) {
 825                         /* List entry */
 826                         char dtbuf[sizeof("mm-dd-yyyy hh:mm")];
 827                         sprintf(dtbuf, "%02u-%02u-%04u %02u:%02u",
 828                                 (zip.fmt.moddate >> 5) & 0xf,  // mm: 0x01e0
 829                                 (zip.fmt.moddate)      & 0x1f, // dd: 0x001f
 830                                 (zip.fmt.moddate >> 9) + 1980, // yy: 0xfe00
 831                                 (zip.fmt.modtime >> 11),       // hh: 0xf800
 832                                 (zip.fmt.modtime >> 5) & 0x3f  // mm: 0x07e0
 833                                 // seconds/2 not shown, encoded in -- 0x001f
 834                         );
 835                         if (!verbose) {
 836                                 //      "  Length      Date    Time    Name\n"
 837                                 //      "---------  ---------- -----   ----"
 838                                 printf(       "%9u  " "%s   "         "%s\n",
 839                                         (unsigned)zip.fmt.ucmpsize,
 840                                         dtbuf,
 841                                         printable_string(dst_fn)
 842                                 );
 843                         } else {
 844                                 char method6[7];
 845                                 unsigned long percents;
 846
 847                                 sprintf(method6, "%6u", zip.fmt.method);
 848                                 if (zip.fmt.method == 0) {
 849                                         strcpy(method6, "Stored");
 850                                 }
 851                                 if (zip.fmt.method == 8) {
 852                                         strcpy(method6, "Defl:N");
 853                                         /* normal, maximum, fast, superfast */
 854                                         IF_DESKTOP(method6[5] = "NXFS"[(zip.fmt.zip_flags >> 1) & 3];)
 855                                 }
 856                                 percents = zip.fmt.ucmpsize - zip.fmt.cmpsize;
 857                                 if ((int32_t)percents < 0)
 858                                         percents = 0; /* happens if ucmpsize < cmpsize */
 859                                 percents = percents * 100;
 860                                 if (zip.fmt.ucmpsize)
 861                                         percents /= zip.fmt.ucmpsize;
 862                                 //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 863                                 //      "--------  ------  ------- ---- ---------- ----- --------  ----"
 864                                 printf(      "%8u  %s"        "%9u%4u%% " "%s "         "%08x  "  "%s\n",
 865                                         (unsigned)zip.fmt.ucmpsize,
 866                                         method6,
 867                                         (unsigned)zip.fmt.cmpsize,
 868                                         (unsigned)percents,
 869                                         dtbuf,
 870                                         zip.fmt.crc32,
 871                                         printable_string(dst_fn)
 872                                 );
 873                                 total_size += zip.fmt.cmpsize;
 874                         }
 875                         total_usize += zip.fmt.ucmpsize;
 876                         goto skip_cmpsize;
 877                 }
 878
 879                 if (dst_fd == STDOUT_FILENO) {
 880                         /* Extracting to STDOUT */
 881                         goto do_extract;
 882                 }
 883
 884                 /* Strip paths (after -l: unzip -lj a.zip lists full names) */
 885                 if (opts & OPT_j)
 886                         overlapping_strcpy(dst_fn, bb_basename(dst_fn));
 887                 /* Did this strip everything ("DIR/" case)? Then skip */
 888                 if (!dst_fn[0])
 889                         goto skip_cmpsize;
 890
 891                 if (last_char_is(dst_fn, '/')) {
 892                         int mode;
 893
 894                         /* Extract directory */
 895                         mode = get_lstat_mode(dst_fn);
 896                         if (mode == -1) { /* ENOENT */
 897                                 if (!quiet) {
 898                                         printf("   creating: %s\n", printable_string(dst_fn));
 899                                 }
 900                                 unzip_create_leading_dirs(dst_fn);
 901                                 if (bb_make_directory(dst_fn, dir_mode, FILEUTILS_IGNORE_CHMOD_ERR)) {
 902                                         xfunc_die();
 903                                 }
 904                         } else {
 905                                 if (!S_ISDIR(mode)) {
 906                                         bb_error_msg_and_die("'%s' exists but is not a %s",
 907                                                 printable_string(dst_fn),
 908                                                 "directory"
 909                                         );
 910                                 }
 911                         }
 912                         goto skip_cmpsize;
 913                 }
 914  check_file:
 915                 /* Does target file already exist? */
 916                 {
 917                         int mode = get_lstat_mode(dst_fn);
 918                         if (mode == -1) {
 919                                 /* ENOENT: does not exist */
 920                                 goto do_open_and_extract;
 921                         }
 922                         if (overwrite == O_NEVER) {
 923                                 goto skip_cmpsize;
 924                         }
 925                         if (!S_ISREG(mode)) {
 926  fishy:
 927                                 bb_error_msg_and_die("'%s' exists but is not a %s",
 928                                         printable_string(dst_fn),
 929                                         "regular file"
 930                                 );
 931                         }
 932                         if (overwrite == O_ALWAYS) {
 933                                 goto do_open_and_extract;
 934                         }
 935                         printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ",
 936                                 printable_string(dst_fn)
 937                         );
 938                         my_fgets80(key_buf);
 939                         /* User input could take a long time. Is it still a regular file? */
 940                         mode = get_lstat_mode(dst_fn);
 941                         if (!S_ISREG(mode))
 942                                 goto fishy;
 943                 }
 944
 945                 /* Extract (or skip) it */
 946                 switch (key_buf[0]) {
 947                 case 'A':
 948                         overwrite = O_ALWAYS;
 949                 case 'y': /* Open file and fall into unzip */
 950  do_open_and_extract:
 951                         unzip_create_leading_dirs(dst_fn);
 952 #if ENABLE_FEATURE_UNZIP_CDF
 953                         dst_fd = -1;
 954                         if (!S_ISLNK(file_mode)) {
 955                                 dst_fd = xopen3(dst_fn,
 956                                         O_WRONLY | O_CREAT | O_TRUNC | O_NOFOLLOW,
 957                                         file_mode);
 958                         }
 959 #else
 960                         /* O_NOFOLLOW defends against symlink attacks */
 961                         dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC | O_NOFOLLOW);
 962 #endif
 963  do_extract:
 964                         if (!quiet) {
 965                                 printf(/* zip.fmt.method == 0
 966                                         ? " extracting: %s\n"
 967                                         : */ "  inflating: %s\n",
 968                                         printable_string(dst_fn)
 969                                 );
 970                         }
 971 #if ENABLE_FEATURE_UNZIP_CDF
 972                         if (S_ISLNK(file_mode)) {
 973                                 if (dst_fd != STDOUT_FILENO) /* not -p? */
 974                                         unzip_extract_symlink(&symlink_placeholders, &zip, dst_fn);
 975                         } else
 976 #endif
 977                         {
 978                                 unzip_extract(&zip, dst_fd);
 979                                 if (dst_fd != STDOUT_FILENO) {
 980                                         /* closing STDOUT is potentially bad for future business */
 981                                         close(dst_fd);
 982                                 }
 983                         }
 984                         break;
 985
 986                 case 'N':
 987                         overwrite = O_NEVER;
 988                 case 'n': /* Skip entry data */
 989  skip_cmpsize:
 990                         unzip_skip(zip.fmt.cmpsize);
 991                         break;
 992
 993                 case 'r':
 994                         /* Prompt for new name */
 995                         printf("new name: ");
 996                         my_fgets80(key_buf);
 997                         free(dst_fn);
 998                         dst_fn = xstrdup(key_buf);
 999                         chomp(dst_fn);
1000                         goto check_file;
1001
1002                 default:
1003                         printf("error: invalid response [%c]\n", (char)key_buf[0]);
1004                         goto check_file;
1005                 }
1006
1007                 total_entries++;
1008         }
1009
1010 #if ENABLE_FEATURE_UNZIP_CDF
1011         create_links_from_list(symlink_placeholders);
1012 #endif
1013
1014         if ((opts & OPT_l) && quiet <= 1) {
1015                 if (!verbose) {
1016                         //      "  Length      Date    Time    Name\n"
1017                         //      "---------  ---------- -----   ----"
1018                         printf( " --------%21s"               "-------\n"
1019                                      "%9lu%21s"               "%u files\n",
1020                                 "",
1021                                 total_usize, "", total_entries);
1022                 } else {
1023                         unsigned long percents = total_usize - total_size;
1024                         if ((long)percents < 0)
1025                                 percents = 0; /* happens if usize < size */
1026                         percents = percents * 100;
1027                         if (total_usize)
1028                                 percents /= total_usize;
1029                         //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
1030                         //      "--------  ------  ------- ---- ---------- ----- --------  ----"
1031                         printf( "--------          ------- ----%28s"                      "----\n"
1032                                 "%8lu"              "%17lu%4u%%%28s"                      "%u files\n",
1033                                 "",
1034                                 total_usize, total_size, (unsigned)percents, "",
1035                                 total_entries);
1036                 }
1037         }
1038
1039         return 0;
1040 }