4 * Header for decompression code shared by multiple compression formats.
7 #ifndef _WIMLIB_DECOMPRESS_COMMON_H
8 #define _WIMLIB_DECOMPRESS_COMMON_H
10 #include "wimlib/assert.h"
11 #include "wimlib/compiler.h"
12 #include "wimlib/endianness.h"
13 #include "wimlib/types.h"
15 #ifndef INPUT_IDX_T_DEFINED
16 #define INPUT_IDX_T_DEFINED
17 typedef u32 input_idx_t;
20 /* Structure to encapsulate a block of in-memory data that is being interpreted
21 * as a stream of bits.
23 * This is geared specifically towards the XPRESS and LZX compression formats
24 * with regards to the actual ordering the bits within the byte sequence. */
25 struct input_bitstream {
27 /* A variable of length at least 32 bits that is used to hold bits that
28 * have been read from the stream. The bits are ordered from high-order
29 * to low-order, and the next bit is always the high-order bit. */
32 /* Number of bits in @bitbuf that are valid. */
35 /* Pointer to the next byte to be retrieved from the input. */
38 /* Number of bytes of data that are left. */
39 input_idx_t data_bytes_left;
42 /* Initializes a bitstream to receive its input from @data. */
44 init_input_bitstream(struct input_bitstream *istream,
45 const void *data, input_idx_t num_data_bytes)
48 istream->bitsleft = 0;
50 istream->data_bytes_left = num_data_bytes;
53 /* Ensures the bit buffer variable for the bitstream contains at least @num_bits
54 * bits. Following this, bitstream_peek_bits() and/or bitstream_remove_bits()
55 * may be called on the bitstream to peek or remove up to @num_bits bits.
57 * If the input data is exhausted, any further bits are assumed to be 0. */
59 bitstream_ensure_bits(struct input_bitstream *istream, unsigned num_bits)
61 for (int nbits = num_bits; (int)istream->bitsleft < nbits; nbits -= 16) {
65 if (unlikely(istream->data_bytes_left < 2)) {
66 istream->bitsleft = num_bits;
70 nextword = le16_to_cpu(*(const le16*)istream->data);
71 shift = sizeof(istream->bitbuf) * 8 - 16 - istream->bitsleft;
72 istream->bitbuf |= (u32)nextword << shift;
74 istream->bitsleft += 16;
75 istream->data_bytes_left -= 2;
79 /* Returns the next @num_bits bits from the bitstream, without removing them.
80 * There must be at least @num_bits remaining in the buffer variable, from a
81 * previous call to bitstream_ensure_bits(). */
83 bitstream_peek_bits(const struct input_bitstream *istream, unsigned num_bits)
85 if (unlikely(num_bits == 0))
87 return istream->bitbuf >> (sizeof(istream->bitbuf) * 8 - num_bits);
90 /* Removes @num_bits from the bitstream. There must be at least @num_bits
91 * remaining in the buffer variable, from a previous call to
92 * bitstream_ensure_bits(). */
94 bitstream_remove_bits(struct input_bitstream *istream, unsigned num_bits)
96 istream->bitbuf <<= num_bits;
97 istream->bitsleft -= num_bits;
100 /* Removes and returns @num_bits bits from the bitstream. There must be at
101 * least @num_bits remaining in the buffer variable, from a previous call to
102 * bitstream_ensure_bits(). */
104 bitstream_pop_bits(struct input_bitstream *istream, unsigned num_bits)
106 u32 n = bitstream_peek_bits(istream, num_bits);
107 bitstream_remove_bits(istream, num_bits);
111 /* Reads and returns the next @num_bits bits from the bitstream.
112 * If the input data is exhausted, the bits are assumed to be 0. */
114 bitstream_read_bits(struct input_bitstream *istream, unsigned num_bits)
116 bitstream_ensure_bits(istream, num_bits);
117 return bitstream_pop_bits(istream, num_bits);
120 /* Reads and returns the next literal byte embedded in the bitstream.
121 * If the input data is exhausted, the byte is assumed to be 0. */
123 bitstream_read_byte(struct input_bitstream *istream)
125 if (unlikely(istream->data_bytes_left == 0))
127 istream->data_bytes_left--;
128 return *istream->data++;
132 /* Needed alignment of decode_table parameter to make_huffman_decode_table().
134 * Reason: We may fill the entries with SSE instructions without worrying
135 * about dealing with the unaligned case. */
136 #define DECODE_TABLE_ALIGNMENT 16
138 /* Maximum supported symbol count for make_huffman_decode_table().
140 * Reason: In direct mapping entries, we store the symbol in 11 bits. */
141 #define DECODE_TABLE_MAX_SYMBOLS 2048
143 /* Maximum supported table bits for make_huffman_decode_table().
145 * Reason: In internal binary tree nodes, offsets are encoded in 14 bits.
146 * But the real limit is 13, because we allocate entries past the end of
147 * the direct lookup part of the table for binary tree nodes. (Note: if
148 * needed this limit could be removed by encoding the offsets relative to
149 * &decode_table[1 << table_bits].) */
150 #define DECODE_TABLE_MAX_TABLE_BITS 13
152 /* Maximum supported codeword length for make_huffman_decode_table().
154 * Reason: In direct mapping entries, we encode the codeword length in 5
155 * bits, and the top 2 bits can't both be set because that has special
157 #define DECODE_TABLE_MAX_CODEWORD_LEN 23
159 /* Reads and returns the next Huffman-encoded symbol from a bitstream. If the
160 * input data is exhausted, the Huffman symbol is decoded as if the missing bits
163 * XXX: This is mostly duplicated in lzms_huffman_decode_symbol() in
164 * lzms-decompress.c. */
166 read_huffsym(struct input_bitstream *istream, const u16 decode_table[],
167 unsigned table_bits, unsigned max_codeword_len)
172 bitstream_ensure_bits(istream, max_codeword_len);
174 /* Index the decode table by the next table_bits bits of the input. */
175 key_bits = bitstream_peek_bits(istream, table_bits);
176 entry = decode_table[key_bits];
177 if (likely(entry < 0xC000)) {
178 /* Fast case: The decode table directly provided the
179 * symbol and codeword length. The low 11 bits are the
180 * symbol, and the high 5 bits are the codeword length. */
181 bitstream_remove_bits(istream, entry >> 11);
182 return entry & 0x7FF;
184 /* Slow case: The codeword for the symbol is longer than
185 * table_bits, so the symbol does not have an entry
186 * directly in the first (1 << table_bits) entries of the
187 * decode table. Traverse the appropriate binary tree
188 * bit-by-bit to decode the symbol. */
189 bitstream_remove_bits(istream, table_bits);
191 key_bits = (entry & 0x3FFF) + bitstream_pop_bits(istream, 1);
192 } while ((entry = decode_table[key_bits]) >= 0xC000);
198 make_huffman_decode_table(u16 decode_table[], unsigned num_syms,
199 unsigned num_bits, const u8 lens[],
200 unsigned max_codeword_len);
202 #endif /* _WIMLIB_DECOMPRESS_COMMON_H */