ferret 0.9.6 → 0.10.0
Sign up to get free protection for your applications and to get access to all the features.
- data/MIT-LICENSE +1 -1
- data/README +12 -24
- data/Rakefile +38 -54
- data/TODO +14 -17
- data/ext/analysis.c +982 -823
- data/ext/analysis.h +133 -76
- data/ext/array.c +96 -58
- data/ext/array.h +40 -13
- data/ext/bitvector.c +476 -118
- data/ext/bitvector.h +264 -22
- data/ext/compound_io.c +217 -229
- data/ext/defines.h +49 -0
- data/ext/document.c +107 -317
- data/ext/document.h +31 -65
- data/ext/except.c +81 -36
- data/ext/except.h +117 -55
- data/ext/extconf.rb +2 -9
- data/ext/ferret.c +211 -104
- data/ext/ferret.h +22 -11
- data/ext/filter.c +97 -82
- data/ext/fs_store.c +348 -367
- data/ext/global.c +226 -188
- data/ext/global.h +44 -26
- data/ext/hash.c +474 -391
- data/ext/hash.h +441 -68
- data/ext/hashset.c +124 -96
- data/ext/hashset.h +169 -20
- data/ext/helper.c +56 -5
- data/ext/helper.h +7 -0
- data/ext/inc/lang.h +29 -49
- data/ext/inc/threading.h +31 -0
- data/ext/ind.c +288 -278
- data/ext/ind.h +68 -0
- data/ext/index.c +5688 -0
- data/ext/index.h +663 -616
- data/ext/lang.h +29 -49
- data/ext/libstemmer.c +3 -3
- data/ext/mem_pool.c +84 -0
- data/ext/mem_pool.h +35 -0
- data/ext/posh.c +1006 -0
- data/ext/posh.h +1007 -0
- data/ext/priorityqueue.c +117 -194
- data/ext/priorityqueue.h +135 -39
- data/ext/q_boolean.c +1305 -1108
- data/ext/q_const_score.c +106 -93
- data/ext/q_filtered_query.c +138 -135
- data/ext/q_fuzzy.c +206 -242
- data/ext/q_match_all.c +94 -80
- data/ext/q_multi_term.c +663 -0
- data/ext/q_parser.c +667 -593
- data/ext/q_phrase.c +992 -555
- data/ext/q_prefix.c +72 -61
- data/ext/q_range.c +235 -210
- data/ext/q_span.c +1480 -1166
- data/ext/q_term.c +273 -246
- data/ext/q_wildcard.c +127 -114
- data/ext/r_analysis.c +1720 -711
- data/ext/r_index.c +3049 -0
- data/ext/r_qparser.c +433 -146
- data/ext/r_search.c +2934 -1993
- data/ext/r_store.c +372 -143
- data/ext/r_utils.c +941 -0
- data/ext/ram_store.c +330 -326
- data/ext/search.c +1291 -668
- data/ext/search.h +403 -702
- data/ext/similarity.c +91 -113
- data/ext/similarity.h +45 -30
- data/ext/sort.c +721 -484
- data/ext/stopwords.c +361 -273
- data/ext/store.c +556 -58
- data/ext/store.h +706 -126
- data/ext/tags +3578 -2780
- data/ext/term_vectors.c +352 -0
- data/ext/threading.h +31 -0
- data/ext/win32.h +54 -0
- data/lib/ferret.rb +5 -17
- data/lib/ferret/document.rb +130 -2
- data/lib/ferret/index.rb +577 -26
- data/lib/ferret/number_tools.rb +157 -0
- data/lib/ferret_version.rb +3 -0
- data/test/test_helper.rb +5 -13
- data/test/unit/analysis/tc_analyzer.rb +513 -1
- data/test/unit/analysis/{ctc_tokenstream.rb → tc_token_stream.rb} +23 -0
- data/test/unit/index/tc_index.rb +183 -240
- data/test/unit/index/tc_index_reader.rb +312 -479
- data/test/unit/index/tc_index_writer.rb +397 -13
- data/test/unit/index/th_doc.rb +269 -206
- data/test/unit/query_parser/tc_query_parser.rb +40 -33
- data/test/unit/search/tc_filter.rb +59 -71
- data/test/unit/search/tc_fuzzy_query.rb +24 -16
- data/test/unit/search/tc_index_searcher.rb +23 -201
- data/test/unit/search/tc_multi_searcher.rb +78 -226
- data/test/unit/search/tc_search_and_sort.rb +93 -81
- data/test/unit/search/tc_sort.rb +23 -23
- data/test/unit/search/tc_sort_field.rb +7 -7
- data/test/unit/search/tc_spans.rb +51 -47
- data/test/unit/search/tm_searcher.rb +339 -0
- data/test/unit/store/tc_fs_store.rb +1 -1
- data/test/unit/store/tm_store_lock.rb +3 -3
- data/test/unit/tc_document.rb +81 -0
- data/test/unit/ts_analysis.rb +1 -1
- data/test/unit/ts_utils.rb +1 -1
- data/test/unit/utils/tc_bit_vector.rb +288 -0
- data/test/unit/utils/tc_number_tools.rb +117 -0
- data/test/unit/utils/tc_priority_queue.rb +106 -0
- metadata +140 -301
- data/CHANGELOG +0 -9
- data/ext/dummy.exe +0 -0
- data/ext/field.c +0 -408
- data/ext/frtio.h +0 -13
- data/ext/inc/except.h +0 -90
- data/ext/index_io.c +0 -382
- data/ext/index_rw.c +0 -2658
- data/ext/lang.c +0 -41
- data/ext/nix_io.c +0 -134
- data/ext/q_multi_phrase.c +0 -380
- data/ext/r_doc.c +0 -582
- data/ext/r_index_io.c +0 -1021
- data/ext/r_term.c +0 -219
- data/ext/term.c +0 -820
- data/ext/termdocs.c +0 -611
- data/ext/vector.c +0 -637
- data/ext/w32_io.c +0 -150
- data/lib/ferret/analysis.rb +0 -11
- data/lib/ferret/analysis/analyzers.rb +0 -112
- data/lib/ferret/analysis/standard_tokenizer.rb +0 -71
- data/lib/ferret/analysis/token.rb +0 -100
- data/lib/ferret/analysis/token_filters.rb +0 -86
- data/lib/ferret/analysis/token_stream.rb +0 -26
- data/lib/ferret/analysis/tokenizers.rb +0 -112
- data/lib/ferret/analysis/word_list_loader.rb +0 -27
- data/lib/ferret/document/document.rb +0 -152
- data/lib/ferret/document/field.rb +0 -312
- data/lib/ferret/index/compound_file_io.rb +0 -338
- data/lib/ferret/index/document_writer.rb +0 -289
- data/lib/ferret/index/field_infos.rb +0 -279
- data/lib/ferret/index/fields_io.rb +0 -181
- data/lib/ferret/index/index.rb +0 -675
- data/lib/ferret/index/index_file_names.rb +0 -33
- data/lib/ferret/index/index_reader.rb +0 -503
- data/lib/ferret/index/index_writer.rb +0 -534
- data/lib/ferret/index/multi_reader.rb +0 -377
- data/lib/ferret/index/multiple_term_doc_pos_enum.rb +0 -98
- data/lib/ferret/index/segment_infos.rb +0 -130
- data/lib/ferret/index/segment_merge_info.rb +0 -49
- data/lib/ferret/index/segment_merge_queue.rb +0 -16
- data/lib/ferret/index/segment_merger.rb +0 -358
- data/lib/ferret/index/segment_reader.rb +0 -412
- data/lib/ferret/index/segment_term_enum.rb +0 -169
- data/lib/ferret/index/segment_term_vector.rb +0 -58
- data/lib/ferret/index/term.rb +0 -53
- data/lib/ferret/index/term_buffer.rb +0 -83
- data/lib/ferret/index/term_doc_enum.rb +0 -291
- data/lib/ferret/index/term_enum.rb +0 -52
- data/lib/ferret/index/term_info.rb +0 -37
- data/lib/ferret/index/term_infos_io.rb +0 -321
- data/lib/ferret/index/term_vector_offset_info.rb +0 -20
- data/lib/ferret/index/term_vectors_io.rb +0 -553
- data/lib/ferret/query_parser.rb +0 -312
- data/lib/ferret/query_parser/query_parser.tab.rb +0 -928
- data/lib/ferret/search.rb +0 -50
- data/lib/ferret/search/boolean_clause.rb +0 -100
- data/lib/ferret/search/boolean_query.rb +0 -299
- data/lib/ferret/search/boolean_scorer.rb +0 -294
- data/lib/ferret/search/caching_wrapper_filter.rb +0 -40
- data/lib/ferret/search/conjunction_scorer.rb +0 -99
- data/lib/ferret/search/disjunction_sum_scorer.rb +0 -205
- data/lib/ferret/search/exact_phrase_scorer.rb +0 -32
- data/lib/ferret/search/explanation.rb +0 -41
- data/lib/ferret/search/field_cache.rb +0 -215
- data/lib/ferret/search/field_doc.rb +0 -31
- data/lib/ferret/search/field_sorted_hit_queue.rb +0 -184
- data/lib/ferret/search/filter.rb +0 -11
- data/lib/ferret/search/filtered_query.rb +0 -130
- data/lib/ferret/search/filtered_term_enum.rb +0 -79
- data/lib/ferret/search/fuzzy_query.rb +0 -154
- data/lib/ferret/search/fuzzy_term_enum.rb +0 -247
- data/lib/ferret/search/hit_collector.rb +0 -34
- data/lib/ferret/search/hit_queue.rb +0 -11
- data/lib/ferret/search/index_searcher.rb +0 -200
- data/lib/ferret/search/match_all_query.rb +0 -104
- data/lib/ferret/search/multi_phrase_query.rb +0 -216
- data/lib/ferret/search/multi_searcher.rb +0 -261
- data/lib/ferret/search/multi_term_query.rb +0 -65
- data/lib/ferret/search/non_matching_scorer.rb +0 -22
- data/lib/ferret/search/phrase_positions.rb +0 -55
- data/lib/ferret/search/phrase_query.rb +0 -214
- data/lib/ferret/search/phrase_scorer.rb +0 -152
- data/lib/ferret/search/prefix_query.rb +0 -54
- data/lib/ferret/search/query.rb +0 -140
- data/lib/ferret/search/query_filter.rb +0 -51
- data/lib/ferret/search/range_filter.rb +0 -103
- data/lib/ferret/search/range_query.rb +0 -139
- data/lib/ferret/search/req_excl_scorer.rb +0 -125
- data/lib/ferret/search/req_opt_sum_scorer.rb +0 -70
- data/lib/ferret/search/score_doc.rb +0 -38
- data/lib/ferret/search/score_doc_comparator.rb +0 -114
- data/lib/ferret/search/scorer.rb +0 -91
- data/lib/ferret/search/similarity.rb +0 -278
- data/lib/ferret/search/sloppy_phrase_scorer.rb +0 -47
- data/lib/ferret/search/sort.rb +0 -112
- data/lib/ferret/search/sort_comparator.rb +0 -60
- data/lib/ferret/search/sort_field.rb +0 -91
- data/lib/ferret/search/spans.rb +0 -12
- data/lib/ferret/search/spans/near_spans_enum.rb +0 -304
- data/lib/ferret/search/spans/span_first_query.rb +0 -79
- data/lib/ferret/search/spans/span_near_query.rb +0 -108
- data/lib/ferret/search/spans/span_not_query.rb +0 -130
- data/lib/ferret/search/spans/span_or_query.rb +0 -176
- data/lib/ferret/search/spans/span_query.rb +0 -25
- data/lib/ferret/search/spans/span_scorer.rb +0 -74
- data/lib/ferret/search/spans/span_term_query.rb +0 -105
- data/lib/ferret/search/spans/span_weight.rb +0 -84
- data/lib/ferret/search/spans/spans_enum.rb +0 -44
- data/lib/ferret/search/term_query.rb +0 -128
- data/lib/ferret/search/term_scorer.rb +0 -183
- data/lib/ferret/search/top_docs.rb +0 -36
- data/lib/ferret/search/top_field_docs.rb +0 -17
- data/lib/ferret/search/weight.rb +0 -54
- data/lib/ferret/search/wildcard_query.rb +0 -26
- data/lib/ferret/search/wildcard_term_enum.rb +0 -61
- data/lib/ferret/stemmers.rb +0 -1
- data/lib/ferret/stemmers/porter_stemmer.rb +0 -218
- data/lib/ferret/store.rb +0 -5
- data/lib/ferret/store/buffered_index_io.rb +0 -190
- data/lib/ferret/store/directory.rb +0 -141
- data/lib/ferret/store/fs_store.rb +0 -381
- data/lib/ferret/store/index_io.rb +0 -245
- data/lib/ferret/store/ram_store.rb +0 -286
- data/lib/ferret/utils.rb +0 -8
- data/lib/ferret/utils/bit_vector.rb +0 -123
- data/lib/ferret/utils/date_tools.rb +0 -138
- data/lib/ferret/utils/number_tools.rb +0 -91
- data/lib/ferret/utils/parameter.rb +0 -41
- data/lib/ferret/utils/priority_queue.rb +0 -120
- data/lib/ferret/utils/string_helper.rb +0 -47
- data/lib/ferret/utils/thread_local.rb +0 -28
- data/lib/ferret/utils/weak_key_hash.rb +0 -60
- data/lib/rferret.rb +0 -37
- data/rake_utils/code_statistics.rb +0 -106
- data/test/benchmark/tb_ram_store.rb +0 -76
- data/test/benchmark/tb_rw_vint.rb +0 -26
- data/test/functional/thread_safety_index_test.rb +0 -81
- data/test/functional/thread_safety_test.rb +0 -137
- data/test/longrunning/tc_numbertools.rb +0 -60
- data/test/longrunning/tm_store.rb +0 -19
- data/test/unit/analysis/ctc_analyzer.rb +0 -532
- data/test/unit/analysis/data/wordfile +0 -6
- data/test/unit/analysis/rtc_letter_tokenizer.rb +0 -20
- data/test/unit/analysis/rtc_lower_case_filter.rb +0 -20
- data/test/unit/analysis/rtc_lower_case_tokenizer.rb +0 -27
- data/test/unit/analysis/rtc_per_field_analyzer_wrapper.rb +0 -39
- data/test/unit/analysis/rtc_porter_stem_filter.rb +0 -16
- data/test/unit/analysis/rtc_standard_analyzer.rb +0 -20
- data/test/unit/analysis/rtc_standard_tokenizer.rb +0 -20
- data/test/unit/analysis/rtc_stop_analyzer.rb +0 -20
- data/test/unit/analysis/rtc_stop_filter.rb +0 -14
- data/test/unit/analysis/rtc_white_space_analyzer.rb +0 -21
- data/test/unit/analysis/rtc_white_space_tokenizer.rb +0 -20
- data/test/unit/analysis/rtc_word_list_loader.rb +0 -32
- data/test/unit/analysis/tc_token.rb +0 -25
- data/test/unit/document/rtc_field.rb +0 -28
- data/test/unit/document/tc_document.rb +0 -47
- data/test/unit/document/tc_field.rb +0 -98
- data/test/unit/index/rtc_compound_file_io.rb +0 -107
- data/test/unit/index/rtc_field_infos.rb +0 -127
- data/test/unit/index/rtc_fields_io.rb +0 -167
- data/test/unit/index/rtc_multiple_term_doc_pos_enum.rb +0 -83
- data/test/unit/index/rtc_segment_infos.rb +0 -74
- data/test/unit/index/rtc_segment_term_docs.rb +0 -17
- data/test/unit/index/rtc_segment_term_enum.rb +0 -60
- data/test/unit/index/rtc_segment_term_vector.rb +0 -71
- data/test/unit/index/rtc_term_buffer.rb +0 -57
- data/test/unit/index/rtc_term_info.rb +0 -19
- data/test/unit/index/rtc_term_infos_io.rb +0 -192
- data/test/unit/index/rtc_term_vectors_io.rb +0 -108
- data/test/unit/index/tc_term.rb +0 -27
- data/test/unit/index/tc_term_voi.rb +0 -18
- data/test/unit/search/rtc_similarity.rb +0 -37
- data/test/unit/search/rtc_sort_field.rb +0 -14
- data/test/unit/search/tc_multi_searcher2.rb +0 -126
- data/test/unit/store/rtc_fs_store.rb +0 -62
- data/test/unit/store/rtc_ram_store.rb +0 -15
- data/test/unit/store/rtm_store.rb +0 -150
- data/test/unit/store/rtm_store_lock.rb +0 -2
- data/test/unit/ts_document.rb +0 -2
- data/test/unit/utils/rtc_bit_vector.rb +0 -73
- data/test/unit/utils/rtc_date_tools.rb +0 -50
- data/test/unit/utils/rtc_number_tools.rb +0 -59
- data/test/unit/utils/rtc_parameter.rb +0 -40
- data/test/unit/utils/rtc_priority_queue.rb +0 -62
- data/test/unit/utils/rtc_string_helper.rb +0 -21
- data/test/unit/utils/rtc_thread.rb +0 -61
- data/test/unit/utils/rtc_weak_key_hash.rb +0 -25
- data/test/utils/number_to_spoken.rb +0 -132
data/ext/q_const_score.c
CHANGED
@@ -3,146 +3,159 @@
|
|
3
3
|
|
4
4
|
/***************************************************************************
|
5
5
|
*
|
6
|
-
*
|
6
|
+
* ConstantScoreScorer
|
7
7
|
*
|
8
8
|
***************************************************************************/
|
9
9
|
|
10
|
-
|
10
|
+
#define CScQ(query) ((ConstantScoreQuery *)(query))
|
11
|
+
#define CScSc(scorer) ((ConstantScoreScorer *)(scorer))
|
12
|
+
|
13
|
+
typedef struct ConstantScoreScorer
|
14
|
+
{
|
15
|
+
Scorer super;
|
16
|
+
BitVector *bv;
|
17
|
+
float score;
|
18
|
+
} ConstantScoreScorer;
|
19
|
+
|
20
|
+
static float cssc_score(Scorer *self)
|
11
21
|
{
|
12
|
-
|
22
|
+
return CScSc(self)->score;
|
13
23
|
}
|
14
24
|
|
15
|
-
|
25
|
+
static bool cssc_next(Scorer *self)
|
16
26
|
{
|
17
|
-
|
18
|
-
Explanation *expl;
|
19
|
-
char *filter_str = filter->to_s(filter);
|
20
|
-
BitVector *bv = filt_get_bv(filter, ir);
|
21
|
-
|
22
|
-
if (bv_get(bv, doc_num)) {
|
23
|
-
expl = expl_create(self->value,
|
24
|
-
strfmt("ConstantScoreQuery(%s), product of:", filter_str));
|
25
|
-
expl_add_detail(expl, expl_create(self->query->boost, estrdup("boost")));
|
26
|
-
expl_add_detail(expl, expl_create(self->qnorm, estrdup("query_norm")));
|
27
|
-
} else {
|
28
|
-
expl = expl_create(self->value,
|
29
|
-
strfmt("ConstantScoreQuery(%s), does not match id %d",
|
30
|
-
filter_str, doc_num));
|
31
|
-
}
|
32
|
-
free(filter_str);
|
33
|
-
return expl;
|
27
|
+
return ((self->doc = bv_scan_next(CScSc(self)->bv)) >= 0);
|
34
28
|
}
|
35
29
|
|
36
|
-
|
30
|
+
static bool cssc_skip_to(Scorer *self, int doc_num)
|
37
31
|
{
|
38
|
-
|
39
|
-
|
40
|
-
self->explain = &csw_explain;
|
41
|
-
self->to_s = &csw_to_s;
|
42
|
-
self->sum_of_squared_weights = &w_sum_of_squared_weights;
|
32
|
+
return ((self->doc = bv_scan_next_from(CScSc(self)->bv, doc_num)) >= 0);
|
33
|
+
}
|
43
34
|
|
44
|
-
|
45
|
-
|
35
|
+
static Explanation *cssc_explain(Scorer *self, int doc_num)
|
36
|
+
{
|
37
|
+
(void)self; (void)doc_num;
|
38
|
+
return expl_new(1.0, "ConstantScoreScorer");
|
39
|
+
}
|
46
40
|
|
47
|
-
|
41
|
+
static Scorer *cssc_new(Weight *weight, IndexReader *ir)
|
42
|
+
{
|
43
|
+
Scorer *self = scorer_new(ConstantScoreScorer, weight->similarity);
|
44
|
+
Filter *filter = CScQ(weight->query)->filter;
|
45
|
+
|
46
|
+
CScSc(self)->score = weight->value;
|
47
|
+
CScSc(self)->bv = filt_get_bv(filter, ir);
|
48
|
+
|
49
|
+
self->score = &cssc_score;
|
50
|
+
self->next = &cssc_next;
|
51
|
+
self->skip_to = &cssc_skip_to;
|
52
|
+
self->explain = &cssc_explain;
|
53
|
+
self->destroy = &scorer_destroy_i;
|
54
|
+
return self;
|
48
55
|
}
|
49
56
|
|
50
57
|
/***************************************************************************
|
51
58
|
*
|
52
|
-
*
|
59
|
+
* ConstantScoreWeight
|
53
60
|
*
|
54
61
|
***************************************************************************/
|
55
62
|
|
56
|
-
char *
|
63
|
+
static char *csw_to_s(Weight *self)
|
57
64
|
{
|
58
|
-
|
59
|
-
char *filter_str = filter->to_s(filter);
|
60
|
-
char *buffer;
|
61
|
-
if (self->boost == 1.0) {
|
62
|
-
buffer = strfmt("ConstantScore(%s)", filter_str);
|
63
|
-
} else {
|
64
|
-
buffer = strfmt("ConstantScore(%s)^%f", filter_str, self->boost);
|
65
|
-
}
|
66
|
-
free(filter_str);
|
67
|
-
return buffer;;
|
65
|
+
return strfmt("ConstantScoreWeight(%f)", self->value);
|
68
66
|
}
|
69
67
|
|
70
|
-
|
68
|
+
static Explanation *csw_explain(Weight *self, IndexReader *ir, int doc_num)
|
71
69
|
{
|
72
|
-
|
73
|
-
|
74
|
-
filter->
|
75
|
-
|
76
|
-
|
70
|
+
Filter *filter = CScQ(self->query)->filter;
|
71
|
+
Explanation *expl;
|
72
|
+
char *filter_str = filter->to_s(filter);
|
73
|
+
BitVector *bv = filt_get_bv(filter, ir);
|
74
|
+
|
75
|
+
if (bv_get(bv, doc_num)) {
|
76
|
+
expl = expl_new(self->value,
|
77
|
+
"ConstantScoreQuery(%s), product of:", filter_str);
|
78
|
+
expl_add_detail(expl, expl_new(self->query->boost, "boost"));
|
79
|
+
expl_add_detail(expl, expl_new(self->qnorm, "query_norm"));
|
80
|
+
}
|
81
|
+
else {
|
82
|
+
expl = expl_new(self->value,
|
83
|
+
"ConstantScoreQuery(%s), does not match id %d",
|
84
|
+
filter_str, doc_num);
|
85
|
+
}
|
86
|
+
free(filter_str);
|
87
|
+
return expl;
|
77
88
|
}
|
78
89
|
|
79
|
-
static
|
90
|
+
static Weight *csw_new(Query *query, Searcher *searcher)
|
80
91
|
{
|
81
|
-
|
82
|
-
}
|
92
|
+
Weight *self = w_new(Weight, query);
|
83
93
|
|
84
|
-
|
85
|
-
|
86
|
-
|
87
|
-
}
|
88
|
-
|
89
|
-
Query *csq_create(Filter *filter)
|
90
|
-
{
|
91
|
-
Query *self = q_create();
|
92
|
-
self->data = filter;
|
94
|
+
self->scorer = &cssc_new;
|
95
|
+
self->explain = &csw_explain;
|
96
|
+
self->to_s = &csw_to_s;
|
93
97
|
|
94
|
-
|
95
|
-
|
96
|
-
self->hash = &csq_hash;
|
97
|
-
self->eq = &csq_eq;
|
98
|
-
self->destroy_i = &csq_destroy;
|
99
|
-
self->create_weight_i = &csw_create;
|
98
|
+
self->similarity = query->get_similarity(query, searcher);
|
99
|
+
self->idf = 1.0;
|
100
100
|
|
101
|
-
|
101
|
+
return self;
|
102
102
|
}
|
103
103
|
|
104
104
|
/***************************************************************************
|
105
105
|
*
|
106
|
-
*
|
106
|
+
* ConstantScoreQuery
|
107
107
|
*
|
108
108
|
***************************************************************************/
|
109
109
|
|
110
|
-
|
110
|
+
static char *csq_to_s(Query *self, const char *field)
|
111
111
|
{
|
112
|
-
|
112
|
+
Filter *filter = CScQ(self)->filter;
|
113
|
+
char *filter_str = filter->to_s(filter);
|
114
|
+
char *buffer;
|
115
|
+
(void)field;
|
116
|
+
if (self->boost == 1.0) {
|
117
|
+
buffer = strfmt("ConstantScore(%s)", filter_str);
|
118
|
+
}
|
119
|
+
else {
|
120
|
+
buffer = strfmt("ConstantScore(%s)^%f", filter_str, self->boost);
|
121
|
+
}
|
122
|
+
free(filter_str);
|
123
|
+
return buffer;;
|
113
124
|
}
|
114
125
|
|
115
|
-
|
126
|
+
static void csq_destroy(Query *self)
|
116
127
|
{
|
117
|
-
|
118
|
-
|
128
|
+
filt_deref(CScQ(self)->filter);
|
129
|
+
q_destroy_i(self);
|
119
130
|
}
|
120
131
|
|
121
|
-
|
132
|
+
static ulong csq_hash(Query *self)
|
122
133
|
{
|
123
|
-
|
124
|
-
return ((self->doc = bv_scan_next_from(bv, doc_num)) >= 0);
|
134
|
+
return filt_hash(CScQ(self)->filter);
|
125
135
|
}
|
126
136
|
|
127
|
-
|
137
|
+
static int csq_eq(Query *self, Query *o)
|
128
138
|
{
|
129
|
-
|
139
|
+
return filt_eq(CScQ(self)->filter, CScQ(o)->filter);
|
140
|
+
}
|
141
|
+
|
142
|
+
Query *csq_new_nr(Filter *filter)
|
143
|
+
{
|
144
|
+
Query *self = q_new(ConstantScoreQuery);
|
145
|
+
CScQ(self)->filter = filter;
|
146
|
+
|
147
|
+
self->type = CONSTANT_QUERY;
|
148
|
+
self->to_s = &csq_to_s;
|
149
|
+
self->hash = &csq_hash;
|
150
|
+
self->eq = &csq_eq;
|
151
|
+
self->destroy_i = &csq_destroy;
|
152
|
+
self->create_weight_i = &csw_new;
|
153
|
+
|
154
|
+
return self;
|
130
155
|
}
|
131
156
|
|
132
|
-
|
157
|
+
Query *csq_new(Filter *filter)
|
133
158
|
{
|
134
|
-
|
135
|
-
|
136
|
-
ConstantScoreScorer *cssc = ALLOC(ConstantScoreScorer);
|
137
|
-
ZEROSET(cssc, ConstantScoreScorer, 1);
|
138
|
-
self->data = cssc;
|
139
|
-
cssc->score = weight->value;
|
140
|
-
cssc->bv = filt_get_bv(filter, ir);
|
141
|
-
|
142
|
-
self->score = &cssc_score;
|
143
|
-
self->next = &cssc_next;
|
144
|
-
self->skip_to = &cssc_skip_to;
|
145
|
-
self->explain = &cssc_explain;
|
146
|
-
self->destroy = &scorer_destroy_i;
|
147
|
-
return self;
|
159
|
+
REF(filter);
|
160
|
+
return csq_new_nr(filter);
|
148
161
|
}
|
data/ext/q_filtered_query.c
CHANGED
@@ -1,206 +1,209 @@
|
|
1
1
|
#include "search.h"
|
2
2
|
#include <string.h>
|
3
3
|
|
4
|
-
Scorer *fqsc_create(Scorer *scorer, BitVector *bv, Similarity *sim);
|
5
|
-
|
6
4
|
/***************************************************************************
|
7
5
|
*
|
8
|
-
*
|
6
|
+
* FilteredQueryScorer
|
9
7
|
*
|
10
8
|
***************************************************************************/
|
11
9
|
|
12
|
-
|
13
|
-
|
14
|
-
return strfmt("FilteredQueryWeight(%f)", self->value);
|
15
|
-
}
|
10
|
+
#define FQSc(scorer) ((FilteredQueryScorer *)(scorer))
|
11
|
+
#define FQQ(query) ((FilteredQuery *)(query))
|
16
12
|
|
17
|
-
|
13
|
+
typedef struct FilteredQueryScorer
|
18
14
|
{
|
19
|
-
|
20
|
-
|
21
|
-
|
15
|
+
Scorer super;
|
16
|
+
Scorer *sub_scorer;
|
17
|
+
BitVector *bv;
|
18
|
+
} FilteredQueryScorer;
|
22
19
|
|
23
|
-
|
20
|
+
float fqsc_score(Scorer *self)
|
24
21
|
{
|
25
|
-
|
26
|
-
|
22
|
+
Scorer *sub_sc = FQSc(self)->sub_scorer;
|
23
|
+
return sub_sc->score(sub_sc);
|
27
24
|
}
|
28
25
|
|
29
|
-
|
26
|
+
bool fqsc_next(Scorer *self)
|
30
27
|
{
|
31
|
-
|
32
|
-
|
28
|
+
Scorer *sub_sc = FQSc(self)->sub_scorer;
|
29
|
+
BitVector *bv = FQSc(self)->bv;
|
30
|
+
while (sub_sc->next(sub_sc)) {
|
31
|
+
self->doc = sub_sc->doc;
|
32
|
+
if (bv_get(bv, self->doc)) return true;
|
33
|
+
}
|
34
|
+
return false;
|
33
35
|
}
|
34
36
|
|
35
|
-
|
37
|
+
bool fqsc_skip_to(Scorer *self, int doc_num)
|
36
38
|
{
|
37
|
-
|
38
|
-
|
39
|
+
Scorer *sub_sc = FQSc(self)->sub_scorer;
|
40
|
+
BitVector *bv = FQSc(self)->bv;
|
41
|
+
if (sub_sc->skip_to(sub_sc, doc_num)) {
|
42
|
+
self->doc = sub_sc->doc;
|
43
|
+
do {
|
44
|
+
if (bv_get(bv, self->doc)) return true;
|
45
|
+
} while (sub_sc->next(sub_sc));
|
46
|
+
}
|
47
|
+
return false;
|
39
48
|
}
|
40
49
|
|
41
|
-
|
50
|
+
Explanation *fqsc_explain(Scorer *self, int doc_num)
|
42
51
|
{
|
43
|
-
|
44
|
-
|
45
|
-
Filter *filter = ((FilteredQuery *)self->query->data)->filter;
|
46
|
-
|
47
|
-
return fqsc_create(scorer, filter->get_bv(filter, ir), self->similarity);
|
52
|
+
Scorer *sub_sc = FQSc(self)->sub_scorer;
|
53
|
+
return sub_sc->explain(sub_sc, doc_num);
|
48
54
|
}
|
49
55
|
|
50
|
-
void
|
56
|
+
void fqsc_destroy(Scorer *self)
|
51
57
|
{
|
52
|
-
|
53
|
-
|
54
|
-
|
58
|
+
FilteredQueryScorer *fqsc = FQSc(self);
|
59
|
+
fqsc->sub_scorer->destroy(fqsc->sub_scorer);
|
60
|
+
scorer_destroy_i(self);
|
55
61
|
}
|
56
62
|
|
57
|
-
|
63
|
+
Scorer *fqsc_new(Scorer *scorer, BitVector *bv, Similarity *sim)
|
58
64
|
{
|
59
|
-
|
60
|
-
|
61
|
-
self->data = sub_weight;
|
65
|
+
Scorer *self = scorer_new(FilteredQueryScorer, sim);
|
62
66
|
|
63
|
-
|
64
|
-
|
65
|
-
self->scorer = &fqw_scorer;
|
66
|
-
self->explain = &fqw_explain;
|
67
|
-
self->to_s = &fqw_to_s;
|
68
|
-
self->destroy = &fqw_destroy;
|
69
|
-
self->sum_of_squared_weights = &fqw_sum_of_squared_weights;
|
67
|
+
FQSc(self)->sub_scorer = scorer;
|
68
|
+
FQSc(self)->bv = bv;
|
70
69
|
|
71
|
-
|
72
|
-
|
73
|
-
|
70
|
+
self->score = &fqsc_score;
|
71
|
+
self->next = &fqsc_next;
|
72
|
+
self->skip_to = &fqsc_skip_to;
|
73
|
+
self->explain = &fqsc_explain;
|
74
|
+
self->destroy = &fqsc_destroy;
|
74
75
|
|
75
|
-
|
76
|
+
return self;
|
76
77
|
}
|
77
78
|
|
78
79
|
/***************************************************************************
|
79
80
|
*
|
80
|
-
*
|
81
|
+
* Weight
|
81
82
|
*
|
82
83
|
***************************************************************************/
|
83
84
|
|
84
|
-
|
85
|
+
#define FQW(weight) ((FilteredQueryWeight *)(weight))
|
86
|
+
typedef struct FilteredQueryWeight
|
85
87
|
{
|
86
|
-
|
87
|
-
|
88
|
-
|
89
|
-
|
90
|
-
|
91
|
-
|
92
|
-
|
93
|
-
} else {
|
94
|
-
buffer = strfmt("FilteredQuery(query:%s, filter:%s)^%f",
|
95
|
-
query_str, filter_str, self->boost);
|
96
|
-
}
|
97
|
-
free(filter_str);
|
98
|
-
free(query_str);
|
99
|
-
return buffer;;
|
88
|
+
Weight super;
|
89
|
+
Weight *sub_weight;
|
90
|
+
} FilteredQueryWeight;
|
91
|
+
|
92
|
+
static char *fqw_to_s(Weight *self)
|
93
|
+
{
|
94
|
+
return strfmt("FilteredQueryWeight(%f)", self->value);
|
100
95
|
}
|
101
96
|
|
102
|
-
|
97
|
+
static float fqw_sum_of_squared_weights(Weight *self)
|
103
98
|
{
|
104
|
-
|
105
|
-
|
106
|
-
fq->filter->destroy(fq->filter);
|
107
|
-
q_deref(fq->query);
|
108
|
-
}
|
109
|
-
free(self->data);
|
110
|
-
q_destroy_i(self);
|
99
|
+
Weight *sub_weight = FQW(self)->sub_weight;
|
100
|
+
return sub_weight->sum_of_squared_weights(sub_weight);
|
111
101
|
}
|
112
102
|
|
113
|
-
|
103
|
+
static void fqw_normalize(Weight *self, float normalization_factor)
|
114
104
|
{
|
115
|
-
|
116
|
-
|
117
|
-
searcher->similarity);
|
105
|
+
Weight *sub_weight = FQW(self)->sub_weight;
|
106
|
+
sub_weight->normalize(sub_weight, normalization_factor);
|
118
107
|
}
|
119
108
|
|
120
|
-
|
109
|
+
static float fqw_get_value(Weight *self)
|
121
110
|
{
|
122
|
-
|
111
|
+
Weight *sub_weight = FQW(self)->sub_weight;
|
112
|
+
return sub_weight->get_value(sub_weight);
|
113
|
+
}
|
123
114
|
|
124
|
-
|
125
|
-
|
126
|
-
|
127
|
-
|
115
|
+
static Explanation *fqw_explain(Weight *self, IndexReader *ir, int doc_num)
|
116
|
+
{
|
117
|
+
Weight *sub_weight = FQW(self)->sub_weight;
|
118
|
+
return sub_weight->explain(sub_weight, ir, doc_num);
|
119
|
+
}
|
128
120
|
|
129
|
-
|
130
|
-
|
131
|
-
|
132
|
-
|
121
|
+
static Scorer *fqw_scorer(Weight *self, IndexReader *ir)
|
122
|
+
{
|
123
|
+
Weight *sub_weight = FQW(self)->sub_weight;
|
124
|
+
Scorer *scorer = sub_weight->scorer(sub_weight, ir);
|
125
|
+
Filter *filter = FQQ(self->query)->filter;
|
133
126
|
|
134
|
-
|
127
|
+
return fqsc_new(scorer, filt_get_bv(filter, ir), self->similarity);
|
128
|
+
}
|
129
|
+
|
130
|
+
static void fqw_destroy(Weight *self)
|
131
|
+
{
|
132
|
+
Weight *sub_weight = FQW(self)->sub_weight;
|
133
|
+
sub_weight->destroy(sub_weight);
|
134
|
+
w_destroy(self);
|
135
|
+
}
|
136
|
+
|
137
|
+
static Weight *fqw_new(Query *query, Weight *sub_weight, Similarity *sim)
|
138
|
+
{
|
139
|
+
Weight *self = w_new(FilteredQueryWeight, query);
|
140
|
+
|
141
|
+
FQW(self)->sub_weight = sub_weight;
|
142
|
+
|
143
|
+
self->get_value = &fqw_get_value;
|
144
|
+
self->normalize = &fqw_normalize;
|
145
|
+
self->scorer = &fqw_scorer;
|
146
|
+
self->explain = &fqw_explain;
|
147
|
+
self->to_s = &fqw_to_s;
|
148
|
+
self->destroy = &fqw_destroy;
|
149
|
+
self->sum_of_squared_weights = &fqw_sum_of_squared_weights;
|
150
|
+
|
151
|
+
self->similarity = sim;
|
152
|
+
self->idf = 1.0;
|
153
|
+
self->value = sub_weight->value;
|
154
|
+
|
155
|
+
return self;
|
135
156
|
}
|
136
157
|
|
137
158
|
/***************************************************************************
|
138
159
|
*
|
139
|
-
*
|
160
|
+
* FilteredQuery
|
140
161
|
*
|
141
162
|
***************************************************************************/
|
142
163
|
|
143
|
-
|
144
|
-
Scorer *sub_scorer;
|
145
|
-
BitVector *bv;
|
146
|
-
} FilteredQueryScorer;
|
147
|
-
|
148
|
-
float fqsc_score(Scorer *self)
|
164
|
+
static char *fq_to_s(Query *self, const char *field)
|
149
165
|
{
|
150
|
-
|
151
|
-
|
166
|
+
FilteredQuery *fq = FQQ(self);
|
167
|
+
char *filter_str = fq->filter->to_s(fq->filter);
|
168
|
+
char *query_str = fq->query->to_s(fq->query, field);
|
169
|
+
char *buffer;
|
170
|
+
if (self->boost == 1.0) {
|
171
|
+
buffer = strfmt("FilteredQuery(query:%s, filter:%s)",
|
172
|
+
query_str, filter_str);
|
173
|
+
} else {
|
174
|
+
buffer = strfmt("FilteredQuery(query:%s, filter:%s)^%f",
|
175
|
+
query_str, filter_str, self->boost);
|
176
|
+
}
|
177
|
+
free(filter_str);
|
178
|
+
free(query_str);
|
179
|
+
return buffer;;
|
152
180
|
}
|
153
181
|
|
154
|
-
|
182
|
+
void fq_destroy(Query *self)
|
155
183
|
{
|
156
|
-
|
157
|
-
|
158
|
-
|
159
|
-
self->doc = sub_sc->doc;
|
160
|
-
if (bv_get(bv, self->doc)) return true;
|
161
|
-
}
|
162
|
-
return false;
|
184
|
+
filt_deref(FQQ(self)->filter);
|
185
|
+
q_deref(FQQ(self)->query);
|
186
|
+
q_destroy_i(self);
|
163
187
|
}
|
164
188
|
|
165
|
-
|
189
|
+
Weight *fq_new_weight(Query *self, Searcher *searcher)
|
166
190
|
{
|
167
|
-
|
168
|
-
|
169
|
-
|
170
|
-
self->doc = sub_sc->doc;
|
171
|
-
do {
|
172
|
-
if (bv_get(bv, self->doc)) return true;
|
173
|
-
} while (sub_sc->next(sub_sc));
|
174
|
-
}
|
175
|
-
return false;
|
191
|
+
Query *sub_query = FQQ(self)->query;
|
192
|
+
return fqw_new(self, q_weight(sub_query, searcher),
|
193
|
+
searcher->similarity);
|
176
194
|
}
|
177
195
|
|
178
|
-
|
196
|
+
Query *fq_new(Query *query, Filter *filter)
|
179
197
|
{
|
180
|
-
|
181
|
-
return sub_sc->explain(sub_sc, doc_num);
|
182
|
-
}
|
198
|
+
Query *self = q_new(FilteredQuery);
|
183
199
|
|
184
|
-
|
185
|
-
|
186
|
-
FilteredQueryScorer *fqsc = (FilteredQueryScorer *)self->data;
|
187
|
-
bv_destroy(fqsc->bv);
|
188
|
-
fqsc->sub_scorer->destroy(fqsc->sub_scorer);
|
189
|
-
scorer_destroy_i(self);
|
190
|
-
}
|
200
|
+
FQQ(self)->query = query;
|
201
|
+
FQQ(self)->filter = filter;
|
191
202
|
|
192
|
-
|
193
|
-
|
194
|
-
|
195
|
-
|
196
|
-
fqsc->sub_scorer = scorer;
|
197
|
-
fqsc->bv = bv;
|
198
|
-
self->data = fqsc;
|
203
|
+
self->type = FILTERED_QUERY;
|
204
|
+
self->to_s = &fq_to_s;
|
205
|
+
self->destroy_i = &fq_destroy;
|
206
|
+
self->create_weight_i = &fq_new_weight;
|
199
207
|
|
200
|
-
|
201
|
-
self->next = &fqsc_next;
|
202
|
-
self->skip_to = &fqsc_skip_to;
|
203
|
-
self->explain = &fqsc_explain;
|
204
|
-
self->destroy = &fqsc_destroy;
|
205
|
-
return self;
|
208
|
+
return self;
|
206
209
|
}
|