isomorfeus-ferret 0.12.6 → 0.13.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/LICENSE +101 -19
- data/README.md +85 -16
- data/ext/isomorfeus_ferret_ext/bm_bitvector.c +22 -30
- data/ext/isomorfeus_ferret_ext/bm_hash.c +6 -12
- data/ext/isomorfeus_ferret_ext/bm_micro_string.c +3 -6
- data/ext/isomorfeus_ferret_ext/bm_store.c +11 -22
- data/ext/isomorfeus_ferret_ext/brotli_common_constants.c +15 -0
- data/ext/isomorfeus_ferret_ext/brotli_common_constants.h +200 -0
- data/ext/isomorfeus_ferret_ext/brotli_common_context.c +156 -0
- data/ext/isomorfeus_ferret_ext/brotli_common_context.h +113 -0
- data/ext/isomorfeus_ferret_ext/brotli_common_dictionary.c +5914 -0
- data/ext/isomorfeus_ferret_ext/brotli_common_dictionary.h +64 -0
- data/ext/isomorfeus_ferret_ext/brotli_common_platform.c +22 -0
- data/ext/isomorfeus_ferret_ext/brotli_common_platform.h +594 -0
- data/ext/isomorfeus_ferret_ext/brotli_common_transform.c +291 -0
- data/ext/isomorfeus_ferret_ext/brotli_common_transform.h +85 -0
- data/ext/isomorfeus_ferret_ext/brotli_common_version.h +26 -0
- data/ext/isomorfeus_ferret_ext/brotli_dec_bit_reader.c +76 -0
- data/ext/isomorfeus_ferret_ext/brotli_dec_bit_reader.h +351 -0
- data/ext/isomorfeus_ferret_ext/brotli_dec_decode.c +2608 -0
- data/ext/isomorfeus_ferret_ext/brotli_dec_huffman.c +339 -0
- data/ext/isomorfeus_ferret_ext/brotli_dec_huffman.h +121 -0
- data/ext/isomorfeus_ferret_ext/brotli_dec_prefix.h +732 -0
- data/ext/isomorfeus_ferret_ext/brotli_dec_state.c +159 -0
- data/ext/isomorfeus_ferret_ext/brotli_dec_state.h +365 -0
- data/ext/isomorfeus_ferret_ext/brotli_decode.h +344 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_backward_references.c +145 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_backward_references.h +39 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_backward_references_hq.c +843 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_backward_references_hq.h +95 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_backward_references_inc.h +163 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_bit_cost.c +35 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_bit_cost.h +63 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_bit_cost_inc.h +127 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_block_encoder_inc.h +34 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_block_splitter.c +194 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_block_splitter.h +51 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_block_splitter_inc.h +440 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_brotli_bit_stream.c +1314 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_brotli_bit_stream.h +84 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_cluster.c +56 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_cluster.h +48 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_cluster_inc.h +320 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_command.c +28 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_command.h +190 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_compress_fragment.c +790 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_compress_fragment.h +61 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_compress_fragment_two_pass.c +645 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_compress_fragment_two_pass.h +54 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_dictionary_hash.c +1846 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_dictionary_hash.h +25 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_encode.c +1927 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_encoder_dict.c +33 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_encoder_dict.h +43 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_entropy_encode.c +503 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_entropy_encode.h +122 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_entropy_encode_static.h +539 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_fast_log.c +105 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_fast_log.h +66 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_find_match_length.h +79 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_hash.h +488 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_hash_composite_inc.h +125 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_hash_forgetful_chain_inc.h +293 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_hash_longest_match64_inc.h +267 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_hash_longest_match_inc.h +262 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_hash_longest_match_quickly_inc.h +266 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_hash_rolling_inc.h +212 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_hash_to_binary_tree_inc.h +329 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_histogram.c +100 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_histogram.h +63 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_histogram_inc.h +51 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_literal_cost.c +175 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_literal_cost.h +30 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_memory.c +170 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_memory.h +114 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_metablock.c +663 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_metablock.h +105 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_metablock_inc.h +183 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_params.h +46 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_prefix.h +53 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_quality.h +165 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_ringbuffer.h +167 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_static_dict.c +486 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_static_dict.h +40 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_static_dict_lut.h +5864 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_utf8_util.c +85 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_utf8_util.h +32 -0
- data/ext/isomorfeus_ferret_ext/brotli_enc_write_bits.h +87 -0
- data/ext/isomorfeus_ferret_ext/brotli_encode.h +448 -0
- data/ext/isomorfeus_ferret_ext/brotli_port.h +288 -0
- data/ext/isomorfeus_ferret_ext/brotli_types.h +83 -0
- data/ext/isomorfeus_ferret_ext/bzlib.c +1572 -0
- data/ext/isomorfeus_ferret_ext/bzlib.h +282 -0
- data/ext/isomorfeus_ferret_ext/bzlib_blocksort.c +1094 -0
- data/ext/isomorfeus_ferret_ext/bzlib_compress.c +672 -0
- data/ext/isomorfeus_ferret_ext/bzlib_crctable.c +104 -0
- data/ext/isomorfeus_ferret_ext/bzlib_decompress.c +652 -0
- data/ext/isomorfeus_ferret_ext/bzlib_huffman.c +205 -0
- data/ext/isomorfeus_ferret_ext/bzlib_private.h +509 -0
- data/ext/isomorfeus_ferret_ext/bzlib_randtable.c +84 -0
- data/ext/isomorfeus_ferret_ext/fio_tmpfile.h +53 -53
- data/ext/isomorfeus_ferret_ext/frb_analysis.c +785 -1192
- data/ext/isomorfeus_ferret_ext/frb_index.c +513 -464
- data/ext/isomorfeus_ferret_ext/frb_qparser.c +48 -60
- data/ext/isomorfeus_ferret_ext/frb_search.c +1520 -1002
- data/ext/isomorfeus_ferret_ext/frb_store.c +96 -96
- data/ext/isomorfeus_ferret_ext/frb_threading.h +0 -1
- data/ext/isomorfeus_ferret_ext/frb_utils.c +147 -196
- data/ext/isomorfeus_ferret_ext/frt_analysis.c +695 -1090
- data/ext/isomorfeus_ferret_ext/frt_analysis.h +174 -170
- data/ext/isomorfeus_ferret_ext/frt_array.c +2 -4
- data/ext/isomorfeus_ferret_ext/frt_bitvector.c +9 -16
- data/ext/isomorfeus_ferret_ext/frt_bitvector.h +32 -81
- data/ext/isomorfeus_ferret_ext/frt_document.c +15 -20
- data/ext/isomorfeus_ferret_ext/frt_document.h +10 -9
- data/ext/isomorfeus_ferret_ext/frt_except.c +5 -12
- data/ext/isomorfeus_ferret_ext/frt_field_index.c +3 -3
- data/ext/isomorfeus_ferret_ext/frt_field_index.h +6 -7
- data/ext/isomorfeus_ferret_ext/frt_filter.c +35 -46
- data/ext/isomorfeus_ferret_ext/frt_fs_store.c +2 -0
- data/ext/isomorfeus_ferret_ext/frt_global.c +91 -200
- data/ext/isomorfeus_ferret_ext/frt_global.h +7 -18
- data/ext/isomorfeus_ferret_ext/frt_hash.c +1 -2
- data/ext/isomorfeus_ferret_ext/frt_ind.c +32 -35
- data/ext/isomorfeus_ferret_ext/frt_ind.h +9 -9
- data/ext/isomorfeus_ferret_ext/frt_index.c +714 -384
- data/ext/isomorfeus_ferret_ext/frt_index.h +274 -290
- data/ext/isomorfeus_ferret_ext/frt_lang.c +0 -2
- data/ext/isomorfeus_ferret_ext/frt_mempool.c +1 -2
- data/ext/isomorfeus_ferret_ext/frt_multimapper.c +4 -7
- data/ext/isomorfeus_ferret_ext/frt_q_boolean.c +68 -91
- data/ext/isomorfeus_ferret_ext/frt_q_const_score.c +35 -38
- data/ext/isomorfeus_ferret_ext/frt_q_filtered_query.c +53 -72
- data/ext/isomorfeus_ferret_ext/frt_q_fuzzy.c +25 -32
- data/ext/isomorfeus_ferret_ext/frt_q_match_all.c +21 -23
- data/ext/isomorfeus_ferret_ext/frt_q_multi_term.c +66 -103
- data/ext/isomorfeus_ferret_ext/frt_q_parser.c +207 -195
- data/ext/isomorfeus_ferret_ext/frt_q_phrase.c +20 -16
- data/ext/isomorfeus_ferret_ext/frt_q_prefix.c +17 -14
- data/ext/isomorfeus_ferret_ext/frt_q_range.c +102 -131
- data/ext/isomorfeus_ferret_ext/frt_q_span.c +179 -178
- data/ext/isomorfeus_ferret_ext/frt_q_term.c +47 -60
- data/ext/isomorfeus_ferret_ext/frt_q_wildcard.c +18 -16
- data/ext/isomorfeus_ferret_ext/frt_ram_store.c +46 -84
- data/ext/isomorfeus_ferret_ext/frt_search.c +105 -146
- data/ext/isomorfeus_ferret_ext/frt_search.h +331 -320
- data/ext/isomorfeus_ferret_ext/frt_similarity.c +5 -13
- data/ext/isomorfeus_ferret_ext/frt_similarity.h +7 -12
- data/ext/isomorfeus_ferret_ext/frt_sort.c +105 -149
- data/ext/isomorfeus_ferret_ext/frt_store.c +13 -7
- data/ext/isomorfeus_ferret_ext/frt_store.h +10 -2
- data/ext/isomorfeus_ferret_ext/frt_threading.h +0 -1
- data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.c +22 -112
- data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.h +2 -32
- data/ext/isomorfeus_ferret_ext/lz4.c +2495 -0
- data/ext/isomorfeus_ferret_ext/lz4.h +774 -0
- data/ext/isomorfeus_ferret_ext/lz4frame.c +1899 -0
- data/ext/isomorfeus_ferret_ext/lz4frame.h +623 -0
- data/ext/isomorfeus_ferret_ext/lz4hc.c +1615 -0
- data/ext/isomorfeus_ferret_ext/lz4hc.h +413 -0
- data/ext/isomorfeus_ferret_ext/lz4xxhash.c +1030 -0
- data/ext/isomorfeus_ferret_ext/lz4xxhash.h +328 -0
- data/ext/isomorfeus_ferret_ext/stem_modules.h +0 -86
- data/ext/isomorfeus_ferret_ext/test.c +0 -17
- data/ext/isomorfeus_ferret_ext/test_1710.c +11 -12
- data/ext/isomorfeus_ferret_ext/test_analysis.c +590 -583
- data/ext/isomorfeus_ferret_ext/test_compound_io.c +1 -1
- data/ext/isomorfeus_ferret_ext/test_document.c +19 -15
- data/ext/isomorfeus_ferret_ext/test_except.c +1 -2
- data/ext/isomorfeus_ferret_ext/test_fields.c +111 -100
- data/ext/isomorfeus_ferret_ext/test_file_deleter.c +10 -27
- data/ext/isomorfeus_ferret_ext/test_filter.c +11 -8
- data/ext/isomorfeus_ferret_ext/test_global.c +0 -46
- data/ext/isomorfeus_ferret_ext/test_hash.c +2 -2
- data/ext/isomorfeus_ferret_ext/test_hashset.c +1 -1
- data/ext/isomorfeus_ferret_ext/test_highlighter.c +15 -11
- data/ext/isomorfeus_ferret_ext/test_index.c +373 -363
- data/ext/isomorfeus_ferret_ext/test_q_const_score.c +5 -3
- data/ext/isomorfeus_ferret_ext/test_q_filtered.c +5 -3
- data/ext/isomorfeus_ferret_ext/test_q_fuzzy.c +13 -10
- data/ext/isomorfeus_ferret_ext/test_q_parser.c +45 -7
- data/ext/isomorfeus_ferret_ext/test_q_span.c +15 -12
- data/ext/isomorfeus_ferret_ext/test_ram_store.c +3 -3
- data/ext/isomorfeus_ferret_ext/test_search.c +60 -64
- data/ext/isomorfeus_ferret_ext/test_segments.c +5 -4
- data/ext/isomorfeus_ferret_ext/test_sort.c +17 -14
- data/ext/isomorfeus_ferret_ext/test_store.c +2 -0
- data/ext/isomorfeus_ferret_ext/test_term.c +3 -1
- data/ext/isomorfeus_ferret_ext/test_term_vectors.c +9 -10
- data/ext/isomorfeus_ferret_ext/test_test.c +1 -2
- data/ext/isomorfeus_ferret_ext/test_threading.c +9 -10
- data/ext/isomorfeus_ferret_ext/testhelper.c +1 -2
- data/lib/isomorfeus/ferret/version.rb +1 -1
- metadata +113 -58
- data/ext/isomorfeus_ferret_ext/email.rl +0 -21
- data/ext/isomorfeus_ferret_ext/frt_scanner.c +0 -900
- data/ext/isomorfeus_ferret_ext/frt_scanner.h +0 -28
- data/ext/isomorfeus_ferret_ext/frt_scanner_mb.c +0 -6706
- data/ext/isomorfeus_ferret_ext/frt_scanner_utf8.c +0 -4420
- data/ext/isomorfeus_ferret_ext/scanner.h +0 -28
- data/ext/isomorfeus_ferret_ext/scanner.in +0 -43
- data/ext/isomorfeus_ferret_ext/scanner.rl +0 -84
- data/ext/isomorfeus_ferret_ext/scanner_mb.rl +0 -200
- data/ext/isomorfeus_ferret_ext/scanner_utf8.rl +0 -85
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_basque.c +0 -1167
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_basque.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_catalan.c +0 -1433
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_catalan.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_danish.c +0 -301
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_danish.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_dutch.c +0 -590
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_dutch.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_english.c +0 -1049
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_english.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_finnish.c +0 -705
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_finnish.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_french.c +0 -1239
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_french.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_german.c +0 -477
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_german.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_hungarian.c +0 -1217
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_hungarian.h +0 -7
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_indonesian.c +0 -394
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_indonesian.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_irish.c +0 -457
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_irish.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_italian.c +0 -1009
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_italian.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_norwegian.c +0 -259
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_norwegian.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_porter.c +0 -704
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_porter.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_portuguese.c +0 -948
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_portuguese.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_spanish.c +0 -1028
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_spanish.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_swedish.c +0 -275
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_swedish.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_hungarian.c +0 -849
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_hungarian.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_romanian.c +0 -952
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_romanian.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_KOI8_R_russian.c +0 -669
- data/ext/isomorfeus_ferret_ext/stem_KOI8_R_russian.h +0 -6
- data/ext/isomorfeus_ferret_ext/stem_modules.txt +0 -63
- data/ext/isomorfeus_ferret_ext/uchar-ucs4.rl +0 -1854
- data/ext/isomorfeus_ferret_ext/uchar-utf8.rl +0 -1999
- data/ext/isomorfeus_ferret_ext/url.rl +0 -27
@@ -0,0 +1,125 @@
|
|
1
|
+
/* NOLINT(build/header_guard) */
|
2
|
+
/* Copyright 2018 Google Inc. All Rights Reserved.
|
3
|
+
|
4
|
+
Distributed under MIT license.
|
5
|
+
See file LICENSE for detail or copy at https://opensource.org/licenses/MIT
|
6
|
+
*/
|
7
|
+
|
8
|
+
/* template parameters: FN, HASHER_A, HASHER_B */
|
9
|
+
|
10
|
+
/* Composite hasher: This hasher allows to combine two other hashers, HASHER_A
|
11
|
+
and HASHER_B. */
|
12
|
+
|
13
|
+
#define HashComposite HASHER()
|
14
|
+
|
15
|
+
#define FN_A(X) EXPAND_CAT(X, HASHER_A)
|
16
|
+
#define FN_B(X) EXPAND_CAT(X, HASHER_B)
|
17
|
+
|
18
|
+
static BROTLI_INLINE size_t FN(HashTypeLength)(void) {
|
19
|
+
size_t a = FN_A(HashTypeLength)();
|
20
|
+
size_t b = FN_B(HashTypeLength)();
|
21
|
+
return a > b ? a : b;
|
22
|
+
}
|
23
|
+
|
24
|
+
static BROTLI_INLINE size_t FN(StoreLookahead)(void) {
|
25
|
+
size_t a = FN_A(StoreLookahead)();
|
26
|
+
size_t b = FN_B(StoreLookahead)();
|
27
|
+
return a > b ? a : b;
|
28
|
+
}
|
29
|
+
|
30
|
+
typedef struct HashComposite {
|
31
|
+
HASHER_A ha;
|
32
|
+
HASHER_B hb;
|
33
|
+
HasherCommon hb_common;
|
34
|
+
|
35
|
+
/* Shortcuts. */
|
36
|
+
void* extra;
|
37
|
+
HasherCommon* common;
|
38
|
+
|
39
|
+
BROTLI_BOOL fresh;
|
40
|
+
const BrotliEncoderParams* params;
|
41
|
+
} HashComposite;
|
42
|
+
|
43
|
+
static void FN(Initialize)(HasherCommon* common,
|
44
|
+
HashComposite* BROTLI_RESTRICT self, const BrotliEncoderParams* params) {
|
45
|
+
self->common = common;
|
46
|
+
self->extra = common->extra;
|
47
|
+
|
48
|
+
self->hb_common = *self->common;
|
49
|
+
self->fresh = BROTLI_TRUE;
|
50
|
+
self->params = params;
|
51
|
+
/* TODO: Initialize of the hashers is defered to Prepare (and params
|
52
|
+
remembered here) because we don't get the one_shot and input_size params
|
53
|
+
here that are needed to know the memory size of them. Instead provide
|
54
|
+
those params to all hashers FN(Initialize) */
|
55
|
+
}
|
56
|
+
|
57
|
+
static void FN(Prepare)(
|
58
|
+
HashComposite* BROTLI_RESTRICT self, BROTLI_BOOL one_shot,
|
59
|
+
size_t input_size, const uint8_t* BROTLI_RESTRICT data) {
|
60
|
+
if (self->fresh) {
|
61
|
+
self->fresh = BROTLI_FALSE;
|
62
|
+
self->hb_common.extra = (uint8_t*)self->extra +
|
63
|
+
FN_A(HashMemAllocInBytes)(self->params, one_shot, input_size);
|
64
|
+
|
65
|
+
FN_A(Initialize)(self->common, &self->ha, self->params);
|
66
|
+
FN_B(Initialize)(&self->hb_common, &self->hb, self->params);
|
67
|
+
}
|
68
|
+
FN_A(Prepare)(&self->ha, one_shot, input_size, data);
|
69
|
+
FN_B(Prepare)(&self->hb, one_shot, input_size, data);
|
70
|
+
}
|
71
|
+
|
72
|
+
static BROTLI_INLINE size_t FN(HashMemAllocInBytes)(
|
73
|
+
const BrotliEncoderParams* params, BROTLI_BOOL one_shot,
|
74
|
+
size_t input_size) {
|
75
|
+
return FN_A(HashMemAllocInBytes)(params, one_shot, input_size) +
|
76
|
+
FN_B(HashMemAllocInBytes)(params, one_shot, input_size);
|
77
|
+
}
|
78
|
+
|
79
|
+
static BROTLI_INLINE void FN(Store)(HashComposite* BROTLI_RESTRICT self,
|
80
|
+
const uint8_t* BROTLI_RESTRICT data, const size_t mask, const size_t ix) {
|
81
|
+
FN_A(Store)(&self->ha, data, mask, ix);
|
82
|
+
FN_B(Store)(&self->hb, data, mask, ix);
|
83
|
+
}
|
84
|
+
|
85
|
+
static BROTLI_INLINE void FN(StoreRange)(
|
86
|
+
HashComposite* BROTLI_RESTRICT self, const uint8_t* BROTLI_RESTRICT data,
|
87
|
+
const size_t mask, const size_t ix_start,
|
88
|
+
const size_t ix_end) {
|
89
|
+
FN_A(StoreRange)(&self->ha, data, mask, ix_start, ix_end);
|
90
|
+
FN_B(StoreRange)(&self->hb, data, mask, ix_start, ix_end);
|
91
|
+
}
|
92
|
+
|
93
|
+
static BROTLI_INLINE void FN(StitchToPreviousBlock)(
|
94
|
+
HashComposite* BROTLI_RESTRICT self,
|
95
|
+
size_t num_bytes, size_t position, const uint8_t* ringbuffer,
|
96
|
+
size_t ring_buffer_mask) {
|
97
|
+
FN_A(StitchToPreviousBlock)(&self->ha, num_bytes, position,
|
98
|
+
ringbuffer, ring_buffer_mask);
|
99
|
+
FN_B(StitchToPreviousBlock)(&self->hb, num_bytes, position,
|
100
|
+
ringbuffer, ring_buffer_mask);
|
101
|
+
}
|
102
|
+
|
103
|
+
static BROTLI_INLINE void FN(PrepareDistanceCache)(
|
104
|
+
HashComposite* BROTLI_RESTRICT self, int* BROTLI_RESTRICT distance_cache) {
|
105
|
+
FN_A(PrepareDistanceCache)(&self->ha, distance_cache);
|
106
|
+
FN_B(PrepareDistanceCache)(&self->hb, distance_cache);
|
107
|
+
}
|
108
|
+
|
109
|
+
static BROTLI_INLINE void FN(FindLongestMatch)(
|
110
|
+
HashComposite* BROTLI_RESTRICT self,
|
111
|
+
const BrotliEncoderDictionary* dictionary,
|
112
|
+
const uint8_t* BROTLI_RESTRICT data, const size_t ring_buffer_mask,
|
113
|
+
const int* BROTLI_RESTRICT distance_cache, const size_t cur_ix,
|
114
|
+
const size_t max_length, const size_t max_backward,
|
115
|
+
const size_t dictionary_distance, const size_t max_distance,
|
116
|
+
HasherSearchResult* BROTLI_RESTRICT out) {
|
117
|
+
FN_A(FindLongestMatch)(&self->ha, dictionary, data, ring_buffer_mask,
|
118
|
+
distance_cache, cur_ix, max_length, max_backward, dictionary_distance,
|
119
|
+
max_distance, out);
|
120
|
+
FN_B(FindLongestMatch)(&self->hb, dictionary, data, ring_buffer_mask,
|
121
|
+
distance_cache, cur_ix, max_length, max_backward, dictionary_distance,
|
122
|
+
max_distance, out);
|
123
|
+
}
|
124
|
+
|
125
|
+
#undef HashComposite
|
@@ -0,0 +1,293 @@
|
|
1
|
+
/* NOLINT(build/header_guard) */
|
2
|
+
/* Copyright 2016 Google Inc. All Rights Reserved.
|
3
|
+
|
4
|
+
Distributed under MIT license.
|
5
|
+
See file LICENSE for detail or copy at https://opensource.org/licenses/MIT
|
6
|
+
*/
|
7
|
+
|
8
|
+
/* template parameters: FN, BUCKET_BITS, NUM_BANKS, BANK_BITS,
|
9
|
+
NUM_LAST_DISTANCES_TO_CHECK */
|
10
|
+
|
11
|
+
/* A (forgetful) hash table to the data seen by the compressor, to
|
12
|
+
help create backward references to previous data.
|
13
|
+
|
14
|
+
Hashes are stored in chains which are bucketed to groups. Group of chains
|
15
|
+
share a storage "bank". When more than "bank size" chain nodes are added,
|
16
|
+
oldest nodes are replaced; this way several chains may share a tail. */
|
17
|
+
|
18
|
+
#define HashForgetfulChain HASHER()
|
19
|
+
|
20
|
+
#define BANK_SIZE (1 << BANK_BITS)
|
21
|
+
|
22
|
+
/* Number of hash buckets. */
|
23
|
+
#define BUCKET_SIZE (1 << BUCKET_BITS)
|
24
|
+
|
25
|
+
#define CAPPED_CHAINS 0
|
26
|
+
|
27
|
+
static BROTLI_INLINE size_t FN(HashTypeLength)(void) { return 4; }
|
28
|
+
static BROTLI_INLINE size_t FN(StoreLookahead)(void) { return 4; }
|
29
|
+
|
30
|
+
/* HashBytes is the function that chooses the bucket to place the address in.*/
|
31
|
+
static BROTLI_INLINE size_t FN(HashBytes)(const uint8_t* BROTLI_RESTRICT data) {
|
32
|
+
const uint32_t h = BROTLI_UNALIGNED_LOAD32LE(data) * kHashMul32;
|
33
|
+
/* The higher bits contain more mixture from the multiplication,
|
34
|
+
so we take our results from there. */
|
35
|
+
return h >> (32 - BUCKET_BITS);
|
36
|
+
}
|
37
|
+
|
38
|
+
typedef struct FN(Slot) {
|
39
|
+
uint16_t delta;
|
40
|
+
uint16_t next;
|
41
|
+
} FN(Slot);
|
42
|
+
|
43
|
+
typedef struct FN(Bank) {
|
44
|
+
FN(Slot) slots[BANK_SIZE];
|
45
|
+
} FN(Bank);
|
46
|
+
|
47
|
+
typedef struct HashForgetfulChain {
|
48
|
+
uint16_t free_slot_idx[NUM_BANKS]; /* Up to 1KiB. Move to dynamic? */
|
49
|
+
size_t max_hops;
|
50
|
+
|
51
|
+
/* Shortcuts. */
|
52
|
+
void* extra;
|
53
|
+
HasherCommon* common;
|
54
|
+
|
55
|
+
/* --- Dynamic size members --- */
|
56
|
+
|
57
|
+
/* uint32_t addr[BUCKET_SIZE]; */
|
58
|
+
|
59
|
+
/* uint16_t head[BUCKET_SIZE]; */
|
60
|
+
|
61
|
+
/* Truncated hash used for quick rejection of "distance cache" candidates. */
|
62
|
+
/* uint8_t tiny_hash[65536];*/
|
63
|
+
|
64
|
+
/* FN(Bank) banks[NUM_BANKS]; */
|
65
|
+
} HashForgetfulChain;
|
66
|
+
|
67
|
+
static uint32_t* FN(Addr)(void* extra) {
|
68
|
+
return (uint32_t*)extra;
|
69
|
+
}
|
70
|
+
|
71
|
+
static uint16_t* FN(Head)(void* extra) {
|
72
|
+
return (uint16_t*)(&FN(Addr)(extra)[BUCKET_SIZE]);
|
73
|
+
}
|
74
|
+
|
75
|
+
static uint8_t* FN(TinyHash)(void* extra) {
|
76
|
+
return (uint8_t*)(&FN(Head)(extra)[BUCKET_SIZE]);
|
77
|
+
}
|
78
|
+
|
79
|
+
static FN(Bank)* FN(Banks)(void* extra) {
|
80
|
+
return (FN(Bank)*)(&FN(TinyHash)(extra)[65536]);
|
81
|
+
}
|
82
|
+
|
83
|
+
static void FN(Initialize)(
|
84
|
+
HasherCommon* common, HashForgetfulChain* BROTLI_RESTRICT self,
|
85
|
+
const BrotliEncoderParams* params) {
|
86
|
+
self->common = common;
|
87
|
+
self->extra = common->extra;
|
88
|
+
|
89
|
+
self->max_hops = (params->quality > 6 ? 7u : 8u) << (params->quality - 4);
|
90
|
+
}
|
91
|
+
|
92
|
+
static void FN(Prepare)(
|
93
|
+
HashForgetfulChain* BROTLI_RESTRICT self, BROTLI_BOOL one_shot,
|
94
|
+
size_t input_size, const uint8_t* BROTLI_RESTRICT data) {
|
95
|
+
uint32_t* BROTLI_RESTRICT addr = FN(Addr)(self->extra);
|
96
|
+
uint16_t* BROTLI_RESTRICT head = FN(Head)(self->extra);
|
97
|
+
uint8_t* BROTLI_RESTRICT tiny_hash = FN(TinyHash)(self->extra);
|
98
|
+
/* Partial preparation is 100 times slower (per socket). */
|
99
|
+
size_t partial_prepare_threshold = BUCKET_SIZE >> 6;
|
100
|
+
if (one_shot && input_size <= partial_prepare_threshold) {
|
101
|
+
size_t i;
|
102
|
+
for (i = 0; i < input_size; ++i) {
|
103
|
+
size_t bucket = FN(HashBytes)(&data[i]);
|
104
|
+
/* See InitEmpty comment. */
|
105
|
+
addr[bucket] = 0xCCCCCCCC;
|
106
|
+
head[bucket] = 0xCCCC;
|
107
|
+
}
|
108
|
+
} else {
|
109
|
+
/* Fill |addr| array with 0xCCCCCCCC value. Because of wrapping, position
|
110
|
+
processed by hasher never reaches 3GB + 64M; this makes all new chains
|
111
|
+
to be terminated after the first node. */
|
112
|
+
memset(addr, 0xCC, sizeof(uint32_t) * BUCKET_SIZE);
|
113
|
+
memset(head, 0, sizeof(uint16_t) * BUCKET_SIZE);
|
114
|
+
}
|
115
|
+
memset(tiny_hash, 0, sizeof(uint8_t) * 65536);
|
116
|
+
memset(self->free_slot_idx, 0, sizeof(self->free_slot_idx));
|
117
|
+
}
|
118
|
+
|
119
|
+
static BROTLI_INLINE size_t FN(HashMemAllocInBytes)(
|
120
|
+
const BrotliEncoderParams* params, BROTLI_BOOL one_shot,
|
121
|
+
size_t input_size) {
|
122
|
+
BROTLI_UNUSED(params);
|
123
|
+
BROTLI_UNUSED(one_shot);
|
124
|
+
BROTLI_UNUSED(input_size);
|
125
|
+
return sizeof(uint32_t) * BUCKET_SIZE + sizeof(uint16_t) * BUCKET_SIZE +
|
126
|
+
sizeof(uint8_t) * 65536 + sizeof(FN(Bank)) * NUM_BANKS;
|
127
|
+
}
|
128
|
+
|
129
|
+
/* Look at 4 bytes at &data[ix & mask]. Compute a hash from these, and prepend
|
130
|
+
node to corresponding chain; also update tiny_hash for current position. */
|
131
|
+
static BROTLI_INLINE void FN(Store)(HashForgetfulChain* BROTLI_RESTRICT self,
|
132
|
+
const uint8_t* BROTLI_RESTRICT data, const size_t mask, const size_t ix) {
|
133
|
+
uint32_t* BROTLI_RESTRICT addr = FN(Addr)(self->extra);
|
134
|
+
uint16_t* BROTLI_RESTRICT head = FN(Head)(self->extra);
|
135
|
+
uint8_t* BROTLI_RESTRICT tiny_hash = FN(TinyHash)(self->extra);
|
136
|
+
FN(Bank)* BROTLI_RESTRICT banks = FN(Banks)(self->extra);
|
137
|
+
const size_t key = FN(HashBytes)(&data[ix & mask]);
|
138
|
+
const size_t bank = key & (NUM_BANKS - 1);
|
139
|
+
const size_t idx = self->free_slot_idx[bank]++ & (BANK_SIZE - 1);
|
140
|
+
size_t delta = ix - addr[key];
|
141
|
+
tiny_hash[(uint16_t)ix] = (uint8_t)key;
|
142
|
+
if (delta > 0xFFFF) delta = CAPPED_CHAINS ? 0 : 0xFFFF;
|
143
|
+
banks[bank].slots[idx].delta = (uint16_t)delta;
|
144
|
+
banks[bank].slots[idx].next = head[key];
|
145
|
+
addr[key] = (uint32_t)ix;
|
146
|
+
head[key] = (uint16_t)idx;
|
147
|
+
}
|
148
|
+
|
149
|
+
static BROTLI_INLINE void FN(StoreRange)(
|
150
|
+
HashForgetfulChain* BROTLI_RESTRICT self,
|
151
|
+
const uint8_t* BROTLI_RESTRICT data, const size_t mask,
|
152
|
+
const size_t ix_start, const size_t ix_end) {
|
153
|
+
size_t i;
|
154
|
+
for (i = ix_start; i < ix_end; ++i) {
|
155
|
+
FN(Store)(self, data, mask, i);
|
156
|
+
}
|
157
|
+
}
|
158
|
+
|
159
|
+
static BROTLI_INLINE void FN(StitchToPreviousBlock)(
|
160
|
+
HashForgetfulChain* BROTLI_RESTRICT self,
|
161
|
+
size_t num_bytes, size_t position, const uint8_t* ringbuffer,
|
162
|
+
size_t ring_buffer_mask) {
|
163
|
+
if (num_bytes >= FN(HashTypeLength)() - 1 && position >= 3) {
|
164
|
+
/* Prepare the hashes for three last bytes of the last write.
|
165
|
+
These could not be calculated before, since they require knowledge
|
166
|
+
of both the previous and the current block. */
|
167
|
+
FN(Store)(self, ringbuffer, ring_buffer_mask, position - 3);
|
168
|
+
FN(Store)(self, ringbuffer, ring_buffer_mask, position - 2);
|
169
|
+
FN(Store)(self, ringbuffer, ring_buffer_mask, position - 1);
|
170
|
+
}
|
171
|
+
}
|
172
|
+
|
173
|
+
static BROTLI_INLINE void FN(PrepareDistanceCache)(
|
174
|
+
HashForgetfulChain* BROTLI_RESTRICT self,
|
175
|
+
int* BROTLI_RESTRICT distance_cache) {
|
176
|
+
BROTLI_UNUSED(self);
|
177
|
+
PrepareDistanceCache(distance_cache, NUM_LAST_DISTANCES_TO_CHECK);
|
178
|
+
}
|
179
|
+
|
180
|
+
/* Find a longest backward match of &data[cur_ix] up to the length of
|
181
|
+
max_length and stores the position cur_ix in the hash table.
|
182
|
+
|
183
|
+
REQUIRES: FN(PrepareDistanceCache) must be invoked for current distance cache
|
184
|
+
values; if this method is invoked repeatedly with the same distance
|
185
|
+
cache values, it is enough to invoke FN(PrepareDistanceCache) once.
|
186
|
+
|
187
|
+
Does not look for matches longer than max_length.
|
188
|
+
Does not look for matches further away than max_backward.
|
189
|
+
Writes the best match into |out|.
|
190
|
+
|out|->score is updated only if a better match is found. */
|
191
|
+
static BROTLI_INLINE void FN(FindLongestMatch)(
|
192
|
+
HashForgetfulChain* BROTLI_RESTRICT self,
|
193
|
+
const BrotliEncoderDictionary* dictionary,
|
194
|
+
const uint8_t* BROTLI_RESTRICT data, const size_t ring_buffer_mask,
|
195
|
+
const int* BROTLI_RESTRICT distance_cache,
|
196
|
+
const size_t cur_ix, const size_t max_length, const size_t max_backward,
|
197
|
+
const size_t dictionary_distance, const size_t max_distance,
|
198
|
+
HasherSearchResult* BROTLI_RESTRICT out) {
|
199
|
+
uint32_t* BROTLI_RESTRICT addr = FN(Addr)(self->extra);
|
200
|
+
uint16_t* BROTLI_RESTRICT head = FN(Head)(self->extra);
|
201
|
+
uint8_t* BROTLI_RESTRICT tiny_hashes = FN(TinyHash)(self->extra);
|
202
|
+
FN(Bank)* BROTLI_RESTRICT banks = FN(Banks)(self->extra);
|
203
|
+
const size_t cur_ix_masked = cur_ix & ring_buffer_mask;
|
204
|
+
/* Don't accept a short copy from far away. */
|
205
|
+
score_t min_score = out->score;
|
206
|
+
score_t best_score = out->score;
|
207
|
+
size_t best_len = out->len;
|
208
|
+
size_t i;
|
209
|
+
const size_t key = FN(HashBytes)(&data[cur_ix_masked]);
|
210
|
+
const uint8_t tiny_hash = (uint8_t)(key);
|
211
|
+
out->len = 0;
|
212
|
+
out->len_code_delta = 0;
|
213
|
+
/* Try last distance first. */
|
214
|
+
for (i = 0; i < NUM_LAST_DISTANCES_TO_CHECK; ++i) {
|
215
|
+
const size_t backward = (size_t)distance_cache[i];
|
216
|
+
size_t prev_ix = (cur_ix - backward);
|
217
|
+
/* For distance code 0 we want to consider 2-byte matches. */
|
218
|
+
if (i > 0 && tiny_hashes[(uint16_t)prev_ix] != tiny_hash) continue;
|
219
|
+
if (prev_ix >= cur_ix || backward > max_backward) {
|
220
|
+
continue;
|
221
|
+
}
|
222
|
+
prev_ix &= ring_buffer_mask;
|
223
|
+
{
|
224
|
+
const size_t len = FindMatchLengthWithLimit(&data[prev_ix],
|
225
|
+
&data[cur_ix_masked],
|
226
|
+
max_length);
|
227
|
+
if (len >= 2) {
|
228
|
+
score_t score = BackwardReferenceScoreUsingLastDistance(len);
|
229
|
+
if (best_score < score) {
|
230
|
+
if (i != 0) score -= BackwardReferencePenaltyUsingLastDistance(i);
|
231
|
+
if (best_score < score) {
|
232
|
+
best_score = score;
|
233
|
+
best_len = len;
|
234
|
+
out->len = best_len;
|
235
|
+
out->distance = backward;
|
236
|
+
out->score = best_score;
|
237
|
+
}
|
238
|
+
}
|
239
|
+
}
|
240
|
+
}
|
241
|
+
}
|
242
|
+
{
|
243
|
+
const size_t bank = key & (NUM_BANKS - 1);
|
244
|
+
size_t backward = 0;
|
245
|
+
size_t hops = self->max_hops;
|
246
|
+
size_t delta = cur_ix - addr[key];
|
247
|
+
size_t slot = head[key];
|
248
|
+
while (hops--) {
|
249
|
+
size_t prev_ix;
|
250
|
+
size_t last = slot;
|
251
|
+
backward += delta;
|
252
|
+
if (backward > max_backward || (CAPPED_CHAINS && !delta)) break;
|
253
|
+
prev_ix = (cur_ix - backward) & ring_buffer_mask;
|
254
|
+
slot = banks[bank].slots[last].next;
|
255
|
+
delta = banks[bank].slots[last].delta;
|
256
|
+
if (cur_ix_masked + best_len > ring_buffer_mask ||
|
257
|
+
prev_ix + best_len > ring_buffer_mask ||
|
258
|
+
data[cur_ix_masked + best_len] != data[prev_ix + best_len]) {
|
259
|
+
continue;
|
260
|
+
}
|
261
|
+
{
|
262
|
+
const size_t len = FindMatchLengthWithLimit(&data[prev_ix],
|
263
|
+
&data[cur_ix_masked],
|
264
|
+
max_length);
|
265
|
+
if (len >= 4) {
|
266
|
+
/* Comparing for >= 3 does not change the semantics, but just saves
|
267
|
+
for a few unnecessary binary logarithms in backward reference
|
268
|
+
score, since we are not interested in such short matches. */
|
269
|
+
score_t score = BackwardReferenceScore(len, backward);
|
270
|
+
if (best_score < score) {
|
271
|
+
best_score = score;
|
272
|
+
best_len = len;
|
273
|
+
out->len = best_len;
|
274
|
+
out->distance = backward;
|
275
|
+
out->score = best_score;
|
276
|
+
}
|
277
|
+
}
|
278
|
+
}
|
279
|
+
}
|
280
|
+
FN(Store)(self, data, ring_buffer_mask, cur_ix);
|
281
|
+
}
|
282
|
+
if (out->score == min_score) {
|
283
|
+
SearchInStaticDictionary(dictionary,
|
284
|
+
self->common, &data[cur_ix_masked], max_length, dictionary_distance,
|
285
|
+
max_distance, out, BROTLI_FALSE);
|
286
|
+
}
|
287
|
+
}
|
288
|
+
|
289
|
+
#undef BANK_SIZE
|
290
|
+
#undef BUCKET_SIZE
|
291
|
+
#undef CAPPED_CHAINS
|
292
|
+
|
293
|
+
#undef HashForgetfulChain
|
@@ -0,0 +1,267 @@
|
|
1
|
+
/* NOLINT(build/header_guard) */
|
2
|
+
/* Copyright 2010 Google Inc. All Rights Reserved.
|
3
|
+
|
4
|
+
Distributed under MIT license.
|
5
|
+
See file LICENSE for detail or copy at https://opensource.org/licenses/MIT
|
6
|
+
*/
|
7
|
+
|
8
|
+
/* template parameters: FN */
|
9
|
+
|
10
|
+
/* A (forgetful) hash table to the data seen by the compressor, to
|
11
|
+
help create backward references to previous data.
|
12
|
+
|
13
|
+
This is a hash map of fixed size (bucket_size_) to a ring buffer of
|
14
|
+
fixed size (block_size_). The ring buffer contains the last block_size_
|
15
|
+
index positions of the given hash key in the compressed data. */
|
16
|
+
|
17
|
+
#define HashLongestMatch HASHER()
|
18
|
+
|
19
|
+
static BROTLI_INLINE size_t FN(HashTypeLength)(void) { return 8; }
|
20
|
+
static BROTLI_INLINE size_t FN(StoreLookahead)(void) { return 8; }
|
21
|
+
|
22
|
+
/* HashBytes is the function that chooses the bucket to place the address in. */
|
23
|
+
static BROTLI_INLINE uint32_t FN(HashBytes)(const uint8_t* BROTLI_RESTRICT data,
|
24
|
+
const uint64_t mask,
|
25
|
+
const int shift) {
|
26
|
+
const uint64_t h = (BROTLI_UNALIGNED_LOAD64LE(data) & mask) * kHashMul64Long;
|
27
|
+
/* The higher bits contain more mixture from the multiplication,
|
28
|
+
so we take our results from there. */
|
29
|
+
return (uint32_t)(h >> shift);
|
30
|
+
}
|
31
|
+
|
32
|
+
typedef struct HashLongestMatch {
|
33
|
+
/* Number of hash buckets. */
|
34
|
+
size_t bucket_size_;
|
35
|
+
/* Only block_size_ newest backward references are kept,
|
36
|
+
and the older are forgotten. */
|
37
|
+
size_t block_size_;
|
38
|
+
/* Left-shift for computing hash bucket index from hash value. */
|
39
|
+
int hash_shift_;
|
40
|
+
/* Mask for selecting the next 4-8 bytes of input */
|
41
|
+
uint64_t hash_mask_;
|
42
|
+
/* Mask for accessing entries in a block (in a ring-buffer manner). */
|
43
|
+
uint32_t block_mask_;
|
44
|
+
|
45
|
+
int block_bits_;
|
46
|
+
int num_last_distances_to_check_;
|
47
|
+
|
48
|
+
/* Shortcuts. */
|
49
|
+
HasherCommon* common_;
|
50
|
+
|
51
|
+
/* --- Dynamic size members --- */
|
52
|
+
|
53
|
+
/* Number of entries in a particular bucket. */
|
54
|
+
uint16_t* num_; /* uint16_t[bucket_size]; */
|
55
|
+
|
56
|
+
/* Buckets containing block_size_ of backward references. */
|
57
|
+
uint32_t* buckets_; /* uint32_t[bucket_size * block_size]; */
|
58
|
+
} HashLongestMatch;
|
59
|
+
|
60
|
+
static void FN(Initialize)(
|
61
|
+
HasherCommon* common, HashLongestMatch* BROTLI_RESTRICT self,
|
62
|
+
const BrotliEncoderParams* params) {
|
63
|
+
self->common_ = common;
|
64
|
+
|
65
|
+
BROTLI_UNUSED(params);
|
66
|
+
self->hash_shift_ = 64 - common->params.bucket_bits;
|
67
|
+
self->hash_mask_ = (~((uint64_t)0U)) >> (64 - 8 * common->params.hash_len);
|
68
|
+
self->bucket_size_ = (size_t)1 << common->params.bucket_bits;
|
69
|
+
self->block_bits_ = common->params.block_bits;
|
70
|
+
self->block_size_ = (size_t)1 << common->params.block_bits;
|
71
|
+
self->block_mask_ = (uint32_t)(self->block_size_ - 1);
|
72
|
+
self->num_last_distances_to_check_ =
|
73
|
+
common->params.num_last_distances_to_check;
|
74
|
+
self->num_ = (uint16_t*)common->extra;
|
75
|
+
self->buckets_ = (uint32_t*)&self->num_[self->bucket_size_];
|
76
|
+
}
|
77
|
+
|
78
|
+
static void FN(Prepare)(
|
79
|
+
HashLongestMatch* BROTLI_RESTRICT self, BROTLI_BOOL one_shot,
|
80
|
+
size_t input_size, const uint8_t* BROTLI_RESTRICT data) {
|
81
|
+
uint16_t* BROTLI_RESTRICT num = self->num_;
|
82
|
+
/* Partial preparation is 100 times slower (per socket). */
|
83
|
+
size_t partial_prepare_threshold = self->bucket_size_ >> 6;
|
84
|
+
if (one_shot && input_size <= partial_prepare_threshold) {
|
85
|
+
size_t i;
|
86
|
+
for (i = 0; i < input_size; ++i) {
|
87
|
+
const uint32_t key = FN(HashBytes)(&data[i], self->hash_mask_,
|
88
|
+
self->hash_shift_);
|
89
|
+
num[key] = 0;
|
90
|
+
}
|
91
|
+
} else {
|
92
|
+
memset(num, 0, self->bucket_size_ * sizeof(num[0]));
|
93
|
+
}
|
94
|
+
}
|
95
|
+
|
96
|
+
static BROTLI_INLINE size_t FN(HashMemAllocInBytes)(
|
97
|
+
const BrotliEncoderParams* params, BROTLI_BOOL one_shot,
|
98
|
+
size_t input_size) {
|
99
|
+
size_t bucket_size = (size_t)1 << params->hasher.bucket_bits;
|
100
|
+
size_t block_size = (size_t)1 << params->hasher.block_bits;
|
101
|
+
BROTLI_UNUSED(one_shot);
|
102
|
+
BROTLI_UNUSED(input_size);
|
103
|
+
return sizeof(uint16_t) * bucket_size +
|
104
|
+
sizeof(uint32_t) * bucket_size * block_size;
|
105
|
+
}
|
106
|
+
|
107
|
+
/* Look at 4 bytes at &data[ix & mask].
|
108
|
+
Compute a hash from these, and store the value of ix at that position. */
|
109
|
+
static BROTLI_INLINE void FN(Store)(
|
110
|
+
HashLongestMatch* BROTLI_RESTRICT self, const uint8_t* BROTLI_RESTRICT data,
|
111
|
+
const size_t mask, const size_t ix) {
|
112
|
+
uint16_t* BROTLI_RESTRICT num = self->num_;
|
113
|
+
uint32_t* BROTLI_RESTRICT buckets = self->buckets_;
|
114
|
+
const uint32_t key = FN(HashBytes)(&data[ix & mask], self->hash_mask_,
|
115
|
+
self->hash_shift_);
|
116
|
+
const size_t minor_ix = num[key] & self->block_mask_;
|
117
|
+
const size_t offset = minor_ix + (key << self->block_bits_);
|
118
|
+
++num[key];
|
119
|
+
buckets[offset] = (uint32_t)ix;
|
120
|
+
}
|
121
|
+
|
122
|
+
static BROTLI_INLINE void FN(StoreRange)(HashLongestMatch* BROTLI_RESTRICT self,
|
123
|
+
const uint8_t* BROTLI_RESTRICT data, const size_t mask,
|
124
|
+
const size_t ix_start, const size_t ix_end) {
|
125
|
+
size_t i;
|
126
|
+
for (i = ix_start; i < ix_end; ++i) {
|
127
|
+
FN(Store)(self, data, mask, i);
|
128
|
+
}
|
129
|
+
}
|
130
|
+
|
131
|
+
static BROTLI_INLINE void FN(StitchToPreviousBlock)(
|
132
|
+
HashLongestMatch* BROTLI_RESTRICT self,
|
133
|
+
size_t num_bytes, size_t position, const uint8_t* ringbuffer,
|
134
|
+
size_t ringbuffer_mask) {
|
135
|
+
if (num_bytes >= FN(HashTypeLength)() - 1 && position >= 3) {
|
136
|
+
/* Prepare the hashes for three last bytes of the last write.
|
137
|
+
These could not be calculated before, since they require knowledge
|
138
|
+
of both the previous and the current block. */
|
139
|
+
FN(Store)(self, ringbuffer, ringbuffer_mask, position - 3);
|
140
|
+
FN(Store)(self, ringbuffer, ringbuffer_mask, position - 2);
|
141
|
+
FN(Store)(self, ringbuffer, ringbuffer_mask, position - 1);
|
142
|
+
}
|
143
|
+
}
|
144
|
+
|
145
|
+
static BROTLI_INLINE void FN(PrepareDistanceCache)(
|
146
|
+
HashLongestMatch* BROTLI_RESTRICT self,
|
147
|
+
int* BROTLI_RESTRICT distance_cache) {
|
148
|
+
PrepareDistanceCache(distance_cache, self->num_last_distances_to_check_);
|
149
|
+
}
|
150
|
+
|
151
|
+
/* Find a longest backward match of &data[cur_ix] up to the length of
|
152
|
+
max_length and stores the position cur_ix in the hash table.
|
153
|
+
|
154
|
+
REQUIRES: FN(PrepareDistanceCache) must be invoked for current distance cache
|
155
|
+
values; if this method is invoked repeatedly with the same distance
|
156
|
+
cache values, it is enough to invoke FN(PrepareDistanceCache) once.
|
157
|
+
|
158
|
+
Does not look for matches longer than max_length.
|
159
|
+
Does not look for matches further away than max_backward.
|
160
|
+
Writes the best match into |out|.
|
161
|
+
|out|->score is updated only if a better match is found. */
|
162
|
+
static BROTLI_INLINE void FN(FindLongestMatch)(
|
163
|
+
HashLongestMatch* BROTLI_RESTRICT self,
|
164
|
+
const BrotliEncoderDictionary* dictionary,
|
165
|
+
const uint8_t* BROTLI_RESTRICT data, const size_t ring_buffer_mask,
|
166
|
+
const int* BROTLI_RESTRICT distance_cache, const size_t cur_ix,
|
167
|
+
const size_t max_length, const size_t max_backward,
|
168
|
+
const size_t dictionary_distance, const size_t max_distance,
|
169
|
+
HasherSearchResult* BROTLI_RESTRICT out) {
|
170
|
+
uint16_t* BROTLI_RESTRICT num = self->num_;
|
171
|
+
uint32_t* BROTLI_RESTRICT buckets = self->buckets_;
|
172
|
+
const size_t cur_ix_masked = cur_ix & ring_buffer_mask;
|
173
|
+
/* Don't accept a short copy from far away. */
|
174
|
+
score_t min_score = out->score;
|
175
|
+
score_t best_score = out->score;
|
176
|
+
size_t best_len = out->len;
|
177
|
+
size_t i;
|
178
|
+
out->len = 0;
|
179
|
+
out->len_code_delta = 0;
|
180
|
+
/* Try last distance first. */
|
181
|
+
for (i = 0; i < (size_t)self->num_last_distances_to_check_; ++i) {
|
182
|
+
const size_t backward = (size_t)distance_cache[i];
|
183
|
+
size_t prev_ix = (size_t)(cur_ix - backward);
|
184
|
+
if (prev_ix >= cur_ix) {
|
185
|
+
continue;
|
186
|
+
}
|
187
|
+
if (BROTLI_PREDICT_FALSE(backward > max_backward)) {
|
188
|
+
continue;
|
189
|
+
}
|
190
|
+
prev_ix &= ring_buffer_mask;
|
191
|
+
|
192
|
+
if (cur_ix_masked + best_len > ring_buffer_mask ||
|
193
|
+
prev_ix + best_len > ring_buffer_mask ||
|
194
|
+
data[cur_ix_masked + best_len] != data[prev_ix + best_len]) {
|
195
|
+
continue;
|
196
|
+
}
|
197
|
+
{
|
198
|
+
const size_t len = FindMatchLengthWithLimit(&data[prev_ix],
|
199
|
+
&data[cur_ix_masked],
|
200
|
+
max_length);
|
201
|
+
if (len >= 3 || (len == 2 && i < 2)) {
|
202
|
+
/* Comparing for >= 2 does not change the semantics, but just saves for
|
203
|
+
a few unnecessary binary logarithms in backward reference score,
|
204
|
+
since we are not interested in such short matches. */
|
205
|
+
score_t score = BackwardReferenceScoreUsingLastDistance(len);
|
206
|
+
if (best_score < score) {
|
207
|
+
if (i != 0) score -= BackwardReferencePenaltyUsingLastDistance(i);
|
208
|
+
if (best_score < score) {
|
209
|
+
best_score = score;
|
210
|
+
best_len = len;
|
211
|
+
out->len = best_len;
|
212
|
+
out->distance = backward;
|
213
|
+
out->score = best_score;
|
214
|
+
}
|
215
|
+
}
|
216
|
+
}
|
217
|
+
}
|
218
|
+
}
|
219
|
+
{
|
220
|
+
const uint32_t key = FN(HashBytes)(
|
221
|
+
&data[cur_ix_masked], self->hash_mask_, self->hash_shift_);
|
222
|
+
uint32_t* BROTLI_RESTRICT bucket = &buckets[key << self->block_bits_];
|
223
|
+
const size_t down =
|
224
|
+
(num[key] > self->block_size_) ?
|
225
|
+
(num[key] - self->block_size_) : 0u;
|
226
|
+
for (i = num[key]; i > down;) {
|
227
|
+
size_t prev_ix = bucket[--i & self->block_mask_];
|
228
|
+
const size_t backward = cur_ix - prev_ix;
|
229
|
+
if (BROTLI_PREDICT_FALSE(backward > max_backward)) {
|
230
|
+
break;
|
231
|
+
}
|
232
|
+
prev_ix &= ring_buffer_mask;
|
233
|
+
if (cur_ix_masked + best_len > ring_buffer_mask ||
|
234
|
+
prev_ix + best_len > ring_buffer_mask ||
|
235
|
+
data[cur_ix_masked + best_len] != data[prev_ix + best_len]) {
|
236
|
+
continue;
|
237
|
+
}
|
238
|
+
{
|
239
|
+
const size_t len = FindMatchLengthWithLimit(&data[prev_ix],
|
240
|
+
&data[cur_ix_masked],
|
241
|
+
max_length);
|
242
|
+
if (len >= 4) {
|
243
|
+
/* Comparing for >= 3 does not change the semantics, but just saves
|
244
|
+
for a few unnecessary binary logarithms in backward reference
|
245
|
+
score, since we are not interested in such short matches. */
|
246
|
+
score_t score = BackwardReferenceScore(len, backward);
|
247
|
+
if (best_score < score) {
|
248
|
+
best_score = score;
|
249
|
+
best_len = len;
|
250
|
+
out->len = best_len;
|
251
|
+
out->distance = backward;
|
252
|
+
out->score = best_score;
|
253
|
+
}
|
254
|
+
}
|
255
|
+
}
|
256
|
+
}
|
257
|
+
bucket[num[key] & self->block_mask_] = (uint32_t)cur_ix;
|
258
|
+
++num[key];
|
259
|
+
}
|
260
|
+
if (min_score == out->score) {
|
261
|
+
SearchInStaticDictionary(dictionary,
|
262
|
+
self->common_, &data[cur_ix_masked], max_length, dictionary_distance,
|
263
|
+
max_distance, out, BROTLI_FALSE);
|
264
|
+
}
|
265
|
+
}
|
266
|
+
|
267
|
+
#undef HashLongestMatch
|