# Filters #
###########
-m4_define([SUPPORTED_FILTERS], [lzma,lzma2,subblock,delta,x86,powerpc,ia64,arm,armthumb,sparc])dnl
+m4_define([SUPPORTED_FILTERS], [lzma1,lzma2,subblock,delta,x86,powerpc,ia64,arm,armthumb,sparc])dnl
m4_define([SIMPLE_FILTERS], [x86,powerpc,ia64,arm,armthumb,sparc])
-m4_define([LZ_FILTERS], [lzma,lzma2])
+m4_define([LZ_FILTERS], [lzma1,lzma2])
m4_foreach([NAME], [SUPPORTED_FILTERS],
[enable_filter_[]NAME=no
esac
done
- # LZMA2 requires that LZMA is enabled.
- test "x$enable_encoder_lzma2" = xyes && enable_encoder_lzma=yes
- test "x$enable_decoder_lzma2" = xyes && enable_decoder_lzma=yes
+ # LZMA2 requires that LZMA1 is enabled.
+ test "x$enable_encoder_lzma2" = xyes && enable_encoder_lzma1=yes
+ test "x$enable_decoder_lzma2" = xyes && enable_decoder_lzma1=yes
AC_MSG_RESULT([$enable_decoders])
fi
-if test "x$enable_encoder_lzma2$enable_encoder_lzma" = xyesno \
- || test "x$enable_decoder_lzma2$enable_decoder_lzma" = xyesno; then
- AC_MSG_ERROR([LZMA2 requires that LZMA is also enabled.])
+if test "x$enable_encoder_lzma2$enable_encoder_lzma1" = xyesno \
+ || test "x$enable_decoder_lzma2$enable_decoder_lzma1" = xyesno; then
+ AC_MSG_ERROR([LZMA2 requires that LZMA1 is also enabled.])
fi
m4_foreach([NAME], [SUPPORTED_FILTERS],
AC_ARG_ENABLE(match-finders, AC_HELP_STRING([--enable-match-finders=LIST],
[Comma-separated list of match finders to build. Default=all.
At least one match finder is required for encoding with
- the LZMA filter. Available match finders:]
+ the LZMA1 and LZMA2 filters. Available match finders:]
m4_translit(m4_defn([SUPPORTED_MATCH_FINDERS]), [,], [ ])), [],
[enable_match_finders=SUPPORTED_MATCH_FINDERS])
enable_match_finders=`echo "$enable_match_finders" | sed 's/,/ /g'`
// Config
+ lzma_options_lzma opt_lzma;
+ if (lzma_lzma_preset(&opt_lzma, 0)) {
+ fprintf(stderr, "preset failed\n");
+ exit(1);
+ }
lzma_filter filters[LZMA_BLOCK_FILTERS_MAX + 1];
filters[0].id = LZMA_FILTER_LZMA2;
- filters[0].options = (void *)&lzma_preset_lzma[0];
+ filters[0].options = &opt_lzma;
filters[1].id = LZMA_VLI_UNKNOWN;
// Init
const size_t in_size = fread(in, 1, BUFFER_SIZE, stdin);
// Filter setup
+ lzma_options_lzma opt_lzma;
+ if (lzma_lzma_preset(&opt_lzma, 0))
+ return 1;
+
lzma_filter filters[] = {
{
.id = LZMA_FILTER_LZMA2,
- .options = (void *)(&lzma_preset_lzma[0])
+ .options = &opt_lzma
},
{
.id = LZMA_VLI_UNKNOWN
lzma_init();
lzma_options_lzma lzma = {
- .dictionary_size = (1 << 27) + (1 << 26),
- .literal_context_bits = 3,
- .literal_pos_bits = 0,
- .pos_bits = 2,
- .preset_dictionary = NULL,
- .preset_dictionary_size = 0,
+ .dict_size = (1U << 27) + (1U << 26),
+ .lc = 3,
+ .lp = 0,
+ .pb = 2,
+ .preset_dict = NULL,
+ .preset_dict_size = 0,
.mode = LZMA_MODE_NORMAL,
- .fast_bytes = 48,
- .match_finder = LZMA_MF_BT4,
- .match_finder_cycles = 0,
+ .nice_len = 48,
+ .mf = LZMA_MF_BT4,
+ .depth = 0,
};
/*
lzma_options_filter filters[] = {
- { LZMA_FILTER_LZMA,
+ { LZMA_FILTER_LZMA1,
(lzma_options_lzma *)&lzma_preset_lzma[6 - 1] },
{ UINT64_MAX, NULL }
};
*/
lzma_filter filters[] = {
- { LZMA_FILTER_LZMA, &lzma },
+ { LZMA_FILTER_LZMA1, &lzma },
{ UINT64_MAX, NULL }
};
// Config
lzma_options_lzma opt_lzma = {
- .dictionary_size = 1 << 16,
- .literal_context_bits = LZMA_LITERAL_CONTEXT_BITS_DEFAULT,
- .literal_pos_bits = LZMA_LITERAL_POS_BITS_DEFAULT,
- .pos_bits = LZMA_POS_BITS_DEFAULT,
- .preset_dictionary = NULL,
+ .dict_size = 1U << 16,
+ .lc = LZMA_LC_DEFAULT,
+ .lp = LZMA_LP_DEFAULT,
+ .pb = LZMA_PB_DEFAULT,
+ .preset_dict = NULL,
.persistent = true,
.mode = LZMA_MODE_NORMAL,
- .fast_bytes = 32,
- .match_finder = LZMA_MF_HC3,
- .match_finder_cycles = 0,
+ .nice_len = 32,
+ .mf = LZMA_MF_HC3,
+ .depth = 0,
};
lzma_options_delta opt_delta = {
.rle = 1, // LZMA_SUBBLOCK_RLE_OFF,
.subfilter_mode = LZMA_SUBFILTER_SET,
};
- opt_subblock.subfilter_options.id = LZMA_FILTER_LZMA;
+ opt_subblock.subfilter_options.id = LZMA_FILTER_LZMA1;
opt_subblock.subfilter_options.options = &opt_lzma;
opt_subblock.subfilter_options.id = LZMA_FILTER_DELTA;
opt_subblock.subfilter_options.options = &opt_delta;
liblzma_la_LIBADD += lz/liblz.la
endif
-if COND_FILTER_LZMA
+if COND_FILTER_LZMA1
SUBDIRS += lzma rangecoder
liblzma_la_LIBADD += \
lzma/liblzma2.la \
* - 16-bit stereo audio: distance = 4 bytes
* - 24-bit RGB image data: distance = 3 bytes
*/
- uint32_t distance;
-# define LZMA_DELTA_DISTANCE_MIN 1
-# define LZMA_DELTA_DISTANCE_MAX 256
+ uint32_t dist;
+# define LZMA_DELTA_DIST_MIN 1
+# define LZMA_DELTA_DIST_MAX 256
/**
* \brief Reserved space for possible future extensions
*/
uint32_t reserved_int1;
uint32_t reserved_int2;
+ uint32_t reserved_int3;
+ uint32_t reserved_int4;
void *reserved_ptr1;
void *reserved_ptr2;
/**
* \file lzma/lzma.h
- * \brief LZMA filter
+ * \brief LZMA1 and LZMA2 filters
*
* \author Copyright (C) 1999-2006 Igor Pavlov
* \author Copyright (C) 2007 Lasse Collin
/**
- * \brief Filter ID
+ * \brief LZMA1 Filter ID
*
- * Filter ID of the LZMA filter. This is used as lzma_filter.id.
+ * LZMA1 is the very same thing as what was called just LZMA in earlier
+ * LZMA Utils, 7-Zip, and LZMA SDK. It's called LZMA1 here to prevent
+ * developers from accidentally using LZMA when they actually want LZMA2.
*/
-#define LZMA_FILTER_LZMA LZMA_VLI_C(0x20)
+#define LZMA_FILTER_LZMA1 LZMA_VLI_C(0x4000000000000001)
+/**
+ * \brief LZMA2 Filter ID
+ *
+ * Usually you want this instead of LZMA1. Compared to LZMA1, LZMA2 adds
+ * support for LZMA_SYNC_FLUSH, uncompressed chunks (expands uncompressible
+ * data less), possibility to change lc/lp/pb in the middle of encoding, and
+ * some other internal improvements.
+ */
#define LZMA_FILTER_LZMA2 LZMA_VLI_C(0x21)
*
* Match finder has major effect on both speed and compression ratio.
* Usually hash chains are faster than binary trees.
+ *
+ * The memory usage formulas are only rough estimates, which are closest to
+ * reality when dict_size is a power of two. The formulas are more complex
+ * in reality, and can also change a little between liblzma versions. Use
+ * lzma_memusage_encoder() to get more accurate estimate of memory usage.
*/
typedef enum {
LZMA_MF_HC3 = 0x03,
/**<
- * \brief Hash Chain with 3 bytes hashing
+ * \brief Hash Chain with 2- and 3-byte hashing
*
- * \todo Memory requirements
+ * Minimum nice_len: 3
*
- * \note It's possible that this match finder gets
- * removed in future. The definition will stay
- * in this header, but liblzma may return
- * LZMA_OPTIONS_ERROR if it is specified (just
- * like it would if the match finder had been
- * disabled at compile time).
+ * Memory usage:
+ * - dict_size <= 16 MiB: dict_size * 7.5
+ * - dict_size > 16 MiB: dict_size * 5.5 + 64 MiB
*/
LZMA_MF_HC4 = 0x04,
/**<
- * \brief Hash Chain with 4 bytes hashing
+ * \brief Hash Chain with 2-, 3-, and 4-byte hashing
*
- * Memory requirements: 7.5 * dictionary_size + 4 MiB
+ * Minimum nice_len: 4
*
- * \note It's possible that this match finder gets
- * removed in future. The definition will stay
- * in this header, but liblzma may return
- * LZMA_OPTIONS_ERROR if it is specified (just
- * like it would if the match finder had been
- * disabled at compile time).
+ * Memory usage: dict_size * 7.5
*/
LZMA_MF_BT2 = 0x12,
/**<
- * \brief Binary Tree with 2 bytes hashing
+ * \brief Binary Tree with 2-byte hashing
+ *
+ * Minimum nice_len: 2
*
- * Memory requirements: 9.5 * dictionary_size + 4 MiB
+ * Memory usage: dict_size * 9.5
*/
LZMA_MF_BT3 = 0x13,
/**<
- * \brief Binary Tree with 3 bytes hashing
+ * \brief Binary Tree with 2- and 3-byte hashing
+ *
+ * Minimum nice_len: 3
*
- * Memory requirements: 11.5 * dictionary_size + 4 MiB
+ * Memory usage:
+ * - dict_size <= 16 MiB: dict_size * 11.5
+ * - dict_size > 16 MiB: dict_size * 9.5 + 64 MiB
*/
LZMA_MF_BT4 = 0x14
/**<
- * \brief Binary Tree with 4 bytes hashing
+ * \brief Binary Tree with 2-, 3-, and 4-byte hashing
*
- * Memory requirements: 11.5 * dictionary_size + 4 MiB
+ * Minimum nice_len: 4
+ *
+ * Memory usage: dict_size * 11.5
*/
} lzma_match_finder;
* finder.
*/
typedef enum {
- LZMA_MODE_FAST = 0,
+ LZMA_MODE_FAST = 1,
/**<
* \brief Fast compression
*
* a hash chain match finder.
*/
- LZMA_MODE_NORMAL = 1
+ LZMA_MODE_NORMAL = 2
/**<
* \brief Normal compression
*
/**
- * \brief Options specific to the LZMA method handler
+ * \brief Options specific to the LZMA1 and LZMA2 filters
*/
typedef struct {
/**********************************
* indicate what data to repeat from the dictionary buffer. Thus,
* the bigger the dictionary, the better compression ratio usually is.
*
- * Raw decoding: Too big dictionary does no other harm than
- * wasting memory. This value is ignored by lzma_raw_decode_buffer(),
- * because it uses the target buffer as the dictionary.
+ * Maximum size of the dictionary depends on multiple things:
+ * - Memory usage limit
+ * - Available address space (not a problem on 64-bit systems)
+ * - Selected match finder (encoder only)
+ *
+ * Currently the maximum dictionary size for encoding is 1.5 GiB
+ * (i.e. (UINT32_C(1) << 30) + (UINT32_C(1) << 29)) even on 64-bit
+ * systems for certain match finder implementation reasons. In future,
+ * there may be match finders that support bigger dictionaries (3 GiB
+ * will probably be the maximum).
+ *
+ * Decoder already supports dictionaries up to 4 GiB - 1 B (i.e.
+ * UINT32_MAX), so increasing the maximum dictionary size of the
+ * encoder won't cause problems for old decoders.
+ *
+ * Because extremely small dictionaries sizes would have unneeded
+ * overhead in the decoder, the minimum dictionary size is 4096 bytes.
+ *
+ * \note When decoding, too big dictionary does no other harm
+ * than wasting memory.
*/
- uint32_t dictionary_size;
-# define LZMA_DICTIONARY_SIZE_MIN (UINT32_C(1) << 12)
-# define LZMA_DICTIONARY_SIZE_MAX (UINT32_C(1) << 30)
-# define LZMA_DICTIONARY_SIZE_DEFAULT (UINT32_C(1) << 23)
+ uint32_t dict_size;
+# define LZMA_DICT_SIZE_MIN UINT32_C(4096)
+# define LZMA_DICT_SIZE_DEFAULT (UINT32_C(1) << 23)
/**
* \brief Pointer to an initial dictionary
*
* \todo This feature is not implemented yet.
*/
- const uint8_t *preset_dictionary;
+ const uint8_t *preset_dict;
/**
* \brief Size of the preset dictionary
*
* Specifies the size of the preset dictionary. If the size is
- * bigger than dictionary_size, only the last dictionary_size
- * bytes are processed.
+ * bigger than dict_size, only the last dict_size bytes are processed.
*
- * This variable is read only when preset_dictionary is not NULL.
+ * This variable is read only when preset_dict is not NULL.
*/
- uint32_t preset_dictionary_size;
+ uint32_t preset_dict_size;
/**
* \brief Number of literal context bits
* account when predicting the bits of the next literal.
*
* \todo Example
+ *
+ * There is a limit that applies to literal context bits and literal
+ * position bits together: lc + lp <= 4. Without this limit the
+ * decoding could become very slow, which could have security related
+ * results in some cases like email servers doing virus scanning.
+ * This limit also simplifies the internal implementation in liblzma.
+ *
+ * There may be LZMA streams that have lc + lp > 4 (maximum lc
+ * possible would be 8). It is not possible to decode such streams
+ * with liblzma.
*/
- uint32_t literal_context_bits;
-# define LZMA_LITERAL_CONTEXT_BITS_MIN 0
-# define LZMA_LITERAL_CONTEXT_BITS_MAX 4
-# define LZMA_LITERAL_CONTEXT_BITS_DEFAULT 3
+ uint32_t lc;
+# define LZMA_LCLP_MIN 0
+# define LZMA_LCLP_MAX 4
+# define LZMA_LC_DEFAULT 3
/**
* \brief Number of literal position bits
*
* \todo Example
*/
- uint32_t literal_pos_bits;
-# define LZMA_LITERAL_POS_BITS_MIN 0
-# define LZMA_LITERAL_POS_BITS_MAX 4
-# define LZMA_LITERAL_POS_BITS_DEFAULT 0
+ uint32_t lp;
+# define LZMA_LP_DEFAULT 0
/**
* \brief Number of position bits
* which a matching sequence is found from the dictionary and
* thus can be stored as distance-length pair.
*
- * Example: If most of the matches occur at byte positions
- * of 8 * n + 3, that is, 3, 11, 19, ... set pos_bits to 3,
- * because 2**3 == 8.
+ * Example: If most of the matches occur at byte positions of
+ * 8 * n + 3, that is, 3, 11, 19, ... set pb to 3, because 2**3 == 8.
*/
- uint32_t pos_bits;
-# define LZMA_POS_BITS_MIN 0
-# define LZMA_POS_BITS_MAX 4
-# define LZMA_POS_BITS_DEFAULT 2
+ uint32_t pb;
+# define LZMA_PB_MIN 0
+# define LZMA_PB_MAX 4
+# define LZMA_PB_DEFAULT 2
/******************************************
* LZMA options needed only when encoding *
* in the middle of the encoding process without resetting the encoder.
*
* This option is used only by LZMA2. LZMA1 ignores this and it is
- * safeto not initialize this when encoding with LZMA1.
+ * safe to not initialize this when encoding with LZMA1.
*/
lzma_bool persistent;
lzma_mode mode;
/**
- * \brief Number of fast bytes
+ * \brief Nice length of a match
+ *
+ * This determines how many bytes the encoder compares from the match
+ * candidates when looking for the best match. Once a match of at
+ * least nice_len bytes long is found, the encoder stops looking for
+ * better condidates and encodes the match. (Naturally, if the found
+ * match is actually longer than nice_len, the actual length is
+ * encoded; it's not truncated to nice_len.)
*
- * Number of fast bytes determines how many bytes the encoder
- * compares from the match candidates when looking for the best
- * match. Bigger fast bytes value usually increase both compression
- * ratio and time.
+ * Bigger values usually increase the compression ratio and
+ * compression time. For most files, 30 to 100 is a good value,
+ * which gives very good compression ratio at good speed.
+ *
+ * The exact minimum value depends on the match finder. The maximum is
+ * 273, which is the maximum length of a match that LZMA can encode.
*/
- uint32_t fast_bytes;
-# define LZMA_FAST_BYTES_MIN 5
-# define LZMA_FAST_BYTES_MAX 273
-# define LZMA_FAST_BYTES_DEFAULT 128
+ uint32_t nice_len;
/** Match finder ID */
- lzma_match_finder match_finder;
+ lzma_match_finder mf;
/**
- * \brief Match finder cycles
+ * \brief Maximum search depth in the match finder
+ *
+ * For every input byte, match finder searches through the hash chain
+ * or binary tree in a loop, each iteration going one step deeper in
+ * the chain or tree. The searching stops if
+ * - a match of at least nice_len bytes long is found;
+ * - all match candidates from the hash chain or binary tree have
+ * been checked; or
+ * - maximum search depth is reached.
+ *
+ * Maximum search depth is needed to prevent the match finder from
+ * wasting too much time in case there are lots of short match
+ * candidates. On the other hand, stopping the search before all
+ * candidates have been checked can reduce compression ratio.
*
- * Higher values give slightly better compression ratio but
- * decrease speed. Use special value 0 to let liblzma use
- * match-finder-dependent default value.
+ * Setting depth to zero tells liblzma to use an automatic default
+ * value, that depends on the selected match finder and nice_len.
+ * The default is in the range [10, 200] or so (it may vary between
+ * liblzma versions).
*
- * \todo Write much better description.
+ * Using a bigger depth value than the default can increase
+ * compression ratio in some cases. There is no strict maximum value,
+ * but high values (thousands or millions) should be used with care:
+ * the encoder could remain fast enough with typical input, but
+ * malicious input could cause the match finder to slow down
+ * dramatically, possibly creating a denial of service attack.
*/
- uint32_t match_finder_cycles;
+ uint32_t depth;
/**
* \brief Reserved space for possible future extensions
uint32_t reserved_int2;
uint32_t reserved_int3;
uint32_t reserved_int4;
+ uint32_t reserved_int5;
+ uint32_t reserved_int6;
+ uint32_t reserved_int7;
+ uint32_t reserved_int8;
void *reserved_ptr1;
void *reserved_ptr2;
/**
- * \brief Maximum sum of literal_context_bits and literal_pos_bits
- *
- * literal_context_bits + literal_pos_bits <= LZMA_LITERAL_BITS_MAX
- */
-#define LZMA_LITERAL_BITS_MAX 4
-
-
-/**
- * \brief Table of presets for the LZMA filter
+ * \brief Set a compression level preset to lzma_options_lzma structure
*
- * lzma_preset_lzma[0] is the fastest and lzma_preset_lzma[8] is the slowest.
- * These presets match the switches -1 .. -9 of the lzma command line tool
+ * level = 0 is the fastest and level = 8 is the slowest. These presets match
+ * the switches -1 .. -9 of the command line tool.
*
* The preset values are subject to changes between liblzma versions.
*
- * This variable is available only if LZMA encoder has been enabled.
+ * This function is available only if LZMA encoder has been enabled.
*/
-extern const lzma_options_lzma lzma_preset_lzma[9];
+extern lzma_bool lzma_lzma_preset(lzma_options_lzma *options, uint32_t level);
case LZMA_FILTER_IA64:
return 16;
- case LZMA_FILTER_LZMA: {
+ case LZMA_FILTER_LZMA1: {
const lzma_options_lzma *lzma = filters[i].options;
- return 1 << MAX(lzma->pos_bits,
- lzma->literal_pos_bits);
+ return 1 << MAX(lzma->pb, lzma->lp);
}
default:
filters[i].options))->alignment;
case LZMA_FILTER_X86:
- case LZMA_FILTER_LZMA:
+ case LZMA_FILTER_LZMA1:
return 1;
case LZMA_FILTER_ARMTHUMB:
break;
case SEQ_DICTIONARY_SIZE:
- coder->options.dictionary_size
+ coder->options.dict_size
|= (size_t)(in[*in_pos]) << (coder->pos * 8);
if (++coder->pos == 4) {
- if (coder->options.dictionary_size
- > LZMA_DICTIONARY_SIZE_MAX)
+ if (coder->options.dict_size > (UINT32_C(1) << 30))
return LZMA_FORMAT_ERROR;
// A hack to ditch tons of false positives: We allow
// LZMA_Alone created only files with 2^n, but accepts
// any dictionary size. If someone complains, this
// will be reconsidered.
- uint32_t d = coder->options.dictionary_size - 1;
+ uint32_t d = coder->options.dict_size - 1;
d |= d >> 2;
d |= d >> 3;
d |= d >> 4;
d |= d >> 16;
++d;
- if (d != coder->options.dictionary_size)
+ if (d != coder->options.dict_size)
return LZMA_FORMAT_ERROR;
coder->pos = 0;
next->coder->sequence = SEQ_PROPERTIES;
next->coder->pos = 0;
- next->coder->options.dictionary_size = 0;
+ next->coder->options.dict_size = 0;
next->coder->uncompressed_size = 0;
next->coder->memlimit = memlimit;
if (lzma_lzma_lclppb_encode(options, next->coder->header))
return LZMA_PROG_ERROR;
- // - Dictionary size (4 bytes)
- if (options->dictionary_size < LZMA_DICTIONARY_SIZE_MIN
- || options->dictionary_size > LZMA_DICTIONARY_SIZE_MAX)
+ // - Dictionary size (4 bytes); limit to 1 GiB since that's what
+ // LZMA SDK currently does for encoding.
+ if (options->dict_size < LZMA_DICT_SIZE_MIN
+ || options->dict_size > (UINT32_C(1) << 30))
return LZMA_PROG_ERROR;
// Round up to to the next 2^n or 2^n + 2^(n - 1) depending on which
//
// FIXME Maybe LZMA_Alone needs some lower limit for maximum
// dictionary size? Must check decoders from old LZMA SDK version.
- uint32_t d = options->dictionary_size - 1;
+ uint32_t d = options->dict_size - 1;
d |= d >> 2;
d |= d >> 3;
d |= d >> 4;
// splitting the data in smaller blocks.
break;
- case LZMA_FILTER_LZMA:
+ case LZMA_FILTER_LZMA1:
// The block sizes of the possible next filters in
// the chain are irrelevant after the LZMA filter.
return ((lzma_options_lzma *)(filters->options))
struct lzma_coder_s {
lzma_next_coder stream_encoder;
+ /// Options for LZMA2
+ lzma_options_lzma opt_lzma;
+
/// We need to keep the filters array available in case
/// LZMA_FULL_FLUSH is used.
lzma_filter filters[5];
static bool
-easy_set_filters(lzma_filter *filters, uint32_t level)
+easy_set_filters(lzma_coder *coder, uint32_t level)
{
bool error = false;
#ifdef HAVE_ENCODER_LZMA2
} else if (level <= 9) {
- filters[0].id = LZMA_FILTER_LZMA2;
- filters[0].options = (void *)(&lzma_preset_lzma[level - 1]);
- filters[1].id = LZMA_VLI_UNKNOWN;
+ error = lzma_lzma_preset(&coder->opt_lzma, level - 1);
+ coder->filters[0].id = LZMA_FILTER_LZMA2;
+ coder->filters[0].options = &coder->opt_lzma;
+ coder->filters[1].id = LZMA_VLI_UNKNOWN;
#endif
} else {
next->coder->stream_encoder = LZMA_NEXT_CODER_INIT;
}
- if (easy_set_filters(next->coder->filters, level))
+ if (easy_set_filters(next->coder, level))
return LZMA_OPTIONS_ERROR;
return lzma_stream_encoder_init(&next->coder->stream_encoder,
extern LZMA_API uint64_t
lzma_easy_memory_usage(lzma_easy_level level)
{
- lzma_filter filters[5];
- if (easy_set_filters(filters, level))
+ lzma_coder coder;
+ if (easy_set_filters(&coder, level))
return UINT32_MAX;
- return lzma_memusage_encoder(filters);
+ return lzma_memusage_encoder(coder.filters);
}
bool changes_size;
} features[] = {
-#if defined (HAVE_ENCODER_LZMA) || defined(HAVE_DECODER_LZMA)
+#if defined (HAVE_ENCODER_LZMA1) || defined(HAVE_DECODER_LZMA1)
{
- .id = LZMA_FILTER_LZMA,
+ .id = LZMA_FILTER_LZMA1,
.non_last_ok = false,
.last_ok = true,
.changes_size = true,
static const lzma_filter_decoder decoders[] = {
-#ifdef HAVE_DECODER_LZMA
+#ifdef HAVE_DECODER_LZMA1
{
- .id = LZMA_FILTER_LZMA,
+ .id = LZMA_FILTER_LZMA1,
.init = &lzma_lzma_decoder_init,
.memusage = &lzma_lzma_decoder_memusage,
.props_decode = &lzma_lzma_props_decode,
static const lzma_filter_encoder encoders[] = {
-#ifdef HAVE_ENCODER_LZMA
+#ifdef HAVE_ENCODER_LZMA1
{
- .id = LZMA_FILTER_LZMA,
+ .id = LZMA_FILTER_LZMA1,
.init = &lzma_lzma_encoder_init,
.memusage = &lzma_lzma_encoder_memusage,
.chunk_size = NULL, // FIXME
lzma_init_check();
-#if defined(HAVE_SMALL) && defined(HAVE_ENCODER_LZMA)
+#if defined(HAVE_SMALL) && defined(HAVE_ENCODER_LZMA1)
lzma_rc_init();
#endif
// Set the delta distance.
if (filters[0].options == NULL)
return LZMA_PROG_ERROR;
- next->coder->distance = ((lzma_options_delta *)(filters[0].options))
- ->distance;
- if (next->coder->distance < LZMA_DELTA_DISTANCE_MIN
- || next->coder->distance > LZMA_DELTA_DISTANCE_MAX)
+ next->coder->distance
+ = ((lzma_options_delta *)(filters[0].options))->dist;
+ if (next->coder->distance < LZMA_DELTA_DIST_MIN
+ || next->coder->distance > LZMA_DELTA_DIST_MAX)
return LZMA_OPTIONS_ERROR;
// Initialize the rest of the variables.
next->coder->pos = 0;
- memzero(next->coder->history, LZMA_DELTA_DISTANCE_MAX);
+ memzero(next->coder->history, LZMA_DELTA_DIST_MAX);
// Initialize the next decoder in the chain, if any.
return lzma_next_filter_init(&next->coder->next,
- allocator, filters + 1);
+ allocator, filters + 1);
}
uint8_t pos;
/// Buffer to hold history of the original data
- uint8_t history[LZMA_DELTA_DISTANCE_MAX];
+ uint8_t history[LZMA_DELTA_DIST_MAX];
};
return LZMA_MEM_ERROR;
opt->type = LZMA_DELTA_TYPE_BYTE;
- opt->distance = props[0] + 1;
+ opt->dist = props[0] + 1;
*options = opt;
// It's possible that newer liblzma versions will support larger
// distance values.
if (opt->type != LZMA_DELTA_TYPE_BYTE
- || opt->distance < LZMA_DELTA_DISTANCE_MIN
- || opt->distance > LZMA_DELTA_DISTANCE_MAX)
+ || opt->dist < LZMA_DELTA_DIST_MIN
+ || opt->dist > LZMA_DELTA_DIST_MAX)
return LZMA_OPTIONS_ERROR;
- out[0] = opt->distance - LZMA_DELTA_DISTANCE_MIN;
+ out[0] = opt->dist - LZMA_DELTA_DIST_MIN;
return LZMA_OK;
}
// For now, the dictionary size is limited to 1.5 GiB. This may grow
// in the future if needed, but it needs a little more work than just
// changing this check.
- if (lz_options->dictionary_size < LZMA_DICTIONARY_SIZE_MIN
- || lz_options->dictionary_size
+ if (lz_options->dict_size < LZMA_DICT_SIZE_MIN
+ || lz_options->dict_size
> (UINT32_C(1) << 30) + (UINT32_C(1) << 29)
- || lz_options->find_len_max
- > lz_options->match_len_max)
+ || lz_options->nice_len > lz_options->match_len_max)
return true;
- mf->keep_size_before = lz_options->before_size
- + lz_options->dictionary_size;
+ mf->keep_size_before = lz_options->before_size + lz_options->dict_size;
mf->keep_size_after = lz_options->after_size
+ lz_options->match_len_max;
// to size_t.
// - Memory usage calculation needs something too, e.g. use uint64_t
// for mf->size.
- uint32_t reserve = lz_options->dictionary_size / 2;
+ uint32_t reserve = lz_options->dict_size / 2;
if (reserve > (UINT32_C(1) << 30))
reserve /= 2;
// Match finder options
mf->match_len_max = lz_options->match_len_max;
- mf->find_len_max = lz_options->find_len_max;
+ mf->nice_len = lz_options->nice_len;
// cyclic_size has to stay smaller than 2 Gi. Note that this doesn't
// mean limitting dictionary size to less than 2 GiB. With a match
// memory to keep the code simpler. The current way is simple and
// still allows pretty big dictionaries, so I don't expect these
// limits to change.
- mf->cyclic_size = lz_options->dictionary_size + 1;
+ mf->cyclic_size = lz_options->dict_size + 1;
// Validate the match finder ID and setup the function pointers.
switch (lz_options->match_finder) {
}
// Calculate the sizes of mf->hash and mf->son and check that
- // find_len_max is big enough for the selected match finder.
+ // nice_len is big enough for the selected match finder.
const uint32_t hash_bytes = lz_options->match_finder & 0x0F;
- if (hash_bytes > mf->find_len_max)
+ if (hash_bytes > mf->nice_len)
return true;
const bool is_bt = (lz_options->match_finder & 0x10) != 0;
} else {
// Round dictionary size up to the next 2^n - 1 so it can
// be used as a hash mask.
- hs = lz_options->dictionary_size - 1;
+ hs = lz_options->dict_size - 1;
hs |= hs >> 1;
hs |= hs >> 2;
hs |= hs >> 4;
}
// Maximum number of match finder cycles
- mf->loops = lz_options->match_finder_cycles;
- if (mf->loops == 0) {
- mf->loops = 16 + (mf->find_len_max / 2);
+ mf->depth = lz_options->depth;
+ if (mf->depth == 0) {
+ mf->depth = 16 + (mf->nice_len / 2);
if (!is_bt)
- mf->loops /= 2;
+ mf->depth /= 2;
}
return false;
uint32_t hash_mask;
/// Maximum number of loops in the match finder
- uint32_t loops;
+ uint32_t depth;
/// Maximum length of a match that the match finder will try to find.
- uint32_t find_len_max;
+ uint32_t nice_len;
/// Maximum length of a match supported by the LZ-based encoder.
- /// If the longest match found by the match finder is find_len_max,
+ /// If the longest match found by the match finder is nice_len,
/// mf_find() tries to expand it up to match_len_max bytes.
uint32_t match_len_max;
size_t before_size;
/// Size of the history buffer
- size_t dictionary_size;
+ size_t dict_size;
/// Extra amount of data to keep available after the "actual"
/// dictionary.
size_t after_size;
/// Maximum length of a match that the LZ-based encoder can accept.
- /// This is used to extend matches of length find_len_max to the
+ /// This is used to extend matches of length nice_len to the
/// maximum possible length.
size_t match_len_max;
/// Match finder will search matches of at maximum of this length.
/// This must be less than or equal to match_len_max.
- size_t find_len_max;
+ size_t nice_len;
/// Type of the match finder to use
lzma_match_finder match_finder;
- /// TODO: Comment
- uint32_t match_finder_cycles;
+ /// Maximum search depth
+ uint32_t depth;
/// TODO: Comment
- const uint8_t *preset_dictionary;
+ const uint8_t *preset_dict;
- uint32_t preset_dictionary_size;
+ uint32_t preset_dict_size;
} lzma_lz_options;
// The total usable buffer space at any moment outside the match finder:
-// before_size + dictionary_size + after_size + match_len_max
+// before_size + dict_size + after_size + match_len_max
//
// In reality, there's some extra space allocated to prevent the number of
-// memmove() calls reasonable. The bigger the dictionary_size is, the bigger
+// memmove() calls reasonable. The bigger the dict_size is, the bigger
// this extra buffer will be since with bigger dictionaries memmove() would
// also take longer.
//
// In other words, a single encoder loop may advance lzma_mf.read_pos at
// maximum of after_size times. Since matches are looked up to
// lzma_mf.buffer[lzma_mf.read_pos + match_len_max - 1], the total
-// amount of extra buffer needed after dictionary_size becomes
+// amount of extra buffer needed after dict_size becomes
// after_size + match_len_max.
//
// before_size has two uses. The first one is to keep literals available
#ifndef NDEBUG
// Validate the matches.
for (uint32_t i = 0; i < count; ++i) {
- assert(matches[i].len <= mf->find_len_max);
+ assert(matches[i].len <= mf->nice_len);
assert(matches[i].dist < mf->read_pos);
assert(memcmp(mf_ptr(mf) - 1,
mf_ptr(mf) - matches[i].dist - 2,
// If a match of maximum search length was found, try to
// extend the match to maximum possible length.
- if (len_best == mf->find_len_max) {
+ if (len_best == mf->nice_len) {
// The limit for the match length is either the
// maximum match length supported by the LZ-based
// encoder or the number of bytes left in the
/// Hash value to indicate unused element in the hash. Since we start the
-/// positions from dictionary_size + 1, zero is always too far to qualify
+/// positions from dict_size + 1, zero is always too far to qualify
/// as usable match position.
#define EMPTY_HASH_VALUE 0
}
-/// When flushing, we cannot run the match finder unless there is find_len_max
+/// When flushing, we cannot run the match finder unless there is nice_len
/// bytes available in the dictionary. Instead, we skip running the match
/// finder (indicating that no match was found), and count how many bytes we
/// have ignored this way.
/// in them.
#define header(is_bt, len_min, ret_op) \
uint32_t len_limit = mf_avail(mf); \
- if (mf->find_len_max <= len_limit) { \
- len_limit = mf->find_len_max; \
+ if (mf->nice_len <= len_limit) { \
+ len_limit = mf->nice_len; \
} else if (len_limit < (len_min) \
|| (is_bt && mf->action == LZMA_SYNC_FLUSH)) { \
assert(mf->action != LZMA_RUN); \
/// of matches found.
#define call_find(func, len_best) \
do { \
- matches_count = func(len_limit, pos, cur, cur_match, mf->loops, \
+ matches_count = func(len_limit, pos, cur, cur_match, mf->depth, \
mf->son, mf->cyclic_pos, mf->cyclic_size, \
matches + matches_count, len_best) \
- matches; \
/// \param pos lzma_mf.read_pos + lzma_mf.offset
/// \param cur Pointer to current byte (mf_ptr(mf))
/// \param cur_match Start position of the current match candidate
-/// \param loops Maximum length of the hash chain
+/// \param depth Maximum length of the hash chain
/// \param son lzma_mf.son (contains the hash chain)
/// \param cyclic_pos
/// \param cyclic_size
const uint32_t pos,
const uint8_t *const cur,
uint32_t cur_match,
- uint32_t loops,
+ uint32_t depth,
uint32_t *const son,
const uint32_t cyclic_pos,
const uint32_t cyclic_size,
while (true) {
const uint32_t delta = pos - cur_match;
- if (loops-- == 0 || delta >= cyclic_size)
+ if (depth-- == 0 || delta >= cyclic_size)
return matches;
const uint8_t *const pb = cur - delta;
const uint32_t pos,
const uint8_t *const cur,
uint32_t cur_match,
- uint32_t loops,
+ uint32_t depth,
uint32_t *const son,
const uint32_t cyclic_pos,
const uint32_t cyclic_size,
while (true) {
const uint32_t delta = pos - cur_match;
- if (loops-- == 0 || delta >= cyclic_size) {
+ if (depth-- == 0 || delta >= cyclic_size) {
*ptr0 = EMPTY_HASH_VALUE;
*ptr1 = EMPTY_HASH_VALUE;
return matches;
const uint32_t pos,
const uint8_t *const cur,
uint32_t cur_match,
- uint32_t loops,
+ uint32_t depth,
uint32_t *const son,
const uint32_t cyclic_pos,
const uint32_t cyclic_size)
while (true) {
const uint32_t delta = pos - cur_match;
- if (loops-- == 0 || delta >= cyclic_size) {
+ if (depth-- == 0 || delta >= cyclic_size) {
*ptr0 = EMPTY_HASH_VALUE;
*ptr1 = EMPTY_HASH_VALUE;
return;
#define bt_skip() \
do { \
- bt_skip_func(len_limit, pos, cur, cur_match, mf->loops, \
+ bt_skip_func(len_limit, pos, cur, cur_match, mf->depth, \
mf->son, mf->cyclic_pos, \
mf->cyclic_size); \
move_pos(mf); \
liblzma2_la_SOURCES = lzma_common.h
-if COND_ENCODER_LZMA
+if COND_ENCODER_LZMA1
liblzma2_la_SOURCES += \
fastpos.h \
lzma_encoder.h \
endif
endif
-if COND_DECODER_LZMA
+if COND_DECODER_LZMA1
liblzma2_la_SOURCES += \
lzma_decoder.c \
lzma_decoder.h
return LZMA_MEM_ERROR;
if (props[0] == 40) {
- opt->dictionary_size = UINT32_MAX;
+ opt->dict_size = UINT32_MAX;
} else {
- opt->dictionary_size = 2 | (props[0] & 1);
- opt->dictionary_size <<= props[0] / 2 + 11;
+ opt->dict_size = 2 | (props[0] & 1);
+ opt->dict_size <<= props[0] / 2 + 11;
}
- opt->preset_dictionary = NULL;
- opt->preset_dictionary_size = 0;
+ opt->preset_dict = NULL;
+ opt->preset_dict_size = 0;
*options = opt;
// Look if there are new options. At least for now,
// only lc/lp/pb can be changed.
if (coder->opt_new != NULL
- && (coder->opt_cur.literal_context_bits
- != coder->opt_new->literal_context_bits
- || coder->opt_cur.literal_pos_bits
- != coder->opt_new->literal_pos_bits
- || coder->opt_cur.pos_bits
- != coder->opt_new->pos_bits)) {
+ && (coder->opt_cur.lc != coder->opt_new->lc
+ || coder->opt_cur.lp != coder->opt_new->lp
+ || coder->opt_cur.pb != coder->opt_new->pb)) {
// Options have been changed, copy them to opt_cur.
- coder->opt_cur.literal_context_bits
- = coder->opt_new->literal_context_bits;
- coder->opt_cur.literal_pos_bits
- = coder->opt_new->literal_pos_bits;
- coder->opt_cur.pos_bits
- = coder->opt_new->pos_bits;
+ coder->opt_cur.lc = coder->opt_new->lc;
+ coder->opt_cur.lp = coder->opt_new->lp;
+ coder->opt_cur.pb = coder->opt_new->pb;
// We need to write the new options and reset
// the encoder state.
// compressed size of a chunk is not smaller than the uncompressed
// size, so we need to have at least LZMA2_COMPRESSED_MAX bytes
// history available.
- if (lz_options->before_size + lz_options->dictionary_size
- < LZMA2_CHUNK_MAX)
- lz_options->before_size = LZMA2_CHUNK_MAX
- - lz_options->dictionary_size;
+ if (lz_options->before_size + lz_options->dict_size < LZMA2_CHUNK_MAX)
+ lz_options->before_size
+ = LZMA2_CHUNK_MAX - lz_options->dict_size;
return LZMA_OK;
}
lzma_lzma2_props_encode(const void *options, uint8_t *out)
{
const lzma_options_lzma *const opt = options;
- uint32_t d = MAX(opt->dictionary_size, LZMA_DICTIONARY_SIZE_MIN);
+ uint32_t d = MAX(opt->dict_size, LZMA_DICT_SIZE_MIN);
// Round up to to the next 2^n - 1 or 2^n + 2^(n - 1) - 1 depending
// on which one is the next:
/// Maximum number of position states. A position state is the lowest pos bits
/// number of bits of the current uncompressed offset. In some places there
/// are different sets of probabilities for different pos states.
-#define POS_STATES_MAX (1 << LZMA_POS_BITS_MAX)
+#define POS_STATES_MAX (1 << LZMA_PB_MAX)
-/// Validates literal_context_bits, literal_pos_bits, and pos_bits.
+/// Validates lc, lp, and pb.
static inline bool
is_lclppb_valid(const lzma_options_lzma *options)
{
- return options->literal_context_bits <= LZMA_LITERAL_CONTEXT_BITS_MAX
- && options->literal_pos_bits
- <= LZMA_LITERAL_POS_BITS_MAX
- && options->literal_context_bits
- + options->literal_pos_bits
- <= LZMA_LITERAL_BITS_MAX
- && options->pos_bits <= LZMA_POS_BITS_MAX;
+ return options->lc <= LZMA_LCLP_MAX && options->lp <= LZMA_LCLP_MAX
+ && options->lc + options->lp <= LZMA_LCLP_MAX
+ && options->pb <= LZMA_PB_MAX;
}
#define LITERAL_CODER_SIZE 0x300
/// Maximum number of literal coders
-#define LITERAL_CODERS_MAX (1 << LZMA_LITERAL_BITS_MAX)
+#define LITERAL_CODERS_MAX (1 << LZMA_LCLP_MAX)
/// Locate the literal coder for the next literal byte. The choice depends on
/// - the lowest literal_pos_bits bits of the position of the current
static inline void
literal_init(probability (*probs)[LITERAL_CODER_SIZE],
- uint32_t literal_context_bits, uint32_t literal_pos_bits)
+ uint32_t lc, uint32_t lp)
{
- assert(literal_context_bits + literal_pos_bits
- <= LZMA_LITERAL_BITS_MAX);
+ assert(lc + lp <= LZMA_LCLP_MAX);
- const uint32_t coders
- = 1U << (literal_context_bits + literal_pos_bits);
+ const uint32_t coders = 1U << (lc + lp);
for (uint32_t i = 0; i < coders; ++i)
for (uint32_t j = 0; j < LITERAL_CODER_SIZE; ++j)
// fastpos.h to understand why).
#define END_POS_MODEL_INDEX 14
-// Seven-bit distances use the full FIXME
+// Pos slots that indicate a distance <= 127.
#define FULL_DISTANCES_BITS (END_POS_MODEL_INDEX / 2)
#define FULL_DISTANCES (1 << FULL_DISTANCES_BITS)
uint32_t rep2; ///< Distance of third latest match
uint32_t rep3; ///< Distance of fourth latest match
- uint32_t pos_mask; // (1U << pos_bits) - 1
+ uint32_t pos_mask; // (1U << pb) - 1
uint32_t literal_context_bits;
uint32_t literal_pos_mask;
// FIXME?
// Calculate pos_mask. We don't need pos_bits as is for anything.
- coder->pos_mask = (1U << options->pos_bits) - 1;
+ coder->pos_mask = (1U << options->pb) - 1;
// Initialize the literal decoder.
- literal_init(coder->literal, options->literal_context_bits,
- options->literal_pos_bits);
+ literal_init(coder->literal, options->lc, options->lp);
- coder->literal_context_bits = options->literal_context_bits;
- coder->literal_pos_mask = (1 << options->literal_pos_bits) - 1;
+ coder->literal_context_bits = options->lc;
+ coder->literal_pos_mask = (1U << options->lp) - 1;
// State
coder->state = STATE_LIT_LIT;
coder->rep1 = 0;
coder->rep2 = 0;
coder->rep3 = 0;
- coder->pos_mask = (1 << options->pos_bits) - 1;
+ coder->pos_mask = (1U << options->pb) - 1;
// Range decoder
rc_reset(coder->rc);
bittree_reset(coder->pos_align, ALIGN_BITS);
// Len decoders (also bit/bittree)
- const uint32_t num_pos_states = 1 << options->pos_bits;
+ const uint32_t num_pos_states = 1U << options->pb;
bit_reset(coder->match_len_decoder.choice);
bit_reset(coder->match_len_decoder.choice2);
bit_reset(coder->rep_len_decoder.choice);
// All dictionary sizes are OK here. LZ decoder will take care of
// the special cases.
const lzma_options_lzma *options = opt;
- *dict_size = options->dictionary_size;
+ *dict_size = options->dict_size;
return LZMA_OK;
}
return true;
// See the file format specification to understand this.
- options->pos_bits = byte / (9 * 5);
- byte -= options->pos_bits * 9 * 5;
- options->literal_pos_bits = byte / 9;
- options->literal_context_bits = byte - options->literal_pos_bits * 9;
+ options->pb = byte / (9 * 5);
+ byte -= options->pb * 9 * 5;
+ options->lp = byte / 9;
+ options->lc = byte - options->lp * 9;
- return options->literal_context_bits + options->literal_pos_bits
- > LZMA_LITERAL_BITS_MAX;
+ return options->lc + options->lp > LZMA_LCLP_MAX;
}
lzma_lzma_decoder_memusage(const void *options)
{
const lzma_options_lzma *const opt = options;
- const uint64_t lz_memusage
- = lzma_lz_decoder_memusage(opt->dictionary_size);
+ const uint64_t lz_memusage = lzma_lz_decoder_memusage(opt->dict_size);
if (lz_memusage == UINT64_MAX)
return UINT64_MAX;
// All dictionary sizes are accepted, including zero. LZ decoder
// will automatically use a dictionary at least a few KiB even if
// a smaller dictionary is requested.
- opt->dictionary_size = integer_read_32(props + 1);
+ opt->dict_size = integer_read_32(props + 1);
- opt->preset_dictionary = NULL;
- opt->preset_dictionary_size = 0;
+ opt->preset_dict = NULL;
+ opt->preset_dict_size = 0;
*options = opt;
// LZ encoder initialization does the validation, also when just
// calculating memory usage, so we don't need to validate here.
lz_options->before_size = OPTS;
- lz_options->dictionary_size = options->dictionary_size;
+ lz_options->dict_size = options->dict_size;
lz_options->after_size = LOOP_INPUT_MAX;
lz_options->match_len_max = MATCH_LEN_MAX;
- lz_options->find_len_max = options->fast_bytes;
- lz_options->match_finder = options->match_finder;
- lz_options->match_finder_cycles = options->match_finder_cycles;
- lz_options->preset_dictionary = options->preset_dictionary;
- lz_options->preset_dictionary_size = options->preset_dictionary_size;
+ lz_options->nice_len = options->nice_len;
+ lz_options->match_finder = options->mf;
+ lz_options->depth = options->depth;
+ lz_options->preset_dict = options->preset_dict;
+ lz_options->preset_dict_size = options->preset_dict_size;
}
{
assert(!coder->is_flushed);
- coder->pos_mask = (1U << options->pos_bits) - 1;
- coder->literal_context_bits = options->literal_context_bits;
- coder->literal_pos_mask = (1U << options->literal_pos_bits) - 1;
+ coder->pos_mask = (1U << options->pb) - 1;
+ coder->literal_context_bits = options->lc;
+ coder->literal_pos_mask = (1U << options->lp) - 1;
// Range coder
rc_reset(&coder->rc);
for (size_t i = 0; i < REP_DISTANCES; ++i)
coder->reps[i] = 0;
- literal_init(coder->literal, options->literal_context_bits,
- options->literal_pos_bits);
+ literal_init(coder->literal, options->lc, options->lp);
// Bit encoders
for (size_t i = 0; i < STATES; ++i) {
// Length encoders
length_encoder_reset(&coder->match_len_encoder,
- 1U << options->pos_bits, coder->fast_mode);
+ 1U << options->pb, coder->fast_mode);
length_encoder_reset(&coder->rep_len_encoder,
- 1U << options->pos_bits, coder->fast_mode);
+ 1U << options->pb, coder->fast_mode);
// Price counts are incremented every time appropriate probabilities
// are changed. price counts are set to zero when the price tables
// Validate some of the options. LZ encoder validates fast_bytes too
// but we need a valid value here earlier.
- if (!is_lclppb_valid(options) || options->fast_bytes < MATCH_LEN_MIN
- || options->fast_bytes > MATCH_LEN_MAX)
+ if (!is_lclppb_valid(options) || options->nice_len < MATCH_LEN_MIN
+ || options->nice_len > MATCH_LEN_MAX)
return LZMA_OPTIONS_ERROR;
// Set compression mode.
// Set dist_table_size.
// Round the dictionary size up to next 2^n.
uint32_t log_size = 0;
- while ((UINT32_C(1) << log_size)
- < options->dictionary_size)
+ while ((UINT32_C(1) << log_size) < options->dict_size)
++log_size;
coder->dist_table_size = log_size * 2;
// Length encoders' price table size
coder->match_len_encoder.table_size
- = options->fast_bytes + 1 - MATCH_LEN_MIN;
+ = options->nice_len + 1 - MATCH_LEN_MIN;
coder->rep_len_encoder.table_size
- = options->fast_bytes + 1 - MATCH_LEN_MIN;
+ = options->nice_len + 1 - MATCH_LEN_MIN;
break;
}
extern bool
lzma_lzma_lclppb_encode(const lzma_options_lzma *options, uint8_t *byte)
{
- if (options->literal_context_bits > LZMA_LITERAL_CONTEXT_BITS_MAX
- || options->literal_pos_bits
- > LZMA_LITERAL_POS_BITS_MAX
- || options->pos_bits > LZMA_POS_BITS_MAX
- || options->literal_context_bits
- + options->literal_pos_bits
- > LZMA_LITERAL_BITS_MAX)
+ if (!is_lclppb_valid(options))
return true;
- *byte = (options->pos_bits * 5 + options->literal_pos_bits) * 9
- + options->literal_context_bits;
+ *byte = (options->pb * 5 + options->lp) * 9 + options->lc;
assert(*byte <= (4 * 5 + 4) * 9 + 8);
return false;
}
-#ifdef HAVE_ENCODER_LZMA
+#ifdef HAVE_ENCODER_LZMA1
extern lzma_ret
lzma_lzma_props_encode(const void *options, uint8_t *out)
{
if (lzma_lzma_lclppb_encode(opt, out))
return LZMA_PROG_ERROR;
- integer_write_32(out + 1, opt->dictionary_size);
+ integer_write_32(out + 1, opt->dict_size);
return LZMA_OK;
}
lzma_lzma_optimum_fast(lzma_coder *restrict coder, lzma_mf *restrict mf,
uint32_t *restrict back_res, uint32_t *restrict len_res)
{
- const uint32_t fast_bytes = mf->find_len_max;
+ const uint32_t nice_len = mf->nice_len;
uint32_t len_main;
uint32_t matches_count;
&& buf[len] == buf_back[len]; ++len) ;
// If we have found a repeated match that is at least
- // fast_bytes long, return it immediatelly.
- if (len >= fast_bytes) {
+ // nice_len long, return it immediatelly.
+ if (len >= nice_len) {
*back_res = i;
*len_res = len;
mf_skip(mf, len - 1);
}
// We didn't find a long enough repeated match. Encode it as a normal
- // match if the match length is at least fast_bytes.
- if (len_main >= fast_bytes) {
+ // match if the match length is at least nice_len.
+ if (len_main >= nice_len) {
*back_res = coder->matches[matches_count - 1].dist
+ REP_DISTANCES;
*len_res = len_main;
uint32_t *restrict back_res, uint32_t *restrict len_res,
uint32_t position)
{
- const uint32_t fast_bytes = mf->find_len_max;
+ const uint32_t nice_len = mf->nice_len;
uint32_t len_main;
uint32_t matches_count;
rep_max_index = i;
}
- if (rep_lens[rep_max_index] >= fast_bytes) {
+ if (rep_lens[rep_max_index] >= nice_len) {
*back_res = rep_max_index;
*len_res = rep_lens[rep_max_index];
mf_skip(mf, *len_res - 1);
}
- if (len_main >= fast_bytes) {
+ if (len_main >= nice_len) {
*back_res = coder->matches[matches_count - 1].dist
+ REP_DISTANCES;
*len_res = len_main;
static inline uint32_t
helper2(lzma_coder *coder, uint32_t *reps, const uint8_t *buf,
uint32_t len_end, uint32_t position, const uint32_t cur,
- const uint32_t fast_bytes, const uint32_t buf_avail_full)
+ const uint32_t nice_len, const uint32_t buf_avail_full)
{
uint32_t matches_count = coder->matches_count;
uint32_t new_len = coder->longest_match_length;
if (buf_avail_full < 2)
return len_end;
- const uint32_t buf_avail = MIN(buf_avail_full, fast_bytes);
+ const uint32_t buf_avail = MIN(buf_avail_full, nice_len);
if (!next_is_literal && match_byte != current_byte) { // speed optimization
// try literal + rep0
const uint8_t *const buf_back = buf - reps[0] - 1;
- const uint32_t limit = MIN(buf_avail_full, fast_bytes + 1);
+ const uint32_t limit = MIN(buf_avail_full, nice_len + 1);
uint32_t len_test = 1;
while (len_test < limit && buf[len_test] == buf_back[len_test])
uint32_t len_test_2 = len_test + 1;
const uint32_t limit = MIN(buf_avail_full,
- len_test_2 + fast_bytes);
+ len_test_2 + nice_len);
for (; len_test_2 < limit
&& buf[len_test_2] == buf_back[len_test_2];
++len_test_2) ;
const uint8_t *const buf_back = buf - cur_back - 1;
uint32_t len_test_2 = len_test + 1;
const uint32_t limit = MIN(buf_avail_full,
- len_test_2 + fast_bytes);
+ len_test_2 + nice_len);
for (; len_test_2 < limit &&
buf[len_test_2] == buf_back[len_test_2];
coder->longest_match_length = mf_find(
mf, &coder->matches_count, coder->matches);
- if (coder->longest_match_length >= mf->find_len_max)
+ if (coder->longest_match_length >= mf->nice_len)
break;
len_end = helper2(coder, reps, mf_ptr(mf) - 1, len_end,
- position + cur, cur, mf->find_len_max,
+ position + cur, cur, mf->nice_len,
MIN(mf_avail(mf) + 1, OPTS - 1 - cur));
}
#include "common.h"
+/*
#define pow2(e) (UINT32_C(1) << (e))
-LZMA_API const lzma_options_lzma lzma_preset_lzma[9] = {
+static const lzma_options_lzma presets[9] = {
// dict lc lp pb mode fb mf mfc
{ pow2(16), NULL, 0, 3, 0, 2, false, LZMA_MODE_FAST, 64, LZMA_MF_HC3, 0, 0, 0, 0, 0, NULL, NULL },
{ pow2(20), NULL, 0, 3, 0, 0, false, LZMA_MODE_FAST, 64, LZMA_MF_HC4, 0, 0, 0, 0, 0, NULL, NULL },
};
-/*
extern LZMA_API lzma_bool
-lzma_preset_lzma(lzma_options_lzma *options, uint32_t level)
+lzma_lzma_preset(lzma_options_lzma *options, uint32_t level)
{
- *options = (lzma_options_lzma){
+ if (level >= ARRAY_SIZE(presetes))
+ return true;
- };
+ *options = presets[level];
+ return false;
+}
+*/
- options->literal_context_bits = LZMA_LITERAL_CONTEXT_BITS_DEFAULT
- options->literal_pos_bits = LZMA_LITERAL_POS_BITS_DEFAULT;
- options->pos_bits = LZMA_POS_BITS_DEFAULT;
- options->preset_dictionary = NULL;
- options->preset_dictionary_size = 0;
- options->persistent = false;
- options->mode = level <= 2 ? LZMA_MODE_FAST : LZMA_MODE_NORMAL;
- options->fast_bytes = level <=
+extern LZMA_API lzma_bool
+lzma_lzma_preset(lzma_options_lzma *options, uint32_t level)
+{
+ if (level >= 9)
+ return true;
+
+ memzero(options, sizeof(*options));
- options->match_finder = level == 1 ? LZMA_MF_HC3
- : (level == 2 ? LZMA_MF_HC4 : LZMA_MF_BT4);
- options->match_finder_cycles = 0;
+ static const uint8_t shift[9] = { 16, 20, 19, 20, 21, 22, 23, 24, 25 };
+ options->dict_size = UINT32_C(1) << shift[level];
+ options->preset_dict = NULL;
+ options->preset_dict_size = 0;
+ options->lc = LZMA_LC_DEFAULT;
+ options->lp = LZMA_LP_DEFAULT;
+ options->pb = LZMA_PB_DEFAULT;
- options->dictionary_size =
+ options->persistent = false;
+ options->mode = level <= 2 ? LZMA_MODE_FAST : LZMA_MODE_NORMAL;
+
+ options->nice_len = level <= 5 ? 32 : 64;
+ options->mf = level <= 1 ? LZMA_MF_HC3 : level <= 2 ? LZMA_MF_HC4
+ : LZMA_MF_BT4;
+ options->depth = 0;
+
+ return false;
}
-*/
-I@top_srcdir@/src/liblzma/api \
-I@top_srcdir@/src/liblzma/common
-if COND_ENCODER_LZMA
+if COND_ENCODER_LZMA1
librangecoder_la_SOURCES += \
range_encoder.h \
price.h
endif
endif
-if COND_DECODER_LZMA
+if COND_DECODER_LZMA1
librangecoder_la_SOURCES += range_decoder.h
endif
// Optimization: We know that LZMA uses End of Payload Marker
// (not End of Input), so we can omit the helper filter.
- if (filters[0].id == LZMA_FILTER_LZMA)
+ if (filters[0].id == LZMA_FILTER_LZMA1)
filters[1].id = LZMA_VLI_UNKNOWN;
return_if_error(lzma_raw_decoder_init(
static void
add_filter(lzma_vli id, const char *opt_str)
{
- if (filter_count == 7) {
+ if (filter_count == LZMA_BLOCK_FILTERS_MAX) {
errmsg(V_ERROR, _("Maximum number of filters is seven"));
my_exit(ERROR);
}
= parse_options_delta(opt_str);
break;
- case LZMA_FILTER_LZMA:
+ case LZMA_FILTER_LZMA1:
case LZMA_FILTER_LZMA2:
opt_filters[filter_count].options
= parse_options_lzma(opt_str);
break;
case OPT_LZMA1:
- add_filter(LZMA_FILTER_LZMA, optarg);
+ add_filter(LZMA_FILTER_LZMA1, optarg);
break;
case OPT_LZMA2:
static void
set_compression_settings(void)
{
+ static lzma_options_lzma opt_lzma;
+
if (filter_count == 0) {
+ if (lzma_lzma_preset(&opt_lzma, preset_number)) {
+ errmsg(V_ERROR, _("Internal error (bug)"));
+ my_exit(ERROR);
+ }
+
opt_filters[0].id = opt_header == HEADER_ALONE
- ? LZMA_FILTER_LZMA : LZMA_FILTER_LZMA2;
- opt_filters[0].options = (lzma_options_lzma *)(
- lzma_preset_lzma + preset_number);
+ ? LZMA_FILTER_LZMA1 : LZMA_FILTER_LZMA2;
+ opt_filters[0].options = &opt_lzma;
filter_count = 1;
}
// If we are using the LZMA_Alone format, allow exactly one filter
// which has to be LZMA.
if (opt_header == HEADER_ALONE && (filter_count != 1
- || opt_filters[0].id != LZMA_FILTER_LZMA)) {
- errmsg(V_ERROR, _("With --format=alone only the LZMA filter "
+ || opt_filters[0].id != LZMA_FILTER_LZMA1)) {
+ errmsg(V_ERROR, _("With --format=alone only the LZMA1 filter "
"is supported"));
my_exit(ERROR);
}
+ // TODO: liblzma probably needs an API to validate the filter chain.
+
// If using --format=raw, we can be decoding.
uint64_t memory_usage = opt_mode == MODE_COMPRESS
? lzma_memusage_encoder(opt_filters)
my_exit(ERROR);
}
- --preset_number;
- opt_filters[0].options = (lzma_options_lzma *)(
- lzma_preset_lzma
- + preset_number);
+ if (lzma_lzma_preset(&opt_lzma, --preset_number)) {
+ errmsg(V_ERROR, _("Internal error (bug)"));
+ my_exit(ERROR);
+ }
+
memory_usage = lzma_memusage_encoder(opt_filters);
}
} else {
" --lzma1=[OPTS] LZMA1 or LZMA2; OPTS is a comma-separated list of zero or\n"
" --lzma2=[OPTS] more of the following options (valid values; default):\n"
" dict=NUM dictionary size in bytes (1 - 1GiB; 8MiB)\n"
-" lc=NUM number of literal context bits (0-8; 3)\n"
+" lc=NUM number of literal context bits (0-4; 3)\n"
" lp=NUM number of literal position bits (0-4; 0)\n"
" pb=NUM number of position bits (0-4; 2)\n"
" mode=MODE compression mode (`fast' or `best'; `best')\n"
-" fb=NUM number of fast bytes (5-273; 128)\n"
+" nice=NUM nice length of a match (2-273; 64)\n"
" mf=NAME match finder (hc3, hc4, bt2, bt3, bt4; bt4)\n"
-" mfc=NUM match finder cycles; 0=automatic (default)\n"
+" depth=NUM maximum search depth; 0=automatic (default)\n"
"\n"
" --x86 x86 filter (sometimes called BCJ filter)\n"
" --powerpc PowerPC (big endian) filter\n"
" --sparc SPARC filter\n"
"\n"
" --delta=[OPTS] Delta filter; valid OPTS (valid values; default):\n"
-" distance=NUM Distance between bytes being\n"
-" subtracted from each other (1-256; 1)\n"
+" dist=NUM distance between bytes being subtracted\n"
+" from each other (1-256; 1)\n"
"\n"
-" --copy No filtering (useful only when specified alone)\n"
" --subblock=[OPTS] Subblock filter; valid OPTS (valid values; default):\n"
-" size=NUM number of bytes of data per subblock\n"
-" (1 - 256Mi; 4Ki)\n"
-" rle=NUM run-length encoder chunk size (0-256; 0)\n"
+" size=NUM number of bytes of data per subblock\n"
+" (1 - 256Mi; 4Ki)\n"
+" rle=NUM run-length encoder chunk size (0-256; 0)\n"
));
puts(_(
if (value == NULL || value[0] == '\0') {
errmsg(V_ERROR, _("%s: Options must be `name=value' "
- "pairs separated with commas"),
- str);
+ "pairs separated with commas"), str);
my_exit(ERROR);
}
///////////
enum {
- OPT_DISTANCE,
+ OPT_DIST,
};
{
lzma_options_delta *opt = options;
switch (key) {
- case OPT_DISTANCE:
- opt->distance = value;
+ case OPT_DIST:
+ opt->dist = value;
break;
}
}
parse_options_delta(const char *str)
{
static const option_map opts[] = {
- { "distance", NULL, LZMA_DELTA_DISTANCE_MIN,
- LZMA_DELTA_DISTANCE_MAX },
+ { "dist", NULL, LZMA_DELTA_DIST_MIN,
+ LZMA_DELTA_DIST_MAX },
{ NULL, NULL, 0, 0 }
};
lzma_options_delta *options = xmalloc(sizeof(lzma_options_subblock));
*options = (lzma_options_delta){
// It's hard to give a useful default for this.
- .distance = LZMA_DELTA_DISTANCE_MIN,
+ .type = LZMA_DELTA_TYPE_BYTE,
+ .dist = LZMA_DELTA_DIST_MIN,
};
parse_options(str, opts, &set_delta, options);
OPT_LP,
OPT_PB,
OPT_MODE,
- OPT_FB,
+ OPT_NICE,
OPT_MF,
- OPT_MC
+ OPT_DEPTH,
};
switch (key) {
case OPT_DICT:
- opt->dictionary_size = value;
+ opt->dict_size = value;
break;
case OPT_LC:
- opt->literal_context_bits = value;
+ opt->lc = value;
break;
case OPT_LP:
- opt->literal_pos_bits = value;
+ opt->lp = value;
break;
case OPT_PB:
- opt->pos_bits = value;
+ opt->pb = value;
break;
case OPT_MODE:
opt->mode = value;
break;
- case OPT_FB:
- opt->fast_bytes = value;
+ case OPT_NICE:
+ opt->nice_len = value;
break;
case OPT_MF:
- opt->match_finder = value;
+ opt->mf = value;
break;
- case OPT_MC:
- opt->match_finder_cycles = value;
+ case OPT_DEPTH:
+ opt->depth = value;
break;
}
}
};
static const option_map opts[] = {
- { "dict", NULL, LZMA_DICTIONARY_SIZE_MIN,
- LZMA_DICTIONARY_SIZE_MAX },
- { "lc", NULL, LZMA_LITERAL_CONTEXT_BITS_MIN,
- LZMA_LITERAL_CONTEXT_BITS_MAX },
- { "lp", NULL, LZMA_LITERAL_POS_BITS_MIN,
- LZMA_LITERAL_POS_BITS_MAX },
- { "pb", NULL, LZMA_POS_BITS_MIN, LZMA_POS_BITS_MAX },
- { "mode", modes, 0, 0 },
- { "fb", NULL, LZMA_FAST_BYTES_MIN, LZMA_FAST_BYTES_MAX },
- { "mf", mfs, 0, 0 },
- { "mc", NULL, 0, UINT32_MAX },
- { NULL, NULL, 0, 0 }
+ { "dict", NULL, LZMA_DICT_SIZE_MIN,
+ (UINT32_C(1) << 30) + (UINT32_C(1) << 29) },
+ { "lc", NULL, LZMA_LCLP_MIN, LZMA_LCLP_MAX },
+ { "lp", NULL, LZMA_LCLP_MIN, LZMA_LCLP_MAX },
+ { "pb", NULL, LZMA_PB_MIN, LZMA_PB_MAX },
+ { "mode", modes, 0, 0 },
+ { "nice", NULL, 2, 273 },
+ { "mf", mfs, 0, 0 },
+ { "depth", NULL, 0, UINT32_MAX },
+ { NULL, NULL, 0, 0 }
};
+ // TODO There should be a way to take some preset as the base for
+ // custom settings.
lzma_options_lzma *options = xmalloc(sizeof(lzma_options_lzma));
*options = (lzma_options_lzma){
- .dictionary_size = LZMA_DICTIONARY_SIZE_DEFAULT,
- .literal_context_bits = LZMA_LITERAL_CONTEXT_BITS_DEFAULT,
- .literal_pos_bits = LZMA_LITERAL_POS_BITS_DEFAULT,
- .pos_bits = LZMA_POS_BITS_DEFAULT,
- .preset_dictionary = NULL,
+ .dict_size = LZMA_DICT_SIZE_DEFAULT,
+ .preset_dict = NULL,
+ .preset_dict_size = 0,
+ .lc = LZMA_LC_DEFAULT,
+ .lp = LZMA_LP_DEFAULT,
+ .pb = LZMA_PB_DEFAULT,
.persistent = false,
.mode = LZMA_MODE_NORMAL,
- .fast_bytes = LZMA_FAST_BYTES_DEFAULT,
- .match_finder = LZMA_MF_BT4,
- .match_finder_cycles = 0,
+ .nice_len = 64,
+ .mf = LZMA_MF_BT4,
+ .depth = 0,
};
parse_options(str, opts, &set_lzma, options);
+ if (options->lc + options->lp > LZMA_LCLP_MAX) {
+ errmsg(V_ERROR, "The sum of lc and lp must be at "
+ "maximum of 4");
+ exit(ERROR);
+ }
+
+ const uint32_t nice_len_min = options->mf & 0x0F;
+ if (options->nice_len < nice_len_min) {
+ errmsg(V_ERROR, "The selected match finder requires at "
+ "least nice=%" PRIu32, nice_len_min);
+ exit(ERROR);
+ }
+
return options;
}
static lzma_block known_options;
static lzma_block decoded_options;
+static lzma_options_lzma opt_lzma;
+
static lzma_filter filters_none[1] = {
{
.id = LZMA_VLI_UNKNOWN,
static lzma_filter filters_one[2] = {
{
.id = LZMA_FILTER_LZMA2,
- .options = (void *)(&lzma_preset_lzma[0]),
+ .options = &opt_lzma,
}, {
.id = LZMA_VLI_UNKNOWN,
}
.options = NULL,
}, {
.id = LZMA_FILTER_LZMA2,
- .options = (void *)(&lzma_preset_lzma[0]),
+ .options = &opt_lzma,
}, {
.id = LZMA_VLI_UNKNOWN,
}
.options = NULL,
}, {
.id = LZMA_FILTER_LZMA2,
- .options = (void *)(&lzma_preset_lzma[0]),
+ .options = &opt_lzma,
}, {
.id = LZMA_VLI_UNKNOWN,
}
main(void)
{
lzma_init();
+ succeed(lzma_lzma_preset(&opt_lzma, 0));
test1();
test2();
--armthumb \
--sparc
do
- test_lzma $ARGS --lzma2=dict=64KiB,fb=32,mode=fast
- test_lzma --subblock $ARGS --lzma2=dict=64KiB,fb=32,mode=fast
+ test_lzma $ARGS --lzma2=dict=64KiB,nice=32,mode=fast
+ test_lzma --subblock $ARGS --lzma2=dict=64KiB,nice=32,mode=fast
done
echo
test_lzma(void)
{
// Test 1
- known_flags.id = LZMA_FILTER_LZMA;
+ known_flags.id = LZMA_FILTER_LZMA1;
known_flags.options = NULL;
expect(encode(99));
projects / xz / commitdiff