*
* Also, a caveat--- according to Microsoft's documentation for XPRESS,
*
- * "Some implementation of the decompression algorithm expect an extra
- * symbol to mark the end of the data. Specifically, some implementations
- * fail during decompression if the Huffman symbol 256 is not found after
- * the actual data."
+ * "Some implementation of the decompression algorithm expect an extra
+ * symbol to mark the end of the data. Specifically, some implementations
+ * fail during decompression if the Huffman symbol 256 is not found after
+ * the actual data."
*
* This is the case for the implementation in WIMGAPI. However, wimlib's
* decompressor in this file currently does not care if this extra symbol is
#endif
#include "wimlib.h"
-#include "wimlib/decompress.h"
+#include "wimlib/decompressor_ops.h"
+#include "wimlib/decompress_common.h"
#include "wimlib/xpress.h"
/*
* Returns the match length, or -1 if the data is invalid.
*/
static int
-xpress_decode_match(unsigned sym, input_idx_t window_pos,
- input_idx_t window_len, u8 window[restrict],
+xpress_decode_match(unsigned sym, u32 window_pos,
+ u32 window_len, u8 window[restrict],
struct input_bitstream * restrict istream)
{
-
- u8 len_hdr;
- u8 offset_bsr;
- int ret;
- u8 *match_dest;
- u8 *match_src;
- unsigned i;
+ unsigned len_hdr;
+ unsigned offset_bsr;
unsigned match_len;
unsigned match_offset;
len_hdr = sym & 0xf;
offset_bsr = sym >> 4;
- if (bitstream_ensure_bits(istream, 16))
- return -1;
+ bitstream_ensure_bits(istream, 16);
match_offset = (1U << offset_bsr) | bitstream_pop_bits(istream, offset_bsr);
if (len_hdr == 0xf) {
- ret = bitstream_read_byte(istream);
- if (ret < 0)
- return ret;
- match_len = ret;
+ match_len = bitstream_read_byte(istream);
if (unlikely(match_len == 0xff)) {
- ret = bitstream_read_byte(istream);
- if (ret < 0)
- return ret;
- match_len = ret;
-
- ret = bitstream_read_byte(istream);
- if (ret < 0)
- return ret;
-
- match_len |= (ret << 8);
+ match_len = bitstream_read_byte(istream);
+ match_len |= (unsigned)bitstream_read_byte(istream) << 8;
} else {
match_len += 0xf;
}
}
match_len += XPRESS_MIN_MATCH_LEN;
-
- /* Verify the match is in bounds, then copy its data to the the current
- * position. */
-
- if (window_pos + match_len > window_len)
+ if (unlikely(match_len > window_len - window_pos))
return -1;
- if (match_offset > window_pos)
+ if (unlikely(match_offset > window_pos))
return -1;
- match_dest = window + window_pos;
- match_src = match_dest - match_offset;
-
- for (i = 0; i < match_len; i++)
- match_dest[i] = match_src[i];
+ lz_copy(&window[window_pos], match_len, match_offset,
+ &window[window_len]);
return match_len;
}
xpress_lz_decode(struct input_bitstream * restrict istream,
u8 uncompressed_data[restrict],
unsigned uncompressed_len,
- const u8 lens[restrict],
const u16 decode_table[restrict])
{
- input_idx_t curpos;
+ u32 curpos;
unsigned match_len;
for (curpos = 0; curpos < uncompressed_len; curpos += match_len) {
unsigned sym;
int ret;
- if (unlikely(bitstream_ensure_bits(istream, 16)))
- return -1;
-
- if (unlikely(read_huffsym(istream, decode_table, lens,
- XPRESS_NUM_SYMBOLS, XPRESS_TABLEBITS,
- &sym, XPRESS_MAX_CODEWORD_LEN)))
- return -1;
+ bitstream_ensure_bits(istream, 16);
+ sym = read_huffsym(istream, decode_table,
+ XPRESS_TABLEBITS, XPRESS_MAX_CODEWORD_LEN);
if (sym < XPRESS_NUM_CHARS) {
/* Literal */
uncompressed_data[curpos] = sym;
}
-/* API function documented in wimlib.h */
-WIMLIBAPI int
-wimlib_xpress_decompress(const void * const restrict _compressed_data,
- const unsigned compressed_len,
- void * const restrict uncompressed_data,
- const unsigned uncompressed_len)
+static int
+xpress_decompress(const void *compressed_data, size_t compressed_size,
+ void *uncompressed_data, size_t uncompressed_size, void *_ctx)
{
- const u8 *compressed_data = _compressed_data;
+ const u8 *cdata = compressed_data;
u8 lens[XPRESS_NUM_SYMBOLS];
u8 *lens_p;
u16 decode_table[(1 << XPRESS_TABLEBITS) + 2 * XPRESS_NUM_SYMBOLS]
/* XPRESS uses only one Huffman code. It contains 512 symbols, and the
* code lengths of these symbols are given literally as 4-bit integers
* in the first 256 bytes of the compressed data. */
- if (compressed_len < XPRESS_NUM_SYMBOLS / 2)
+ if (compressed_size < XPRESS_NUM_SYMBOLS / 2)
return -1;
lens_p = lens;
for (unsigned i = 0; i < XPRESS_NUM_SYMBOLS / 2; i++) {
- *lens_p++ = compressed_data[i] & 0xf;
- *lens_p++ = compressed_data[i] >> 4;
+ *lens_p++ = cdata[i] & 0xf;
+ *lens_p++ = cdata[i] >> 4;
}
if (make_huffman_decode_table(decode_table, XPRESS_NUM_SYMBOLS,
XPRESS_MAX_CODEWORD_LEN))
return -1;
- init_input_bitstream(&istream, compressed_data + XPRESS_NUM_SYMBOLS / 2,
- compressed_len - XPRESS_NUM_SYMBOLS / 2);
+ init_input_bitstream(&istream, cdata + XPRESS_NUM_SYMBOLS / 2,
+ compressed_size - XPRESS_NUM_SYMBOLS / 2);
return xpress_lz_decode(&istream, uncompressed_data,
- uncompressed_len, lens, decode_table);
+ uncompressed_size, decode_table);
}
+
+const struct decompressor_ops xpress_decompressor_ops = {
+ .decompress = xpress_decompress,
+};
git://wimlib.net / wimlib / blobdiff