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 +1,672 @@
|
|
1
|
+
#include <string.h>
|
2
|
+
#include "frt_global.h"
|
3
|
+
#include "frt_search.h"
|
4
|
+
#include "frt_helper.h"
|
5
|
+
|
6
|
+
#define MTQ(query) ((FrtMultiTermQuery *)(query))
|
7
|
+
|
8
|
+
/***************************************************************************
|
9
|
+
*
|
10
|
+
* MultiTerm
|
11
|
+
*
|
12
|
+
***************************************************************************/
|
13
|
+
|
14
|
+
/***************************************************************************
|
15
|
+
* BoostedTerm
|
16
|
+
***************************************************************************/
|
17
|
+
|
18
|
+
typedef struct BoostedTerm
|
19
|
+
{
|
20
|
+
char *term;
|
21
|
+
float boost;
|
22
|
+
} BoostedTerm;
|
23
|
+
|
24
|
+
static bool boosted_term_less_than(const BoostedTerm *bt1, const BoostedTerm *bt2)
|
25
|
+
{
|
26
|
+
// if (bt1->boost == bt2->boost) { return (strcmp(bt1->term, bt2->term) < 0); }
|
27
|
+
// return (bt1->boost < bt2->boost);
|
28
|
+
if (bt1->boost < bt2->boost) {
|
29
|
+
return true;
|
30
|
+
}
|
31
|
+
if (bt1->boost > bt2->boost) {
|
32
|
+
return false;
|
33
|
+
}
|
34
|
+
return (strcmp(bt1->term, bt2->term) < 0);
|
35
|
+
}
|
36
|
+
|
37
|
+
static void boosted_term_destroy(BoostedTerm *self)
|
38
|
+
{
|
39
|
+
free(self->term);
|
40
|
+
free(self);
|
41
|
+
}
|
42
|
+
|
43
|
+
static BoostedTerm *boosted_term_new(const char *term, float boost)
|
44
|
+
{
|
45
|
+
BoostedTerm *self = FRT_ALLOC(BoostedTerm);
|
46
|
+
self->term = frt_estrdup(term);
|
47
|
+
self->boost = boost;
|
48
|
+
return self;
|
49
|
+
}
|
50
|
+
|
51
|
+
/***************************************************************************
|
52
|
+
* TermDocEnumWrapper
|
53
|
+
***************************************************************************/
|
54
|
+
|
55
|
+
#define TDE_READ_SIZE 16
|
56
|
+
|
57
|
+
typedef struct TermDocEnumWrapper
|
58
|
+
{
|
59
|
+
const char *term;
|
60
|
+
FrtTermDocEnum *tde;
|
61
|
+
float boost;
|
62
|
+
int doc;
|
63
|
+
int freq;
|
64
|
+
int docs[TDE_READ_SIZE];
|
65
|
+
int freqs[TDE_READ_SIZE];
|
66
|
+
int pointer;
|
67
|
+
int pointer_max;
|
68
|
+
} TermDocEnumWrapper;
|
69
|
+
|
70
|
+
static bool tdew_less_than(const TermDocEnumWrapper *tdew1,
|
71
|
+
const TermDocEnumWrapper *tdew2)
|
72
|
+
{
|
73
|
+
return (tdew1->doc < tdew2->doc);
|
74
|
+
}
|
75
|
+
|
76
|
+
static bool tdew_next(TermDocEnumWrapper *self)
|
77
|
+
{
|
78
|
+
self->pointer++;
|
79
|
+
if (self->pointer >= self->pointer_max) {
|
80
|
+
/* refill buffer */
|
81
|
+
self->pointer_max = self->tde->read(self->tde, self->docs, self->freqs, TDE_READ_SIZE);
|
82
|
+
if (self->pointer_max != 0) {
|
83
|
+
self->pointer = 0;
|
84
|
+
}
|
85
|
+
else {
|
86
|
+
return false;
|
87
|
+
}
|
88
|
+
}
|
89
|
+
self->doc = self->docs[self->pointer];
|
90
|
+
self->freq = self->freqs[self->pointer];
|
91
|
+
return true;
|
92
|
+
}
|
93
|
+
|
94
|
+
static bool tdew_skip_to(TermDocEnumWrapper *self, int doc_num)
|
95
|
+
{
|
96
|
+
FrtTermDocEnum *tde = self->tde;
|
97
|
+
|
98
|
+
while (++(self->pointer) < self->pointer_max) {
|
99
|
+
if (self->docs[self->pointer] >= doc_num) {
|
100
|
+
self->doc = self->docs[self->pointer];
|
101
|
+
self->freq = self->freqs[self->pointer];
|
102
|
+
return true;
|
103
|
+
}
|
104
|
+
}
|
105
|
+
|
106
|
+
/* not found in cache, seek underlying stream */
|
107
|
+
if (tde->skip_to(tde, doc_num)) {
|
108
|
+
self->pointer_max = 1;
|
109
|
+
self->pointer = 0;
|
110
|
+
self->docs[0] = self->doc = tde->doc_num(tde);
|
111
|
+
self->freqs[0] = self->freq = tde->freq(tde);
|
112
|
+
return true;
|
113
|
+
} else {
|
114
|
+
return false;
|
115
|
+
}
|
116
|
+
}
|
117
|
+
|
118
|
+
static void tdew_destroy(TermDocEnumWrapper *self)
|
119
|
+
{
|
120
|
+
self->tde->close(self->tde);
|
121
|
+
free(self);
|
122
|
+
}
|
123
|
+
|
124
|
+
static TermDocEnumWrapper *tdew_new(const char *term, FrtTermDocEnum *tde,
|
125
|
+
float boost)
|
126
|
+
{
|
127
|
+
TermDocEnumWrapper *self = FRT_ALLOC_AND_ZERO(TermDocEnumWrapper);
|
128
|
+
self->term = term;
|
129
|
+
self->tde = tde;
|
130
|
+
self->boost = boost;
|
131
|
+
self->doc = -1;
|
132
|
+
return self;
|
133
|
+
}
|
134
|
+
|
135
|
+
/***************************************************************************
|
136
|
+
* MultiTermScorer
|
137
|
+
***************************************************************************/
|
138
|
+
|
139
|
+
#define SCORE_CACHE_SIZE 32
|
140
|
+
#define MTSc(scorer) ((MultiTermScorer *)(scorer))
|
141
|
+
|
142
|
+
typedef struct MultiTermScorer
|
143
|
+
{
|
144
|
+
FrtScorer super;
|
145
|
+
FrtSymbol field;
|
146
|
+
frt_uchar *norms;
|
147
|
+
FrtWeight *weight;
|
148
|
+
TermDocEnumWrapper **tdew_a;
|
149
|
+
int tdew_cnt;
|
150
|
+
FrtPriorityQueue *tdew_pq;
|
151
|
+
float weight_value;
|
152
|
+
float score_cache[SCORE_CACHE_SIZE];
|
153
|
+
float total_score;
|
154
|
+
} MultiTermScorer;
|
155
|
+
|
156
|
+
static float multi_tsc_score(FrtScorer *self)
|
157
|
+
{
|
158
|
+
return MTSc(self)->total_score * MTSc(self)->weight_value
|
159
|
+
* frt_sim_decode_norm(self->similarity, MTSc(self)->norms[self->doc]);
|
160
|
+
}
|
161
|
+
|
162
|
+
static bool multi_tsc_next(FrtScorer *self)
|
163
|
+
{
|
164
|
+
int curr_doc;
|
165
|
+
float total_score = 0.0f;
|
166
|
+
TermDocEnumWrapper *tdew;
|
167
|
+
MultiTermScorer *mtsc = MTSc(self);
|
168
|
+
FrtPriorityQueue *tdew_pq = mtsc->tdew_pq;
|
169
|
+
if (tdew_pq == NULL) {
|
170
|
+
TermDocEnumWrapper **tdew_a = mtsc->tdew_a;
|
171
|
+
int i;
|
172
|
+
tdew_pq = frt_pq_new(mtsc->tdew_cnt, (frt_lt_ft)tdew_less_than, (frt_free_ft)NULL);
|
173
|
+
for (i = mtsc->tdew_cnt - 1; i >= 0; i--) {
|
174
|
+
if (tdew_next(tdew_a[i])) {
|
175
|
+
frt_pq_push(tdew_pq, tdew_a[i]);
|
176
|
+
}
|
177
|
+
}
|
178
|
+
mtsc->tdew_pq = tdew_pq;
|
179
|
+
}
|
180
|
+
|
181
|
+
tdew = (TermDocEnumWrapper *)frt_pq_top(tdew_pq);
|
182
|
+
if (tdew == NULL) {
|
183
|
+
return false;
|
184
|
+
}
|
185
|
+
|
186
|
+
self->doc = curr_doc = tdew->doc;
|
187
|
+
do {
|
188
|
+
int freq = tdew->freq;
|
189
|
+
if (freq < SCORE_CACHE_SIZE) {
|
190
|
+
total_score += mtsc->score_cache[freq] * tdew->boost;
|
191
|
+
}
|
192
|
+
else {
|
193
|
+
total_score += frt_sim_tf(self->similarity, (float)freq) * tdew->boost;
|
194
|
+
}
|
195
|
+
|
196
|
+
if (tdew_next(tdew)) {
|
197
|
+
frt_pq_down(tdew_pq);
|
198
|
+
}
|
199
|
+
else {
|
200
|
+
frt_pq_pop(tdew_pq);
|
201
|
+
}
|
202
|
+
|
203
|
+
} while (((tdew = (TermDocEnumWrapper *)frt_pq_top(tdew_pq)) != NULL)
|
204
|
+
&& tdew->doc == curr_doc);
|
205
|
+
mtsc->total_score = total_score;
|
206
|
+
return true;
|
207
|
+
}
|
208
|
+
|
209
|
+
static bool multi_tsc_advance_to(FrtScorer *self, int target_doc_num)
|
210
|
+
{
|
211
|
+
FrtPriorityQueue *tdew_pq = MTSc(self)->tdew_pq;
|
212
|
+
TermDocEnumWrapper *tdew;
|
213
|
+
if (tdew_pq == NULL) {
|
214
|
+
MultiTermScorer *mtsc = MTSc(self);
|
215
|
+
TermDocEnumWrapper **tdew_a = mtsc->tdew_a;
|
216
|
+
int i;
|
217
|
+
tdew_pq = frt_pq_new(mtsc->tdew_cnt, (frt_lt_ft)tdew_less_than, (frt_free_ft)NULL);
|
218
|
+
for (i = mtsc->tdew_cnt - 1; i >= 0; i--) {
|
219
|
+
if (tdew_skip_to(tdew_a[i], target_doc_num)) {
|
220
|
+
frt_pq_push(tdew_pq, tdew_a[i]);
|
221
|
+
}
|
222
|
+
}
|
223
|
+
MTSc(self)->tdew_pq = tdew_pq;
|
224
|
+
}
|
225
|
+
if (tdew_pq->size == 0) {
|
226
|
+
self->doc = -1;
|
227
|
+
return false;
|
228
|
+
}
|
229
|
+
while ((tdew = (TermDocEnumWrapper *)frt_pq_top(tdew_pq)) != NULL
|
230
|
+
&& (target_doc_num > tdew->doc)) {
|
231
|
+
if (tdew_skip_to(tdew, target_doc_num)) {
|
232
|
+
frt_pq_down(tdew_pq);
|
233
|
+
}
|
234
|
+
else {
|
235
|
+
frt_pq_pop(tdew_pq);
|
236
|
+
}
|
237
|
+
}
|
238
|
+
tdew = (TermDocEnumWrapper *)frt_pq_top(tdew_pq);
|
239
|
+
return (frt_pq_top(tdew_pq) == NULL) ? false : true;
|
240
|
+
}
|
241
|
+
|
242
|
+
static bool multi_tsc_skip_to(FrtScorer *self, int target_doc_num)
|
243
|
+
{
|
244
|
+
return multi_tsc_advance_to(self, target_doc_num) && multi_tsc_next(self);
|
245
|
+
}
|
246
|
+
|
247
|
+
static FrtExplanation *multi_tsc_explain(FrtScorer *self, int doc_num)
|
248
|
+
{
|
249
|
+
MultiTermScorer *mtsc = MTSc(self);
|
250
|
+
TermDocEnumWrapper *tdew;
|
251
|
+
|
252
|
+
if (multi_tsc_advance_to(self, doc_num) &&
|
253
|
+
(tdew = (TermDocEnumWrapper *)frt_pq_top(mtsc->tdew_pq))->doc == doc_num) {
|
254
|
+
|
255
|
+
FrtPriorityQueue *tdew_pq = MTSc(self)->tdew_pq;
|
256
|
+
FrtExplanation *expl = frt_expl_new(0.0f, "The sum of:");
|
257
|
+
int curr_doc = self->doc = tdew->doc;
|
258
|
+
float total_score = 0.0f;
|
259
|
+
|
260
|
+
do {
|
261
|
+
int freq = tdew->freq;
|
262
|
+
frt_expl_add_detail(expl,
|
263
|
+
frt_expl_new(frt_sim_tf(self->similarity, (float)freq) * tdew->boost,
|
264
|
+
"tf(term_freq(%s:%s)=%d)^%f",
|
265
|
+
rb_id2name(mtsc->field), tdew->term, freq, tdew->boost));
|
266
|
+
|
267
|
+
total_score += frt_sim_tf(self->similarity, (float)freq) * tdew->boost;
|
268
|
+
|
269
|
+
/* maintain tdew queue, even though it probably won't get used
|
270
|
+
* again */
|
271
|
+
if (tdew_next(tdew)) {
|
272
|
+
frt_pq_down(tdew_pq);
|
273
|
+
}
|
274
|
+
else {
|
275
|
+
frt_pq_pop(tdew_pq);
|
276
|
+
}
|
277
|
+
|
278
|
+
} while (((tdew = (TermDocEnumWrapper *)frt_pq_top(tdew_pq)) != NULL)
|
279
|
+
&& tdew->doc == curr_doc);
|
280
|
+
expl->value = total_score;
|
281
|
+
return expl;
|
282
|
+
}
|
283
|
+
else {
|
284
|
+
return frt_expl_new(0.0f, "None of the required terms exist in the index");
|
285
|
+
}
|
286
|
+
}
|
287
|
+
|
288
|
+
static void multi_tsc_destroy(FrtScorer *self)
|
289
|
+
{
|
290
|
+
int i;
|
291
|
+
TermDocEnumWrapper **tdew_a = MTSc(self)->tdew_a;
|
292
|
+
for (i = MTSc(self)->tdew_cnt - 1; i >= 0; i--) {
|
293
|
+
tdew_destroy(tdew_a[i]);
|
294
|
+
}
|
295
|
+
free(tdew_a);
|
296
|
+
if (MTSc(self)->tdew_pq) frt_pq_destroy(MTSc(self)->tdew_pq);
|
297
|
+
frt_scorer_destroy_i(self);
|
298
|
+
}
|
299
|
+
|
300
|
+
static FrtScorer *multi_tsc_new(FrtWeight *weight, FrtSymbol field,
|
301
|
+
TermDocEnumWrapper **tdew_a, int tdew_cnt,
|
302
|
+
frt_uchar *norms)
|
303
|
+
{
|
304
|
+
int i;
|
305
|
+
FrtScorer *self = frt_scorer_new(MultiTermScorer, weight->similarity);
|
306
|
+
|
307
|
+
MTSc(self)->weight = weight;
|
308
|
+
MTSc(self)->field = field;
|
309
|
+
MTSc(self)->weight_value = weight->value;
|
310
|
+
MTSc(self)->tdew_a = tdew_a;
|
311
|
+
MTSc(self)->tdew_cnt = tdew_cnt;
|
312
|
+
MTSc(self)->norms = norms;
|
313
|
+
|
314
|
+
for (i = 0; i < SCORE_CACHE_SIZE; i++) {
|
315
|
+
MTSc(self)->score_cache[i] = frt_sim_tf(self->similarity, (float)i);
|
316
|
+
}
|
317
|
+
|
318
|
+
self->score = &multi_tsc_score;
|
319
|
+
self->next = &multi_tsc_next;
|
320
|
+
self->skip_to = &multi_tsc_skip_to;
|
321
|
+
self->explain = &multi_tsc_explain;
|
322
|
+
self->destroy = &multi_tsc_destroy;
|
323
|
+
|
324
|
+
return self;
|
325
|
+
}
|
326
|
+
|
327
|
+
/***************************************************************************
|
328
|
+
* MultiTermWeight
|
329
|
+
***************************************************************************/
|
330
|
+
|
331
|
+
static char *multi_tw_to_s(FrtWeight *self)
|
332
|
+
{
|
333
|
+
return frt_strfmt("MultiTermWeight(%f)", self->value);
|
334
|
+
}
|
335
|
+
|
336
|
+
static FrtScorer *multi_tw_scorer(FrtWeight *self, FrtIndexReader *ir)
|
337
|
+
{
|
338
|
+
FrtScorer *multi_tsc = NULL;
|
339
|
+
FrtPriorityQueue *boosted_terms = MTQ(self->query)->boosted_terms;
|
340
|
+
const int field_num = frt_fis_get_field_num(ir->fis, MTQ(self->query)->field);
|
341
|
+
|
342
|
+
if (boosted_terms->size > 0 && field_num >= 0) {
|
343
|
+
int i;
|
344
|
+
FrtTermDocEnum *tde;
|
345
|
+
FrtTermEnum *te = ir->terms(ir, field_num);
|
346
|
+
TermDocEnumWrapper **tdew_a = FRT_ALLOC_N(TermDocEnumWrapper *,
|
347
|
+
boosted_terms->size);
|
348
|
+
int tdew_cnt = 0;
|
349
|
+
/* Priority queues skip the first element */
|
350
|
+
for (i = boosted_terms->size; i > 0; i--) {
|
351
|
+
char *term;
|
352
|
+
BoostedTerm *bt = (BoostedTerm *)boosted_terms->heap[i];
|
353
|
+
if (((term = te->skip_to(te, bt->term)) != NULL) && (strcmp(term, bt->term) == 0)) {
|
354
|
+
tde = ir->term_docs(ir);
|
355
|
+
tde->seek_te(tde, te);
|
356
|
+
tdew_a[tdew_cnt++] = tdew_new(bt->term, tde, bt->boost);
|
357
|
+
}
|
358
|
+
}
|
359
|
+
te->close(te);
|
360
|
+
if (tdew_cnt) {
|
361
|
+
multi_tsc = multi_tsc_new(self, MTQ(self->query)->field, tdew_a,
|
362
|
+
tdew_cnt, frt_ir_get_norms_i(ir, field_num));
|
363
|
+
}
|
364
|
+
else {
|
365
|
+
free(tdew_a);
|
366
|
+
}
|
367
|
+
}
|
368
|
+
return multi_tsc;
|
369
|
+
}
|
370
|
+
|
371
|
+
static FrtExplanation *multi_tw_explain(FrtWeight *self, FrtIndexReader *ir, int doc_num)
|
372
|
+
{
|
373
|
+
FrtExplanation *expl;
|
374
|
+
FrtExplanation *idf_expl1;
|
375
|
+
FrtExplanation *idf_expl2;
|
376
|
+
FrtExplanation *query_expl;
|
377
|
+
FrtExplanation *qnorm_expl;
|
378
|
+
FrtExplanation *field_expl;
|
379
|
+
FrtExplanation *tf_expl;
|
380
|
+
FrtScorer *scorer;
|
381
|
+
frt_uchar *field_norms;
|
382
|
+
float field_norm;
|
383
|
+
FrtExplanation *field_norm_expl;
|
384
|
+
|
385
|
+
char *query_str;
|
386
|
+
FrtMultiTermQuery *mtq = MTQ(self->query);
|
387
|
+
const char *field_name = rb_id2name(mtq->field);
|
388
|
+
FrtPriorityQueue *bt_pq = mtq->boosted_terms;
|
389
|
+
int i;
|
390
|
+
int total_doc_freqs = 0;
|
391
|
+
char *doc_freqs = NULL;
|
392
|
+
size_t len = 0, pos = 0;
|
393
|
+
const int field_num = frt_fis_get_field_num(ir->fis, mtq->field);
|
394
|
+
|
395
|
+
if (field_num < 0) {
|
396
|
+
return frt_expl_new(0.0f, "field \"%s\" does not exist in the index",
|
397
|
+
field_name);
|
398
|
+
}
|
399
|
+
|
400
|
+
query_str = self->query->to_s(self->query, (FrtSymbol)NULL);
|
401
|
+
|
402
|
+
expl = frt_expl_new(0.0f, "weight(%s in %d), product of:", query_str, doc_num);
|
403
|
+
|
404
|
+
len = 30;
|
405
|
+
for (i = bt_pq->size; i > 0; i--) {
|
406
|
+
len += strlen(((BoostedTerm *)bt_pq->heap[i])->term) + 30;
|
407
|
+
}
|
408
|
+
doc_freqs = FRT_ALLOC_N(char, len);
|
409
|
+
for (i = bt_pq->size; i > 0; i--) {
|
410
|
+
char *term = ((BoostedTerm *)bt_pq->heap[i])->term;
|
411
|
+
int doc_freq = ir->doc_freq(ir, field_num, term);
|
412
|
+
pos += sprintf(doc_freqs + pos, "(%s=%d) + ", term, doc_freq);
|
413
|
+
total_doc_freqs += doc_freq;
|
414
|
+
}
|
415
|
+
pos -= 2; /* remove " + " from the end */
|
416
|
+
sprintf(doc_freqs + pos, "= %d", total_doc_freqs);
|
417
|
+
|
418
|
+
idf_expl1 = frt_expl_new(self->idf, "idf(%s:<%s>)", field_name, doc_freqs);
|
419
|
+
idf_expl2 = frt_expl_new(self->idf, "idf(%s:<%s>)", field_name, doc_freqs);
|
420
|
+
free(doc_freqs);
|
421
|
+
|
422
|
+
/* explain query weight */
|
423
|
+
query_expl = frt_expl_new(0.0f, "query_weight(%s), product of:", query_str);
|
424
|
+
|
425
|
+
if (self->query->boost != 1.0f) {
|
426
|
+
frt_expl_add_detail(query_expl, frt_expl_new(self->query->boost, "boost"));
|
427
|
+
}
|
428
|
+
frt_expl_add_detail(query_expl, idf_expl1);
|
429
|
+
|
430
|
+
qnorm_expl = frt_expl_new(self->qnorm, "query_norm");
|
431
|
+
frt_expl_add_detail(query_expl, qnorm_expl);
|
432
|
+
|
433
|
+
query_expl->value = self->query->boost * self->idf * self->qnorm;
|
434
|
+
|
435
|
+
frt_expl_add_detail(expl, query_expl);
|
436
|
+
|
437
|
+
/* explain field weight */
|
438
|
+
field_expl = frt_expl_new(0.0f, "field_weight(%s in %d), product of:",
|
439
|
+
query_str, doc_num);
|
440
|
+
free(query_str);
|
441
|
+
|
442
|
+
if ((scorer = self->scorer(self, ir)) != NULL) {
|
443
|
+
tf_expl = scorer->explain(scorer, doc_num);
|
444
|
+
scorer->destroy(scorer);
|
445
|
+
}
|
446
|
+
else {
|
447
|
+
tf_expl = frt_expl_new(0.0f, "no terms were found");
|
448
|
+
}
|
449
|
+
frt_expl_add_detail(field_expl, tf_expl);
|
450
|
+
frt_expl_add_detail(field_expl, idf_expl2);
|
451
|
+
|
452
|
+
field_norms = ir->get_norms(ir, field_num);
|
453
|
+
field_norm = (field_norms != NULL)
|
454
|
+
? frt_sim_decode_norm(self->similarity, field_norms[doc_num])
|
455
|
+
: (float)0.0f;
|
456
|
+
field_norm_expl = frt_expl_new(field_norm, "field_norm(field=%s, doc=%d)",
|
457
|
+
field_name, doc_num);
|
458
|
+
|
459
|
+
frt_expl_add_detail(field_expl, field_norm_expl);
|
460
|
+
|
461
|
+
field_expl->value = tf_expl->value * self->idf * field_norm;
|
462
|
+
|
463
|
+
/* combine them */
|
464
|
+
if (query_expl->value == 1.0f) {
|
465
|
+
frt_expl_destroy(expl);
|
466
|
+
return field_expl;
|
467
|
+
}
|
468
|
+
else {
|
469
|
+
expl->value = (query_expl->value * field_expl->value);
|
470
|
+
frt_expl_add_detail(expl, field_expl);
|
471
|
+
return expl;
|
472
|
+
}
|
473
|
+
}
|
474
|
+
|
475
|
+
static FrtWeight *multi_tw_new(FrtQuery *query, FrtSearcher *searcher)
|
476
|
+
{
|
477
|
+
int i;
|
478
|
+
int doc_freq = 0;
|
479
|
+
FrtWeight *self = w_new(FrtWeight, query);
|
480
|
+
FrtPriorityQueue *bt_pq = MTQ(query)->boosted_terms;
|
481
|
+
|
482
|
+
self->scorer = &multi_tw_scorer;
|
483
|
+
self->explain = &multi_tw_explain;
|
484
|
+
self->to_s = &multi_tw_to_s;
|
485
|
+
|
486
|
+
self->similarity = query->get_similarity(query, searcher);
|
487
|
+
self->value = query->boost;
|
488
|
+
self->idf = 0.0f;
|
489
|
+
|
490
|
+
for (i = bt_pq->size; i > 0; i--) {
|
491
|
+
doc_freq += searcher->doc_freq(searcher, MTQ(query)->field,
|
492
|
+
((BoostedTerm *)bt_pq->heap[i])->term);
|
493
|
+
}
|
494
|
+
self->idf += frt_sim_idf(self->similarity, doc_freq,
|
495
|
+
searcher->max_doc(searcher));
|
496
|
+
|
497
|
+
return self;
|
498
|
+
}
|
499
|
+
|
500
|
+
|
501
|
+
/***************************************************************************
|
502
|
+
* MultiTermQuery
|
503
|
+
***************************************************************************/
|
504
|
+
|
505
|
+
static char *multi_tq_to_s(FrtQuery *self, FrtSymbol default_field)
|
506
|
+
{
|
507
|
+
int i;
|
508
|
+
FrtPriorityQueue *boosted_terms = MTQ(self)->boosted_terms, *bt_pq_clone;
|
509
|
+
BoostedTerm *bt;
|
510
|
+
char *buffer, *bptr;
|
511
|
+
const char *field_name = rb_id2name(MTQ(self)->field);
|
512
|
+
int flen = strlen(field_name);
|
513
|
+
int tlen = 0;
|
514
|
+
|
515
|
+
/* Priority queues skip the first element */
|
516
|
+
for (i = boosted_terms->size; i > 0; i--) {
|
517
|
+
tlen += strlen(((BoostedTerm *)boosted_terms->heap[i])->term) + 35;
|
518
|
+
}
|
519
|
+
|
520
|
+
bptr = buffer = FRT_ALLOC_N(char, tlen + flen + 35);
|
521
|
+
|
522
|
+
if (default_field != MTQ(self)->field) {
|
523
|
+
bptr += sprintf(bptr, "%s:", field_name);
|
524
|
+
}
|
525
|
+
|
526
|
+
*(bptr++) = '"';
|
527
|
+
bt_pq_clone = frt_pq_clone(boosted_terms);
|
528
|
+
while ((bt = (BoostedTerm *)frt_pq_pop(bt_pq_clone)) != NULL) {
|
529
|
+
bptr += sprintf(bptr, "%s", bt->term);
|
530
|
+
|
531
|
+
if (bt->boost != 1.0f) {
|
532
|
+
*bptr = '^';
|
533
|
+
frt_dbl_to_s(++bptr, bt->boost);
|
534
|
+
bptr += (int)strlen(bptr);
|
535
|
+
}
|
536
|
+
|
537
|
+
*(bptr++) = '|';
|
538
|
+
}
|
539
|
+
frt_pq_destroy(bt_pq_clone);
|
540
|
+
|
541
|
+
if (bptr[-1] == '"') {
|
542
|
+
bptr++; /* handle zero term case */
|
543
|
+
}
|
544
|
+
bptr[-1] = '"'; /* delete last '|' char */
|
545
|
+
bptr[ 0] = '\0';
|
546
|
+
|
547
|
+
if (self->boost != 1.0f) {
|
548
|
+
*bptr = '^';
|
549
|
+
frt_dbl_to_s(++bptr, self->boost);
|
550
|
+
}
|
551
|
+
return buffer;
|
552
|
+
}
|
553
|
+
|
554
|
+
static void multi_tq_destroy_i(FrtQuery *self)
|
555
|
+
{
|
556
|
+
frt_pq_destroy(MTQ(self)->boosted_terms);
|
557
|
+
frt_q_destroy_i(self);
|
558
|
+
}
|
559
|
+
|
560
|
+
static void multi_tq_extract_terms(FrtQuery *self, FrtHashSet *terms)
|
561
|
+
{
|
562
|
+
int i;
|
563
|
+
FrtPriorityQueue *boosted_terms = MTQ(self)->boosted_terms;
|
564
|
+
for (i = boosted_terms->size; i > 0; i--) {
|
565
|
+
BoostedTerm *bt = (BoostedTerm *)boosted_terms->heap[i];
|
566
|
+
frt_hs_add(terms, frt_term_new(MTQ(self)->field, bt->term));
|
567
|
+
}
|
568
|
+
}
|
569
|
+
|
570
|
+
static unsigned long long multi_tq_hash(FrtQuery *self)
|
571
|
+
{
|
572
|
+
int i;
|
573
|
+
unsigned long long hash = frt_str_hash(rb_id2name(MTQ(self)->field));
|
574
|
+
FrtPriorityQueue *boosted_terms = MTQ(self)->boosted_terms;
|
575
|
+
for (i = boosted_terms->size; i > 0; i--) {
|
576
|
+
BoostedTerm *bt = (BoostedTerm *)boosted_terms->heap[i];
|
577
|
+
hash ^= frt_str_hash(bt->term) ^ frt_float2int(bt->boost);
|
578
|
+
}
|
579
|
+
return hash;
|
580
|
+
}
|
581
|
+
|
582
|
+
static int multi_tq_eq(FrtQuery *self, FrtQuery *o)
|
583
|
+
{
|
584
|
+
int i;
|
585
|
+
FrtPriorityQueue *boosted_terms1 = MTQ(self)->boosted_terms;
|
586
|
+
FrtPriorityQueue *boosted_terms2 = MTQ(o)->boosted_terms;
|
587
|
+
|
588
|
+
if ((MTQ(self)->field != MTQ(o)->field)
|
589
|
+
|| boosted_terms1->size != boosted_terms2->size) {
|
590
|
+
return false;
|
591
|
+
}
|
592
|
+
for (i = boosted_terms1->size; i > 0; i--) {
|
593
|
+
BoostedTerm *bt1 = (BoostedTerm *)boosted_terms1->heap[i];
|
594
|
+
BoostedTerm *bt2 = (BoostedTerm *)boosted_terms2->heap[i];
|
595
|
+
if ((strcmp(bt1->term, bt2->term) != 0) || (bt1->boost != bt2->boost)) {
|
596
|
+
return false;
|
597
|
+
}
|
598
|
+
}
|
599
|
+
return true;
|
600
|
+
}
|
601
|
+
|
602
|
+
static FrtMatchVector *multi_tq_get_matchv_i(FrtQuery *self, FrtMatchVector *mv,
|
603
|
+
FrtTermVector *tv)
|
604
|
+
{
|
605
|
+
if (tv->field == MTQ(self)->field) {
|
606
|
+
int i;
|
607
|
+
FrtPriorityQueue *boosted_terms = MTQ(self)->boosted_terms;
|
608
|
+
for (i = boosted_terms->size; i > 0; i--) {
|
609
|
+
int j;
|
610
|
+
BoostedTerm *bt = (BoostedTerm *)boosted_terms->heap[i];
|
611
|
+
FrtTVTerm *tv_term = frt_tv_get_tv_term(tv, bt->term);
|
612
|
+
if (tv_term) {
|
613
|
+
for (j = 0; j < tv_term->freq; j++) {
|
614
|
+
int pos = tv_term->positions[j];
|
615
|
+
frt_matchv_add(mv, pos, pos);
|
616
|
+
}
|
617
|
+
}
|
618
|
+
}
|
619
|
+
}
|
620
|
+
return mv;
|
621
|
+
}
|
622
|
+
|
623
|
+
FrtQuery *frt_multi_tq_new_conf(FrtSymbol field, int max_terms, float min_boost)
|
624
|
+
{
|
625
|
+
FrtQuery *self;
|
626
|
+
|
627
|
+
if (max_terms <= 0) {
|
628
|
+
FRT_RAISE(FRT_ARG_ERROR, ":max_terms must be greater than or equal to zero. "
|
629
|
+
"%d < 0. ", max_terms);
|
630
|
+
}
|
631
|
+
|
632
|
+
self = frt_q_new(FrtMultiTermQuery);
|
633
|
+
|
634
|
+
MTQ(self)->field = field;
|
635
|
+
MTQ(self)->boosted_terms = frt_pq_new(max_terms,
|
636
|
+
(frt_lt_ft)&boosted_term_less_than,
|
637
|
+
(frt_free_ft)&boosted_term_destroy);
|
638
|
+
MTQ(self)->min_boost = min_boost;
|
639
|
+
|
640
|
+
self->type = MULTI_TERM_QUERY;
|
641
|
+
self->to_s = &multi_tq_to_s;
|
642
|
+
self->extract_terms = &multi_tq_extract_terms;
|
643
|
+
self->hash = &multi_tq_hash;
|
644
|
+
self->eq = &multi_tq_eq;
|
645
|
+
self->destroy_i = &multi_tq_destroy_i;
|
646
|
+
self->create_weight_i = &multi_tw_new;
|
647
|
+
self->get_matchv_i = &multi_tq_get_matchv_i;
|
648
|
+
|
649
|
+
return self;
|
650
|
+
}
|
651
|
+
|
652
|
+
FrtQuery *frt_multi_tq_new(FrtSymbol field)
|
653
|
+
{
|
654
|
+
return frt_multi_tq_new_conf(field, MULTI_TERM_QUERY_MAX_TERMS, 0.0f);
|
655
|
+
}
|
656
|
+
|
657
|
+
void frt_multi_tq_add_term_boost(FrtQuery *self, const char *term, float boost)
|
658
|
+
{
|
659
|
+
if (boost > MTQ(self)->min_boost && term && term[0]) {
|
660
|
+
BoostedTerm *bt = boosted_term_new(term, boost);
|
661
|
+
FrtPriorityQueue *bt_pq = MTQ(self)->boosted_terms;
|
662
|
+
frt_pq_insert(bt_pq, bt);
|
663
|
+
if (frt_pq_full(bt_pq)) {
|
664
|
+
MTQ(self)->min_boost = ((BoostedTerm *)frt_pq_top(bt_pq))->boost;
|
665
|
+
}
|
666
|
+
}
|
667
|
+
}
|
668
|
+
|
669
|
+
void frt_multi_tq_add_term(FrtQuery *self, const char *term)
|
670
|
+
{
|
671
|
+
frt_multi_tq_add_term_boost(self, term, 1.0f);
|
672
|
+
}
|