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 #define ZIP_FILEHEADER_MAGIC            SWAP_LE32(0x04034b50)
  36 #define ZIP_CDS_MAGIC                   SWAP_LE32(0x02014b50)
  37 #define ZIP_CDS_END_MAGIC               SWAP_LE32(0x06054b50)
  38 #define ZIP_DD_MAGIC                    SWAP_LE32(0x08074b50)
  39
  40 extern unsigned int gunzip_crc;
  41 extern unsigned int gunzip_bytes_out;
  42
  43 typedef union {
  44         unsigned char raw[26];
  45         struct {
  46                 unsigned short version; /* 0-1 */
  47                 unsigned short flags;   /* 2-3 */
  48                 unsigned short method;  /* 4-5 */
  49                 unsigned short modtime; /* 6-7 */
  50                 unsigned short moddate; /* 8-9 */
  51                 unsigned int crc32 ATTRIBUTE_PACKED;    /* 10-13 */
  52                 unsigned int cmpsize ATTRIBUTE_PACKED;  /* 14-17 */
  53                 unsigned int ucmpsize ATTRIBUTE_PACKED; /* 18-21 */
  54                 unsigned short filename_len;    /* 22-23 */
  55                 unsigned short extra_len;               /* 24-25 */
  56         } formated ATTRIBUTE_PACKED;
  57 } zip_header_t;
  58
  59 static void unzip_skip(int fd, off_t skip)
  60 {
  61         if (lseek(fd, skip, SEEK_CUR) == (off_t)-1) {
  62                 if ((errno != ESPIPE) || (bb_copyfd_size(fd, -1, skip) != skip)) {
  63                         bb_error_msg_and_die("Seek failure");
  64                 }
  65         }
  66 }
  67
  68 static void unzip_create_leading_dirs(char *fn)
  69 {
  70         /* Create all leading directories */
  71         char *name = bb_xstrdup(fn);
  72         if (bb_make_directory(dirname(name), 0777, FILEUTILS_RECUR)) {
  73                 bb_error_msg_and_die("Exiting"); /* bb_make_directory is noisy */
  74         }
  75         free(name);
  76 }
  77
  78 static int unzip_extract(zip_header_t *zip_header, int src_fd, int dst_fd)
  79 {
  80         if (zip_header->formated.method == 0) {
  81                 /* Method 0 - stored (not compressed) */
  82                 int size = zip_header->formated.ucmpsize;
  83                 if (size && (bb_copyfd_size(src_fd, dst_fd, size) != size)) {
  84                         bb_error_msg_and_die("Cannot complete extraction");
  85                 }
  86
  87         } else {
  88                 /* Method 8 - inflate */
  89                 inflate_init(zip_header->formated.cmpsize);
  90                 inflate_unzip(src_fd, dst_fd);
  91                 inflate_cleanup();
  92                 /* Validate decompression - crc */
  93                 if (zip_header->formated.crc32 != (gunzip_crc ^ 0xffffffffL)) {
  94                         bb_error_msg("Invalid compressed data--crc error");
  95                         return 1;
  96                 }
  97                 /* Validate decompression - size */
  98                 if (zip_header->formated.ucmpsize != gunzip_bytes_out) {
  99                         bb_error_msg("Invalid compressed data--length error");
 100                         return 1;
 101                 }
 102         }
 103         return 0;
 104 }
 105
 106 int unzip_main(int argc, char **argv)
 107 {
 108         zip_header_t zip_header;
 109         enum {v_silent, v_normal, v_list} verbosity = v_normal;
 110         enum {o_prompt, o_never, o_always} overwrite = o_prompt;
 111         unsigned int total_size = 0;
 112         unsigned int total_entries = 0;
 113         int src_fd = -1, dst_fd = -1;
 114         char *src_fn = NULL, *dst_fn = NULL;
 115         llist_t *zaccept = NULL;
 116         llist_t *zreject = NULL;
 117         char *base_dir = NULL;
 118         int failed, i, opt, opt_range = 0, list_header_done = 0;
 119         char key_buf[512];
 120         struct stat stat_buf;
 121
 122         while((opt = getopt(argc, argv, "-d:lnopqx")) != -1) {
 123                 switch(opt_range) {
 124                 case 0: /* Options */
 125                         switch(opt) {
 126                         case 'l': /* List */
 127                                 verbosity = v_list;
 128                                 break;
 129
 130                         case 'n': /* Never overwrite existing files */
 131                                 overwrite = o_never;
 132                                 break;
 133
 134                         case 'o': /* Always overwrite existing files */
 135                                 overwrite = o_always;
 136                                 break;
 137
 138                         case 'p': /* Extract files to stdout and fall through to set verbosity */
 139                                 dst_fd = STDOUT_FILENO;
 140
 141                         case 'q': /* Be quiet */
 142                                 verbosity = (verbosity == v_normal) ? v_silent : verbosity;
 143                                 break;
 144
 145                         case 1 : /* The zip file */
 146                                 src_fn = bb_xstrndup(optarg, strlen(optarg)+4);
 147                                 opt_range++;
 148                                 break;
 149
 150                         default:
 151                                 bb_show_usage();
 152
 153                         }
 154                         break;
 155
 156                 case 1: /* Include files */
 157                         if (opt == 1) {
 158                                 llist_add_to(&zaccept, optarg);
 159
 160                         } else if (opt == 'd') {
 161                                 base_dir = optarg;
 162                                 opt_range += 2;
 163
 164                         } else if (opt == 'x') {
 165                                 opt_range++;
 166
 167                         } else {
 168                                 bb_show_usage();
 169                         }
 170                         break;
 171
 172                 case 2 : /* Exclude files */
 173                         if (opt == 1) {
 174                                 llist_add_to(&zreject, optarg);
 175
 176                         } else if (opt == 'd') { /* Extract to base directory */
 177                                 base_dir = optarg;
 178                                 opt_range++;
 179
 180                         } else {
 181                                 bb_show_usage();
 182                         }
 183                         break;
 184
 185                 default:
 186                         bb_show_usage();
 187                 }
 188         }
 189
 190         if (src_fn == NULL) {
 191                 bb_show_usage();
 192         }
 193
 194         /* Open input file */
 195         if (strcmp("-", src_fn) == 0) {
 196                 src_fd = STDIN_FILENO;
 197                 /* Cannot use prompt mode since zip data is arriving on STDIN */
 198                 overwrite = (overwrite == o_prompt) ? o_never : overwrite;
 199
 200         } else {
 201                 static const char *const extn[] = {"", ".zip", ".ZIP"};
 202                 int orig_src_fn_len = strlen(src_fn);
 203                 for(i = 0; (i < 3) && (src_fd == -1); i++) {
 204                         strcpy(src_fn + orig_src_fn_len, extn[i]);
 205                         src_fd = open(src_fn, O_RDONLY);
 206                 }
 207                 if (src_fd == -1) {
 208                         src_fn[orig_src_fn_len] = 0;
 209                         bb_error_msg_and_die("Cannot open %s, %s.zip, %s.ZIP", src_fn, src_fn, src_fn);
 210                 }
 211         }
 212
 213         /* Change dir if necessary */
 214         if (base_dir)
 215                 bb_xchdir(base_dir);
 216
 217         if (verbosity != v_silent)
 218                 printf("Archive:  %s\n", src_fn);
 219
 220         failed = 0;
 221
 222         while (1) {
 223                 unsigned int magic;
 224
 225                 /* Check magic number */
 226                 xread(src_fd, &magic, 4);
 227                 if (magic == ZIP_CDS_MAGIC) {
 228                         break;
 229                 } else if (magic != ZIP_FILEHEADER_MAGIC) {
 230                         bb_error_msg_and_die("Invalid zip magic %08X", magic);
 231                 }
 232
 233                 /* Read the file header */
 234                 xread(src_fd, zip_header.raw, 26);
 235                 zip_header.formated.version = SWAP_LE32(zip_header.formated.version);
 236                 zip_header.formated.flags = SWAP_LE32(zip_header.formated.flags);
 237                 zip_header.formated.method = SWAP_LE32(zip_header.formated.method);
 238                 zip_header.formated.modtime = SWAP_LE32(zip_header.formated.modtime);
 239                 zip_header.formated.moddate = SWAP_LE32(zip_header.formated.moddate);
 240                 zip_header.formated.crc32 = SWAP_LE32(zip_header.formated.crc32);
 241                 zip_header.formated.cmpsize = SWAP_LE32(zip_header.formated.cmpsize);
 242                 zip_header.formated.ucmpsize = SWAP_LE32(zip_header.formated.ucmpsize);
 243                 zip_header.formated.filename_len = SWAP_LE32(zip_header.formated.filename_len);
 244                 zip_header.formated.extra_len = SWAP_LE32(zip_header.formated.extra_len);
 245                 if ((zip_header.formated.method != 0) && (zip_header.formated.method != 8)) {
 246                         bb_error_msg_and_die("Unsupported compression method %d", zip_header.formated.method);
 247                 }
 248
 249                 /* Read filename */
 250                 free(dst_fn);
 251                 dst_fn = xzalloc(zip_header.formated.filename_len + 1);
 252                 xread(src_fd, dst_fn, zip_header.formated.filename_len);
 253
 254                 /* Skip extra header bytes */
 255                 unzip_skip(src_fd, zip_header.formated.extra_len);
 256
 257                 if ((verbosity == v_list) && !list_header_done){
 258                         printf("  Length     Date   Time    Name\n"
 259                                    " --------    ----   ----    ----\n");
 260                         list_header_done = 1;
 261                 }
 262
 263                 /* Filter zip entries */
 264                 if (find_list_entry(zreject, dst_fn) ||
 265                         (zaccept && !find_list_entry(zaccept, dst_fn))) { /* Skip entry */
 266                         i = 'n';
 267
 268                 } else { /* Extract entry */
 269                         total_size += zip_header.formated.ucmpsize;
 270
 271                         if (verbosity == v_list) { /* List entry */
 272                                 unsigned int dostime = zip_header.formated.modtime | (zip_header.formated.moddate << 16);
 273                                 printf("%9u  %02u-%02u-%02u %02u:%02u   %s\n",
 274                                            zip_header.formated.ucmpsize,
 275                                            (dostime & 0x01e00000) >> 21,
 276                                            (dostime & 0x001f0000) >> 16,
 277                                            (((dostime & 0xfe000000) >> 25) + 1980) % 100,
 278                                            (dostime & 0x0000f800) >> 11,
 279                                            (dostime & 0x000007e0) >> 5,
 280                                            dst_fn);
 281                                 total_entries++;
 282                                 i = 'n';
 283
 284                         } else if (dst_fd == STDOUT_FILENO) { /* Extracting to STDOUT */
 285                                 i = -1;
 286
 287                         } else if (last_char_is(dst_fn, '/')) { /* Extract directory */
 288                                 if (stat(dst_fn, &stat_buf) == -1) {
 289                                         if (errno != ENOENT) {
 290                                                 bb_perror_msg_and_die("Cannot stat '%s'",dst_fn);
 291                                         }
 292                                         if (verbosity == v_normal) {
 293                                                 printf("   creating: %s\n", dst_fn);
 294                                         }
 295                                         unzip_create_leading_dirs(dst_fn);
 296                                         if (bb_make_directory(dst_fn, 0777, 0)) {
 297                                                 bb_error_msg_and_die("Exiting");
 298                                         }
 299                                 } else {
 300                                         if (!S_ISDIR(stat_buf.st_mode)) {
 301                                                 bb_error_msg_and_die("'%s' exists but is not directory", dst_fn);
 302                                         }
 303                                 }
 304                                 i = 'n';
 305
 306                         } else {  /* Extract file */
 307                         _check_file:
 308                                 if (stat(dst_fn, &stat_buf) == -1) { /* File does not exist */
 309                                         if (errno != ENOENT) {
 310                                                 bb_perror_msg_and_die("Cannot stat '%s'",dst_fn);
 311                                         }
 312                                         i = 'y';
 313
 314                                 } else { /* File already exists */
 315                                         if (overwrite == o_never) {
 316                                                 i = 'n';
 317
 318                                         } else if (S_ISREG(stat_buf.st_mode)) { /* File is regular file */
 319                                                 if (overwrite == o_always) {
 320                                                         i = 'y';
 321                                                 } else {
 322                                                         printf("replace %s? [y]es, [n]o, [A]ll, [N]one, [r]ename: ", dst_fn);
 323                                                         if (!fgets(key_buf, 512, stdin)) {
 324                                                                 bb_perror_msg_and_die("Cannot read input");
 325                                                         }
 326                                                         i = key_buf[0];
 327                                                 }
 328
 329                                         } else { /* File is not regular file */
 330                                                 bb_error_msg_and_die("'%s' exists but is not regular file",dst_fn);
 331                                         }
 332                                 }
 333                         }
 334                 }
 335
 336                 switch (i) {
 337                 case 'A':
 338                         overwrite = o_always;
 339                 case 'y': /* Open file and fall into unzip */
 340                         unzip_create_leading_dirs(dst_fn);
 341                         dst_fd = bb_xopen(dst_fn, O_WRONLY | O_CREAT);
 342                 case -1: /* Unzip */
 343                         if (verbosity == v_normal) {
 344                                 printf("  inflating: %s\n", dst_fn);
 345                         }
 346                         if (unzip_extract(&zip_header, src_fd, dst_fd)) {
 347                             failed = 1;
 348                         }
 349                         if (dst_fd != STDOUT_FILENO) {
 350                                 /* closing STDOUT is potentially bad for future business */
 351                                 close(dst_fd);
 352                         }
 353                         break;
 354
 355                 case 'N':
 356                         overwrite = o_never;
 357                 case 'n':
 358                         /* Skip entry data */
 359                         unzip_skip(src_fd, zip_header.formated.cmpsize);
 360                         break;
 361
 362                 case 'r':
 363                         /* Prompt for new name */
 364                         printf("new name: ");
 365                         if (!fgets(key_buf, 512, stdin)) {
 366                                 bb_perror_msg_and_die("Cannot read input");
 367                         }
 368                         free(dst_fn);
 369                         dst_fn = bb_xstrdup(key_buf);
 370                         chomp(dst_fn);
 371                         goto _check_file;
 372
 373                 default:
 374                         printf("error: invalid response [%c]\n",(char)i);
 375                         goto _check_file;
 376                 }
 377
 378                 /* Data descriptor section */
 379                 if (zip_header.formated.flags & 4) {
 380                         /* skip over duplicate crc, compressed size and uncompressed size */
 381                         unzip_skip(src_fd, 12);
 382                 }
 383         }
 384
 385         if (verbosity == v_list) {
 386                 printf(" --------                   -------\n"
 387                        "%9d                   %d files\n", total_size, total_entries);
 388         }
 389
 390         return failed;
 391 }