4 #include "wimlib/assert.h"
5 #include "wimlib/types.h"
7 //#define ENABLE_LZX_DEBUG
8 #ifdef ENABLE_LZX_DEBUG
9 # define LZX_DEBUG DEBUG
10 # define LZX_ASSERT wimlib_assert
12 # define LZX_DEBUG(format, ...)
13 # define LZX_ASSERT(...)
16 /* Constants, most of which are defined by the LZX specification: */
18 /* The smallest and largest allowed match lengths. */
19 #define LZX_MIN_MATCH_LEN 2
20 #define LZX_MAX_MATCH_LEN 257
22 /* Number of values an uncompressed literal byte can represent. */
23 #define LZX_NUM_CHARS 256
25 /* Each LZX block begins with 3 bits that determines the block type. Below are
26 * the valid block types. Values 0, and 4 through 7, are invalid. */
27 #define LZX_BLOCKTYPE_VERBATIM 1
28 #define LZX_BLOCKTYPE_ALIGNED 2
29 #define LZX_BLOCKTYPE_UNCOMPRESSED 3
31 #define LZX_NUM_PRIMARY_LENS 7 /* this one missing from spec! */
33 /* NOTE: There are really 51 position slots in the LZX format as a whole, but
34 * only 30 are needed to allow for the window to be up to 32768 bytes long,
35 * which is the maximum in the WIM format. */
36 #define LZX_NUM_POSITION_SLOTS 30
38 /* Read the LZX specification for information about the Huffman trees used in
39 * the LZX compression format. Basically there are 4 of them: The main tree,
40 * the length tree, the pre tree, and the aligned tree. The main tree and
41 * length tree are given at the beginning of VERBATIM and ALIGNED blocks as a
42 * list of *_NUM_SYMBOLS code length values. They are read using the
43 * read_code_lens() function and built using the make_decode_table() function.
44 * The decode table is not a real tree but rather a table that we can index by
45 * some number of bits (*_TABLEBITS) of the input to quickly look up the symbol
46 * corresponding to a Huffman code.
48 * The ALIGNED tree is only present on ALIGNED blocks.
50 * A PRECODE is used to encode the code lengths for the main tree and the length
51 * tree. There is a separate pretree for each half of the main tree. */
53 #define LZX_MAINCODE_NUM_SYMBOLS (LZX_NUM_CHARS + \
54 (LZX_NUM_POSITION_SLOTS << 3))
55 #define LZX_MAINCODE_TABLEBITS 11
57 #define LZX_LENCODE_NUM_SYMBOLS 249
58 #define LZX_LENCODE_TABLEBITS 10
60 #define LZX_PRECODE_NUM_SYMBOLS 20
61 #define LZX_PRECODE_TABLEBITS 6
62 #define LZX_PRECODE_ELEMENT_SIZE 4
64 #define LZX_ALIGNEDCODE_NUM_SYMBOLS 8
65 #define LZX_ALIGNEDCODE_TABLEBITS 7
66 #define LZX_ALIGNEDCODE_ELEMENT_SIZE 3
68 /* Maximum allowed length of Huffman codewords. */
69 #define LZX_MAX_MAIN_CODEWORD_LEN 16
70 #define LZX_MAX_LEN_CODEWORD_LEN 16
71 #define LZX_MAX_PRE_CODEWORD_LEN 16
72 #define LZX_MAX_ALIGNED_CODEWORD_LEN 8
74 /* For the LZX-compressed blocks in WIM files, this value is always used as the
75 * filesize parameter for the call instruction (0xe8 byte) preprocessing, even
76 * though the blocks themselves are not this size, and the size of the actual
77 * file resource in the WIM file is very likely to be something entirely
78 * different as well. */
79 #define LZX_WIM_MAGIC_FILESIZE 12000000
81 #define LZX_BLOCKTYPE_NBITS 3
82 #define LZX_BLOCKSIZE_NBITS 16
84 #define USE_LZX_EXTRA_BITS_ARRAY
86 #ifdef USE_LZX_EXTRA_BITS_ARRAY
87 extern const u8 lzx_extra_bits[];
90 /* Given the number of a LZX position slot, return the number of extra bits that
91 * are needed to encode the match offset. */
92 static inline unsigned
93 lzx_get_num_extra_bits(unsigned position_slot)
95 #ifdef USE_LZX_EXTRA_BITS_ARRAY
97 return lzx_extra_bits[position_slot];
99 /* Calculate directly using a shift and subtraction. */
100 LZX_ASSERT(position_slot >= 2 && position_slot <= 37);
101 return (position_slot >> 1) - 1;
105 extern const u32 lzx_position_base[];
107 #define LZX_NUM_RECENT_OFFSETS 3
109 /* Least-recently used queue for match offsets. */
110 struct lzx_lru_queue {
111 u32 R[LZX_NUM_RECENT_OFFSETS];
114 /* In the LZX format, an offset of n bytes is actually encoded
115 * as (n + LZX_OFFSET_OFFSET). */
116 #define LZX_OFFSET_OFFSET (LZX_NUM_RECENT_OFFSETS - 1)
119 lzx_lru_queue_init(struct lzx_lru_queue *queue)
121 for (unsigned i = 0; i < LZX_NUM_RECENT_OFFSETS; i++)
125 #endif /* _WIMLIB_LZX_H */