isomorfeus-ferret 0.12.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/LICENSE +612 -0
- data/README.md +44 -0
- data/ext/isomorfeus_ferret_ext/benchmark.c +223 -0
- data/ext/isomorfeus_ferret_ext/benchmark.h +45 -0
- data/ext/isomorfeus_ferret_ext/benchmarks_all.h +25 -0
- data/ext/isomorfeus_ferret_ext/bm_bitvector.c +123 -0
- data/ext/isomorfeus_ferret_ext/bm_hash.c +118 -0
- data/ext/isomorfeus_ferret_ext/bm_micro_string.c +40 -0
- data/ext/isomorfeus_ferret_ext/bm_store.c +93 -0
- data/ext/isomorfeus_ferret_ext/email.rl +21 -0
- data/ext/isomorfeus_ferret_ext/extconf.rb +5 -0
- data/ext/isomorfeus_ferret_ext/fio_tmpfile.h +53 -0
- data/ext/isomorfeus_ferret_ext/frb_analysis.c +2577 -0
- data/ext/isomorfeus_ferret_ext/frb_index.c +3457 -0
- data/ext/isomorfeus_ferret_ext/frb_lang.c +9 -0
- data/ext/isomorfeus_ferret_ext/frb_lang.h +17 -0
- data/ext/isomorfeus_ferret_ext/frb_qparser.c +629 -0
- data/ext/isomorfeus_ferret_ext/frb_search.c +4460 -0
- data/ext/isomorfeus_ferret_ext/frb_store.c +515 -0
- data/ext/isomorfeus_ferret_ext/frb_threading.h +30 -0
- data/ext/isomorfeus_ferret_ext/frb_utils.c +1127 -0
- data/ext/isomorfeus_ferret_ext/frt_analysis.c +1644 -0
- data/ext/isomorfeus_ferret_ext/frt_analysis.h +247 -0
- data/ext/isomorfeus_ferret_ext/frt_array.c +124 -0
- data/ext/isomorfeus_ferret_ext/frt_array.h +54 -0
- data/ext/isomorfeus_ferret_ext/frt_bitvector.c +95 -0
- data/ext/isomorfeus_ferret_ext/frt_bitvector.h +586 -0
- data/ext/isomorfeus_ferret_ext/frt_compound_io.c +374 -0
- data/ext/isomorfeus_ferret_ext/frt_config.h +44 -0
- data/ext/isomorfeus_ferret_ext/frt_document.c +134 -0
- data/ext/isomorfeus_ferret_ext/frt_document.h +52 -0
- data/ext/isomorfeus_ferret_ext/frt_except.c +95 -0
- data/ext/isomorfeus_ferret_ext/frt_except.h +188 -0
- data/ext/isomorfeus_ferret_ext/frt_field_index.c +233 -0
- data/ext/isomorfeus_ferret_ext/frt_field_index.h +42 -0
- data/ext/isomorfeus_ferret_ext/frt_filter.c +157 -0
- data/ext/isomorfeus_ferret_ext/frt_fs_store.c +502 -0
- data/ext/isomorfeus_ferret_ext/frt_global.c +427 -0
- data/ext/isomorfeus_ferret_ext/frt_global.h +290 -0
- data/ext/isomorfeus_ferret_ext/frt_hash.c +518 -0
- data/ext/isomorfeus_ferret_ext/frt_hash.h +466 -0
- data/ext/isomorfeus_ferret_ext/frt_hashset.c +191 -0
- data/ext/isomorfeus_ferret_ext/frt_hashset.h +206 -0
- data/ext/isomorfeus_ferret_ext/frt_helper.c +62 -0
- data/ext/isomorfeus_ferret_ext/frt_helper.h +13 -0
- data/ext/isomorfeus_ferret_ext/frt_ind.c +353 -0
- data/ext/isomorfeus_ferret_ext/frt_ind.h +54 -0
- data/ext/isomorfeus_ferret_ext/frt_index.c +6377 -0
- data/ext/isomorfeus_ferret_ext/frt_index.h +880 -0
- data/ext/isomorfeus_ferret_ext/frt_lang.c +104 -0
- data/ext/isomorfeus_ferret_ext/frt_lang.h +44 -0
- data/ext/isomorfeus_ferret_ext/frt_mempool.c +87 -0
- data/ext/isomorfeus_ferret_ext/frt_mempool.h +33 -0
- data/ext/isomorfeus_ferret_ext/frt_multimapper.c +349 -0
- data/ext/isomorfeus_ferret_ext/frt_multimapper.h +52 -0
- data/ext/isomorfeus_ferret_ext/frt_posh.c +1006 -0
- data/ext/isomorfeus_ferret_ext/frt_posh.h +973 -0
- data/ext/isomorfeus_ferret_ext/frt_priorityqueue.c +147 -0
- data/ext/isomorfeus_ferret_ext/frt_priorityqueue.h +147 -0
- data/ext/isomorfeus_ferret_ext/frt_q_boolean.c +1612 -0
- data/ext/isomorfeus_ferret_ext/frt_q_const_score.c +157 -0
- data/ext/isomorfeus_ferret_ext/frt_q_filtered_query.c +209 -0
- data/ext/isomorfeus_ferret_ext/frt_q_fuzzy.c +281 -0
- data/ext/isomorfeus_ferret_ext/frt_q_match_all.c +147 -0
- data/ext/isomorfeus_ferret_ext/frt_q_multi_term.c +672 -0
- data/ext/isomorfeus_ferret_ext/frt_q_parser.c +3084 -0
- data/ext/isomorfeus_ferret_ext/frt_q_phrase.c +1182 -0
- data/ext/isomorfeus_ferret_ext/frt_q_prefix.c +98 -0
- data/ext/isomorfeus_ferret_ext/frt_q_range.c +665 -0
- data/ext/isomorfeus_ferret_ext/frt_q_span.c +2386 -0
- data/ext/isomorfeus_ferret_ext/frt_q_term.c +311 -0
- data/ext/isomorfeus_ferret_ext/frt_q_wildcard.c +166 -0
- data/ext/isomorfeus_ferret_ext/frt_ram_store.c +460 -0
- data/ext/isomorfeus_ferret_ext/frt_scanner.c +899 -0
- data/ext/isomorfeus_ferret_ext/frt_scanner.h +28 -0
- data/ext/isomorfeus_ferret_ext/frt_scanner_mb.c +6705 -0
- data/ext/isomorfeus_ferret_ext/frt_scanner_utf8.c +4419 -0
- data/ext/isomorfeus_ferret_ext/frt_search.c +1824 -0
- data/ext/isomorfeus_ferret_ext/frt_search.h +924 -0
- data/ext/isomorfeus_ferret_ext/frt_similarity.c +150 -0
- data/ext/isomorfeus_ferret_ext/frt_similarity.h +79 -0
- data/ext/isomorfeus_ferret_ext/frt_sort.c +796 -0
- data/ext/isomorfeus_ferret_ext/frt_stopwords.c +395 -0
- data/ext/isomorfeus_ferret_ext/frt_store.c +680 -0
- data/ext/isomorfeus_ferret_ext/frt_store.h +789 -0
- data/ext/isomorfeus_ferret_ext/frt_term_vectors.c +72 -0
- data/ext/isomorfeus_ferret_ext/frt_threading.h +23 -0
- data/ext/isomorfeus_ferret_ext/frt_win32.h +54 -0
- data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.c +409 -0
- data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.h +95 -0
- data/ext/isomorfeus_ferret_ext/libstemmer.c +93 -0
- data/ext/isomorfeus_ferret_ext/libstemmer.h +73 -0
- data/ext/isomorfeus_ferret_ext/q_parser.y +1366 -0
- data/ext/isomorfeus_ferret_ext/scanner.h +28 -0
- data/ext/isomorfeus_ferret_ext/scanner.in +43 -0
- data/ext/isomorfeus_ferret_ext/scanner.rl +84 -0
- data/ext/isomorfeus_ferret_ext/scanner_mb.rl +200 -0
- data/ext/isomorfeus_ferret_ext/scanner_utf8.rl +85 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_danish.c +324 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_danish.h +7 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_dutch.c +610 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_dutch.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_english.c +1104 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_english.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_finnish.c +749 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_finnish.h +7 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_french.c +1233 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_french.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_german.c +490 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_german.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_hungarian.c +1217 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_hungarian.h +7 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_italian.c +1052 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_italian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_norwegian.c +283 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_norwegian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_porter.c +735 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_porter.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_portuguese.c +1003 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_portuguese.h +7 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_spanish.c +1079 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_spanish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_swedish.c +293 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_swedish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_romanian.c +984 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_romanian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_KOI8_R_russian.c +686 -0
- data/ext/isomorfeus_ferret_ext/stem_KOI8_R_russian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_danish.c +325 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_danish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_dutch.c +620 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_dutch.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_english.c +1111 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_english.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_finnish.c +754 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_finnish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_french.c +1242 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_french.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_german.c +495 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_german.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_hungarian.c +1220 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_hungarian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_italian.c +1059 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_italian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_norwegian.c +285 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_norwegian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_porter.c +741 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_porter.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_portuguese.c +1009 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_portuguese.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_romanian.c +990 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_romanian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_russian.c +680 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_russian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_spanish.c +1083 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_spanish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_swedish.c +294 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_swedish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_turkish.c +2191 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_turkish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_api.c +66 -0
- data/ext/isomorfeus_ferret_ext/stem_api.h +26 -0
- data/ext/isomorfeus_ferret_ext/stem_header.h +57 -0
- data/ext/isomorfeus_ferret_ext/stem_modules.h +190 -0
- data/ext/isomorfeus_ferret_ext/stem_modules.txt +50 -0
- data/ext/isomorfeus_ferret_ext/stem_utilities.c +478 -0
- data/ext/isomorfeus_ferret_ext/test.c +850 -0
- data/ext/isomorfeus_ferret_ext/test.h +416 -0
- data/ext/isomorfeus_ferret_ext/test_1710.c +63 -0
- data/ext/isomorfeus_ferret_ext/test_analysis.c +1221 -0
- data/ext/isomorfeus_ferret_ext/test_array.c +272 -0
- data/ext/isomorfeus_ferret_ext/test_bitvector.c +600 -0
- data/ext/isomorfeus_ferret_ext/test_compound_io.c +170 -0
- data/ext/isomorfeus_ferret_ext/test_document.c +156 -0
- data/ext/isomorfeus_ferret_ext/test_except.c +244 -0
- data/ext/isomorfeus_ferret_ext/test_fields.c +522 -0
- data/ext/isomorfeus_ferret_ext/test_file_deleter.c +185 -0
- data/ext/isomorfeus_ferret_ext/test_filter.c +331 -0
- data/ext/isomorfeus_ferret_ext/test_fs_store.c +25 -0
- data/ext/isomorfeus_ferret_ext/test_global.c +299 -0
- data/ext/isomorfeus_ferret_ext/test_hash.c +485 -0
- data/ext/isomorfeus_ferret_ext/test_hashset.c +288 -0
- data/ext/isomorfeus_ferret_ext/test_helper.c +47 -0
- data/ext/isomorfeus_ferret_ext/test_highlighter.c +548 -0
- data/ext/isomorfeus_ferret_ext/test_index.c +2323 -0
- data/ext/isomorfeus_ferret_ext/test_lang.c +74 -0
- data/ext/isomorfeus_ferret_ext/test_mempool.c +102 -0
- data/ext/isomorfeus_ferret_ext/test_multimapper.c +64 -0
- data/ext/isomorfeus_ferret_ext/test_priorityqueue.c +213 -0
- data/ext/isomorfeus_ferret_ext/test_q_const_score.c +84 -0
- data/ext/isomorfeus_ferret_ext/test_q_filtered.c +61 -0
- data/ext/isomorfeus_ferret_ext/test_q_fuzzy.c +241 -0
- data/ext/isomorfeus_ferret_ext/test_q_parser.c +464 -0
- data/ext/isomorfeus_ferret_ext/test_q_span.c +575 -0
- data/ext/isomorfeus_ferret_ext/test_ram_store.c +77 -0
- data/ext/isomorfeus_ferret_ext/test_search.c +1874 -0
- data/ext/isomorfeus_ferret_ext/test_segments.c +167 -0
- data/ext/isomorfeus_ferret_ext/test_similarity.c +25 -0
- data/ext/isomorfeus_ferret_ext/test_sort.c +333 -0
- data/ext/isomorfeus_ferret_ext/test_store.c +591 -0
- data/ext/isomorfeus_ferret_ext/test_store.h +3 -0
- data/ext/isomorfeus_ferret_ext/test_term.c +351 -0
- data/ext/isomorfeus_ferret_ext/test_term_vectors.c +373 -0
- data/ext/isomorfeus_ferret_ext/test_test.c +83 -0
- data/ext/isomorfeus_ferret_ext/test_threading.c +188 -0
- data/ext/isomorfeus_ferret_ext/testhelper.c +561 -0
- data/ext/isomorfeus_ferret_ext/testhelper.h +25 -0
- data/ext/isomorfeus_ferret_ext/tests_all.h +87 -0
- data/ext/isomorfeus_ferret_ext/uchar-ucs4.rl +1854 -0
- data/ext/isomorfeus_ferret_ext/uchar-utf8.rl +1999 -0
- data/ext/isomorfeus_ferret_ext/url.rl +27 -0
- data/ext/isomorfeus_ferret_ext/word_list.h +15156 -0
- data/lib/isomorfeus/ferret/document.rb +132 -0
- data/lib/isomorfeus/ferret/field_symbol.rb +85 -0
- data/lib/isomorfeus/ferret/index/field_infos.rb +48 -0
- data/lib/isomorfeus/ferret/index/index.rb +970 -0
- data/lib/isomorfeus/ferret/monitor.rb +323 -0
- data/lib/isomorfeus/ferret/stdlib_patches.rb +151 -0
- data/lib/isomorfeus/ferret/version.rb +5 -0
- data/lib/isomorfeus-ferret.rb +8 -0
- metadata +307 -0
@@ -0,0 +1,150 @@
|
|
1
|
+
#include "frt_global.h"
|
2
|
+
#include "frt_similarity.h"
|
3
|
+
#include "frt_search.h"
|
4
|
+
#include "frt_array.h"
|
5
|
+
#include "frt_helper.h"
|
6
|
+
#include <math.h>
|
7
|
+
#include <stdlib.h>
|
8
|
+
#include <string.h>
|
9
|
+
|
10
|
+
/****************************************************************************
|
11
|
+
*
|
12
|
+
* Term
|
13
|
+
*
|
14
|
+
****************************************************************************/
|
15
|
+
|
16
|
+
FrtTerm *frt_term_new(FrtSymbol field, const char *text)
|
17
|
+
{
|
18
|
+
FrtTerm *t = FRT_ALLOC(FrtTerm);
|
19
|
+
t->field = field;
|
20
|
+
t->text = frt_estrdup(text);
|
21
|
+
return t;
|
22
|
+
}
|
23
|
+
|
24
|
+
void frt_term_destroy(FrtTerm *self)
|
25
|
+
{
|
26
|
+
free(self->text);
|
27
|
+
free(self);
|
28
|
+
}
|
29
|
+
|
30
|
+
int frt_term_eq(const void *t1, const void *t2)
|
31
|
+
{
|
32
|
+
return ((strcmp(((FrtTerm *)t1)->text, ((FrtTerm *)t2)->text) == 0) &&
|
33
|
+
(((FrtTerm *)t1)->field == ((FrtTerm *)t2)->field));
|
34
|
+
}
|
35
|
+
|
36
|
+
unsigned long long frt_term_hash(const void *t)
|
37
|
+
{
|
38
|
+
return frt_str_hash(((FrtTerm *)t)->text) * frt_str_hash(rb_id2name(((FrtTerm *)t)->field));
|
39
|
+
}
|
40
|
+
|
41
|
+
/****************************************************************************
|
42
|
+
*
|
43
|
+
* Similarity
|
44
|
+
*
|
45
|
+
****************************************************************************/
|
46
|
+
|
47
|
+
static float simdef_length_norm(FrtSimilarity *s, FrtSymbol field, int num_terms)
|
48
|
+
{
|
49
|
+
(void)s;
|
50
|
+
(void)field;
|
51
|
+
return (float)(1.0 / sqrt(num_terms));
|
52
|
+
}
|
53
|
+
|
54
|
+
static float simdef_query_norm(struct FrtSimilarity *s, float sum_of_squared_weights)
|
55
|
+
{
|
56
|
+
(void)s;
|
57
|
+
return (float)(1.0 / sqrt(sum_of_squared_weights));
|
58
|
+
}
|
59
|
+
|
60
|
+
static float simdef_tf(struct FrtSimilarity *s, float freq)
|
61
|
+
{
|
62
|
+
(void)s;
|
63
|
+
return (float)sqrt(freq);
|
64
|
+
}
|
65
|
+
|
66
|
+
static float simdef_sloppy_freq(struct FrtSimilarity *s, int distance)
|
67
|
+
{
|
68
|
+
(void)s;
|
69
|
+
return (float)(1.0 / (double)(distance + 1));
|
70
|
+
}
|
71
|
+
|
72
|
+
static float simdef_idf_term(struct FrtSimilarity *s, FrtSymbol field, char *term,
|
73
|
+
FrtSearcher *searcher)
|
74
|
+
{
|
75
|
+
return s->idf(s, searcher->doc_freq(searcher, field, term),
|
76
|
+
searcher->max_doc(searcher));
|
77
|
+
}
|
78
|
+
|
79
|
+
static float simdef_idf_phrase(struct FrtSimilarity *s, FrtSymbol field,
|
80
|
+
FrtPhrasePosition *positions,
|
81
|
+
int pp_cnt, FrtSearcher *searcher)
|
82
|
+
{
|
83
|
+
float idf = 0.0f;
|
84
|
+
int i, j;
|
85
|
+
for (i = 0; i < pp_cnt; i++) {
|
86
|
+
char **terms = positions[i].terms;
|
87
|
+
for (j = frt_ary_size(terms) - 1; j >= 0; j--) {
|
88
|
+
idf += frt_sim_idf_term(s, field, terms[j], searcher);
|
89
|
+
}
|
90
|
+
}
|
91
|
+
return idf;
|
92
|
+
}
|
93
|
+
|
94
|
+
static float simdef_idf(struct FrtSimilarity *s, int doc_freq, int num_docs)
|
95
|
+
{
|
96
|
+
(void)s;
|
97
|
+
return (float)(log((float)num_docs/(float)(doc_freq+1)) + 1.0);
|
98
|
+
}
|
99
|
+
|
100
|
+
static float simdef_coord(struct FrtSimilarity *s, int overlap, int max_overlap)
|
101
|
+
{
|
102
|
+
(void)s;
|
103
|
+
return (float)((double)overlap / (double)max_overlap);
|
104
|
+
}
|
105
|
+
|
106
|
+
static float simdef_decode_norm(struct FrtSimilarity *s, frt_uchar b)
|
107
|
+
{
|
108
|
+
return s->norm_table[b];
|
109
|
+
}
|
110
|
+
|
111
|
+
static frt_uchar simdef_encode_norm(struct FrtSimilarity *s, float f)
|
112
|
+
{
|
113
|
+
(void)s;
|
114
|
+
return frt_float2byte(f);
|
115
|
+
}
|
116
|
+
|
117
|
+
static void simdef_destroy(FrtSimilarity *s)
|
118
|
+
{
|
119
|
+
(void)s;
|
120
|
+
/* nothing to do here */
|
121
|
+
}
|
122
|
+
|
123
|
+
static FrtSimilarity default_similarity = {
|
124
|
+
NULL,
|
125
|
+
{0},
|
126
|
+
&simdef_length_norm,
|
127
|
+
&simdef_query_norm,
|
128
|
+
&simdef_tf,
|
129
|
+
&simdef_sloppy_freq,
|
130
|
+
&simdef_idf_term,
|
131
|
+
&simdef_idf_phrase,
|
132
|
+
&simdef_idf,
|
133
|
+
&simdef_coord,
|
134
|
+
&simdef_decode_norm,
|
135
|
+
&simdef_encode_norm,
|
136
|
+
&simdef_destroy
|
137
|
+
};
|
138
|
+
|
139
|
+
FrtSimilarity *frt_sim_create_default()
|
140
|
+
{
|
141
|
+
int i;
|
142
|
+
if (!default_similarity.data) {
|
143
|
+
for (i = 0; i < 256; i++) {
|
144
|
+
default_similarity.norm_table[i] = frt_byte2float((unsigned char)i);
|
145
|
+
}
|
146
|
+
|
147
|
+
default_similarity.data = &default_similarity;
|
148
|
+
}
|
149
|
+
return &default_similarity;
|
150
|
+
}
|
@@ -0,0 +1,79 @@
|
|
1
|
+
#ifndef FRT_SIMILARITY_H
|
2
|
+
#define FRT_SIMILARITY_H
|
3
|
+
|
4
|
+
typedef struct FrtSearcher FrtSearcher;
|
5
|
+
|
6
|
+
/****************************************************************************
|
7
|
+
*
|
8
|
+
* FrtTerm
|
9
|
+
*
|
10
|
+
****************************************************************************/
|
11
|
+
|
12
|
+
typedef struct FrtTerm
|
13
|
+
{
|
14
|
+
FrtSymbol field;
|
15
|
+
char *text;
|
16
|
+
} FrtTerm;
|
17
|
+
|
18
|
+
extern FrtTerm *frt_term_new(FrtSymbol field, const char *text);
|
19
|
+
extern void frt_term_destroy(FrtTerm *self);
|
20
|
+
extern int frt_term_eq(const void *t1, const void *t2);
|
21
|
+
extern unsigned long long frt_term_hash(const void *t);
|
22
|
+
|
23
|
+
/***************************************************************************
|
24
|
+
*
|
25
|
+
* FrtPhrasePosition
|
26
|
+
*
|
27
|
+
***************************************************************************/
|
28
|
+
|
29
|
+
typedef struct FrtPhrasePosition
|
30
|
+
{
|
31
|
+
int pos;
|
32
|
+
char **terms;
|
33
|
+
} FrtPhrasePosition;
|
34
|
+
|
35
|
+
/***************************************************************************
|
36
|
+
*
|
37
|
+
* FrtSimilarity
|
38
|
+
*
|
39
|
+
***************************************************************************/
|
40
|
+
|
41
|
+
typedef struct FrtSimilarity FrtSimilarity;
|
42
|
+
|
43
|
+
struct FrtSimilarity
|
44
|
+
{
|
45
|
+
void *data;
|
46
|
+
float norm_table[256];
|
47
|
+
float (*length_norm)(FrtSimilarity *self, FrtSymbol field, int num_terms);
|
48
|
+
float (*query_norm)(FrtSimilarity *self, float sum_of_squared_weights);
|
49
|
+
float (*tf)(FrtSimilarity *self, float freq);
|
50
|
+
float (*sloppy_freq)(FrtSimilarity *self, int distance);
|
51
|
+
float (*idf_term)(FrtSimilarity *self, FrtSymbol field, char *term,
|
52
|
+
FrtSearcher *searcher);
|
53
|
+
float (*idf_phrase)(FrtSimilarity *self, FrtSymbol field,
|
54
|
+
FrtPhrasePosition *positions,
|
55
|
+
int pp_cnt, FrtSearcher *searcher);
|
56
|
+
float (*idf)(FrtSimilarity *self, int doc_freq, int num_docs);
|
57
|
+
float (*coord)(FrtSimilarity *self, int overlap, int max_overlap);
|
58
|
+
float (*decode_norm)(FrtSimilarity *self, unsigned char b);
|
59
|
+
unsigned char (*encode_norm)(FrtSimilarity *self, float f);
|
60
|
+
void (*destroy)(FrtSimilarity *self);
|
61
|
+
};
|
62
|
+
|
63
|
+
#define frt_sim_length_norm(msim, field, num_terms) msim->length_norm(msim, field, num_terms)
|
64
|
+
#define frt_sim_query_norm(msim, sosw) msim->query_norm(msim, sosw)
|
65
|
+
#define frt_sim_tf(msim, freq) msim->tf(msim, freq)
|
66
|
+
#define frt_sim_sloppy_freq(msim, distance) msim->sloppy_freq(msim, distance)
|
67
|
+
#define frt_sim_idf_term(msim, field, term, searcher)\
|
68
|
+
msim->idf_term(msim, field, term, searcher)
|
69
|
+
#define frt_sim_idf_phrase(msim, field, positions, pos_cnt, searcher)\
|
70
|
+
msim->idf_phrase(msim, field, positions, pos_cnt, searcher)
|
71
|
+
#define frt_sim_idf(msim, doc_freq, num_docs) msim->idf(msim, doc_freq, num_docs)
|
72
|
+
#define frt_sim_coord(msim, overlap, max_overlap) msim->coord(msim, overlap, max_overlap)
|
73
|
+
#define frt_sim_decode_norm(msim, b) msim->decode_norm(msim, b)
|
74
|
+
#define frt_sim_encode_norm(msim, f) msim->encode_norm(msim, f)
|
75
|
+
#define frt_sim_destroy(msim) msim->destroy(msim)
|
76
|
+
|
77
|
+
FrtSimilarity *frt_sim_create_default();
|
78
|
+
|
79
|
+
#endif
|