/*
* LZX is a LZ77 and Huffman-code based compression format that has many
* similarities to the DEFLATE format used in zlib. The compression ratio is as
- * good or better than DEFLATE. However, in WIM files only up to 32768 bytes of
- * data can ever compressed be in the same LZX block, so a .tar.gz file could
- * potentially be smaller than a WIM file that uses LZX compression because it
- * can use a larger LZ77 window size.
+ * good or better than DEFLATE.
*
* Some notes on the LZX compression format as used in Windows Imaging (WIM)
* files:
*
* A compressed WIM resource consists of a table of chunk offsets followed by
* the compressed chunks themselves. All compressed chunks except possibly the
- * last decompress to WIM_CHUNK_SIZE (= 32768) bytes. This is quite similar to
- * the cabinet (.cab) file format, but they are not the same. According to the
- * cabinet format documentation, the LZX block size is independent from the
- * CFDATA blocks, and a LZX block may span several CFDATA blocks. However, in
- * WIMs, LZX blocks do not appear to ever span multiple WIM chunks. Note that
- * this means any WIM chunk may be decompressed or compressed independently from
- * any other chunk, which is convenient.
+ * last decompress to a fixed number of bytes, by default 32768. This is quite
+ * similar to the cabinet (.cab) file format, but they are not the same.
+ * According to the cabinet format documentation, the LZX block size is
+ * independent from the CFDATA blocks, and a LZX block may span several CFDATA
+ * blocks. However, in WIMs, LZX blocks do not appear to ever span multiple WIM
+ * chunks. Note that this means any WIM chunk may be decompressed or compressed
+ * independently from any other chunk, which allows random access.
*
* A LZX compressed WIM chunk contains one or more LZX blocks of the aligned,
* verbatim, or uncompressed block types. For aligned and verbatim blocks, the
* size of the block in uncompressed bytes is specified by a bit following the 3
* bits that specify the block type, possibly followed by an additional 16 bits.
- * '1' means to use the default block size (equal to 32768, the size of a WIM
- * chunk--- and this seems to only be valid for the first LZX block in a WIM
- * chunk), while '0' means that the block size is provided by the next 16 bits.
+ * '1' means to use the default block size (equal to 32768, the default size of
+ * a WIM chunk), while '0' means that the block size is provided by the next 16
+ * bits.
*
* The cabinet format, as documented, allows for the possibility that a
* compressed CFDATA chunk is up to 6144 bytes larger than the data it
* defined in the specification.
*
* The LZX document states that aligned offset blocks have their aligned offset
- * huffman tree AFTER the main and length trees. The implementation suggests
+ * Huffman tree AFTER the main and length trees. The implementation suggests
* that the aligned offset tree is BEFORE the main and length trees.
*
* The LZX document decoding algorithm states that, in an aligned offset block,
* if an extra_bits value is 1, 2 or 3, then that number of bits should be read
* and the result added to the match offset. This is correct for 1 and 2, but
- * not 3, where just a huffman symbol (using the aligned tree) should be read.
+ * not 3, where just a Huffman symbol (using the aligned tree) should be read.
*
* Regarding the E8 preprocessing, the LZX document states 'No translation may
* be performed on the last 6 bytes of the input block'. This is correct.
* would cause the next four bytes to be modified, at least one of which would
* be in the last 6 bytes, which is not allowed according to the spec.
*
- * The specification states that the huffman trees must always contain at least
+ * The specification states that the Huffman trees must always contain at least
* one element. However, many CAB files contain blocks where the length tree is
* completely empty (because there are no matches), and this is expected to
* succeed.
#endif
#include "wimlib.h"
-#include "wimlib/decompress.h"
+#include "wimlib/decompressor_ops.h"
+#include "wimlib/decompress_common.h"
#include "wimlib/lzx.h"
#include "wimlib/util.h"
struct lzx_tables {
u16 maintree_decode_table[(1 << LZX_MAINCODE_TABLEBITS) +
- (LZX_MAINCODE_NUM_SYMBOLS * 2)]
+ (LZX_MAINCODE_MAX_NUM_SYMBOLS * 2)]
_aligned_attribute(DECODE_TABLE_ALIGNMENT);
- u8 maintree_lens[LZX_MAINCODE_NUM_SYMBOLS];
+ u8 maintree_lens[LZX_MAINCODE_MAX_NUM_SYMBOLS];
u16 lentree_decode_table[(1 << LZX_LENCODE_TABLEBITS) +
u8 alignedtree_lens[LZX_ALIGNEDCODE_NUM_SYMBOLS];
} _aligned_attribute(DECODE_TABLE_ALIGNMENT);
+struct lzx_decompressor {
+ u32 max_window_size;
+ unsigned num_main_syms;
+ struct lzx_tables tables;
+};
/*
* Reads a Huffman-encoded symbol using the pre-tree.
static inline int
read_huffsym_using_maintree(struct input_bitstream *istream,
const struct lzx_tables *tables,
- unsigned *n)
+ unsigned *n,
+ unsigned num_main_syms)
{
return read_huffsym(istream, tables->maintree_decode_table,
- tables->maintree_lens, LZX_MAINCODE_NUM_SYMBOLS,
+ tables->maintree_lens, num_main_syms,
LZX_MAINCODE_TABLEBITS, n, LZX_MAX_MAIN_CODEWORD_LEN);
}
* code length values from the input.
*
* @istream: The bit stream for the input. It is positioned on the beginning
- * of the pretree for the code length values.
+ * of the pretree for the code length values.
* @lens: An array that contains the length values from the previous time
- * the code lengths for this Huffman tree were read, or all
- * 0's if this is the first time.
+ * the code lengths for this Huffman tree were read, or all
+ * 0's if this is the first time.
* @num_lens: Number of length values to decode and return.
*
*/
_aligned_attribute(DECODE_TABLE_ALIGNMENT);
u8 pretree_lens[LZX_PRECODE_NUM_SYMBOLS];
unsigned i;
- unsigned len;
+ u32 len;
int ret;
/* Read the code lengths of the pretree codes. There are 20 lengths of
* the next lengths are all equal to the next symbol in the
* input. */
unsigned tree_code;
- unsigned num_zeroes;
+ u32 num_zeroes;
unsigned code;
- unsigned num_same;
+ u32 num_same;
signed char value;
ret = read_huffsym_using_pretree(istream, pretree_decode_table,
*
* @istream: The input bitstream.
* @block_size_ret: A pointer to an int into which the size of the block,
- * in bytes, will be returned.
+ * in bytes, will be returned.
* @block_type_ret: A pointer to an int into which the type of the block
- * (LZX_BLOCKTYPE_*) will be returned.
+ * (LZX_BLOCKTYPE_*) will be returned.
* @tables: A pointer to a lzx_tables structure in which the
- * main tree, the length tree, and possibly the
- * aligned offset tree will be constructed.
+ * main tree, the length tree, and possibly the
+ * aligned offset tree will be constructed.
* @queue: A pointer to the least-recently-used queue into which
- * R0, R1, and R2 will be written (only for uncompressed
- * blocks, which contain this information in the header)
+ * R0, R1, and R2 will be written (only for uncompressed
+ * blocks, which contain this information in the header)
*/
static int
lzx_read_block_header(struct input_bitstream *istream,
+ unsigned num_main_syms,
+ unsigned max_window_size,
unsigned *block_size_ret,
unsigned *block_type_ret,
struct lzx_tables *tables,
int ret;
unsigned block_type;
unsigned block_size;
- unsigned s;
- unsigned i;
- unsigned len;
- ret = bitstream_ensure_bits(istream, LZX_BLOCKTYPE_NBITS + 1);
- if (ret) {
- LZX_DEBUG("LZX input stream overrun");
+ ret = bitstream_ensure_bits(istream, 4);
+ if (ret)
return ret;
- }
/* The first three bits tell us what kind of block it is, and are one
* of the LZX_BLOCKTYPE_* values. */
- block_type = bitstream_read_bits_nocheck(istream, LZX_BLOCKTYPE_NBITS);
-
- /* The next bit indicates whether the block size is the default (32768),
- * indicated by a 1 bit, or whether the block size is given by the next
- * 16 bits, indicated by a 0 bit. */
- s = bitstream_read_bits_nocheck(istream, 1);
+ block_type = bitstream_read_bits_nocheck(istream, 3);
- if (s) {
- block_size = 32768;
+ /* Read the block size. This mirrors the behavior
+ * lzx_write_compressed_block() in lzx-compress.c; see that for more
+ * details. */
+ if (bitstream_read_bits_nocheck(istream, 1)) {
+ block_size = LZX_DEFAULT_BLOCK_SIZE;
} else {
- ret = bitstream_read_bits(istream, LZX_BLOCKSIZE_NBITS, &block_size);
+ u32 tmp;
+ block_size = 0;
+
+ ret = bitstream_read_bits(istream, 8, &tmp);
+ if (ret)
+ return ret;
+ block_size |= tmp;
+
+ ret = bitstream_read_bits(istream, 8, &tmp);
if (ret)
return ret;
- block_size = le16_to_cpu(block_size);
+ block_size <<= 8;
+ block_size |= tmp;
+
+ if (max_window_size >= 65536) {
+ ret = bitstream_read_bits(istream, 8, &tmp);
+ if (ret)
+ return ret;
+ block_size <<= 8;
+ block_size |= tmp;
+ }
}
switch (block_type) {
/* Read the path lengths for the elements of the aligned tree,
* then build it. */
- for (i = 0; i < LZX_ALIGNEDCODE_NUM_SYMBOLS; i++) {
+ for (unsigned i = 0; i < LZX_ALIGNEDCODE_NUM_SYMBOLS; i++) {
+ u32 len;
+
ret = bitstream_read_bits(istream,
LZX_ALIGNEDCODE_ELEMENT_SIZE,
&len);
* tree. */
LZX_DEBUG("Reading path lengths for remaining elements of "
"main tree (%d elements).",
- LZX_MAINCODE_NUM_SYMBOLS - LZX_NUM_CHARS);
+ num_main_syms - LZX_NUM_CHARS);
ret = lzx_read_code_lens(istream,
tables->maintree_lens + LZX_NUM_CHARS,
- LZX_MAINCODE_NUM_SYMBOLS - LZX_NUM_CHARS);
+ num_main_syms - LZX_NUM_CHARS);
if (ret) {
LZX_DEBUG("Failed to read the path lengths for the "
"remaining elements of the main tree");
"table for the main tree.");
ret = make_huffman_decode_table(tables->maintree_decode_table,
- LZX_MAINCODE_NUM_SYMBOLS,
+ num_main_syms,
LZX_MAINCODE_TABLEBITS,
tables->maintree_lens,
LZX_MAX_MAIN_CODEWORD_LEN);
istream->bitsleft = 0;
istream->bitbuf = 0;
}
- queue->R[0] = le32_to_cpu(*(u32*)(istream->data + 0));
- queue->R[1] = le32_to_cpu(*(u32*)(istream->data + 4));
- queue->R[2] = le32_to_cpu(*(u32*)(istream->data + 8));
+ queue->R[0] = le32_to_cpu(*(le32*)(istream->data + 0));
+ queue->R[1] = le32_to_cpu(*(le32*)(istream->data + 4));
+ queue->R[2] = le32_to_cpu(*(le32*)(istream->data + 8));
istream->data += 12;
istream->data_bytes_left -= 12;
/* The uncompressed data of this block directly follows and will
* tree.
*
* @block_type: The type of the block (LZX_BLOCKTYPE_ALIGNED or
- * LZX_BLOCKTYPE_VERBATIM)
+ * LZX_BLOCKTYPE_VERBATIM)
*
* @bytes_remaining: The amount of uncompressed data remaining to be
- * uncompressed in this block. It is an error if the match
- * is longer than this number.
+ * uncompressed in this block. It is an error if the match
+ * is longer than this number.
*
* @window: A pointer to the window into which the uncompressed
- * data is being written.
+ * data is being written.
*
* @window_pos: The current byte offset in the window.
*
*
* Returns the length of the match, or a negative number on error. The possible
* error cases are:
- * - Match would exceed the amount of data remaining to be uncompressed.
- * - Match refers to data before the window.
- * - The input bitstream ended unexpectedly.
+ * - Match would exceed the amount of data remaining to be uncompressed.
+ * - Match refers to data before the window.
+ * - The input bitstream ended unexpectedly.
*/
static int
lzx_decode_match(unsigned main_element, int block_type,
unsigned match_offset;
unsigned additional_len;
unsigned num_extra_bits;
- unsigned verbatim_bits;
- unsigned aligned_bits;
+ u32 verbatim_bits;
+ u32 aligned_bits;
unsigned i;
int ret;
u8 *match_dest;
}
static void
-undo_call_insn_translation(u32 *call_insn_target, int input_pos,
+undo_call_insn_translation(u32 *call_insn_target, s32 input_pos,
s32 file_size)
{
s32 abs_offset;
* as it is used in calculating the translated jump targets. But in WIM files,
* this file size is always the same (LZX_WIM_MAGIC_FILESIZE == 12000000).*/
static void
-undo_call_insn_preprocessing(u8 uncompressed_data[], int uncompressed_data_len)
+undo_call_insn_preprocessing(u8 *uncompressed_data, s32 uncompressed_size)
{
- for (int i = 0; i < uncompressed_data_len - 10; i++) {
+ for (s32 i = 0; i < uncompressed_size - 10; i++) {
if (uncompressed_data[i] == 0xe8) {
undo_call_insn_translation((u32*)&uncompressed_data[i + 1],
i,
* been read.
*
* @block_type: The type of the block (LZX_BLOCKTYPE_VERBATIM or
- * LZX_BLOCKTYPE_ALIGNED)
+ * LZX_BLOCKTYPE_ALIGNED)
* @block_size: The size of the block, in bytes.
+ * @num_main_syms: Number of symbols in the main alphabet.
* @window: Pointer to the decompression window.
* @window_pos: The current position in the window. Will be 0 for the first
- * block.
+ * block.
* @tables: The Huffman decoding tables for the block (main, length, and
- * aligned offset, the latter only for LZX_BLOCKTYPE_ALIGNED)
+ * aligned offset, the latter only for LZX_BLOCKTYPE_ALIGNED)
* @queue: The least-recently-used queue for match offsets.
* @istream: The input bitstream for the compressed literals.
*/
static int
lzx_decompress_block(int block_type, unsigned block_size,
+ unsigned num_main_syms,
u8 *window,
unsigned window_pos,
const struct lzx_tables *tables,
end = window_pos + block_size;
while (window_pos < end) {
ret = read_huffsym_using_maintree(istream, tables,
- &main_element);
+ &main_element,
+ num_main_syms);
if (ret)
return ret;
/* literal: 0 to LZX_NUM_CHARS - 1 */
window[window_pos++] = main_element;
} else {
- /* match: LZX_NUM_CHARS to LZX_MAINCODE_NUM_SYMBOLS - 1 */
+ /* match: LZX_NUM_CHARS to num_main_syms - 1 */
match_len = lzx_decode_match(main_element,
block_type,
end - window_pos,
return 0;
}
-/* API function documented in wimlib.h */
-WIMLIBAPI int
-wimlib_lzx_decompress(const void *compressed_data, unsigned compressed_len,
- void *uncompressed_data, unsigned uncompressed_len)
+static int
+lzx_decompress(const void *compressed_data, size_t compressed_size,
+ void *uncompressed_data, size_t uncompressed_size,
+ void *_ctx)
{
- struct lzx_tables tables;
+ struct lzx_decompressor *ctx = _ctx;
struct input_bitstream istream;
struct lzx_lru_queue queue;
unsigned window_pos;
int ret;
bool e8_preprocessing_done;
- LZX_DEBUG("lzx_decompress (compressed_data = %p, compressed_len = %d, "
- "uncompressed_data = %p, uncompressed_len = %d).",
- compressed_data, compressed_len,
- uncompressed_data, uncompressed_len);
-
- wimlib_assert(uncompressed_len <= 32768);
+ LZX_DEBUG("compressed_data = %p, compressed_size = %zu, "
+ "uncompressed_data = %p, uncompressed_size = %zu, "
+ "max_window_size=%u).",
+ compressed_data, compressed_size,
+ uncompressed_data, uncompressed_size,
+ ctx->max_window_size);
+
+ if (uncompressed_size > ctx->max_window_size) {
+ LZX_DEBUG("Uncompressed size of %zu exceeds "
+ "window size of %u!",
+ uncompressed_size, ctx->max_window_size);
+ return -1;
+ }
- memset(tables.maintree_lens, 0, sizeof(tables.maintree_lens));
- memset(tables.lentree_lens, 0, sizeof(tables.lentree_lens));
+ memset(ctx->tables.maintree_lens, 0, sizeof(ctx->tables.maintree_lens));
+ memset(ctx->tables.lentree_lens, 0, sizeof(ctx->tables.lentree_lens));
lzx_lru_queue_init(&queue);
- init_input_bitstream(&istream, compressed_data, compressed_len);
+ init_input_bitstream(&istream, compressed_data, compressed_size);
e8_preprocessing_done = false; /* Set to true if there may be 0xe8 bytes
in the uncompressed data. */
* blocks. */
for (window_pos = 0;
- window_pos < uncompressed_len;
+ window_pos < uncompressed_size;
window_pos += block_size)
{
LZX_DEBUG("Reading block header.");
- ret = lzx_read_block_header(&istream, &block_size,
- &block_type, &tables, &queue);
+ ret = lzx_read_block_header(&istream, ctx->num_main_syms,
+ ctx->max_window_size, &block_size,
+ &block_type, &ctx->tables, &queue);
if (ret)
return ret;
LZX_DEBUG("block_size = %u, window_pos = %u",
block_size, window_pos);
- if (block_size > uncompressed_len - window_pos) {
+ if (block_size > uncompressed_size - window_pos) {
LZX_DEBUG("Expected a block size of at "
- "most %u bytes (found %u bytes)",
- uncompressed_len - window_pos, block_size);
+ "most %zu bytes (found %u bytes)",
+ uncompressed_size - window_pos, block_size);
return -1;
}
LZX_DEBUG("LZX_BLOCKTYPE_ALIGNED");
ret = lzx_decompress_block(block_type,
block_size,
+ ctx->num_main_syms,
uncompressed_data,
window_pos,
- &tables,
+ &ctx->tables,
&queue,
&istream);
if (ret)
return ret;
- if (tables.maintree_lens[0xe8] != 0)
+
+ if (ctx->tables.maintree_lens[0xe8] != 0)
e8_preprocessing_done = true;
break;
case LZX_BLOCKTYPE_UNCOMPRESSED:
}
}
if (e8_preprocessing_done)
- undo_call_insn_preprocessing(uncompressed_data, uncompressed_len);
+ undo_call_insn_preprocessing(uncompressed_data, uncompressed_size);
return 0;
}
+
+static void
+lzx_free_decompressor(void *_ctx)
+{
+ struct lzx_decompressor *ctx = _ctx;
+
+ FREE(ctx);
+}
+
+static int
+lzx_create_decompressor(size_t max_window_size,
+ const struct wimlib_decompressor_params_header *params,
+ void **ctx_ret)
+{
+ struct lzx_decompressor *ctx;
+
+ if (!lzx_window_size_valid(max_window_size))
+ return WIMLIB_ERR_INVALID_PARAM;
+
+ ctx = MALLOC(sizeof(struct lzx_decompressor));
+ if (ctx == NULL)
+ return WIMLIB_ERR_NOMEM;
+
+ ctx->max_window_size = max_window_size;
+ ctx->num_main_syms = lzx_get_num_main_syms(max_window_size);
+
+ *ctx_ret = ctx;
+ return 0;
+}
+
+const struct decompressor_ops lzx_decompressor_ops = {
+ .create_decompressor = lzx_create_decompressor,
+ .decompress = lzx_decompress,
+ .free_decompressor = lzx_free_decompressor,
+};