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 +1,2386 @@
|
|
1
|
+
#include <string.h>
|
2
|
+
#include <limits.h>
|
3
|
+
#include "frt_global.h"
|
4
|
+
#include "frt_search.h"
|
5
|
+
#include "frt_hashset.h"
|
6
|
+
|
7
|
+
#define CLAUSE_INIT_CAPA 4
|
8
|
+
|
9
|
+
/*****************************************************************************
|
10
|
+
*
|
11
|
+
* SpanQuery
|
12
|
+
*
|
13
|
+
*****************************************************************************/
|
14
|
+
|
15
|
+
#define SpQ(query) ((FrtSpanQuery *)(query))
|
16
|
+
|
17
|
+
static unsigned long long spanq_hash(FrtQuery *self)
|
18
|
+
{
|
19
|
+
return SpQ(self)->field ? frt_str_hash(rb_id2name(SpQ(self)->field)) : 0;
|
20
|
+
}
|
21
|
+
|
22
|
+
static int spanq_eq(FrtQuery *self, FrtQuery *o)
|
23
|
+
{
|
24
|
+
return SpQ(self)->field == SpQ(o)->field;
|
25
|
+
}
|
26
|
+
|
27
|
+
static void spanq_destroy_i(FrtQuery *self)
|
28
|
+
{
|
29
|
+
frt_q_destroy_i(self);
|
30
|
+
}
|
31
|
+
|
32
|
+
static FrtMatchVector *mv_to_term_mv(FrtMatchVector *term_mv, FrtMatchVector *full_mv,
|
33
|
+
FrtHashSet *terms, FrtTermVector *tv)
|
34
|
+
{
|
35
|
+
FrtHashSetEntry *hse;
|
36
|
+
for (hse = terms->first; hse; hse = hse->next) {
|
37
|
+
char *term = (char *)hse->elem;
|
38
|
+
FrtTVTerm *tv_term = frt_tv_get_tv_term(tv, term);
|
39
|
+
if (tv_term) {
|
40
|
+
int i, m_idx = 0;
|
41
|
+
for (i = 0; i < tv_term->freq; i++) {
|
42
|
+
int pos = tv_term->positions[i];
|
43
|
+
for (; m_idx < full_mv->size; m_idx++) {
|
44
|
+
if (pos <= full_mv->matches[m_idx].end) {
|
45
|
+
if (pos >= full_mv->matches[m_idx].start) {
|
46
|
+
frt_matchv_add(term_mv, pos, pos);
|
47
|
+
}
|
48
|
+
break;
|
49
|
+
}
|
50
|
+
}
|
51
|
+
}
|
52
|
+
}
|
53
|
+
}
|
54
|
+
|
55
|
+
return term_mv;
|
56
|
+
}
|
57
|
+
|
58
|
+
/***************************************************************************
|
59
|
+
* TVTermDocEnum
|
60
|
+
* dummy TermDocEnum used by the highlighter to find matches
|
61
|
+
***************************************************************************/
|
62
|
+
|
63
|
+
#define TV_TDE(tde) ((TVTermDocEnum *)(tde))
|
64
|
+
|
65
|
+
typedef struct TVTermDocEnum
|
66
|
+
{
|
67
|
+
FrtTermDocEnum super;
|
68
|
+
int doc;
|
69
|
+
int index;
|
70
|
+
int freq;
|
71
|
+
int *positions;
|
72
|
+
FrtTermVector *tv;
|
73
|
+
} TVTermDocEnum;
|
74
|
+
|
75
|
+
static void tv_tde_seek(FrtTermDocEnum *tde, int field_num, const char *term)
|
76
|
+
{
|
77
|
+
TVTermDocEnum *tv_tde = TV_TDE(tde);
|
78
|
+
FrtTVTerm *tv_term = frt_tv_get_tv_term(tv_tde->tv, term);
|
79
|
+
(void)field_num;
|
80
|
+
if (tv_term) {
|
81
|
+
tv_tde->doc = -1;
|
82
|
+
tv_tde->index = 0;
|
83
|
+
tv_tde->freq = tv_term->freq;
|
84
|
+
tv_tde->positions = tv_term->positions;
|
85
|
+
}
|
86
|
+
else {
|
87
|
+
tv_tde->doc = INT_MAX;
|
88
|
+
}
|
89
|
+
}
|
90
|
+
|
91
|
+
static bool tv_tde_next(FrtTermDocEnum *tde)
|
92
|
+
{
|
93
|
+
if (TV_TDE(tde)->doc == -1) {
|
94
|
+
TV_TDE(tde)->doc = 0;
|
95
|
+
return true;
|
96
|
+
}
|
97
|
+
else {
|
98
|
+
TV_TDE(tde)->doc = INT_MAX;
|
99
|
+
return false;
|
100
|
+
}
|
101
|
+
}
|
102
|
+
|
103
|
+
static bool tv_tde_skip_to(FrtTermDocEnum *tde, int doc_num)
|
104
|
+
{
|
105
|
+
if (doc_num == 0) {
|
106
|
+
TV_TDE(tde)->doc = 0;
|
107
|
+
return true;
|
108
|
+
}
|
109
|
+
else {
|
110
|
+
TV_TDE(tde)->doc = INT_MAX;
|
111
|
+
return false;
|
112
|
+
}
|
113
|
+
}
|
114
|
+
|
115
|
+
static int tv_tde_next_position(FrtTermDocEnum *tde)
|
116
|
+
{
|
117
|
+
return TV_TDE(tde)->positions[TV_TDE(tde)->index++];
|
118
|
+
}
|
119
|
+
|
120
|
+
static int tv_tde_freq(FrtTermDocEnum *tde)
|
121
|
+
{
|
122
|
+
return TV_TDE(tde)->freq;
|
123
|
+
}
|
124
|
+
|
125
|
+
static int tv_tde_doc_num(FrtTermDocEnum *tde)
|
126
|
+
{
|
127
|
+
return TV_TDE(tde)->doc;
|
128
|
+
}
|
129
|
+
|
130
|
+
static FrtTermDocEnum *spanq_ir_term_positions(FrtIndexReader *ir)
|
131
|
+
{
|
132
|
+
TVTermDocEnum *tv_tde = FRT_ALLOC(TVTermDocEnum);
|
133
|
+
FrtTermDocEnum *tde = (FrtTermDocEnum *)tv_tde;
|
134
|
+
tv_tde->tv = (FrtTermVector *)ir->store;
|
135
|
+
tde->seek = &tv_tde_seek;
|
136
|
+
tde->doc_num = &tv_tde_doc_num;
|
137
|
+
tde->freq = &tv_tde_freq;
|
138
|
+
tde->next = &tv_tde_next;
|
139
|
+
tde->skip_to = &tv_tde_skip_to;
|
140
|
+
tde->next_position = &tv_tde_next_position;
|
141
|
+
tde->close = (void (*)(FrtTermDocEnum *tde))&free;
|
142
|
+
|
143
|
+
return tde;
|
144
|
+
}
|
145
|
+
|
146
|
+
static FrtMatchVector *spanq_get_matchv_i(FrtQuery *self, FrtMatchVector *mv,
|
147
|
+
FrtTermVector *tv)
|
148
|
+
{
|
149
|
+
if (SpQ(self)->field == tv->field) {
|
150
|
+
FrtSpanEnum *sp_enum;
|
151
|
+
FrtIndexReader *ir = FRT_ALLOC(FrtIndexReader);
|
152
|
+
FrtMatchVector *full_mv = frt_matchv_new();
|
153
|
+
FrtHashSet *terms = SpQ(self)->get_terms(self);
|
154
|
+
/* FIXME What is going on here? Need to document this! */
|
155
|
+
ir->fis = frt_fis_new(FRT_STORE_NO, FRT_INDEX_NO, FRT_TERM_VECTOR_NO);
|
156
|
+
frt_fis_add_field(ir->fis,
|
157
|
+
frt_fi_new(tv->field, FRT_STORE_NO, FRT_INDEX_NO, FRT_TERM_VECTOR_NO));
|
158
|
+
ir->store = (FrtStore *)tv;
|
159
|
+
ir->term_positions = &spanq_ir_term_positions;
|
160
|
+
sp_enum = SpQ(self)->get_spans(self, ir);
|
161
|
+
while (sp_enum->next(sp_enum)) {
|
162
|
+
frt_matchv_add(full_mv,
|
163
|
+
sp_enum->start(sp_enum),
|
164
|
+
sp_enum->end(sp_enum) - 1);
|
165
|
+
}
|
166
|
+
sp_enum->destroy(sp_enum);
|
167
|
+
|
168
|
+
frt_fis_deref(ir->fis);
|
169
|
+
free(ir);
|
170
|
+
|
171
|
+
frt_matchv_compact(full_mv);
|
172
|
+
mv_to_term_mv(mv, full_mv, terms, tv);
|
173
|
+
frt_matchv_destroy(full_mv);
|
174
|
+
frt_hs_destroy(terms);
|
175
|
+
}
|
176
|
+
return mv;
|
177
|
+
}
|
178
|
+
|
179
|
+
/***************************************************************************
|
180
|
+
*
|
181
|
+
* SpanScorer
|
182
|
+
*
|
183
|
+
***************************************************************************/
|
184
|
+
|
185
|
+
#define SpSc(scorer) ((SpanScorer *)(scorer))
|
186
|
+
typedef struct SpanScorer
|
187
|
+
{
|
188
|
+
FrtScorer super;
|
189
|
+
FrtIndexReader *ir;
|
190
|
+
FrtSpanEnum *spans;
|
191
|
+
FrtSimilarity *sim;
|
192
|
+
frt_uchar *norms;
|
193
|
+
FrtWeight *weight;
|
194
|
+
float value;
|
195
|
+
float freq;
|
196
|
+
bool first_time : 1;
|
197
|
+
bool more : 1;
|
198
|
+
} SpanScorer;
|
199
|
+
|
200
|
+
static float spansc_score(FrtScorer *self)
|
201
|
+
{
|
202
|
+
SpanScorer *spansc = SpSc(self);
|
203
|
+
float raw = frt_sim_tf(spansc->sim, spansc->freq) * spansc->value;
|
204
|
+
|
205
|
+
/* normalize */
|
206
|
+
return raw * frt_sim_decode_norm(self->similarity, spansc->norms[self->doc]);
|
207
|
+
}
|
208
|
+
|
209
|
+
static bool spansc_next(FrtScorer *self)
|
210
|
+
{
|
211
|
+
SpanScorer *spansc = SpSc(self);
|
212
|
+
FrtSpanEnum *se = spansc->spans;
|
213
|
+
int match_length;
|
214
|
+
|
215
|
+
if (spansc->first_time) {
|
216
|
+
spansc->more = se->next(se);
|
217
|
+
spansc->first_time = false;
|
218
|
+
}
|
219
|
+
|
220
|
+
if (!spansc->more) {
|
221
|
+
return false;
|
222
|
+
}
|
223
|
+
|
224
|
+
spansc->freq = 0.0f;
|
225
|
+
self->doc = se->doc(se);
|
226
|
+
|
227
|
+
do {
|
228
|
+
match_length = se->end(se) - se->start(se);
|
229
|
+
spansc->freq += frt_sim_sloppy_freq(spansc->sim, match_length);
|
230
|
+
spansc->more = se->next(se);
|
231
|
+
} while (spansc->more && (self->doc == se->doc(se)));
|
232
|
+
|
233
|
+
return (spansc->more || (spansc->freq != 0.0));
|
234
|
+
}
|
235
|
+
|
236
|
+
static bool spansc_skip_to(FrtScorer *self, int target)
|
237
|
+
{
|
238
|
+
SpanScorer *spansc = SpSc(self);
|
239
|
+
FrtSpanEnum *se = spansc->spans;
|
240
|
+
|
241
|
+
spansc->more = se->skip_to(se, target);
|
242
|
+
if (!spansc->more) {
|
243
|
+
return false;
|
244
|
+
}
|
245
|
+
|
246
|
+
spansc->freq = 0.0f;
|
247
|
+
self->doc = se->doc(se);
|
248
|
+
|
249
|
+
while (spansc->more && (se->doc(se) == target)) {
|
250
|
+
spansc->freq += frt_sim_sloppy_freq(spansc->sim, se->end(se) - se->start(se));
|
251
|
+
spansc->more = se->next(se);
|
252
|
+
if (spansc->first_time) {
|
253
|
+
spansc->first_time = false;
|
254
|
+
}
|
255
|
+
}
|
256
|
+
|
257
|
+
return (spansc->more || (spansc->freq != 0.0));
|
258
|
+
}
|
259
|
+
|
260
|
+
static FrtExplanation *spansc_explain(FrtScorer *self, int target)
|
261
|
+
{
|
262
|
+
FrtExplanation *tf_explanation;
|
263
|
+
SpanScorer *spansc = SpSc(self);
|
264
|
+
float phrase_freq;
|
265
|
+
self->skip_to(self, target);
|
266
|
+
phrase_freq = (self->doc == target) ? spansc->freq : (float)0.0;
|
267
|
+
|
268
|
+
tf_explanation = frt_expl_new(frt_sim_tf(self->similarity, phrase_freq),
|
269
|
+
"tf(phrase_freq(%f)", phrase_freq);
|
270
|
+
|
271
|
+
return tf_explanation;
|
272
|
+
}
|
273
|
+
|
274
|
+
static void spansc_destroy(FrtScorer *self)
|
275
|
+
{
|
276
|
+
SpanScorer *spansc = SpSc(self);
|
277
|
+
if (spansc->spans) {
|
278
|
+
spansc->spans->destroy(spansc->spans);
|
279
|
+
}
|
280
|
+
frt_scorer_destroy_i(self);
|
281
|
+
}
|
282
|
+
|
283
|
+
static FrtScorer *spansc_new(FrtWeight *weight, FrtIndexReader *ir)
|
284
|
+
{
|
285
|
+
FrtScorer *self = NULL;
|
286
|
+
const int field_num = frt_fis_get_field_num(ir->fis, SpQ(weight->query)->field);
|
287
|
+
if (field_num >= 0) {
|
288
|
+
FrtQuery *spanq = weight->query;
|
289
|
+
self = frt_scorer_new(SpanScorer, weight->similarity);
|
290
|
+
|
291
|
+
SpSc(self)->first_time = true;
|
292
|
+
SpSc(self)->more = true;
|
293
|
+
SpSc(self)->spans = SpQ(spanq)->get_spans(spanq, ir);
|
294
|
+
SpSc(self)->sim = weight->similarity;
|
295
|
+
SpSc(self)->norms = ir->get_norms(ir, field_num);
|
296
|
+
SpSc(self)->weight = weight;
|
297
|
+
SpSc(self)->value = weight->value;
|
298
|
+
SpSc(self)->freq = 0.0f;
|
299
|
+
|
300
|
+
self->score = &spansc_score;
|
301
|
+
self->next = &spansc_next;
|
302
|
+
self->skip_to = &spansc_skip_to;
|
303
|
+
self->explain = &spansc_explain;
|
304
|
+
self->destroy = &spansc_destroy;
|
305
|
+
}
|
306
|
+
return self;
|
307
|
+
}
|
308
|
+
|
309
|
+
/*****************************************************************************
|
310
|
+
* SpanTermEnum
|
311
|
+
*****************************************************************************/
|
312
|
+
|
313
|
+
#define SpTEn(span_enum) ((SpanTermEnum *)(span_enum))
|
314
|
+
#define SpTQ(query) ((FrtSpanTermQuery *)(query))
|
315
|
+
|
316
|
+
typedef struct SpanTermEnum
|
317
|
+
{
|
318
|
+
FrtSpanEnum super;
|
319
|
+
FrtTermDocEnum *positions;
|
320
|
+
int position;
|
321
|
+
int doc;
|
322
|
+
int count;
|
323
|
+
int freq;
|
324
|
+
} SpanTermEnum;
|
325
|
+
|
326
|
+
|
327
|
+
static bool spante_next(FrtSpanEnum *self)
|
328
|
+
{
|
329
|
+
SpanTermEnum *ste = SpTEn(self);
|
330
|
+
FrtTermDocEnum *tde = ste->positions;
|
331
|
+
|
332
|
+
if (ste->count == ste->freq) {
|
333
|
+
if (! tde->next(tde)) {
|
334
|
+
ste->doc = INT_MAX;
|
335
|
+
return false;
|
336
|
+
}
|
337
|
+
ste->doc = tde->doc_num(tde);
|
338
|
+
ste->freq = tde->freq(tde);
|
339
|
+
ste->count = 0;
|
340
|
+
}
|
341
|
+
ste->position = tde->next_position(tde);
|
342
|
+
ste->count++;
|
343
|
+
return true;
|
344
|
+
}
|
345
|
+
|
346
|
+
static bool spante_skip_to(FrtSpanEnum *self, int target)
|
347
|
+
{
|
348
|
+
SpanTermEnum *ste = SpTEn(self);
|
349
|
+
FrtTermDocEnum *tde = ste->positions;
|
350
|
+
|
351
|
+
/* are we already at the correct position? */
|
352
|
+
/* FIXME: perhaps this the the better solution but currently it ->skip_to
|
353
|
+
* does a ->next not matter what
|
354
|
+
if (ste->doc >= target) {
|
355
|
+
return true;
|
356
|
+
}
|
357
|
+
*/
|
358
|
+
|
359
|
+
if (! tde->skip_to(tde, target)) {
|
360
|
+
ste->doc = INT_MAX;
|
361
|
+
return false;
|
362
|
+
}
|
363
|
+
|
364
|
+
ste->doc = tde->doc_num(tde);
|
365
|
+
ste->freq = tde->freq(tde);
|
366
|
+
ste->count = 0;
|
367
|
+
|
368
|
+
ste->position = tde->next_position(tde);
|
369
|
+
ste->count++;
|
370
|
+
return true;
|
371
|
+
}
|
372
|
+
|
373
|
+
static int spante_doc(FrtSpanEnum *self)
|
374
|
+
{
|
375
|
+
return SpTEn(self)->doc;
|
376
|
+
}
|
377
|
+
|
378
|
+
static int spante_start(FrtSpanEnum *self)
|
379
|
+
{
|
380
|
+
return SpTEn(self)->position;
|
381
|
+
}
|
382
|
+
|
383
|
+
static int spante_end(FrtSpanEnum *self)
|
384
|
+
{
|
385
|
+
return SpTEn(self)->position + 1;
|
386
|
+
}
|
387
|
+
|
388
|
+
static char *spante_to_s(FrtSpanEnum *self)
|
389
|
+
{
|
390
|
+
char *query_str = self->query->to_s(self->query, (FrtSymbol)NULL);
|
391
|
+
char pos_str[20];
|
392
|
+
size_t len = strlen(query_str);
|
393
|
+
int pos;
|
394
|
+
char *str = FRT_ALLOC_N(char, len + 40);
|
395
|
+
|
396
|
+
if (self->doc(self) < 0) {
|
397
|
+
sprintf(pos_str, "START");
|
398
|
+
}
|
399
|
+
else {
|
400
|
+
if (self->doc(self) == INT_MAX) {
|
401
|
+
sprintf(pos_str, "END");
|
402
|
+
}
|
403
|
+
else {
|
404
|
+
pos = SpTEn(self)->position;
|
405
|
+
sprintf(pos_str, "%d", self->doc(self) - pos);
|
406
|
+
}
|
407
|
+
}
|
408
|
+
sprintf(str, "SpanTermEnum(%s)@%s", query_str, pos_str);
|
409
|
+
free(query_str);
|
410
|
+
return str;
|
411
|
+
}
|
412
|
+
|
413
|
+
static void spante_destroy(FrtSpanEnum *self)
|
414
|
+
{
|
415
|
+
FrtTermDocEnum *tde = SpTEn(self)->positions;
|
416
|
+
tde->close(tde);
|
417
|
+
free(self);
|
418
|
+
}
|
419
|
+
|
420
|
+
static FrtSpanEnum *spante_new(FrtQuery *query, FrtIndexReader *ir)
|
421
|
+
{
|
422
|
+
char *term = SpTQ(query)->term;
|
423
|
+
FrtSpanEnum *self = (FrtSpanEnum *)FRT_ALLOC(SpanTermEnum);
|
424
|
+
|
425
|
+
SpTEn(self)->positions = frt_ir_term_positions_for(ir, SpQ(query)->field,
|
426
|
+
term);
|
427
|
+
SpTEn(self)->position = -1;
|
428
|
+
SpTEn(self)->doc = -1;
|
429
|
+
SpTEn(self)->count = 0;
|
430
|
+
SpTEn(self)->freq = 0;
|
431
|
+
|
432
|
+
self->query = query;
|
433
|
+
self->next = &spante_next;
|
434
|
+
self->skip_to = &spante_skip_to;
|
435
|
+
self->doc = &spante_doc;
|
436
|
+
self->start = &spante_start;
|
437
|
+
self->end = &spante_end;
|
438
|
+
self->destroy = &spante_destroy;
|
439
|
+
self->to_s = &spante_to_s;
|
440
|
+
|
441
|
+
return self;
|
442
|
+
}
|
443
|
+
|
444
|
+
/*****************************************************************************
|
445
|
+
* SpanMultiTermEnum
|
446
|
+
*****************************************************************************/
|
447
|
+
|
448
|
+
/* * TermPosEnumWrapper * */
|
449
|
+
#define TPE_READ_SIZE 16
|
450
|
+
|
451
|
+
typedef struct TermPosEnumWrapper
|
452
|
+
{
|
453
|
+
const char *term;
|
454
|
+
FrtTermDocEnum *tpe;
|
455
|
+
int doc;
|
456
|
+
int pos;
|
457
|
+
} TermPosEnumWrapper;
|
458
|
+
|
459
|
+
static bool tpew_less_than(const TermPosEnumWrapper *tpew1,
|
460
|
+
const TermPosEnumWrapper *tpew2)
|
461
|
+
{
|
462
|
+
return (tpew1->doc < tpew2->doc)
|
463
|
+
|| (tpew1->doc == tpew2->doc && tpew1->pos < tpew2->pos);
|
464
|
+
}
|
465
|
+
|
466
|
+
static bool tpew_next(TermPosEnumWrapper *self)
|
467
|
+
{
|
468
|
+
FrtTermDocEnum *tpe = self->tpe;
|
469
|
+
if (0 > (self->pos = tpe->next_position(tpe))) {
|
470
|
+
if (!tpe->next(tpe)) return false;
|
471
|
+
self->doc = tpe->doc_num(tpe);
|
472
|
+
self->pos = tpe->next_position(tpe);
|
473
|
+
}
|
474
|
+
return true;
|
475
|
+
}
|
476
|
+
|
477
|
+
static bool tpew_skip_to(TermPosEnumWrapper *self, int doc_num)
|
478
|
+
{
|
479
|
+
FrtTermDocEnum *tpe = self->tpe;
|
480
|
+
|
481
|
+
if (tpe->skip_to(tpe, doc_num)) {
|
482
|
+
self->doc = tpe->doc_num(tpe);
|
483
|
+
self->pos = tpe->next_position(tpe);
|
484
|
+
return true;
|
485
|
+
}
|
486
|
+
else {
|
487
|
+
return false;
|
488
|
+
}
|
489
|
+
}
|
490
|
+
|
491
|
+
static void tpew_destroy(TermPosEnumWrapper *self)
|
492
|
+
{
|
493
|
+
self->tpe->close(self->tpe);
|
494
|
+
free(self);
|
495
|
+
}
|
496
|
+
|
497
|
+
static TermPosEnumWrapper *tpew_new(const char *term, FrtTermDocEnum *tpe)
|
498
|
+
{
|
499
|
+
TermPosEnumWrapper *self = FRT_ALLOC_AND_ZERO(TermPosEnumWrapper);
|
500
|
+
self->term = term;
|
501
|
+
self->tpe = tpe;
|
502
|
+
self->doc = -1;
|
503
|
+
self->pos = -1;
|
504
|
+
return self;
|
505
|
+
}
|
506
|
+
#define SpMTEn(span_enum) ((SpanMultiTermEnum *)(span_enum))
|
507
|
+
#define SpMTQ(query) ((FrtSpanMultiTermQuery *)(query))
|
508
|
+
|
509
|
+
typedef struct SpanMultiTermEnum
|
510
|
+
{
|
511
|
+
FrtSpanEnum super;
|
512
|
+
FrtPriorityQueue *tpew_pq;
|
513
|
+
TermPosEnumWrapper **tpews;
|
514
|
+
int tpew_cnt;
|
515
|
+
int pos;
|
516
|
+
int doc;
|
517
|
+
} SpanMultiTermEnum;
|
518
|
+
|
519
|
+
static bool spanmte_next(FrtSpanEnum *self)
|
520
|
+
{
|
521
|
+
int curr_doc, curr_pos;
|
522
|
+
TermPosEnumWrapper *tpew;
|
523
|
+
SpanMultiTermEnum *mte = SpMTEn(self);
|
524
|
+
FrtPriorityQueue *tpew_pq = mte->tpew_pq;
|
525
|
+
if (tpew_pq == NULL) {
|
526
|
+
TermPosEnumWrapper **tpews = mte->tpews;
|
527
|
+
int i;
|
528
|
+
tpew_pq = frt_pq_new(mte->tpew_cnt, (frt_lt_ft)tpew_less_than, (frt_free_ft)NULL);
|
529
|
+
for (i = mte->tpew_cnt - 1; i >= 0; i--) {
|
530
|
+
if (tpew_next(tpews[i])) {
|
531
|
+
frt_pq_push(tpew_pq, tpews[i]);
|
532
|
+
}
|
533
|
+
}
|
534
|
+
mte->tpew_pq = tpew_pq;
|
535
|
+
}
|
536
|
+
|
537
|
+
tpew = (TermPosEnumWrapper *)frt_pq_top(tpew_pq);
|
538
|
+
if (tpew == NULL) {
|
539
|
+
return false;
|
540
|
+
}
|
541
|
+
|
542
|
+
mte->doc = curr_doc = tpew->doc;
|
543
|
+
mte->pos = curr_pos = tpew->pos;
|
544
|
+
|
545
|
+
do {
|
546
|
+
if (tpew_next(tpew)) {
|
547
|
+
frt_pq_down(tpew_pq);
|
548
|
+
}
|
549
|
+
else {
|
550
|
+
frt_pq_pop(tpew_pq);
|
551
|
+
}
|
552
|
+
} while (((tpew = (TermPosEnumWrapper *)frt_pq_top(tpew_pq)) != NULL)
|
553
|
+
&& tpew->doc == curr_doc && tpew->pos == curr_pos);
|
554
|
+
return true;
|
555
|
+
}
|
556
|
+
|
557
|
+
static bool spanmte_skip_to(FrtSpanEnum *self, int target)
|
558
|
+
{
|
559
|
+
SpanMultiTermEnum *mte = SpMTEn(self);
|
560
|
+
FrtPriorityQueue *tpew_pq = mte->tpew_pq;
|
561
|
+
TermPosEnumWrapper *tpew;
|
562
|
+
if (tpew_pq == NULL) {
|
563
|
+
TermPosEnumWrapper **tpews = mte->tpews;
|
564
|
+
int i;
|
565
|
+
tpew_pq = frt_pq_new(mte->tpew_cnt, (frt_lt_ft)tpew_less_than, (frt_free_ft)NULL);
|
566
|
+
for (i = mte->tpew_cnt - 1; i >= 0; i--) {
|
567
|
+
tpew_skip_to(tpews[i], target);
|
568
|
+
frt_pq_push(tpew_pq, tpews[i]);
|
569
|
+
}
|
570
|
+
mte->tpew_pq = tpew_pq;
|
571
|
+
}
|
572
|
+
if (tpew_pq->size == 0) {
|
573
|
+
mte->doc = -1;
|
574
|
+
return false;
|
575
|
+
}
|
576
|
+
while ((tpew = (TermPosEnumWrapper *)frt_pq_top(tpew_pq)) != NULL
|
577
|
+
&& (target > tpew->doc)) {
|
578
|
+
if (tpew_skip_to(tpew, target)) {
|
579
|
+
frt_pq_down(tpew_pq);
|
580
|
+
}
|
581
|
+
else {
|
582
|
+
frt_pq_pop(tpew_pq);
|
583
|
+
}
|
584
|
+
}
|
585
|
+
return spanmte_next(self);
|
586
|
+
}
|
587
|
+
|
588
|
+
static int spanmte_doc(FrtSpanEnum *self)
|
589
|
+
{
|
590
|
+
return SpMTEn(self)->doc;
|
591
|
+
}
|
592
|
+
|
593
|
+
static int spanmte_start(FrtSpanEnum *self)
|
594
|
+
{
|
595
|
+
return SpMTEn(self)->pos;
|
596
|
+
}
|
597
|
+
|
598
|
+
static int spanmte_end(FrtSpanEnum *self)
|
599
|
+
{
|
600
|
+
return SpMTEn(self)->pos + 1;
|
601
|
+
}
|
602
|
+
|
603
|
+
static void spanmte_destroy(FrtSpanEnum *self)
|
604
|
+
{
|
605
|
+
SpanMultiTermEnum *mte = SpMTEn(self);
|
606
|
+
int i;
|
607
|
+
if (mte->tpew_pq) frt_pq_destroy(mte->tpew_pq);
|
608
|
+
for (i = 0; i < mte->tpew_cnt; i++) {
|
609
|
+
tpew_destroy(mte->tpews[i]);
|
610
|
+
}
|
611
|
+
free(mte->tpews);
|
612
|
+
free(self);
|
613
|
+
}
|
614
|
+
|
615
|
+
static FrtSpanEnum *spanmte_new(FrtQuery *query, FrtIndexReader *ir)
|
616
|
+
{
|
617
|
+
FrtSpanEnum *self = (FrtSpanEnum *)FRT_ALLOC(SpanMultiTermEnum);
|
618
|
+
SpanMultiTermEnum *smte = SpMTEn(self);
|
619
|
+
FrtSpanMultiTermQuery *smtq = SpMTQ(query);
|
620
|
+
int i;
|
621
|
+
|
622
|
+
|
623
|
+
smte->tpews = FRT_ALLOC_N(TermPosEnumWrapper *, smtq->term_cnt);
|
624
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
625
|
+
char *term = smtq->terms[i];
|
626
|
+
smte->tpews[i] = tpew_new(term,
|
627
|
+
frt_ir_term_positions_for(ir, SpQ(query)->field, term));
|
628
|
+
}
|
629
|
+
smte->tpew_cnt = smtq->term_cnt;
|
630
|
+
smte->tpew_pq = NULL;
|
631
|
+
smte->pos = -1;
|
632
|
+
smte->doc = -1;
|
633
|
+
|
634
|
+
self->query = query;
|
635
|
+
self->next = &spanmte_next;
|
636
|
+
self->skip_to = &spanmte_skip_to;
|
637
|
+
self->doc = &spanmte_doc;
|
638
|
+
self->start = &spanmte_start;
|
639
|
+
self->end = &spanmte_end;
|
640
|
+
self->destroy = &spanmte_destroy;
|
641
|
+
self->to_s = &spante_to_s;
|
642
|
+
|
643
|
+
return self;
|
644
|
+
}
|
645
|
+
|
646
|
+
|
647
|
+
/*****************************************************************************
|
648
|
+
* SpanFirstEnum
|
649
|
+
*****************************************************************************/
|
650
|
+
|
651
|
+
#define SpFEn(span_enum) ((SpanFirstEnum *)(span_enum))
|
652
|
+
#define SpFQ(query) ((FrtSpanFirstQuery *)(query))
|
653
|
+
|
654
|
+
typedef struct SpanFirstEnum
|
655
|
+
{
|
656
|
+
FrtSpanEnum super;
|
657
|
+
FrtSpanEnum *sub_enum;
|
658
|
+
} SpanFirstEnum;
|
659
|
+
|
660
|
+
|
661
|
+
static bool spanfe_next(FrtSpanEnum *self)
|
662
|
+
{
|
663
|
+
FrtSpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
664
|
+
int end = SpFQ(self->query)->end;
|
665
|
+
while (sub_enum->next(sub_enum)) { /* scan to next match */
|
666
|
+
if (sub_enum->end(sub_enum) <= end) {
|
667
|
+
return true;
|
668
|
+
}
|
669
|
+
}
|
670
|
+
return false;
|
671
|
+
}
|
672
|
+
|
673
|
+
static bool spanfe_skip_to(FrtSpanEnum *self, int target)
|
674
|
+
{
|
675
|
+
FrtSpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
676
|
+
int end = SpFQ(self->query)->end;
|
677
|
+
|
678
|
+
if (! sub_enum->skip_to(sub_enum, target)) {
|
679
|
+
return false;
|
680
|
+
}
|
681
|
+
|
682
|
+
if (sub_enum->end(sub_enum) <= end) { /* there is a match */
|
683
|
+
return true;
|
684
|
+
}
|
685
|
+
|
686
|
+
return spanfe_next(self); /* scan to next match */
|
687
|
+
}
|
688
|
+
|
689
|
+
static int spanfe_doc(FrtSpanEnum *self)
|
690
|
+
{
|
691
|
+
FrtSpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
692
|
+
return sub_enum->doc(sub_enum);
|
693
|
+
}
|
694
|
+
|
695
|
+
static int spanfe_start(FrtSpanEnum *self)
|
696
|
+
{
|
697
|
+
FrtSpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
698
|
+
return sub_enum->start(sub_enum);
|
699
|
+
}
|
700
|
+
|
701
|
+
static int spanfe_end(FrtSpanEnum *self)
|
702
|
+
{
|
703
|
+
FrtSpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
704
|
+
return sub_enum->end(sub_enum);
|
705
|
+
}
|
706
|
+
|
707
|
+
static char *spanfe_to_s(FrtSpanEnum *self)
|
708
|
+
{
|
709
|
+
char *query_str = self->query->to_s(self->query, (FrtSymbol)NULL);
|
710
|
+
char *res = frt_strfmt("SpanFirstEnum(%s)", query_str);
|
711
|
+
free(query_str);
|
712
|
+
return res;
|
713
|
+
}
|
714
|
+
|
715
|
+
static void spanfe_destroy(FrtSpanEnum *self)
|
716
|
+
{
|
717
|
+
FrtSpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
718
|
+
sub_enum->destroy(sub_enum);
|
719
|
+
free(self);
|
720
|
+
}
|
721
|
+
|
722
|
+
static FrtSpanEnum *spanfe_new(FrtQuery *query, FrtIndexReader *ir)
|
723
|
+
{
|
724
|
+
FrtSpanEnum *self = (FrtSpanEnum *)FRT_ALLOC(SpanFirstEnum);
|
725
|
+
FrtSpanFirstQuery *sfq = SpFQ(query);
|
726
|
+
|
727
|
+
SpFEn(self)->sub_enum = SpQ(sfq->match)->get_spans(sfq->match, ir);
|
728
|
+
|
729
|
+
self->query = query;
|
730
|
+
self->next = &spanfe_next;
|
731
|
+
self->skip_to = &spanfe_skip_to;
|
732
|
+
self->doc = &spanfe_doc;
|
733
|
+
self->start = &spanfe_start;
|
734
|
+
self->end = &spanfe_end;
|
735
|
+
self->destroy = &spanfe_destroy;
|
736
|
+
self->to_s = &spanfe_to_s;
|
737
|
+
|
738
|
+
return self;
|
739
|
+
}
|
740
|
+
|
741
|
+
|
742
|
+
/*****************************************************************************
|
743
|
+
* SpanOrEnum
|
744
|
+
*****************************************************************************/
|
745
|
+
|
746
|
+
#define SpOEn(span_enum) ((SpanOrEnum *)(span_enum))
|
747
|
+
#define SpOQ(query) ((FrtSpanOrQuery *)(query))
|
748
|
+
|
749
|
+
typedef struct SpanOrEnum
|
750
|
+
{
|
751
|
+
FrtSpanEnum super;
|
752
|
+
FrtPriorityQueue *queue;
|
753
|
+
FrtSpanEnum **span_enums;
|
754
|
+
int s_cnt;
|
755
|
+
bool first_time : 1;
|
756
|
+
} SpanOrEnum;
|
757
|
+
|
758
|
+
|
759
|
+
static bool span_less_than(FrtSpanEnum *s1, FrtSpanEnum *s2)
|
760
|
+
{
|
761
|
+
int doc_diff, start_diff;
|
762
|
+
doc_diff = s1->doc(s1) - s2->doc(s2);
|
763
|
+
if (doc_diff == 0) {
|
764
|
+
start_diff = s1->start(s1) - s2->start(s2);
|
765
|
+
if (start_diff == 0) {
|
766
|
+
return s1->end(s1) < s2->end(s2);
|
767
|
+
}
|
768
|
+
else {
|
769
|
+
return start_diff < 0;
|
770
|
+
}
|
771
|
+
}
|
772
|
+
else {
|
773
|
+
return doc_diff < 0;
|
774
|
+
}
|
775
|
+
}
|
776
|
+
|
777
|
+
static bool spanoe_next(FrtSpanEnum *self)
|
778
|
+
{
|
779
|
+
SpanOrEnum *soe = SpOEn(self);
|
780
|
+
FrtSpanEnum *se;
|
781
|
+
int i;
|
782
|
+
|
783
|
+
if (soe->first_time) { /* first time -- initialize */
|
784
|
+
for (i = 0; i < soe->s_cnt; i++) {
|
785
|
+
se = soe->span_enums[i];
|
786
|
+
if (se->next(se)) { /* move to first entry */
|
787
|
+
frt_pq_push(soe->queue, se);
|
788
|
+
}
|
789
|
+
}
|
790
|
+
soe->first_time = false;
|
791
|
+
return soe->queue->size != 0;
|
792
|
+
}
|
793
|
+
|
794
|
+
if (soe->queue->size == 0) {
|
795
|
+
return false; /* all done */
|
796
|
+
}
|
797
|
+
|
798
|
+
se = (FrtSpanEnum *)frt_pq_top(soe->queue);
|
799
|
+
if (se->next(se)) { /* move to next */
|
800
|
+
frt_pq_down(soe->queue);
|
801
|
+
return true;
|
802
|
+
}
|
803
|
+
|
804
|
+
frt_pq_pop(soe->queue); /* exhausted a clause */
|
805
|
+
|
806
|
+
return soe->queue->size != 0;
|
807
|
+
}
|
808
|
+
|
809
|
+
static bool spanoe_skip_to(FrtSpanEnum *self, int target)
|
810
|
+
{
|
811
|
+
SpanOrEnum *soe = SpOEn(self);
|
812
|
+
FrtSpanEnum *se;
|
813
|
+
int i;
|
814
|
+
|
815
|
+
if (soe->first_time) { /* first time -- initialize */
|
816
|
+
for (i = 0; i < soe->s_cnt; i++) {
|
817
|
+
se = soe->span_enums[i];
|
818
|
+
if (se->skip_to(se, target)) {/* move to target */
|
819
|
+
frt_pq_push(soe->queue, se);
|
820
|
+
}
|
821
|
+
}
|
822
|
+
soe->first_time = false;
|
823
|
+
}
|
824
|
+
else {
|
825
|
+
while ((soe->queue->size != 0) &&
|
826
|
+
((se = (FrtSpanEnum *)frt_pq_top(soe->queue)) != NULL) &&
|
827
|
+
(se->doc(se) < target)) {
|
828
|
+
if (se->skip_to(se, target)) {
|
829
|
+
frt_pq_down(soe->queue);
|
830
|
+
}
|
831
|
+
else {
|
832
|
+
frt_pq_pop(soe->queue);
|
833
|
+
}
|
834
|
+
}
|
835
|
+
}
|
836
|
+
|
837
|
+
return soe->queue->size != 0;
|
838
|
+
}
|
839
|
+
|
840
|
+
#define SpOEn_Top_SE(self) (FrtSpanEnum *)frt_pq_top(SpOEn(self)->queue)
|
841
|
+
|
842
|
+
static int spanoe_doc(FrtSpanEnum *self)
|
843
|
+
{
|
844
|
+
FrtSpanEnum *se = SpOEn_Top_SE(self);
|
845
|
+
return se->doc(se);
|
846
|
+
}
|
847
|
+
|
848
|
+
static int spanoe_start(FrtSpanEnum *self)
|
849
|
+
{
|
850
|
+
FrtSpanEnum *se = SpOEn_Top_SE(self);
|
851
|
+
return se->start(se);
|
852
|
+
}
|
853
|
+
|
854
|
+
static int spanoe_end(FrtSpanEnum *self)
|
855
|
+
{
|
856
|
+
FrtSpanEnum *se = SpOEn_Top_SE(self);
|
857
|
+
return se->end(se);
|
858
|
+
}
|
859
|
+
|
860
|
+
static char *spanoe_to_s(FrtSpanEnum *self)
|
861
|
+
{
|
862
|
+
SpanOrEnum *soe = SpOEn(self);
|
863
|
+
char *query_str = self->query->to_s(self->query, (FrtSymbol)NULL);
|
864
|
+
char doc_str[62];
|
865
|
+
size_t len = strlen(query_str);
|
866
|
+
char *str = FRT_ALLOC_N(char, len + 80);
|
867
|
+
|
868
|
+
if (soe->first_time) {
|
869
|
+
sprintf(doc_str, "START");
|
870
|
+
}
|
871
|
+
else {
|
872
|
+
if (soe->queue->size == 0) {
|
873
|
+
sprintf(doc_str, "END");
|
874
|
+
}
|
875
|
+
else {
|
876
|
+
sprintf(doc_str, "%d:%d-%d", self->doc(self),
|
877
|
+
self->start(self), self->end(self));
|
878
|
+
}
|
879
|
+
}
|
880
|
+
sprintf(str, "SpanOrEnum(%s)@%s", query_str, doc_str);
|
881
|
+
free(query_str);
|
882
|
+
return str;
|
883
|
+
}
|
884
|
+
|
885
|
+
static void spanoe_destroy(FrtSpanEnum *self)
|
886
|
+
{
|
887
|
+
FrtSpanEnum *se;
|
888
|
+
SpanOrEnum *soe = SpOEn(self);
|
889
|
+
int i;
|
890
|
+
frt_pq_destroy(soe->queue);
|
891
|
+
for (i = 0; i < soe->s_cnt; i++) {
|
892
|
+
se = soe->span_enums[i];
|
893
|
+
se->destroy(se);
|
894
|
+
}
|
895
|
+
free(soe->span_enums);
|
896
|
+
free(self);
|
897
|
+
}
|
898
|
+
|
899
|
+
static FrtSpanEnum *spanoe_new(FrtQuery *query, FrtIndexReader *ir)
|
900
|
+
{
|
901
|
+
FrtQuery *clause;
|
902
|
+
FrtSpanEnum *self = (FrtSpanEnum *)FRT_ALLOC(SpanOrEnum);
|
903
|
+
FrtSpanOrQuery *soq = SpOQ(query);
|
904
|
+
int i;
|
905
|
+
|
906
|
+
SpOEn(self)->first_time = true;
|
907
|
+
SpOEn(self)->s_cnt = soq->c_cnt;
|
908
|
+
SpOEn(self)->span_enums = FRT_ALLOC_N(FrtSpanEnum *, SpOEn(self)->s_cnt);
|
909
|
+
|
910
|
+
for (i = 0; i < SpOEn(self)->s_cnt; i++) {
|
911
|
+
clause = soq->clauses[i];
|
912
|
+
SpOEn(self)->span_enums[i] = SpQ(clause)->get_spans(clause, ir);
|
913
|
+
}
|
914
|
+
|
915
|
+
SpOEn(self)->queue = frt_pq_new(SpOEn(self)->s_cnt, (frt_lt_ft)&span_less_than,
|
916
|
+
(frt_free_ft)NULL);
|
917
|
+
|
918
|
+
self->query = query;
|
919
|
+
self->next = &spanoe_next;
|
920
|
+
self->skip_to = &spanoe_skip_to;
|
921
|
+
self->doc = &spanoe_doc;
|
922
|
+
self->start = &spanoe_start;
|
923
|
+
self->end = &spanoe_end;
|
924
|
+
self->destroy = &spanoe_destroy;
|
925
|
+
self->to_s = &spanoe_to_s;
|
926
|
+
|
927
|
+
return self;
|
928
|
+
}
|
929
|
+
|
930
|
+
/*****************************************************************************
|
931
|
+
* SpanNearEnum
|
932
|
+
*****************************************************************************/
|
933
|
+
|
934
|
+
#define SpNEn(span_enum) ((SpanNearEnum *)(span_enum))
|
935
|
+
#define SpNQ(query) ((FrtSpanNearQuery *)(query))
|
936
|
+
|
937
|
+
typedef struct SpanNearEnum
|
938
|
+
{
|
939
|
+
FrtSpanEnum super;
|
940
|
+
FrtSpanEnum **span_enums;
|
941
|
+
int s_cnt;
|
942
|
+
int slop;
|
943
|
+
int current;
|
944
|
+
int doc;
|
945
|
+
int start;
|
946
|
+
int end;
|
947
|
+
bool first_time : 1;
|
948
|
+
bool in_order : 1;
|
949
|
+
} SpanNearEnum;
|
950
|
+
|
951
|
+
|
952
|
+
#define SpNEn_NEXT() do {\
|
953
|
+
sne->current = (sne->current+1) % sne->s_cnt;\
|
954
|
+
se = sne->span_enums[sne->current];\
|
955
|
+
} while (0);
|
956
|
+
|
957
|
+
static bool sne_init(SpanNearEnum *sne)
|
958
|
+
{
|
959
|
+
FrtSpanEnum *se = sne->span_enums[sne->current];
|
960
|
+
int prev_doc = se->doc(se);
|
961
|
+
int i;
|
962
|
+
|
963
|
+
for (i = 1; i < sne->s_cnt; i++) {
|
964
|
+
SpNEn_NEXT();
|
965
|
+
if (!se->skip_to(se, prev_doc)) {
|
966
|
+
return false;
|
967
|
+
}
|
968
|
+
prev_doc = se->doc(se);
|
969
|
+
}
|
970
|
+
return true;
|
971
|
+
}
|
972
|
+
|
973
|
+
static bool sne_goto_next_doc(SpanNearEnum *sne)
|
974
|
+
{
|
975
|
+
FrtSpanEnum *se = sne->span_enums[sne->current];
|
976
|
+
int prev_doc = se->doc(se);
|
977
|
+
|
978
|
+
SpNEn_NEXT();
|
979
|
+
|
980
|
+
while (se->doc(se) < prev_doc) {
|
981
|
+
if (! se->skip_to(se, prev_doc)) {
|
982
|
+
return false;
|
983
|
+
}
|
984
|
+
prev_doc = se->doc(se);
|
985
|
+
SpNEn_NEXT();
|
986
|
+
}
|
987
|
+
return true;
|
988
|
+
}
|
989
|
+
|
990
|
+
static bool sne_next_unordered_match(FrtSpanEnum *self)
|
991
|
+
{
|
992
|
+
SpanNearEnum *sne = SpNEn(self);
|
993
|
+
FrtSpanEnum *se, *min_se = NULL;
|
994
|
+
int i;
|
995
|
+
int max_end, end, min_start, start, doc;
|
996
|
+
int lengths_sum;
|
997
|
+
|
998
|
+
while (true) {
|
999
|
+
max_end = 0;
|
1000
|
+
min_start = INT_MAX;
|
1001
|
+
lengths_sum = 0;
|
1002
|
+
|
1003
|
+
for (i = 0; i < sne->s_cnt; i++) {
|
1004
|
+
se = sne->span_enums[i];
|
1005
|
+
if ((end=se->end(se)) > max_end) {
|
1006
|
+
max_end = end;
|
1007
|
+
}
|
1008
|
+
if ((start=se->start(se)) < min_start) {
|
1009
|
+
min_start = start;
|
1010
|
+
min_se = se;
|
1011
|
+
sne->current = i; /* current should point to the minimum span */
|
1012
|
+
}
|
1013
|
+
lengths_sum += end - start;
|
1014
|
+
}
|
1015
|
+
|
1016
|
+
if ((max_end - min_start - lengths_sum) <= sne->slop) {
|
1017
|
+
/* we have a match */
|
1018
|
+
sne->start = min_start;
|
1019
|
+
sne->end = max_end;
|
1020
|
+
sne->doc = min_se->doc(min_se);
|
1021
|
+
return true;
|
1022
|
+
}
|
1023
|
+
|
1024
|
+
/* increment the minimum span_enum and try again */
|
1025
|
+
doc = min_se->doc(min_se);
|
1026
|
+
if (!min_se->next(min_se)) {
|
1027
|
+
return false;
|
1028
|
+
}
|
1029
|
+
if (doc < min_se->doc(min_se)) {
|
1030
|
+
if (!sne_goto_next_doc(sne)) return false;
|
1031
|
+
}
|
1032
|
+
}
|
1033
|
+
}
|
1034
|
+
|
1035
|
+
static bool sne_next_ordered_match(FrtSpanEnum *self)
|
1036
|
+
{
|
1037
|
+
SpanNearEnum *sne = SpNEn(self);
|
1038
|
+
FrtSpanEnum *se;
|
1039
|
+
int i;
|
1040
|
+
int prev_doc, prev_start, prev_end;
|
1041
|
+
int doc=0, start=0, end=0;
|
1042
|
+
int lengths_sum;
|
1043
|
+
|
1044
|
+
while (true) {
|
1045
|
+
se = sne->span_enums[0];
|
1046
|
+
|
1047
|
+
prev_doc = se->doc(se);
|
1048
|
+
sne->start = prev_start = se->start(se);
|
1049
|
+
prev_end = se->end(se);
|
1050
|
+
|
1051
|
+
i = 1;
|
1052
|
+
lengths_sum = prev_end - prev_start;
|
1053
|
+
|
1054
|
+
while (i < sne->s_cnt) {
|
1055
|
+
se = sne->span_enums[i];
|
1056
|
+
doc = se->doc(se);
|
1057
|
+
start = se->start(se);
|
1058
|
+
end = se->end(se);
|
1059
|
+
while ((doc == prev_doc) && ((start < prev_start) ||
|
1060
|
+
((start == prev_start) && (end < prev_end)))) {
|
1061
|
+
if (!se->next(se)) {
|
1062
|
+
return false;
|
1063
|
+
}
|
1064
|
+
doc = se->doc(se);
|
1065
|
+
start = se->start(se);
|
1066
|
+
end = se->end(se);
|
1067
|
+
}
|
1068
|
+
if (doc != prev_doc) {
|
1069
|
+
sne->current = i;
|
1070
|
+
if (!sne_goto_next_doc(sne)) {
|
1071
|
+
return false;
|
1072
|
+
}
|
1073
|
+
break;
|
1074
|
+
}
|
1075
|
+
i++;
|
1076
|
+
lengths_sum += end - start;
|
1077
|
+
prev_doc = doc;
|
1078
|
+
prev_start = start;
|
1079
|
+
prev_end = end;
|
1080
|
+
}
|
1081
|
+
if (i == sne->s_cnt) {
|
1082
|
+
if ((end - sne->start - lengths_sum) <= sne->slop) {
|
1083
|
+
/* we have a match */
|
1084
|
+
sne->end = end;
|
1085
|
+
sne->doc = doc;
|
1086
|
+
|
1087
|
+
/* the minimum span is always the first span so it needs to be
|
1088
|
+
* incremented next time around */
|
1089
|
+
sne->current = 0;
|
1090
|
+
return true;
|
1091
|
+
|
1092
|
+
}
|
1093
|
+
else {
|
1094
|
+
se = sne->span_enums[0];
|
1095
|
+
if (!se->next(se)) {
|
1096
|
+
return false;
|
1097
|
+
}
|
1098
|
+
if (se->doc(se) != prev_doc) {
|
1099
|
+
sne->current = 0;
|
1100
|
+
if (!sne_goto_next_doc(sne)) {
|
1101
|
+
return false;
|
1102
|
+
}
|
1103
|
+
}
|
1104
|
+
}
|
1105
|
+
}
|
1106
|
+
}
|
1107
|
+
}
|
1108
|
+
|
1109
|
+
static bool sne_next_match(FrtSpanEnum *self)
|
1110
|
+
{
|
1111
|
+
SpanNearEnum *sne = SpNEn(self);
|
1112
|
+
FrtSpanEnum *se_curr, *se_next;
|
1113
|
+
|
1114
|
+
if (!sne->first_time) {
|
1115
|
+
if (!sne_init(sne)) {
|
1116
|
+
return false;
|
1117
|
+
}
|
1118
|
+
sne->first_time = false;
|
1119
|
+
}
|
1120
|
+
se_curr = sne->span_enums[sne->current];
|
1121
|
+
se_next = sne->span_enums[(sne->current+1)%sne->s_cnt];
|
1122
|
+
if (se_curr->doc(se_curr) > se_next->doc(se_next)) {
|
1123
|
+
if (!sne_goto_next_doc(sne)) {
|
1124
|
+
return false;
|
1125
|
+
}
|
1126
|
+
}
|
1127
|
+
|
1128
|
+
if (sne->in_order) {
|
1129
|
+
return sne_next_ordered_match(self);
|
1130
|
+
}
|
1131
|
+
else {
|
1132
|
+
return sne_next_unordered_match(self);
|
1133
|
+
}
|
1134
|
+
}
|
1135
|
+
|
1136
|
+
static bool spanne_next(FrtSpanEnum *self)
|
1137
|
+
{
|
1138
|
+
SpanNearEnum *sne = SpNEn(self);
|
1139
|
+
FrtSpanEnum *se;
|
1140
|
+
|
1141
|
+
se = sne->span_enums[sne->current];
|
1142
|
+
if (!se->next(se)) return false;
|
1143
|
+
|
1144
|
+
return sne_next_match(self);
|
1145
|
+
}
|
1146
|
+
|
1147
|
+
static bool spanne_skip_to(FrtSpanEnum *self, int target)
|
1148
|
+
{
|
1149
|
+
FrtSpanEnum *se = SpNEn(self)->span_enums[SpNEn(self)->current];
|
1150
|
+
if (!se->skip_to(se, target)) {
|
1151
|
+
return false;
|
1152
|
+
}
|
1153
|
+
|
1154
|
+
return sne_next_match(self);
|
1155
|
+
}
|
1156
|
+
|
1157
|
+
static int spanne_doc(FrtSpanEnum *self)
|
1158
|
+
{
|
1159
|
+
return SpNEn(self)->doc;
|
1160
|
+
}
|
1161
|
+
|
1162
|
+
static int spanne_start(FrtSpanEnum *self)
|
1163
|
+
{
|
1164
|
+
return SpNEn(self)->start;
|
1165
|
+
}
|
1166
|
+
|
1167
|
+
static int spanne_end(FrtSpanEnum *self)
|
1168
|
+
{
|
1169
|
+
return SpNEn(self)->end;
|
1170
|
+
}
|
1171
|
+
|
1172
|
+
static char *spanne_to_s(FrtSpanEnum *self)
|
1173
|
+
{
|
1174
|
+
SpanNearEnum *sne = SpNEn(self);
|
1175
|
+
char *query_str = self->query->to_s(self->query, (FrtSymbol)NULL);
|
1176
|
+
char doc_str[62];
|
1177
|
+
size_t len = strlen(query_str);
|
1178
|
+
char *str = FRT_ALLOC_N(char, len + 80);
|
1179
|
+
|
1180
|
+
if (sne->first_time) {
|
1181
|
+
sprintf(doc_str, "START");
|
1182
|
+
}
|
1183
|
+
else {
|
1184
|
+
sprintf(doc_str, "%d:%d-%d", self->doc(self),
|
1185
|
+
self->start(self), self->end(self));
|
1186
|
+
}
|
1187
|
+
sprintf(str, "SpanNearEnum(%s)@%s", query_str, doc_str);
|
1188
|
+
free(query_str);
|
1189
|
+
return str;
|
1190
|
+
}
|
1191
|
+
|
1192
|
+
static void spanne_destroy(FrtSpanEnum *self)
|
1193
|
+
{
|
1194
|
+
FrtSpanEnum *se;
|
1195
|
+
SpanNearEnum *sne = SpNEn(self);
|
1196
|
+
int i;
|
1197
|
+
for (i = 0; i < sne->s_cnt; i++) {
|
1198
|
+
se = sne->span_enums[i];
|
1199
|
+
se->destroy(se);
|
1200
|
+
}
|
1201
|
+
free(sne->span_enums);
|
1202
|
+
free(self);
|
1203
|
+
}
|
1204
|
+
|
1205
|
+
static FrtSpanEnum *spanne_new(FrtQuery *query, FrtIndexReader *ir)
|
1206
|
+
{
|
1207
|
+
int i;
|
1208
|
+
FrtQuery *clause;
|
1209
|
+
FrtSpanEnum *self = (FrtSpanEnum *)FRT_ALLOC(SpanNearEnum);
|
1210
|
+
FrtSpanNearQuery *snq = SpNQ(query);
|
1211
|
+
|
1212
|
+
SpNEn(self)->first_time = true;
|
1213
|
+
SpNEn(self)->in_order = snq->in_order;
|
1214
|
+
SpNEn(self)->slop = snq->slop;
|
1215
|
+
SpNEn(self)->s_cnt = snq->c_cnt;
|
1216
|
+
SpNEn(self)->span_enums = FRT_ALLOC_N(FrtSpanEnum *, SpNEn(self)->s_cnt);
|
1217
|
+
|
1218
|
+
for (i = 0; i < SpNEn(self)->s_cnt; i++) {
|
1219
|
+
clause = snq->clauses[i];
|
1220
|
+
SpNEn(self)->span_enums[i] = SpQ(clause)->get_spans(clause, ir);
|
1221
|
+
}
|
1222
|
+
SpNEn(self)->current = 0;
|
1223
|
+
|
1224
|
+
SpNEn(self)->doc = -1;
|
1225
|
+
SpNEn(self)->start = -1;
|
1226
|
+
SpNEn(self)->end = -1;
|
1227
|
+
|
1228
|
+
self->query = query;
|
1229
|
+
self->next = &spanne_next;
|
1230
|
+
self->skip_to = &spanne_skip_to;
|
1231
|
+
self->doc = &spanne_doc;
|
1232
|
+
self->start = &spanne_start;
|
1233
|
+
self->end = &spanne_end;
|
1234
|
+
self->destroy = &spanne_destroy;
|
1235
|
+
self->to_s = &spanne_to_s;
|
1236
|
+
|
1237
|
+
return self;
|
1238
|
+
}
|
1239
|
+
|
1240
|
+
/*****************************************************************************
|
1241
|
+
*
|
1242
|
+
* SpanNotEnum
|
1243
|
+
*
|
1244
|
+
*****************************************************************************/
|
1245
|
+
|
1246
|
+
#define SpXEn(span_enum) ((SpanNotEnum *)(span_enum))
|
1247
|
+
#define SpXQ(query) ((FrtSpanNotQuery *)(query))
|
1248
|
+
|
1249
|
+
typedef struct SpanNotEnum
|
1250
|
+
{
|
1251
|
+
FrtSpanEnum super;
|
1252
|
+
FrtSpanEnum *inc;
|
1253
|
+
FrtSpanEnum *exc;
|
1254
|
+
bool more_inc : 1;
|
1255
|
+
bool more_exc : 1;
|
1256
|
+
} SpanNotEnum;
|
1257
|
+
|
1258
|
+
|
1259
|
+
static bool spanxe_next(FrtSpanEnum *self)
|
1260
|
+
{
|
1261
|
+
SpanNotEnum *sxe = SpXEn(self);
|
1262
|
+
FrtSpanEnum *inc = sxe->inc, *exc = sxe->exc;
|
1263
|
+
if (sxe->more_inc) { /* move to next incl */
|
1264
|
+
sxe->more_inc = inc->next(inc);
|
1265
|
+
}
|
1266
|
+
|
1267
|
+
while (sxe->more_inc && sxe->more_exc) {
|
1268
|
+
if (inc->doc(inc) > exc->doc(exc)) { /* skip excl */
|
1269
|
+
sxe->more_exc = exc->skip_to(exc, inc->doc(inc));
|
1270
|
+
}
|
1271
|
+
|
1272
|
+
while (sxe->more_exc /* while excl is before */
|
1273
|
+
&& (inc->doc(inc) == exc->doc(exc))
|
1274
|
+
&& (exc->end(exc) <= inc->start(inc))) {
|
1275
|
+
sxe->more_exc = exc->next(exc); /* increment excl */
|
1276
|
+
}
|
1277
|
+
|
1278
|
+
if (! sxe->more_exc || /* if no intersection */
|
1279
|
+
(inc->doc(inc) != exc->doc(exc)) ||
|
1280
|
+
inc->end(inc) <= exc->start(exc)) {
|
1281
|
+
break; /* we found a match */
|
1282
|
+
}
|
1283
|
+
|
1284
|
+
sxe->more_inc = inc->next(inc); /* intersected: keep scanning */
|
1285
|
+
}
|
1286
|
+
return sxe->more_inc;
|
1287
|
+
}
|
1288
|
+
|
1289
|
+
static bool spanxe_skip_to(FrtSpanEnum *self, int target)
|
1290
|
+
{
|
1291
|
+
SpanNotEnum *sxe = SpXEn(self);
|
1292
|
+
FrtSpanEnum *inc = sxe->inc, *exc = sxe->exc;
|
1293
|
+
int doc;
|
1294
|
+
|
1295
|
+
if (sxe->more_inc) { /* move to next incl */
|
1296
|
+
if (!(sxe->more_inc=sxe->inc->skip_to(sxe->inc, target))) return false;
|
1297
|
+
}
|
1298
|
+
|
1299
|
+
if (sxe->more_inc && ((doc=inc->doc(inc)) > exc->doc(exc))) {
|
1300
|
+
sxe->more_exc = exc->skip_to(exc, doc);
|
1301
|
+
}
|
1302
|
+
|
1303
|
+
while (sxe->more_exc /* while excl is before */
|
1304
|
+
&& inc->doc(inc) == exc->doc(exc)
|
1305
|
+
&& exc->end(exc) <= inc->start(inc)) {
|
1306
|
+
sxe->more_exc = exc->next(exc); /* increment excl */
|
1307
|
+
}
|
1308
|
+
|
1309
|
+
if (!sxe->more_exc || /* if no intersection */
|
1310
|
+
inc->doc(inc) != exc->doc(exc) ||
|
1311
|
+
inc->end(inc) <= exc->start(exc)) {
|
1312
|
+
return true; /* we found a match */
|
1313
|
+
}
|
1314
|
+
|
1315
|
+
return spanxe_next(self); /* scan to next match */
|
1316
|
+
}
|
1317
|
+
|
1318
|
+
static int spanxe_doc(FrtSpanEnum *self)
|
1319
|
+
{
|
1320
|
+
FrtSpanEnum *inc = SpXEn(self)->inc;
|
1321
|
+
return inc->doc(inc);
|
1322
|
+
}
|
1323
|
+
|
1324
|
+
static int spanxe_start(FrtSpanEnum *self)
|
1325
|
+
{
|
1326
|
+
FrtSpanEnum *inc = SpXEn(self)->inc;
|
1327
|
+
return inc->start(inc);
|
1328
|
+
}
|
1329
|
+
|
1330
|
+
static int spanxe_end(FrtSpanEnum *self)
|
1331
|
+
{
|
1332
|
+
FrtSpanEnum *inc = SpXEn(self)->inc;
|
1333
|
+
return inc->end(inc);
|
1334
|
+
}
|
1335
|
+
|
1336
|
+
static char *spanxe_to_s(FrtSpanEnum *self)
|
1337
|
+
{
|
1338
|
+
char *query_str = self->query->to_s(self->query, (FrtSymbol)NULL);
|
1339
|
+
char *res = frt_strfmt("SpanNotEnum(%s)", query_str);
|
1340
|
+
free(query_str);
|
1341
|
+
return res;
|
1342
|
+
}
|
1343
|
+
|
1344
|
+
static void spanxe_destroy(FrtSpanEnum *self)
|
1345
|
+
{
|
1346
|
+
SpanNotEnum *sxe = SpXEn(self);
|
1347
|
+
sxe->inc->destroy(sxe->inc);
|
1348
|
+
sxe->exc->destroy(sxe->exc);
|
1349
|
+
free(self);
|
1350
|
+
}
|
1351
|
+
|
1352
|
+
static FrtSpanEnum *spanxe_new(FrtQuery *query, FrtIndexReader *ir)
|
1353
|
+
{
|
1354
|
+
FrtSpanEnum *self = (FrtSpanEnum *)FRT_ALLOC(SpanNotEnum);
|
1355
|
+
SpanNotEnum *sxe = SpXEn(self);
|
1356
|
+
FrtSpanNotQuery *sxq = SpXQ(query);
|
1357
|
+
|
1358
|
+
sxe->inc = SpQ(sxq->inc)->get_spans(sxq->inc, ir);
|
1359
|
+
sxe->exc = SpQ(sxq->exc)->get_spans(sxq->exc, ir);
|
1360
|
+
sxe->more_inc = true;
|
1361
|
+
sxe->more_exc = sxe->exc->next(sxe->exc);
|
1362
|
+
|
1363
|
+
self->query = query;
|
1364
|
+
self->next = &spanxe_next;
|
1365
|
+
self->skip_to = &spanxe_skip_to;
|
1366
|
+
self->doc = &spanxe_doc;
|
1367
|
+
self->start = &spanxe_start;
|
1368
|
+
self->end = &spanxe_end;
|
1369
|
+
self->destroy = &spanxe_destroy;
|
1370
|
+
self->to_s = &spanxe_to_s;
|
1371
|
+
|
1372
|
+
return self;
|
1373
|
+
}
|
1374
|
+
|
1375
|
+
/*****************************************************************************
|
1376
|
+
*
|
1377
|
+
* SpanWeight
|
1378
|
+
*
|
1379
|
+
*****************************************************************************/
|
1380
|
+
|
1381
|
+
#define SpW(weight) ((SpanWeight *)(weight))
|
1382
|
+
typedef struct SpanWeight
|
1383
|
+
{
|
1384
|
+
FrtWeight super;
|
1385
|
+
FrtHashSet *terms;
|
1386
|
+
} SpanWeight;
|
1387
|
+
|
1388
|
+
static FrtExplanation *spanw_explain(FrtWeight *self, FrtIndexReader *ir, int target)
|
1389
|
+
{
|
1390
|
+
FrtExplanation *expl;
|
1391
|
+
FrtExplanation *idf_expl1;
|
1392
|
+
FrtExplanation *idf_expl2;
|
1393
|
+
FrtExplanation *query_expl;
|
1394
|
+
FrtExplanation *qnorm_expl;
|
1395
|
+
FrtExplanation *field_expl;
|
1396
|
+
FrtExplanation *tf_expl;
|
1397
|
+
FrtScorer *scorer;
|
1398
|
+
frt_uchar *field_norms;
|
1399
|
+
float field_norm;
|
1400
|
+
FrtExplanation *field_norm_expl;
|
1401
|
+
const char *field_name = rb_id2name(SpQ(self->query)->field);
|
1402
|
+
|
1403
|
+
char *query_str;
|
1404
|
+
FrtHashSet *terms = SpW(self)->terms;
|
1405
|
+
const int field_num = frt_fis_get_field_num(ir->fis, SpQ(self->query)->field);
|
1406
|
+
char *doc_freqs = NULL;
|
1407
|
+
size_t df_i = 0;
|
1408
|
+
FrtHashSetEntry *hse;
|
1409
|
+
|
1410
|
+
if (field_num < 0) {
|
1411
|
+
return frt_expl_new(0.0, "field \"%s\" does not exist in the index", field_name);
|
1412
|
+
}
|
1413
|
+
|
1414
|
+
query_str = self->query->to_s(self->query, (FrtSymbol)NULL);
|
1415
|
+
|
1416
|
+
for (hse = terms->first; hse; hse = hse->next) {
|
1417
|
+
char *term = (char *)hse->elem;
|
1418
|
+
FRT_REALLOC_N(doc_freqs, char, df_i + strlen(term) + 23);
|
1419
|
+
df_i += sprintf(doc_freqs + df_i, "%s=%d, ", term,
|
1420
|
+
ir->doc_freq(ir, field_num, term));
|
1421
|
+
}
|
1422
|
+
/* remove the ',' at the end of the string if it exists */
|
1423
|
+
if (terms->size > 0) {
|
1424
|
+
df_i -= 2;
|
1425
|
+
doc_freqs[df_i] = '\0';
|
1426
|
+
}
|
1427
|
+
else {
|
1428
|
+
doc_freqs = frt_estrdup("");
|
1429
|
+
}
|
1430
|
+
|
1431
|
+
expl = frt_expl_new(0.0, "weight(%s in %d), product of:", query_str, target);
|
1432
|
+
|
1433
|
+
/* We need two of these as it's included in both the query explanation
|
1434
|
+
* and the field explanation */
|
1435
|
+
idf_expl1 = frt_expl_new(self->idf, "idf(%s: %s)", field_name, doc_freqs);
|
1436
|
+
idf_expl2 = frt_expl_new(self->idf, "idf(%s: %s)", field_name, doc_freqs);
|
1437
|
+
if (terms->size > 0) {
|
1438
|
+
free(doc_freqs); /* only free if allocated */
|
1439
|
+
}
|
1440
|
+
|
1441
|
+
/* explain query weight */
|
1442
|
+
query_expl = frt_expl_new(0.0, "query_weight(%s), product of:", query_str);
|
1443
|
+
|
1444
|
+
if (self->query->boost != 1.0) {
|
1445
|
+
frt_expl_add_detail(query_expl, frt_expl_new(self->query->boost, "boost"));
|
1446
|
+
}
|
1447
|
+
|
1448
|
+
frt_expl_add_detail(query_expl, idf_expl1);
|
1449
|
+
|
1450
|
+
qnorm_expl = frt_expl_new(self->qnorm, "query_norm");
|
1451
|
+
frt_expl_add_detail(query_expl, qnorm_expl);
|
1452
|
+
|
1453
|
+
query_expl->value = self->query->boost * idf_expl1->value * qnorm_expl->value;
|
1454
|
+
|
1455
|
+
frt_expl_add_detail(expl, query_expl);
|
1456
|
+
|
1457
|
+
/* explain field weight */
|
1458
|
+
field_expl = frt_expl_new(0.0, "field_weight(%s:%s in %d), product of:", field_name, query_str, target);
|
1459
|
+
free(query_str);
|
1460
|
+
|
1461
|
+
scorer = self->scorer(self, ir);
|
1462
|
+
tf_expl = scorer->explain(scorer, target);
|
1463
|
+
scorer->destroy(scorer);
|
1464
|
+
frt_expl_add_detail(field_expl, tf_expl);
|
1465
|
+
frt_expl_add_detail(field_expl, idf_expl2);
|
1466
|
+
|
1467
|
+
field_norms = ir->get_norms(ir, field_num);
|
1468
|
+
field_norm = (field_norms
|
1469
|
+
? frt_sim_decode_norm(self->similarity, field_norms[target])
|
1470
|
+
: (float)0.0);
|
1471
|
+
field_norm_expl = frt_expl_new(field_norm, "field_norm(field=%s, doc=%d)",
|
1472
|
+
field_name, target);
|
1473
|
+
frt_expl_add_detail(field_expl, field_norm_expl);
|
1474
|
+
|
1475
|
+
field_expl->value = tf_expl->value * idf_expl2->value * field_norm_expl->value;
|
1476
|
+
|
1477
|
+
/* combine them */
|
1478
|
+
if (query_expl->value == 1.0) {
|
1479
|
+
frt_expl_destroy(expl);
|
1480
|
+
return field_expl;
|
1481
|
+
}
|
1482
|
+
else {
|
1483
|
+
expl->value = (query_expl->value * field_expl->value);
|
1484
|
+
frt_expl_add_detail(expl, field_expl);
|
1485
|
+
return expl;
|
1486
|
+
}
|
1487
|
+
}
|
1488
|
+
|
1489
|
+
static char *spanw_to_s(FrtWeight *self)
|
1490
|
+
{
|
1491
|
+
return frt_strfmt("SpanWeight(%f)", self->value);
|
1492
|
+
}
|
1493
|
+
|
1494
|
+
static void spanw_destroy(FrtWeight *self)
|
1495
|
+
{
|
1496
|
+
frt_hs_destroy(SpW(self)->terms);
|
1497
|
+
frt_w_destroy(self);
|
1498
|
+
}
|
1499
|
+
|
1500
|
+
static FrtWeight *spanw_new(FrtQuery *query, FrtSearcher *searcher)
|
1501
|
+
{
|
1502
|
+
FrtHashSetEntry *hse;
|
1503
|
+
FrtWeight *self = w_new(SpanWeight, query);
|
1504
|
+
FrtHashSet *terms = SpQ(query)->get_terms(query);
|
1505
|
+
|
1506
|
+
SpW(self)->terms = terms;
|
1507
|
+
self->scorer = &spansc_new;
|
1508
|
+
self->explain = &spanw_explain;
|
1509
|
+
self->to_s = &spanw_to_s;
|
1510
|
+
self->destroy = &spanw_destroy;
|
1511
|
+
|
1512
|
+
self->similarity = query->get_similarity(query, searcher);
|
1513
|
+
|
1514
|
+
self->idf = 0.0f;
|
1515
|
+
|
1516
|
+
for (hse = terms->first; hse; hse = hse->next) {
|
1517
|
+
self->idf += frt_sim_idf_term(self->similarity, SpQ(query)->field,
|
1518
|
+
(char *)hse->elem, searcher);
|
1519
|
+
}
|
1520
|
+
|
1521
|
+
return self;
|
1522
|
+
}
|
1523
|
+
|
1524
|
+
/*****************************************************************************
|
1525
|
+
* FrtSpanTermQuery
|
1526
|
+
*****************************************************************************/
|
1527
|
+
|
1528
|
+
static char *spantq_to_s(FrtQuery *self, FrtSymbol default_field)
|
1529
|
+
{
|
1530
|
+
if (default_field && default_field == SpQ(self)->field) {
|
1531
|
+
return frt_strfmt("span_terms(%s)", SpTQ(self)->term);
|
1532
|
+
} else {
|
1533
|
+
return frt_strfmt("span_terms(%s:%s)", rb_id2name(SpQ(self)->field), SpTQ(self)->term);
|
1534
|
+
}
|
1535
|
+
}
|
1536
|
+
|
1537
|
+
static void spantq_destroy_i(FrtQuery *self)
|
1538
|
+
{
|
1539
|
+
free(SpTQ(self)->term);
|
1540
|
+
spanq_destroy_i(self);
|
1541
|
+
}
|
1542
|
+
|
1543
|
+
static void spantq_extract_terms(FrtQuery *self, FrtHashSet *terms)
|
1544
|
+
{
|
1545
|
+
frt_hs_add(terms, frt_term_new(SpQ(self)->field, SpTQ(self)->term));
|
1546
|
+
}
|
1547
|
+
|
1548
|
+
static FrtHashSet *spantq_get_terms(FrtQuery *self)
|
1549
|
+
{
|
1550
|
+
FrtHashSet *terms = frt_hs_new_str(&free);
|
1551
|
+
frt_hs_add(terms, frt_estrdup(SpTQ(self)->term));
|
1552
|
+
return terms;
|
1553
|
+
}
|
1554
|
+
|
1555
|
+
static unsigned long long spantq_hash(FrtQuery *self)
|
1556
|
+
{
|
1557
|
+
return spanq_hash(self) ^ frt_str_hash(SpTQ(self)->term);
|
1558
|
+
}
|
1559
|
+
|
1560
|
+
static int spantq_eq(FrtQuery *self, FrtQuery *o)
|
1561
|
+
{
|
1562
|
+
return spanq_eq(self, o) && strcmp(SpTQ(self)->term, SpTQ(o)->term) == 0;
|
1563
|
+
}
|
1564
|
+
|
1565
|
+
FrtQuery *frt_spantq_new(FrtSymbol field, const char *term)
|
1566
|
+
{
|
1567
|
+
FrtQuery *self = frt_q_new(FrtSpanTermQuery);
|
1568
|
+
|
1569
|
+
SpTQ(self)->term = frt_estrdup(term);
|
1570
|
+
SpQ(self)->field = field;
|
1571
|
+
SpQ(self)->get_spans = &spante_new;
|
1572
|
+
SpQ(self)->get_terms = &spantq_get_terms;
|
1573
|
+
|
1574
|
+
self->type = SPAN_TERM_QUERY;
|
1575
|
+
self->extract_terms = &spantq_extract_terms;
|
1576
|
+
self->to_s = &spantq_to_s;
|
1577
|
+
self->hash = &spantq_hash;
|
1578
|
+
self->eq = &spantq_eq;
|
1579
|
+
self->destroy_i = &spantq_destroy_i;
|
1580
|
+
self->create_weight_i = &spanw_new;
|
1581
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
1582
|
+
return self;
|
1583
|
+
}
|
1584
|
+
|
1585
|
+
/*****************************************************************************
|
1586
|
+
* SpanMultiTermQuery
|
1587
|
+
*****************************************************************************/
|
1588
|
+
|
1589
|
+
static char *spanmtq_to_s(FrtQuery *self, FrtSymbol field)
|
1590
|
+
{
|
1591
|
+
char *terms = NULL, *p;
|
1592
|
+
int len = 3, i;
|
1593
|
+
FrtSpanMultiTermQuery *smtq = SpMTQ(self);
|
1594
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1595
|
+
len += strlen(smtq->terms[i]) + 2;
|
1596
|
+
}
|
1597
|
+
p = terms = FRT_ALLOC_N(char, len);
|
1598
|
+
*(p++) = '[';
|
1599
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1600
|
+
if (i != 0) *(p++) = ',';
|
1601
|
+
strcpy(p, smtq->terms[i]);
|
1602
|
+
p += strlen(smtq->terms[i]);
|
1603
|
+
}
|
1604
|
+
*(p++) = ']';
|
1605
|
+
*p = '\0';
|
1606
|
+
|
1607
|
+
if (field == SpQ(self)->field) {
|
1608
|
+
p = frt_strfmt("span_terms(%s)", terms);
|
1609
|
+
}
|
1610
|
+
else {
|
1611
|
+
p = frt_strfmt("span_terms(%s:%s)", rb_id2name(SpQ(self)->field), terms);
|
1612
|
+
}
|
1613
|
+
free(terms);
|
1614
|
+
return p;
|
1615
|
+
}
|
1616
|
+
|
1617
|
+
static void spanmtq_destroy_i(FrtQuery *self)
|
1618
|
+
{
|
1619
|
+
FrtSpanMultiTermQuery *smtq = SpMTQ(self);
|
1620
|
+
int i;
|
1621
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1622
|
+
free(smtq->terms[i]);
|
1623
|
+
}
|
1624
|
+
free(smtq->terms);
|
1625
|
+
spanq_destroy_i(self);
|
1626
|
+
}
|
1627
|
+
|
1628
|
+
static void spanmtq_extract_terms(FrtQuery *self, FrtHashSet *terms)
|
1629
|
+
{
|
1630
|
+
FrtSpanMultiTermQuery *smtq = SpMTQ(self);
|
1631
|
+
int i;
|
1632
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1633
|
+
frt_hs_add(terms, frt_term_new(SpQ(self)->field, smtq->terms[i]));
|
1634
|
+
}
|
1635
|
+
}
|
1636
|
+
|
1637
|
+
static FrtHashSet *spanmtq_get_terms(FrtQuery *self)
|
1638
|
+
{
|
1639
|
+
FrtHashSet *terms = frt_hs_new_str(&free);
|
1640
|
+
FrtSpanMultiTermQuery *smtq = SpMTQ(self);
|
1641
|
+
int i;
|
1642
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1643
|
+
frt_hs_add(terms, frt_estrdup(smtq->terms[i]));
|
1644
|
+
}
|
1645
|
+
return terms;
|
1646
|
+
}
|
1647
|
+
|
1648
|
+
static unsigned long long spanmtq_hash(FrtQuery *self)
|
1649
|
+
{
|
1650
|
+
unsigned long long hash = spanq_hash(self);
|
1651
|
+
FrtSpanMultiTermQuery *smtq = SpMTQ(self);
|
1652
|
+
int i;
|
1653
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1654
|
+
hash ^= frt_str_hash(smtq->terms[i]);
|
1655
|
+
}
|
1656
|
+
return hash;
|
1657
|
+
}
|
1658
|
+
|
1659
|
+
static int spanmtq_eq(FrtQuery *self, FrtQuery *o)
|
1660
|
+
{
|
1661
|
+
FrtSpanMultiTermQuery *smtq = SpMTQ(self);
|
1662
|
+
FrtSpanMultiTermQuery *smtqo = SpMTQ(o);
|
1663
|
+
int i;
|
1664
|
+
if (!spanq_eq(self, o)) return false;
|
1665
|
+
if (smtq->term_cnt != smtqo->term_cnt) return false;
|
1666
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1667
|
+
if (strcmp(smtq->terms[i], smtqo->terms[i]) != 0) return false;
|
1668
|
+
}
|
1669
|
+
return true;;
|
1670
|
+
}
|
1671
|
+
|
1672
|
+
FrtQuery *frt_spanmtq_new_conf(FrtSymbol field, int max_terms)
|
1673
|
+
{
|
1674
|
+
FrtQuery *self = frt_q_new(FrtSpanMultiTermQuery);
|
1675
|
+
|
1676
|
+
SpMTQ(self)->terms = FRT_ALLOC_N(char *, max_terms);
|
1677
|
+
SpMTQ(self)->term_cnt = 0;
|
1678
|
+
SpMTQ(self)->term_capa = max_terms;
|
1679
|
+
|
1680
|
+
SpQ(self)->field = field;
|
1681
|
+
SpQ(self)->get_spans = &spanmte_new;
|
1682
|
+
SpQ(self)->get_terms = &spanmtq_get_terms;
|
1683
|
+
|
1684
|
+
self->type = SPAN_MULTI_TERM_QUERY;
|
1685
|
+
self->extract_terms = &spanmtq_extract_terms;
|
1686
|
+
self->to_s = &spanmtq_to_s;
|
1687
|
+
self->hash = &spanmtq_hash;
|
1688
|
+
self->eq = &spanmtq_eq;
|
1689
|
+
self->destroy_i = &spanmtq_destroy_i;
|
1690
|
+
self->create_weight_i = &spanw_new;
|
1691
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
1692
|
+
|
1693
|
+
return self;
|
1694
|
+
}
|
1695
|
+
|
1696
|
+
FrtQuery *frt_spanmtq_new(FrtSymbol field)
|
1697
|
+
{
|
1698
|
+
return frt_spanmtq_new_conf(field, SPAN_MULTI_TERM_QUERY_CAPA);
|
1699
|
+
}
|
1700
|
+
|
1701
|
+
void frt_spanmtq_add_term(FrtQuery *self, const char *term)
|
1702
|
+
{
|
1703
|
+
FrtSpanMultiTermQuery *smtq = SpMTQ(self);
|
1704
|
+
if (smtq->term_cnt < smtq->term_capa) {
|
1705
|
+
smtq->terms[smtq->term_cnt++] = frt_estrdup(term);
|
1706
|
+
}
|
1707
|
+
}
|
1708
|
+
|
1709
|
+
/*****************************************************************************
|
1710
|
+
*
|
1711
|
+
* SpanFirstQuery
|
1712
|
+
*
|
1713
|
+
*****************************************************************************/
|
1714
|
+
|
1715
|
+
static char *spanfq_to_s(FrtQuery *self, FrtSymbol field)
|
1716
|
+
{
|
1717
|
+
FrtQuery *match = SpFQ(self)->match;
|
1718
|
+
char *q_str = match->to_s(match, field);
|
1719
|
+
char *res = frt_strfmt("span_first(%s, %d)", q_str, SpFQ(self)->end);
|
1720
|
+
free(q_str);
|
1721
|
+
return res;
|
1722
|
+
}
|
1723
|
+
|
1724
|
+
static void spanfq_extract_terms(FrtQuery *self, FrtHashSet *terms)
|
1725
|
+
{
|
1726
|
+
SpFQ(self)->match->extract_terms(SpFQ(self)->match, terms);
|
1727
|
+
}
|
1728
|
+
|
1729
|
+
static FrtHashSet *spanfq_get_terms(FrtQuery *self)
|
1730
|
+
{
|
1731
|
+
FrtSpanFirstQuery *sfq = SpFQ(self);
|
1732
|
+
return SpQ(sfq->match)->get_terms(sfq->match);
|
1733
|
+
}
|
1734
|
+
|
1735
|
+
static FrtQuery *spanfq_rewrite(FrtQuery *self, FrtIndexReader *ir)
|
1736
|
+
{
|
1737
|
+
FrtQuery *q, *rq;
|
1738
|
+
|
1739
|
+
q = SpFQ(self)->match;
|
1740
|
+
rq = q->rewrite(q, ir);
|
1741
|
+
frt_q_deref(q);
|
1742
|
+
SpFQ(self)->match = rq;
|
1743
|
+
|
1744
|
+
self->ref_cnt++;
|
1745
|
+
return self; /* no clauses rewrote */
|
1746
|
+
}
|
1747
|
+
|
1748
|
+
static void spanfq_destroy_i(FrtQuery *self)
|
1749
|
+
{
|
1750
|
+
frt_q_deref(SpFQ(self)->match);
|
1751
|
+
spanq_destroy_i(self);
|
1752
|
+
}
|
1753
|
+
|
1754
|
+
static unsigned long long spanfq_hash(FrtQuery *self)
|
1755
|
+
{
|
1756
|
+
return spanq_hash(self) ^ SpFQ(self)->match->hash(SpFQ(self)->match)
|
1757
|
+
^ SpFQ(self)->end;
|
1758
|
+
}
|
1759
|
+
|
1760
|
+
static int spanfq_eq(FrtQuery *self, FrtQuery *o)
|
1761
|
+
{
|
1762
|
+
FrtSpanFirstQuery *sfq1 = SpFQ(self);
|
1763
|
+
FrtSpanFirstQuery *sfq2 = SpFQ(o);
|
1764
|
+
return spanq_eq(self, o) && sfq1->match->eq(sfq1->match, sfq2->match)
|
1765
|
+
&& (sfq1->end == sfq2->end);
|
1766
|
+
}
|
1767
|
+
|
1768
|
+
FrtQuery *frt_spanfq_new_nr(FrtQuery *match, int end)
|
1769
|
+
{
|
1770
|
+
FrtQuery *self = frt_q_new(FrtSpanFirstQuery);
|
1771
|
+
|
1772
|
+
SpFQ(self)->match = match;
|
1773
|
+
SpFQ(self)->end = end;
|
1774
|
+
|
1775
|
+
SpQ(self)->field = SpQ(match)->field;
|
1776
|
+
SpQ(self)->get_spans = &spanfe_new;
|
1777
|
+
SpQ(self)->get_terms = &spanfq_get_terms;
|
1778
|
+
|
1779
|
+
self->type = SPAN_FIRST_QUERY;
|
1780
|
+
self->rewrite = &spanfq_rewrite;
|
1781
|
+
self->extract_terms = &spanfq_extract_terms;
|
1782
|
+
self->to_s = &spanfq_to_s;
|
1783
|
+
self->hash = &spanfq_hash;
|
1784
|
+
self->eq = &spanfq_eq;
|
1785
|
+
self->destroy_i = &spanfq_destroy_i;
|
1786
|
+
self->create_weight_i = &spanw_new;
|
1787
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
1788
|
+
|
1789
|
+
return self;
|
1790
|
+
}
|
1791
|
+
|
1792
|
+
FrtQuery *frt_spanfq_new(FrtQuery *match, int end)
|
1793
|
+
{
|
1794
|
+
FRT_REF(match);
|
1795
|
+
return frt_spanfq_new_nr(match, end);
|
1796
|
+
}
|
1797
|
+
|
1798
|
+
/*****************************************************************************
|
1799
|
+
*
|
1800
|
+
* FrtSpanOrQuery
|
1801
|
+
*
|
1802
|
+
*****************************************************************************/
|
1803
|
+
|
1804
|
+
static char *spanoq_to_s(FrtQuery *self, FrtSymbol field)
|
1805
|
+
{
|
1806
|
+
int i;
|
1807
|
+
FrtSpanOrQuery *soq = SpOQ(self);
|
1808
|
+
char *res, *res_p;
|
1809
|
+
char **q_strs = FRT_ALLOC_N(char *, soq->c_cnt);
|
1810
|
+
int len = 50;
|
1811
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1812
|
+
FrtQuery *clause = soq->clauses[i];
|
1813
|
+
q_strs[i] = clause->to_s(clause, field);
|
1814
|
+
len += strlen(q_strs[i]) + 2;
|
1815
|
+
}
|
1816
|
+
|
1817
|
+
res_p = res = FRT_ALLOC_N(char, len);
|
1818
|
+
res_p += sprintf(res_p, "span_or[");
|
1819
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1820
|
+
if (i != 0) *(res_p++) = ',';
|
1821
|
+
res_p += sprintf(res_p, "%s", q_strs[i]);
|
1822
|
+
free(q_strs[i]);
|
1823
|
+
}
|
1824
|
+
free(q_strs);
|
1825
|
+
|
1826
|
+
*(res_p)++ = ']';
|
1827
|
+
*res_p = 0;
|
1828
|
+
return res;
|
1829
|
+
}
|
1830
|
+
|
1831
|
+
static void spanoq_extract_terms(FrtQuery *self, FrtHashSet *terms)
|
1832
|
+
{
|
1833
|
+
FrtSpanOrQuery *soq = SpOQ(self);
|
1834
|
+
int i;
|
1835
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1836
|
+
FrtQuery *clause = soq->clauses[i];
|
1837
|
+
clause->extract_terms(clause, terms);
|
1838
|
+
}
|
1839
|
+
}
|
1840
|
+
|
1841
|
+
static FrtHashSet *spanoq_get_terms(FrtQuery *self)
|
1842
|
+
{
|
1843
|
+
FrtSpanOrQuery *soq = SpOQ(self);
|
1844
|
+
FrtHashSet *terms = frt_hs_new_str(&free);
|
1845
|
+
int i;
|
1846
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1847
|
+
FrtQuery *clause = soq->clauses[i];
|
1848
|
+
FrtHashSet *sub_terms = SpQ(clause)->get_terms(clause);
|
1849
|
+
frt_hs_merge(terms, sub_terms);
|
1850
|
+
}
|
1851
|
+
|
1852
|
+
return terms;
|
1853
|
+
}
|
1854
|
+
|
1855
|
+
static FrtSpanEnum *spanoq_get_spans(FrtQuery *self, FrtIndexReader *ir)
|
1856
|
+
{
|
1857
|
+
FrtSpanOrQuery *soq = SpOQ(self);
|
1858
|
+
if (soq->c_cnt == 1) {
|
1859
|
+
FrtQuery *q = soq->clauses[0];
|
1860
|
+
return SpQ(q)->get_spans(q, ir);
|
1861
|
+
}
|
1862
|
+
|
1863
|
+
return spanoe_new(self, ir);
|
1864
|
+
}
|
1865
|
+
|
1866
|
+
static FrtQuery *spanoq_rewrite(FrtQuery *self, FrtIndexReader *ir)
|
1867
|
+
{
|
1868
|
+
FrtSpanOrQuery *soq = SpOQ(self);
|
1869
|
+
int i;
|
1870
|
+
|
1871
|
+
/* replace clauses with their rewritten queries */
|
1872
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1873
|
+
FrtQuery *clause = soq->clauses[i];
|
1874
|
+
FrtQuery *rewritten = clause->rewrite(clause, ir);
|
1875
|
+
frt_q_deref(clause);
|
1876
|
+
soq->clauses[i] = rewritten;
|
1877
|
+
}
|
1878
|
+
|
1879
|
+
self->ref_cnt++;
|
1880
|
+
return self;
|
1881
|
+
}
|
1882
|
+
|
1883
|
+
static void spanoq_destroy_i(FrtQuery *self)
|
1884
|
+
{
|
1885
|
+
FrtSpanOrQuery *soq = SpOQ(self);
|
1886
|
+
|
1887
|
+
int i;
|
1888
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1889
|
+
FrtQuery *clause = soq->clauses[i];
|
1890
|
+
frt_q_deref(clause);
|
1891
|
+
}
|
1892
|
+
free(soq->clauses);
|
1893
|
+
|
1894
|
+
spanq_destroy_i(self);
|
1895
|
+
}
|
1896
|
+
|
1897
|
+
static unsigned long long spanoq_hash(FrtQuery *self)
|
1898
|
+
{
|
1899
|
+
int i;
|
1900
|
+
unsigned long long hash = spanq_hash(self);
|
1901
|
+
FrtSpanOrQuery *soq = SpOQ(self);
|
1902
|
+
|
1903
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1904
|
+
FrtQuery *q = soq->clauses[i];
|
1905
|
+
hash ^= q->hash(q);
|
1906
|
+
}
|
1907
|
+
return hash;
|
1908
|
+
}
|
1909
|
+
|
1910
|
+
static int spanoq_eq(FrtQuery *self, FrtQuery *o)
|
1911
|
+
{
|
1912
|
+
int i;
|
1913
|
+
FrtQuery *q1, *q2;
|
1914
|
+
FrtSpanOrQuery *soq1 = SpOQ(self);
|
1915
|
+
FrtSpanOrQuery *soq2 = SpOQ(o);
|
1916
|
+
|
1917
|
+
if (!spanq_eq(self, o) || soq1->c_cnt != soq2->c_cnt) {
|
1918
|
+
return false;
|
1919
|
+
}
|
1920
|
+
for (i = 0; i < soq1->c_cnt; i++) {
|
1921
|
+
q1 = soq1->clauses[i];
|
1922
|
+
q2 = soq2->clauses[i];
|
1923
|
+
if (!q1->eq(q1, q2)) {
|
1924
|
+
return false;
|
1925
|
+
}
|
1926
|
+
}
|
1927
|
+
return true;
|
1928
|
+
}
|
1929
|
+
|
1930
|
+
FrtQuery *frt_spanoq_new()
|
1931
|
+
{
|
1932
|
+
FrtQuery *self = frt_q_new(FrtSpanOrQuery);
|
1933
|
+
SpOQ(self)->clauses = FRT_ALLOC_N(FrtQuery *, CLAUSE_INIT_CAPA);
|
1934
|
+
SpOQ(self)->c_capa = CLAUSE_INIT_CAPA;
|
1935
|
+
|
1936
|
+
SpQ(self)->field = (FrtSymbol)NULL;
|
1937
|
+
SpQ(self)->get_spans = &spanoq_get_spans;
|
1938
|
+
SpQ(self)->get_terms = &spanoq_get_terms;
|
1939
|
+
|
1940
|
+
self->type = SPAN_OR_QUERY;
|
1941
|
+
self->rewrite = &spanoq_rewrite;
|
1942
|
+
self->extract_terms = &spanoq_extract_terms;
|
1943
|
+
self->to_s = &spanoq_to_s;
|
1944
|
+
self->hash = &spanoq_hash;
|
1945
|
+
self->eq = &spanoq_eq;
|
1946
|
+
self->destroy_i = &spanoq_destroy_i;
|
1947
|
+
self->create_weight_i = &spanw_new;
|
1948
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
1949
|
+
|
1950
|
+
return self;
|
1951
|
+
}
|
1952
|
+
|
1953
|
+
FrtQuery *frt_spanoq_add_clause_nr(FrtQuery *self, FrtQuery *clause)
|
1954
|
+
{
|
1955
|
+
const int curr_index = SpOQ(self)->c_cnt++;
|
1956
|
+
if (clause->type < SPAN_TERM_QUERY || clause->type > SPAN_NEAR_QUERY) {
|
1957
|
+
FRT_RAISE(FRT_ARG_ERROR, "Tried to add a %s to a SpanOrQuery. This is not a "
|
1958
|
+
"SpanQuery.", frt_q_get_query_name(clause->type));
|
1959
|
+
}
|
1960
|
+
if (curr_index == 0) {
|
1961
|
+
SpQ(self)->field = SpQ(clause)->field;
|
1962
|
+
}
|
1963
|
+
else if (SpQ(self)->field != SpQ(clause)->field) {
|
1964
|
+
FRT_RAISE(FRT_ARG_ERROR, "All clauses in a SpanQuery must have the same field. "
|
1965
|
+
"Attempted to add a SpanQuery with field \"%s\" to a SpanOrQuery "
|
1966
|
+
"with field \"%s\"", rb_id2name(SpQ(clause)->field), rb_id2name(SpQ(self)->field));
|
1967
|
+
}
|
1968
|
+
if (curr_index >= SpOQ(self)->c_capa) {
|
1969
|
+
SpOQ(self)->c_capa <<= 1;
|
1970
|
+
FRT_REALLOC_N(SpOQ(self)->clauses, FrtQuery *, SpOQ(self)->c_capa);
|
1971
|
+
}
|
1972
|
+
SpOQ(self)->clauses[curr_index] = clause;
|
1973
|
+
return clause;
|
1974
|
+
}
|
1975
|
+
|
1976
|
+
FrtQuery *frt_spanoq_add_clause(FrtQuery *self, FrtQuery *clause)
|
1977
|
+
{
|
1978
|
+
FRT_REF(clause);
|
1979
|
+
return frt_spanoq_add_clause_nr(self, clause);
|
1980
|
+
}
|
1981
|
+
|
1982
|
+
/*****************************************************************************
|
1983
|
+
*
|
1984
|
+
* SpanNearQuery
|
1985
|
+
*
|
1986
|
+
*****************************************************************************/
|
1987
|
+
|
1988
|
+
static char *spannq_to_s(FrtQuery *self, FrtSymbol field)
|
1989
|
+
{
|
1990
|
+
int i;
|
1991
|
+
FrtSpanNearQuery *snq = SpNQ(self);
|
1992
|
+
char *res, *res_p;
|
1993
|
+
char **q_strs = FRT_ALLOC_N(char *, snq->c_cnt);
|
1994
|
+
int len = 50;
|
1995
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
1996
|
+
FrtQuery *clause = snq->clauses[i];
|
1997
|
+
q_strs[i] = clause->to_s(clause, field);
|
1998
|
+
len += strlen(q_strs[i]);
|
1999
|
+
}
|
2000
|
+
|
2001
|
+
res_p = res = FRT_ALLOC_N(char, len);
|
2002
|
+
res_p += sprintf(res_p, "span_near[");
|
2003
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2004
|
+
if (i != 0) *(res_p)++ = ',';
|
2005
|
+
res_p += sprintf(res_p, "%s", q_strs[i]);
|
2006
|
+
free(q_strs[i]);
|
2007
|
+
}
|
2008
|
+
free(q_strs);
|
2009
|
+
|
2010
|
+
*(res_p++) = ']';
|
2011
|
+
*res_p = 0;
|
2012
|
+
return res;
|
2013
|
+
}
|
2014
|
+
|
2015
|
+
static void spannq_extract_terms(FrtQuery *self, FrtHashSet *terms)
|
2016
|
+
{
|
2017
|
+
FrtSpanNearQuery *snq = SpNQ(self);
|
2018
|
+
int i;
|
2019
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2020
|
+
FrtQuery *clause = snq->clauses[i];
|
2021
|
+
clause->extract_terms(clause, terms);
|
2022
|
+
}
|
2023
|
+
}
|
2024
|
+
|
2025
|
+
static FrtHashSet *spannq_get_terms(FrtQuery *self)
|
2026
|
+
{
|
2027
|
+
FrtSpanNearQuery *snq = SpNQ(self);
|
2028
|
+
FrtHashSet *terms = frt_hs_new_str(&free);
|
2029
|
+
int i;
|
2030
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2031
|
+
FrtQuery *clause = snq->clauses[i];
|
2032
|
+
FrtHashSet *sub_terms = SpQ(clause)->get_terms(clause);
|
2033
|
+
frt_hs_merge(terms, sub_terms);
|
2034
|
+
}
|
2035
|
+
|
2036
|
+
return terms;
|
2037
|
+
}
|
2038
|
+
|
2039
|
+
static FrtSpanEnum *spannq_get_spans(FrtQuery *self, FrtIndexReader *ir)
|
2040
|
+
{
|
2041
|
+
FrtSpanNearQuery *snq = SpNQ(self);
|
2042
|
+
|
2043
|
+
if (snq->c_cnt == 1) {
|
2044
|
+
FrtQuery *q = snq->clauses[0];
|
2045
|
+
return SpQ(q)->get_spans(q, ir);
|
2046
|
+
}
|
2047
|
+
|
2048
|
+
return spanne_new(self, ir);
|
2049
|
+
}
|
2050
|
+
|
2051
|
+
static FrtQuery *spannq_rewrite(FrtQuery *self, FrtIndexReader *ir)
|
2052
|
+
{
|
2053
|
+
FrtSpanNearQuery *snq = SpNQ(self);
|
2054
|
+
int i;
|
2055
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2056
|
+
FrtQuery *clause = snq->clauses[i];
|
2057
|
+
FrtQuery *rewritten = clause->rewrite(clause, ir);
|
2058
|
+
frt_q_deref(clause);
|
2059
|
+
snq->clauses[i] = rewritten;
|
2060
|
+
}
|
2061
|
+
|
2062
|
+
self->ref_cnt++;
|
2063
|
+
return self;
|
2064
|
+
}
|
2065
|
+
|
2066
|
+
static void spannq_destroy(FrtQuery *self)
|
2067
|
+
{
|
2068
|
+
FrtSpanNearQuery *snq = SpNQ(self);
|
2069
|
+
|
2070
|
+
int i;
|
2071
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2072
|
+
FrtQuery *clause = snq->clauses[i];
|
2073
|
+
frt_q_deref(clause);
|
2074
|
+
}
|
2075
|
+
free(snq->clauses);
|
2076
|
+
|
2077
|
+
spanq_destroy_i(self);
|
2078
|
+
}
|
2079
|
+
|
2080
|
+
static unsigned long long spannq_hash(FrtQuery *self)
|
2081
|
+
{
|
2082
|
+
int i;
|
2083
|
+
unsigned long long hash = spanq_hash(self);
|
2084
|
+
FrtSpanNearQuery *snq = SpNQ(self);
|
2085
|
+
|
2086
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2087
|
+
FrtQuery *q = snq->clauses[i];
|
2088
|
+
hash ^= q->hash(q);
|
2089
|
+
}
|
2090
|
+
return ((hash ^ snq->slop) << 1) | snq->in_order;
|
2091
|
+
}
|
2092
|
+
|
2093
|
+
static int spannq_eq(FrtQuery *self, FrtQuery *o)
|
2094
|
+
{
|
2095
|
+
int i;
|
2096
|
+
FrtQuery *q1, *q2;
|
2097
|
+
FrtSpanNearQuery *snq1 = SpNQ(self);
|
2098
|
+
FrtSpanNearQuery *snq2 = SpNQ(o);
|
2099
|
+
if (! spanq_eq(self, o)
|
2100
|
+
|| (snq1->c_cnt != snq2->c_cnt)
|
2101
|
+
|| (snq1->slop != snq2->slop)
|
2102
|
+
|| (snq1->in_order != snq2->in_order)) {
|
2103
|
+
return false;
|
2104
|
+
}
|
2105
|
+
|
2106
|
+
for (i = 0; i < snq1->c_cnt; i++) {
|
2107
|
+
q1 = snq1->clauses[i];
|
2108
|
+
q2 = snq2->clauses[i];
|
2109
|
+
if (!q1->eq(q1, q2)) {
|
2110
|
+
return false;
|
2111
|
+
}
|
2112
|
+
}
|
2113
|
+
|
2114
|
+
return true;
|
2115
|
+
}
|
2116
|
+
|
2117
|
+
FrtQuery *frt_spannq_new(int slop, bool in_order)
|
2118
|
+
{
|
2119
|
+
FrtQuery *self = frt_q_new(FrtSpanNearQuery);
|
2120
|
+
|
2121
|
+
SpNQ(self)->clauses = FRT_ALLOC_N(FrtQuery *, CLAUSE_INIT_CAPA);
|
2122
|
+
SpNQ(self)->c_capa = CLAUSE_INIT_CAPA;
|
2123
|
+
SpNQ(self)->slop = slop;
|
2124
|
+
SpNQ(self)->in_order = in_order;
|
2125
|
+
|
2126
|
+
SpQ(self)->get_spans = &spannq_get_spans;
|
2127
|
+
SpQ(self)->get_terms = &spannq_get_terms;
|
2128
|
+
SpQ(self)->field = (FrtSymbol)NULL;
|
2129
|
+
|
2130
|
+
self->type = SPAN_NEAR_QUERY;
|
2131
|
+
self->rewrite = &spannq_rewrite;
|
2132
|
+
self->extract_terms = &spannq_extract_terms;
|
2133
|
+
self->to_s = &spannq_to_s;
|
2134
|
+
self->hash = &spannq_hash;
|
2135
|
+
self->eq = &spannq_eq;
|
2136
|
+
self->destroy_i = &spannq_destroy;
|
2137
|
+
self->create_weight_i = &spanw_new;
|
2138
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
2139
|
+
|
2140
|
+
return self;
|
2141
|
+
}
|
2142
|
+
|
2143
|
+
FrtQuery *frt_spannq_add_clause_nr(FrtQuery *self, FrtQuery *clause)
|
2144
|
+
{
|
2145
|
+
const int curr_index = SpNQ(self)->c_cnt++;
|
2146
|
+
if (clause->type < SPAN_TERM_QUERY || clause->type > SPAN_NEAR_QUERY) {
|
2147
|
+
FRT_RAISE(FRT_ARG_ERROR, "Tried to add a %s to a SpanNearQuery. This is not a "
|
2148
|
+
"SpanQuery.", frt_q_get_query_name(clause->type));
|
2149
|
+
}
|
2150
|
+
if (curr_index == 0) {
|
2151
|
+
SpQ(self)->field = SpQ(clause)->field;
|
2152
|
+
}
|
2153
|
+
else if (SpQ(self)->field != SpQ(clause)->field) {
|
2154
|
+
FRT_RAISE(FRT_ARG_ERROR, "All clauses in a SpanQuery must have the same field. "
|
2155
|
+
"Attempted to add a SpanQuery with field \"%s\" to SpanNearQuery "
|
2156
|
+
"with field \"%s\"", rb_id2name(SpQ(clause)->field), rb_id2name(SpQ(self)->field));
|
2157
|
+
}
|
2158
|
+
if (curr_index >= SpNQ(self)->c_capa) {
|
2159
|
+
SpNQ(self)->c_capa <<= 1;
|
2160
|
+
FRT_REALLOC_N(SpNQ(self)->clauses, FrtQuery *, SpNQ(self)->c_capa);
|
2161
|
+
}
|
2162
|
+
SpNQ(self)->clauses[curr_index] = clause;
|
2163
|
+
return clause;
|
2164
|
+
}
|
2165
|
+
|
2166
|
+
FrtQuery *frt_spannq_add_clause(FrtQuery *self, FrtQuery *clause)
|
2167
|
+
{
|
2168
|
+
FRT_REF(clause);
|
2169
|
+
return frt_spannq_add_clause_nr(self, clause);
|
2170
|
+
}
|
2171
|
+
|
2172
|
+
/*****************************************************************************
|
2173
|
+
*
|
2174
|
+
* FrtSpanNotQuery
|
2175
|
+
*
|
2176
|
+
*****************************************************************************/
|
2177
|
+
|
2178
|
+
static char *spanxq_to_s(FrtQuery *self, FrtSymbol field)
|
2179
|
+
{
|
2180
|
+
FrtSpanNotQuery *sxq = SpXQ(self);
|
2181
|
+
char *inc_s = sxq->inc->to_s(sxq->inc, field);
|
2182
|
+
char *exc_s = sxq->exc->to_s(sxq->exc, field);
|
2183
|
+
char *res = frt_strfmt("span_not(inc:<%s>, exc:<%s>)", inc_s, exc_s);
|
2184
|
+
|
2185
|
+
free(inc_s);
|
2186
|
+
free(exc_s);
|
2187
|
+
return res;
|
2188
|
+
}
|
2189
|
+
|
2190
|
+
static void spanxq_extract_terms(FrtQuery *self, FrtHashSet *terms)
|
2191
|
+
{
|
2192
|
+
SpXQ(self)->inc->extract_terms(SpXQ(self)->inc, terms);
|
2193
|
+
}
|
2194
|
+
|
2195
|
+
static FrtHashSet *spanxq_get_terms(FrtQuery *self)
|
2196
|
+
{
|
2197
|
+
return SpQ(SpXQ(self)->inc)->get_terms(SpXQ(self)->inc);
|
2198
|
+
}
|
2199
|
+
|
2200
|
+
static FrtQuery *spanxq_rewrite(FrtQuery *self, FrtIndexReader *ir)
|
2201
|
+
{
|
2202
|
+
FrtSpanNotQuery *sxq = SpXQ(self);
|
2203
|
+
FrtQuery *q, *rq;
|
2204
|
+
|
2205
|
+
/* rewrite inclusive query */
|
2206
|
+
q = sxq->inc;
|
2207
|
+
rq = q->rewrite(q, ir);
|
2208
|
+
frt_q_deref(q);
|
2209
|
+
sxq->inc = rq;
|
2210
|
+
|
2211
|
+
/* rewrite exclusive query */
|
2212
|
+
q = sxq->exc;
|
2213
|
+
rq = q->rewrite(q, ir);
|
2214
|
+
frt_q_deref(q);
|
2215
|
+
sxq->exc = rq;
|
2216
|
+
|
2217
|
+
self->ref_cnt++;
|
2218
|
+
return self;
|
2219
|
+
}
|
2220
|
+
|
2221
|
+
static void spanxq_destroy(FrtQuery *self)
|
2222
|
+
{
|
2223
|
+
FrtSpanNotQuery *sxq = SpXQ(self);
|
2224
|
+
|
2225
|
+
frt_q_deref(sxq->inc);
|
2226
|
+
frt_q_deref(sxq->exc);
|
2227
|
+
|
2228
|
+
spanq_destroy_i(self);
|
2229
|
+
}
|
2230
|
+
|
2231
|
+
static unsigned long long spanxq_hash(FrtQuery *self)
|
2232
|
+
{
|
2233
|
+
FrtSpanNotQuery *sxq = SpXQ(self);
|
2234
|
+
return spanq_hash(self) ^ sxq->inc->hash(sxq->inc)
|
2235
|
+
^ sxq->exc->hash(sxq->exc);
|
2236
|
+
}
|
2237
|
+
|
2238
|
+
static int spanxq_eq(FrtQuery *self, FrtQuery *o)
|
2239
|
+
{
|
2240
|
+
FrtSpanNotQuery *sxq1 = SpXQ(self);
|
2241
|
+
FrtSpanNotQuery *sxq2 = SpXQ(o);
|
2242
|
+
return spanq_eq(self, o) && sxq1->inc->eq(sxq1->inc, sxq2->inc)
|
2243
|
+
&& sxq1->exc->eq(sxq1->exc, sxq2->exc);
|
2244
|
+
}
|
2245
|
+
|
2246
|
+
|
2247
|
+
FrtQuery *frt_spanxq_new_nr(FrtQuery *inc, FrtQuery *exc)
|
2248
|
+
{
|
2249
|
+
FrtQuery *self;
|
2250
|
+
if (SpQ(inc)->field != SpQ(exc)->field) {
|
2251
|
+
FRT_RAISE(FRT_ARG_ERROR, "All clauses in a SpanQuery must have the same field. "
|
2252
|
+
"Attempted to add a SpanQuery with field \"%s\" along with a "
|
2253
|
+
"SpanQuery with field \"%s\" to an SpanNotQuery",
|
2254
|
+
rb_id2name(SpQ(inc)->field), rb_id2name(SpQ(exc)->field));
|
2255
|
+
}
|
2256
|
+
self = frt_q_new(FrtSpanNotQuery);
|
2257
|
+
|
2258
|
+
SpXQ(self)->inc = inc;
|
2259
|
+
SpXQ(self)->exc = exc;
|
2260
|
+
|
2261
|
+
SpQ(self)->field = SpQ(inc)->field;
|
2262
|
+
SpQ(self)->get_spans = &spanxe_new;
|
2263
|
+
SpQ(self)->get_terms = &spanxq_get_terms;
|
2264
|
+
|
2265
|
+
self->type = SPAN_NOT_QUERY;
|
2266
|
+
self->rewrite = &spanxq_rewrite;
|
2267
|
+
self->extract_terms = &spanxq_extract_terms;
|
2268
|
+
self->to_s = &spanxq_to_s;
|
2269
|
+
self->hash = &spanxq_hash;
|
2270
|
+
self->eq = &spanxq_eq;
|
2271
|
+
self->destroy_i = &spanxq_destroy;
|
2272
|
+
self->create_weight_i = &spanw_new;
|
2273
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
2274
|
+
|
2275
|
+
return self;
|
2276
|
+
}
|
2277
|
+
|
2278
|
+
FrtQuery *frt_spanxq_new(FrtQuery *inc, FrtQuery *exc)
|
2279
|
+
{
|
2280
|
+
FRT_REF(inc);
|
2281
|
+
FRT_REF(exc);
|
2282
|
+
return frt_spanxq_new_nr(inc, exc);
|
2283
|
+
}
|
2284
|
+
|
2285
|
+
|
2286
|
+
/*****************************************************************************
|
2287
|
+
*
|
2288
|
+
* Rewritables
|
2289
|
+
*
|
2290
|
+
*****************************************************************************/
|
2291
|
+
|
2292
|
+
/*****************************************************************************
|
2293
|
+
*
|
2294
|
+
* FrtSpanPrefixQuery
|
2295
|
+
*
|
2296
|
+
*****************************************************************************/
|
2297
|
+
|
2298
|
+
#define SpPfxQ(query) ((FrtSpanPrefixQuery *)(query))
|
2299
|
+
|
2300
|
+
static char *spanprq_to_s(FrtQuery *self, FrtSymbol default_field)
|
2301
|
+
{
|
2302
|
+
char *buffer, *bptr;
|
2303
|
+
const char *prefix = SpPfxQ(self)->prefix;
|
2304
|
+
size_t plen = strlen(prefix);
|
2305
|
+
FrtSymbol field = SpQ(self)->field;
|
2306
|
+
const char *field_name = rb_id2name(field);
|
2307
|
+
size_t flen = strlen(field_name);
|
2308
|
+
|
2309
|
+
|
2310
|
+
bptr = buffer = FRT_ALLOC_N(char, plen + flen + 35);
|
2311
|
+
|
2312
|
+
if (default_field == (FrtSymbol)NULL || (field != default_field)) {
|
2313
|
+
bptr += sprintf(bptr, "%s:", field_name);
|
2314
|
+
}
|
2315
|
+
|
2316
|
+
bptr += sprintf(bptr, "%s*", prefix);
|
2317
|
+
if (self->boost != 1.0) {
|
2318
|
+
*bptr = '^';
|
2319
|
+
frt_dbl_to_s(++bptr, self->boost);
|
2320
|
+
}
|
2321
|
+
|
2322
|
+
return buffer;
|
2323
|
+
}
|
2324
|
+
|
2325
|
+
static FrtQuery *spanprq_rewrite(FrtQuery *self, FrtIndexReader *ir)
|
2326
|
+
{
|
2327
|
+
const int field_num = frt_fis_get_field_num(ir->fis, SpQ(self)->field);
|
2328
|
+
FrtQuery *volatile q = frt_spanmtq_new_conf(SpQ(self)->field, SpPfxQ(self)->max_terms);
|
2329
|
+
q->boost = self->boost; /* set the boost */
|
2330
|
+
|
2331
|
+
if (field_num >= 0) {
|
2332
|
+
const char *prefix = SpPfxQ(self)->prefix;
|
2333
|
+
FrtTermEnum *te = ir->terms_from(ir, field_num, prefix);
|
2334
|
+
const char *term = te->curr_term;
|
2335
|
+
size_t prefix_len = strlen(prefix);
|
2336
|
+
|
2337
|
+
FRT_TRY
|
2338
|
+
do {
|
2339
|
+
if (strncmp(term, prefix, prefix_len) != 0) {
|
2340
|
+
break;
|
2341
|
+
}
|
2342
|
+
frt_spanmtq_add_term(q, term); /* found a match */
|
2343
|
+
} while (te->next(te));
|
2344
|
+
FRT_XFINALLY
|
2345
|
+
te->close(te);
|
2346
|
+
FRT_XENDTRY
|
2347
|
+
}
|
2348
|
+
|
2349
|
+
return q;
|
2350
|
+
}
|
2351
|
+
|
2352
|
+
static void spanprq_destroy(FrtQuery *self)
|
2353
|
+
{
|
2354
|
+
free(SpPfxQ(self)->prefix);
|
2355
|
+
spanq_destroy_i(self);
|
2356
|
+
}
|
2357
|
+
|
2358
|
+
static unsigned long long spanprq_hash(FrtQuery *self)
|
2359
|
+
{
|
2360
|
+
return frt_str_hash(rb_id2name(SpQ(self)->field)) ^ frt_str_hash(SpPfxQ(self)->prefix);
|
2361
|
+
}
|
2362
|
+
|
2363
|
+
static int spanprq_eq(FrtQuery *self, FrtQuery *o)
|
2364
|
+
{
|
2365
|
+
return (strcmp(SpPfxQ(self)->prefix, SpPfxQ(o)->prefix) == 0)
|
2366
|
+
&& (SpQ(self)->field == SpQ(o)->field);
|
2367
|
+
}
|
2368
|
+
|
2369
|
+
FrtQuery *frt_spanprq_new(FrtSymbol field, const char *prefix)
|
2370
|
+
{
|
2371
|
+
FrtQuery *self = frt_q_new(FrtSpanPrefixQuery);
|
2372
|
+
|
2373
|
+
SpQ(self)->field = field;
|
2374
|
+
SpPfxQ(self)->prefix = frt_estrdup(prefix);
|
2375
|
+
SpPfxQ(self)->max_terms = FRT_SPAN_PREFIX_QUERY_MAX_TERMS;
|
2376
|
+
|
2377
|
+
self->type = SPAN_PREFIX_QUERY;
|
2378
|
+
self->rewrite = &spanprq_rewrite;
|
2379
|
+
self->to_s = &spanprq_to_s;
|
2380
|
+
self->hash = &spanprq_hash;
|
2381
|
+
self->eq = &spanprq_eq;
|
2382
|
+
self->destroy_i = &spanprq_destroy;
|
2383
|
+
self->create_weight_i = &frt_q_create_weight_unsup;
|
2384
|
+
|
2385
|
+
return self;
|
2386
|
+
}
|