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
  74 #if ARCHIVE_VERSION_NUMBER < 4000000
  75 /* Deprecated; remove in libarchive 4.0 */
  76 int
  77 archive_read_support_compression_bzip2(struct archive *a)
  78 {
  79         return archive_read_support_filter_bzip2(a);
  80 }
  81 #endif
  82
  83 static const struct archive_read_filter_bidder_vtable
  84 bzip2_bidder_vtable = {
  85         .bid = bzip2_reader_bid,
  86         .init = bzip2_reader_init,
  87 };
  88
  89 int
  90 archive_read_support_filter_bzip2(struct archive *_a)
  91 {
  92         struct archive_read *a = (struct archive_read *)_a;
  93
  94         if (__archive_read_register_bidder(a, NULL, "bzip2",
  95                                 &bzip2_bidder_vtable) != ARCHIVE_OK)
  96                 return (ARCHIVE_FATAL);
  97
  98 #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR)
  99         return (ARCHIVE_OK);
 100 #else
 101         archive_set_error(_a, ARCHIVE_ERRNO_MISC,
 102             "Using external bzip2 program");
 103         return (ARCHIVE_WARN);
 104 #endif
 105 }
 106
 107 /*
 108  * Test whether we can handle this data.
 109  *
 110  * This logic returns zero if any part of the signature fails.  It
 111  * also tries to Do The Right Thing if a very short buffer prevents us
 112  * from verifying as much as we would like.
 113  */
 114 static int
 115 bzip2_reader_bid(struct archive_read_filter_bidder *self, struct archive_read_filter *filter)
 116 {
 117         const unsigned char *buffer;
 118         ssize_t avail;
 119         int bits_checked;
 120
 121         (void)self; /* UNUSED */
 122
 123         /* Minimal bzip2 archive is 14 bytes. */
 124         buffer = __archive_read_filter_ahead(filter, 14, &avail);
 125         if (buffer == NULL)
 126                 return (0);
 127
 128         /* First three bytes must be "BZh" */
 129         bits_checked = 0;
 130         if (memcmp(buffer, "BZh", 3) != 0)
 131                 return (0);
 132         bits_checked += 24;
 133
 134         /* Next follows a compression flag which must be an ASCII digit. */
 135         if (buffer[3] < '1' || buffer[3] > '9')
 136                 return (0);
 137         bits_checked += 5;
 138
 139         /* After BZh[1-9], there must be either a data block
 140          * which begins with 0x314159265359 or an end-of-data
 141          * marker of 0x177245385090. */
 142         if (memcmp(buffer + 4, "\x31\x41\x59\x26\x53\x59", 6) == 0)
 143                 bits_checked += 48;
 144         else if (memcmp(buffer + 4, "\x17\x72\x45\x38\x50\x90", 6) == 0)
 145                 bits_checked += 48;
 146         else
 147                 return (0);
 148
 149         return (bits_checked);
 150 }
 151
 152 #if !defined(HAVE_BZLIB_H) || !defined(BZ_CONFIG_ERROR)
 153
 154 /*
 155  * If we don't have the library on this system, we can't actually do the
 156  * decompression.  We can, however, still detect compressed archives
 157  * and emit a useful message.
 158  */
 159 static int
 160 bzip2_reader_init(struct archive_read_filter *self)
 161 {
 162         int r;
 163
 164         r = __archive_read_program(self, "bzip2 -d");
 165         /* Note: We set the format here even if __archive_read_program()
 166          * above fails.  We do, after all, know what the format is
 167          * even if we weren't able to read it. */
 168         self->code = ARCHIVE_FILTER_BZIP2;
 169         self->name = "bzip2";
 170         return (r);
 171 }
 172
 173
 174 #else
 175
 176 static const struct archive_read_filter_vtable
 177 bzip2_reader_vtable = {
 178         .read = bzip2_filter_read,
 179         .close = bzip2_filter_close,
 180 };
 181
 182 /*
 183  * Setup the callbacks.
 184  */
 185 static int
 186 bzip2_reader_init(struct archive_read_filter *self)
 187 {
 188         static const size_t out_block_size = 64 * 1024;
 189         void *out_block;
 190         struct private_data *state;
 191
 192         self->code = ARCHIVE_FILTER_BZIP2;
 193         self->name = "bzip2";
 194
 195         state = (struct private_data *)calloc(sizeof(*state), 1);
 196         out_block = (unsigned char *)malloc(out_block_size);
 197         if (state == NULL || out_block == NULL) {
 198                 archive_set_error(&self->archive->archive, ENOMEM,
 199                     "Can't allocate data for bzip2 decompression");
 200                 free(out_block);
 201                 free(state);
 202                 return (ARCHIVE_FATAL);
 203         }
 204
 205         self->data = state;
 206         state->out_block_size = out_block_size;
 207         state->out_block = out_block;
 208         self->vtable = &bzip2_reader_vtable;
 209
 210         return (ARCHIVE_OK);
 211 }
 212
 213 /*
 214  * Return the next block of decompressed data.
 215  */
 216 static ssize_t
 217 bzip2_filter_read(struct archive_read_filter *self, const void **p)
 218 {
 219         struct private_data *state;
 220         size_t decompressed;
 221         const char *read_buf;
 222         ssize_t ret;
 223
 224         state = (struct private_data *)self->data;
 225
 226         if (state->eof) {
 227                 *p = NULL;
 228                 return (0);
 229         }
 230
 231         /* Empty our output buffer. */
 232         state->stream.next_out = state->out_block;
 233         state->stream.avail_out = state->out_block_size;
 234
 235         /* Try to fill the output buffer. */
 236         for (;;) {
 237                 if (!state->valid) {
 238                         if (bzip2_reader_bid(self->bidder, self->upstream) == 0) {
 239                                 state->eof = 1;
 240                                 *p = state->out_block;
 241                                 decompressed = state->stream.next_out
 242                                     - state->out_block;
 243                                 return (decompressed);
 244                         }
 245                         /* Initialize compression library. */
 246                         ret = BZ2_bzDecompressInit(&(state->stream),
 247                                            0 /* library verbosity */,
 248                                            0 /* don't use low-mem algorithm */);
 249
 250                         /* If init fails, try low-memory algorithm instead. */
 251                         if (ret == BZ_MEM_ERROR)
 252                                 ret = BZ2_bzDecompressInit(&(state->stream),
 253                                            0 /* library verbosity */,
 254                                            1 /* do use low-mem algo */);
 255
 256                         if (ret != BZ_OK) {
 257                                 const char *detail = NULL;
 258                                 int err = ARCHIVE_ERRNO_MISC;
 259                                 switch (ret) {
 260                                 case BZ_PARAM_ERROR:
 261                                         detail = "invalid setup parameter";
 262                                         break;
 263                                 case BZ_MEM_ERROR:
 264                                         err = ENOMEM;
 265                                         detail = "out of memory";
 266                                         break;
 267                                 case BZ_CONFIG_ERROR:
 268                                         detail = "mis-compiled library";
 269                                         break;
 270                                 }
 271                                 archive_set_error(&self->archive->archive, err,
 272                                     "Internal error initializing decompressor%s%s",
 273                                     detail == NULL ? "" : ": ",
 274                                     detail);
 275                                 return (ARCHIVE_FATAL);
 276                         }
 277                         state->valid = 1;
 278                 }
 279
 280                 /* stream.next_in is really const, but bzlib
 281                  * doesn't declare it so. <sigh> */
 282                 read_buf =
 283                     __archive_read_filter_ahead(self->upstream, 1, &ret);
 284                 if (read_buf == NULL) {
 285                         archive_set_error(&self->archive->archive,
 286                             ARCHIVE_ERRNO_MISC,
 287                             "truncated bzip2 input");
 288                         return (ARCHIVE_FATAL);
 289                 }
 290                 state->stream.next_in = (char *)(uintptr_t)read_buf;
 291                 state->stream.avail_in = ret;
 292                 /* There is no more data, return whatever we have. */
 293                 if (ret == 0) {
 294                         state->eof = 1;
 295                         *p = state->out_block;
 296                         decompressed = state->stream.next_out
 297                             - state->out_block;
 298                         return (decompressed);
 299                 }
 300
 301                 /* Decompress as much as we can in one pass. */
 302                 ret = BZ2_bzDecompress(&(state->stream));
 303                 __archive_read_filter_consume(self->upstream,
 304                     state->stream.next_in - read_buf);
 305
 306                 switch (ret) {
 307                 case BZ_STREAM_END: /* Found end of stream. */
 308                         switch (BZ2_bzDecompressEnd(&(state->stream))) {
 309                         case BZ_OK:
 310                                 break;
 311                         default:
 312                                 archive_set_error(&(self->archive->archive),
 313                                           ARCHIVE_ERRNO_MISC,
 314                                           "Failed to clean up decompressor");
 315                                 return (ARCHIVE_FATAL);
 316                         }
 317                         state->valid = 0;
 318                         /* FALLTHROUGH */
 319                 case BZ_OK: /* Decompressor made some progress. */
 320                         /* If we filled our buffer, update stats and return. */
 321                         if (state->stream.avail_out == 0) {
 322                                 *p = state->out_block;
 323                                 decompressed = state->stream.next_out
 324                                     - state->out_block;
 325                                 return (decompressed);
 326                         }
 327                         break;
 328                 default: /* Return an error. */
 329                         archive_set_error(&self->archive->archive,
 330                             ARCHIVE_ERRNO_MISC, "bzip decompression failed");
 331                         return (ARCHIVE_FATAL);
 332                 }
 333         }
 334 }
 335
 336 /*
 337  * Clean up the decompressor.
 338  */
 339 static int
 340 bzip2_filter_close(struct archive_read_filter *self)
 341 {
 342         struct private_data *state;
 343         int ret = ARCHIVE_OK;
 344
 345         state = (struct private_data *)self->data;
 346
 347         if (state->valid) {
 348                 switch (BZ2_bzDecompressEnd(&state->stream)) {
 349                 case BZ_OK:
 350                         break;
 351                 default:
 352                         archive_set_error(&self->archive->archive,
 353                                           ARCHIVE_ERRNO_MISC,
 354                                           "Failed to clean up decompressor");
 355                         ret = ARCHIVE_FATAL;
 356                 }
 357                 state->valid = 0;
 358         }
 359
 360         free(state->out_block);
 361         free(state);
 362         return (ret);
 363 }
 364
 365 #endif /* HAVE_BZLIB_H && BZ_CONFIG_ERROR */