lib/libarchive/archive_write_set_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("$FreeBSD$");
  29
  30 #ifdef HAVE_ERRNO_H
  31 #include <errno.h>
  32 #endif
  33 #ifdef HAVE_STDLIB_H
  34 #include <stdlib.h>
  35 #endif
  36 #ifdef HAVE_STRING_H
  37 #include <string.h>
  38 #endif
  39 #include <time.h>
  40 #ifdef HAVE_ZLIB_H
  41 #include <zlib.h>
  42 #endif
  43
  44 #include "archive.h"
  45 #include "archive_private.h"
  46 #include "archive_write_private.h"
  47
  48 #ifndef HAVE_ZLIB_H
  49 int
  50 archive_write_set_compression_gzip(struct archive *_a)
  51 {
  52         /* Unsupported gzip compression, we don't have zlib */
  53         return (ARCHIVE_FATAL);
  54 }
  55 #else
  56 /* Don't compile this if we don't have zlib. */
  57
  58 struct private_data {
  59         z_stream         stream;
  60         int64_t          total_in;
  61         unsigned char   *compressed;
  62         size_t           compressed_buffer_size;
  63         unsigned long    crc;
  64 };
  65
  66
  67 /*
  68  * Yuck.  zlib.h is not const-correct, so I need this one bit
  69  * of ugly hackery to convert a const * pointer to a non-const pointer.
  70  */
  71 #define SET_NEXT_IN(st,src)                                     \
  72         (st)->stream.next_in = (Bytef *)(uintptr_t)(const void *)(src)
  73
  74 static int      archive_compressor_gzip_finish(struct archive_write *);
  75 static int      archive_compressor_gzip_init(struct archive_write *);
  76 static int      archive_compressor_gzip_write(struct archive_write *,
  77                     const void *, size_t);
  78 static int      drive_compressor(struct archive_write *, struct private_data *,
  79                     int finishing);
  80
  81
  82 /*
  83  * Allocate, initialize and return a archive object.
  84  */
  85 int
  86 archive_write_set_compression_gzip(struct archive *_a)
  87 {
  88         struct archive_write *a = (struct archive_write *)_a;
  89         __archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC,
  90             ARCHIVE_STATE_NEW, "archive_write_set_compression_gzip");
  91         a->compressor.init = &archive_compressor_gzip_init;
  92         a->archive.compression_code = ARCHIVE_COMPRESSION_GZIP;
  93         a->archive.compression_name = "gzip";
  94         return (ARCHIVE_OK);
  95 }
  96
  97 /*
  98  * Setup callback.
  99  */
 100 static int
 101 archive_compressor_gzip_init(struct archive_write *a)
 102 {
 103         int ret;
 104         struct private_data *state;
 105         time_t t;
 106
 107         a->archive.compression_code = ARCHIVE_COMPRESSION_GZIP;
 108         a->archive.compression_name = "gzip";
 109
 110         if (a->client_opener != NULL) {
 111                 ret = (a->client_opener)(&a->archive, a->client_data);
 112                 if (ret != ARCHIVE_OK)
 113                         return (ret);
 114         }
 115
 116         /*
 117          * The next check is a temporary workaround until the gzip
 118          * code can be overhauled some.  The code should not require
 119          * that compressed_buffer_size == bytes_per_block.  Removing
 120          * this assumption will allow us to compress larger chunks at
 121          * a time, which should improve overall performance
 122          * marginally.  As a minor side-effect, such a cleanup would
 123          * allow us to support truly arbitrary block sizes.
 124          */
 125         if (a->bytes_per_block < 10) {
 126                 archive_set_error(&a->archive, EINVAL,
 127                     "GZip compressor requires a minimum 10 byte block size");
 128                 return (ARCHIVE_FATAL);
 129         }
 130
 131         state = (struct private_data *)malloc(sizeof(*state));
 132         if (state == NULL) {
 133                 archive_set_error(&a->archive, ENOMEM,
 134                     "Can't allocate data for compression");
 135                 return (ARCHIVE_FATAL);
 136         }
 137         memset(state, 0, sizeof(*state));
 138
 139         /*
 140          * See comment above.  We should set compressed_buffer_size to
 141          * max(bytes_per_block, 65536), but the code can't handle that yet.
 142          */
 143         state->compressed_buffer_size = a->bytes_per_block;
 144         state->compressed = (unsigned char *)malloc(state->compressed_buffer_size);
 145         state->crc = crc32(0L, NULL, 0);
 146
 147         if (state->compressed == NULL) {
 148                 archive_set_error(&a->archive, ENOMEM,
 149                     "Can't allocate data for compression buffer");
 150                 free(state);
 151                 return (ARCHIVE_FATAL);
 152         }
 153
 154         state->stream.next_out = state->compressed;
 155         state->stream.avail_out = state->compressed_buffer_size;
 156
 157         /* Prime output buffer with a gzip header. */
 158         t = time(NULL);
 159         state->compressed[0] = 0x1f; /* GZip signature bytes */
 160         state->compressed[1] = 0x8b;
 161         state->compressed[2] = 0x08; /* "Deflate" compression */
 162         state->compressed[3] = 0; /* No options */
 163         state->compressed[4] = (t)&0xff;  /* Timestamp */
 164         state->compressed[5] = (t>>8)&0xff;
 165         state->compressed[6] = (t>>16)&0xff;
 166         state->compressed[7] = (t>>24)&0xff;
 167         state->compressed[8] = 0; /* No deflate options */
 168         state->compressed[9] = 3; /* OS=Unix */
 169         state->stream.next_out += 10;
 170         state->stream.avail_out -= 10;
 171
 172         a->compressor.write = archive_compressor_gzip_write;
 173         a->compressor.finish = archive_compressor_gzip_finish;
 174
 175         /* Initialize compression library. */
 176         ret = deflateInit2(&(state->stream),
 177             Z_DEFAULT_COMPRESSION,
 178             Z_DEFLATED,
 179             -15 /* < 0 to suppress zlib header */,
 180             8,
 181             Z_DEFAULT_STRATEGY);
 182
 183         if (ret == Z_OK) {
 184                 a->compressor.data = state;
 185                 return (0);
 186         }
 187
 188         /* Library setup failed: clean up. */
 189         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Internal error "
 190             "initializing compression library");
 191         free(state->compressed);
 192         free(state);
 193
 194         /* Override the error message if we know what really went wrong. */
 195         switch (ret) {
 196         case Z_STREAM_ERROR:
 197                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 198                     "Internal error initializing "
 199                     "compression library: invalid setup parameter");
 200                 break;
 201         case Z_MEM_ERROR:
 202                 archive_set_error(&a->archive, ENOMEM, "Internal error initializing "
 203                     "compression library");
 204                 break;
 205         case Z_VERSION_ERROR:
 206                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 207                     "Internal error initializing "
 208                     "compression library: invalid library version");
 209                 break;
 210         }
 211
 212         return (ARCHIVE_FATAL);
 213 }
 214
 215 /*
 216  * Write data to the compressed stream.
 217  */
 218 static int
 219 archive_compressor_gzip_write(struct archive_write *a, const void *buff,
 220     size_t length)
 221 {
 222         struct private_data *state;
 223         int ret;
 224
 225         state = (struct private_data *)a->compressor.data;
 226         if (a->client_writer == NULL) {
 227                 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
 228                     "No write callback is registered?  "
 229                     "This is probably an internal programming error.");
 230                 return (ARCHIVE_FATAL);
 231         }
 232
 233         /* Update statistics */
 234         state->crc = crc32(state->crc, (const Bytef *)buff, length);
 235         state->total_in += length;
 236
 237         /* Compress input data to output buffer */
 238         SET_NEXT_IN(state, buff);
 239         state->stream.avail_in = length;
 240         if ((ret = drive_compressor(a, state, 0)) != ARCHIVE_OK)
 241                 return (ret);
 242
 243         a->archive.file_position += length;
 244         return (ARCHIVE_OK);
 245 }
 246
 247
 248 /*
 249  * Finish the compression...
 250  */
 251 static int
 252 archive_compressor_gzip_finish(struct archive_write *a)
 253 {
 254         ssize_t block_length, target_block_length, bytes_written;
 255         int ret;
 256         struct private_data *state;
 257         unsigned tocopy;
 258         unsigned char trailer[8];
 259
 260         state = (struct private_data *)a->compressor.data;
 261         ret = 0;
 262         if (a->client_writer == NULL) {
 263                 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
 264                     "No write callback is registered?  "
 265                     "This is probably an internal programming error.");
 266                 ret = ARCHIVE_FATAL;
 267                 goto cleanup;
 268         }
 269
 270         /* By default, always pad the uncompressed data. */
 271         if (a->pad_uncompressed) {
 272                 tocopy = a->bytes_per_block -
 273                     (state->total_in % a->bytes_per_block);
 274                 while (tocopy > 0 && tocopy < (unsigned)a->bytes_per_block) {
 275                         SET_NEXT_IN(state, a->nulls);
 276                         state->stream.avail_in = tocopy < a->null_length ?
 277                             tocopy : a->null_length;
 278                         state->crc = crc32(state->crc, a->nulls,
 279                             state->stream.avail_in);
 280                         state->total_in += state->stream.avail_in;
 281                         tocopy -= state->stream.avail_in;
 282                         ret = drive_compressor(a, state, 0);
 283                         if (ret != ARCHIVE_OK)
 284                                 goto cleanup;
 285                 }
 286         }
 287
 288         /* Finish compression cycle */
 289         if (((ret = drive_compressor(a, state, 1))) != ARCHIVE_OK)
 290                 goto cleanup;
 291
 292         /* Build trailer: 4-byte CRC and 4-byte length. */
 293         trailer[0] = (state->crc)&0xff;
 294         trailer[1] = (state->crc >> 8)&0xff;
 295         trailer[2] = (state->crc >> 16)&0xff;
 296         trailer[3] = (state->crc >> 24)&0xff;
 297         trailer[4] = (state->total_in)&0xff;
 298         trailer[5] = (state->total_in >> 8)&0xff;
 299         trailer[6] = (state->total_in >> 16)&0xff;
 300         trailer[7] = (state->total_in >> 24)&0xff;
 301
 302         /* Add trailer to current block. */
 303         tocopy = 8;
 304         if (tocopy > state->stream.avail_out)
 305                 tocopy = state->stream.avail_out;
 306         memcpy(state->stream.next_out, trailer, tocopy);
 307         state->stream.next_out += tocopy;
 308         state->stream.avail_out -= tocopy;
 309
 310         /* If it overflowed, flush and start a new block. */
 311         if (tocopy < 8) {
 312                 bytes_written = (a->client_writer)(&a->archive, a->client_data,
 313                     state->compressed, state->compressed_buffer_size);
 314                 if (bytes_written <= 0) {
 315                         ret = ARCHIVE_FATAL;
 316                         goto cleanup;
 317                 }
 318                 a->archive.raw_position += bytes_written;
 319                 state->stream.next_out = state->compressed;
 320                 state->stream.avail_out = state->compressed_buffer_size;
 321                 memcpy(state->stream.next_out, trailer + tocopy, 8-tocopy);
 322                 state->stream.next_out += 8-tocopy;
 323                 state->stream.avail_out -= 8-tocopy;
 324         }
 325
 326         /* Optionally, pad the final compressed block. */
 327         block_length = state->stream.next_out - state->compressed;
 328
 329
 330         /* Tricky calculation to determine size of last block. */
 331         target_block_length = block_length;
 332         if (a->bytes_in_last_block <= 0)
 333                 /* Default or Zero: pad to full block */
 334                 target_block_length = a->bytes_per_block;
 335         else
 336                 /* Round length to next multiple of bytes_in_last_block. */
 337                 target_block_length = a->bytes_in_last_block *
 338                     ( (block_length + a->bytes_in_last_block - 1) /
 339                         a->bytes_in_last_block);
 340         if (target_block_length > a->bytes_per_block)
 341                 target_block_length = a->bytes_per_block;
 342         if (block_length < target_block_length) {
 343                 memset(state->stream.next_out, 0,
 344                     target_block_length - block_length);
 345                 block_length = target_block_length;
 346         }
 347
 348         /* Write the last block */
 349         bytes_written = (a->client_writer)(&a->archive, a->client_data,
 350             state->compressed, block_length);
 351         if (bytes_written <= 0) {
 352                 ret = ARCHIVE_FATAL;
 353                 goto cleanup;
 354         }
 355         a->archive.raw_position += bytes_written;
 356
 357         /* Cleanup: shut down compressor, release memory, etc. */
 358 cleanup:
 359         switch (deflateEnd(&(state->stream))) {
 360         case Z_OK:
 361                 break;
 362         default:
 363                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 364                     "Failed to clean up compressor");
 365                 ret = ARCHIVE_FATAL;
 366         }
 367         free(state->compressed);
 368         free(state);
 369         return (ret);
 370 }
 371
 372 /*
 373  * Utility function to push input data through compressor,
 374  * writing full output blocks as necessary.
 375  *
 376  * Note that this handles both the regular write case (finishing ==
 377  * false) and the end-of-archive case (finishing == true).
 378  */
 379 static int
 380 drive_compressor(struct archive_write *a, struct private_data *state, int finishing)
 381 {
 382         ssize_t bytes_written;
 383         int ret;
 384
 385         for (;;) {
 386                 if (state->stream.avail_out == 0) {
 387                         bytes_written = (a->client_writer)(&a->archive,
 388                             a->client_data, state->compressed,
 389                             state->compressed_buffer_size);
 390                         if (bytes_written <= 0) {
 391                                 /* TODO: Handle this write failure */
 392                                 return (ARCHIVE_FATAL);
 393                         } else if ((size_t)bytes_written < state->compressed_buffer_size) {
 394                                 /* Short write: Move remaining to
 395                                  * front of block and keep filling */
 396                                 memmove(state->compressed,
 397                                     state->compressed + bytes_written,
 398                                     state->compressed_buffer_size - bytes_written);
 399                         }
 400                         a->archive.raw_position += bytes_written;
 401                         state->stream.next_out
 402                             = state->compressed +
 403                             state->compressed_buffer_size - bytes_written;
 404                         state->stream.avail_out = bytes_written;
 405                 }
 406
 407                 /* If there's nothing to do, we're done. */
 408                 if (!finishing && state->stream.avail_in == 0)
 409                         return (ARCHIVE_OK);
 410
 411                 ret = deflate(&(state->stream),
 412                     finishing ? Z_FINISH : Z_NO_FLUSH );
 413
 414                 switch (ret) {
 415                 case Z_OK:
 416                         /* In non-finishing case, check if compressor
 417                          * consumed everything */
 418                         if (!finishing && state->stream.avail_in == 0)
 419                                 return (ARCHIVE_OK);
 420                         /* In finishing case, this return always means
 421                          * there's more work */
 422                         break;
 423                 case Z_STREAM_END:
 424                         /* This return can only occur in finishing case. */
 425                         return (ARCHIVE_OK);
 426                 default:
 427                         /* Any other return value indicates an error. */
 428                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 429                             "GZip compression failed:"
 430                             " deflate() call returned status %d",
 431                             ret);
 432                         return (ARCHIVE_FATAL);
 433                 }
 434         }
 435 }
 436
 437 #endif /* HAVE_ZLIB_H */