contrib/libarchive/libarchive/archive_read_support_compression_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: src/lib/libarchive/archive_read_support_compression_gzip.c,v 1.17 2008/12/06 06:45:15 kientzle Exp $");
  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 int
  82 archive_read_support_compression_gzip(struct archive *_a)
  83 {
  84         struct archive_read *a = (struct archive_read *)_a;
  85         struct archive_read_filter_bidder *bidder = __archive_read_get_bidder(a);
  86
  87         if (bidder == NULL)
  88                 return (ARCHIVE_FATAL);
  89
  90         bidder->data = NULL;
  91         bidder->bid = gzip_bidder_bid;
  92         bidder->init = gzip_bidder_init;
  93         bidder->options = NULL;
  94         bidder->free = NULL; /* No data, so no cleanup necessary. */
  95         /* Signal the extent of gzip support with the return value here. */
  96 #if HAVE_ZLIB_H
  97         return (ARCHIVE_OK);
  98 #else
  99         archive_set_error(_a, ARCHIVE_ERRNO_MISC,
 100             "Using external gunzip program");
 101         return (ARCHIVE_WARN);
 102 #endif
 103 }
 104
 105 /*
 106  * Read and verify the header.
 107  *
 108  * Returns zero if the header couldn't be validated, else returns
 109  * number of bytes in header.  If pbits is non-NULL, it receives a
 110  * count of bits verified, suitable for use by bidder.
 111  */
 112 static int
 113 peek_at_header(struct archive_read_filter *filter, int *pbits)
 114 {
 115         const unsigned char *p;
 116         ssize_t avail, len;
 117         int bits = 0;
 118         int header_flags;
 119
 120         /* Start by looking at the first ten bytes of the header, which
 121          * is all fixed layout. */
 122         len = 10;
 123         p = __archive_read_filter_ahead(filter, len, &avail);
 124         if (p == NULL || avail == 0)
 125                 return (0);
 126         if (p[0] != 037)
 127                 return (0);
 128         bits += 8;
 129         if (p[1] != 0213)
 130                 return (0);
 131         bits += 8;
 132         if (p[2] != 8) /* We only support deflation. */
 133                 return (0);
 134         bits += 8;
 135         if ((p[3] & 0xE0)!= 0)  /* No reserved flags set. */
 136                 return (0);
 137         bits += 3;
 138         header_flags = p[3];
 139         /* Bytes 4-7 are mod time. */
 140         /* Byte 8 is deflate flags. */
 141         /* XXXX TODO: return deflate flags back to consume_header for use
 142            in initializing the decompressor. */
 143         /* Byte 9 is OS. */
 144
 145         /* Optional extra data:  2 byte length plus variable body. */
 146         if (header_flags & 4) {
 147                 p = __archive_read_filter_ahead(filter, len + 2, &avail);
 148                 if (p == NULL)
 149                         return (0);
 150                 len += ((int)p[len + 1] << 8) | (int)p[len];
 151         }
 152
 153         /* Null-terminated optional filename. */
 154         if (header_flags & 8) {
 155                 do {
 156                         ++len;
 157                         if (avail < len)
 158                                 p = __archive_read_filter_ahead(filter,
 159                                     len, &avail);
 160                         if (p == NULL)
 161                                 return (0);
 162                 } while (p[len - 1] != 0);
 163         }
 164
 165         /* Null-terminated optional comment. */
 166         if (header_flags & 16) {
 167                 do {
 168                         ++len;
 169                         if (avail < len)
 170                                 p = __archive_read_filter_ahead(filter,
 171                                     len, &avail);
 172                         if (p == NULL)
 173                                 return (0);
 174                 } while (p[len - 1] != 0);
 175         }
 176
 177         /* Optional header CRC */
 178         if ((header_flags & 2)) {
 179                 p = __archive_read_filter_ahead(filter, len + 2, &avail);
 180                 if (p == NULL)
 181                         return (0);
 182 #if 0
 183         int hcrc = ((int)p[len + 1] << 8) | (int)p[len];
 184         int crc = /* XXX TODO: Compute header CRC. */;
 185         if (crc != hcrc)
 186                 return (0);
 187         bits += 16;
 188 #endif
 189                 len += 2;
 190         }
 191
 192         if (pbits != NULL)
 193                 *pbits = bits;
 194         return (len);
 195 }
 196
 197 /*
 198  * Bidder just verifies the header and returns the number of verified bits.
 199  */
 200 static int
 201 gzip_bidder_bid(struct archive_read_filter_bidder *self,
 202     struct archive_read_filter *filter)
 203 {
 204         int bits_checked;
 205
 206         (void)self; /* UNUSED */
 207
 208         if (peek_at_header(filter, &bits_checked))
 209                 return (bits_checked);
 210         return (0);
 211 }
 212
 213
 214 #ifndef HAVE_ZLIB_H
 215
 216 /*
 217  * If we don't have the library on this system, we can't do the
 218  * decompression directly.  We can, however, try to run gunzip
 219  * in case that's available.
 220  */
 221 static int
 222 gzip_bidder_init(struct archive_read_filter *self)
 223 {
 224         int r;
 225
 226         r = __archive_read_program(self, "gunzip");
 227         /* Note: We set the format here even if __archive_read_program()
 228          * above fails.  We do, after all, know what the format is
 229          * even if we weren't able to read it. */
 230         self->code = ARCHIVE_COMPRESSION_GZIP;
 231         self->name = "gzip";
 232         return (r);
 233 }
 234
 235 #else
 236
 237 /*
 238  * Initialize the filter object.
 239  */
 240 static int
 241 gzip_bidder_init(struct archive_read_filter *self)
 242 {
 243         struct private_data *state;
 244         static const size_t out_block_size = 64 * 1024;
 245         void *out_block;
 246
 247         self->code = ARCHIVE_COMPRESSION_GZIP;
 248         self->name = "gzip";
 249
 250         state = (struct private_data *)calloc(sizeof(*state), 1);
 251         out_block = (unsigned char *)malloc(out_block_size);
 252         if (state == NULL || out_block == NULL) {
 253                 free(out_block);
 254                 free(state);
 255                 archive_set_error(&self->archive->archive, ENOMEM,
 256                     "Can't allocate data for gzip decompression");
 257                 return (ARCHIVE_FATAL);
 258         }
 259
 260         self->data = state;
 261         state->out_block_size = out_block_size;
 262         state->out_block = out_block;
 263         self->read = gzip_filter_read;
 264         self->skip = NULL; /* not supported */
 265         self->close = gzip_filter_close;
 266
 267         state->in_stream = 0; /* We're not actually within a stream yet. */
 268
 269         return (ARCHIVE_OK);
 270 }
 271
 272 static int
 273 consume_header(struct archive_read_filter *self)
 274 {
 275         struct private_data *state;
 276         ssize_t avail;
 277         size_t len;
 278         int ret;
 279
 280         state = (struct private_data *)self->data;
 281
 282         /* If this is a real header, consume it. */
 283         len = peek_at_header(self->upstream, NULL);
 284         if (len == 0)
 285                 return (ARCHIVE_EOF);
 286         __archive_read_filter_consume(self->upstream, len);
 287
 288         /* Initialize CRC accumulator. */
 289         state->crc = crc32(0L, NULL, 0);
 290
 291         /* Initialize compression library. */
 292         state->stream.next_in = (unsigned char *)(uintptr_t)
 293             __archive_read_filter_ahead(self->upstream, 1, &avail);
 294         state->stream.avail_in = avail;
 295         ret = inflateInit2(&(state->stream),
 296             -15 /* Don't check for zlib header */);
 297
 298         /* Decipher the error code. */
 299         switch (ret) {
 300         case Z_OK:
 301                 state->in_stream = 1;
 302                 return (ARCHIVE_OK);
 303         case Z_STREAM_ERROR:
 304                 archive_set_error(&self->archive->archive,
 305                     ARCHIVE_ERRNO_MISC,
 306                     "Internal error initializing compression library: "
 307                     "invalid setup parameter");
 308                 break;
 309         case Z_MEM_ERROR:
 310                 archive_set_error(&self->archive->archive, ENOMEM,
 311                     "Internal error initializing compression library: "
 312                     "out of memory");
 313                 break;
 314         case Z_VERSION_ERROR:
 315                 archive_set_error(&self->archive->archive,
 316                     ARCHIVE_ERRNO_MISC,
 317                     "Internal error initializing compression library: "
 318                     "invalid library version");
 319                 break;
 320         default:
 321                 archive_set_error(&self->archive->archive,
 322                     ARCHIVE_ERRNO_MISC,
 323                     "Internal error initializing compression library: "
 324                     " Zlib error %d", ret);
 325                 break;
 326         }
 327         return (ARCHIVE_FATAL);
 328 }
 329
 330 static int
 331 consume_trailer(struct archive_read_filter *self)
 332 {
 333         struct private_data *state;
 334         const unsigned char *p;
 335         ssize_t avail;
 336
 337         state = (struct private_data *)self->data;
 338
 339         state->in_stream = 0;
 340         switch (inflateEnd(&(state->stream))) {
 341         case Z_OK:
 342                 break;
 343         default:
 344                 archive_set_error(&self->archive->archive,
 345                     ARCHIVE_ERRNO_MISC,
 346                     "Failed to clean up gzip decompressor");
 347                 return (ARCHIVE_FATAL);
 348         }
 349
 350         /* GZip trailer is a fixed 8 byte structure. */
 351         p = __archive_read_filter_ahead(self->upstream, 8, &avail);
 352         if (p == NULL || avail == 0)
 353                 return (ARCHIVE_FATAL);
 354
 355         /* XXX TODO: Verify the length and CRC. */
 356
 357         /* We've verified the trailer, so consume it now. */
 358         __archive_read_filter_consume(self->upstream, 8);
 359
 360         return (ARCHIVE_OK);
 361 }
 362
 363 static ssize_t
 364 gzip_filter_read(struct archive_read_filter *self, const void **p)
 365 {
 366         struct private_data *state;
 367         size_t decompressed;
 368         ssize_t avail_in;
 369         int ret;
 370
 371         state = (struct private_data *)self->data;
 372
 373         /* Empty our output buffer. */
 374         state->stream.next_out = state->out_block;
 375         state->stream.avail_out = state->out_block_size;
 376
 377         /* Try to fill the output buffer. */
 378         while (state->stream.avail_out > 0 && !state->eof) {
 379                 /* If we're not in a stream, read a header
 380                  * and initialize the decompression library. */
 381                 if (!state->in_stream) {
 382                         ret = consume_header(self);
 383                         if (ret == ARCHIVE_EOF) {
 384                                 state->eof = 1;
 385                                 break;
 386                         }
 387                         if (ret < ARCHIVE_OK)
 388                                 return (ret);
 389                 }
 390
 391                 /* Peek at the next available data. */
 392                 /* ZLib treats stream.next_in as const but doesn't declare
 393                  * it so, hence this ugly cast. */
 394                 state->stream.next_in = (unsigned char *)(uintptr_t)
 395                     __archive_read_filter_ahead(self->upstream, 1, &avail_in);
 396                 if (state->stream.next_in == NULL)
 397                         return (ARCHIVE_FATAL);
 398                 state->stream.avail_in = avail_in;
 399
 400                 /* Decompress and consume some of that data. */
 401                 ret = inflate(&(state->stream), 0);
 402                 switch (ret) {
 403                 case Z_OK: /* Decompressor made some progress. */
 404                         __archive_read_filter_consume(self->upstream,
 405                             avail_in - state->stream.avail_in);
 406                         break;
 407                 case Z_STREAM_END: /* Found end of stream. */
 408                         __archive_read_filter_consume(self->upstream,
 409                             avail_in - state->stream.avail_in);
 410                         /* Consume the stream trailer; release the
 411                          * decompression library. */
 412                         ret = consume_trailer(self);
 413                         break;
 414                 default:
 415                         /* Return an error. */
 416                         archive_set_error(&self->archive->archive,
 417                             ARCHIVE_ERRNO_MISC,
 418                             "gzip decompression failed");
 419                         return (ARCHIVE_FATAL);
 420                 }
 421         }
 422
 423         /* We've read as much as we can. */
 424         decompressed = state->stream.next_out - state->out_block;
 425         state->total_out += decompressed;
 426         if (decompressed == 0)
 427                 *p = NULL;
 428         else
 429                 *p = state->out_block;
 430         return (decompressed);
 431 }
 432
 433 /*
 434  * Clean up the decompressor.
 435  */
 436 static int
 437 gzip_filter_close(struct archive_read_filter *self)
 438 {
 439         struct private_data *state;
 440         int ret;
 441
 442         state = (struct private_data *)self->data;
 443         ret = ARCHIVE_OK;
 444
 445         if (state->in_stream) {
 446                 switch (inflateEnd(&(state->stream))) {
 447                 case Z_OK:
 448                         break;
 449                 default:
 450                         archive_set_error(&(self->archive->archive),
 451                             ARCHIVE_ERRNO_MISC,
 452                             "Failed to clean up gzip compressor");
 453                         ret = ARCHIVE_FATAL;
 454                 }
 455         }
 456
 457         free(state->out_block);
 458         free(state);
 459         return (ret);
 460 }
 461
 462 #endif /* HAVE_ZLIB_H */