contrib/libarchive/libarchive/archive_read_support_filter_gzip.c

   1 /*-
   2  * Copyright (c) 2003-2007 Tim Kientzle
   3  * All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *    notice, this list of conditions and the following disclaimer in the
  12  *    documentation and/or other materials provided with the distribution.
  13  *
  14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR(S) ``AS IS'' AND ANY EXPRESS OR
  15  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
  16  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
  17  * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
  18  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
  19  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  20  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  21  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  22  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
  23  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  24  */
  25
  26 #include "archive_platform.h"
  27
  28 __FBSDID("$FreeBSD$");
  29
  30
  31 #ifdef HAVE_ERRNO_H
  32 #include <errno.h>
  33 #endif
  34 #ifdef HAVE_STDLIB_H
  35 #include <stdlib.h>
  36 #endif
  37 #ifdef HAVE_STRING_H
  38 #include <string.h>
  39 #endif
  40 #ifdef HAVE_UNISTD_H
  41 #include <unistd.h>
  42 #endif
  43 #ifdef HAVE_ZLIB_H
  44 #include <zlib.h>
  45 #endif
  46
  47 #include "archive.h"
  48 #include "archive_private.h"
  49 #include "archive_read_private.h"
  50
  51 #ifdef HAVE_ZLIB_H
  52 struct private_data {
  53         z_stream         stream;
  54         char             in_stream;
  55         unsigned char   *out_block;
  56         size_t           out_block_size;
  57         int64_t          total_out;
  58         unsigned long    crc;
  59         char             eof; /* True = found end of compressed data. */
  60 };
  61
  62 /* Gzip Filter. */
  63 static ssize_t  gzip_filter_read(struct archive_read_filter *, const void **);
  64 static int      gzip_filter_close(struct archive_read_filter *);
  65 #endif
  66
  67 /*
  68  * Note that we can detect gzip archives even if we can't decompress
  69  * them.  (In fact, we like detecting them because we can give better
  70  * error messages.)  So the bid framework here gets compiled even
  71  * if zlib is unavailable.
  72  *
  73  * TODO: If zlib is unavailable, gzip_bidder_init() should
  74  * use the compress_program framework to try to fire up an external
  75  * gunzip program.
  76  */
  77 static int      gzip_bidder_bid(struct archive_read_filter_bidder *,
  78                     struct archive_read_filter *);
  79 static int      gzip_bidder_init(struct archive_read_filter *);
  80
  81 #if ARCHIVE_VERSION_NUMBER < 4000000
  82 /* Deprecated; remove in libarchive 4.0 */
  83 int
  84 archive_read_support_compression_gzip(struct archive *a)
  85 {
  86         return archive_read_support_filter_gzip(a);
  87 }
  88 #endif
  89
  90 int
  91 archive_read_support_filter_gzip(struct archive *_a)
  92 {
  93         struct archive_read *a = (struct archive_read *)_a;
  94         struct archive_read_filter_bidder *bidder;
  95
  96         archive_check_magic(_a, ARCHIVE_READ_MAGIC,
  97             ARCHIVE_STATE_NEW, "archive_read_support_filter_gzip");
  98
  99         if (__archive_read_get_bidder(a, &bidder) != ARCHIVE_OK)
 100                 return (ARCHIVE_FATAL);
 101
 102         bidder->data = NULL;
 103         bidder->bid = gzip_bidder_bid;
 104         bidder->init = gzip_bidder_init;
 105         bidder->options = NULL;
 106         bidder->free = NULL; /* No data, so no cleanup necessary. */
 107         /* Signal the extent of gzip support with the return value here. */
 108 #if HAVE_ZLIB_H
 109         return (ARCHIVE_OK);
 110 #else
 111         archive_set_error(_a, ARCHIVE_ERRNO_MISC,
 112             "Using external gunzip program");
 113         return (ARCHIVE_WARN);
 114 #endif
 115 }
 116
 117 /*
 118  * Read and verify the header.
 119  *
 120  * Returns zero if the header couldn't be validated, else returns
 121  * number of bytes in header.  If pbits is non-NULL, it receives a
 122  * count of bits verified, suitable for use by bidder.
 123  */
 124 static int
 125 peek_at_header(struct archive_read_filter *filter, int *pbits)
 126 {
 127         const unsigned char *p;
 128         ssize_t avail, len;
 129         int bits = 0;
 130         int header_flags;
 131
 132         /* Start by looking at the first ten bytes of the header, which
 133          * is all fixed layout. */
 134         len = 10;
 135         p = __archive_read_filter_ahead(filter, len, &avail);
 136         if (p == NULL || avail == 0)
 137                 return (0);
 138         /* We only support deflation- third byte must be 0x08. */
 139         if (memcmp(p, "\x1F\x8B\x08", 3) != 0)
 140                 return (0);
 141         bits += 24;
 142         if ((p[3] & 0xE0)!= 0)  /* No reserved flags set. */
 143                 return (0);
 144         bits += 3;
 145         header_flags = p[3];
 146         /* Bytes 4-7 are mod time. */
 147         /* Byte 8 is deflate flags. */
 148         /* XXXX TODO: return deflate flags back to consume_header for use
 149            in initializing the decompressor. */
 150         /* Byte 9 is OS. */
 151
 152         /* Optional extra data:  2 byte length plus variable body. */
 153         if (header_flags & 4) {
 154                 p = __archive_read_filter_ahead(filter, len + 2, &avail);
 155                 if (p == NULL)
 156                         return (0);
 157                 len += ((int)p[len + 1] << 8) | (int)p[len];
 158                 len += 2;
 159         }
 160
 161         /* Null-terminated optional filename. */
 162         if (header_flags & 8) {
 163                 do {
 164                         ++len;
 165                         if (avail < len)
 166                                 p = __archive_read_filter_ahead(filter,
 167                                     len, &avail);
 168                         if (p == NULL)
 169                                 return (0);
 170                 } while (p[len - 1] != 0);
 171         }
 172
 173         /* Null-terminated optional comment. */
 174         if (header_flags & 16) {
 175                 do {
 176                         ++len;
 177                         if (avail < len)
 178                                 p = __archive_read_filter_ahead(filter,
 179                                     len, &avail);
 180                         if (p == NULL)
 181                                 return (0);
 182                 } while (p[len - 1] != 0);
 183         }
 184
 185         /* Optional header CRC */
 186         if ((header_flags & 2)) {
 187                 p = __archive_read_filter_ahead(filter, len + 2, &avail);
 188                 if (p == NULL)
 189                         return (0);
 190 #if 0
 191         int hcrc = ((int)p[len + 1] << 8) | (int)p[len];
 192         int crc = /* XXX TODO: Compute header CRC. */;
 193         if (crc != hcrc)
 194                 return (0);
 195         bits += 16;
 196 #endif
 197                 len += 2;
 198         }
 199
 200         if (pbits != NULL)
 201                 *pbits = bits;
 202         return (len);
 203 }
 204
 205 /*
 206  * Bidder just verifies the header and returns the number of verified bits.
 207  */
 208 static int
 209 gzip_bidder_bid(struct archive_read_filter_bidder *self,
 210     struct archive_read_filter *filter)
 211 {
 212         int bits_checked;
 213
 214         (void)self; /* UNUSED */
 215
 216         if (peek_at_header(filter, &bits_checked))
 217                 return (bits_checked);
 218         return (0);
 219 }
 220
 221
 222 #ifndef HAVE_ZLIB_H
 223
 224 /*
 225  * If we don't have the library on this system, we can't do the
 226  * decompression directly.  We can, however, try to run gunzip
 227  * in case that's available.
 228  */
 229 static int
 230 gzip_bidder_init(struct archive_read_filter *self)
 231 {
 232         int r;
 233
 234         r = __archive_read_program(self, "gunzip");
 235         /* Note: We set the format here even if __archive_read_program()
 236          * above fails.  We do, after all, know what the format is
 237          * even if we weren't able to read it. */
 238         self->code = ARCHIVE_COMPRESSION_GZIP;
 239         self->name = "gzip";
 240         return (r);
 241 }
 242
 243 #else
 244
 245 /*
 246  * Initialize the filter object.
 247  */
 248 static int
 249 gzip_bidder_init(struct archive_read_filter *self)
 250 {
 251         struct private_data *state;
 252         static const size_t out_block_size = 64 * 1024;
 253         void *out_block;
 254
 255         self->code = ARCHIVE_COMPRESSION_GZIP;
 256         self->name = "gzip";
 257
 258         state = (struct private_data *)calloc(sizeof(*state), 1);
 259         out_block = (unsigned char *)malloc(out_block_size);
 260         if (state == NULL || out_block == NULL) {
 261                 free(out_block);
 262                 free(state);
 263                 archive_set_error(&self->archive->archive, ENOMEM,
 264                     "Can't allocate data for gzip decompression");
 265                 return (ARCHIVE_FATAL);
 266         }
 267
 268         self->data = state;
 269         state->out_block_size = out_block_size;
 270         state->out_block = out_block;
 271         self->read = gzip_filter_read;
 272         self->skip = NULL; /* not supported */
 273         self->close = gzip_filter_close;
 274
 275         state->in_stream = 0; /* We're not actually within a stream yet. */
 276
 277         return (ARCHIVE_OK);
 278 }
 279
 280 static int
 281 consume_header(struct archive_read_filter *self)
 282 {
 283         struct private_data *state;
 284         ssize_t avail;
 285         size_t len;
 286         int ret;
 287
 288         state = (struct private_data *)self->data;
 289
 290         /* If this is a real header, consume it. */
 291         len = peek_at_header(self->upstream, NULL);
 292         if (len == 0)
 293                 return (ARCHIVE_EOF);
 294         __archive_read_filter_consume(self->upstream, len);
 295
 296         /* Initialize CRC accumulator. */
 297         state->crc = crc32(0L, NULL, 0);
 298
 299         /* Initialize compression library. */
 300         state->stream.next_in = (unsigned char *)(uintptr_t)
 301             __archive_read_filter_ahead(self->upstream, 1, &avail);
 302         state->stream.avail_in = avail;
 303         ret = inflateInit2(&(state->stream),
 304             -15 /* Don't check for zlib header */);
 305
 306         /* Decipher the error code. */
 307         switch (ret) {
 308         case Z_OK:
 309                 state->in_stream = 1;
 310                 return (ARCHIVE_OK);
 311         case Z_STREAM_ERROR:
 312                 archive_set_error(&self->archive->archive,
 313                     ARCHIVE_ERRNO_MISC,
 314                     "Internal error initializing compression library: "
 315                     "invalid setup parameter");
 316                 break;
 317         case Z_MEM_ERROR:
 318                 archive_set_error(&self->archive->archive, ENOMEM,
 319                     "Internal error initializing compression library: "
 320                     "out of memory");
 321                 break;
 322         case Z_VERSION_ERROR:
 323                 archive_set_error(&self->archive->archive,
 324                     ARCHIVE_ERRNO_MISC,
 325                     "Internal error initializing compression library: "
 326                     "invalid library version");
 327                 break;
 328         default:
 329                 archive_set_error(&self->archive->archive,
 330                     ARCHIVE_ERRNO_MISC,
 331                     "Internal error initializing compression library: "
 332                     " Zlib error %d", ret);
 333                 break;
 334         }
 335         return (ARCHIVE_FATAL);
 336 }
 337
 338 static int
 339 consume_trailer(struct archive_read_filter *self)
 340 {
 341         struct private_data *state;
 342         const unsigned char *p;
 343         ssize_t avail;
 344
 345         state = (struct private_data *)self->data;
 346
 347         state->in_stream = 0;
 348         switch (inflateEnd(&(state->stream))) {
 349         case Z_OK:
 350                 break;
 351         default:
 352                 archive_set_error(&self->archive->archive,
 353                     ARCHIVE_ERRNO_MISC,
 354                     "Failed to clean up gzip decompressor");
 355                 return (ARCHIVE_FATAL);
 356         }
 357
 358         /* GZip trailer is a fixed 8 byte structure. */
 359         p = __archive_read_filter_ahead(self->upstream, 8, &avail);
 360         if (p == NULL || avail == 0)
 361                 return (ARCHIVE_FATAL);
 362
 363         /* XXX TODO: Verify the length and CRC. */
 364
 365         /* We've verified the trailer, so consume it now. */
 366         __archive_read_filter_consume(self->upstream, 8);
 367
 368         return (ARCHIVE_OK);
 369 }
 370
 371 static ssize_t
 372 gzip_filter_read(struct archive_read_filter *self, const void **p)
 373 {
 374         struct private_data *state;
 375         size_t decompressed;
 376         ssize_t avail_in;
 377         int ret;
 378
 379         state = (struct private_data *)self->data;
 380
 381         /* Empty our output buffer. */
 382         state->stream.next_out = state->out_block;
 383         state->stream.avail_out = state->out_block_size;
 384
 385         /* Try to fill the output buffer. */
 386         while (state->stream.avail_out > 0 && !state->eof) {
 387                 /* If we're not in a stream, read a header
 388                  * and initialize the decompression library. */
 389                 if (!state->in_stream) {
 390                         ret = consume_header(self);
 391                         if (ret == ARCHIVE_EOF) {
 392                                 state->eof = 1;
 393                                 break;
 394                         }
 395                         if (ret < ARCHIVE_OK)
 396                                 return (ret);
 397                 }
 398
 399                 /* Peek at the next available data. */
 400                 /* ZLib treats stream.next_in as const but doesn't declare
 401                  * it so, hence this ugly cast. */
 402                 state->stream.next_in = (unsigned char *)(uintptr_t)
 403                     __archive_read_filter_ahead(self->upstream, 1, &avail_in);
 404                 if (state->stream.next_in == NULL) {
 405                         archive_set_error(&self->archive->archive,
 406                             ARCHIVE_ERRNO_MISC,
 407                             "truncated gzip input");
 408                         return (ARCHIVE_FATAL);
 409                 }
 410                 state->stream.avail_in = avail_in;
 411
 412                 /* Decompress and consume some of that data. */
 413                 ret = inflate(&(state->stream), 0);
 414                 switch (ret) {
 415                 case Z_OK: /* Decompressor made some progress. */
 416                         __archive_read_filter_consume(self->upstream,
 417                             avail_in - state->stream.avail_in);
 418                         break;
 419                 case Z_STREAM_END: /* Found end of stream. */
 420                         __archive_read_filter_consume(self->upstream,
 421                             avail_in - state->stream.avail_in);
 422                         /* Consume the stream trailer; release the
 423                          * decompression library. */
 424                         ret = consume_trailer(self);
 425                         if (ret < ARCHIVE_OK)
 426                                 return (ret);
 427                         break;
 428                 default:
 429                         /* Return an error. */
 430                         archive_set_error(&self->archive->archive,
 431                             ARCHIVE_ERRNO_MISC,
 432                             "gzip decompression failed");
 433                         return (ARCHIVE_FATAL);
 434                 }
 435         }
 436
 437         /* We've read as much as we can. */
 438         decompressed = state->stream.next_out - state->out_block;
 439         state->total_out += decompressed;
 440         if (decompressed == 0)
 441                 *p = NULL;
 442         else
 443                 *p = state->out_block;
 444         return (decompressed);
 445 }
 446
 447 /*
 448  * Clean up the decompressor.
 449  */
 450 static int
 451 gzip_filter_close(struct archive_read_filter *self)
 452 {
 453         struct private_data *state;
 454         int ret;
 455
 456         state = (struct private_data *)self->data;
 457         ret = ARCHIVE_OK;
 458
 459         if (state->in_stream) {
 460                 switch (inflateEnd(&(state->stream))) {
 461                 case Z_OK:
 462                         break;
 463                 default:
 464                         archive_set_error(&(self->archive->archive),
 465                             ARCHIVE_ERRNO_MISC,
 466                             "Failed to clean up gzip compressor");
 467                         ret = ARCHIVE_FATAL;
 468                 }
 469         }
 470
 471         free(state->out_block);
 472         free(state);
 473         return (ret);
 474 }
 475
 476 #endif /* HAVE_ZLIB_H */