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 the GPL v2 or later, see the file LICENSE in this tarball.
  11  */
  12
  13 /* For reference see
  14  * http://www.pkware.com/company/standards/appnote/
  15  * http://www.info-zip.org/pub/infozip/doc/appnote-iz-latest.zip
  16  */
  17
  18 /* TODO
  19  * Endian issues
  20  * Zip64 + other methods
  21  * Improve handling of zip format, ie.
  22  * - deferred CRC, comp. & uncomp. lengths (zip header flags bit 3)
  23  * - unix file permissions, etc.
  24  * - central directory
  25  */
  26
  27 #include <fcntl.h>
  28 #include <stdlib.h>
  29 #include <string.h>
  30 #include <unistd.h>
  31 #include <errno.h>
  32 #include "unarchive.h"
  33 #include "busybox.h"
  34
  35 #if BB_BIG_ENDIAN
  36 static inline unsigned short
  37 __swap16(unsigned short x) {
  38         return (((uint16_t)(x) & 0xFF) << 8) | (((uint16_t)(x) & 0xFF00) >> 8);
  39 }
  40
  41 static inline uint32_t
  42 __swap32(uint32_t x) {
  43          return (((x & 0xFF) << 24) |
  44                 ((x & 0xFF00) << 8) |
  45                 ((x & 0xFF0000) >> 8) |
  46                 ((x & 0xFF000000) >> 24));
  47 }
  48 #else /* it's little-endian */
  49 # define __swap16(x) (x)
  50 # define __swap32(x) (x)
  51 #endif /* BB_BIG_ENDIAN */
  52
  53 #define ZIP_FILEHEADER_MAGIC            __swap32(0x04034b50)
  54 #define ZIP_CDS_MAGIC                   __swap32(0x02014b50)
  55 #define ZIP_CDS_END_MAGIC               __swap32(0x06054b50)
  56 #define ZIP_DD_MAGIC                    __swap32(0x08074b50)
  57
  58 extern unsigned int gunzip_crc;
  59 extern unsigned int gunzip_bytes_out;
  60
  61 typedef union {
  62         unsigned char raw[26];
  63         struct {
  64                 unsigned short version; /* 0-1 */
  65                 unsigned short flags;   /* 2-3 */
  66                 unsigned short method;  /* 4-5 */
  67                 unsigned short modtime; /* 6-7 */
  68                 unsigned short moddate; /* 8-9 */
  69                 unsigned int crc32 ATTRIBUTE_PACKED;    /* 10-13 */
  70                 unsigned int cmpsize ATTRIBUTE_PACKED;  /* 14-17 */
  71                 unsigned int ucmpsize ATTRIBUTE_PACKED; /* 18-21 */
  72                 unsigned short filename_len;    /* 22-23 */
  73                 unsigned short extra_len;               /* 24-25 */
  74         } formated ATTRIBUTE_PACKED;
  75 } zip_header_t;
  76
  77 static void unzip_skip(int fd, off_t skip)
  78 {
  79         if (lseek(fd, skip, SEEK_CUR) == (off_t)-1) {
  80                 if ((errno != ESPIPE) || (bb_copyfd_size(fd, -1, skip) != skip)) {
  81                         bb_error_msg_and_die("Seek failure");
  82                 }
  83         }
  84 }
  85
  86 static void unzip_read(int fd, void *buf, size_t count)
  87 {
  88         if (bb_xread(fd, buf, count) != count) {
  89                 bb_error_msg_and_die(bb_msg_read_error);
  90         }
  91 }
  92
  93 static void unzip_create_leading_dirs(char *fn)
  94 {
  95         /* Create all leading directories */
  96         char *name = bb_xstrdup(fn);
  97         if (bb_make_directory(dirname(name), 0777, FILEUTILS_RECUR)) {
  98                 bb_error_msg_and_die("Exiting"); /* bb_make_directory is noisy */
  99         }
 100         free(name);
 101 }
 102
 103 static int unzip_extract(zip_header_t *zip_header, int src_fd, int dst_fd)
 104 {
 105         if (zip_header->formated.method == 0) {
 106                 /* Method 0 - stored (not compressed) */
 107                 int size = zip_header->formated.ucmpsize;
 108                 if (size && (bb_copyfd_size(src_fd, dst_fd, size) != size)) {
 109                         bb_error_msg_and_die("Cannot complete extraction");
 110                 }
 111
 112         } else {
 113                 /* Method 8 - inflate */
 114                 inflate_init(zip_header->formated.cmpsize);
 115                 inflate_unzip(src_fd, dst_fd);
 116                 inflate_cleanup();
 117                 /* Validate decompression - crc */
 118                 if (zip_header->formated.crc32 != (gunzip_crc ^ 0xffffffffL)) {
 119                         bb_error_msg("Invalid compressed data--crc error");
 120                         return 1;
 121                 }
 122                 /* Validate decompression - size */
 123                 if (zip_header->formated.ucmpsize != gunzip_bytes_out) {
 124                         bb_error_msg("Invalid compressed data--length error");
 125                         return 1;
 126                 }
 127         }
 128         return 0;
 129 }
 130
 131 int unzip_main(int argc, char **argv)
 132 {
 133         zip_header_t zip_header;
 134         enum {v_silent, v_normal, v_list} verbosity = v_normal;
 135         enum {o_prompt, o_never, o_always} overwrite = o_prompt;
 136         unsigned int total_size = 0;
 137         unsigned int total_entries = 0;
 138         int src_fd = -1, dst_fd = -1;
 139         char *src_fn = NULL, *dst_fn = NULL;
 140         llist_t *zaccept = NULL;
 141         llist_t *zreject = NULL;
 142         char *base_dir = NULL;
 143         int failed, i, opt, opt_range = 0, list_header_done = 0;
 144         char key_buf[512];
 145         struct stat stat_buf;
 146
 147         while((opt = getopt(argc, argv, "-d:lnopqx")) != -1) {
 148                 switch(opt_range) {
 149                 case 0: /* Options */
 150                         switch(opt) {
 151                         case 'l': /* List */
 152                                 verbosity = v_list;
 153                                 break;
 154
 155                         case 'n': /* Never overwrite existing files */
 156                                 overwrite = o_never;
 157                                 break;
 158
 159                         case 'o': /* Always overwrite existing files */
 160                                 overwrite = o_always;
 161                                 break;
 162
 163                         case 'p': /* Extract files to stdout and fall through to set verbosity */
 164                                 dst_fd = STDOUT_FILENO;
 165
 166                         case 'q': /* Be quiet */
 167                                 verbosity = (verbosity == v_normal) ? v_silent : verbosity;
 168                                 break;
 169
 170                         case 1 : /* The zip file */
 171                                 src_fn = bb_xstrndup(optarg, strlen(optarg)+4);
 172                                 opt_range++;
 173                                 break;
 174
 175                         default:
 176                                 bb_show_usage();
 177
 178                         }
 179                         break;
 180
 181                 case 1: /* Include files */
 182                         if (opt == 1) {
 183                                 llist_add_to(&zaccept, optarg);
 184
 185                         } else if (opt == 'd') {
 186                                 base_dir = optarg;
 187                                 opt_range += 2;
 188
 189                         } else if (opt == 'x') {
 190                                 opt_range++;
 191
 192                         } else {
 193                                 bb_show_usage();
 194                         }
 195                         break;
 196
 197                 case 2 : /* Exclude files */
 198                         if (opt == 1) {
 199                                 llist_add_to(&zreject, optarg);
 200
 201                         } else if (opt == 'd') { /* Extract to base directory */
 202                                 base_dir = optarg;
 203                                 opt_range++;
 204
 205                         } else {
 206                                 bb_show_usage();
 207                         }
 208                         break;
 209
 210                 default:
 211                         bb_show_usage();
 212                 }
 213         }
 214
 215         if (src_fn == NULL) {
 216                 bb_show_usage();
 217         }
 218
 219         /* Open input file */
 220         if (strcmp("-", src_fn) == 0) {
 221                 src_fd = STDIN_FILENO;
 222                 /* Cannot use prompt mode since zip data is arriving on STDIN */
 223                 overwrite = (overwrite == o_prompt) ? o_never : overwrite;
 224
 225         } else {
 226                 static const char *const extn[] = {"", ".zip", ".ZIP"};
 227                 int orig_src_fn_len = strlen(src_fn);
 228                 for(i = 0; (i < 3) && (src_fd == -1); i++) {
 229                         strcpy(src_fn + orig_src_fn_len, extn[i]);
 230                         src_fd = open(src_fn, O_RDONLY);
 231                 }
 232                 if (src_fd == -1) {
 233                         src_fn[orig_src_fn_len] = 0;
 234                         bb_error_msg_and_die("Cannot open %s, %s.zip, %s.ZIP", src_fn, src_fn, src_fn);
 235                 }
 236         }
 237
 238         /* Change dir if necessary */
 239         if (base_dir)
 240                 bb_xchdir(base_dir);
 241
 242         if (verbosity != v_silent)
 243                 printf("Archive:  %s\n", src_fn);
 244
 245         failed = 0;
 246
 247         while (1) {
 248                 unsigned int magic;
 249
 250                 /* Check magic number */
 251                 unzip_read(src_fd, &magic, 4);
 252                 if (magic == ZIP_CDS_MAGIC) {
 253                         break;
 254                 } else if (magic != ZIP_FILEHEADER_MAGIC) {
 255                         bb_error_msg_and_die("Invalid zip magic %08X", magic);
 256                 }
 257
 258                 /* Read the file header */
 259                 unzip_read(src_fd, zip_header.raw, 26);
 260 #if BB_BIG_ENDIAN
 261                 zip_header.formated.version = __swap16(zip_header.formated.version);
 262                 zip_header.formated.flags = __swap16(zip_header.formated.flags);
 263                 zip_header.formated.method = __swap16(zip_header.formated.method);
 264                 zip_header.formated.modtime = __swap16(zip_header.formated.modtime);
 265                 zip_header.formated.moddate = __swap16(zip_header.formated.moddate);
 266                 zip_header.formated.crc32 = __swap32(zip_header.formated.crc32);
 267                 zip_header.formated.cmpsize = __swap32(zip_header.formated.cmpsize);
 268                 zip_header.formated.ucmpsize = __swap32(zip_header.formated.ucmpsize);
 269                 zip_header.formated.filename_len = __swap16(zip_header.formated.filename_len);
 270                 zip_header.formated.extra_len = __swap16(zip_header.formated.extra_len);
 271 #endif /* BB_BIG_ENDIAN */
 272                 if ((zip_header.formated.method != 0) && (zip_header.formated.method != 8)) {
 273                         bb_error_msg_and_die("Unsupported compression method %d", zip_header.formated.method);
 274                 }
 275
 276                 /* Read filename */
 277                 free(dst_fn);
 278                 dst_fn = xzalloc(zip_header.formated.filename_len + 1);
 279                 unzip_read(src_fd, dst_fn, zip_header.formated.filename_len);
 280
 281                 /* Skip extra header bytes */
 282                 unzip_skip(src_fd, zip_header.formated.extra_len);
 283
 284                 if ((verbosity == v_list) && !list_header_done){
 285                         printf("  Length     Date   Time    Name\n"
 286                                    " --------    ----   ----    ----\n");
 287                         list_header_done = 1;
 288                 }
 289
 290                 /* Filter zip entries */
 291                 if (find_list_entry(zreject, dst_fn) ||
 292                         (zaccept && !find_list_entry(zaccept, dst_fn))) { /* Skip entry */
 293                         i = 'n';
 294
 295                 } else { /* Extract entry */
 296                         total_size += zip_header.formated.ucmpsize;
 297
 298                         if (verbosity == v_list) { /* List entry */
 299                                 unsigned int dostime = zip_header.formated.modtime | (zip_header.formated.moddate << 16);
 300                                 printf("%9u  %02u-%02u-%02u %02u:%02u   %s\n",
 301                                            zip_header.formated.ucmpsize,
 302                                            (dostime & 0x01e00000) >> 21,
 303                                            (dostime & 0x001f0000) >> 16,
 304                                            (((dostime & 0xfe000000) >> 25) + 1980) % 100,
 305                                            (dostime & 0x0000f800) >> 11,
 306                                            (dostime & 0x000007e0) >> 5,
 307                                            dst_fn);
 308                                 total_entries++;
 309                                 i = 'n';
 310
 311                         } else if (dst_fd == STDOUT_FILENO) { /* Extracting to STDOUT */
 312                                 i = -1;
 313
 314                         } else if (last_char_is(dst_fn, '/')) { /* Extract directory */
 315                                 if (stat(dst_fn, &stat_buf) == -1) {
 316                                         if (errno != ENOENT) {
 317                                                 bb_perror_msg_and_die("Cannot stat '%s'",dst_fn);
 318                                         }
 319                                         if (verbosity == v_normal) {
 320                                                 printf("   creating: %s\n", dst_fn);
 321                                         }
 322                                         unzip_create_leading_dirs(dst_fn);
 323                                         if (bb_make_directory(dst_fn, 0777, 0)) {
 324                                                 bb_error_msg_and_die("Exiting");
 325                                         }
 326                                 } else {
 327                                         if (!S_ISDIR(stat_buf.st_mode)) {
 328                                                 bb_error_msg_and_die("'%s' exists but is not directory", dst_fn);
 329                                         }
 330                                 }
 331                                 i = 'n';
 332
 333                         } else {  /* Extract file */
 334                         _check_file:
 335                                 if (stat(dst_fn, &stat_buf) == -1) { /* File does not exist */
 336                                         if (errno != ENOENT) {
 337                                                 bb_perror_msg_and_die("Cannot stat '%s'",dst_fn);
 338                                         }
 339                                         i = 'y';
 340
 341                                 } else { /* File already exists */
 342                                         if (overwrite == o_never) {
 343                                                 i = 'n';
 344
 345                                         } else if (S_ISREG(stat_buf.st_mode)) { /* File is regular file */
 346                                                 if (overwrite == o_always) {
 347                                                         i = 'y';
 348                                                 } else {
 349                                                         printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
 350                                                         if (!fgets(key_buf, 512, stdin)) {
 351                                                                 bb_perror_msg_and_die("Cannot read input");
 352                                                         }
 353                                                         i = key_buf[0];
 354                                                 }
 355
 356                                         } else { /* File is not regular file */
 357                                                 bb_error_msg_and_die("'%s' exists but is not regular file",dst_fn);
 358                                         }
 359                                 }
 360                         }
 361                 }
 362
 363                 switch (i) {
 364                 case 'A':
 365                         overwrite = o_always;
 366                 case 'y': /* Open file and fall into unzip */
 367                         unzip_create_leading_dirs(dst_fn);
 368                         dst_fd = bb_xopen(dst_fn, O_WRONLY | O_CREAT);
 369                 case -1: /* Unzip */
 370                         if (verbosity == v_normal) {
 371                                 printf("  inflating: %s\n", dst_fn);
 372                         }
 373                         if (unzip_extract(&zip_header, src_fd, dst_fd)) {
 374                             failed = 1;
 375                         }
 376                         if (dst_fd != STDOUT_FILENO) {
 377                                 /* closing STDOUT is potentially bad for future business */
 378                                 close(dst_fd);
 379                         }
 380                         break;
 381
 382                 case 'N':
 383                         overwrite = o_never;
 384                 case 'n':
 385                         /* Skip entry data */
 386                         unzip_skip(src_fd, zip_header.formated.cmpsize);
 387                         break;
 388
 389                 case 'r':
 390                         /* Prompt for new name */
 391                         printf("new name: ");
 392                         if (!fgets(key_buf, 512, stdin)) {
 393                                 bb_perror_msg_and_die("Cannot read input");
 394                         }
 395                         free(dst_fn);
 396                         dst_fn = bb_xstrdup(key_buf);
 397                         chomp(dst_fn);
 398                         goto _check_file;
 399
 400                 default:
 401                         printf("error: invalid response [%c]\n",(char)i);
 402                         goto _check_file;
 403                 }
 404
 405                 /* Data descriptor section */
 406                 if (zip_header.formated.flags & 4) {
 407                         /* skip over duplicate crc, compressed size and uncompressed size */
 408                         unzip_skip(src_fd, 12);
 409                 }
 410         }
 411
 412         if (verbosity == v_list) {
 413                 printf(" --------                   -------\n"
 414                        "%9d                   %d files\n", total_size, total_entries);
 415         }
 416
 417         return failed;
 418 }