contrib/libarchive/libarchive/archive_read_support_filter_bzip2.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 #ifdef HAVE_ERRNO_H
  31 #include <errno.h>
  32 #endif
  33 #include <stdio.h>
  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_BZLIB_H
  44 #include <bzlib.h>
  45 #endif
  46
  47 #include "archive.h"
  48 #include "archive_private.h"
  49 #include "archive_read_private.h"
  50
  51 #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR)
  52 struct private_data {
  53         bz_stream        stream;
  54         char            *out_block;
  55         size_t           out_block_size;
  56         char             valid; /* True = decompressor is initialized */
  57         char             eof; /* True = found end of compressed data. */
  58 };
  59
  60 /* Bzip2 filter */
  61 static ssize_t  bzip2_filter_read(struct archive_read_filter *, const void **);
  62 static int      bzip2_filter_close(struct archive_read_filter *);
  63 #endif
  64
  65 /*
  66  * Note that we can detect bzip2 archives even if we can't decompress
  67  * them.  (In fact, we like detecting them because we can give better
  68  * error messages.)  So the bid framework here gets compiled even
  69  * if bzlib is unavailable.
  70  */
  71 static int      bzip2_reader_bid(struct archive_read_filter_bidder *, struct archive_read_filter *);
  72 static int      bzip2_reader_init(struct archive_read_filter *);
  73 static int      bzip2_reader_free(struct archive_read_filter_bidder *);
  74
  75 #if ARCHIVE_VERSION_NUMBER < 4000000
  76 /* Deprecated; remove in libarchive 4.0 */
  77 int
  78 archive_read_support_compression_bzip2(struct archive *a)
  79 {
  80         return archive_read_support_filter_bzip2(a);
  81 }
  82 #endif
  83
  84 int
  85 archive_read_support_filter_bzip2(struct archive *_a)
  86 {
  87         struct archive_read *a = (struct archive_read *)_a;
  88         struct archive_read_filter_bidder *reader;
  89
  90         archive_check_magic(_a, ARCHIVE_READ_MAGIC,
  91             ARCHIVE_STATE_NEW, "archive_read_support_filter_bzip2");
  92
  93         if (__archive_read_get_bidder(a, &reader) != ARCHIVE_OK)
  94                 return (ARCHIVE_FATAL);
  95
  96         reader->data = NULL;
  97         reader->bid = bzip2_reader_bid;
  98         reader->init = bzip2_reader_init;
  99         reader->options = NULL;
 100         reader->free = bzip2_reader_free;
 101 #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR)
 102         return (ARCHIVE_OK);
 103 #else
 104         archive_set_error(_a, ARCHIVE_ERRNO_MISC,
 105             "Using external bunzip2 program");
 106         return (ARCHIVE_WARN);
 107 #endif
 108 }
 109
 110 static int
 111 bzip2_reader_free(struct archive_read_filter_bidder *self){
 112         (void)self; /* UNUSED */
 113         return (ARCHIVE_OK);
 114 }
 115
 116 /*
 117  * Test whether we can handle this data.
 118  *
 119  * This logic returns zero if any part of the signature fails.  It
 120  * also tries to Do The Right Thing if a very short buffer prevents us
 121  * from verifying as much as we would like.
 122  */
 123 static int
 124 bzip2_reader_bid(struct archive_read_filter_bidder *self, struct archive_read_filter *filter)
 125 {
 126         const unsigned char *buffer;
 127         ssize_t avail;
 128         int bits_checked;
 129
 130         (void)self; /* UNUSED */
 131
 132         /* Minimal bzip2 archive is 14 bytes. */
 133         buffer = __archive_read_filter_ahead(filter, 14, &avail);
 134         if (buffer == NULL)
 135                 return (0);
 136
 137         /* First three bytes must be "BZh" */
 138         bits_checked = 0;
 139         if (memcmp(buffer, "BZh", 3) != 0)
 140                 return (0);
 141         bits_checked += 24;
 142
 143         /* Next follows a compression flag which must be an ASCII digit. */
 144         if (buffer[3] < '1' || buffer[3] > '9')
 145                 return (0);
 146         bits_checked += 5;
 147
 148         /* After BZh[1-9], there must be either a data block
 149          * which begins with 0x314159265359 or an end-of-data
 150          * marker of 0x177245385090. */
 151         if (memcmp(buffer + 4, "\x31\x41\x59\x26\x53\x59", 6) == 0)
 152                 bits_checked += 48;
 153         else if (memcmp(buffer + 4, "\x17\x72\x45\x38\x50\x90", 6) == 0)
 154                 bits_checked += 48;
 155         else
 156                 return (0);
 157
 158         return (bits_checked);
 159 }
 160
 161 #if !defined(HAVE_BZLIB_H) || !defined(BZ_CONFIG_ERROR)
 162
 163 /*
 164  * If we don't have the library on this system, we can't actually do the
 165  * decompression.  We can, however, still detect compressed archives
 166  * and emit a useful message.
 167  */
 168 static int
 169 bzip2_reader_init(struct archive_read_filter *self)
 170 {
 171         int r;
 172
 173         r = __archive_read_program(self, "bunzip2");
 174         /* Note: We set the format here even if __archive_read_program()
 175          * above fails.  We do, after all, know what the format is
 176          * even if we weren't able to read it. */
 177         self->code = ARCHIVE_COMPRESSION_BZIP2;
 178         self->name = "bzip2";
 179         return (r);
 180 }
 181
 182
 183 #else
 184
 185 /*
 186  * Setup the callbacks.
 187  */
 188 static int
 189 bzip2_reader_init(struct archive_read_filter *self)
 190 {
 191         static const size_t out_block_size = 64 * 1024;
 192         void *out_block;
 193         struct private_data *state;
 194
 195         self->code = ARCHIVE_COMPRESSION_BZIP2;
 196         self->name = "bzip2";
 197
 198         state = (struct private_data *)calloc(sizeof(*state), 1);
 199         out_block = (unsigned char *)malloc(out_block_size);
 200         if (state == NULL || out_block == NULL) {
 201                 archive_set_error(&self->archive->archive, ENOMEM,
 202                     "Can't allocate data for bzip2 decompression");
 203                 free(out_block);
 204                 free(state);
 205                 return (ARCHIVE_FATAL);
 206         }
 207
 208         self->data = state;
 209         state->out_block_size = out_block_size;
 210         state->out_block = out_block;
 211         self->read = bzip2_filter_read;
 212         self->skip = NULL; /* not supported */
 213         self->close = bzip2_filter_close;
 214
 215         return (ARCHIVE_OK);
 216 }
 217
 218 /*
 219  * Return the next block of decompressed data.
 220  */
 221 static ssize_t
 222 bzip2_filter_read(struct archive_read_filter *self, const void **p)
 223 {
 224         struct private_data *state;
 225         size_t decompressed;
 226         const char *read_buf;
 227         ssize_t ret;
 228
 229         state = (struct private_data *)self->data;
 230
 231         if (state->eof) {
 232                 *p = NULL;
 233                 return (0);
 234         }
 235
 236         /* Empty our output buffer. */
 237         state->stream.next_out = state->out_block;
 238         state->stream.avail_out = state->out_block_size;
 239
 240         /* Try to fill the output buffer. */
 241         for (;;) {
 242                 if (!state->valid) {
 243                         if (bzip2_reader_bid(self->bidder, self->upstream) == 0) {
 244                                 state->eof = 1;
 245                                 *p = state->out_block;
 246                                 decompressed = state->stream.next_out
 247                                     - state->out_block;
 248                                 return (decompressed);
 249                         }
 250                         /* Initialize compression library. */
 251                         ret = BZ2_bzDecompressInit(&(state->stream),
 252                                            0 /* library verbosity */,
 253                                            0 /* don't use low-mem algorithm */);
 254
 255                         /* If init fails, try low-memory algorithm instead. */
 256                         if (ret == BZ_MEM_ERROR)
 257                                 ret = BZ2_bzDecompressInit(&(state->stream),
 258                                            0 /* library verbosity */,
 259                                            1 /* do use low-mem algo */);
 260
 261                         if (ret != BZ_OK) {
 262                                 const char *detail = NULL;
 263                                 int err = ARCHIVE_ERRNO_MISC;
 264                                 switch (ret) {
 265                                 case BZ_PARAM_ERROR:
 266                                         detail = "invalid setup parameter";
 267                                         break;
 268                                 case BZ_MEM_ERROR:
 269                                         err = ENOMEM;
 270                                         detail = "out of memory";
 271                                         break;
 272                                 case BZ_CONFIG_ERROR:
 273                                         detail = "mis-compiled library";
 274                                         break;
 275                                 }
 276                                 archive_set_error(&self->archive->archive, err,
 277                                     "Internal error initializing decompressor%s%s",
 278                                     detail == NULL ? "" : ": ",
 279                                     detail);
 280                                 return (ARCHIVE_FATAL);
 281                         }
 282                         state->valid = 1;
 283                 }
 284
 285                 /* stream.next_in is really const, but bzlib
 286                  * doesn't declare it so. <sigh> */
 287                 read_buf =
 288                     __archive_read_filter_ahead(self->upstream, 1, &ret);
 289                 if (read_buf == NULL) {
 290                         archive_set_error(&self->archive->archive,
 291                             ARCHIVE_ERRNO_MISC,
 292                             "truncated bzip2 input");
 293                         return (ARCHIVE_FATAL);
 294                 }
 295                 state->stream.next_in = (char *)(uintptr_t)read_buf;
 296                 state->stream.avail_in = ret;
 297                 /* There is no more data, return whatever we have. */
 298                 if (ret == 0) {
 299                         state->eof = 1;
 300                         *p = state->out_block;
 301                         decompressed = state->stream.next_out
 302                             - state->out_block;
 303                         return (decompressed);
 304                 }
 305
 306                 /* Decompress as much as we can in one pass. */
 307                 ret = BZ2_bzDecompress(&(state->stream));
 308                 __archive_read_filter_consume(self->upstream,
 309                     state->stream.next_in - read_buf);
 310
 311                 switch (ret) {
 312                 case BZ_STREAM_END: /* Found end of stream. */
 313                         switch (BZ2_bzDecompressEnd(&(state->stream))) {
 314                         case BZ_OK:
 315                                 break;
 316                         default:
 317                                 archive_set_error(&(self->archive->archive),
 318                                           ARCHIVE_ERRNO_MISC,
 319                                           "Failed to clean up decompressor");
 320                                 return (ARCHIVE_FATAL);
 321                         }
 322                         state->valid = 0;
 323                         /* FALLTHROUGH */
 324                 case BZ_OK: /* Decompressor made some progress. */
 325                         /* If we filled our buffer, update stats and return. */
 326                         if (state->stream.avail_out == 0) {
 327                                 *p = state->out_block;
 328                                 decompressed = state->stream.next_out
 329                                     - state->out_block;
 330                                 return (decompressed);
 331                         }
 332                         break;
 333                 default: /* Return an error. */
 334                         archive_set_error(&self->archive->archive,
 335                             ARCHIVE_ERRNO_MISC, "bzip decompression failed");
 336                         return (ARCHIVE_FATAL);
 337                 }
 338         }
 339 }
 340
 341 /*
 342  * Clean up the decompressor.
 343  */
 344 static int
 345 bzip2_filter_close(struct archive_read_filter *self)
 346 {
 347         struct private_data *state;
 348         int ret = ARCHIVE_OK;
 349
 350         state = (struct private_data *)self->data;
 351
 352         if (state->valid) {
 353                 switch (BZ2_bzDecompressEnd(&state->stream)) {
 354                 case BZ_OK:
 355                         break;
 356                 default:
 357                         archive_set_error(&self->archive->archive,
 358                                           ARCHIVE_ERRNO_MISC,
 359                                           "Failed to clean up decompressor");
 360                         ret = ARCHIVE_FATAL;
 361                 }
 362                 state->valid = 0;
 363         }
 364
 365         free(state->out_block);
 366         free(state);
 367         return (ret);
 368 }
 369
 370 #endif /* HAVE_BZLIB_H && BZ_CONFIG_ERROR */