busybox/archival/unzip.c
<<
>>
Prefs
   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      Write to stdout"
  67//usage:     "\n        -t      Test"
  68//usage:     "\n        -q      Quiet"
  69//usage:     "\n        -x FILE Exclude FILEs"
  70//usage:     "\n        -d DIR  Extract into DIR"
  71
  72#include "libbb.h"
  73#include "bb_archive.h"
  74
  75#if 0
  76# define dbg(...) bb_error_msg(__VA_ARGS__)
  77#else
  78# define dbg(...) ((void)0)
  79#endif
  80
  81enum {
  82#if BB_BIG_ENDIAN
  83        ZIP_FILEHEADER_MAGIC = 0x504b0304,
  84        ZIP_CDF_MAGIC        = 0x504b0102, /* CDF item */
  85        ZIP_CDE_MAGIC        = 0x504b0506, /* End of CDF */
  86        ZIP64_CDE_MAGIC      = 0x504b0606, /* End of Zip64 CDF */
  87        ZIP_DD_MAGIC         = 0x504b0708,
  88#else
  89        ZIP_FILEHEADER_MAGIC = 0x04034b50,
  90        ZIP_CDF_MAGIC        = 0x02014b50,
  91        ZIP_CDE_MAGIC        = 0x06054b50,
  92        ZIP64_CDE_MAGIC      = 0x06064b50,
  93        ZIP_DD_MAGIC         = 0x08074b50,
  94#endif
  95};
  96
  97#define ZIP_HEADER_LEN 26
  98
  99typedef union {
 100        uint8_t raw[ZIP_HEADER_LEN];
 101        struct {
 102                uint16_t version;               /* 0-1 */
 103                uint16_t zip_flags;             /* 2-3 */
 104                uint16_t method;                /* 4-5 */
 105                uint16_t modtime;               /* 6-7 */
 106                uint16_t moddate;               /* 8-9 */
 107                uint32_t crc32 PACKED;          /* 10-13 */
 108                uint32_t cmpsize PACKED;        /* 14-17 */
 109                uint32_t ucmpsize PACKED;       /* 18-21 */
 110                uint16_t filename_len;          /* 22-23 */
 111                uint16_t extra_len;             /* 24-25 */
 112                /* filename follows (not NUL terminated) */
 113                /* extra field follows */
 114                /* data follows */
 115        } fmt PACKED;
 116} zip_header_t; /* PACKED - gcc 4.2.1 doesn't like it (spews warning) */
 117
 118#define FIX_ENDIANNESS_ZIP(zip) \
 119do { if (BB_BIG_ENDIAN) { \
 120        (zip).fmt.method        = SWAP_LE16((zip).fmt.method      ); \
 121        (zip).fmt.crc32         = SWAP_LE32((zip).fmt.crc32       ); \
 122        (zip).fmt.cmpsize       = SWAP_LE32((zip).fmt.cmpsize     ); \
 123        (zip).fmt.ucmpsize      = SWAP_LE32((zip).fmt.ucmpsize    ); \
 124        (zip).fmt.filename_len  = SWAP_LE16((zip).fmt.filename_len); \
 125        (zip).fmt.extra_len     = SWAP_LE16((zip).fmt.extra_len   ); \
 126}} while (0)
 127
 128#define CDF_HEADER_LEN 42
 129
 130typedef union {
 131        uint8_t raw[CDF_HEADER_LEN];
 132        struct {
 133                /* uint32_t signature; 50 4b 01 02 */
 134                uint16_t version_made_by;       /* 0-1 */
 135                uint16_t version_needed;        /* 2-3 */
 136                uint16_t cdf_flags;             /* 4-5 */
 137                uint16_t method;                /* 6-7 */
 138                uint16_t modtime;               /* 8-9 */
 139                uint16_t moddate;               /* 10-11 */
 140                uint32_t crc32;                 /* 12-15 */
 141                uint32_t cmpsize;               /* 16-19 */
 142                uint32_t ucmpsize;              /* 20-23 */
 143                uint16_t filename_len;          /* 24-25 */
 144                uint16_t extra_len;             /* 26-27 */
 145                uint16_t file_comment_length;   /* 28-29 */
 146                uint16_t disk_number_start;     /* 30-31 */
 147                uint16_t internal_attributes;   /* 32-33 */
 148                uint32_t external_attributes PACKED; /* 34-37 */
 149                uint32_t relative_offset_of_local_header PACKED; /* 38-41 */
 150                /* filename follows (not NUL terminated) */
 151                /* extra field follows */
 152                /* file comment follows */
 153        } fmt PACKED;
 154} cdf_header_t;
 155
 156#define FIX_ENDIANNESS_CDF(cdf) \
 157do { if (BB_BIG_ENDIAN) { \
 158        (cdf).fmt.version_made_by = SWAP_LE16((cdf).fmt.version_made_by); \
 159        (cdf).fmt.version_needed  = SWAP_LE16((cdf).fmt.version_needed ); \
 160        (cdf).fmt.method          = SWAP_LE16((cdf).fmt.method         ); \
 161        (cdf).fmt.modtime         = SWAP_LE16((cdf).fmt.modtime        ); \
 162        (cdf).fmt.moddate         = SWAP_LE16((cdf).fmt.moddate        ); \
 163        (cdf).fmt.crc32           = SWAP_LE32((cdf).fmt.crc32          ); \
 164        (cdf).fmt.cmpsize         = SWAP_LE32((cdf).fmt.cmpsize        ); \
 165        (cdf).fmt.ucmpsize        = SWAP_LE32((cdf).fmt.ucmpsize       ); \
 166        (cdf).fmt.filename_len    = SWAP_LE16((cdf).fmt.filename_len   ); \
 167        (cdf).fmt.extra_len       = SWAP_LE16((cdf).fmt.extra_len      ); \
 168        (cdf).fmt.file_comment_length = SWAP_LE16((cdf).fmt.file_comment_length); \
 169        (cdf).fmt.external_attributes = SWAP_LE32((cdf).fmt.external_attributes); \
 170}} while (0)
 171
 172#define CDE_LEN 16
 173
 174typedef union {
 175        uint8_t raw[CDE_LEN];
 176        struct {
 177                /* uint32_t signature; 50 4b 05 06 */
 178                uint16_t this_disk_no;
 179                uint16_t disk_with_cdf_no;
 180                uint16_t cdf_entries_on_this_disk;
 181                uint16_t cdf_entries_total;
 182                uint32_t cdf_size;
 183                uint32_t cdf_offset;
 184                /* uint16_t archive_comment_length; */
 185                /* archive comment follows */
 186        } fmt PACKED;
 187} cde_t;
 188
 189#define FIX_ENDIANNESS_CDE(cde) \
 190do { if (BB_BIG_ENDIAN) { \
 191        (cde).fmt.cdf_offset = SWAP_LE32((cde).fmt.cdf_offset); \
 192}} while (0)
 193
 194struct BUG {
 195        /* Check the offset of the last element, not the length.  This leniency
 196         * allows for poor packing, whereby the overall struct may be too long,
 197         * even though the elements are all in the right place.
 198         */
 199        char BUG_zip_header_must_be_26_bytes[
 200                offsetof(zip_header_t, fmt.extra_len) + 2
 201                        == ZIP_HEADER_LEN ? 1 : -1];
 202        char BUG_cdf_header_must_be_42_bytes[
 203                offsetof(cdf_header_t, fmt.relative_offset_of_local_header) + 4
 204                        == CDF_HEADER_LEN ? 1 : -1];
 205        char BUG_cde_must_be_16_bytes[
 206                sizeof(cde_t) == CDE_LEN ? 1 : -1];
 207};
 208
 209
 210enum { zip_fd = 3 };
 211
 212
 213/* This value means that we failed to find CDF */
 214#define BAD_CDF_OFFSET ((uint32_t)0xffffffff)
 215
 216#if !ENABLE_FEATURE_UNZIP_CDF
 217
 218# define find_cdf_offset() BAD_CDF_OFFSET
 219
 220#else
 221/* Seen in the wild:
 222 * Self-extracting PRO2K3XP_32.exe contains 19078464 byte zip archive,
 223 * where CDE was nearly 48 kbytes before EOF.
 224 * (Surprisingly, it also apparently has *another* CDE structure
 225 * closer to the end, with bogus cdf_offset).
 226 * To make extraction work, bumped PEEK_FROM_END from 16k to 64k.
 227 */
 228#define PEEK_FROM_END (64*1024)
 229/* NB: does not preserve file position! */
 230static uint32_t find_cdf_offset(void)
 231{
 232        cde_t cde;
 233        unsigned char *buf;
 234        unsigned char *p;
 235        off_t end;
 236        uint32_t found;
 237
 238        end = lseek(zip_fd, 0, SEEK_END);
 239        if (end == (off_t) -1)
 240                return BAD_CDF_OFFSET;
 241
 242        end -= PEEK_FROM_END;
 243        if (end < 0)
 244                end = 0;
 245
 246        dbg("Looking for cdf_offset starting from 0x%"OFF_FMT"x", end);
 247        xlseek(zip_fd, end, SEEK_SET);
 248        buf = xzalloc(PEEK_FROM_END);
 249        full_read(zip_fd, buf, PEEK_FROM_END);
 250
 251        found = BAD_CDF_OFFSET;
 252        p = buf;
 253        while (p <= buf + PEEK_FROM_END - CDE_LEN - 4) {
 254                if (*p != 'P') {
 255                        p++;
 256                        continue;
 257                }
 258                if (*++p != 'K')
 259                        continue;
 260                if (*++p != 5)
 261                        continue;
 262                if (*++p != 6)
 263                        continue;
 264                /* we found CDE! */
 265                memcpy(cde.raw, p + 1, CDE_LEN);
 266                dbg("cde.this_disk_no:%d",             cde.fmt.this_disk_no            );
 267                dbg("cde.disk_with_cdf_no:%d",         cde.fmt.disk_with_cdf_no        );
 268                dbg("cde.cdf_entries_on_this_disk:%d", cde.fmt.cdf_entries_on_this_disk);
 269                dbg("cde.cdf_entries_total:%d",        cde.fmt.cdf_entries_total       );
 270                dbg("cde.cdf_size:%d",                 cde.fmt.cdf_size                );
 271                dbg("cde.cdf_offset:%x",               cde.fmt.cdf_offset              );
 272                FIX_ENDIANNESS_CDE(cde);
 273                /*
 274                 * I've seen .ZIP files with seemingly valid CDEs
 275                 * where cdf_offset points past EOF - ??
 276                 * This check ignores such CDEs:
 277                 */
 278                if (cde.fmt.cdf_offset < end + (p - buf)) {
 279                        found = cde.fmt.cdf_offset;
 280                        dbg("Possible cdf_offset:0x%x at 0x%"OFF_FMT"x",
 281                                (unsigned)found, end + (p-3 - buf));
 282                        dbg("  cdf_offset+cdf_size:0x%x",
 283                                (unsigned)(found + SWAP_LE32(cde.fmt.cdf_size)));
 284                        /*
 285                         * We do not "break" here because only the last CDE is valid.
 286                         * I've seen a .zip archive which contained a .zip file,
 287                         * uncompressed, and taking the first CDE was using
 288                         * the CDE inside that file!
 289                         */
 290                }
 291        }
 292        free(buf);
 293        dbg("Found cdf_offset:0x%x", (unsigned)found);
 294        return found;
 295};
 296
 297static uint32_t read_next_cdf(uint32_t cdf_offset, cdf_header_t *cdf)
 298{
 299        uint32_t magic;
 300
 301        if (cdf_offset == BAD_CDF_OFFSET)
 302                return cdf_offset;
 303
 304        dbg("Reading CDF at 0x%x", (unsigned)cdf_offset);
 305        xlseek(zip_fd, cdf_offset, SEEK_SET);
 306        xread(zip_fd, &magic, 4);
 307        /* Central Directory End? Assume CDF has ended.
 308         * (more correct method is to use cde.cdf_entries_total counter)
 309         */
 310        if (magic == ZIP_CDE_MAGIC) {
 311                dbg("got ZIP_CDE_MAGIC");
 312                return 0; /* EOF */
 313        }
 314        if (magic == ZIP64_CDE_MAGIC) { /* seen in .zip with >4GB files */
 315                dbg("got ZIP64_CDE_MAGIC");
 316                return 0; /* EOF */
 317        }
 318        xread(zip_fd, cdf->raw, CDF_HEADER_LEN);
 319
 320        FIX_ENDIANNESS_CDF(*cdf);
 321        dbg("  magic:%08x filename_len:%u extra_len:%u file_comment_length:%u",
 322                magic,
 323                (unsigned)cdf->fmt.filename_len,
 324                (unsigned)cdf->fmt.extra_len,
 325                (unsigned)cdf->fmt.file_comment_length
 326        );
 327//TODO: require that magic == ZIP_CDF_MAGIC?
 328
 329        cdf_offset += 4 + CDF_HEADER_LEN
 330                + cdf->fmt.filename_len
 331                + cdf->fmt.extra_len
 332                + cdf->fmt.file_comment_length;
 333
 334        dbg("Next cdf_offset 0x%x", cdf_offset);
 335        return cdf_offset;
 336};
 337#endif
 338
 339static void die_if_bad_fnamesize(unsigned sz)
 340{
 341        if (sz > 0xfff) /* more than 4k?! no funny business please */
 342                bb_simple_error_msg_and_die("bad archive");
 343}
 344
 345static void unzip_skip(off_t skip)
 346{
 347        if (skip != 0)
 348                if (lseek(zip_fd, skip, SEEK_CUR) == (off_t)-1)
 349                        bb_copyfd_exact_size(zip_fd, -1, skip);
 350}
 351
 352static void unzip_create_leading_dirs(const char *fn)
 353{
 354        /* Create all leading directories */
 355        char *name = xstrdup(fn);
 356
 357        /* mode of -1: set mode according to umask */
 358        if (bb_make_directory(dirname(name), -1, FILEUTILS_RECUR)) {
 359                xfunc_die(); /* bb_make_directory is noisy */
 360        }
 361        free(name);
 362}
 363
 364#if ENABLE_FEATURE_UNZIP_CDF
 365static void unzip_extract_symlink(llist_t **symlink_placeholders,
 366                zip_header_t *zip,
 367                const char *dst_fn)
 368{
 369        char *target;
 370
 371        die_if_bad_fnamesize(zip->fmt.ucmpsize);
 372
 373        if (zip->fmt.method == 0) {
 374                /* Method 0 - stored (not compressed) */
 375                target = xzalloc(zip->fmt.ucmpsize + 1);
 376                xread(zip_fd, target, zip->fmt.ucmpsize);
 377        } else {
 378#if 1
 379                bb_simple_error_msg_and_die("compressed symlink is not supported");
 380#else
 381                transformer_state_t xstate;
 382                init_transformer_state(&xstate);
 383                xstate.mem_output_size_max = zip->fmt.ucmpsize;
 384                /* ...unpack... */
 385                if (!xstate.mem_output_buf)
 386                        WTF();
 387                target = xstate.mem_output_buf;
 388                target = xrealloc(target, xstate.mem_output_size + 1);
 389                target[xstate.mem_output_size] = '\0';
 390#endif
 391        }
 392        create_or_remember_link(symlink_placeholders,
 393                        target,
 394                        dst_fn,
 395                        0);
 396        free(target);
 397}
 398#endif
 399
 400static void unzip_extract(zip_header_t *zip, int dst_fd)
 401{
 402        transformer_state_t xstate;
 403
 404        if (zip->fmt.method == 0) {
 405                /* Method 0 - stored (not compressed) */
 406                off_t size = zip->fmt.ucmpsize;
 407                if (size)
 408                        bb_copyfd_exact_size(zip_fd, dst_fd, size);
 409                return;
 410        }
 411
 412        init_transformer_state(&xstate);
 413        xstate.bytes_in = zip->fmt.cmpsize;
 414        xstate.src_fd = zip_fd;
 415        xstate.dst_fd = dst_fd;
 416        if (zip->fmt.method == 8) {
 417                /* Method 8 - inflate */
 418                if (inflate_unzip(&xstate) < 0)
 419                        bb_simple_error_msg_and_die("inflate error");
 420                /* Validate decompression - crc */
 421                if (zip->fmt.crc32 != (xstate.crc32 ^ 0xffffffffL)) {
 422                        bb_simple_error_msg_and_die("crc error");
 423                }
 424        }
 425#if ENABLE_FEATURE_UNZIP_BZIP2
 426        else if (zip->fmt.method == 12) {
 427                /* Tested. Unpacker reads too much, but we use CDF
 428                 * and will seek to the correct beginning of next file.
 429                 */
 430                xstate.bytes_out = unpack_bz2_stream(&xstate);
 431                if (xstate.bytes_out < 0)
 432                        bb_simple_error_msg_and_die("inflate error");
 433        }
 434#endif
 435#if ENABLE_FEATURE_UNZIP_LZMA
 436        else if (zip->fmt.method == 14) {
 437                /* Not tested yet */
 438                xstate.bytes_out = unpack_lzma_stream(&xstate);
 439                if (xstate.bytes_out < 0)
 440                        bb_simple_error_msg_and_die("inflate error");
 441        }
 442#endif
 443#if ENABLE_FEATURE_UNZIP_XZ
 444        else if (zip->fmt.method == 95) {
 445                /* Not tested yet */
 446                xstate.bytes_out = unpack_xz_stream(&xstate);
 447                if (xstate.bytes_out < 0)
 448                        bb_simple_error_msg_and_die("inflate error");
 449        }
 450#endif
 451        else {
 452                bb_error_msg_and_die("unsupported method %u", zip->fmt.method);
 453        }
 454
 455        /* Validate decompression - size */
 456        if (zip->fmt.ucmpsize != 0xffffffff /* seen on files with >4GB uncompressed data */
 457         && zip->fmt.ucmpsize != xstate.bytes_out
 458        ) {
 459                /* Don't die. Who knows, maybe len calculation
 460                 * was botched somewhere. After all, crc matched! */
 461                bb_simple_error_msg("bad length");
 462        }
 463}
 464
 465static void my_fgets80(char *buf80)
 466{
 467        fflush_all();
 468        if (!fgets(buf80, 80, stdin)) {
 469                bb_simple_perror_msg_and_die("can't read standard input");
 470        }
 471}
 472
 473static int get_lstat_mode(const char *dst_fn)
 474{
 475        struct stat stat_buf;
 476        if (lstat(dst_fn, &stat_buf) == -1) {
 477                if (errno != ENOENT) {
 478                        bb_perror_msg_and_die("can't stat '%s'",
 479                                dst_fn
 480                        );
 481                }
 482                /* File does not exist */
 483                return -1;
 484        }
 485        return stat_buf.st_mode;
 486}
 487
 488int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 489int unzip_main(int argc, char **argv)
 490{
 491        enum {
 492                OPT_l = (1 << 0),
 493                OPT_x = (1 << 1),
 494                OPT_j = (1 << 2),
 495        };
 496        unsigned opts;
 497        smallint quiet = 0;
 498        IF_NOT_FEATURE_UNZIP_CDF(const) smallint verbose = 0;
 499        enum { O_PROMPT, O_NEVER, O_ALWAYS };
 500        smallint overwrite = O_PROMPT;
 501        uint32_t cdf_offset;
 502        unsigned long total_usize;
 503        unsigned long total_size;
 504        unsigned total_entries;
 505        int dst_fd = -1;
 506        char *src_fn = NULL;
 507        char *dst_fn = NULL;
 508        llist_t *zaccept = NULL;
 509        llist_t *zreject = NULL;
 510        char *base_dir = NULL;
 511#if ENABLE_FEATURE_UNZIP_CDF
 512        llist_t *symlink_placeholders = NULL;
 513#endif
 514        int i;
 515        char key_buf[80]; /* must match size used by my_fgets80 */
 516
 517/* -q, -l and -v: UnZip 5.52 of 28 February 2005, by Info-ZIP:
 518 *
 519 * # /usr/bin/unzip -qq -v decompress_unlzma.i.zip
 520 *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 521 * # /usr/bin/unzip -q -v decompress_unlzma.i.zip
 522 *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 523 * --------  ------  ------- -----   ----   ----   ------    ----
 524 *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 525 * --------          -------  ---                            -------
 526 *   204372            35278  83%                            1 file
 527 * # /usr/bin/unzip -v decompress_unlzma.i.zip
 528 * Archive:  decompress_unlzma.i.zip
 529 *  Length   Method    Size  Ratio   Date   Time   CRC-32    Name
 530 * --------  ------  ------- -----   ----   ----   ------    ----
 531 *   204372  Defl:N    35278  83%  09-06-09 14:23  0d056252  decompress_unlzma.i
 532 * --------          -------  ---                            -------
 533 *   204372            35278  83%                            1 file
 534 * # unzip -v decompress_unlzma.i.zip
 535 * Archive:  decompress_unlzma.i.zip
 536 *   Length     Date   Time    Name
 537 *  --------    ----   ----    ----
 538 *    204372  09-06-09 14:23   decompress_unlzma.i
 539 *  --------                   -------
 540 *    204372                   1 files
 541 * # /usr/bin/unzip -l -qq decompress_unlzma.i.zip
 542 *    204372  09-06-09 14:23   decompress_unlzma.i
 543 * # /usr/bin/unzip -l -q decompress_unlzma.i.zip
 544 *   Length     Date   Time    Name
 545 *  --------    ----   ----    ----
 546 *    204372  09-06-09 14:23   decompress_unlzma.i
 547 *  --------                   -------
 548 *    204372                   1 file
 549 * # /usr/bin/unzip -l decompress_unlzma.i.zip
 550 * Archive:  decompress_unlzma.i.zip
 551 *   Length     Date   Time    Name
 552 *  --------    ----   ----    ----
 553 *    204372  09-06-09 14:23   decompress_unlzma.i
 554 *  --------                   -------
 555 *    204372                   1 file
 556 */
 557
 558        opts = 0;
 559        /* '-' makes getopt return 1 for non-options */
 560        while ((i = getopt(argc, argv, "-d:lnotpqxjv")) != -1) {
 561                switch (i) {
 562                case 'd':  /* Extract to base directory */
 563                        base_dir = optarg;
 564                        break;
 565
 566                case 'l': /* List */
 567                        opts |= OPT_l;
 568                        break;
 569
 570                case 'n': /* Never overwrite existing files */
 571                        overwrite = O_NEVER;
 572                        break;
 573
 574                case 'o': /* Always overwrite existing files */
 575                        overwrite = O_ALWAYS;
 576                        break;
 577
 578                case 't': /* Extract files to /dev/null */
 579                        xmove_fd(xopen("/dev/null", O_WRONLY), STDOUT_FILENO);
 580                        /*fallthrough*/
 581
 582                case 'p': /* Extract files to stdout */
 583                        dst_fd = STDOUT_FILENO;
 584                        /*fallthrough*/
 585
 586                case 'q': /* Be quiet */
 587                        quiet++;
 588                        break;
 589
 590                case 'v': /* Verbose list */
 591                        IF_FEATURE_UNZIP_CDF(verbose++;)
 592                        opts |= OPT_l;
 593                        break;
 594
 595                case 'x':
 596                        opts |= OPT_x;
 597                        break;
 598
 599                case 'j':
 600                        opts |= OPT_j;
 601                        break;
 602
 603                case 1:
 604                        if (!src_fn) {
 605                                /* The zip file */
 606                                /* +5: space for ".zip" and NUL */
 607                                src_fn = xmalloc(strlen(optarg) + 5);
 608                                strcpy(src_fn, optarg);
 609                        } else if (!(opts & OPT_x)) {
 610                                /* Include files */
 611                                llist_add_to(&zaccept, optarg);
 612                        } else {
 613                                /* Exclude files */
 614                                llist_add_to(&zreject, optarg);
 615                        }
 616                        break;
 617
 618                default:
 619                        bb_show_usage();
 620                }
 621        }
 622
 623#ifndef __GLIBC__
 624        /*
 625         * This code is needed for non-GNU getopt
 626         * which doesn't understand "-" in option string.
 627         * The -x option won't work properly in this case:
 628         * "unzip a.zip q -x w e" will be interpreted as
 629         * "unzip a.zip q w e -x" = "unzip a.zip q w e"
 630         */
 631        argv += optind;
 632        if (argv[0]) {
 633                /* +5: space for ".zip" and NUL */
 634                src_fn = xmalloc(strlen(argv[0]) + 5);
 635                strcpy(src_fn, argv[0]);
 636                while (*++argv)
 637                        llist_add_to(&zaccept, *argv);
 638        }
 639#endif
 640
 641        if (!src_fn) {
 642                bb_show_usage();
 643        }
 644
 645        /* Open input file */
 646        if (LONE_DASH(src_fn)) {
 647                xdup2(STDIN_FILENO, zip_fd);
 648                /* Cannot use prompt mode since zip data is arriving on STDIN */
 649                if (overwrite == O_PROMPT)
 650                        overwrite = O_NEVER;
 651        } else {
 652                static const char extn[][5] ALIGN1 = { ".zip", ".ZIP" };
 653                char *ext = src_fn + strlen(src_fn);
 654                int src_fd;
 655
 656                i = 0;
 657                for (;;) {
 658                        src_fd = open(src_fn, O_RDONLY);
 659                        if (src_fd >= 0)
 660                                break;
 661                        if (++i > 2) {
 662                                *ext = '\0';
 663                                bb_error_msg_and_die("can't open %s[.zip]",
 664                                        src_fn
 665                                );
 666                        }
 667                        strcpy(ext, extn[i - 1]);
 668                }
 669                xmove_fd(src_fd, zip_fd);
 670        }
 671
 672        /* Change dir if necessary */
 673        if (base_dir) {
 674                /* -p DIR: try to create, errors don't matter.
 675                 * UnZip 6.00 does no multi-level mkdir (-p DIR1/DIR2 syntax),
 676                 * not using bb_make_directory() here (yet?)
 677                 */
 678                mkdir(base_dir, 0777);
 679                xchdir(base_dir);
 680        }
 681
 682        if (quiet <= 1) { /* not -qq */
 683                if (quiet == 0) {
 684                        printf("Archive:  %s\n",
 685                                printable_string(src_fn)
 686                        );
 687                }
 688                if (opts & OPT_l) {
 689                        puts(verbose ?
 690                                " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 691                                "--------  ------  ------- ---- ---------- ----- --------  ----"
 692                                :
 693                                "  Length      Date    Time    Name\n"
 694                                "---------  ---------- -----   ----"
 695                                );
 696                }
 697        }
 698
 699/* Example of an archive with one 0-byte long file named 'z'
 700 * created by Zip 2.31 on Unix:
 701 * 0000 [50 4b]03 04 0a 00 00 00 00 00 42 1a b8 3c 00 00 |PK........B..<..|
 702 *       sig........ vneed flags compr mtime mdate crc32>
 703 * 0010  00 00 00 00 00 00 00 00 00 00 01 00 15 00 7a 55 |..............zU|
 704 *      >..... csize...... usize...... fnlen exlen fn ex>
 705 * 0020  54 09 00 03 cc d3 f9 4b cc d3 f9 4b 55 78 04 00 |T......K...KUx..|
 706 *      >tra_field......................................
 707 * 0030  00 00 00 00[50 4b]01 02 17 03 0a 00 00 00 00 00 |....PK..........|
 708 *       ........... sig........ vmade vneed flags compr
 709 * 0040  42 1a b8 3c 00 00 00 00 00 00 00 00 00 00 00 00 |B..<............|
 710 *       mtime mdate crc32...... csize...... usize......
 711 * 0050  01 00 0d 00 00 00 00 00 00 00 00 00 a4 81 00 00 |................|
 712 *       fnlen exlen clen. dnum. iattr eattr...... relofs> (eattr = rw-r--r--)
 713 * 0060  00 00 7a 55 54 05 00 03 cc d3 f9 4b 55 78 00 00 |..zUT......KUx..|
 714 *      >..... fn extra_field...........................
 715 * 0070 [50 4b]05 06 00 00 00 00 01 00 01 00 3c 00 00 00 |PK..........<...|
 716 * 0080  34 00 00 00 00 00                               |4.....|
 717 */
 718        total_usize = 0;
 719        total_size = 0;
 720        total_entries = 0;
 721        cdf_offset = find_cdf_offset(); /* try to seek to the end, find CDE and CDF start */
 722        while (1) {
 723                zip_header_t zip;
 724                mode_t dir_mode = 0777;
 725#if ENABLE_FEATURE_UNZIP_CDF
 726                mode_t file_mode = 0666;
 727#endif
 728
 729                if (!ENABLE_FEATURE_UNZIP_CDF || cdf_offset == BAD_CDF_OFFSET) {
 730                        /* Normally happens when input is unseekable.
 731                         *
 732                         * Valid ZIP file has Central Directory at the end
 733                         * with central directory file headers (CDFs).
 734                         * After it, there is a Central Directory End structure.
 735                         * CDFs identify what files are in the ZIP and where
 736                         * they are located. This allows ZIP readers to load
 737                         * the list of files without reading the entire ZIP archive.
 738                         * ZIP files may be appended to, only files specified in
 739                         * the CD are valid. Scanning for local file headers is
 740                         * not a correct algorithm.
 741                         *
 742                         * We try to do the above, and resort to "linear" reading
 743                         * of ZIP file only if seek failed or CDE wasn't found.
 744                         */
 745                        uint32_t magic;
 746
 747                        /* Check magic number */
 748                        xread(zip_fd, &magic, 4);
 749                        /* CDF item? Assume there are no more files, exit */
 750                        if (magic == ZIP_CDF_MAGIC) {
 751                                dbg("got ZIP_CDF_MAGIC");
 752                                break;
 753                        }
 754                        /* Data descriptor? It was a streaming file, go on */
 755                        if (magic == ZIP_DD_MAGIC) {
 756                                dbg("got ZIP_DD_MAGIC");
 757                                /* skip over duplicate crc32, cmpsize and ucmpsize */
 758                                unzip_skip(3 * 4);
 759                                continue;
 760                        }
 761                        if (magic != ZIP_FILEHEADER_MAGIC)
 762                                bb_error_msg_and_die("invalid zip magic %08X", (int)magic);
 763                        dbg("got ZIP_FILEHEADER_MAGIC");
 764
 765                        xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 766                        FIX_ENDIANNESS_ZIP(zip);
 767                        if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 768                                bb_error_msg_and_die("zip flag %s is not supported",
 769                                        "8 (streaming)");
 770                        }
 771                }
 772#if ENABLE_FEATURE_UNZIP_CDF
 773                else {
 774                        /* cdf_offset is valid (and we know the file is seekable) */
 775                        cdf_header_t cdf;
 776                        cdf_offset = read_next_cdf(cdf_offset, &cdf);
 777                        if (cdf_offset == 0) /* EOF? */
 778                                break;
 779# if 1
 780                        xlseek(zip_fd,
 781                                SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4,
 782                                SEEK_SET);
 783                        xread(zip_fd, zip.raw, ZIP_HEADER_LEN);
 784                        FIX_ENDIANNESS_ZIP(zip);
 785                        if (zip.fmt.zip_flags & SWAP_LE16(0x0008)) {
 786                                /* 0x0008 - streaming. [u]cmpsize can be reliably gotten
 787                                 * only from Central Directory.
 788                                 */
 789                                zip.fmt.crc32    = cdf.fmt.crc32;
 790                                zip.fmt.cmpsize  = cdf.fmt.cmpsize;
 791                                zip.fmt.ucmpsize = cdf.fmt.ucmpsize;
 792                        }
 793// Seen in some zipfiles: central directory 9 byte extra field contains
 794// a subfield with ID 0x5455 and 5 data bytes, which is a Unix-style UTC mtime.
 795// Local header version:
 796//  u16 0x5455 ("UT")
 797//  u16 size (1 + 4 * n)
 798//  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 799//  u32 mtime
 800//  u32 atime
 801//  u32 ctime
 802// Central header version:
 803//  u16 0x5455 ("UT")
 804//  u16 size (5 (or 1?))
 805//  u8  flags: bit 0:mtime is present, bit 1:atime is present, bit 2:ctime is present
 806//  u32 mtime (CDF does not store atime/ctime)
 807# else
 808                        /* CDF has the same data as local header, no need to read the latter...
 809                         * ...not really. An archive was seen with cdf.extra_len == 6 but
 810                         * zip.extra_len == 0.
 811                         */
 812                        memcpy(&zip.fmt.version,
 813                                &cdf.fmt.version_needed, ZIP_HEADER_LEN);
 814                        xlseek(zip_fd,
 815                                SWAP_LE32(cdf.fmt.relative_offset_of_local_header) + 4 + ZIP_HEADER_LEN,
 816                                SEEK_SET);
 817# endif
 818                        if ((cdf.fmt.version_made_by >> 8) == 3) {
 819                                /* This archive is created on Unix */
 820                                dir_mode = file_mode = (cdf.fmt.external_attributes >> 16);
 821                        }
 822                }
 823#endif
 824
 825                if (zip.fmt.zip_flags & SWAP_LE16(0x0001)) {
 826                        /* 0x0001 - encrypted */
 827                        bb_error_msg_and_die("zip flag %s is not supported",
 828                                        "1 (encryption)");
 829                }
 830                dbg("File cmpsize:0x%x extra_len:0x%x ucmpsize:0x%x",
 831                        (unsigned)zip.fmt.cmpsize,
 832                        (unsigned)zip.fmt.extra_len,
 833                        (unsigned)zip.fmt.ucmpsize
 834                );
 835
 836                /* Read filename */
 837                free(dst_fn);
 838                die_if_bad_fnamesize(zip.fmt.filename_len);
 839                dst_fn = xzalloc(zip.fmt.filename_len + 1);
 840                xread(zip_fd, dst_fn, zip.fmt.filename_len);
 841                /* Skip extra header bytes */
 842                unzip_skip(zip.fmt.extra_len);
 843
 844                /* Guard against "/abspath", "/../" and similar attacks */
 845                overlapping_strcpy(dst_fn, strip_unsafe_prefix(dst_fn));
 846
 847                /* Filter zip entries */
 848                if (find_list_entry(zreject, dst_fn)
 849                 || (zaccept && !find_list_entry(zaccept, dst_fn))
 850                ) { /* Skip entry */
 851                        goto skip_cmpsize;
 852                }
 853
 854                if (opts & OPT_l) {
 855                        /* List entry */
 856                        char dtbuf[sizeof("mm-dd-yyyy hh:mm")];
 857                        sprintf(dtbuf, "%02u-%02u-%04u %02u:%02u",
 858                                (zip.fmt.moddate >> 5) & 0xf,  // mm: 0x01e0
 859                                (zip.fmt.moddate)      & 0x1f, // dd: 0x001f
 860                                (zip.fmt.moddate >> 9) + 1980, // yy: 0xfe00
 861                                (zip.fmt.modtime >> 11),       // hh: 0xf800
 862                                (zip.fmt.modtime >> 5) & 0x3f  // mm: 0x07e0
 863                                // seconds/2 not shown, encoded in -- 0x001f
 864                        );
 865                        if (!verbose) {
 866                                //      "  Length      Date    Time    Name\n"
 867                                //      "---------  ---------- -----   ----"
 868                                printf(       "%9u  " "%s   "         "%s\n",
 869                                        (unsigned)zip.fmt.ucmpsize,
 870                                        dtbuf,
 871                                        printable_string(dst_fn)
 872                                );
 873                        } else {
 874                                char method6[7];
 875                                unsigned long percents;
 876
 877                                sprintf(method6, "%6u", zip.fmt.method);
 878                                if (zip.fmt.method == 0) {
 879                                        strcpy(method6, "Stored");
 880                                }
 881                                if (zip.fmt.method == 8) {
 882                                        strcpy(method6, "Defl:N");
 883                                        /* normal, maximum, fast, superfast */
 884                                        IF_DESKTOP(method6[5] = "NXFS"[(zip.fmt.zip_flags >> 1) & 3];)
 885                                }
 886                                percents = zip.fmt.ucmpsize - zip.fmt.cmpsize;
 887                                if ((int32_t)percents < 0)
 888                                        percents = 0; /* happens if ucmpsize < cmpsize */
 889                                percents = percents * 100;
 890                                if (zip.fmt.ucmpsize)
 891                                        percents /= zip.fmt.ucmpsize;
 892                                //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
 893                                //      "--------  ------  ------- ---- ---------- ----- --------  ----"
 894                                printf(      "%8u  %s"        "%9u%4u%% " "%s "         "%08x  "  "%s\n",
 895                                        (unsigned)zip.fmt.ucmpsize,
 896                                        method6,
 897                                        (unsigned)zip.fmt.cmpsize,
 898                                        (unsigned)percents,
 899                                        dtbuf,
 900                                        zip.fmt.crc32,
 901                                        printable_string(dst_fn)
 902                                );
 903                                total_size += zip.fmt.cmpsize;
 904                        }
 905                        total_usize += zip.fmt.ucmpsize;
 906                        goto skip_cmpsize;
 907                }
 908
 909                if (dst_fd == STDOUT_FILENO) {
 910                        /* Extracting to STDOUT */
 911                        goto do_extract;
 912                }
 913
 914                /* Strip paths (after -l: unzip -lj a.zip lists full names) */
 915                if (opts & OPT_j)
 916                        overlapping_strcpy(dst_fn, bb_basename(dst_fn));
 917                /* Did this strip everything ("DIR/" case)? Then skip */
 918                if (!dst_fn[0])
 919                        goto skip_cmpsize;
 920
 921                if (last_char_is(dst_fn, '/')) {
 922                        int mode;
 923
 924                        /* Extract directory */
 925                        mode = get_lstat_mode(dst_fn);
 926                        if (mode == -1) { /* ENOENT */
 927                                if (!quiet) {
 928                                        printf("   creating: %s\n", printable_string(dst_fn));
 929                                }
 930                                unzip_create_leading_dirs(dst_fn);
 931                                if (bb_make_directory(dst_fn, dir_mode, FILEUTILS_IGNORE_CHMOD_ERR)) {
 932                                        xfunc_die();
 933                                }
 934                        } else {
 935                                if (!S_ISDIR(mode)) {
 936                                        bb_error_msg_and_die("'%s' exists but is not a %s",
 937                                                printable_string(dst_fn),
 938                                                "directory"
 939                                        );
 940                                }
 941                        }
 942                        goto skip_cmpsize;
 943                }
 944 check_file:
 945                /* Does target file already exist? */
 946                {
 947                        int mode = get_lstat_mode(dst_fn);
 948                        if (mode == -1) {
 949                                /* ENOENT: does not exist */
 950                                goto do_open_and_extract;
 951                        }
 952                        if (overwrite == O_NEVER) {
 953                                goto skip_cmpsize;
 954                        }
 955                        if (!S_ISREG(mode)) {
 956 fishy:
 957                                bb_error_msg_and_die("'%s' exists but is not a %s",
 958                                        printable_string(dst_fn),
 959                                        "regular file"
 960                                );
 961                        }
 962                        if (overwrite == O_ALWAYS) {
 963                                goto do_open_and_extract;
 964                        }
 965                        printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ",
 966                                printable_string(dst_fn)
 967                        );
 968                        my_fgets80(key_buf);
 969                        /* User input could take a long time. Is it still a regular file? */
 970                        mode = get_lstat_mode(dst_fn);
 971                        if (!S_ISREG(mode))
 972                                goto fishy;
 973                }
 974
 975                /* Extract (or skip) it */
 976                switch (key_buf[0]) {
 977                case 'A':
 978                        overwrite = O_ALWAYS;
 979                case 'y': /* Open file and fall into unzip */
 980 do_open_and_extract:
 981                        unzip_create_leading_dirs(dst_fn);
 982#if ENABLE_FEATURE_UNZIP_CDF
 983                        dst_fd = -1;
 984                        if (!S_ISLNK(file_mode)) {
 985                                dst_fd = xopen3(dst_fn,
 986                                        O_WRONLY | O_CREAT | O_TRUNC | O_NOFOLLOW,
 987                                        file_mode);
 988                        }
 989#else
 990                        /* O_NOFOLLOW defends against symlink attacks */
 991                        dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC | O_NOFOLLOW);
 992#endif
 993                        if (!quiet) {
 994                                printf(/* zip.fmt.method == 0
 995                                        ? " extracting: %s\n"
 996                                        : */ "  inflating: %s\n",
 997                                        printable_string(dst_fn)
 998                                );
 999                        }
1000 do_extract:
1001#if ENABLE_FEATURE_UNZIP_CDF
1002                        if (S_ISLNK(file_mode)) {
1003                                if (dst_fd != STDOUT_FILENO) /* not -p? */
1004                                        unzip_extract_symlink(&symlink_placeholders, &zip, dst_fn);
1005                        } else
1006#endif
1007                        {
1008                                unzip_extract(&zip, dst_fd);
1009                                if (dst_fd != STDOUT_FILENO) {
1010                                        /* closing STDOUT is potentially bad for future business */
1011                                        close(dst_fd);
1012                                }
1013                        }
1014                        break;
1015
1016                case 'N':
1017                        overwrite = O_NEVER;
1018                case 'n': /* Skip entry data */
1019 skip_cmpsize:
1020                        unzip_skip(zip.fmt.cmpsize);
1021                        break;
1022
1023                case 'r':
1024                        /* Prompt for new name */
1025                        printf("new name: ");
1026                        my_fgets80(key_buf);
1027                        free(dst_fn);
1028                        dst_fn = xstrdup(key_buf);
1029                        chomp(dst_fn);
1030                        goto check_file;
1031
1032                default:
1033                        printf("error: invalid response [%c]\n", (char)key_buf[0]);
1034                        goto check_file;
1035                }
1036
1037                total_entries++;
1038        }
1039
1040#if ENABLE_FEATURE_UNZIP_CDF
1041        create_links_from_list(symlink_placeholders);
1042#endif
1043
1044        if ((opts & OPT_l) && quiet <= 1) {
1045                if (!verbose) {
1046                        //      "  Length      Date    Time    Name\n"
1047                        //      "---------  ---------- -----   ----"
1048                        printf( " --------%21s"               "-------\n"
1049                                     "%9lu%21s"               "%u files\n",
1050                                "",
1051                                total_usize, "", total_entries);
1052                } else {
1053                        unsigned long percents = total_usize - total_size;
1054                        if ((long)percents < 0)
1055                                percents = 0; /* happens if usize < size */
1056                        percents = percents * 100;
1057                        if (total_usize)
1058                                percents /= total_usize;
1059                        //      " Length   Method    Size  Cmpr    Date    Time   CRC-32   Name\n"
1060                        //      "--------  ------  ------- ---- ---------- ----- --------  ----"
1061                        printf( "--------          ------- ----%28s"                      "----\n"
1062                                "%8lu"              "%17lu%4u%%%28s"                      "%u files\n",
1063                                "",
1064                                total_usize, total_size, (unsigned)percents, "",
1065                                total_entries);
1066                }
1067        }
1068
1069        return 0;
1070}
1071