wimlib.net Git - wimlib/blob - src/xpress-compress.c

   1 /*
   2  * xpress-compress.c
   3  *
   4  * XPRESS compression routines.
   5  *
   6  * See the comments in xpress-decompress.c about the XPRESS format.
   7  */
   8
   9 /*
  10  * Copyright (C) 2012, 2013 Eric Biggers
  11  *
  12  * This file is part of wimlib, a library for working with WIM files.
  13  *
  14  * wimlib is free software; you can redistribute it and/or modify it under the
  15  * terms of the GNU General Public License as published by the Free
  16  * Software Foundation; either version 3 of the License, or (at your option)
  17  * any later version.
  18  *
  19  * wimlib is distributed in the hope that it will be useful, but WITHOUT ANY
  20  * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
  21  * A PARTICULAR PURPOSE. See the GNU General Public License for more
  22  * details.
  23  *
  24  * You should have received a copy of the GNU General Public License
  25  * along with wimlib; if not, see http://www.gnu.org/licenses/.
  26  */
  27
  28 #ifdef HAVE_CONFIG_H
  29 #  include "config.h"
  30 #endif
  31
  32 #include "wimlib.h"
  33 #include "wimlib/assert.h"
  34 #include "wimlib/compress.h"
  35 #include "wimlib/error.h"
  36 #include "wimlib/util.h"
  37 #include "wimlib/xpress.h"
  38
  39 #ifdef HAVE_ALLOCA_H
  40 #  include <alloca.h>
  41 #endif
  42
  43 #include <string.h>
  44
  45 /* Intermediate XPRESS match/literal representation.  */
  46 struct xpress_match {
  47         u16 adjusted_len;  /* Match length minus XPRESS_MIN_MATCH_LEN */
  48         u16 offset;        /* Match offset */
  49         /* For literals, offset == 0 and adjusted_len is the literal.  */
  50 };
  51
  52 /*
  53  * Writes @match, which is a match given in the intermediate representation for
  54  * XPRESS matches, to the output stream @ostream.
  55  *
  56  * @codewords and @lens provide the Huffman code that is being used.
  57  */
  58 static void
  59 xpress_write_match(struct xpress_match match,
  60                    struct output_bitstream *restrict ostream,
  61                    const u16 codewords[restrict],
  62                    const u8 lens[restrict])
  63 {
  64         u8 len_hdr = min(match.adjusted_len, 0xf);
  65         u8 offset_bsr = bsr32(match.offset);
  66         unsigned sym = XPRESS_NUM_CHARS + ((offset_bsr << 4) | len_hdr);
  67
  68         bitstream_put_bits(ostream, codewords[sym], lens[sym]);
  69
  70         if (match.adjusted_len >= 0xf) {
  71                 u8 byte1 = min(match.adjusted_len - 0xf, 0xff);
  72                 bitstream_put_byte(ostream, byte1);
  73                 if (byte1 == 0xff) {
  74                         bitstream_put_byte(ostream, match.adjusted_len & 0xff);
  75                         bitstream_put_byte(ostream, match.adjusted_len >> 8);
  76                 }
  77         }
  78         bitstream_put_bits(ostream, match.offset ^ (1U << offset_bsr), offset_bsr);
  79 }
  80
  81 static void
  82 xpress_write_matches_and_literals(struct output_bitstream *ostream,
  83                                   const struct xpress_match matches[restrict],
  84                                   input_idx_t num_matches,
  85                                   const u16 codewords[restrict],
  86                                   const u8 lens[restrict])
  87 {
  88         for (input_idx_t i = 0; i < num_matches; i++) {
  89                 if (matches[i].offset) {
  90                         /* Real match  */
  91                         xpress_write_match(matches[i], ostream, codewords, lens);
  92                 } else {
  93                         /* Literal byte  */
  94                         u8 lit = matches[i].adjusted_len;
  95                         bitstream_put_bits(ostream, codewords[lit], lens[lit]);
  96                 }
  97         }
  98         bitstream_put_bits(ostream, codewords[XPRESS_END_OF_DATA], lens[XPRESS_END_OF_DATA]);
  99 }
 100
 101 struct xpress_record_ctx {
 102         input_idx_t freqs[XPRESS_NUM_SYMBOLS];
 103         struct xpress_match *matches;
 104 };
 105
 106 static void
 107 xpress_record_literal(u8 lit, void *_ctx)
 108 {
 109         struct xpress_record_ctx *ctx = _ctx;
 110         ctx->freqs[lit]++;
 111         *(ctx->matches++) = (struct xpress_match) { .offset = 0, .adjusted_len = lit };
 112 }
 113
 114 static void
 115 xpress_record_match(unsigned len, unsigned offset, void *_ctx)
 116 {
 117         struct xpress_record_ctx *ctx = _ctx;
 118
 119         XPRESS_ASSERT(len >= XPRESS_MIN_MATCH_LEN);
 120         XPRESS_ASSERT(len <= XPRESS_MAX_MATCH_LEN);
 121         XPRESS_ASSERT(offset >= XPRESS_MIN_OFFSET);
 122         XPRESS_ASSERT(offset <= XPRESS_MAX_OFFSET);
 123
 124         unsigned adjusted_len = len - XPRESS_MIN_MATCH_LEN;
 125         unsigned len_hdr = min(adjusted_len, 0xf);
 126         unsigned sym = XPRESS_NUM_CHARS + ((bsr32(offset) << 4) | len_hdr);
 127
 128         XPRESS_ASSERT(sym >= XPRESS_NUM_CHARS);
 129         XPRESS_ASSERT(sym < XPRESS_NUM_SYMBOLS);
 130
 131         ctx->freqs[sym]++;
 132         *(ctx->matches++) = (struct xpress_match) { .offset = offset,
 133                                                     .adjusted_len = adjusted_len };
 134 }
 135
 136 static const struct lz_params xpress_lz_params = {
 137         .min_match      = XPRESS_MIN_MATCH_LEN,
 138         .max_match      = XPRESS_MAX_MATCH_LEN,
 139         .max_offset     = XPRESS_MAX_OFFSET,
 140         .good_match     = 16,
 141         .nice_match     = 32,
 142         .max_chain_len  = 16,
 143         .max_lazy_match = 16,
 144         .too_far        = 4096,
 145 };
 146
 147 /* API function documented in wimlib.h  */
 148 WIMLIBAPI unsigned
 149 wimlib_xpress_compress(const void * restrict uncompressed_data,
 150                        unsigned uncompressed_len,
 151                        void * restrict compressed_data)
 152 {
 153         u8 *cptr = compressed_data;
 154         struct output_bitstream ostream;
 155
 156         struct xpress_record_ctx record_ctx;
 157
 158         struct xpress_match *matches;
 159         input_idx_t *prev_tab;
 160         u8 *udata;
 161
 162         u16 codewords[XPRESS_NUM_SYMBOLS];
 163         u8 lens[XPRESS_NUM_SYMBOLS];
 164         input_idx_t num_matches;
 165         input_idx_t compressed_len;
 166         input_idx_t i;
 167         const size_t stack_max = 65536;
 168
 169         /* XPRESS requires 256 bytes of overhead for the Huffman code, so it's
 170          * impossible to compress 256 bytes or less of data to less than the
 171          * input size.
 172          *
 173          * +1 to take into account that the buffer for compressed data is 1 byte
 174          * smaller than the buffer for uncompressed data.
 175          *
 176          * +4 to take into account that init_output_bitstream() requires at
 177          * least 4 bytes of data.  */
 178         if (uncompressed_len < XPRESS_NUM_SYMBOLS / 2 + 1 + 4)
 179                 return 0;
 180
 181         if (uncompressed_len <= stack_max) {
 182                 matches = alloca(uncompressed_len * sizeof(matches[0]));
 183                 udata = alloca(uncompressed_len + 8);
 184                 prev_tab = alloca(uncompressed_len * sizeof(prev_tab[0]));
 185         } else {
 186                 matches = MALLOC(uncompressed_len * sizeof(matches[0]));
 187                 udata = MALLOC(uncompressed_len + 8);
 188                 prev_tab = MALLOC(uncompressed_len * sizeof(prev_tab[0]));
 189                 if (matches == NULL || udata == NULL || prev_tab == NULL) {
 190                         WARNING("Failed to allocate memory for compression...");
 191                         compressed_len = 0;
 192                         goto out_free;
 193                 }
 194         }
 195
 196         /* Copy the data to a temporary buffer, but only to avoid
 197          * inconsequential accesses of uninitialized memory in
 198          * lz_analyze_block().  */
 199         memcpy(udata, uncompressed_data, uncompressed_len);
 200         memset(udata + uncompressed_len, 0, 8);
 201
 202         /* Determine match/literal sequence to divide the data into.  */
 203         ZERO_ARRAY(record_ctx.freqs);
 204         record_ctx.matches = matches;
 205         lz_analyze_block(udata,
 206                          uncompressed_len,
 207                          xpress_record_match,
 208                          xpress_record_literal,
 209                          &record_ctx,
 210                          &xpress_lz_params,
 211                          prev_tab);
 212
 213         num_matches = (record_ctx.matches - matches);
 214
 215         /* Account for end of data symbol.  */
 216         record_ctx.freqs[XPRESS_END_OF_DATA]++;
 217
 218         /* Build the Huffman code.  */
 219         make_canonical_huffman_code(XPRESS_NUM_SYMBOLS, XPRESS_MAX_CODEWORD_LEN,
 220                                     record_ctx.freqs, lens, codewords);
 221
 222         /* Output the Huffman code as a series of 512 4-bit lengths.  */
 223         for (i = 0; i < XPRESS_NUM_SYMBOLS; i += 2)
 224                 *cptr++ = (lens[i] & 0xf) | (lens[i + 1] << 4);
 225
 226         /* Output the encoded matches/literals.  */
 227         init_output_bitstream(&ostream, cptr,
 228                               uncompressed_len - XPRESS_NUM_SYMBOLS / 2 - 1);
 229         xpress_write_matches_and_literals(&ostream, matches,
 230                                           num_matches, codewords, lens);
 231
 232         /* Flush any pending data and get the length of the compressed data.  */
 233         compressed_len = flush_output_bitstream(&ostream);
 234         if (compressed_len == ~(input_idx_t)0) {
 235                 compressed_len = 0;
 236                 goto out_free;
 237         }
 238         compressed_len += XPRESS_NUM_SYMBOLS / 2;
 239
 240 #if defined(ENABLE_XPRESS_DEBUG) || defined(ENABLE_VERIFY_COMPRESSION)
 241         /* Verify that we really get the same thing back when decompressing.  */
 242         if (wimlib_xpress_decompress(compressed_data, compressed_len,
 243                                      udata, uncompressed_len))
 244         {
 245                 ERROR("Failed to decompress data we "
 246                       "compressed using XPRESS algorithm");
 247                 wimlib_assert(0);
 248                 compressed_len = 0;
 249                 goto out_free;
 250         }
 251
 252         if (memcmp(uncompressed_data, udata, uncompressed_len)) {
 253                 ERROR("Data we compressed using XPRESS algorithm "
 254                       "didn't decompress to original");
 255                 wimlib_assert(0);
 256                 compressed_len = 0;
 257                 goto out_free;
 258         }
 259 #endif
 260
 261 out_free:
 262         if (uncompressed_len > stack_max) {
 263                 FREE(matches);
 264                 FREE(udata);
 265                 FREE(prev_tab);
 266         }
 267         return compressed_len;
 268 }