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 (BYTE_ORDER == 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
  49 #define __swap16(x) (x)
  50 #define __swap32(x) (x)
  51 #endif
  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("Read failure");
  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("Failed to create directory");
  99         }
 100         free(name);
 101 }
 102
 103 static void 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                 }
 121                 /* Validate decompression - size */
 122                 if (zip_header->formated.ucmpsize != gunzip_bytes_out) {
 123                         bb_error_msg("Invalid compressed data--length error");
 124                 }
 125         }
 126 }
 127
 128 extern int unzip_main(int argc, char **argv)
 129 {
 130         zip_header_t zip_header;
 131         enum {v_silent, v_normal, v_list} verbosity = v_normal;
 132         enum {o_prompt, o_never, o_always} overwrite = o_prompt;
 133         unsigned int total_size = 0;
 134         unsigned int total_entries = 0;
 135         int src_fd = -1, dst_fd = -1;
 136         char *src_fn = NULL, *dst_fn = NULL;
 137         llist_t *zaccept = NULL;
 138         llist_t *zreject = NULL;
 139         char *base_dir = NULL;
 140         int i, opt, opt_range = 0, list_header_done = 0;
 141         char key_buf[512];
 142         struct stat stat_buf;
 143
 144         while((opt = getopt(argc, argv, "-d:lnopqx")) != -1) {
 145                 switch(opt_range) {
 146                 case 0: /* Options */
 147                         switch(opt) {
 148                         case 'l': /* List */
 149                                 verbosity = v_list;
 150                                 break;
 151
 152                         case 'n': /* Never overwrite existing files */
 153                                 overwrite = o_never;
 154                                 break;
 155
 156                         case 'o': /* Always overwrite existing files */
 157                                 overwrite = o_always;
 158                                 break;
 159
 160                         case 'p': /* Extract files to stdout and fall through to set verbosity */
 161                                 dst_fd = STDOUT_FILENO;
 162
 163                         case 'q': /* Be quiet */
 164                                 verbosity = (verbosity == v_normal) ? v_silent : verbosity;
 165                                 break;
 166
 167                         case 1 : /* The zip file */
 168                                 src_fn = bb_xstrndup(optarg, strlen(optarg)+4);
 169                                 opt_range++;
 170                                 break;
 171
 172                         default:
 173                                 bb_show_usage();
 174
 175                         }
 176                         break;
 177
 178                 case 1: /* Include files */
 179                         if (opt == 1) {
 180                                 zaccept = llist_add_to(zaccept, optarg);
 181
 182                         } else if (opt == 'd') {
 183                                 base_dir = optarg;
 184                                 opt_range += 2;
 185
 186                         } else if (opt == 'x') {
 187                                 opt_range++;
 188
 189                         } else {
 190                                 bb_show_usage();
 191                         }
 192                         break;
 193
 194                 case 2 : /* Exclude files */
 195                         if (opt == 1) {
 196                                 zreject = llist_add_to(zreject, optarg);
 197
 198                         } else if (opt == 'd') { /* Extract to base directory */
 199                                 base_dir = optarg;
 200                                 opt_range++;
 201
 202                         } else {
 203                                 bb_show_usage();
 204                         }
 205                         break;
 206
 207                 default:
 208                         bb_show_usage();
 209                 }
 210         }
 211
 212         if (src_fn == NULL) {
 213                 bb_show_usage();
 214         }
 215
 216         /* Open input file */
 217         if (strcmp("-", src_fn) == 0) {
 218                 src_fd = STDIN_FILENO;
 219                 /* Cannot use prompt mode since zip data is arriving on STDIN */
 220                 overwrite = (overwrite == o_prompt) ? o_never : overwrite;
 221
 222         } else {
 223                 static const char *const extn[] = {"", ".zip", ".ZIP"};
 224                 int orig_src_fn_len = strlen(src_fn);
 225                 for(i = 0; (i < 3) && (src_fd == -1); i++) {
 226                         strcpy(src_fn + orig_src_fn_len, extn[i]);
 227                         src_fd = open(src_fn, O_RDONLY);
 228                 }
 229                 if (src_fd == -1) {
 230                         src_fn[orig_src_fn_len] = 0;
 231                         bb_error_msg_and_die("Cannot open %s, %s.zip, %s.ZIP", src_fn, src_fn, src_fn);
 232                 }
 233         }
 234
 235         /* Change dir if necessary */
 236         if (base_dir && chdir(base_dir)) {
 237                 bb_perror_msg_and_die("Cannot chdir");
 238         }
 239
 240         if (verbosity != v_silent)
 241                 printf("Archive:  %s\n", src_fn);
 242
 243         while (1) {
 244                 unsigned int magic;
 245
 246                 /* Check magic number */
 247                 unzip_read(src_fd, &magic, 4);
 248                 if (magic == ZIP_CDS_MAGIC) {
 249                         break;
 250                 } else if (magic != ZIP_FILEHEADER_MAGIC) {
 251                         bb_error_msg_and_die("Invalid zip magic %08X", magic);
 252                 }
 253
 254                 /* Read the file header */
 255                 unzip_read(src_fd, zip_header.raw, 26);
 256 #if (BYTE_ORDER == BIG_ENDIAN)
 257                 zip_header.formated.version = __swap16(zip_header.formated.version);
 258                 zip_header.formated.flags = __swap16(zip_header.formated.flags);
 259                 zip_header.formated.method = __swap16(zip_header.formated.method);
 260                 zip_header.formated.modtime = __swap16(zip_header.formated.modtime);
 261                 zip_header.formated.moddate = __swap16(zip_header.formated.moddate);
 262                 zip_header.formated.crc32 = __swap32(zip_header.formated.crc32);
 263                 zip_header.formated.cmpsize = __swap32(zip_header.formated.cmpsize);
 264                 zip_header.formated.ucmpsize = __swap32(zip_header.formated.ucmpsize);
 265                 zip_header.formated.filename_len = __swap16(zip_header.formated.filename_len);
 266                 zip_header.formated.extra_len = __swap16(zip_header.formated.extra_len);
 267 #endif
 268                 if ((zip_header.formated.method != 0) && (zip_header.formated.method != 8)) {
 269                         bb_error_msg_and_die("Unsupported compression method %d", zip_header.formated.method);
 270                 }
 271
 272                 /* Read filename */
 273                 free(dst_fn);
 274                 dst_fn = xmalloc(zip_header.formated.filename_len + 1);
 275                 unzip_read(src_fd, dst_fn, zip_header.formated.filename_len);
 276                 dst_fn[zip_header.formated.filename_len] = 0;
 277
 278                 /* Skip extra header bytes */
 279                 unzip_skip(src_fd, zip_header.formated.extra_len);
 280
 281                 if ((verbosity == v_list) && !list_header_done){
 282                         printf("  Length     Date   Time    Name\n");
 283                         printf(" --------    ----   ----    ----\n");
 284                         list_header_done = 1;
 285                 }
 286
 287                 /* Filter zip entries */
 288                 if (find_list_entry(zreject, dst_fn) ||
 289                         (zaccept && !find_list_entry(zaccept, dst_fn))) { /* Skip entry */
 290                         i = 'n';
 291
 292                 } else { /* Extract entry */
 293                         total_size += zip_header.formated.ucmpsize;
 294
 295                         if (verbosity == v_list) { /* List entry */
 296                                 unsigned int dostime = zip_header.formated.modtime | (zip_header.formated.moddate << 16);
 297                                 printf("%9u  %02u-%02u-%02u %02u:%02u   %s\n",
 298                                            zip_header.formated.ucmpsize,
 299                                            (dostime & 0x01e00000) >> 21,
 300                                            (dostime & 0x001f0000) >> 16,
 301                                            (((dostime & 0xfe000000) >> 25) + 1980) % 100,
 302                                            (dostime & 0x0000f800) >> 11,
 303                                            (dostime & 0x000007e0) >> 5,
 304                                            dst_fn);
 305                                 total_entries++;
 306                                 i = 'n';
 307
 308                         } else if (dst_fd == STDOUT_FILENO) { /* Extracting to STDOUT */
 309                                 i = -1;
 310
 311                         } else if (last_char_is(dst_fn, '/')) { /* Extract directory */
 312                                 if (stat(dst_fn, &stat_buf) == -1) {
 313                                         if (errno != ENOENT) {
 314                                                 bb_perror_msg_and_die("Cannot stat '%s'",dst_fn);
 315                                         }
 316                                         if (verbosity == v_normal) {
 317                                                 printf("   creating: %s\n", dst_fn);
 318                                         }
 319                                         unzip_create_leading_dirs(dst_fn);
 320                                         if (bb_make_directory(dst_fn, 0777, 0)) {
 321                                                 bb_error_msg_and_die("Failed to create directory");
 322                                         }
 323                                 } else {
 324                                         if (!S_ISDIR(stat_buf.st_mode)) {
 325                                                 bb_error_msg_and_die("'%s' exists but is not directory", dst_fn);
 326                                         }
 327                                 }
 328                                 i = 'n';
 329
 330                         } else {  /* Extract file */
 331                         _check_file:
 332                                 if (stat(dst_fn, &stat_buf) == -1) { /* File does not exist */
 333                                         if (errno != ENOENT) {
 334                                                 bb_perror_msg_and_die("Cannot stat '%s'",dst_fn);
 335                                         }
 336                                         i = 'y';
 337
 338                                 } else { /* File already exists */
 339                                         if (overwrite == o_never) {
 340                                                 i = 'n';
 341
 342                                         } else if (S_ISREG(stat_buf.st_mode)) { /* File is regular file */
 343                                                 if (overwrite == o_always) {
 344                                                         i = 'y';
 345                                                 } else {
 346                                                         printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
 347                                                         if (!fgets(key_buf, 512, stdin)) {
 348                                                                 bb_perror_msg_and_die("Cannot read input");
 349                                                         }
 350                                                         i = key_buf[0];
 351                                                 }
 352
 353                                         } else { /* File is not regular file */
 354                                                 bb_error_msg_and_die("'%s' exists but is not regular file",dst_fn);
 355                                         }
 356                                 }
 357                         }
 358                 }
 359
 360                 switch (i) {
 361                 case 'A':
 362                         overwrite = o_always;
 363                 case 'y': /* Open file and fall into unzip */
 364                         unzip_create_leading_dirs(dst_fn);
 365                         dst_fd = bb_xopen(dst_fn, O_WRONLY | O_CREAT);
 366                 case -1: /* Unzip */
 367                         if (verbosity == v_normal) {
 368                                 printf("  inflating: %s\n", dst_fn);
 369                         }
 370                         unzip_extract(&zip_header, src_fd, dst_fd);
 371                         if (dst_fd != STDOUT_FILENO) {
 372                                 /* closing STDOUT is potentially bad for future business */
 373                                 close(dst_fd);
 374                         }
 375                         break;
 376
 377                 case 'N':
 378                         overwrite = o_never;
 379                 case 'n':
 380                         /* Skip entry data */
 381                         unzip_skip(src_fd, zip_header.formated.cmpsize);
 382                         break;
 383
 384                 case 'r':
 385                         /* Prompt for new name */
 386                         printf("new name: ");
 387                         if (!fgets(key_buf, 512, stdin)) {
 388                                 bb_perror_msg_and_die("Cannot read input");
 389                         }
 390                         free(dst_fn);
 391                         dst_fn = bb_xstrdup(key_buf);
 392                         chomp(dst_fn);
 393                         goto _check_file;
 394
 395                 default:
 396                         printf("error: invalid response [%c]\n",(char)i);
 397                         goto _check_file;
 398                 }
 399
 400                 /* Data descriptor section */
 401                 if (zip_header.formated.flags & 4) {
 402                         /* skip over duplicate crc, compressed size and uncompressed size */
 403                         unzip_skip(src_fd, 12);
 404                 }
 405         }
 406
 407         if (verbosity == v_list) {
 408                 printf(" --------                   -------\n"
 409                        "%9d                   %d files\n", total_size, total_entries);
 410         }
 411
 412         return(EXIT_SUCCESS);
 413 }
 414
 415 /* END CODE */
 416 /*
 417 Local Variables:
 418 c-file-style: "linux"
 419 c-basic-offset: 4
 420 tab-width: 4
 421 End:
 422 */