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