* of the bits within the 16-bit coding units is such that the first bit is the
* high-order bit and the last bit is the low-order bit.
*
- * From these two logical bitstreams, a LZMS decompressor can reconstitute the
+ * From these two logical bitstreams, an LZMS decompressor can reconstitute the
* series of items that make up the LZMS data representation. Each such item
* may be a literal byte or a match. Matches may be either traditional LZ77
* matches or "delta" matches, either of which can have its offset encoded
* sequence of bytes beginning at the current position and extending for the
* length is exactly equal to the equal-length sequence of bytes at the offset
* back in the window. On the other hand, a delta match consists of a length,
- * raw offset, and power. It asserts that the sequence of bytes of beginning at
+ * raw offset, and power. It asserts that the sequence of bytes beginning at
* the current position and extending for the length is equal to the bytewise
* sum of the two equal-length sequences of bytes (2**power) and (raw_offset *
* 2**power) bytes before the current position, minus bytewise the sequence of
* filled in with the next 16 bits from the forwards bitstream.
*
* To decode each bit, the range decoder requires a probability that is
- * logically a real number between 0 and 1. Multiplying this
- * probability by the current range and taking the floor gives the bound between
- * the 0-bit region of the range and the 1-bit region of the range. However, in
- * LZMS, probabilities are restricted to values of n/64 where n is an integer is
+ * logically a real number between 0 and 1. Multiplying this probability by the
+ * current range and taking the floor gives the bound between the 0-bit region
+ * of the range and the 1-bit region of the range. However, in LZMS,
+ * probabilities are restricted to values of n/64 where n is an integer is
* between 1 and 63 inclusively, so the implementation may use integer
* operations instead. Following calculation of the bound, if the current code
* is in the 0-bit region, the new range becomes the current code and the
* decoded bit is 0; otherwise, the bound must be subtracted from both the range
* and the code, and the decoded bit is 1. More information about range coding
- * can be found https://en.wikipedia.org/wiki/Range_encoding. Furthermore, note
- * that the LZMA format also uses range coding and has public domain code
+ * can be found at https://en.wikipedia.org/wiki/Range_encoding. Furthermore,
+ * note that the LZMA format also uses range coding and has public domain code
* available for it.
*
* The probability used to range-decode each bit must be taken from a table, of
* bitstream. For this, there are 5 different Huffman codes used:
*
* - The literal code, used for decoding literal bytes. Each of the 256
- * symbols represents literal byte. This code must be rebuilt whenever 1024
- * symbols have been decoded with it.
+ * symbols represents a literal byte. This code must be rebuilt whenever
+ * 1024 symbols have been decoded with it.
*
* - The LZ offset code, used for decoding the offsets of standard LZ77
* matches. Each symbol represents a position slot, which corresponds to a
*
* Codewords in all the LZMS Huffman codes are limited to 15 bits. If the
* canonical code for a given set of symbol frequencies has any codewords longer
- * than 15 bits, all frequencies must be divided by 2, rounding up, and the code
- * construction must be attempted again.
+ * than 15 bits, then all frequencies must be divided by 2, rounding up, and the
+ * code construction must be attempted again.
*
- * A LZMS-compressed block seemingly cannot have a size greater than or equal to
- * the original uncompressed size. In such cases the block must be stored
+ * A LZMS-compressed block seemingly cannot have a compressed size greater than
+ * or equal to the uncompressed size. In such cases the block must be stored
* uncompressed.
*
* After all LZMS items have been decoded, the data must be postprocessed to
#endif
#include "wimlib.h"
-#include "wimlib/compress.h"
-#include "wimlib/decompress.h"
+#include "wimlib/compress_common.h"
+#include "wimlib/decompressor_ops.h"
+#include "wimlib/decompress_common.h"
#include "wimlib/error.h"
#include "wimlib/lzms.h"
#include "wimlib/util.h"
u32 upcoming_lz_offset;
u32 upcoming_delta_power;
u32 upcoming_delta_offset;
+
+ /* Used for postprocessing */
+ s32 last_target_usages[65536];
};
/* A table that maps position slots to their base values. These are constants
{
int ret;
- /* XXX: This implementation that makes use of code already implemented
- * for the XPRESS and LZX compression formats. However, since for the
+ /* XXX: This implementation makes use of code already implemented for
+ * the XPRESS and LZX compression formats. However, since for the
* adaptive codes used in LZMS we don't actually need the explicit codes
* themselves, only the decode tables, it may be possible to optimize
* this by somehow directly building or updating the Huffman decode
return 0;
}
-/* Validate a LZ match and copy it to the output buffer. */
+/* Validate an LZ match and copy it to the output buffer. */
static int
lzms_copy_lz_match(struct lzms_decompressor *ctx, u32 length, u32 offset)
{
for (int i = LZMS_NUM_RECENT_OFFSETS - 1; i >= 0; i--)
ctx->recent_lz_offsets[i + 1] = ctx->recent_lz_offsets[i];
ctx->recent_lz_offsets[0] = ctx->prev_lz_offset;
-
}
if (ctx->prev_delta_offset != 0) {
dec->sym_freqs[i] = 1;
}
-/* Prepare to decode items from a LZMS-compressed block. */
+/* Prepare to decode items from an LZMS-compressed block. */
static void
lzms_init_decompressor(struct lzms_decompressor *ctx,
const void *cdata, unsigned clen,
/* Initialize position and length slot bases if not done already. */
lzms_init_slot_bases();
- /* Like in other compression formats such as LZX and DEFLATE, match
- * offsets in LZMS are represented as a position slot, which corresponds
- * to a fixed lesser or equal match offset, followed by a
- * position-slot-dependent number of extra bits that gives an additional
- * offset from that position slot. Because the full number of position
- * slots may exceed the length of the compressed block, here we
- * calculate the number of position slots that will actually be used in
- * the compressed representation. */
+ /* Calculate the number of position slots needed for this compressed
+ * block. */
num_position_slots = lzms_get_position_slot_raw(ulen - 1) + 1;
LZMS_DEBUG("Using %u position slots", num_position_slots);
LZMS_DELTA_POWER_CODE_REBUILD_FREQ);
- /* Initialize range decoders (all of which wrap around the same
- * lzms_range_decoder_raw). */
+ /* Initialize range decoders, all of which wrap around the same
+ * lzms_range_decoder_raw. */
lzms_init_range_decoder(&ctx->main_range_decoder,
&ctx->rd, LZMS_NUM_MAIN_STATES);
lzms_init_range_decoder(&ctx->delta_repeat_match_range_decoders[i],
&ctx->rd, LZMS_NUM_DELTA_REPEAT_MATCH_STATES);
-
/* Initialize the LRU queue for recent match offsets. */
for (size_t i = 0; i < LZMS_NUM_RECENT_OFFSETS + 1; i++)
ctx->recent_lz_offsets[i] = i + 1;
/* Decode the series of literals and matches from the LZMS-compressed data.
* Returns 0 on success; nonzero if the compressed data is invalid. */
static int
-lzms_decode_items(const u8 *cdata, size_t clen, u8 *ubuf, size_t ulen)
+lzms_decode_items(const u8 *cdata, size_t clen, u8 *ubuf, size_t ulen,
+ struct lzms_decompressor *ctx)
{
- /* XXX: The context could be allocated on the heap. */
- struct lzms_decompressor ctx;
-
/* Initialize the LZMS decompressor. */
- lzms_init_decompressor(&ctx, cdata, clen, ubuf, ulen);
+ lzms_init_decompressor(ctx, cdata, clen, ubuf, ulen);
/* Decode the sequence of items. */
- while (ctx.out_next != ctx.out_end) {
- LZMS_DEBUG("Position %u", ctx.out_next - ctx.out_begin);
- if (lzms_decode_item(&ctx))
+ while (ctx->out_next != ctx->out_end) {
+ LZMS_DEBUG("Position %u", ctx->out_next - ctx->out_begin);
+ if (lzms_decode_item(ctx))
return -1;
}
return 0;
* actually needs to be done (or to plug in alternate filters, like in LZMA),
* and the corresponding preprocessing seems to be done unconditionally. */
static void
-lzms_postprocess_data(u8 *ubuf, s32 ulen)
+lzms_postprocess_data(u8 *ubuf, s32 ulen, s32 *last_target_usages)
{
/* Offset (from beginning of buffer) of the most recent reference to a
* seemingly valid target address. */
s32 closest_target_usage = -LZMS_X86_MAX_TRANSLATION_OFFSET - 1;
- /* Offset (from beginning of buffer) of the most recently used target
- * address beginning with two bytes equal to the array index.
- *
- * XXX: This array could be allocated on the heap. */
- s32 last_target_usages[65536];
+ /* Initialize the last_target_usages array. Each entry will contain the
+ * offset (from beginning of buffer) of the most recently used target
+ * address beginning with two bytes equal to the array index. */
for (s32 i = 0; i < 65536; i++)
last_target_usages[i] = -LZMS_X86_MAX_GOOD_TARGET_OFFSET - 1;
last_target_usages);
}
-/* API function documented in wimlib.h */
-WIMLIBAPI int
-wimlib_lzms_decompress(const void *cdata, unsigned clen,
- void *ubuf, unsigned ulen)
+static int
+lzms_decompress(const void *compressed_data, size_t compressed_size,
+ void *uncompressed_data, size_t uncompressed_size, void *_ctx)
{
+ struct lzms_decompressor *ctx = _ctx;
+
/* The range decoder requires that a minimum of 4 bytes of compressed
* data be initially available. */
- if (clen < 4) {
- LZMS_DEBUG("Compressed length too small (got %u, expected >= 4)",
- clen);
+ if (compressed_size < 4) {
+ LZMS_DEBUG("Compressed size too small (got %zu, expected >= 4)",
+ compressed_size);
return -1;
}
/* A LZMS-compressed data block should be evenly divisible into 16-bit
* integers. */
- if (clen % 2 != 0) {
- LZMS_DEBUG("Compressed length not divisible by 2 (got %u)", clen);
+ if (compressed_size % 2 != 0) {
+ LZMS_DEBUG("Compressed size not divisible by 2 (got %zu)",
+ compressed_size);
return -1;
}
/* Handle the trivial case where nothing needs to be decompressed.
* (Necessary because a window of size 0 does not have a valid position
* slot.) */
- if (ulen == 0)
+ if (uncompressed_size == 0)
return 0;
/* The x86 post-processor requires that the uncompressed length fit into
* a signed 32-bit integer. Also, the position slot table cannot be
* searched for a position of INT32_MAX or greater. */
- if (ulen >= INT32_MAX) {
+ if (uncompressed_size >= INT32_MAX) {
LZMS_DEBUG("Uncompressed length too large "
"(got %u, expected < INT32_MAX)", ulen);
return -1;
}
/* Decode the literals and matches. */
- if (lzms_decode_items(cdata, clen, ubuf, ulen))
+ if (lzms_decode_items(compressed_data, compressed_size,
+ uncompressed_data, uncompressed_size, ctx))
return -1;
/* Postprocess the data. */
- lzms_postprocess_data(ubuf, ulen);
+ lzms_postprocess_data(uncompressed_data, uncompressed_size,
+ ctx->last_target_usages);
LZMS_DEBUG("Decompression successful.");
return 0;
}
+
+static void
+lzms_free_decompressor(void *_ctx)
+{
+ struct lzms_decompressor *ctx = _ctx;
+
+ FREE(ctx);
+}
+
+static int
+lzms_create_decompressor(size_t max_block_size,
+ const struct wimlib_decompressor_params_header *params,
+ void **ctx_ret)
+{
+ struct lzms_decompressor *ctx;
+
+ ctx = MALLOC(sizeof(struct lzms_decompressor));
+ if (ctx == NULL)
+ return WIMLIB_ERR_NOMEM;
+
+ *ctx_ret = ctx;
+ return 0;
+}
+
+const struct decompressor_ops lzms_decompressor_ops = {
+ .create_decompressor = lzms_create_decompressor,
+ .decompress = lzms_decompress,
+ .free_decompressor = lzms_free_decompressor,
+};