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 +1,311 @@
|
|
1
|
+
#include "frt_global.h"
|
2
|
+
#include <string.h>
|
3
|
+
#include "frt_search.h"
|
4
|
+
|
5
|
+
#define TQ(query) ((FrtTermQuery *)(query))
|
6
|
+
#define TSc(scorer) ((TermScorer *)(scorer))
|
7
|
+
|
8
|
+
/***************************************************************************
|
9
|
+
*
|
10
|
+
* TermScorer
|
11
|
+
*
|
12
|
+
***************************************************************************/
|
13
|
+
|
14
|
+
#define SCORE_CACHE_SIZE 32
|
15
|
+
#define TDE_READ_SIZE 32
|
16
|
+
|
17
|
+
typedef struct TermScorer
|
18
|
+
{
|
19
|
+
FrtScorer super;
|
20
|
+
int docs[TDE_READ_SIZE];
|
21
|
+
int freqs[TDE_READ_SIZE];
|
22
|
+
int pointer;
|
23
|
+
int pointer_max;
|
24
|
+
float score_cache[SCORE_CACHE_SIZE];
|
25
|
+
FrtWeight *weight;
|
26
|
+
FrtTermDocEnum *tde;
|
27
|
+
frt_uchar *norms;
|
28
|
+
float weight_value;
|
29
|
+
} TermScorer;
|
30
|
+
|
31
|
+
static float tsc_score(FrtScorer *self)
|
32
|
+
{
|
33
|
+
TermScorer *ts = TSc(self);
|
34
|
+
int freq = ts->freqs[ts->pointer];
|
35
|
+
float score;
|
36
|
+
/* compute tf(f)*weight */
|
37
|
+
if (freq < SCORE_CACHE_SIZE) { /* check cache */
|
38
|
+
score = ts->score_cache[freq]; /* cache hit */
|
39
|
+
}
|
40
|
+
else {
|
41
|
+
/* cache miss */
|
42
|
+
score = frt_sim_tf(self->similarity, (float)freq) * ts->weight_value;
|
43
|
+
}
|
44
|
+
/* normalize for field */
|
45
|
+
score *= frt_sim_decode_norm(self->similarity, ts->norms[self->doc]);
|
46
|
+
return score;
|
47
|
+
}
|
48
|
+
|
49
|
+
static bool tsc_next(FrtScorer *self)
|
50
|
+
{
|
51
|
+
TermScorer *ts = TSc(self);
|
52
|
+
|
53
|
+
ts->pointer++;
|
54
|
+
if (ts->pointer >= ts->pointer_max) {
|
55
|
+
/* refill buffer */
|
56
|
+
ts->pointer_max = ts->tde->read(ts->tde, ts->docs, ts->freqs,
|
57
|
+
TDE_READ_SIZE);
|
58
|
+
if (ts->pointer_max != 0) {
|
59
|
+
ts->pointer = 0;
|
60
|
+
}
|
61
|
+
else {
|
62
|
+
return false;
|
63
|
+
}
|
64
|
+
}
|
65
|
+
self->doc = ts->docs[ts->pointer];
|
66
|
+
return true;
|
67
|
+
}
|
68
|
+
|
69
|
+
static bool tsc_skip_to(FrtScorer *self, int doc_num)
|
70
|
+
{
|
71
|
+
TermScorer *ts = TSc(self);
|
72
|
+
FrtTermDocEnum *tde = ts->tde;
|
73
|
+
|
74
|
+
/* first scan in cache */
|
75
|
+
while (++(ts->pointer) < ts->pointer_max) {
|
76
|
+
if (ts->docs[ts->pointer] >= doc_num) {
|
77
|
+
self->doc = ts->docs[ts->pointer];
|
78
|
+
return true;
|
79
|
+
}
|
80
|
+
}
|
81
|
+
|
82
|
+
/* not found in cache, seek underlying stream */
|
83
|
+
if (tde->skip_to(tde, doc_num)) {
|
84
|
+
ts->pointer_max = 1;
|
85
|
+
ts->pointer = 0;
|
86
|
+
ts->docs[0] = self->doc = tde->doc_num(tde);
|
87
|
+
ts->freqs[0] = tde->freq(tde);
|
88
|
+
return true;
|
89
|
+
}
|
90
|
+
else {
|
91
|
+
return false;
|
92
|
+
}
|
93
|
+
}
|
94
|
+
|
95
|
+
static FrtExplanation *tsc_explain(FrtScorer *self, int doc_num)
|
96
|
+
{
|
97
|
+
TermScorer *ts = TSc(self);
|
98
|
+
FrtQuery *query = ts->weight->get_query(ts->weight);
|
99
|
+
int tf = 0;
|
100
|
+
|
101
|
+
tsc_skip_to(self, doc_num);
|
102
|
+
if (self->doc == doc_num) {
|
103
|
+
tf = ts->freqs[ts->pointer];
|
104
|
+
}
|
105
|
+
return frt_expl_new(frt_sim_tf(self->similarity, (float)tf),
|
106
|
+
"tf(term_freq(%s:%s)=%d)",
|
107
|
+
rb_id2name(TQ(query)->field), TQ(query)->term, tf);
|
108
|
+
}
|
109
|
+
|
110
|
+
static void tsc_destroy(FrtScorer *self)
|
111
|
+
{
|
112
|
+
TSc(self)->tde->close(TSc(self)->tde);
|
113
|
+
frt_scorer_destroy_i(self);
|
114
|
+
}
|
115
|
+
|
116
|
+
static FrtScorer *tsc_new(FrtWeight *weight, FrtTermDocEnum *tde, frt_uchar *norms)
|
117
|
+
{
|
118
|
+
int i;
|
119
|
+
FrtScorer *self = frt_scorer_new(TermScorer, weight->similarity);
|
120
|
+
TSc(self)->weight = weight;
|
121
|
+
TSc(self)->tde = tde;
|
122
|
+
TSc(self)->norms = norms;
|
123
|
+
TSc(self)->weight_value = weight->value;
|
124
|
+
|
125
|
+
for (i = 0; i < SCORE_CACHE_SIZE; i++) {
|
126
|
+
TSc(self)->score_cache[i]
|
127
|
+
= frt_sim_tf(self->similarity, (float)i) * TSc(self)->weight_value;
|
128
|
+
}
|
129
|
+
|
130
|
+
self->score = &tsc_score;
|
131
|
+
self->next = &tsc_next;
|
132
|
+
self->skip_to = &tsc_skip_to;
|
133
|
+
self->explain = &tsc_explain;
|
134
|
+
self->destroy = &tsc_destroy;
|
135
|
+
return self;
|
136
|
+
}
|
137
|
+
|
138
|
+
/***************************************************************************
|
139
|
+
*
|
140
|
+
* TermWeight
|
141
|
+
*
|
142
|
+
***************************************************************************/
|
143
|
+
|
144
|
+
static FrtScorer *tw_scorer(FrtWeight *self, FrtIndexReader *ir)
|
145
|
+
{
|
146
|
+
FrtTermQuery *tq = TQ(self->query);
|
147
|
+
FrtTermDocEnum *tde = ir_term_docs_for(ir, tq->field, tq->term);
|
148
|
+
/* ir_term_docs_for should always return a TermDocEnum */
|
149
|
+
assert(NULL != tde);
|
150
|
+
|
151
|
+
return tsc_new(self, tde, frt_ir_get_norms(ir, tq->field));
|
152
|
+
}
|
153
|
+
|
154
|
+
static FrtExplanation *tw_explain(FrtWeight *self, FrtIndexReader *ir, int doc_num)
|
155
|
+
{
|
156
|
+
FrtExplanation *qnorm_expl;
|
157
|
+
FrtExplanation *field_expl;
|
158
|
+
FrtScorer *scorer;
|
159
|
+
FrtExplanation *tf_expl;
|
160
|
+
frt_uchar *field_norms;
|
161
|
+
float field_norm;
|
162
|
+
FrtExplanation *field_norm_expl;
|
163
|
+
char *query_str = self->query->to_s(self->query, (FrtSymbol)NULL);
|
164
|
+
FrtTermQuery *tq = TQ(self->query);
|
165
|
+
char *term = tq->term;
|
166
|
+
FrtExplanation *expl = frt_expl_new(0.0, "weight(%s in %d), product of:", query_str, doc_num);
|
167
|
+
/* We need two of these as it's included in both the query explanation
|
168
|
+
* and the field explanation */
|
169
|
+
FrtExplanation *idf_expl1 = frt_expl_new(self->idf, "idf(doc_freq=%d)", frt_ir_doc_freq(ir, tq->field, term));
|
170
|
+
FrtExplanation *idf_expl2 = frt_expl_new(self->idf, "idf(doc_freq=%d)", frt_ir_doc_freq(ir, tq->field, term));
|
171
|
+
/* explain query weight */
|
172
|
+
FrtExplanation *query_expl = frt_expl_new(0.0, "query_weight(%s), product of:", query_str);
|
173
|
+
free(query_str);
|
174
|
+
if (self->query->boost != 1.0) {
|
175
|
+
frt_expl_add_detail(query_expl, frt_expl_new(self->query->boost, "boost"));
|
176
|
+
}
|
177
|
+
frt_expl_add_detail(query_expl, idf_expl1);
|
178
|
+
qnorm_expl = frt_expl_new(self->qnorm, "query_norm");
|
179
|
+
frt_expl_add_detail(query_expl, qnorm_expl);
|
180
|
+
query_expl->value = self->query->boost
|
181
|
+
* idf_expl1->value * qnorm_expl->value;
|
182
|
+
frt_expl_add_detail(expl, query_expl);
|
183
|
+
/* explain field weight */
|
184
|
+
field_expl = frt_expl_new(0.0, "field_weight(%s:%s in %d), product of:", rb_id2name(tq->field), term, doc_num);
|
185
|
+
scorer = self->scorer(self, ir);
|
186
|
+
tf_expl = scorer->explain(scorer, doc_num);
|
187
|
+
scorer->destroy(scorer);
|
188
|
+
frt_expl_add_detail(field_expl, tf_expl);
|
189
|
+
frt_expl_add_detail(field_expl, idf_expl2);
|
190
|
+
field_norms = frt_ir_get_norms(ir, tq->field);
|
191
|
+
field_norm = (field_norms ? frt_sim_decode_norm(self->similarity, field_norms[doc_num]) : (float)0.0);
|
192
|
+
field_norm_expl = frt_expl_new(field_norm, "field_norm(field=%s, doc=%d)", rb_id2name(tq->field), doc_num);
|
193
|
+
frt_expl_add_detail(field_expl, field_norm_expl);
|
194
|
+
field_expl->value = tf_expl->value * idf_expl2->value * field_norm_expl->value;
|
195
|
+
/* combine them */
|
196
|
+
if (query_expl->value == 1.0) {
|
197
|
+
frt_expl_destroy(expl);
|
198
|
+
return field_expl;
|
199
|
+
} else {
|
200
|
+
expl->value = (query_expl->value * field_expl->value);
|
201
|
+
frt_expl_add_detail(expl, field_expl);
|
202
|
+
return expl;
|
203
|
+
}
|
204
|
+
}
|
205
|
+
|
206
|
+
static char *tw_to_s(FrtWeight *self)
|
207
|
+
{
|
208
|
+
return frt_strfmt("TermWeight(%f)", self->value);
|
209
|
+
}
|
210
|
+
|
211
|
+
static FrtWeight *tw_new(FrtQuery *query, FrtSearcher *searcher)
|
212
|
+
{
|
213
|
+
FrtWeight *self = w_new(FrtWeight, query);
|
214
|
+
self->scorer = &tw_scorer;
|
215
|
+
self->explain = &tw_explain;
|
216
|
+
self->to_s = &tw_to_s;
|
217
|
+
|
218
|
+
self->similarity = query->get_similarity(query, searcher);
|
219
|
+
self->idf = frt_sim_idf(self->similarity,
|
220
|
+
searcher->doc_freq(searcher,
|
221
|
+
TQ(query)->field,
|
222
|
+
TQ(query)->term),
|
223
|
+
searcher->max_doc(searcher)); /* compute idf */
|
224
|
+
|
225
|
+
return self;
|
226
|
+
}
|
227
|
+
|
228
|
+
/***************************************************************************
|
229
|
+
*
|
230
|
+
* TermQuery
|
231
|
+
*
|
232
|
+
***************************************************************************/
|
233
|
+
|
234
|
+
static void tq_destroy(FrtQuery *self)
|
235
|
+
{
|
236
|
+
free(TQ(self)->term);
|
237
|
+
frt_q_destroy_i(self);
|
238
|
+
}
|
239
|
+
|
240
|
+
static char *tq_to_s(FrtQuery *self, FrtSymbol default_field)
|
241
|
+
{
|
242
|
+
const char *field_name = rb_id2name(TQ(self)->field);
|
243
|
+
size_t flen = strlen(field_name);
|
244
|
+
const char *term = TQ(self)->term;
|
245
|
+
size_t tlen = strlen(term);
|
246
|
+
char *buffer = FRT_ALLOC_N(char, 34 + flen + tlen);
|
247
|
+
char *b = buffer;
|
248
|
+
if (default_field != TQ(self)->field) {
|
249
|
+
memcpy(b, field_name, sizeof(char) * flen);
|
250
|
+
b[flen] = ':';
|
251
|
+
b += flen + 1;
|
252
|
+
}
|
253
|
+
memcpy(b, term, tlen);
|
254
|
+
b += tlen;
|
255
|
+
*b = 0;
|
256
|
+
if (self->boost != 1.0) {
|
257
|
+
*b = '^';
|
258
|
+
frt_dbl_to_s(b+1, self->boost);
|
259
|
+
}
|
260
|
+
return buffer;
|
261
|
+
}
|
262
|
+
|
263
|
+
static void tq_extract_terms(FrtQuery *self, FrtHashSet *terms)
|
264
|
+
{
|
265
|
+
frt_hs_add(terms, frt_term_new(TQ(self)->field, TQ(self)->term));
|
266
|
+
}
|
267
|
+
|
268
|
+
static unsigned long long tq_hash(FrtQuery *self)
|
269
|
+
{
|
270
|
+
return frt_str_hash(TQ(self)->term) ^ frt_str_hash(rb_id2name(TQ(self)->field));
|
271
|
+
}
|
272
|
+
|
273
|
+
static int tq_eq(FrtQuery *self, FrtQuery *o)
|
274
|
+
{
|
275
|
+
return (strcmp(TQ(self)->term, TQ(o)->term) == 0)
|
276
|
+
&& (TQ(self)->field == TQ(o)->field);
|
277
|
+
}
|
278
|
+
|
279
|
+
static FrtMatchVector *tq_get_matchv_i(FrtQuery *self, FrtMatchVector *mv,
|
280
|
+
FrtTermVector *tv)
|
281
|
+
{
|
282
|
+
if (tv->field == TQ(self)->field) {
|
283
|
+
int i;
|
284
|
+
FrtTVTerm *tv_term = frt_tv_get_tv_term(tv, TQ(self)->term);
|
285
|
+
if (tv_term) {
|
286
|
+
for (i = 0; i < tv_term->freq; i++) {
|
287
|
+
int pos = tv_term->positions[i];
|
288
|
+
frt_matchv_add(mv, pos, pos);
|
289
|
+
}
|
290
|
+
}
|
291
|
+
}
|
292
|
+
return mv;
|
293
|
+
}
|
294
|
+
|
295
|
+
FrtQuery *frt_tq_new(FrtSymbol field, const char *term)
|
296
|
+
{
|
297
|
+
FrtQuery *self = frt_q_new(FrtTermQuery);
|
298
|
+
TQ(self)->field = field;
|
299
|
+
TQ(self)->term = frt_estrdup(term);
|
300
|
+
self->type = TERM_QUERY;
|
301
|
+
self->extract_terms = &tq_extract_terms;
|
302
|
+
self->to_s = &tq_to_s;
|
303
|
+
self->hash = &tq_hash;
|
304
|
+
self->eq = &tq_eq;
|
305
|
+
|
306
|
+
self->destroy_i = &tq_destroy;
|
307
|
+
self->create_weight_i = &tw_new;
|
308
|
+
self->get_matchv_i = &tq_get_matchv_i;
|
309
|
+
|
310
|
+
return self;
|
311
|
+
}
|
@@ -0,0 +1,166 @@
|
|
1
|
+
#include <string.h>
|
2
|
+
#include "frt_global.h"
|
3
|
+
#include "frt_search.h"
|
4
|
+
|
5
|
+
/****************************************************************************
|
6
|
+
*
|
7
|
+
* FrtWildCardQuery
|
8
|
+
*
|
9
|
+
****************************************************************************/
|
10
|
+
|
11
|
+
#define WCQ(query) ((FrtWildCardQuery *)(query))
|
12
|
+
|
13
|
+
static char *wcq_to_s(FrtQuery *self, FrtSymbol default_field)
|
14
|
+
{
|
15
|
+
char *buffer, *bptr;
|
16
|
+
const char *field_name = rb_id2name(WCQ(self)->field);
|
17
|
+
const char *pattern = WCQ(self)->pattern;
|
18
|
+
bptr = buffer = FRT_ALLOC_N(char, strlen(pattern) + strlen(field_name) + 35);
|
19
|
+
|
20
|
+
if (WCQ(self)->field != default_field) {
|
21
|
+
bptr += sprintf(bptr, "%s:", field_name);
|
22
|
+
}
|
23
|
+
bptr += sprintf(bptr, "%s", pattern);
|
24
|
+
|
25
|
+
if (self->boost != 1.0) {
|
26
|
+
*bptr = '^';
|
27
|
+
frt_dbl_to_s(++bptr, self->boost);
|
28
|
+
}
|
29
|
+
|
30
|
+
return buffer;
|
31
|
+
}
|
32
|
+
|
33
|
+
bool frt_wc_match(const char *pattern, const char *text)
|
34
|
+
{
|
35
|
+
const char *p = pattern, *t = text, *xt;
|
36
|
+
|
37
|
+
/* include '\0' as we need to match empty string */
|
38
|
+
const char *text_last = t + strlen(t);
|
39
|
+
|
40
|
+
for (;; p++, t++) {
|
41
|
+
|
42
|
+
/* end of text so make sure end of pattern doesn't matter */
|
43
|
+
if (*t == '\0') {
|
44
|
+
while (*p) {
|
45
|
+
if (*p != FRT_WILD_STRING) {
|
46
|
+
return false;
|
47
|
+
}
|
48
|
+
p++;
|
49
|
+
}
|
50
|
+
return true;
|
51
|
+
}
|
52
|
+
|
53
|
+
/* If we've gone past the end of the pattern, return false. */
|
54
|
+
if (*p == '\0') {
|
55
|
+
return false;
|
56
|
+
}
|
57
|
+
|
58
|
+
/* Match a single character, so continue. */
|
59
|
+
if (*p == FRT_WILD_CHAR) {
|
60
|
+
continue;
|
61
|
+
}
|
62
|
+
|
63
|
+
if (*p == FRT_WILD_STRING) {
|
64
|
+
/* Look at the character beyond the '*'. */
|
65
|
+
p++;
|
66
|
+
/* Examine the string, starting at the last character. */
|
67
|
+
for (xt = text_last; xt >= t; xt--) {
|
68
|
+
if (frt_wc_match(p, xt)) return true;
|
69
|
+
}
|
70
|
+
return false;
|
71
|
+
}
|
72
|
+
if (*p != *t) {
|
73
|
+
return false;
|
74
|
+
}
|
75
|
+
}
|
76
|
+
|
77
|
+
return false;
|
78
|
+
}
|
79
|
+
|
80
|
+
static FrtQuery *wcq_rewrite(FrtQuery *self, FrtIndexReader *ir)
|
81
|
+
{
|
82
|
+
FrtQuery *q;
|
83
|
+
const char *pattern = WCQ(self)->pattern;
|
84
|
+
const char *first_star = strchr(pattern, FRT_WILD_STRING);
|
85
|
+
const char *first_ques = strchr(pattern, FRT_WILD_CHAR);
|
86
|
+
|
87
|
+
if (NULL == first_star && NULL == first_ques) {
|
88
|
+
q = frt_tq_new(WCQ(self)->field, pattern);
|
89
|
+
q->boost = self->boost;
|
90
|
+
}
|
91
|
+
else {
|
92
|
+
const int field_num = frt_fis_get_field_num(ir->fis, WCQ(self)->field);
|
93
|
+
q = frt_multi_tq_new_conf(WCQ(self)->field, FrtMTQMaxTerms(self), 0.0);
|
94
|
+
|
95
|
+
if (field_num >= 0) {
|
96
|
+
FrtTermEnum *te;
|
97
|
+
char prefix[FRT_MAX_WORD_SIZE] = "";
|
98
|
+
int prefix_len;
|
99
|
+
|
100
|
+
pattern = (first_ques && (!first_star || first_star > first_ques))
|
101
|
+
? first_ques : first_star;
|
102
|
+
|
103
|
+
prefix_len = (int)(pattern - WCQ(self)->pattern);
|
104
|
+
|
105
|
+
if (prefix_len > 0) {
|
106
|
+
memcpy(prefix, WCQ(self)->pattern, prefix_len);
|
107
|
+
prefix[prefix_len] = '\0';
|
108
|
+
}
|
109
|
+
|
110
|
+
te = ir->terms_from(ir, field_num, prefix);
|
111
|
+
|
112
|
+
if (te != NULL) {
|
113
|
+
const char *term = te->curr_term;
|
114
|
+
const char *pat_term = term + prefix_len;
|
115
|
+
do {
|
116
|
+
if (prefix[0] && strncmp(term, prefix, prefix_len) != 0) {
|
117
|
+
break;
|
118
|
+
}
|
119
|
+
|
120
|
+
if (frt_wc_match(pattern, pat_term)) {
|
121
|
+
frt_multi_tq_add_term(q, term);
|
122
|
+
}
|
123
|
+
} while (te->next(te) != NULL);
|
124
|
+
te->close(te);
|
125
|
+
}
|
126
|
+
}
|
127
|
+
}
|
128
|
+
|
129
|
+
return q;
|
130
|
+
}
|
131
|
+
|
132
|
+
static void wcq_destroy(FrtQuery *self)
|
133
|
+
{
|
134
|
+
free(WCQ(self)->pattern);
|
135
|
+
frt_q_destroy_i(self);
|
136
|
+
}
|
137
|
+
|
138
|
+
static unsigned long long wcq_hash(FrtQuery *self)
|
139
|
+
{
|
140
|
+
return frt_str_hash(rb_id2name(WCQ(self)->field)) ^ frt_str_hash(WCQ(self)->pattern);
|
141
|
+
}
|
142
|
+
|
143
|
+
static int wcq_eq(FrtQuery *self, FrtQuery *o)
|
144
|
+
{
|
145
|
+
return (strcmp(WCQ(self)->pattern, WCQ(o)->pattern) == 0)
|
146
|
+
&& (WCQ(self)->field == WCQ(o)->field);
|
147
|
+
}
|
148
|
+
|
149
|
+
FrtQuery *frt_wcq_new(FrtSymbol field, const char *pattern)
|
150
|
+
{
|
151
|
+
FrtQuery *self = frt_q_new(FrtWildCardQuery);
|
152
|
+
|
153
|
+
WCQ(self)->field = field;
|
154
|
+
WCQ(self)->pattern = frt_estrdup(pattern);
|
155
|
+
FrtMTQMaxTerms(self) = FRT_WILD_CARD_QUERY_MAX_TERMS;
|
156
|
+
|
157
|
+
self->type = WILD_CARD_QUERY;
|
158
|
+
self->rewrite = &wcq_rewrite;
|
159
|
+
self->to_s = &wcq_to_s;
|
160
|
+
self->hash = &wcq_hash;
|
161
|
+
self->eq = &wcq_eq;
|
162
|
+
self->destroy_i = &wcq_destroy;
|
163
|
+
self->create_weight_i = &frt_q_create_weight_unsup;
|
164
|
+
|
165
|
+
return self;
|
166
|
+
}
|