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 "libbb.h"
  28 #include "unarchive.h"
  29
  30 #define ZIP_FILEHEADER_MAGIC            SWAP_LE32(0x04034b50)
  31 #define ZIP_CDS_MAGIC                   SWAP_LE32(0x02014b50)
  32 #define ZIP_CDS_END_MAGIC               SWAP_LE32(0x06054b50)
  33 #define ZIP_DD_MAGIC                    SWAP_LE32(0x08074b50)
  34
  35 typedef union {
  36         unsigned char raw[26];
  37         struct {
  38                 unsigned short version; /* 0-1 */
  39                 unsigned short flags;   /* 2-3 */
  40                 unsigned short method;  /* 4-5 */
  41                 unsigned short modtime; /* 6-7 */
  42                 unsigned short moddate; /* 8-9 */
  43                 unsigned int crc32 ATTRIBUTE_PACKED;    /* 10-13 */
  44                 unsigned int cmpsize ATTRIBUTE_PACKED;  /* 14-17 */
  45                 unsigned int ucmpsize ATTRIBUTE_PACKED; /* 18-21 */
  46                 unsigned short filename_len;    /* 22-23 */
  47                 unsigned short extra_len;               /* 24-25 */
  48         } formatted ATTRIBUTE_PACKED;
  49 } zip_header_t;
  50
  51 static void unzip_skip(int fd, off_t skip)
  52 {
  53         if (lseek(fd, skip, SEEK_CUR) == (off_t)-1) {
  54                 if (errno != ESPIPE)
  55                         bb_error_msg_and_die("seek failure");
  56                 bb_copyfd_exact_size(fd, -1, skip);
  57         }
  58 }
  59
  60 static void unzip_create_leading_dirs(char *fn)
  61 {
  62         /* Create all leading directories */
  63         char *name = xstrdup(fn);
  64         if (bb_make_directory(dirname(name), 0777, FILEUTILS_RECUR)) {
  65                 bb_error_msg_and_die("exiting"); /* bb_make_directory is noisy */
  66         }
  67         free(name);
  68 }
  69
  70 static int unzip_extract(zip_header_t *zip_header, int src_fd, int dst_fd)
  71 {
  72         if (zip_header->formatted.method == 0) {
  73                 /* Method 0 - stored (not compressed) */
  74                 off_t size = zip_header->formatted.ucmpsize;
  75                 if (size)
  76                         bb_copyfd_exact_size(src_fd, dst_fd, size);
  77         } else {
  78                 /* Method 8 - inflate */
  79                 inflate_unzip_result res;
  80                 /* err = */ inflate_unzip(&res, zip_header->formatted.cmpsize, src_fd, dst_fd);
  81 // we should check for -1 error return
  82                 /* Validate decompression - crc */
  83                 if (zip_header->formatted.crc32 != (res.crc ^ 0xffffffffL)) {
  84                         bb_error_msg("invalid compressed data--%s error", "crc");
  85                         return 1;
  86                 }
  87                 /* Validate decompression - size */
  88                 if (zip_header->formatted.ucmpsize != res.bytes_out) {
  89                         bb_error_msg("invalid compressed data--%s error", "length");
  90                         return 1;
  91                 }
  92         }
  93         return 0;
  94 }
  95
  96 int unzip_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  97 int unzip_main(int argc, char **argv)
  98 {
  99         zip_header_t zip_header;
 100         smallint verbose = 1;
 101         smallint listing = 0;
 102         smallint list_header_done = 0;
 103         smallint failed;
 104         enum {o_prompt, o_never, o_always} overwrite = o_prompt;
 105         unsigned int total_size = 0;
 106         unsigned int total_entries = 0;
 107         int src_fd = -1, dst_fd = -1;
 108         char *src_fn = NULL, *dst_fn = NULL;
 109         llist_t *zaccept = NULL;
 110         llist_t *zreject = NULL;
 111         char *base_dir = NULL;
 112         int i, opt, opt_range = 0;
 113         char key_buf[512];
 114         struct stat stat_buf;
 115
 116         while ((opt = getopt(argc, argv, "-d:lnopqx")) != -1) {
 117                 switch (opt_range) {
 118                 case 0: /* Options */
 119                         switch (opt) {
 120                         case 'l': /* List */
 121                                 listing = 1;
 122                                 break;
 123
 124                         case 'n': /* Never overwrite existing files */
 125                                 overwrite = o_never;
 126                                 break;
 127
 128                         case 'o': /* Always overwrite existing files */
 129                                 overwrite = o_always;
 130                                 break;
 131
 132                         case 'p': /* Extract files to stdout and fall through to set verbosity */
 133                                 dst_fd = STDOUT_FILENO;
 134
 135                         case 'q': /* Be quiet */
 136                                 verbose = 0;
 137                                 break;
 138
 139                         case 1 : /* The zip file */
 140                                 src_fn = xmalloc(strlen(optarg)+4);
 141                                 strcpy(src_fn, optarg);
 142                                 opt_range++;
 143                                 break;
 144
 145                         default:
 146                                 bb_show_usage();
 147
 148                         }
 149                         break;
 150
 151                 case 1: /* Include files */
 152                         if (opt == 1) {
 153                                 llist_add_to(&zaccept, optarg);
 154
 155                         } else if (opt == 'd') {
 156                                 base_dir = optarg;
 157                                 opt_range += 2;
 158
 159                         } else if (opt == 'x') {
 160                                 opt_range++;
 161
 162                         } else {
 163                                 bb_show_usage();
 164                         }
 165                         break;
 166
 167                 case 2 : /* Exclude files */
 168                         if (opt == 1) {
 169                                 llist_add_to(&zreject, optarg);
 170
 171                         } else if (opt == 'd') { /* Extract to base directory */
 172                                 base_dir = optarg;
 173                                 opt_range++;
 174
 175                         } else {
 176                                 bb_show_usage();
 177                         }
 178                         break;
 179
 180                 default:
 181                         bb_show_usage();
 182                 }
 183         }
 184
 185         if (src_fn == NULL) {
 186                 bb_show_usage();
 187         }
 188
 189         /* Open input file */
 190         if (LONE_DASH(src_fn)) {
 191                 src_fd = STDIN_FILENO;
 192                 /* Cannot use prompt mode since zip data is arriving on STDIN */
 193                 overwrite = (overwrite == o_prompt) ? o_never : overwrite;
 194         } else {
 195                 static const char *const extn[] = {"", ".zip", ".ZIP"};
 196                 int orig_src_fn_len = strlen(src_fn);
 197                 for (i = 0; (i < 3) && (src_fd == -1); i++) {
 198                         strcpy(src_fn + orig_src_fn_len, extn[i]);
 199                         src_fd = open(src_fn, O_RDONLY);
 200                 }
 201                 if (src_fd == -1) {
 202                         src_fn[orig_src_fn_len] = '\0';
 203                         bb_error_msg_and_die("cannot open %s, %s.zip, %s.ZIP", src_fn, src_fn, src_fn);
 204                 }
 205         }
 206
 207         /* Change dir if necessary */
 208         if (base_dir)
 209                 xchdir(base_dir);
 210
 211         if (verbose)
 212                 printf("Archive:  %s\n", src_fn);
 213
 214         failed = 0;
 215
 216         while (1) {
 217                 unsigned int magic;
 218
 219                 /* Check magic number */
 220                 xread(src_fd, &magic, 4);
 221                 if (magic == ZIP_CDS_MAGIC) {
 222                         break;
 223                 } else if (magic != ZIP_FILEHEADER_MAGIC) {
 224                         bb_error_msg_and_die("invalid zip magic %08X", magic);
 225                 }
 226
 227                 /* Read the file header */
 228                 xread(src_fd, zip_header.raw, 26);
 229                 zip_header.formatted.version = SWAP_LE32(zip_header.formatted.version);
 230                 zip_header.formatted.flags = SWAP_LE32(zip_header.formatted.flags);
 231                 zip_header.formatted.method = SWAP_LE32(zip_header.formatted.method);
 232                 zip_header.formatted.modtime = SWAP_LE32(zip_header.formatted.modtime);
 233                 zip_header.formatted.moddate = SWAP_LE32(zip_header.formatted.moddate);
 234                 zip_header.formatted.crc32 = SWAP_LE32(zip_header.formatted.crc32);
 235                 zip_header.formatted.cmpsize = SWAP_LE32(zip_header.formatted.cmpsize);
 236                 zip_header.formatted.ucmpsize = SWAP_LE32(zip_header.formatted.ucmpsize);
 237                 zip_header.formatted.filename_len = SWAP_LE32(zip_header.formatted.filename_len);
 238                 zip_header.formatted.extra_len = SWAP_LE32(zip_header.formatted.extra_len);
 239                 if ((zip_header.formatted.method != 0) && (zip_header.formatted.method != 8)) {
 240                         bb_error_msg_and_die("unsupported compression method %d", zip_header.formatted.method);
 241                 }
 242
 243                 /* Read filename */
 244                 free(dst_fn);
 245                 dst_fn = xzalloc(zip_header.formatted.filename_len + 1);
 246                 xread(src_fd, dst_fn, zip_header.formatted.filename_len);
 247
 248                 /* Skip extra header bytes */
 249                 unzip_skip(src_fd, zip_header.formatted.extra_len);
 250
 251                 if (listing && verbose && !list_header_done){
 252                         puts("  Length     Date   Time    Name\n"
 253                              " --------    ----   ----    ----");
 254                         list_header_done = 1;
 255                 }
 256
 257                 /* Filter zip entries */
 258                 if (find_list_entry(zreject, dst_fn) ||
 259                         (zaccept && !find_list_entry(zaccept, dst_fn))) { /* Skip entry */
 260                         i = 'n';
 261
 262                 } else { /* Extract entry */
 263                         total_size += zip_header.formatted.ucmpsize;
 264
 265                         if (listing) { /* List entry */
 266                                 if (verbose) {
 267                                         unsigned int dostime = zip_header.formatted.modtime | (zip_header.formatted.moddate << 16);
 268                                         printf("%9u  %02u-%02u-%02u %02u:%02u   %s\n",
 269                                            zip_header.formatted.ucmpsize,
 270                                            (dostime & 0x01e00000) >> 21,
 271                                            (dostime & 0x001f0000) >> 16,
 272                                            (((dostime & 0xfe000000) >> 25) + 1980) % 100,
 273                                            (dostime & 0x0000f800) >> 11,
 274                                            (dostime & 0x000007e0) >> 5,
 275                                            dst_fn);
 276                                         total_entries++;
 277                                 } else {
 278                                         /* short listing -- filenames only */
 279                                         puts(dst_fn);
 280                                 }
 281                                 i = 'n';
 282                         } else if (dst_fd == STDOUT_FILENO) { /* Extracting to STDOUT */
 283                                 i = -1;
 284                         } else if (last_char_is(dst_fn, '/')) { /* Extract directory */
 285                                 if (stat(dst_fn, &stat_buf) == -1) {
 286                                         if (errno != ENOENT) {
 287                                                 bb_perror_msg_and_die("cannot stat '%s'",dst_fn);
 288                                         }
 289                                         if (verbose) {
 290                                                 printf("   creating: %s\n", dst_fn);
 291                                         }
 292                                         unzip_create_leading_dirs(dst_fn);
 293                                         if (bb_make_directory(dst_fn, 0777, 0)) {
 294                                                 bb_error_msg_and_die("exiting");
 295                                         }
 296                                 } else {
 297                                         if (!S_ISDIR(stat_buf.st_mode)) {
 298                                                 bb_error_msg_and_die("'%s' exists but is not directory", dst_fn);
 299                                         }
 300                                 }
 301                                 i = 'n';
 302
 303                         } else {  /* Extract file */
 304  _check_file:
 305                                 if (stat(dst_fn, &stat_buf) == -1) { /* File does not exist */
 306                                         if (errno != ENOENT) {
 307                                                 bb_perror_msg_and_die("cannot stat '%s'",dst_fn);
 308                                         }
 309                                         i = 'y';
 310                                 } else { /* File already exists */
 311                                         if (overwrite == o_never) {
 312                                                 i = 'n';
 313                                         } else if (S_ISREG(stat_buf.st_mode)) { /* File is regular file */
 314                                                 if (overwrite == o_always) {
 315                                                         i = 'y';
 316                                                 } else {
 317                                                         printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
 318                                                         if (!fgets(key_buf, 512, stdin)) {
 319                                                                 bb_perror_msg_and_die("cannot read input");
 320                                                         }
 321                                                         i = key_buf[0];
 322                                                 }
 323                                         } else { /* File is not regular file */
 324                                                 bb_error_msg_and_die("'%s' exists but is not regular file",dst_fn);
 325                                         }
 326                                 }
 327                         }
 328                 }
 329
 330                 switch (i) {
 331                 case 'A':
 332                         overwrite = o_always;
 333                 case 'y': /* Open file and fall into unzip */
 334                         unzip_create_leading_dirs(dst_fn);
 335                         dst_fd = xopen(dst_fn, O_WRONLY | O_CREAT | O_TRUNC);
 336                 case -1: /* Unzip */
 337                         if (verbose) {
 338                                 printf("  inflating: %s\n", dst_fn);
 339                         }
 340                         if (unzip_extract(&zip_header, src_fd, dst_fd)) {
 341                                 failed = 1;
 342                         }
 343                         if (dst_fd != STDOUT_FILENO) {
 344                                 /* closing STDOUT is potentially bad for future business */
 345                                 close(dst_fd);
 346                         }
 347                         break;
 348
 349                 case 'N':
 350                         overwrite = o_never;
 351                 case 'n':
 352                         /* Skip entry data */
 353                         unzip_skip(src_fd, zip_header.formatted.cmpsize);
 354                         break;
 355
 356                 case 'r':
 357                         /* Prompt for new name */
 358                         printf("new name: ");
 359                         if (!fgets(key_buf, 512, stdin)) {
 360                                 bb_perror_msg_and_die("cannot read input");
 361                         }
 362                         free(dst_fn);
 363                         dst_fn = xstrdup(key_buf);
 364                         chomp(dst_fn);
 365                         goto _check_file;
 366
 367                 default:
 368                         printf("error: invalid response [%c]\n",(char)i);
 369                         goto _check_file;
 370                 }
 371
 372                 /* Data descriptor section */
 373                 if (zip_header.formatted.flags & 4) {
 374                         /* skip over duplicate crc, compressed size and uncompressed size */
 375                         unzip_skip(src_fd, 12);
 376                 }
 377         }
 378
 379         if (listing && verbose) {
 380                 printf(" --------                   -------\n"
 381                        "%9d                   %d files\n", total_size, total_entries);
 382         }
 383
 384         return failed;
 385 }