contrib/libarchive-2/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 /* Don't compile this if we don't have zlib. */
  29 #if HAVE_ZLIB_H
  30
  31 __FBSDID("$FreeBSD: src/lib/libarchive/archive_write_set_compression_gzip.c,v 1.16 2008/02/21 03:21:50 kientzle Exp $");
  32
  33 #ifdef HAVE_ERRNO_H
  34 #include <errno.h>
  35 #endif
  36 #ifdef HAVE_STDLIB_H
  37 #include <stdlib.h>
  38 #endif
  39 #ifdef HAVE_STRING_H
  40 #include <string.h>
  41 #endif
  42 #include <time.h>
  43 #ifdef HAVE_ZLIB_H
  44 #include <zlib.h>
  45 #endif
  46
  47 #include "archive.h"
  48 #include "archive_private.h"
  49 #include "archive_write_private.h"
  50
  51 struct private_data {
  52         z_stream         stream;
  53         int64_t          total_in;
  54         unsigned char   *compressed;
  55         size_t           compressed_buffer_size;
  56         unsigned long    crc;
  57 };
  58
  59
  60 /*
  61  * Yuck.  zlib.h is not const-correct, so I need this one bit
  62  * of ugly hackery to convert a const * pointer to a non-const pointer.
  63  */
  64 #define SET_NEXT_IN(st,src)                                     \
  65         (st)->stream.next_in = (Bytef *)(uintptr_t)(const void *)(src)
  66
  67 static int      archive_compressor_gzip_finish(struct archive_write *);
  68 static int      archive_compressor_gzip_init(struct archive_write *);
  69 static int      archive_compressor_gzip_write(struct archive_write *,
  70                     const void *, size_t);
  71 static int      drive_compressor(struct archive_write *, struct private_data *,
  72                     int finishing);
  73
  74
  75 /*
  76  * Allocate, initialize and return a archive object.
  77  */
  78 int
  79 archive_write_set_compression_gzip(struct archive *_a)
  80 {
  81         struct archive_write *a = (struct archive_write *)_a;
  82         __archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC,
  83             ARCHIVE_STATE_NEW, "archive_write_set_compression_gzip");
  84         a->compressor.init = &archive_compressor_gzip_init;
  85         a->archive.compression_code = ARCHIVE_COMPRESSION_GZIP;
  86         a->archive.compression_name = "gzip";
  87         return (ARCHIVE_OK);
  88 }
  89
  90 /*
  91  * Setup callback.
  92  */
  93 static int
  94 archive_compressor_gzip_init(struct archive_write *a)
  95 {
  96         int ret;
  97         struct private_data *state;
  98         time_t t;
  99
 100         a->archive.compression_code = ARCHIVE_COMPRESSION_GZIP;
 101         a->archive.compression_name = "gzip";
 102
 103         if (a->client_opener != NULL) {
 104                 ret = (a->client_opener)(&a->archive, a->client_data);
 105                 if (ret != ARCHIVE_OK)
 106                         return (ret);
 107         }
 108
 109         /*
 110          * The next check is a temporary workaround until the gzip
 111          * code can be overhauled some.  The code should not require
 112          * that compressed_buffer_size == bytes_per_block.  Removing
 113          * this assumption will allow us to compress larger chunks at
 114          * a time, which should improve overall performance
 115          * marginally.  As a minor side-effect, such a cleanup would
 116          * allow us to support truly arbitrary block sizes.
 117          */
 118         if (a->bytes_per_block < 10) {
 119                 archive_set_error(&a->archive, EINVAL,
 120                     "GZip compressor requires a minimum 10 byte block size");
 121                 return (ARCHIVE_FATAL);
 122         }
 123
 124         state = (struct private_data *)malloc(sizeof(*state));
 125         if (state == NULL) {
 126                 archive_set_error(&a->archive, ENOMEM,
 127                     "Can't allocate data for compression");
 128                 return (ARCHIVE_FATAL);
 129         }
 130         memset(state, 0, sizeof(*state));
 131
 132         /*
 133          * See comment above.  We should set compressed_buffer_size to
 134          * max(bytes_per_block, 65536), but the code can't handle that yet.
 135          */
 136         state->compressed_buffer_size = a->bytes_per_block;
 137         state->compressed = (unsigned char *)malloc(state->compressed_buffer_size);
 138         state->crc = crc32(0L, NULL, 0);
 139
 140         if (state->compressed == NULL) {
 141                 archive_set_error(&a->archive, ENOMEM,
 142                     "Can't allocate data for compression buffer");
 143                 free(state);
 144                 return (ARCHIVE_FATAL);
 145         }
 146
 147         state->stream.next_out = state->compressed;
 148         state->stream.avail_out = state->compressed_buffer_size;
 149
 150         /* Prime output buffer with a gzip header. */
 151         t = time(NULL);
 152         state->compressed[0] = 0x1f; /* GZip signature bytes */
 153         state->compressed[1] = 0x8b;
 154         state->compressed[2] = 0x08; /* "Deflate" compression */
 155         state->compressed[3] = 0; /* No options */
 156         state->compressed[4] = (t)&0xff;  /* Timestamp */
 157         state->compressed[5] = (t>>8)&0xff;
 158         state->compressed[6] = (t>>16)&0xff;
 159         state->compressed[7] = (t>>24)&0xff;
 160         state->compressed[8] = 0; /* No deflate options */
 161         state->compressed[9] = 3; /* OS=Unix */
 162         state->stream.next_out += 10;
 163         state->stream.avail_out -= 10;
 164
 165         a->compressor.write = archive_compressor_gzip_write;
 166         a->compressor.finish = archive_compressor_gzip_finish;
 167
 168         /* Initialize compression library. */
 169         ret = deflateInit2(&(state->stream),
 170             Z_DEFAULT_COMPRESSION,
 171             Z_DEFLATED,
 172             -15 /* < 0 to suppress zlib header */,
 173             8,
 174             Z_DEFAULT_STRATEGY);
 175
 176         if (ret == Z_OK) {
 177                 a->compressor.data = state;
 178                 return (0);
 179         }
 180
 181         /* Library setup failed: clean up. */
 182         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Internal error "
 183             "initializing compression library");
 184         free(state->compressed);
 185         free(state);
 186
 187         /* Override the error message if we know what really went wrong. */
 188         switch (ret) {
 189         case Z_STREAM_ERROR:
 190                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 191                     "Internal error initializing "
 192                     "compression library: invalid setup parameter");
 193                 break;
 194         case Z_MEM_ERROR:
 195                 archive_set_error(&a->archive, ENOMEM, "Internal error initializing "
 196                     "compression library");
 197                 break;
 198         case Z_VERSION_ERROR:
 199                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 200                     "Internal error initializing "
 201                     "compression library: invalid library version");
 202                 break;
 203         }
 204
 205         return (ARCHIVE_FATAL);
 206 }
 207
 208 /*
 209  * Write data to the compressed stream.
 210  */
 211 static int
 212 archive_compressor_gzip_write(struct archive_write *a, const void *buff,
 213     size_t length)
 214 {
 215         struct private_data *state;
 216         int ret;
 217
 218         state = (struct private_data *)a->compressor.data;
 219         if (a->client_writer == NULL) {
 220                 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
 221                     "No write callback is registered?  "
 222                     "This is probably an internal programming error.");
 223                 return (ARCHIVE_FATAL);
 224         }
 225
 226         /* Update statistics */
 227         state->crc = crc32(state->crc, (const Bytef *)buff, length);
 228         state->total_in += length;
 229
 230         /* Compress input data to output buffer */
 231         SET_NEXT_IN(state, buff);
 232         state->stream.avail_in = length;
 233         if ((ret = drive_compressor(a, state, 0)) != ARCHIVE_OK)
 234                 return (ret);
 235
 236         a->archive.file_position += length;
 237         return (ARCHIVE_OK);
 238 }
 239
 240
 241 /*
 242  * Finish the compression...
 243  */
 244 static int
 245 archive_compressor_gzip_finish(struct archive_write *a)
 246 {
 247         ssize_t block_length, target_block_length, bytes_written;
 248         int ret;
 249         struct private_data *state;
 250         unsigned tocopy;
 251         unsigned char trailer[8];
 252
 253         state = (struct private_data *)a->compressor.data;
 254         ret = 0;
 255         if (a->client_writer == NULL) {
 256                 archive_set_error(&a->archive, ARCHIVE_ERRNO_PROGRAMMER,
 257                     "No write callback is registered?  "
 258                     "This is probably an internal programming error.");
 259                 ret = ARCHIVE_FATAL;
 260                 goto cleanup;
 261         }
 262
 263         /* By default, always pad the uncompressed data. */
 264         if (a->pad_uncompressed) {
 265                 tocopy = a->bytes_per_block -
 266                     (state->total_in % a->bytes_per_block);
 267                 while (tocopy > 0 && tocopy < (unsigned)a->bytes_per_block) {
 268                         SET_NEXT_IN(state, a->nulls);
 269                         state->stream.avail_in = tocopy < a->null_length ?
 270                             tocopy : a->null_length;
 271                         state->crc = crc32(state->crc, a->nulls,
 272                             state->stream.avail_in);
 273                         state->total_in += state->stream.avail_in;
 274                         tocopy -= state->stream.avail_in;
 275                         ret = drive_compressor(a, state, 0);
 276                         if (ret != ARCHIVE_OK)
 277                                 goto cleanup;
 278                 }
 279         }
 280
 281         /* Finish compression cycle */
 282         if (((ret = drive_compressor(a, state, 1))) != ARCHIVE_OK)
 283                 goto cleanup;
 284
 285         /* Build trailer: 4-byte CRC and 4-byte length. */
 286         trailer[0] = (state->crc)&0xff;
 287         trailer[1] = (state->crc >> 8)&0xff;
 288         trailer[2] = (state->crc >> 16)&0xff;
 289         trailer[3] = (state->crc >> 24)&0xff;
 290         trailer[4] = (state->total_in)&0xff;
 291         trailer[5] = (state->total_in >> 8)&0xff;
 292         trailer[6] = (state->total_in >> 16)&0xff;
 293         trailer[7] = (state->total_in >> 24)&0xff;
 294
 295         /* Add trailer to current block. */
 296         tocopy = 8;
 297         if (tocopy > state->stream.avail_out)
 298                 tocopy = state->stream.avail_out;
 299         memcpy(state->stream.next_out, trailer, tocopy);
 300         state->stream.next_out += tocopy;
 301         state->stream.avail_out -= tocopy;
 302
 303         /* If it overflowed, flush and start a new block. */
 304         if (tocopy < 8) {
 305                 bytes_written = (a->client_writer)(&a->archive, a->client_data,
 306                     state->compressed, state->compressed_buffer_size);
 307                 if (bytes_written <= 0) {
 308                         ret = ARCHIVE_FATAL;
 309                         goto cleanup;
 310                 }
 311                 a->archive.raw_position += bytes_written;
 312                 state->stream.next_out = state->compressed;
 313                 state->stream.avail_out = state->compressed_buffer_size;
 314                 memcpy(state->stream.next_out, trailer + tocopy, 8-tocopy);
 315                 state->stream.next_out += 8-tocopy;
 316                 state->stream.avail_out -= 8-tocopy;
 317         }
 318
 319         /* Optionally, pad the final compressed block. */
 320         block_length = state->stream.next_out - state->compressed;
 321
 322
 323         /* Tricky calculation to determine size of last block. */
 324         target_block_length = block_length;
 325         if (a->bytes_in_last_block <= 0)
 326                 /* Default or Zero: pad to full block */
 327                 target_block_length = a->bytes_per_block;
 328         else
 329                 /* Round length to next multiple of bytes_in_last_block. */
 330                 target_block_length = a->bytes_in_last_block *
 331                     ( (block_length + a->bytes_in_last_block - 1) /
 332                         a->bytes_in_last_block);
 333         if (target_block_length > a->bytes_per_block)
 334                 target_block_length = a->bytes_per_block;
 335         if (block_length < target_block_length) {
 336                 memset(state->stream.next_out, 0,
 337                     target_block_length - block_length);
 338                 block_length = target_block_length;
 339         }
 340
 341         /* Write the last block */
 342         bytes_written = (a->client_writer)(&a->archive, a->client_data,
 343             state->compressed, block_length);
 344         if (bytes_written <= 0) {
 345                 ret = ARCHIVE_FATAL;
 346                 goto cleanup;
 347         }
 348         a->archive.raw_position += bytes_written;
 349
 350         /* Cleanup: shut down compressor, release memory, etc. */
 351 cleanup:
 352         switch (deflateEnd(&(state->stream))) {
 353         case Z_OK:
 354                 break;
 355         default:
 356                 archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 357                     "Failed to clean up compressor");
 358                 ret = ARCHIVE_FATAL;
 359         }
 360         free(state->compressed);
 361         free(state);
 362         return (ret);
 363 }
 364
 365 /*
 366  * Utility function to push input data through compressor,
 367  * writing full output blocks as necessary.
 368  *
 369  * Note that this handles both the regular write case (finishing ==
 370  * false) and the end-of-archive case (finishing == true).
 371  */
 372 static int
 373 drive_compressor(struct archive_write *a, struct private_data *state, int finishing)
 374 {
 375         ssize_t bytes_written;
 376         int ret;
 377
 378         for (;;) {
 379                 if (state->stream.avail_out == 0) {
 380                         bytes_written = (a->client_writer)(&a->archive,
 381                             a->client_data, state->compressed,
 382                             state->compressed_buffer_size);
 383                         if (bytes_written <= 0) {
 384                                 /* TODO: Handle this write failure */
 385                                 return (ARCHIVE_FATAL);
 386                         } else if ((size_t)bytes_written < state->compressed_buffer_size) {
 387                                 /* Short write: Move remaining to
 388                                  * front of block and keep filling */
 389                                 memmove(state->compressed,
 390                                     state->compressed + bytes_written,
 391                                     state->compressed_buffer_size - bytes_written);
 392                         }
 393                         a->archive.raw_position += bytes_written;
 394                         state->stream.next_out
 395                             = state->compressed +
 396                             state->compressed_buffer_size - bytes_written;
 397                         state->stream.avail_out = bytes_written;
 398                 }
 399
 400                 /* If there's nothing to do, we're done. */
 401                 if (!finishing && state->stream.avail_in == 0)
 402                         return (ARCHIVE_OK);
 403
 404                 ret = deflate(&(state->stream),
 405                     finishing ? Z_FINISH : Z_NO_FLUSH );
 406
 407                 switch (ret) {
 408                 case Z_OK:
 409                         /* In non-finishing case, check if compressor
 410                          * consumed everything */
 411                         if (!finishing && state->stream.avail_in == 0)
 412                                 return (ARCHIVE_OK);
 413                         /* In finishing case, this return always means
 414                          * there's more work */
 415                         break;
 416                 case Z_STREAM_END:
 417                         /* This return can only occur in finishing case. */
 418                         return (ARCHIVE_OK);
 419                 default:
 420                         /* Any other return value indicates an error. */
 421                         archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC,
 422                             "GZip compression failed:"
 423                             " deflate() call returned status %d",
 424                             ret);
 425                         return (ARCHIVE_FATAL);
 426                 }
 427         }
 428 }
 429
 430 #endif /* HAVE_ZLIB_H */