1 ///////////////////////////////////////////////////////////////////////////////
3 /// \file lzma_encoder_optimum_fast.c
7 // This file has been put into the public domain.
8 // You can do whatever you want with this file.
10 ///////////////////////////////////////////////////////////////////////////////
12 #include "lzma_encoder_private.h"
13 #include "memcmplen.h"
16 #define change_pair(small_dist, big_dist) \
17 (((big_dist) >> 7) > (small_dist))
21 lzma_lzma_optimum_fast(lzma_lzma1_encoder *restrict coder,
23 uint32_t *restrict back_res, uint32_t *restrict len_res)
25 const uint32_t nice_len = mf->nice_len;
28 uint32_t matches_count;
29 if (mf->read_ahead == 0) {
30 len_main = mf_find(mf, &matches_count, coder->matches);
32 assert(mf->read_ahead == 1);
33 len_main = coder->longest_match_length;
34 matches_count = coder->matches_count;
37 const uint8_t *buf = mf_ptr(mf) - 1;
38 const uint32_t buf_avail = my_min(mf_avail(mf) + 1, MATCH_LEN_MAX);
41 // There's not enough input left to encode a match.
42 *back_res = UINT32_MAX;
47 // Look for repeated matches; scan the previous four match distances
49 uint32_t rep_index = 0;
51 for (uint32_t i = 0; i < REPS; ++i) {
52 // Pointer to the beginning of the match candidate
53 const uint8_t *const buf_back = buf - coder->reps[i] - 1;
55 // If the first two bytes (2 == MATCH_LEN_MIN) do not match,
56 // this rep is not useful.
57 if (not_equal_16(buf, buf_back))
60 // The first two bytes matched.
61 // Calculate the length of the match.
62 const uint32_t len = lzma_memcmplen(
63 buf, buf_back, 2, buf_avail);
65 // If we have found a repeated match that is at least
66 // nice_len long, return it immediately.
67 if (len >= nice_len) {
80 // We didn't find a long enough repeated match. Encode it as a normal
81 // match if the match length is at least nice_len.
82 if (len_main >= nice_len) {
83 *back_res = coder->matches[matches_count - 1].dist + REPS;
85 mf_skip(mf, len_main - 1);
89 uint32_t back_main = 0;
91 back_main = coder->matches[matches_count - 1].dist;
93 while (matches_count > 1 && len_main ==
94 coder->matches[matches_count - 2].len + 1) {
95 if (!change_pair(coder->matches[
96 matches_count - 2].dist,
101 len_main = coder->matches[matches_count - 1].len;
102 back_main = coder->matches[matches_count - 1].dist;
105 if (len_main == 2 && back_main >= 0x80)
110 if (rep_len + 1 >= len_main
111 || (rep_len + 2 >= len_main
112 && back_main > (UINT32_C(1) << 9))
113 || (rep_len + 3 >= len_main
114 && back_main > (UINT32_C(1) << 15))) {
115 *back_res = rep_index;
117 mf_skip(mf, rep_len - 1);
122 if (len_main < 2 || buf_avail <= 2) {
123 *back_res = UINT32_MAX;
128 // Get the matches for the next byte. If we find a better match,
129 // the current byte is encoded as a literal.
130 coder->longest_match_length = mf_find(mf,
131 &coder->matches_count, coder->matches);
133 if (coder->longest_match_length >= 2) {
134 const uint32_t new_dist = coder->matches[
135 coder->matches_count - 1].dist;
137 if ((coder->longest_match_length >= len_main
138 && new_dist < back_main)
139 || (coder->longest_match_length == len_main + 1
140 && !change_pair(back_main, new_dist))
141 || (coder->longest_match_length > len_main + 1)
142 || (coder->longest_match_length + 1 >= len_main
144 && change_pair(new_dist, back_main))) {
145 *back_res = UINT32_MAX;
151 // In contrast to LZMA SDK, dictionary could not have been moved
152 // between mf_find() calls, thus it is safe to just increment
153 // the old buf pointer instead of recalculating it with mf_ptr().
156 const uint32_t limit = my_max(2, len_main - 1);
158 for (uint32_t i = 0; i < REPS; ++i) {
159 if (memcmp(buf, buf - coder->reps[i] - 1, limit) == 0) {
160 *back_res = UINT32_MAX;
166 *back_res = back_main + REPS;
168 mf_skip(mf, len_main - 2);