sdsykes-ferret 0.11.6.19
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGELOG +24 -0
- data/MIT-LICENSE +20 -0
- data/README +102 -0
- data/Rakefile +338 -0
- data/TODO +17 -0
- data/TUTORIAL +231 -0
- data/bin/ferret-browser +79 -0
- data/ext/analysis.c +1555 -0
- data/ext/analysis.h +219 -0
- data/ext/api.c +69 -0
- data/ext/api.h +27 -0
- data/ext/array.c +123 -0
- data/ext/array.h +53 -0
- data/ext/bitvector.c +540 -0
- data/ext/bitvector.h +272 -0
- data/ext/compound_io.c +383 -0
- data/ext/config.h +42 -0
- data/ext/document.c +156 -0
- data/ext/document.h +53 -0
- data/ext/except.c +120 -0
- data/ext/except.h +168 -0
- data/ext/extconf.rb +14 -0
- data/ext/ferret.c +402 -0
- data/ext/ferret.h +91 -0
- data/ext/filter.c +156 -0
- data/ext/fs_store.c +483 -0
- data/ext/global.c +418 -0
- data/ext/global.h +117 -0
- data/ext/hash.c +567 -0
- data/ext/hash.h +473 -0
- data/ext/hashset.c +170 -0
- data/ext/hashset.h +187 -0
- data/ext/header.h +58 -0
- data/ext/helper.c +62 -0
- data/ext/helper.h +13 -0
- data/ext/inc/lang.h +48 -0
- data/ext/inc/threading.h +31 -0
- data/ext/index.c +6425 -0
- data/ext/index.h +961 -0
- data/ext/lang.h +66 -0
- data/ext/libstemmer.c +92 -0
- data/ext/libstemmer.h +79 -0
- data/ext/mempool.c +87 -0
- data/ext/mempool.h +35 -0
- data/ext/modules.h +162 -0
- data/ext/multimapper.c +310 -0
- data/ext/multimapper.h +51 -0
- data/ext/posh.c +1006 -0
- data/ext/posh.h +1007 -0
- data/ext/priorityqueue.c +151 -0
- data/ext/priorityqueue.h +143 -0
- data/ext/q_boolean.c +1608 -0
- data/ext/q_const_score.c +161 -0
- data/ext/q_filtered_query.c +209 -0
- data/ext/q_fuzzy.c +268 -0
- data/ext/q_match_all.c +148 -0
- data/ext/q_multi_term.c +677 -0
- data/ext/q_parser.c +2825 -0
- data/ext/q_phrase.c +1126 -0
- data/ext/q_prefix.c +100 -0
- data/ext/q_range.c +350 -0
- data/ext/q_span.c +2402 -0
- data/ext/q_term.c +337 -0
- data/ext/q_wildcard.c +171 -0
- data/ext/r_analysis.c +2575 -0
- data/ext/r_index.c +3472 -0
- data/ext/r_qparser.c +585 -0
- data/ext/r_search.c +4105 -0
- data/ext/r_store.c +513 -0
- data/ext/r_utils.c +963 -0
- data/ext/ram_store.c +471 -0
- data/ext/search.c +1741 -0
- data/ext/search.h +885 -0
- data/ext/similarity.c +150 -0
- data/ext/similarity.h +82 -0
- data/ext/sort.c +983 -0
- data/ext/stem_ISO_8859_1_danish.c +338 -0
- data/ext/stem_ISO_8859_1_danish.h +16 -0
- data/ext/stem_ISO_8859_1_dutch.c +635 -0
- data/ext/stem_ISO_8859_1_dutch.h +16 -0
- data/ext/stem_ISO_8859_1_english.c +1156 -0
- data/ext/stem_ISO_8859_1_english.h +16 -0
- data/ext/stem_ISO_8859_1_finnish.c +792 -0
- data/ext/stem_ISO_8859_1_finnish.h +16 -0
- data/ext/stem_ISO_8859_1_french.c +1276 -0
- data/ext/stem_ISO_8859_1_french.h +16 -0
- data/ext/stem_ISO_8859_1_german.c +512 -0
- data/ext/stem_ISO_8859_1_german.h +16 -0
- data/ext/stem_ISO_8859_1_italian.c +1091 -0
- data/ext/stem_ISO_8859_1_italian.h +16 -0
- data/ext/stem_ISO_8859_1_norwegian.c +296 -0
- data/ext/stem_ISO_8859_1_norwegian.h +16 -0
- data/ext/stem_ISO_8859_1_porter.c +776 -0
- data/ext/stem_ISO_8859_1_porter.h +16 -0
- data/ext/stem_ISO_8859_1_portuguese.c +1035 -0
- data/ext/stem_ISO_8859_1_portuguese.h +16 -0
- data/ext/stem_ISO_8859_1_spanish.c +1119 -0
- data/ext/stem_ISO_8859_1_spanish.h +16 -0
- data/ext/stem_ISO_8859_1_swedish.c +307 -0
- data/ext/stem_ISO_8859_1_swedish.h +16 -0
- data/ext/stem_KOI8_R_russian.c +701 -0
- data/ext/stem_KOI8_R_russian.h +16 -0
- data/ext/stem_UTF_8_danish.c +344 -0
- data/ext/stem_UTF_8_danish.h +16 -0
- data/ext/stem_UTF_8_dutch.c +653 -0
- data/ext/stem_UTF_8_dutch.h +16 -0
- data/ext/stem_UTF_8_english.c +1176 -0
- data/ext/stem_UTF_8_english.h +16 -0
- data/ext/stem_UTF_8_finnish.c +808 -0
- data/ext/stem_UTF_8_finnish.h +16 -0
- data/ext/stem_UTF_8_french.c +1296 -0
- data/ext/stem_UTF_8_french.h +16 -0
- data/ext/stem_UTF_8_german.c +526 -0
- data/ext/stem_UTF_8_german.h +16 -0
- data/ext/stem_UTF_8_italian.c +1113 -0
- data/ext/stem_UTF_8_italian.h +16 -0
- data/ext/stem_UTF_8_norwegian.c +302 -0
- data/ext/stem_UTF_8_norwegian.h +16 -0
- data/ext/stem_UTF_8_porter.c +794 -0
- data/ext/stem_UTF_8_porter.h +16 -0
- data/ext/stem_UTF_8_portuguese.c +1055 -0
- data/ext/stem_UTF_8_portuguese.h +16 -0
- data/ext/stem_UTF_8_russian.c +709 -0
- data/ext/stem_UTF_8_russian.h +16 -0
- data/ext/stem_UTF_8_spanish.c +1137 -0
- data/ext/stem_UTF_8_spanish.h +16 -0
- data/ext/stem_UTF_8_swedish.c +313 -0
- data/ext/stem_UTF_8_swedish.h +16 -0
- data/ext/stopwords.c +401 -0
- data/ext/store.c +692 -0
- data/ext/store.h +777 -0
- data/ext/term_vectors.c +352 -0
- data/ext/threading.h +31 -0
- data/ext/utilities.c +446 -0
- data/ext/win32.h +54 -0
- data/lib/ferret.rb +29 -0
- data/lib/ferret/browser.rb +246 -0
- data/lib/ferret/browser/s/global.js +192 -0
- data/lib/ferret/browser/s/style.css +148 -0
- data/lib/ferret/browser/views/document/list.rhtml +49 -0
- data/lib/ferret/browser/views/document/show.rhtml +27 -0
- data/lib/ferret/browser/views/error/index.rhtml +7 -0
- data/lib/ferret/browser/views/help/index.rhtml +8 -0
- data/lib/ferret/browser/views/home/index.rhtml +29 -0
- data/lib/ferret/browser/views/layout.rhtml +22 -0
- data/lib/ferret/browser/views/term-vector/index.rhtml +4 -0
- data/lib/ferret/browser/views/term/index.rhtml +199 -0
- data/lib/ferret/browser/views/term/termdocs.rhtml +1 -0
- data/lib/ferret/browser/webrick.rb +14 -0
- data/lib/ferret/document.rb +130 -0
- data/lib/ferret/field_infos.rb +44 -0
- data/lib/ferret/index.rb +786 -0
- data/lib/ferret/number_tools.rb +157 -0
- data/lib/ferret_version.rb +3 -0
- data/setup.rb +1555 -0
- data/test/test_all.rb +5 -0
- data/test/test_helper.rb +24 -0
- data/test/threading/number_to_spoken.rb +132 -0
- data/test/threading/thread_safety_index_test.rb +79 -0
- data/test/threading/thread_safety_read_write_test.rb +76 -0
- data/test/threading/thread_safety_test.rb +133 -0
- data/test/unit/analysis/tc_analyzer.rb +548 -0
- data/test/unit/analysis/tc_token_stream.rb +646 -0
- data/test/unit/index/tc_index.rb +762 -0
- data/test/unit/index/tc_index_reader.rb +699 -0
- data/test/unit/index/tc_index_writer.rb +437 -0
- data/test/unit/index/th_doc.rb +315 -0
- data/test/unit/largefile/tc_largefile.rb +46 -0
- data/test/unit/query_parser/tc_query_parser.rb +238 -0
- data/test/unit/search/tc_filter.rb +135 -0
- data/test/unit/search/tc_fuzzy_query.rb +147 -0
- data/test/unit/search/tc_index_searcher.rb +61 -0
- data/test/unit/search/tc_multi_searcher.rb +128 -0
- data/test/unit/search/tc_multiple_search_requests.rb +58 -0
- data/test/unit/search/tc_search_and_sort.rb +179 -0
- data/test/unit/search/tc_sort.rb +49 -0
- data/test/unit/search/tc_sort_field.rb +27 -0
- data/test/unit/search/tc_spans.rb +190 -0
- data/test/unit/search/tm_searcher.rb +384 -0
- data/test/unit/store/tc_fs_store.rb +77 -0
- data/test/unit/store/tc_ram_store.rb +35 -0
- data/test/unit/store/tm_store.rb +34 -0
- data/test/unit/store/tm_store_lock.rb +68 -0
- data/test/unit/tc_document.rb +81 -0
- data/test/unit/ts_analysis.rb +2 -0
- data/test/unit/ts_index.rb +2 -0
- data/test/unit/ts_largefile.rb +4 -0
- data/test/unit/ts_query_parser.rb +2 -0
- data/test/unit/ts_search.rb +2 -0
- data/test/unit/ts_store.rb +2 -0
- data/test/unit/ts_utils.rb +2 -0
- data/test/unit/utils/tc_bit_vector.rb +295 -0
- data/test/unit/utils/tc_number_tools.rb +117 -0
- data/test/unit/utils/tc_priority_queue.rb +106 -0
- metadata +285 -0
data/ext/q_prefix.c
ADDED
@@ -0,0 +1,100 @@
|
|
1
|
+
#include <string.h>
|
2
|
+
#include "search.h"
|
3
|
+
|
4
|
+
/****************************************************************************
|
5
|
+
*
|
6
|
+
* PrefixQuery
|
7
|
+
*
|
8
|
+
****************************************************************************/
|
9
|
+
|
10
|
+
#define PfxQ(query) ((PrefixQuery *)(query))
|
11
|
+
|
12
|
+
static char *prq_to_s(Query *self, const char *current_field)
|
13
|
+
{
|
14
|
+
char *buffer, *bptr;
|
15
|
+
const char *prefix = PfxQ(self)->prefix;
|
16
|
+
const char *field = PfxQ(self)->field;
|
17
|
+
size_t plen = strlen(prefix);
|
18
|
+
size_t flen = strlen(field);
|
19
|
+
|
20
|
+
bptr = buffer = ALLOC_N(char, plen + flen + 35);
|
21
|
+
|
22
|
+
if (strcmp(field, current_field) != 0) {
|
23
|
+
sprintf(bptr, "%s:", field);
|
24
|
+
bptr += flen + 1;
|
25
|
+
}
|
26
|
+
|
27
|
+
sprintf(bptr, "%s*", prefix);
|
28
|
+
bptr += plen + 1;
|
29
|
+
if (self->boost != 1.0) {
|
30
|
+
*bptr = '^';
|
31
|
+
dbl_to_s(++bptr, self->boost);
|
32
|
+
}
|
33
|
+
|
34
|
+
return buffer;
|
35
|
+
}
|
36
|
+
|
37
|
+
static Query *prq_rewrite(Query *self, IndexReader *ir)
|
38
|
+
{
|
39
|
+
const char *field = PfxQ(self)->field;
|
40
|
+
const int field_num = fis_get_field_num(ir->fis, field);
|
41
|
+
Query *volatile q = multi_tq_new_conf(field, MTQMaxTerms(self), 0.0);
|
42
|
+
q->boost = self->boost; /* set the boost */
|
43
|
+
|
44
|
+
if (field_num >= 0) {
|
45
|
+
const char *prefix = PfxQ(self)->prefix;
|
46
|
+
TermEnum *te = ir->terms_from(ir, field_num, prefix);
|
47
|
+
const char *term = te->curr_term;
|
48
|
+
size_t prefix_len = strlen(prefix);
|
49
|
+
|
50
|
+
TRY
|
51
|
+
do {
|
52
|
+
if (strncmp(term, prefix, prefix_len) != 0) {
|
53
|
+
break;
|
54
|
+
}
|
55
|
+
multi_tq_add_term(q, term); /* found a match */
|
56
|
+
} while (te->next(te));
|
57
|
+
XFINALLY
|
58
|
+
te->close(te);
|
59
|
+
XENDTRY
|
60
|
+
}
|
61
|
+
|
62
|
+
return q;
|
63
|
+
}
|
64
|
+
|
65
|
+
static void prq_destroy(Query *self)
|
66
|
+
{
|
67
|
+
free(PfxQ(self)->field);
|
68
|
+
free(PfxQ(self)->prefix);
|
69
|
+
q_destroy_i(self);
|
70
|
+
}
|
71
|
+
|
72
|
+
static unsigned long prq_hash(Query *self)
|
73
|
+
{
|
74
|
+
return str_hash(PfxQ(self)->field) ^ str_hash(PfxQ(self)->prefix);
|
75
|
+
}
|
76
|
+
|
77
|
+
static int prq_eq(Query *self, Query *o)
|
78
|
+
{
|
79
|
+
return (strcmp(PfxQ(self)->prefix, PfxQ(o)->prefix) == 0)
|
80
|
+
&& (strcmp(PfxQ(self)->field, PfxQ(o)->field) == 0);
|
81
|
+
}
|
82
|
+
|
83
|
+
Query *prefixq_new(const char *field, const char *prefix)
|
84
|
+
{
|
85
|
+
Query *self = q_new(PrefixQuery);
|
86
|
+
|
87
|
+
PfxQ(self)->field = estrdup(field);
|
88
|
+
PfxQ(self)->prefix = estrdup(prefix);
|
89
|
+
MTQMaxTerms(self) = PREFIX_QUERY_MAX_TERMS;
|
90
|
+
|
91
|
+
self->type = PREFIX_QUERY;
|
92
|
+
self->rewrite = &prq_rewrite;
|
93
|
+
self->to_s = &prq_to_s;
|
94
|
+
self->hash = &prq_hash;
|
95
|
+
self->eq = &prq_eq;
|
96
|
+
self->destroy_i = &prq_destroy;
|
97
|
+
self->create_weight_i = &q_create_weight_unsup;
|
98
|
+
|
99
|
+
return self;
|
100
|
+
}
|
data/ext/q_range.c
ADDED
@@ -0,0 +1,350 @@
|
|
1
|
+
#include <string.h>
|
2
|
+
#include "search.h"
|
3
|
+
|
4
|
+
/*****************************************************************************
|
5
|
+
*
|
6
|
+
* Range
|
7
|
+
*
|
8
|
+
*****************************************************************************/
|
9
|
+
|
10
|
+
typedef struct Range
|
11
|
+
{
|
12
|
+
char *field;
|
13
|
+
char *lower_term;
|
14
|
+
char *upper_term;
|
15
|
+
bool include_lower : 1;
|
16
|
+
bool include_upper : 1;
|
17
|
+
} Range;
|
18
|
+
|
19
|
+
static char *range_to_s(Range *range, const char *field, float boost)
|
20
|
+
{
|
21
|
+
char *buffer, *b;
|
22
|
+
size_t flen, llen, ulen;
|
23
|
+
|
24
|
+
flen = strlen(range->field);
|
25
|
+
llen = range->lower_term ? strlen(range->lower_term) : 0;
|
26
|
+
ulen = range->upper_term ? strlen(range->upper_term) : 0;
|
27
|
+
buffer = ALLOC_N(char, flen + llen + ulen + 40);
|
28
|
+
b = buffer;
|
29
|
+
|
30
|
+
if (strcmp(field, range->field)) {
|
31
|
+
memcpy(buffer, range->field, flen * sizeof(char));
|
32
|
+
b += flen;
|
33
|
+
*b = ':';
|
34
|
+
b++;
|
35
|
+
}
|
36
|
+
|
37
|
+
if (range->lower_term) {
|
38
|
+
*b = range->include_lower ? '[' : '{';
|
39
|
+
b++;
|
40
|
+
memcpy(b, range->lower_term, llen);
|
41
|
+
b += llen;
|
42
|
+
} else {
|
43
|
+
*b = '<';
|
44
|
+
b++;
|
45
|
+
}
|
46
|
+
|
47
|
+
if (range->upper_term && range->lower_term) {
|
48
|
+
*b = ' '; b++;
|
49
|
+
}
|
50
|
+
|
51
|
+
if (range->upper_term) {
|
52
|
+
memcpy(b, range->upper_term, ulen);
|
53
|
+
b += ulen;
|
54
|
+
*b = range->include_upper ? ']' : '}';
|
55
|
+
b++;
|
56
|
+
} else {
|
57
|
+
*b = '>';
|
58
|
+
b++;
|
59
|
+
}
|
60
|
+
|
61
|
+
*b = 0;
|
62
|
+
if (boost != 1.0) {
|
63
|
+
*b = '^';
|
64
|
+
dbl_to_s(b + 1, boost);
|
65
|
+
}
|
66
|
+
return buffer;
|
67
|
+
}
|
68
|
+
|
69
|
+
static void range_destroy(Range *range)
|
70
|
+
{
|
71
|
+
free(range->field);
|
72
|
+
free(range->lower_term);
|
73
|
+
free(range->upper_term);
|
74
|
+
free(range);
|
75
|
+
}
|
76
|
+
|
77
|
+
static unsigned long range_hash(Range *filt)
|
78
|
+
{
|
79
|
+
return filt->include_lower | (filt->include_upper << 1)
|
80
|
+
| ((str_hash(filt->field)
|
81
|
+
^ (filt->lower_term ? str_hash(filt->lower_term) : 0)
|
82
|
+
^ (filt->upper_term ? str_hash(filt->upper_term) : 0)) << 2);
|
83
|
+
}
|
84
|
+
|
85
|
+
static int str_eq(char *s1, char *s2)
|
86
|
+
{
|
87
|
+
return (s1 && s2 && (strcmp(s1, s2) == 0)) || (s1 == s2);
|
88
|
+
}
|
89
|
+
|
90
|
+
static int range_eq(Range *filt, Range *o)
|
91
|
+
{
|
92
|
+
return (str_eq(filt->field, o->field)
|
93
|
+
&& str_eq(filt->lower_term, o->lower_term)
|
94
|
+
&& str_eq(filt->upper_term, o->upper_term)
|
95
|
+
&& (filt->include_lower == o->include_lower)
|
96
|
+
&& (filt->include_upper == o->include_upper));
|
97
|
+
}
|
98
|
+
|
99
|
+
Range *range_new(const char *field, const char *lower_term,
|
100
|
+
const char *upper_term, bool include_lower,
|
101
|
+
bool include_upper)
|
102
|
+
{
|
103
|
+
Range *range;
|
104
|
+
|
105
|
+
if (!lower_term && !upper_term) {
|
106
|
+
RAISE(ARG_ERROR, "Nil bounds for range. A range must include either "
|
107
|
+
"lower bound or an upper bound");
|
108
|
+
}
|
109
|
+
if (include_lower && !lower_term) {
|
110
|
+
RAISE(ARG_ERROR, "Lower bound must be non-nil to be inclusive. That "
|
111
|
+
"is, if you specify :include_lower => true when you create a "
|
112
|
+
"range you must include a :lower_term");
|
113
|
+
}
|
114
|
+
if (include_upper && !upper_term) {
|
115
|
+
RAISE(ARG_ERROR, "Upper bound must be non-nil to be inclusive. That "
|
116
|
+
"is, if you specify :include_upper => true when you create a "
|
117
|
+
"range you must include a :upper_term");
|
118
|
+
}
|
119
|
+
if (upper_term && lower_term && (strcmp(upper_term, lower_term) < 0)) {
|
120
|
+
RAISE(ARG_ERROR, "Upper bound must be greater than lower bound. "
|
121
|
+
"\"%s\" < \"%s\"", upper_term, lower_term);
|
122
|
+
}
|
123
|
+
|
124
|
+
range = ALLOC(Range);
|
125
|
+
|
126
|
+
range->field = estrdup((char *)field);
|
127
|
+
range->lower_term = lower_term ? estrdup(lower_term) : NULL;
|
128
|
+
range->upper_term = upper_term ? estrdup(upper_term) : NULL;
|
129
|
+
range->include_lower = include_lower;
|
130
|
+
range->include_upper = include_upper;
|
131
|
+
return range;
|
132
|
+
}
|
133
|
+
|
134
|
+
/***************************************************************************
|
135
|
+
*
|
136
|
+
* RangeFilter
|
137
|
+
*
|
138
|
+
***************************************************************************/
|
139
|
+
|
140
|
+
typedef struct RangeFilter
|
141
|
+
{
|
142
|
+
Filter super;
|
143
|
+
Range *range;
|
144
|
+
} RangeFilter;
|
145
|
+
|
146
|
+
#define RF(filt) ((RangeFilter *)(filt))
|
147
|
+
|
148
|
+
static void rfilt_destroy_i(Filter *filt)
|
149
|
+
{
|
150
|
+
range_destroy(RF(filt)->range);
|
151
|
+
filt_destroy_i(filt);
|
152
|
+
}
|
153
|
+
|
154
|
+
static char *rfilt_to_s(Filter *filt)
|
155
|
+
{
|
156
|
+
char *rstr = range_to_s(RF(filt)->range, "", 1.0);
|
157
|
+
char *rfstr = strfmt("RangeFilter< %s >", rstr);
|
158
|
+
free(rstr);
|
159
|
+
return rfstr;
|
160
|
+
}
|
161
|
+
|
162
|
+
static BitVector *rfilt_get_bv_i(Filter *filt, IndexReader *ir)
|
163
|
+
{
|
164
|
+
BitVector *bv = bv_new_capa(ir->max_doc(ir));
|
165
|
+
Range *range = RF(filt)->range;
|
166
|
+
FieldInfo *fi = fis_get_field(ir->fis, range->field);
|
167
|
+
/* the field info exists we need to add docs to the bit vector, otherwise
|
168
|
+
* we just return an empty bit vector */
|
169
|
+
if (fi) {
|
170
|
+
const char *lower_term =
|
171
|
+
range->lower_term ? range->lower_term : EMPTY_STRING;
|
172
|
+
const char *upper_term = range->upper_term;
|
173
|
+
const bool include_upper = range->include_upper;
|
174
|
+
const int field_num = fi->number;
|
175
|
+
char *term;
|
176
|
+
TermEnum* te;
|
177
|
+
TermDocEnum *tde;
|
178
|
+
bool check_lower;
|
179
|
+
|
180
|
+
te = ir->terms(ir, field_num);
|
181
|
+
if (te->skip_to(te, lower_term) == NULL) {
|
182
|
+
te->close(te);
|
183
|
+
return bv;
|
184
|
+
}
|
185
|
+
|
186
|
+
check_lower = !(range->include_lower || (lower_term == EMPTY_STRING));
|
187
|
+
|
188
|
+
tde = ir->term_docs(ir);
|
189
|
+
term = te->curr_term;
|
190
|
+
do {
|
191
|
+
if (!check_lower
|
192
|
+
|| (strcmp(term, lower_term) > 0)) {
|
193
|
+
check_lower = false;
|
194
|
+
if (upper_term) {
|
195
|
+
int compare = strcmp(upper_term, term);
|
196
|
+
/* Break if upper term is greater than or equal to upper
|
197
|
+
* term and include_upper is false or ther term is fully
|
198
|
+
* greater than upper term. This is optimized so that only
|
199
|
+
* one check is done except in last check or two */
|
200
|
+
if ((compare <= 0)
|
201
|
+
&& (!include_upper || (compare < 0))) {
|
202
|
+
break;
|
203
|
+
}
|
204
|
+
}
|
205
|
+
/* we have a good term, find the docs */
|
206
|
+
/* text is already pointing to term buffer text */
|
207
|
+
tde->seek_te(tde, te);
|
208
|
+
while (tde->next(tde)) {
|
209
|
+
bv_set(bv, tde->doc_num(tde));
|
210
|
+
/* printf("Setting %d\n", tde->doc_num(tde)); */
|
211
|
+
}
|
212
|
+
}
|
213
|
+
} while (te->next(te));
|
214
|
+
|
215
|
+
tde->close(tde);
|
216
|
+
te->close(te);
|
217
|
+
}
|
218
|
+
|
219
|
+
return bv;
|
220
|
+
}
|
221
|
+
|
222
|
+
static unsigned long rfilt_hash(Filter *filt)
|
223
|
+
{
|
224
|
+
return range_hash(RF(filt)->range);
|
225
|
+
}
|
226
|
+
|
227
|
+
static int rfilt_eq(Filter *filt, Filter *o)
|
228
|
+
{
|
229
|
+
return range_eq(RF(filt)->range, RF(o)->range);
|
230
|
+
}
|
231
|
+
|
232
|
+
Filter *rfilt_new(const char *field,
|
233
|
+
const char *lower_term, const char *upper_term,
|
234
|
+
bool include_lower, bool include_upper)
|
235
|
+
{
|
236
|
+
Filter *filt = filt_new(RangeFilter);
|
237
|
+
RF(filt)->range = range_new(field, lower_term, upper_term,
|
238
|
+
include_lower, include_upper);
|
239
|
+
|
240
|
+
filt->get_bv_i = &rfilt_get_bv_i;
|
241
|
+
filt->hash = &rfilt_hash;
|
242
|
+
filt->eq = &rfilt_eq;
|
243
|
+
filt->to_s = &rfilt_to_s;
|
244
|
+
filt->destroy_i = &rfilt_destroy_i;
|
245
|
+
return filt;
|
246
|
+
}
|
247
|
+
|
248
|
+
/*****************************************************************************
|
249
|
+
*
|
250
|
+
* RangeQuery
|
251
|
+
*
|
252
|
+
*****************************************************************************/
|
253
|
+
|
254
|
+
#define RQ(query) ((RangeQuery *)(query))
|
255
|
+
typedef struct RangeQuery
|
256
|
+
{
|
257
|
+
Query f;
|
258
|
+
Range *range;
|
259
|
+
} RangeQuery;
|
260
|
+
|
261
|
+
static char *rq_to_s(Query *self, const char *field)
|
262
|
+
{
|
263
|
+
return range_to_s(RQ(self)->range, field, self->boost);
|
264
|
+
}
|
265
|
+
|
266
|
+
static void rq_destroy(Query *self)
|
267
|
+
{
|
268
|
+
range_destroy(RQ(self)->range);
|
269
|
+
q_destroy_i(self);
|
270
|
+
}
|
271
|
+
|
272
|
+
static MatchVector *rq_get_matchv_i(Query *self, MatchVector *mv,
|
273
|
+
TermVector *tv)
|
274
|
+
{
|
275
|
+
Range *range = RQ(((ConstantScoreQuery *)self)->original)->range;
|
276
|
+
if (strcmp(tv->field, range->field) == 0) {
|
277
|
+
int i, j;
|
278
|
+
char *upper_text = range->upper_term;
|
279
|
+
char *lower_text = range->lower_term;
|
280
|
+
int upper_limit = range->include_upper ? 1 : 0;
|
281
|
+
int lower_limit = range->include_lower ? 1 : 0;
|
282
|
+
|
283
|
+
for (i = tv->term_cnt - 1; i >= 0; i--) {
|
284
|
+
TVTerm *tv_term = &(tv->terms[i]);
|
285
|
+
char *text = tv_term->text;
|
286
|
+
if ((!upper_text || strcmp(text, upper_text) < upper_limit) &&
|
287
|
+
(!lower_text || strcmp(lower_text, text) < lower_limit)) {
|
288
|
+
|
289
|
+
for (j = 0; j < tv_term->freq; j++) {
|
290
|
+
int pos = tv_term->positions[j];
|
291
|
+
matchv_add(mv, pos, pos);
|
292
|
+
}
|
293
|
+
}
|
294
|
+
}
|
295
|
+
}
|
296
|
+
return mv;
|
297
|
+
}
|
298
|
+
|
299
|
+
static Query *rq_rewrite(Query *self, IndexReader *ir)
|
300
|
+
{
|
301
|
+
Query *csq;
|
302
|
+
Range *r = RQ(self)->range;
|
303
|
+
Filter *filter = rfilt_new(r->field, r->lower_term, r->upper_term,
|
304
|
+
r->include_lower, r->include_upper);
|
305
|
+
(void)ir;
|
306
|
+
csq = csq_new_nr(filter);
|
307
|
+
((ConstantScoreQuery *)csq)->original = self;
|
308
|
+
csq->get_matchv_i = &rq_get_matchv_i;
|
309
|
+
return (Query *)csq;
|
310
|
+
}
|
311
|
+
|
312
|
+
static unsigned long rq_hash(Query *self)
|
313
|
+
{
|
314
|
+
return range_hash(RQ(self)->range);
|
315
|
+
}
|
316
|
+
|
317
|
+
static int rq_eq(Query *self, Query *o)
|
318
|
+
{
|
319
|
+
return range_eq(RQ(self)->range, RQ(o)->range);
|
320
|
+
}
|
321
|
+
|
322
|
+
Query *rq_new_less(const char *field, const char *upper_term,
|
323
|
+
bool include_upper)
|
324
|
+
{
|
325
|
+
return rq_new(field, NULL, upper_term, false, include_upper);
|
326
|
+
}
|
327
|
+
|
328
|
+
Query *rq_new_more(const char *field, const char *lower_term,
|
329
|
+
bool include_lower)
|
330
|
+
{
|
331
|
+
return rq_new(field, lower_term, NULL, include_lower, false);
|
332
|
+
}
|
333
|
+
|
334
|
+
Query *rq_new(const char *field, const char *lower_term,
|
335
|
+
const char *upper_term, bool include_lower, bool include_upper)
|
336
|
+
{
|
337
|
+
Query *self = q_new(RangeQuery);
|
338
|
+
|
339
|
+
RQ(self)->range = range_new(field, lower_term, upper_term,
|
340
|
+
include_lower, include_upper);
|
341
|
+
|
342
|
+
self->type = RANGE_QUERY;
|
343
|
+
self->rewrite = &rq_rewrite;
|
344
|
+
self->to_s = &rq_to_s;
|
345
|
+
self->hash = &rq_hash;
|
346
|
+
self->eq = &rq_eq;
|
347
|
+
self->destroy_i = &rq_destroy;
|
348
|
+
self->create_weight_i = &q_create_weight_unsup;
|
349
|
+
return self;
|
350
|
+
}
|
data/ext/q_span.c
ADDED
@@ -0,0 +1,2402 @@
|
|
1
|
+
#include <string.h>
|
2
|
+
#include <limits.h>
|
3
|
+
#include "search.h"
|
4
|
+
#include "hashset.h"
|
5
|
+
|
6
|
+
#define CLAUSE_INIT_CAPA 4
|
7
|
+
|
8
|
+
/*****************************************************************************
|
9
|
+
*
|
10
|
+
* SpanQuery
|
11
|
+
*
|
12
|
+
*****************************************************************************/
|
13
|
+
|
14
|
+
/***************************************************************************
|
15
|
+
* SpanQuery
|
16
|
+
***************************************************************************/
|
17
|
+
|
18
|
+
#define SpQ(query) ((SpanQuery *)(query))
|
19
|
+
|
20
|
+
static unsigned long spanq_hash(Query *self)
|
21
|
+
{
|
22
|
+
return str_hash(SpQ(self)->field);
|
23
|
+
}
|
24
|
+
|
25
|
+
static int spanq_eq(Query *self, Query *o)
|
26
|
+
{
|
27
|
+
return strcmp(SpQ(self)->field, SpQ(o)->field) == 0;
|
28
|
+
}
|
29
|
+
|
30
|
+
static void spanq_destroy_i(Query *self)
|
31
|
+
{
|
32
|
+
q_destroy_i(self);
|
33
|
+
}
|
34
|
+
|
35
|
+
static MatchVector *mv_to_term_mv(MatchVector *term_mv, MatchVector *full_mv,
|
36
|
+
HashSet *terms, TermVector *tv)
|
37
|
+
{
|
38
|
+
int i;
|
39
|
+
for (i = 0; i < terms->size; i++) {
|
40
|
+
char *term = (char *)terms->elems[i];
|
41
|
+
TVTerm *tv_term = tv_get_tv_term(tv, term);
|
42
|
+
if (tv_term) {
|
43
|
+
int j;
|
44
|
+
int m_idx = 0;
|
45
|
+
for (j = 0; j < tv_term->freq; j++) {
|
46
|
+
int pos = tv_term->positions[j];
|
47
|
+
for (; m_idx < full_mv->size; m_idx++) {
|
48
|
+
if (pos <= full_mv->matches[m_idx].end) {
|
49
|
+
if (pos >= full_mv->matches[m_idx].start) {
|
50
|
+
matchv_add(term_mv, pos, pos);
|
51
|
+
}
|
52
|
+
break;
|
53
|
+
}
|
54
|
+
}
|
55
|
+
}
|
56
|
+
}
|
57
|
+
}
|
58
|
+
|
59
|
+
return term_mv;
|
60
|
+
}
|
61
|
+
|
62
|
+
/***************************************************************************
|
63
|
+
* TVTermDocEnum
|
64
|
+
* dummy TermDocEnum used by the highlighter to find matches
|
65
|
+
***************************************************************************/
|
66
|
+
|
67
|
+
#define TV_TDE(tde) ((TVTermDocEnum *)(tde))
|
68
|
+
|
69
|
+
typedef struct TVTermDocEnum
|
70
|
+
{
|
71
|
+
TermDocEnum super;
|
72
|
+
int doc;
|
73
|
+
int index;
|
74
|
+
int freq;
|
75
|
+
int *positions;
|
76
|
+
TermVector *tv;
|
77
|
+
} TVTermDocEnum;
|
78
|
+
|
79
|
+
static void tv_tde_seek(TermDocEnum *tde, int field_num, const char *term)
|
80
|
+
{
|
81
|
+
TVTermDocEnum *tv_tde = TV_TDE(tde);
|
82
|
+
TVTerm *tv_term = tv_get_tv_term(tv_tde->tv, term);
|
83
|
+
(void)field_num;
|
84
|
+
if (tv_term) {
|
85
|
+
tv_tde->doc = -1;
|
86
|
+
tv_tde->index = 0;
|
87
|
+
tv_tde->freq = tv_term->freq;
|
88
|
+
tv_tde->positions = tv_term->positions;
|
89
|
+
}
|
90
|
+
else {
|
91
|
+
tv_tde->doc = INT_MAX;
|
92
|
+
}
|
93
|
+
}
|
94
|
+
|
95
|
+
static bool tv_tde_next(TermDocEnum *tde)
|
96
|
+
{
|
97
|
+
if (TV_TDE(tde)->doc == -1) {
|
98
|
+
TV_TDE(tde)->doc = 0;
|
99
|
+
return true;
|
100
|
+
}
|
101
|
+
else {
|
102
|
+
TV_TDE(tde)->doc = INT_MAX;
|
103
|
+
return false;
|
104
|
+
}
|
105
|
+
}
|
106
|
+
|
107
|
+
static bool tv_tde_skip_to(TermDocEnum *tde, int doc_num)
|
108
|
+
{
|
109
|
+
if (doc_num == 0) {
|
110
|
+
TV_TDE(tde)->doc = 0;
|
111
|
+
return true;
|
112
|
+
}
|
113
|
+
else {
|
114
|
+
TV_TDE(tde)->doc = INT_MAX;
|
115
|
+
return false;
|
116
|
+
}
|
117
|
+
}
|
118
|
+
|
119
|
+
static int tv_tde_next_position(TermDocEnum *tde)
|
120
|
+
{
|
121
|
+
return TV_TDE(tde)->positions[TV_TDE(tde)->index++];
|
122
|
+
}
|
123
|
+
|
124
|
+
static int tv_tde_freq(TermDocEnum *tde)
|
125
|
+
{
|
126
|
+
return TV_TDE(tde)->freq;
|
127
|
+
}
|
128
|
+
|
129
|
+
static int tv_tde_doc_num(TermDocEnum *tde)
|
130
|
+
{
|
131
|
+
return TV_TDE(tde)->doc;
|
132
|
+
}
|
133
|
+
|
134
|
+
static TermDocEnum *spanq_ir_term_positions(IndexReader *ir)
|
135
|
+
{
|
136
|
+
TVTermDocEnum *tv_tde = ALLOC(TVTermDocEnum);
|
137
|
+
TermDocEnum *tde = (TermDocEnum *)tv_tde;
|
138
|
+
tv_tde->tv = (TermVector *)ir->store;
|
139
|
+
tde->seek = &tv_tde_seek;
|
140
|
+
tde->doc_num = &tv_tde_doc_num;
|
141
|
+
tde->freq = &tv_tde_freq;
|
142
|
+
tde->next = &tv_tde_next;
|
143
|
+
tde->skip_to = &tv_tde_skip_to;
|
144
|
+
tde->next_position = &tv_tde_next_position;
|
145
|
+
tde->close = (void (*)(TermDocEnum *tde))&free;
|
146
|
+
|
147
|
+
return tde;
|
148
|
+
}
|
149
|
+
|
150
|
+
static MatchVector *spanq_get_matchv_i(Query *self, MatchVector *mv,
|
151
|
+
TermVector *tv)
|
152
|
+
{
|
153
|
+
if (strcmp(SpQ(self)->field, tv->field) == 0) {
|
154
|
+
SpanEnum *sp_enum;
|
155
|
+
IndexReader *ir = ALLOC(IndexReader);
|
156
|
+
MatchVector *full_mv = matchv_new();
|
157
|
+
HashSet *terms = SpQ(self)->get_terms(self);
|
158
|
+
ir->fis = fis_new(0, 0, 0);
|
159
|
+
fis_add_field(ir->fis, fi_new(tv->field, 0, 0, 0));
|
160
|
+
ir->store = (Store *)tv;
|
161
|
+
ir->term_positions = &spanq_ir_term_positions;
|
162
|
+
sp_enum = SpQ(self)->get_spans(self, ir);
|
163
|
+
while (sp_enum->next(sp_enum)) {
|
164
|
+
matchv_add(full_mv,
|
165
|
+
sp_enum->start(sp_enum),
|
166
|
+
sp_enum->end(sp_enum) - 1);
|
167
|
+
}
|
168
|
+
sp_enum->destroy(sp_enum);
|
169
|
+
|
170
|
+
fis_deref(ir->fis);
|
171
|
+
free(ir);
|
172
|
+
|
173
|
+
matchv_compact(full_mv);
|
174
|
+
mv_to_term_mv(mv, full_mv, terms, tv);
|
175
|
+
matchv_destroy(full_mv);
|
176
|
+
hs_destroy(terms);
|
177
|
+
}
|
178
|
+
return mv;
|
179
|
+
}
|
180
|
+
|
181
|
+
/***************************************************************************
|
182
|
+
*
|
183
|
+
* SpanScorer
|
184
|
+
*
|
185
|
+
***************************************************************************/
|
186
|
+
|
187
|
+
#define SpSc(scorer) ((SpanScorer *)(scorer))
|
188
|
+
typedef struct SpanScorer
|
189
|
+
{
|
190
|
+
Scorer super;
|
191
|
+
IndexReader *ir;
|
192
|
+
SpanEnum *spans;
|
193
|
+
Similarity *sim;
|
194
|
+
uchar *norms;
|
195
|
+
Weight *weight;
|
196
|
+
float value;
|
197
|
+
float freq;
|
198
|
+
bool first_time : 1;
|
199
|
+
bool more : 1;
|
200
|
+
} SpanScorer;
|
201
|
+
|
202
|
+
static float spansc_score(Scorer *self)
|
203
|
+
{
|
204
|
+
SpanScorer *spansc = SpSc(self);
|
205
|
+
float raw = sim_tf(spansc->sim, spansc->freq) * spansc->value;
|
206
|
+
|
207
|
+
/* normalize */
|
208
|
+
return raw * sim_decode_norm(self->similarity, spansc->norms[self->doc]);
|
209
|
+
}
|
210
|
+
|
211
|
+
static bool spansc_next(Scorer *self)
|
212
|
+
{
|
213
|
+
SpanScorer *spansc = SpSc(self);
|
214
|
+
SpanEnum *se = spansc->spans;
|
215
|
+
int match_length;
|
216
|
+
|
217
|
+
if (spansc->first_time) {
|
218
|
+
spansc->more = se->next(se);
|
219
|
+
spansc->first_time = false;
|
220
|
+
}
|
221
|
+
|
222
|
+
if (!spansc->more) {
|
223
|
+
return false;
|
224
|
+
}
|
225
|
+
|
226
|
+
spansc->freq = 0.0;
|
227
|
+
self->doc = se->doc(se);
|
228
|
+
|
229
|
+
while (spansc->more && (self->doc == se->doc(se))) {
|
230
|
+
match_length = se->end(se) - se->start(se);
|
231
|
+
spansc->freq += sim_sloppy_freq(spansc->sim, match_length);
|
232
|
+
spansc->more = se->next(se);
|
233
|
+
}
|
234
|
+
|
235
|
+
return (spansc->more || (spansc->freq != 0.0));
|
236
|
+
}
|
237
|
+
|
238
|
+
static bool spansc_skip_to(Scorer *self, int target)
|
239
|
+
{
|
240
|
+
SpanScorer *spansc = SpSc(self);
|
241
|
+
SpanEnum *se = spansc->spans;
|
242
|
+
|
243
|
+
spansc->more = se->skip_to(se, target);
|
244
|
+
|
245
|
+
if (!spansc->more) {
|
246
|
+
return false;
|
247
|
+
}
|
248
|
+
|
249
|
+
spansc->freq = 0.0;
|
250
|
+
self->doc = se->doc(se);
|
251
|
+
|
252
|
+
while (spansc->more && (se->doc(se) == target)) {
|
253
|
+
spansc->freq += sim_sloppy_freq(spansc->sim, se->end(se) - se->start(se));
|
254
|
+
spansc->more = se->next(se);
|
255
|
+
}
|
256
|
+
|
257
|
+
return (spansc->more || (spansc->freq != 0.0));
|
258
|
+
}
|
259
|
+
|
260
|
+
static Explanation *spansc_explain(Scorer *self, int target)
|
261
|
+
{
|
262
|
+
Explanation *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 = expl_new(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(Scorer *self)
|
275
|
+
{
|
276
|
+
SpanScorer *spansc = SpSc(self);
|
277
|
+
if (spansc->spans) {
|
278
|
+
spansc->spans->destroy(spansc->spans);
|
279
|
+
}
|
280
|
+
scorer_destroy_i(self);
|
281
|
+
}
|
282
|
+
|
283
|
+
Scorer *spansc_new(Weight *weight, IndexReader *ir)
|
284
|
+
{
|
285
|
+
Scorer *self = NULL;
|
286
|
+
const int field_num = fis_get_field_num(ir->fis, SpQ(weight->query)->field);
|
287
|
+
if (field_num >= 0) {
|
288
|
+
Query *spanq = weight->query;
|
289
|
+
self = 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 = 0.0;
|
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) ((SpanTermQuery *)(query))
|
315
|
+
|
316
|
+
typedef struct SpanTermEnum
|
317
|
+
{
|
318
|
+
SpanEnum super;
|
319
|
+
TermDocEnum *positions;
|
320
|
+
int position;
|
321
|
+
int doc;
|
322
|
+
int count;
|
323
|
+
int freq;
|
324
|
+
} SpanTermEnum;
|
325
|
+
|
326
|
+
|
327
|
+
static bool spante_next(SpanEnum *self)
|
328
|
+
{
|
329
|
+
SpanTermEnum *ste = SpTEn(self);
|
330
|
+
TermDocEnum *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(SpanEnum *self, int target)
|
347
|
+
{
|
348
|
+
SpanTermEnum *ste = SpTEn(self);
|
349
|
+
TermDocEnum *tde = ste->positions;
|
350
|
+
|
351
|
+
/* are we already at the correct position? */
|
352
|
+
if (ste->doc >= target) {
|
353
|
+
return true;
|
354
|
+
}
|
355
|
+
|
356
|
+
if (! tde->skip_to(tde, target)) {
|
357
|
+
ste->doc = INT_MAX;
|
358
|
+
return false;
|
359
|
+
}
|
360
|
+
|
361
|
+
ste->doc = tde->doc_num(tde);
|
362
|
+
ste->freq = tde->freq(tde);
|
363
|
+
ste->count = 0;
|
364
|
+
|
365
|
+
ste->position = tde->next_position(tde);
|
366
|
+
ste->count++;
|
367
|
+
return true;
|
368
|
+
}
|
369
|
+
|
370
|
+
static int spante_doc(SpanEnum *self)
|
371
|
+
{
|
372
|
+
return SpTEn(self)->doc;
|
373
|
+
}
|
374
|
+
|
375
|
+
static int spante_start(SpanEnum *self)
|
376
|
+
{
|
377
|
+
return SpTEn(self)->position;
|
378
|
+
}
|
379
|
+
|
380
|
+
static int spante_end(SpanEnum *self)
|
381
|
+
{
|
382
|
+
return SpTEn(self)->position + 1;
|
383
|
+
}
|
384
|
+
|
385
|
+
static char *spante_to_s(SpanEnum *self)
|
386
|
+
{
|
387
|
+
char *field = SpQ(self->query)->field;
|
388
|
+
char *query_str = self->query->to_s(self->query, field);
|
389
|
+
char pos_str[20];
|
390
|
+
size_t len = strlen(query_str);
|
391
|
+
int pos;
|
392
|
+
char *str = ALLOC_N(char, len + 40);
|
393
|
+
|
394
|
+
if (self->doc(self) < 0) {
|
395
|
+
sprintf(pos_str, "START");
|
396
|
+
}
|
397
|
+
else {
|
398
|
+
if (self->doc(self) == INT_MAX) {
|
399
|
+
sprintf(pos_str, "END");
|
400
|
+
}
|
401
|
+
else {
|
402
|
+
pos = SpTEn(self)->position;
|
403
|
+
sprintf(pos_str, "%d", self->doc(self) - pos);
|
404
|
+
}
|
405
|
+
}
|
406
|
+
sprintf("SpanTermEnum(%s)@%s", query_str, pos_str);
|
407
|
+
free(query_str);
|
408
|
+
return str;
|
409
|
+
}
|
410
|
+
|
411
|
+
static void spante_destroy(SpanEnum *self)
|
412
|
+
{
|
413
|
+
TermDocEnum *tde = SpTEn(self)->positions;
|
414
|
+
tde->close(tde);
|
415
|
+
free(self);
|
416
|
+
}
|
417
|
+
|
418
|
+
static SpanEnum *spante_new(Query *query, IndexReader *ir)
|
419
|
+
{
|
420
|
+
char *term = SpTQ(query)->term;
|
421
|
+
char *field = SpQ(query)->field;
|
422
|
+
SpanEnum *self = (SpanEnum *)emalloc(sizeof(SpanTermEnum));
|
423
|
+
|
424
|
+
SpTEn(self)->positions = ir_term_positions_for(ir, field, term);
|
425
|
+
SpTEn(self)->position = -1;
|
426
|
+
SpTEn(self)->doc = -1;
|
427
|
+
SpTEn(self)->count = 0;
|
428
|
+
SpTEn(self)->freq = 0;
|
429
|
+
|
430
|
+
self->query = query;
|
431
|
+
self->next = &spante_next;
|
432
|
+
self->skip_to = &spante_skip_to;
|
433
|
+
self->doc = &spante_doc;
|
434
|
+
self->start = &spante_start;
|
435
|
+
self->end = &spante_end;
|
436
|
+
self->destroy = &spante_destroy;
|
437
|
+
self->to_s = &spante_to_s;
|
438
|
+
|
439
|
+
return self;
|
440
|
+
}
|
441
|
+
|
442
|
+
/*****************************************************************************
|
443
|
+
* SpanMultiTermEnum
|
444
|
+
*****************************************************************************/
|
445
|
+
|
446
|
+
/* * TermPosEnumWrapper * */
|
447
|
+
#define TPE_READ_SIZE 16
|
448
|
+
|
449
|
+
typedef struct TermPosEnumWrapper
|
450
|
+
{
|
451
|
+
const char *term;
|
452
|
+
TermDocEnum *tpe;
|
453
|
+
int doc;
|
454
|
+
int pos;
|
455
|
+
} TermPosEnumWrapper;
|
456
|
+
|
457
|
+
static bool tpew_less_than(const TermPosEnumWrapper *tpew1,
|
458
|
+
const TermPosEnumWrapper *tpew2)
|
459
|
+
{
|
460
|
+
return (tpew1->doc < tpew2->doc)
|
461
|
+
|| (tpew1->doc == tpew2->doc && tpew1->pos < tpew2->pos);
|
462
|
+
}
|
463
|
+
|
464
|
+
static bool tpew_next(TermPosEnumWrapper *self)
|
465
|
+
{
|
466
|
+
TermDocEnum *tpe = self->tpe;
|
467
|
+
if (0 > (self->pos = tpe->next_position(tpe))) {
|
468
|
+
if (!tpe->next(tpe)) return false;
|
469
|
+
self->doc = tpe->doc_num(tpe);
|
470
|
+
self->pos = tpe->next_position(tpe);
|
471
|
+
}
|
472
|
+
return true;
|
473
|
+
}
|
474
|
+
|
475
|
+
static bool tpew_skip_to(TermPosEnumWrapper *self, int doc_num)
|
476
|
+
{
|
477
|
+
TermDocEnum *tpe = self->tpe;
|
478
|
+
|
479
|
+
if (tpe->skip_to(tpe, doc_num)) {
|
480
|
+
self->doc = tpe->doc_num(tpe);
|
481
|
+
self->pos = tpe->next_position(tpe);
|
482
|
+
return true;
|
483
|
+
}
|
484
|
+
else {
|
485
|
+
return false;
|
486
|
+
}
|
487
|
+
}
|
488
|
+
|
489
|
+
static void tpew_destroy(TermPosEnumWrapper *self)
|
490
|
+
{
|
491
|
+
self->tpe->close(self->tpe);
|
492
|
+
free(self);
|
493
|
+
}
|
494
|
+
|
495
|
+
static TermPosEnumWrapper *tpew_new(const char *term, TermDocEnum *tpe)
|
496
|
+
{
|
497
|
+
TermPosEnumWrapper *self = ALLOC_AND_ZERO(TermPosEnumWrapper);
|
498
|
+
self->term = term;
|
499
|
+
self->tpe = tpe;
|
500
|
+
self->doc = -1;
|
501
|
+
self->pos = -1;
|
502
|
+
return self;
|
503
|
+
}
|
504
|
+
#define SpMTEn(span_enum) ((SpanMultiTermEnum *)(span_enum))
|
505
|
+
#define SpMTQ(query) ((SpanMultiTermQuery *)(query))
|
506
|
+
|
507
|
+
typedef struct SpanMultiTermEnum
|
508
|
+
{
|
509
|
+
SpanEnum super;
|
510
|
+
PriorityQueue *tpew_pq;
|
511
|
+
TermPosEnumWrapper **tpews;
|
512
|
+
int tpew_cnt;
|
513
|
+
int pos;
|
514
|
+
int doc;
|
515
|
+
} SpanMultiTermEnum;
|
516
|
+
|
517
|
+
static bool spanmte_next(SpanEnum *self)
|
518
|
+
{
|
519
|
+
int curr_doc, curr_pos;
|
520
|
+
TermPosEnumWrapper *tpew;
|
521
|
+
SpanMultiTermEnum *mte = SpMTEn(self);
|
522
|
+
PriorityQueue *tpew_pq = mte->tpew_pq;
|
523
|
+
if (tpew_pq == NULL) {
|
524
|
+
TermPosEnumWrapper **tpews = mte->tpews;
|
525
|
+
int i;
|
526
|
+
tpew_pq = pq_new(mte->tpew_cnt, (lt_ft)tpew_less_than, (free_ft)NULL);
|
527
|
+
for (i = mte->tpew_cnt - 1; i >= 0; i--) {
|
528
|
+
if (tpew_next(tpews[i])) {
|
529
|
+
pq_push(tpew_pq, tpews[i]);
|
530
|
+
}
|
531
|
+
}
|
532
|
+
mte->tpew_pq = tpew_pq;
|
533
|
+
}
|
534
|
+
|
535
|
+
tpew = (TermPosEnumWrapper *)pq_top(tpew_pq);
|
536
|
+
if (tpew == NULL) {
|
537
|
+
return false;
|
538
|
+
}
|
539
|
+
|
540
|
+
mte->doc = curr_doc = tpew->doc;
|
541
|
+
mte->pos = curr_pos = tpew->pos;
|
542
|
+
|
543
|
+
do {
|
544
|
+
if (tpew_next(tpew)) {
|
545
|
+
pq_down(tpew_pq);
|
546
|
+
}
|
547
|
+
else {
|
548
|
+
pq_pop(tpew_pq);
|
549
|
+
}
|
550
|
+
} while (((tpew = (TermPosEnumWrapper *)pq_top(tpew_pq)) != NULL)
|
551
|
+
&& tpew->doc == curr_doc && tpew->pos == curr_pos);
|
552
|
+
return true;
|
553
|
+
}
|
554
|
+
|
555
|
+
static bool spanmte_skip_to(SpanEnum *self, int target)
|
556
|
+
{
|
557
|
+
SpanMultiTermEnum *mte = SpMTEn(self);
|
558
|
+
PriorityQueue *tpew_pq = mte->tpew_pq;
|
559
|
+
TermPosEnumWrapper *tpew;
|
560
|
+
if (tpew_pq == NULL) {
|
561
|
+
TermPosEnumWrapper **tpews = mte->tpews;
|
562
|
+
int i;
|
563
|
+
tpew_pq = pq_new(mte->tpew_cnt, (lt_ft)tpew_less_than, (free_ft)NULL);
|
564
|
+
for (i = mte->tpew_cnt - 1; i >= 0; i--) {
|
565
|
+
tpew_skip_to(tpews[i], target);
|
566
|
+
pq_push(tpew_pq, tpews[i]);
|
567
|
+
}
|
568
|
+
mte->tpew_pq = tpew_pq;
|
569
|
+
}
|
570
|
+
if (tpew_pq->size == 0) {
|
571
|
+
mte->doc = -1;
|
572
|
+
return false;
|
573
|
+
}
|
574
|
+
while ((tpew = (TermPosEnumWrapper *)pq_top(tpew_pq)) != NULL
|
575
|
+
&& (target > tpew->doc)) {
|
576
|
+
if (tpew_skip_to(tpew, target)) {
|
577
|
+
pq_down(tpew_pq);
|
578
|
+
}
|
579
|
+
else {
|
580
|
+
pq_pop(tpew_pq);
|
581
|
+
}
|
582
|
+
}
|
583
|
+
return spanmte_next(self);
|
584
|
+
}
|
585
|
+
|
586
|
+
static int spanmte_doc(SpanEnum *self)
|
587
|
+
{
|
588
|
+
return SpMTEn(self)->doc;
|
589
|
+
}
|
590
|
+
|
591
|
+
static int spanmte_start(SpanEnum *self)
|
592
|
+
{
|
593
|
+
return SpMTEn(self)->pos;
|
594
|
+
}
|
595
|
+
|
596
|
+
static int spanmte_end(SpanEnum *self)
|
597
|
+
{
|
598
|
+
return SpMTEn(self)->pos + 1;
|
599
|
+
}
|
600
|
+
|
601
|
+
static void spanmte_destroy(SpanEnum *self)
|
602
|
+
{
|
603
|
+
SpanMultiTermEnum *mte = SpMTEn(self);
|
604
|
+
int i;
|
605
|
+
if (mte->tpew_pq) pq_destroy(mte->tpew_pq);
|
606
|
+
for (i = 0; i < mte->tpew_cnt; i++) {
|
607
|
+
tpew_destroy(mte->tpews[i]);
|
608
|
+
}
|
609
|
+
free(mte->tpews);
|
610
|
+
free(self);
|
611
|
+
}
|
612
|
+
|
613
|
+
static SpanEnum *spanmte_new(Query *query, IndexReader *ir)
|
614
|
+
{
|
615
|
+
char *field = SpQ(query)->field;
|
616
|
+
SpanEnum *self = (SpanEnum *)emalloc(sizeof(SpanMultiTermEnum));
|
617
|
+
SpanMultiTermEnum *smte = SpMTEn(self);
|
618
|
+
SpanMultiTermQuery *smtq = SpMTQ(query);
|
619
|
+
int i;
|
620
|
+
|
621
|
+
|
622
|
+
smte->tpews = ALLOC_N(TermPosEnumWrapper *, smtq->term_cnt);
|
623
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
624
|
+
char *term = smtq->terms[i];
|
625
|
+
smte->tpews[i] = tpew_new(term, ir_term_positions_for(ir, field, term));
|
626
|
+
}
|
627
|
+
smte->tpew_cnt = smtq->term_cnt;
|
628
|
+
smte->tpew_pq = NULL;
|
629
|
+
smte->pos = -1;
|
630
|
+
smte->doc = -1;
|
631
|
+
|
632
|
+
self->query = query;
|
633
|
+
self->next = &spanmte_next;
|
634
|
+
self->skip_to = &spanmte_skip_to;
|
635
|
+
self->doc = &spanmte_doc;
|
636
|
+
self->start = &spanmte_start;
|
637
|
+
self->end = &spanmte_end;
|
638
|
+
self->destroy = &spanmte_destroy;
|
639
|
+
self->to_s = &spante_to_s;
|
640
|
+
|
641
|
+
return self;
|
642
|
+
}
|
643
|
+
|
644
|
+
|
645
|
+
/*****************************************************************************
|
646
|
+
* SpanFirstEnum
|
647
|
+
*****************************************************************************/
|
648
|
+
|
649
|
+
#define SpFEn(span_enum) ((SpanFirstEnum *)(span_enum))
|
650
|
+
#define SpFQ(query) ((SpanFirstQuery *)(query))
|
651
|
+
|
652
|
+
typedef struct SpanFirstEnum
|
653
|
+
{
|
654
|
+
SpanEnum super;
|
655
|
+
SpanEnum *sub_enum;
|
656
|
+
} SpanFirstEnum;
|
657
|
+
|
658
|
+
|
659
|
+
static bool spanfe_next(SpanEnum *self)
|
660
|
+
{
|
661
|
+
SpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
662
|
+
int end = SpFQ(self->query)->end;
|
663
|
+
while (sub_enum->next(sub_enum)) { /* scan to next match */
|
664
|
+
if (sub_enum->end(sub_enum) <= end) {
|
665
|
+
return true;
|
666
|
+
}
|
667
|
+
}
|
668
|
+
return false;
|
669
|
+
}
|
670
|
+
|
671
|
+
static bool spanfe_skip_to(SpanEnum *self, int target)
|
672
|
+
{
|
673
|
+
SpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
674
|
+
int end = SpFQ(self->query)->end;
|
675
|
+
|
676
|
+
if (! sub_enum->skip_to(sub_enum, target)) {
|
677
|
+
return false;
|
678
|
+
}
|
679
|
+
|
680
|
+
if (sub_enum->end(sub_enum) <= end) { /* there is a match */
|
681
|
+
return true;
|
682
|
+
}
|
683
|
+
|
684
|
+
return sub_enum->next(sub_enum); /* scan to next match */
|
685
|
+
}
|
686
|
+
|
687
|
+
static int spanfe_doc(SpanEnum *self)
|
688
|
+
{
|
689
|
+
SpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
690
|
+
return sub_enum->doc(sub_enum);
|
691
|
+
}
|
692
|
+
|
693
|
+
static int spanfe_start(SpanEnum *self)
|
694
|
+
{
|
695
|
+
SpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
696
|
+
return sub_enum->start(sub_enum);
|
697
|
+
}
|
698
|
+
|
699
|
+
static int spanfe_end(SpanEnum *self)
|
700
|
+
{
|
701
|
+
SpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
702
|
+
return sub_enum->end(sub_enum);
|
703
|
+
}
|
704
|
+
|
705
|
+
static char *spanfe_to_s(SpanEnum *self)
|
706
|
+
{
|
707
|
+
char *field = SpQ(self->query)->field;
|
708
|
+
char *query_str = self->query->to_s(self->query, field);
|
709
|
+
char *res = strfmt("SpanFirstEnum(%s)", query_str);
|
710
|
+
free(query_str);
|
711
|
+
return res;
|
712
|
+
}
|
713
|
+
|
714
|
+
static void spanfe_destroy(SpanEnum *self)
|
715
|
+
{
|
716
|
+
SpanEnum *sub_enum = SpFEn(self)->sub_enum;
|
717
|
+
sub_enum->destroy(sub_enum);
|
718
|
+
free(self);
|
719
|
+
}
|
720
|
+
|
721
|
+
static SpanEnum *spanfe_new(Query *query, IndexReader *ir)
|
722
|
+
{
|
723
|
+
SpanEnum *self = (SpanEnum *)emalloc(sizeof(SpanFirstEnum));
|
724
|
+
SpanFirstQuery *sfq = SpFQ(query);
|
725
|
+
|
726
|
+
SpFEn(self)->sub_enum = SpQ(sfq->match)->get_spans(sfq->match, ir);
|
727
|
+
|
728
|
+
self->query = query;
|
729
|
+
self->next = &spanfe_next;
|
730
|
+
self->skip_to = &spanfe_skip_to;
|
731
|
+
self->doc = &spanfe_doc;
|
732
|
+
self->start = &spanfe_start;
|
733
|
+
self->end = &spanfe_end;
|
734
|
+
self->destroy = &spanfe_destroy;
|
735
|
+
self->to_s = &spanfe_to_s;
|
736
|
+
|
737
|
+
return self;
|
738
|
+
}
|
739
|
+
|
740
|
+
|
741
|
+
/*****************************************************************************
|
742
|
+
* SpanOrEnum
|
743
|
+
*****************************************************************************/
|
744
|
+
|
745
|
+
#define SpOEn(span_enum) ((SpanOrEnum *)(span_enum))
|
746
|
+
#define SpOQ(query) ((SpanOrQuery *)(query))
|
747
|
+
|
748
|
+
typedef struct SpanOrEnum
|
749
|
+
{
|
750
|
+
SpanEnum super;
|
751
|
+
PriorityQueue *queue;
|
752
|
+
SpanEnum **span_enums;
|
753
|
+
int s_cnt;
|
754
|
+
bool first_time : 1;
|
755
|
+
} SpanOrEnum;
|
756
|
+
|
757
|
+
|
758
|
+
static bool span_less_than(SpanEnum *s1, SpanEnum *s2)
|
759
|
+
{
|
760
|
+
int doc_diff, start_diff;
|
761
|
+
doc_diff = s1->doc(s1) - s2->doc(s2);
|
762
|
+
if (doc_diff == 0) {
|
763
|
+
start_diff = s1->start(s1) - s2->start(s2);
|
764
|
+
if (start_diff == 0) {
|
765
|
+
return s1->end(s1) < s2->end(s2);
|
766
|
+
}
|
767
|
+
else {
|
768
|
+
return start_diff < 0;
|
769
|
+
}
|
770
|
+
}
|
771
|
+
else {
|
772
|
+
return doc_diff < 0;
|
773
|
+
}
|
774
|
+
}
|
775
|
+
|
776
|
+
static bool spanoe_next(SpanEnum *self)
|
777
|
+
{
|
778
|
+
SpanOrEnum *soe = SpOEn(self);
|
779
|
+
SpanEnum *se;
|
780
|
+
int i;
|
781
|
+
|
782
|
+
if (soe->first_time) { /* first time -- initialize */
|
783
|
+
for (i = 0; i < soe->s_cnt; i++) {
|
784
|
+
se = soe->span_enums[i];
|
785
|
+
if (se->next(se)) { /* move to first entry */
|
786
|
+
pq_push(soe->queue, se);
|
787
|
+
}
|
788
|
+
}
|
789
|
+
soe->first_time = false;
|
790
|
+
return soe->queue->size != 0;
|
791
|
+
}
|
792
|
+
|
793
|
+
if (soe->queue->size == 0) {
|
794
|
+
return false; /* all done */
|
795
|
+
}
|
796
|
+
|
797
|
+
se = (SpanEnum *)pq_top(soe->queue);
|
798
|
+
if (se->next(se)) { /* move to next */
|
799
|
+
pq_down(soe->queue);
|
800
|
+
return true;
|
801
|
+
}
|
802
|
+
|
803
|
+
pq_pop(soe->queue); /* exhausted a clause */
|
804
|
+
|
805
|
+
return soe->queue->size != 0;
|
806
|
+
}
|
807
|
+
|
808
|
+
static bool spanoe_skip_to(SpanEnum *self, int target)
|
809
|
+
{
|
810
|
+
SpanOrEnum *soe = SpOEn(self);
|
811
|
+
SpanEnum *se;
|
812
|
+
int i;
|
813
|
+
|
814
|
+
if (soe->first_time) { /* first time -- initialize */
|
815
|
+
for (i = 0; i < soe->s_cnt; i++) {
|
816
|
+
se = soe->span_enums[i];
|
817
|
+
if (se->skip_to(se, target)) {/* move to target */
|
818
|
+
pq_push(soe->queue, se);
|
819
|
+
}
|
820
|
+
}
|
821
|
+
soe->first_time = false;
|
822
|
+
}
|
823
|
+
else {
|
824
|
+
while ((soe->queue->size != 0) &&
|
825
|
+
((se = (SpanEnum *)pq_top(soe->queue))->doc(se) < target)) {
|
826
|
+
if (se->skip_to(se, target)) {
|
827
|
+
pq_down(soe->queue);
|
828
|
+
}
|
829
|
+
else {
|
830
|
+
pq_pop(soe->queue);
|
831
|
+
}
|
832
|
+
}
|
833
|
+
}
|
834
|
+
|
835
|
+
return soe->queue->size != 0;
|
836
|
+
}
|
837
|
+
|
838
|
+
#define SpOEn_Top_SE(self) (SpanEnum *)pq_top(SpOEn(self)->queue)
|
839
|
+
|
840
|
+
static int spanoe_doc(SpanEnum *self)
|
841
|
+
{
|
842
|
+
SpanEnum *se = SpOEn_Top_SE(self);
|
843
|
+
return se->doc(se);
|
844
|
+
}
|
845
|
+
|
846
|
+
static int spanoe_start(SpanEnum *self)
|
847
|
+
{
|
848
|
+
SpanEnum *se = SpOEn_Top_SE(self);
|
849
|
+
return se->start(se);
|
850
|
+
}
|
851
|
+
|
852
|
+
static int spanoe_end(SpanEnum *self)
|
853
|
+
{
|
854
|
+
SpanEnum *se = SpOEn_Top_SE(self);
|
855
|
+
return se->end(se);
|
856
|
+
}
|
857
|
+
|
858
|
+
static char *spanoe_to_s(SpanEnum *self)
|
859
|
+
{
|
860
|
+
SpanOrEnum *soe = SpOEn(self);
|
861
|
+
char *field = SpQ(self->query)->field;
|
862
|
+
char *query_str = self->query->to_s(self->query, field);
|
863
|
+
char doc_str[62];
|
864
|
+
size_t len = strlen(query_str);
|
865
|
+
char *str = ALLOC_N(char, len + 80);
|
866
|
+
|
867
|
+
if (soe->first_time) {
|
868
|
+
sprintf(doc_str, "START");
|
869
|
+
}
|
870
|
+
else {
|
871
|
+
if (soe->queue->size == 0) {
|
872
|
+
sprintf(doc_str, "END");
|
873
|
+
}
|
874
|
+
else {
|
875
|
+
sprintf(doc_str, "%d:%d-%d", self->doc(self),
|
876
|
+
self->start(self), self->end(self));
|
877
|
+
}
|
878
|
+
}
|
879
|
+
sprintf("SpanOrEnum(%s)@%s", query_str, doc_str);
|
880
|
+
free(query_str);
|
881
|
+
return str;
|
882
|
+
}
|
883
|
+
|
884
|
+
static void spanoe_destroy(SpanEnum *self)
|
885
|
+
{
|
886
|
+
SpanEnum *se;
|
887
|
+
SpanOrEnum *soe = SpOEn(self);
|
888
|
+
int i;
|
889
|
+
pq_destroy(soe->queue);
|
890
|
+
for (i = 0; i < soe->s_cnt; i++) {
|
891
|
+
se = soe->span_enums[i];
|
892
|
+
se->destroy(se);
|
893
|
+
}
|
894
|
+
free(soe->span_enums);
|
895
|
+
free(self);
|
896
|
+
}
|
897
|
+
|
898
|
+
SpanEnum *spanoe_new(Query *query, IndexReader *ir)
|
899
|
+
{
|
900
|
+
Query *clause;
|
901
|
+
SpanEnum *self = (SpanEnum *)emalloc(sizeof(SpanOrEnum));
|
902
|
+
SpanOrQuery *soq = SpOQ(query);
|
903
|
+
int i;
|
904
|
+
|
905
|
+
SpOEn(self)->first_time = true;
|
906
|
+
SpOEn(self)->s_cnt = soq->c_cnt;
|
907
|
+
SpOEn(self)->span_enums = ALLOC_N(SpanEnum *, SpOEn(self)->s_cnt);
|
908
|
+
|
909
|
+
for (i = 0; i < SpOEn(self)->s_cnt; i++) {
|
910
|
+
clause = soq->clauses[i];
|
911
|
+
SpOEn(self)->span_enums[i] = SpQ(clause)->get_spans(clause, ir);
|
912
|
+
}
|
913
|
+
|
914
|
+
SpOEn(self)->queue = pq_new(SpOEn(self)->s_cnt, (lt_ft)&span_less_than,
|
915
|
+
(free_ft)NULL);
|
916
|
+
|
917
|
+
self->query = query;
|
918
|
+
self->next = &spanoe_next;
|
919
|
+
self->skip_to = &spanoe_skip_to;
|
920
|
+
self->doc = &spanoe_doc;
|
921
|
+
self->start = &spanoe_start;
|
922
|
+
self->end = &spanoe_end;
|
923
|
+
self->destroy = &spanoe_destroy;
|
924
|
+
self->to_s = &spanoe_to_s;
|
925
|
+
|
926
|
+
return self;
|
927
|
+
}
|
928
|
+
|
929
|
+
/*****************************************************************************
|
930
|
+
* SpanNearEnum
|
931
|
+
*****************************************************************************/
|
932
|
+
|
933
|
+
#define SpNEn(span_enum) ((SpanNearEnum *)(span_enum))
|
934
|
+
#define SpNQ(query) ((SpanNearQuery *)(query))
|
935
|
+
|
936
|
+
typedef struct SpanNearEnum
|
937
|
+
{
|
938
|
+
SpanEnum super;
|
939
|
+
SpanEnum **span_enums;
|
940
|
+
int s_cnt;
|
941
|
+
int slop;
|
942
|
+
int current;
|
943
|
+
int doc;
|
944
|
+
int start;
|
945
|
+
int end;
|
946
|
+
bool first_time : 1;
|
947
|
+
bool in_order : 1;
|
948
|
+
} SpanNearEnum;
|
949
|
+
|
950
|
+
|
951
|
+
#define SpNEn_NEXT() do {\
|
952
|
+
sne->current = (sne->current+1) % sne->s_cnt;\
|
953
|
+
se = sne->span_enums[sne->current];\
|
954
|
+
} while (0);
|
955
|
+
|
956
|
+
static bool sne_init(SpanNearEnum *sne)
|
957
|
+
{
|
958
|
+
SpanEnum *se = sne->span_enums[sne->current];
|
959
|
+
int prev_doc = se->doc(se);
|
960
|
+
int i;
|
961
|
+
|
962
|
+
for (i = 1; i < sne->s_cnt; i++) {
|
963
|
+
SpNEn_NEXT();
|
964
|
+
if (!se->skip_to(se, prev_doc)) {
|
965
|
+
return false;
|
966
|
+
}
|
967
|
+
prev_doc = se->doc(se);
|
968
|
+
}
|
969
|
+
return true;
|
970
|
+
}
|
971
|
+
|
972
|
+
static bool sne_goto_next_doc(SpanNearEnum *sne)
|
973
|
+
{
|
974
|
+
SpanEnum *se = sne->span_enums[sne->current];
|
975
|
+
int prev_doc = se->doc(se);
|
976
|
+
|
977
|
+
SpNEn_NEXT();
|
978
|
+
|
979
|
+
while (se->doc(se) < prev_doc) {
|
980
|
+
if (! se->skip_to(se, prev_doc)) {
|
981
|
+
return false;
|
982
|
+
}
|
983
|
+
prev_doc = se->doc(se);
|
984
|
+
SpNEn_NEXT();
|
985
|
+
}
|
986
|
+
return true;
|
987
|
+
}
|
988
|
+
|
989
|
+
static bool sne_next_unordered_match(SpanEnum *self)
|
990
|
+
{
|
991
|
+
SpanNearEnum *sne = SpNEn(self);
|
992
|
+
SpanEnum *se, *min_se = NULL;
|
993
|
+
int i;
|
994
|
+
int max_end, end, min_start, start, doc;
|
995
|
+
int lengths_sum;
|
996
|
+
|
997
|
+
while (true) {
|
998
|
+
max_end = 0;
|
999
|
+
min_start = INT_MAX;
|
1000
|
+
lengths_sum = 0;
|
1001
|
+
|
1002
|
+
for (i = 0; i < sne->s_cnt; i++) {
|
1003
|
+
se = sne->span_enums[i];
|
1004
|
+
if ((end=se->end(se)) > max_end) {
|
1005
|
+
max_end = end;
|
1006
|
+
}
|
1007
|
+
if ((start=se->start(se)) < min_start) {
|
1008
|
+
min_start = start;
|
1009
|
+
min_se = se;
|
1010
|
+
sne->current = i; /* current should point to the minimum span */
|
1011
|
+
}
|
1012
|
+
lengths_sum += end - start;
|
1013
|
+
}
|
1014
|
+
|
1015
|
+
if ((max_end - min_start - lengths_sum) <= sne->slop) {
|
1016
|
+
/* we have a match */
|
1017
|
+
sne->start = min_start;
|
1018
|
+
sne->end = max_end;
|
1019
|
+
sne->doc = min_se->doc(min_se);
|
1020
|
+
return true;
|
1021
|
+
}
|
1022
|
+
|
1023
|
+
/* increment the minimum span_enum and try again */
|
1024
|
+
doc = min_se->doc(min_se);
|
1025
|
+
if (!min_se->next(min_se)) {
|
1026
|
+
return false;
|
1027
|
+
}
|
1028
|
+
if (doc < min_se->doc(min_se)) {
|
1029
|
+
if (!sne_goto_next_doc(sne)) return false;
|
1030
|
+
}
|
1031
|
+
}
|
1032
|
+
}
|
1033
|
+
|
1034
|
+
static bool sne_next_ordered_match(SpanEnum *self)
|
1035
|
+
{
|
1036
|
+
SpanNearEnum *sne = SpNEn(self);
|
1037
|
+
SpanEnum *se;
|
1038
|
+
int i;
|
1039
|
+
int prev_doc, prev_start, prev_end;
|
1040
|
+
int doc=0, start=0, end=0;
|
1041
|
+
int lengths_sum;
|
1042
|
+
|
1043
|
+
while (true) {
|
1044
|
+
se = sne->span_enums[0];
|
1045
|
+
|
1046
|
+
prev_doc = se->doc(se);
|
1047
|
+
sne->start = prev_start = se->start(se);
|
1048
|
+
prev_end = se->end(se);
|
1049
|
+
|
1050
|
+
i = 1;
|
1051
|
+
lengths_sum = prev_end - prev_start;
|
1052
|
+
|
1053
|
+
while (i < sne->s_cnt) {
|
1054
|
+
se = sne->span_enums[i];
|
1055
|
+
doc = se->doc(se);
|
1056
|
+
start = se->start(se);
|
1057
|
+
end = se->end(se);
|
1058
|
+
while ((doc == prev_doc) && ((start < prev_start) ||
|
1059
|
+
((start == prev_start) && (end < prev_end)))) {
|
1060
|
+
if (!se->next(se)) {
|
1061
|
+
return false;
|
1062
|
+
}
|
1063
|
+
doc = se->doc(se);
|
1064
|
+
start = se->start(se);
|
1065
|
+
end = se->end(se);
|
1066
|
+
}
|
1067
|
+
if (doc != prev_doc) {
|
1068
|
+
sne->current = i;
|
1069
|
+
if (!sne_goto_next_doc(sne)) {
|
1070
|
+
return false;
|
1071
|
+
}
|
1072
|
+
break;
|
1073
|
+
}
|
1074
|
+
i++;
|
1075
|
+
lengths_sum += end - start;
|
1076
|
+
prev_doc = doc;
|
1077
|
+
prev_start = start;
|
1078
|
+
prev_end = end;
|
1079
|
+
}
|
1080
|
+
if (i == sne->s_cnt) {
|
1081
|
+
if ((end - sne->start - lengths_sum) <= sne->slop) {
|
1082
|
+
/* we have a match */
|
1083
|
+
sne->end = end;
|
1084
|
+
sne->doc = doc;
|
1085
|
+
|
1086
|
+
/* the minimum span is always the first span so it needs to be
|
1087
|
+
* incremented next time around */
|
1088
|
+
sne->current = 0;
|
1089
|
+
return true;
|
1090
|
+
|
1091
|
+
}
|
1092
|
+
else {
|
1093
|
+
se = sne->span_enums[0];
|
1094
|
+
if (!se->next(se)) {
|
1095
|
+
return false;
|
1096
|
+
}
|
1097
|
+
if (se->doc(se) != prev_doc) {
|
1098
|
+
sne->current = 0;
|
1099
|
+
if (!sne_goto_next_doc(sne)) {
|
1100
|
+
return false;
|
1101
|
+
}
|
1102
|
+
}
|
1103
|
+
}
|
1104
|
+
}
|
1105
|
+
}
|
1106
|
+
}
|
1107
|
+
|
1108
|
+
static bool sne_next_match(SpanEnum *self)
|
1109
|
+
{
|
1110
|
+
SpanNearEnum *sne = SpNEn(self);
|
1111
|
+
SpanEnum *se_curr, *se_next;
|
1112
|
+
|
1113
|
+
if (!sne->first_time) {
|
1114
|
+
if (!sne_init(sne)) {
|
1115
|
+
return false;
|
1116
|
+
}
|
1117
|
+
sne->first_time = false;
|
1118
|
+
}
|
1119
|
+
se_curr = sne->span_enums[sne->current];
|
1120
|
+
se_next = sne->span_enums[(sne->current+1)%sne->s_cnt];
|
1121
|
+
if (se_curr->doc(se_curr) > se_next->doc(se_next)) {
|
1122
|
+
if (!sne_goto_next_doc(sne)) {
|
1123
|
+
return false;
|
1124
|
+
}
|
1125
|
+
}
|
1126
|
+
|
1127
|
+
if (sne->in_order) {
|
1128
|
+
return sne_next_ordered_match(self);
|
1129
|
+
}
|
1130
|
+
else {
|
1131
|
+
return sne_next_unordered_match(self);
|
1132
|
+
}
|
1133
|
+
}
|
1134
|
+
|
1135
|
+
static bool spanne_next(SpanEnum *self)
|
1136
|
+
{
|
1137
|
+
SpanNearEnum *sne = SpNEn(self);
|
1138
|
+
SpanEnum *se;
|
1139
|
+
|
1140
|
+
se = sne->span_enums[sne->current];
|
1141
|
+
if (!se->next(se)) return false;
|
1142
|
+
|
1143
|
+
return sne_next_match(self);
|
1144
|
+
}
|
1145
|
+
|
1146
|
+
static bool spanne_skip_to(SpanEnum *self, int target)
|
1147
|
+
{
|
1148
|
+
SpanEnum *se = SpNEn(self)->span_enums[SpNEn(self)->current];
|
1149
|
+
if (!se->skip_to(se, target)) {
|
1150
|
+
return false;
|
1151
|
+
}
|
1152
|
+
|
1153
|
+
return sne_next_match(self);
|
1154
|
+
}
|
1155
|
+
|
1156
|
+
static int spanne_doc(SpanEnum *self)
|
1157
|
+
{
|
1158
|
+
return SpNEn(self)->doc;
|
1159
|
+
}
|
1160
|
+
|
1161
|
+
static int spanne_start(SpanEnum *self)
|
1162
|
+
{
|
1163
|
+
return SpNEn(self)->start;
|
1164
|
+
}
|
1165
|
+
|
1166
|
+
static int spanne_end(SpanEnum *self)
|
1167
|
+
{
|
1168
|
+
return SpNEn(self)->end;
|
1169
|
+
}
|
1170
|
+
|
1171
|
+
static char *spanne_to_s(SpanEnum *self)
|
1172
|
+
{
|
1173
|
+
SpanNearEnum *sne = SpNEn(self);
|
1174
|
+
char *field = SpQ(self->query)->field;
|
1175
|
+
char *query_str = self->query->to_s(self->query, field);
|
1176
|
+
char doc_str[62];
|
1177
|
+
size_t len = strlen(query_str);
|
1178
|
+
char *str = 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("SpanNearEnum(%s)@%s", query_str, doc_str);
|
1188
|
+
free(query_str);
|
1189
|
+
return str;
|
1190
|
+
}
|
1191
|
+
|
1192
|
+
static void spanne_destroy(SpanEnum *self)
|
1193
|
+
{
|
1194
|
+
SpanEnum *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 SpanEnum *spanne_new(Query *query, IndexReader *ir)
|
1206
|
+
{
|
1207
|
+
int i;
|
1208
|
+
Query *clause;
|
1209
|
+
SpanEnum *self = (SpanEnum *)emalloc(sizeof(SpanNearEnum));
|
1210
|
+
SpanNearQuery *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 = ALLOC_N(SpanEnum *, 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) ((SpanNotQuery *)(query))
|
1248
|
+
|
1249
|
+
typedef struct SpanNotEnum
|
1250
|
+
{
|
1251
|
+
SpanEnum super;
|
1252
|
+
SpanEnum *inc;
|
1253
|
+
SpanEnum *exc;
|
1254
|
+
bool more_inc : 1;
|
1255
|
+
bool more_exc : 1;
|
1256
|
+
} SpanNotEnum;
|
1257
|
+
|
1258
|
+
|
1259
|
+
static bool spanxe_next(SpanEnum *self)
|
1260
|
+
{
|
1261
|
+
SpanNotEnum *sxe = SpXEn(self);
|
1262
|
+
SpanEnum *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(SpanEnum *self, int target)
|
1290
|
+
{
|
1291
|
+
SpanNotEnum *sxe = SpXEn(self);
|
1292
|
+
SpanEnum *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(SpanEnum *self)
|
1319
|
+
{
|
1320
|
+
SpanEnum *inc = SpXEn(self)->inc;
|
1321
|
+
return inc->doc(inc);
|
1322
|
+
}
|
1323
|
+
|
1324
|
+
static int spanxe_start(SpanEnum *self)
|
1325
|
+
{
|
1326
|
+
SpanEnum *inc = SpXEn(self)->inc;
|
1327
|
+
return inc->start(inc);
|
1328
|
+
}
|
1329
|
+
|
1330
|
+
static int spanxe_end(SpanEnum *self)
|
1331
|
+
{
|
1332
|
+
SpanEnum *inc = SpXEn(self)->inc;
|
1333
|
+
return inc->end(inc);
|
1334
|
+
}
|
1335
|
+
|
1336
|
+
static char *spanxe_to_s(SpanEnum *self)
|
1337
|
+
{
|
1338
|
+
char *field = SpQ(self->query)->field;
|
1339
|
+
char *query_str = self->query->to_s(self->query, field);
|
1340
|
+
char *res = strfmt("SpanNotEnum(%s)", query_str);
|
1341
|
+
free(query_str);
|
1342
|
+
return res;
|
1343
|
+
}
|
1344
|
+
|
1345
|
+
static void spanxe_destroy(SpanEnum *self)
|
1346
|
+
{
|
1347
|
+
SpanNotEnum *sxe = SpXEn(self);
|
1348
|
+
sxe->inc->destroy(sxe->inc);
|
1349
|
+
sxe->exc->destroy(sxe->exc);
|
1350
|
+
free(self);
|
1351
|
+
}
|
1352
|
+
|
1353
|
+
static SpanEnum *spanxe_new(Query *query, IndexReader *ir)
|
1354
|
+
{
|
1355
|
+
SpanEnum *self = (SpanEnum *)emalloc(sizeof(SpanNotEnum));
|
1356
|
+
SpanNotEnum *sxe = SpXEn(self);
|
1357
|
+
SpanNotQuery *sxq = SpXQ(query);
|
1358
|
+
|
1359
|
+
sxe->inc = SpQ(sxq->inc)->get_spans(sxq->inc, ir);
|
1360
|
+
sxe->exc = SpQ(sxq->exc)->get_spans(sxq->exc, ir);
|
1361
|
+
sxe->more_inc = true;
|
1362
|
+
sxe->more_exc = sxe->exc->next(sxe->exc);
|
1363
|
+
|
1364
|
+
self->query = query;
|
1365
|
+
self->next = &spanxe_next;
|
1366
|
+
self->skip_to = &spanxe_skip_to;
|
1367
|
+
self->doc = &spanxe_doc;
|
1368
|
+
self->start = &spanxe_start;
|
1369
|
+
self->end = &spanxe_end;
|
1370
|
+
self->destroy = &spanxe_destroy;
|
1371
|
+
self->to_s = &spanxe_to_s;
|
1372
|
+
|
1373
|
+
return self;
|
1374
|
+
}
|
1375
|
+
|
1376
|
+
/*****************************************************************************
|
1377
|
+
*
|
1378
|
+
* SpanWeight
|
1379
|
+
*
|
1380
|
+
*****************************************************************************/
|
1381
|
+
|
1382
|
+
#define SpW(weight) ((SpanWeight *)(weight))
|
1383
|
+
typedef struct SpanWeight
|
1384
|
+
{
|
1385
|
+
Weight super;
|
1386
|
+
HashSet *terms;
|
1387
|
+
} SpanWeight;
|
1388
|
+
|
1389
|
+
static Explanation *spanw_explain(Weight *self, IndexReader *ir, int target)
|
1390
|
+
{
|
1391
|
+
Explanation *expl;
|
1392
|
+
Explanation *idf_expl1;
|
1393
|
+
Explanation *idf_expl2;
|
1394
|
+
Explanation *query_expl;
|
1395
|
+
Explanation *qnorm_expl;
|
1396
|
+
Explanation *field_expl;
|
1397
|
+
Explanation *tf_expl;
|
1398
|
+
Scorer *scorer;
|
1399
|
+
uchar *field_norms;
|
1400
|
+
float field_norm;
|
1401
|
+
Explanation *field_norm_expl;
|
1402
|
+
|
1403
|
+
char *query_str;
|
1404
|
+
HashSet *terms = SpW(self)->terms;
|
1405
|
+
char *field = SpQ(self->query)->field;
|
1406
|
+
const int field_num = fis_get_field_num(ir->fis, field);
|
1407
|
+
char *doc_freqs = NULL;
|
1408
|
+
size_t df_i = 0;
|
1409
|
+
int i;
|
1410
|
+
|
1411
|
+
if (field_num < 0) {
|
1412
|
+
return expl_new(0.0, "field \"%s\" does not exist in the index", field);
|
1413
|
+
}
|
1414
|
+
|
1415
|
+
query_str = self->query->to_s(self->query, "");
|
1416
|
+
|
1417
|
+
for (i = 0; i < terms->size; i++) {
|
1418
|
+
char *term = (char *)terms->elems[i];
|
1419
|
+
REALLOC_N(doc_freqs, char, df_i + strlen(term) + 23);
|
1420
|
+
sprintf(doc_freqs + df_i, "%s=%d, ", term,
|
1421
|
+
ir->doc_freq(ir, field_num, term));
|
1422
|
+
df_i = strlen(doc_freqs);
|
1423
|
+
}
|
1424
|
+
/* remove the ',' at the end of the string if it exists */
|
1425
|
+
if (terms->size > 0) {
|
1426
|
+
df_i -= 2;
|
1427
|
+
doc_freqs[df_i] = '\0';
|
1428
|
+
}
|
1429
|
+
else {
|
1430
|
+
doc_freqs = "";
|
1431
|
+
}
|
1432
|
+
|
1433
|
+
expl = expl_new(0.0, "weight(%s in %d), product of:", query_str, target);
|
1434
|
+
|
1435
|
+
/* We need two of these as it's included in both the query explanation
|
1436
|
+
* and the field explanation */
|
1437
|
+
idf_expl1 = expl_new(self->idf, "idf(%s: %s)", field, doc_freqs);
|
1438
|
+
idf_expl2 = expl_new(self->idf, "idf(%s: %s)", field, doc_freqs);
|
1439
|
+
if (terms->size > 0) {
|
1440
|
+
free(doc_freqs); /* only free if allocated */
|
1441
|
+
}
|
1442
|
+
|
1443
|
+
/* explain query weight */
|
1444
|
+
query_expl = expl_new(0.0, "query_weight(%s), product of:", query_str);
|
1445
|
+
|
1446
|
+
if (self->query->boost != 1.0) {
|
1447
|
+
expl_add_detail(query_expl, expl_new(self->query->boost, "boost"));
|
1448
|
+
}
|
1449
|
+
|
1450
|
+
expl_add_detail(query_expl, idf_expl1);
|
1451
|
+
|
1452
|
+
qnorm_expl = expl_new(self->qnorm, "query_norm");
|
1453
|
+
expl_add_detail(query_expl, qnorm_expl);
|
1454
|
+
|
1455
|
+
query_expl->value = self->query->boost * idf_expl1->value * qnorm_expl->value;
|
1456
|
+
|
1457
|
+
expl_add_detail(expl, query_expl);
|
1458
|
+
|
1459
|
+
/* explain field weight */
|
1460
|
+
field_expl = expl_new(0.0, "field_weight(%s:%s in %d), product of:",
|
1461
|
+
field, query_str, target);
|
1462
|
+
free(query_str);
|
1463
|
+
|
1464
|
+
scorer = self->scorer(self, ir);
|
1465
|
+
tf_expl = scorer->explain(scorer, target);
|
1466
|
+
scorer->destroy(scorer);
|
1467
|
+
expl_add_detail(field_expl, tf_expl);
|
1468
|
+
expl_add_detail(field_expl, idf_expl2);
|
1469
|
+
|
1470
|
+
field_norms = ir->get_norms(ir, field_num);
|
1471
|
+
field_norm = (field_norms
|
1472
|
+
? sim_decode_norm(self->similarity, field_norms[target])
|
1473
|
+
: (float)0.0);
|
1474
|
+
field_norm_expl = expl_new(field_norm, "field_norm(field=%s, doc=%d)",
|
1475
|
+
field, target);
|
1476
|
+
expl_add_detail(field_expl, field_norm_expl);
|
1477
|
+
|
1478
|
+
field_expl->value = tf_expl->value * idf_expl2->value * field_norm_expl->value;
|
1479
|
+
|
1480
|
+
/* combine them */
|
1481
|
+
if (query_expl->value == 1.0) {
|
1482
|
+
expl_destroy(expl);
|
1483
|
+
return field_expl;
|
1484
|
+
}
|
1485
|
+
else {
|
1486
|
+
expl->value = (query_expl->value * field_expl->value);
|
1487
|
+
expl_add_detail(expl, field_expl);
|
1488
|
+
return expl;
|
1489
|
+
}
|
1490
|
+
}
|
1491
|
+
|
1492
|
+
static char *spanw_to_s(Weight *self)
|
1493
|
+
{
|
1494
|
+
return strfmt("SpanWeight(%f)", self->value);
|
1495
|
+
}
|
1496
|
+
|
1497
|
+
static void spanw_destroy(Weight *self)
|
1498
|
+
{
|
1499
|
+
hs_destroy(SpW(self)->terms);
|
1500
|
+
w_destroy(self);
|
1501
|
+
}
|
1502
|
+
|
1503
|
+
static Weight *spanw_new(Query *query, Searcher *searcher)
|
1504
|
+
{
|
1505
|
+
int i;
|
1506
|
+
Weight *self = w_new(SpanWeight, query);
|
1507
|
+
HashSet *terms = SpQ(query)->get_terms(query);
|
1508
|
+
|
1509
|
+
SpW(self)->terms = terms;
|
1510
|
+
self->scorer = &spansc_new;
|
1511
|
+
self->explain = &spanw_explain;
|
1512
|
+
self->to_s = &spanw_to_s;
|
1513
|
+
self->destroy = &spanw_destroy;
|
1514
|
+
|
1515
|
+
self->similarity = query->get_similarity(query, searcher);
|
1516
|
+
|
1517
|
+
self->idf = 0.0;
|
1518
|
+
|
1519
|
+
for (i = terms->size - 1; i >= 0; i--) {
|
1520
|
+
self->idf += sim_idf_term(self->similarity, SpQ(query)->field,
|
1521
|
+
(char *)terms->elems[i], searcher);
|
1522
|
+
}
|
1523
|
+
|
1524
|
+
return self;
|
1525
|
+
}
|
1526
|
+
|
1527
|
+
/*****************************************************************************
|
1528
|
+
* SpanTermQuery
|
1529
|
+
*****************************************************************************/
|
1530
|
+
|
1531
|
+
static char *spantq_to_s(Query *self, const char *field)
|
1532
|
+
{
|
1533
|
+
if (field == SpQ(self)->field) {
|
1534
|
+
return strfmt("span_terms(%s)", SpTQ(self)->term);
|
1535
|
+
}
|
1536
|
+
else {
|
1537
|
+
return strfmt("span_terms(%s:%s)", SpQ(self)->field, SpTQ(self)->term);
|
1538
|
+
}
|
1539
|
+
}
|
1540
|
+
|
1541
|
+
static void spantq_destroy_i(Query *self)
|
1542
|
+
{
|
1543
|
+
free(SpTQ(self)->term);
|
1544
|
+
free(SpQ(self)->field);
|
1545
|
+
spanq_destroy_i(self);
|
1546
|
+
}
|
1547
|
+
|
1548
|
+
static void spantq_extract_terms(Query *self, HashSet *terms)
|
1549
|
+
{
|
1550
|
+
hs_add(terms, term_new(SpQ(self)->field, SpTQ(self)->term));
|
1551
|
+
}
|
1552
|
+
|
1553
|
+
static HashSet *spantq_get_terms(Query *self)
|
1554
|
+
{
|
1555
|
+
HashSet *terms = hs_new_str(&free);
|
1556
|
+
hs_add(terms, estrdup(SpTQ(self)->term));
|
1557
|
+
return terms;
|
1558
|
+
}
|
1559
|
+
|
1560
|
+
static unsigned long spantq_hash(Query *self)
|
1561
|
+
{
|
1562
|
+
return spanq_hash(self) ^ str_hash(SpTQ(self)->term);
|
1563
|
+
}
|
1564
|
+
|
1565
|
+
static int spantq_eq(Query *self, Query *o)
|
1566
|
+
{
|
1567
|
+
return spanq_eq(self, o) && strcmp(SpTQ(self)->term, SpTQ(o)->term) == 0;
|
1568
|
+
}
|
1569
|
+
|
1570
|
+
Query *spantq_new(const char *field, const char *term)
|
1571
|
+
{
|
1572
|
+
Query *self = q_new(SpanTermQuery);
|
1573
|
+
|
1574
|
+
SpTQ(self)->term = estrdup(term);
|
1575
|
+
SpQ(self)->field = estrdup(field);
|
1576
|
+
SpQ(self)->get_spans = &spante_new;
|
1577
|
+
SpQ(self)->get_terms = &spantq_get_terms;
|
1578
|
+
|
1579
|
+
self->type = SPAN_TERM_QUERY;
|
1580
|
+
self->extract_terms = &spantq_extract_terms;
|
1581
|
+
self->to_s = &spantq_to_s;
|
1582
|
+
self->hash = &spantq_hash;
|
1583
|
+
self->eq = &spantq_eq;
|
1584
|
+
self->destroy_i = &spantq_destroy_i;
|
1585
|
+
self->create_weight_i = &spanw_new;
|
1586
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
1587
|
+
return self;
|
1588
|
+
}
|
1589
|
+
|
1590
|
+
/*****************************************************************************
|
1591
|
+
* SpanMultiTermQuery
|
1592
|
+
*****************************************************************************/
|
1593
|
+
|
1594
|
+
static char *spanmtq_to_s(Query *self, const char *field)
|
1595
|
+
{
|
1596
|
+
char *terms = NULL, *p;
|
1597
|
+
int len = 2, i;
|
1598
|
+
SpanMultiTermQuery *smtq = SpMTQ(self);
|
1599
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1600
|
+
len += strlen(smtq->terms[i]) + 2;
|
1601
|
+
}
|
1602
|
+
p = terms = ALLOC_N(char, len);
|
1603
|
+
*(p++) = '[';
|
1604
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1605
|
+
strcpy(p, smtq->terms[i]);
|
1606
|
+
p += strlen(smtq->terms[i]);
|
1607
|
+
*(p++) = ',';
|
1608
|
+
}
|
1609
|
+
if (p > terms) p--;
|
1610
|
+
*(p++) = ']';
|
1611
|
+
*p = '\0';
|
1612
|
+
|
1613
|
+
if (field == SpQ(self)->field) {
|
1614
|
+
p = strfmt("span_terms(%s)", terms);
|
1615
|
+
}
|
1616
|
+
else {
|
1617
|
+
p = strfmt("span_terms(%s:%s)", SpQ(self)->field, terms);
|
1618
|
+
}
|
1619
|
+
free(terms);
|
1620
|
+
return p;
|
1621
|
+
}
|
1622
|
+
|
1623
|
+
static void spanmtq_destroy_i(Query *self)
|
1624
|
+
{
|
1625
|
+
SpanMultiTermQuery *smtq = SpMTQ(self);
|
1626
|
+
int i;
|
1627
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1628
|
+
free(smtq->terms[i]);
|
1629
|
+
}
|
1630
|
+
free(smtq->terms);
|
1631
|
+
free(SpQ(self)->field);
|
1632
|
+
spanq_destroy_i(self);
|
1633
|
+
}
|
1634
|
+
|
1635
|
+
static void spanmtq_extract_terms(Query *self, HashSet *terms)
|
1636
|
+
{
|
1637
|
+
SpanMultiTermQuery *smtq = SpMTQ(self);
|
1638
|
+
int i;
|
1639
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1640
|
+
hs_add(terms, term_new(SpQ(self)->field, smtq->terms[i]));
|
1641
|
+
}
|
1642
|
+
}
|
1643
|
+
|
1644
|
+
static HashSet *spanmtq_get_terms(Query *self)
|
1645
|
+
{
|
1646
|
+
HashSet *terms = hs_new_str(&free);
|
1647
|
+
SpanMultiTermQuery *smtq = SpMTQ(self);
|
1648
|
+
int i;
|
1649
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1650
|
+
hs_add(terms, estrdup(smtq->terms[i]));
|
1651
|
+
}
|
1652
|
+
return terms;
|
1653
|
+
}
|
1654
|
+
|
1655
|
+
static unsigned long spanmtq_hash(Query *self)
|
1656
|
+
{
|
1657
|
+
unsigned long hash = spanq_hash(self);
|
1658
|
+
SpanMultiTermQuery *smtq = SpMTQ(self);
|
1659
|
+
int i;
|
1660
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1661
|
+
hash ^= str_hash(smtq->terms[i]);
|
1662
|
+
}
|
1663
|
+
return hash;
|
1664
|
+
}
|
1665
|
+
|
1666
|
+
static int spanmtq_eq(Query *self, Query *o)
|
1667
|
+
{
|
1668
|
+
SpanMultiTermQuery *smtq = SpMTQ(self);
|
1669
|
+
SpanMultiTermQuery *smtqo = SpMTQ(o);
|
1670
|
+
int i;
|
1671
|
+
if (!spanq_eq(self, o)) return false;
|
1672
|
+
if (smtq->term_cnt != smtqo->term_cnt) return false;
|
1673
|
+
for (i = 0; i < smtq->term_cnt; i++) {
|
1674
|
+
if (strcmp(smtq->terms[i], smtqo->terms[i]) != 0) return false;
|
1675
|
+
}
|
1676
|
+
return true;;
|
1677
|
+
}
|
1678
|
+
|
1679
|
+
Query *spanmtq_new_conf(const char *field, int max_terms)
|
1680
|
+
{
|
1681
|
+
Query *self = q_new(SpanMultiTermQuery);
|
1682
|
+
|
1683
|
+
SpMTQ(self)->terms = ALLOC_N(char *, max_terms);
|
1684
|
+
SpMTQ(self)->term_cnt = 0;
|
1685
|
+
SpMTQ(self)->term_capa = max_terms;
|
1686
|
+
|
1687
|
+
SpQ(self)->field = estrdup(field);
|
1688
|
+
SpQ(self)->get_spans = &spanmte_new;
|
1689
|
+
SpQ(self)->get_terms = &spanmtq_get_terms;
|
1690
|
+
|
1691
|
+
self->type = SPAN_MULTI_TERM_QUERY;
|
1692
|
+
self->extract_terms = &spanmtq_extract_terms;
|
1693
|
+
self->to_s = &spanmtq_to_s;
|
1694
|
+
self->hash = &spanmtq_hash;
|
1695
|
+
self->eq = &spanmtq_eq;
|
1696
|
+
self->destroy_i = &spanmtq_destroy_i;
|
1697
|
+
self->create_weight_i = &spanw_new;
|
1698
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
1699
|
+
|
1700
|
+
return self;
|
1701
|
+
}
|
1702
|
+
|
1703
|
+
Query *spanmtq_new(const char *field)
|
1704
|
+
{
|
1705
|
+
return spanmtq_new_conf(field, SPAN_MULTI_TERM_QUERY_CAPA);
|
1706
|
+
}
|
1707
|
+
|
1708
|
+
void spanmtq_add_term(Query *self, const char *term)
|
1709
|
+
{
|
1710
|
+
SpanMultiTermQuery *smtq = SpMTQ(self);
|
1711
|
+
if (smtq->term_cnt < smtq->term_capa) {
|
1712
|
+
smtq->terms[smtq->term_cnt++] = estrdup(term);
|
1713
|
+
}
|
1714
|
+
}
|
1715
|
+
|
1716
|
+
/*****************************************************************************
|
1717
|
+
*
|
1718
|
+
* SpanFirstQuery
|
1719
|
+
*
|
1720
|
+
*****************************************************************************/
|
1721
|
+
|
1722
|
+
static char *spanfq_to_s(Query *self, const char *field)
|
1723
|
+
{
|
1724
|
+
Query *match = SpFQ(self)->match;
|
1725
|
+
char *q_str = match->to_s(match, field);
|
1726
|
+
char *res = strfmt("span_first(%s, %d)", q_str, SpFQ(self)->end);
|
1727
|
+
free(q_str);
|
1728
|
+
return res;
|
1729
|
+
}
|
1730
|
+
|
1731
|
+
static void spanfq_extract_terms(Query *self, HashSet *terms)
|
1732
|
+
{
|
1733
|
+
SpFQ(self)->match->extract_terms(SpFQ(self)->match, terms);
|
1734
|
+
}
|
1735
|
+
|
1736
|
+
static HashSet *spanfq_get_terms(Query *self)
|
1737
|
+
{
|
1738
|
+
SpanFirstQuery *sfq = SpFQ(self);
|
1739
|
+
return SpQ(sfq->match)->get_terms(sfq->match);
|
1740
|
+
}
|
1741
|
+
|
1742
|
+
static Query *spanfq_rewrite(Query *self, IndexReader *ir)
|
1743
|
+
{
|
1744
|
+
Query *q, *rq;
|
1745
|
+
|
1746
|
+
q = SpFQ(self)->match;
|
1747
|
+
rq = q->rewrite(q, ir);
|
1748
|
+
q_deref(q);
|
1749
|
+
SpFQ(self)->match = rq;
|
1750
|
+
|
1751
|
+
self->ref_cnt++;
|
1752
|
+
return self; /* no clauses rewrote */
|
1753
|
+
}
|
1754
|
+
|
1755
|
+
static void spanfq_destroy_i(Query *self)
|
1756
|
+
{
|
1757
|
+
q_deref(SpFQ(self)->match);
|
1758
|
+
free(SpQ(self)->field);
|
1759
|
+
spanq_destroy_i(self);
|
1760
|
+
}
|
1761
|
+
|
1762
|
+
static unsigned long spanfq_hash(Query *self)
|
1763
|
+
{
|
1764
|
+
return spanq_hash(self) ^ SpFQ(self)->match->hash(SpFQ(self)->match)
|
1765
|
+
^ SpFQ(self)->end;
|
1766
|
+
}
|
1767
|
+
|
1768
|
+
static int spanfq_eq(Query *self, Query *o)
|
1769
|
+
{
|
1770
|
+
SpanFirstQuery *sfq1 = SpFQ(self);
|
1771
|
+
SpanFirstQuery *sfq2 = SpFQ(o);
|
1772
|
+
return spanq_eq(self, o) && sfq1->match->eq(sfq1->match, sfq2->match)
|
1773
|
+
&& (sfq1->end == sfq2->end);
|
1774
|
+
}
|
1775
|
+
|
1776
|
+
Query *spanfq_new_nr(Query *match, int end)
|
1777
|
+
{
|
1778
|
+
Query *self = q_new(SpanFirstQuery);
|
1779
|
+
|
1780
|
+
SpFQ(self)->match = match;
|
1781
|
+
SpFQ(self)->end = end;
|
1782
|
+
|
1783
|
+
SpQ(self)->field = estrdup(SpQ(match)->field);
|
1784
|
+
SpQ(self)->get_spans = &spanfe_new;
|
1785
|
+
SpQ(self)->get_terms = &spanfq_get_terms;
|
1786
|
+
|
1787
|
+
self->type = SPAN_FIRST_QUERY;
|
1788
|
+
self->rewrite = &spanfq_rewrite;
|
1789
|
+
self->extract_terms = &spanfq_extract_terms;
|
1790
|
+
self->to_s = &spanfq_to_s;
|
1791
|
+
self->hash = &spanfq_hash;
|
1792
|
+
self->eq = &spanfq_eq;
|
1793
|
+
self->destroy_i = &spanfq_destroy_i;
|
1794
|
+
self->create_weight_i = &spanw_new;
|
1795
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
1796
|
+
|
1797
|
+
return self;
|
1798
|
+
}
|
1799
|
+
|
1800
|
+
Query *spanfq_new(Query *match, int end)
|
1801
|
+
{
|
1802
|
+
REF(match);
|
1803
|
+
return spanfq_new_nr(match, end);
|
1804
|
+
}
|
1805
|
+
|
1806
|
+
/*****************************************************************************
|
1807
|
+
*
|
1808
|
+
* SpanOrQuery
|
1809
|
+
*
|
1810
|
+
*****************************************************************************/
|
1811
|
+
|
1812
|
+
static char *spanoq_to_s(Query *self, const char *field)
|
1813
|
+
{
|
1814
|
+
int i;
|
1815
|
+
SpanOrQuery *soq = SpOQ(self);
|
1816
|
+
char *res, *res_p;
|
1817
|
+
char **q_strs = ALLOC_N(char *, soq->c_cnt);
|
1818
|
+
int len = 50;
|
1819
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1820
|
+
Query *clause = soq->clauses[i];
|
1821
|
+
q_strs[i] = clause->to_s(clause, field);
|
1822
|
+
len += strlen(q_strs[i]) + 2;
|
1823
|
+
}
|
1824
|
+
|
1825
|
+
res_p = res = ALLOC_N(char, len);
|
1826
|
+
sprintf(res_p, "span_or[ ");
|
1827
|
+
res_p += strlen(res_p);
|
1828
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1829
|
+
sprintf(res_p, "%s, ", q_strs[i]);
|
1830
|
+
free(q_strs[i]);
|
1831
|
+
res_p += strlen(res_p);
|
1832
|
+
}
|
1833
|
+
free(q_strs);
|
1834
|
+
|
1835
|
+
sprintf(res_p - 2, " ]");
|
1836
|
+
return res;
|
1837
|
+
}
|
1838
|
+
|
1839
|
+
static void spanoq_extract_terms(Query *self, HashSet *terms)
|
1840
|
+
{
|
1841
|
+
SpanOrQuery *soq = SpOQ(self);
|
1842
|
+
int i;
|
1843
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1844
|
+
Query *clause = soq->clauses[i];
|
1845
|
+
clause->extract_terms(clause, terms);
|
1846
|
+
}
|
1847
|
+
}
|
1848
|
+
|
1849
|
+
static HashSet *spanoq_get_terms(Query *self)
|
1850
|
+
{
|
1851
|
+
SpanOrQuery *soq = SpOQ(self);
|
1852
|
+
HashSet *terms = hs_new_str(&free);
|
1853
|
+
int i;
|
1854
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1855
|
+
Query *clause = soq->clauses[i];
|
1856
|
+
HashSet *sub_terms = SpQ(clause)->get_terms(clause);
|
1857
|
+
hs_merge(terms, sub_terms);
|
1858
|
+
}
|
1859
|
+
|
1860
|
+
return terms;
|
1861
|
+
}
|
1862
|
+
|
1863
|
+
static SpanEnum *spanoq_get_spans(Query *self, IndexReader *ir)
|
1864
|
+
{
|
1865
|
+
SpanOrQuery *soq = SpOQ(self);
|
1866
|
+
if (soq->c_cnt == 1) {
|
1867
|
+
Query *q = soq->clauses[0];
|
1868
|
+
return SpQ(q)->get_spans(q, ir);
|
1869
|
+
}
|
1870
|
+
|
1871
|
+
return spanoe_new(self, ir);
|
1872
|
+
}
|
1873
|
+
|
1874
|
+
static Query *spanoq_rewrite(Query *self, IndexReader *ir)
|
1875
|
+
{
|
1876
|
+
SpanOrQuery *soq = SpOQ(self);
|
1877
|
+
int i;
|
1878
|
+
|
1879
|
+
/* replace clauses with their rewritten queries */
|
1880
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1881
|
+
Query *clause = soq->clauses[i];
|
1882
|
+
Query *rewritten = clause->rewrite(clause, ir);
|
1883
|
+
q_deref(clause);
|
1884
|
+
soq->clauses[i] = rewritten;
|
1885
|
+
}
|
1886
|
+
|
1887
|
+
self->ref_cnt++;
|
1888
|
+
return self;
|
1889
|
+
}
|
1890
|
+
|
1891
|
+
static void spanoq_destroy_i(Query *self)
|
1892
|
+
{
|
1893
|
+
SpanOrQuery *soq = SpOQ(self);
|
1894
|
+
|
1895
|
+
int i;
|
1896
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1897
|
+
Query *clause = soq->clauses[i];
|
1898
|
+
q_deref(clause);
|
1899
|
+
}
|
1900
|
+
free(soq->clauses);
|
1901
|
+
free(SpQ(self)->field);
|
1902
|
+
|
1903
|
+
spanq_destroy_i(self);
|
1904
|
+
}
|
1905
|
+
|
1906
|
+
static unsigned long spanoq_hash(Query *self)
|
1907
|
+
{
|
1908
|
+
int i;
|
1909
|
+
unsigned long hash = spanq_hash(self);
|
1910
|
+
SpanOrQuery *soq = SpOQ(self);
|
1911
|
+
|
1912
|
+
for (i = 0; i < soq->c_cnt; i++) {
|
1913
|
+
Query *q = soq->clauses[i];
|
1914
|
+
hash ^= q->hash(q);
|
1915
|
+
}
|
1916
|
+
return hash;
|
1917
|
+
}
|
1918
|
+
|
1919
|
+
static int spanoq_eq(Query *self, Query *o)
|
1920
|
+
{
|
1921
|
+
int i;
|
1922
|
+
Query *q1, *q2;
|
1923
|
+
SpanOrQuery *soq1 = SpOQ(self);
|
1924
|
+
SpanOrQuery *soq2 = SpOQ(o);
|
1925
|
+
|
1926
|
+
if (!spanq_eq(self, o) || soq1->c_cnt != soq2->c_cnt) {
|
1927
|
+
return false;
|
1928
|
+
}
|
1929
|
+
for (i = 0; i < soq1->c_cnt; i++) {
|
1930
|
+
q1 = soq1->clauses[i];
|
1931
|
+
q2 = soq2->clauses[i];
|
1932
|
+
if (!q1->eq(q1, q2)) {
|
1933
|
+
return false;
|
1934
|
+
}
|
1935
|
+
}
|
1936
|
+
return true;
|
1937
|
+
}
|
1938
|
+
|
1939
|
+
Query *spanoq_new()
|
1940
|
+
{
|
1941
|
+
Query *self = q_new(SpanOrQuery);
|
1942
|
+
SpOQ(self)->clauses = ALLOC_N(Query *, CLAUSE_INIT_CAPA);
|
1943
|
+
SpOQ(self)->c_capa = CLAUSE_INIT_CAPA;
|
1944
|
+
|
1945
|
+
SpQ(self)->field = estrdup((char *)EMPTY_STRING);
|
1946
|
+
SpQ(self)->get_spans = &spanoq_get_spans;
|
1947
|
+
SpQ(self)->get_terms = &spanoq_get_terms;
|
1948
|
+
|
1949
|
+
self->type = SPAN_OR_QUERY;
|
1950
|
+
self->rewrite = &spanoq_rewrite;
|
1951
|
+
self->extract_terms = &spanoq_extract_terms;
|
1952
|
+
self->to_s = &spanoq_to_s;
|
1953
|
+
self->hash = &spanoq_hash;
|
1954
|
+
self->eq = &spanoq_eq;
|
1955
|
+
self->destroy_i = &spanoq_destroy_i;
|
1956
|
+
self->create_weight_i = &spanw_new;
|
1957
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
1958
|
+
|
1959
|
+
return self;
|
1960
|
+
}
|
1961
|
+
|
1962
|
+
Query *spanoq_add_clause_nr(Query *self, Query *clause)
|
1963
|
+
{
|
1964
|
+
const int curr_index = SpOQ(self)->c_cnt++;
|
1965
|
+
if (clause->type < SPAN_TERM_QUERY || clause->type > SPAN_NEAR_QUERY) {
|
1966
|
+
RAISE(ARG_ERROR, "Tried to add a %s to a SpanOrQuery. This is not a "
|
1967
|
+
"SpanQuery.", q_get_query_name(clause->type));
|
1968
|
+
}
|
1969
|
+
if (curr_index == 0) {
|
1970
|
+
free(SpQ(self)->field);
|
1971
|
+
SpQ(self)->field = estrdup(SpQ(clause)->field);
|
1972
|
+
}
|
1973
|
+
else if (strcmp(SpQ(self)->field, SpQ(clause)->field) != 0) {
|
1974
|
+
RAISE(ARG_ERROR, "All clauses in a SpanQuery must have the same field. "
|
1975
|
+
"Attempted to add a SpanQuery with field \"%s\" to a SpanOrQuery "
|
1976
|
+
"with field \"%s\"", SpQ(clause)->field, SpQ(self)->field);
|
1977
|
+
}
|
1978
|
+
if (curr_index >= SpOQ(self)->c_capa) {
|
1979
|
+
SpOQ(self)->c_capa <<= 1;
|
1980
|
+
REALLOC_N(SpOQ(self)->clauses, Query *, SpOQ(self)->c_capa);
|
1981
|
+
}
|
1982
|
+
SpOQ(self)->clauses[curr_index] = clause;
|
1983
|
+
return clause;
|
1984
|
+
}
|
1985
|
+
|
1986
|
+
Query *spanoq_add_clause(Query *self, Query *clause)
|
1987
|
+
{
|
1988
|
+
REF(clause);
|
1989
|
+
return spanoq_add_clause_nr(self, clause);
|
1990
|
+
}
|
1991
|
+
|
1992
|
+
/*****************************************************************************
|
1993
|
+
*
|
1994
|
+
* SpanNearQuery
|
1995
|
+
*
|
1996
|
+
*****************************************************************************/
|
1997
|
+
|
1998
|
+
static char *spannq_to_s(Query *self, const char *field)
|
1999
|
+
{
|
2000
|
+
int i;
|
2001
|
+
SpanNearQuery *snq = SpNQ(self);
|
2002
|
+
char *res, *res_p;
|
2003
|
+
char **q_strs = ALLOC_N(char *, snq->c_cnt);
|
2004
|
+
int len = 50;
|
2005
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2006
|
+
Query *clause = snq->clauses[i];
|
2007
|
+
q_strs[i] = clause->to_s(clause, field);
|
2008
|
+
len += strlen(q_strs[i]);
|
2009
|
+
}
|
2010
|
+
|
2011
|
+
res_p = res = ALLOC_N(char, len);
|
2012
|
+
sprintf(res_p, "span_near[ ");
|
2013
|
+
res_p += strlen(res_p);
|
2014
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2015
|
+
sprintf(res_p, "%s, ", q_strs[i]);
|
2016
|
+
free(q_strs[i]);
|
2017
|
+
res_p += strlen(res_p);
|
2018
|
+
}
|
2019
|
+
free(q_strs);
|
2020
|
+
|
2021
|
+
sprintf(res_p - 2, " ]");
|
2022
|
+
return res;
|
2023
|
+
}
|
2024
|
+
|
2025
|
+
static void spannq_extract_terms(Query *self, HashSet *terms)
|
2026
|
+
{
|
2027
|
+
SpanNearQuery *snq = SpNQ(self);
|
2028
|
+
int i;
|
2029
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2030
|
+
Query *clause = snq->clauses[i];
|
2031
|
+
clause->extract_terms(clause, terms);
|
2032
|
+
}
|
2033
|
+
}
|
2034
|
+
|
2035
|
+
static HashSet *spannq_get_terms(Query *self)
|
2036
|
+
{
|
2037
|
+
SpanNearQuery *snq = SpNQ(self);
|
2038
|
+
HashSet *terms = hs_new_str(&free);
|
2039
|
+
int i;
|
2040
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2041
|
+
Query *clause = snq->clauses[i];
|
2042
|
+
HashSet *sub_terms = SpQ(clause)->get_terms(clause);
|
2043
|
+
hs_merge(terms, sub_terms);
|
2044
|
+
}
|
2045
|
+
|
2046
|
+
return terms;
|
2047
|
+
}
|
2048
|
+
|
2049
|
+
static SpanEnum *spannq_get_spans(Query *self, IndexReader *ir)
|
2050
|
+
{
|
2051
|
+
SpanNearQuery *snq = SpNQ(self);
|
2052
|
+
|
2053
|
+
if (snq->c_cnt == 1) {
|
2054
|
+
Query *q = snq->clauses[0];
|
2055
|
+
return SpQ(q)->get_spans(q, ir);
|
2056
|
+
}
|
2057
|
+
|
2058
|
+
return spanne_new(self, ir);
|
2059
|
+
}
|
2060
|
+
|
2061
|
+
static Query *spannq_rewrite(Query *self, IndexReader *ir)
|
2062
|
+
{
|
2063
|
+
SpanNearQuery *snq = SpNQ(self);
|
2064
|
+
int i;
|
2065
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2066
|
+
Query *clause = snq->clauses[i];
|
2067
|
+
Query *rewritten = clause->rewrite(clause, ir);
|
2068
|
+
q_deref(clause);
|
2069
|
+
snq->clauses[i] = rewritten;
|
2070
|
+
}
|
2071
|
+
|
2072
|
+
self->ref_cnt++;
|
2073
|
+
return self;
|
2074
|
+
}
|
2075
|
+
|
2076
|
+
static void spannq_destroy(Query *self)
|
2077
|
+
{
|
2078
|
+
SpanNearQuery *snq = SpNQ(self);
|
2079
|
+
|
2080
|
+
int i;
|
2081
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2082
|
+
Query *clause = snq->clauses[i];
|
2083
|
+
q_deref(clause);
|
2084
|
+
}
|
2085
|
+
free(snq->clauses);
|
2086
|
+
free(SpQ(self)->field);
|
2087
|
+
|
2088
|
+
spanq_destroy_i(self);
|
2089
|
+
}
|
2090
|
+
|
2091
|
+
static unsigned long spannq_hash(Query *self)
|
2092
|
+
{
|
2093
|
+
int i;
|
2094
|
+
unsigned long hash = spanq_hash(self);
|
2095
|
+
SpanNearQuery *snq = SpNQ(self);
|
2096
|
+
|
2097
|
+
for (i = 0; i < snq->c_cnt; i++) {
|
2098
|
+
Query *q = snq->clauses[i];
|
2099
|
+
hash ^= q->hash(q);
|
2100
|
+
}
|
2101
|
+
return ((hash ^ snq->slop) << 1) | snq->in_order;
|
2102
|
+
}
|
2103
|
+
|
2104
|
+
static int spannq_eq(Query *self, Query *o)
|
2105
|
+
{
|
2106
|
+
int i;
|
2107
|
+
Query *q1, *q2;
|
2108
|
+
SpanNearQuery *snq1 = SpNQ(self);
|
2109
|
+
SpanNearQuery *snq2 = SpNQ(o);
|
2110
|
+
if (! spanq_eq(self, o)
|
2111
|
+
|| (snq1->c_cnt != snq2->c_cnt)
|
2112
|
+
|| (snq1->slop != snq2->slop)
|
2113
|
+
|| (snq1->in_order != snq2->in_order)) {
|
2114
|
+
return false;
|
2115
|
+
}
|
2116
|
+
|
2117
|
+
for (i = 0; i < snq1->c_cnt; i++) {
|
2118
|
+
q1 = snq1->clauses[i];
|
2119
|
+
q2 = snq2->clauses[i];
|
2120
|
+
if (!q1->eq(q1, q2)) {
|
2121
|
+
return false;
|
2122
|
+
}
|
2123
|
+
}
|
2124
|
+
|
2125
|
+
return true;
|
2126
|
+
}
|
2127
|
+
|
2128
|
+
Query *spannq_new(int slop, bool in_order)
|
2129
|
+
{
|
2130
|
+
Query *self = q_new(SpanNearQuery);
|
2131
|
+
|
2132
|
+
SpNQ(self)->clauses = ALLOC_N(Query *, CLAUSE_INIT_CAPA);
|
2133
|
+
SpNQ(self)->c_capa = CLAUSE_INIT_CAPA;
|
2134
|
+
SpNQ(self)->slop = slop;
|
2135
|
+
SpNQ(self)->in_order = in_order;
|
2136
|
+
|
2137
|
+
SpQ(self)->get_spans = &spannq_get_spans;
|
2138
|
+
SpQ(self)->get_terms = &spannq_get_terms;
|
2139
|
+
SpQ(self)->field = estrdup((char *)EMPTY_STRING);
|
2140
|
+
|
2141
|
+
self->type = SPAN_NEAR_QUERY;
|
2142
|
+
self->rewrite = &spannq_rewrite;
|
2143
|
+
self->extract_terms = &spannq_extract_terms;
|
2144
|
+
self->to_s = &spannq_to_s;
|
2145
|
+
self->hash = &spannq_hash;
|
2146
|
+
self->eq = &spannq_eq;
|
2147
|
+
self->destroy_i = &spannq_destroy;
|
2148
|
+
self->create_weight_i = &spanw_new;
|
2149
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
2150
|
+
|
2151
|
+
return self;
|
2152
|
+
}
|
2153
|
+
|
2154
|
+
Query *spannq_add_clause_nr(Query *self, Query *clause)
|
2155
|
+
{
|
2156
|
+
const int curr_index = SpNQ(self)->c_cnt++;
|
2157
|
+
if (clause->type < SPAN_TERM_QUERY || clause->type > SPAN_NEAR_QUERY) {
|
2158
|
+
RAISE(ARG_ERROR, "Tried to add a %s to a SpanNearQuery. This is not a "
|
2159
|
+
"SpanQuery.", q_get_query_name(clause->type));
|
2160
|
+
}
|
2161
|
+
if (curr_index == 0) {
|
2162
|
+
free(SpQ(self)->field);
|
2163
|
+
SpQ(self)->field = estrdup(SpQ(clause)->field);
|
2164
|
+
}
|
2165
|
+
else if (strcmp(SpQ(self)->field, SpQ(clause)->field) != 0) {
|
2166
|
+
RAISE(ARG_ERROR, "All clauses in a SpanQuery must have the same field. "
|
2167
|
+
"Attempted to add a SpanQuery with field \"%s\" to SpanNearQuery "
|
2168
|
+
"with field \"%s\"", SpQ(clause)->field, SpQ(self)->field);
|
2169
|
+
}
|
2170
|
+
if (curr_index >= SpNQ(self)->c_capa) {
|
2171
|
+
SpNQ(self)->c_capa <<= 1;
|
2172
|
+
REALLOC_N(SpNQ(self)->clauses, Query *, SpNQ(self)->c_capa);
|
2173
|
+
}
|
2174
|
+
SpNQ(self)->clauses[curr_index] = clause;
|
2175
|
+
return clause;
|
2176
|
+
}
|
2177
|
+
|
2178
|
+
Query *spannq_add_clause(Query *self, Query *clause)
|
2179
|
+
{
|
2180
|
+
REF(clause);
|
2181
|
+
return spannq_add_clause_nr(self, clause);
|
2182
|
+
}
|
2183
|
+
|
2184
|
+
/*****************************************************************************
|
2185
|
+
*
|
2186
|
+
* SpanNotQuery
|
2187
|
+
*
|
2188
|
+
*****************************************************************************/
|
2189
|
+
|
2190
|
+
static char *spanxq_to_s(Query *self, const char *field)
|
2191
|
+
{
|
2192
|
+
SpanNotQuery *sxq = SpXQ(self);
|
2193
|
+
char *inc_s = sxq->inc->to_s(sxq->inc, field);
|
2194
|
+
char *exc_s = sxq->exc->to_s(sxq->exc, field);
|
2195
|
+
char *res = strfmt("span_not(inc:<%s>, exc:<%s>)", inc_s, exc_s);
|
2196
|
+
|
2197
|
+
free(inc_s);
|
2198
|
+
free(exc_s);
|
2199
|
+
return res;
|
2200
|
+
}
|
2201
|
+
|
2202
|
+
static void spanxq_extract_terms(Query *self, HashSet *terms)
|
2203
|
+
{
|
2204
|
+
SpXQ(self)->inc->extract_terms(SpXQ(self)->inc, terms);
|
2205
|
+
}
|
2206
|
+
|
2207
|
+
static HashSet *spanxq_get_terms(Query *self)
|
2208
|
+
{
|
2209
|
+
return SpQ(SpXQ(self)->inc)->get_terms(SpXQ(self)->inc);
|
2210
|
+
}
|
2211
|
+
|
2212
|
+
static Query *spanxq_rewrite(Query *self, IndexReader *ir)
|
2213
|
+
{
|
2214
|
+
SpanNotQuery *sxq = SpXQ(self);
|
2215
|
+
Query *q, *rq;
|
2216
|
+
|
2217
|
+
/* rewrite inclusive query */
|
2218
|
+
q = sxq->inc;
|
2219
|
+
rq = q->rewrite(q, ir);
|
2220
|
+
q_deref(q);
|
2221
|
+
sxq->inc = rq;
|
2222
|
+
|
2223
|
+
/* rewrite exclusive query */
|
2224
|
+
q = sxq->exc;
|
2225
|
+
rq = q->rewrite(q, ir);
|
2226
|
+
q_deref(q);
|
2227
|
+
sxq->exc = rq;
|
2228
|
+
|
2229
|
+
self->ref_cnt++;
|
2230
|
+
return self;
|
2231
|
+
}
|
2232
|
+
|
2233
|
+
static void spanxq_destroy(Query *self)
|
2234
|
+
{
|
2235
|
+
SpanNotQuery *sxq = SpXQ(self);
|
2236
|
+
|
2237
|
+
q_deref(sxq->inc);
|
2238
|
+
q_deref(sxq->exc);
|
2239
|
+
|
2240
|
+
free(SpQ(self)->field);
|
2241
|
+
|
2242
|
+
spanq_destroy_i(self);
|
2243
|
+
}
|
2244
|
+
|
2245
|
+
static unsigned long spanxq_hash(Query *self)
|
2246
|
+
{
|
2247
|
+
SpanNotQuery *sxq = SpXQ(self);
|
2248
|
+
return spanq_hash(self) ^ sxq->inc->hash(sxq->inc)
|
2249
|
+
^ sxq->exc->hash(sxq->exc);
|
2250
|
+
}
|
2251
|
+
|
2252
|
+
static int spanxq_eq(Query *self, Query *o)
|
2253
|
+
{
|
2254
|
+
SpanNotQuery *sxq1 = SpXQ(self);
|
2255
|
+
SpanNotQuery *sxq2 = SpXQ(o);
|
2256
|
+
return spanq_eq(self, o) && sxq1->inc->eq(sxq1->inc, sxq2->inc)
|
2257
|
+
&& sxq1->exc->eq(sxq1->exc, sxq2->exc);
|
2258
|
+
}
|
2259
|
+
|
2260
|
+
|
2261
|
+
Query *spanxq_new_nr(Query *inc, Query *exc)
|
2262
|
+
{
|
2263
|
+
Query *self;
|
2264
|
+
if (strcmp(SpQ(inc)->field, SpQ(inc)->field) != 0) {
|
2265
|
+
RAISE(ARG_ERROR, "All clauses in a SpanQuery must have the same field. "
|
2266
|
+
"Attempted to add a SpanQuery with field \"%s\" along with a "
|
2267
|
+
"SpanQuery with field \"%s\" to an SpanNotQuery",
|
2268
|
+
SpQ(inc)->field, SpQ(exc)->field);
|
2269
|
+
}
|
2270
|
+
self = q_new(SpanNotQuery);
|
2271
|
+
|
2272
|
+
SpXQ(self)->inc = inc;
|
2273
|
+
SpXQ(self)->exc = exc;
|
2274
|
+
|
2275
|
+
SpQ(self)->field = estrdup(SpQ(inc)->field);
|
2276
|
+
SpQ(self)->get_spans = &spanxe_new;
|
2277
|
+
SpQ(self)->get_terms = &spanxq_get_terms;
|
2278
|
+
|
2279
|
+
self->type = SPAN_NOT_QUERY;
|
2280
|
+
self->rewrite = &spanxq_rewrite;
|
2281
|
+
self->extract_terms = &spanxq_extract_terms;
|
2282
|
+
self->to_s = &spanxq_to_s;
|
2283
|
+
self->hash = &spanxq_hash;
|
2284
|
+
self->eq = &spanxq_eq;
|
2285
|
+
self->destroy_i = &spanxq_destroy;
|
2286
|
+
self->create_weight_i = &spanw_new;
|
2287
|
+
self->get_matchv_i = &spanq_get_matchv_i;
|
2288
|
+
|
2289
|
+
return self;
|
2290
|
+
}
|
2291
|
+
|
2292
|
+
Query *spanxq_new(Query *inc, Query *exc)
|
2293
|
+
{
|
2294
|
+
REF(inc);
|
2295
|
+
REF(exc);
|
2296
|
+
return spanxq_new_nr(inc, exc);
|
2297
|
+
}
|
2298
|
+
|
2299
|
+
|
2300
|
+
/*****************************************************************************
|
2301
|
+
*
|
2302
|
+
* Rewritables
|
2303
|
+
*
|
2304
|
+
*****************************************************************************/
|
2305
|
+
|
2306
|
+
/*****************************************************************************
|
2307
|
+
*
|
2308
|
+
* SpanPrefixQuery
|
2309
|
+
*
|
2310
|
+
*****************************************************************************/
|
2311
|
+
|
2312
|
+
#define SpPfxQ(query) ((SpanPrefixQuery *)(query))
|
2313
|
+
|
2314
|
+
static char *spanprq_to_s(Query *self, const char *current_field)
|
2315
|
+
{
|
2316
|
+
char *buffer, *bptr;
|
2317
|
+
const char *prefix = SpPfxQ(self)->prefix;
|
2318
|
+
const char *field = SpQ(self)->field;
|
2319
|
+
size_t plen = strlen(prefix);
|
2320
|
+
size_t flen = strlen(field);
|
2321
|
+
|
2322
|
+
bptr = buffer = ALLOC_N(char, plen + flen + 35);
|
2323
|
+
|
2324
|
+
if (strcmp(field, current_field) != 0) {
|
2325
|
+
sprintf(bptr, "%s:", field);
|
2326
|
+
bptr += flen + 1;
|
2327
|
+
}
|
2328
|
+
|
2329
|
+
sprintf(bptr, "%s*", prefix);
|
2330
|
+
bptr += plen + 1;
|
2331
|
+
if (self->boost != 1.0) {
|
2332
|
+
*bptr = '^';
|
2333
|
+
dbl_to_s(++bptr, self->boost);
|
2334
|
+
}
|
2335
|
+
|
2336
|
+
return buffer;
|
2337
|
+
}
|
2338
|
+
|
2339
|
+
static Query *spanprq_rewrite(Query *self, IndexReader *ir)
|
2340
|
+
{
|
2341
|
+
const char *field = SpQ(self)->field;
|
2342
|
+
const int field_num = fis_get_field_num(ir->fis, field);
|
2343
|
+
Query *volatile q = spanmtq_new_conf(field, SpPfxQ(self)->max_terms);
|
2344
|
+
q->boost = self->boost; /* set the boost */
|
2345
|
+
|
2346
|
+
if (field_num >= 0) {
|
2347
|
+
const char *prefix = SpPfxQ(self)->prefix;
|
2348
|
+
TermEnum *te = ir->terms_from(ir, field_num, prefix);
|
2349
|
+
const char *term = te->curr_term;
|
2350
|
+
size_t prefix_len = strlen(prefix);
|
2351
|
+
|
2352
|
+
TRY
|
2353
|
+
do {
|
2354
|
+
if (strncmp(term, prefix, prefix_len) != 0) {
|
2355
|
+
break;
|
2356
|
+
}
|
2357
|
+
spanmtq_add_term(q, term); /* found a match */
|
2358
|
+
} while (te->next(te));
|
2359
|
+
XFINALLY
|
2360
|
+
te->close(te);
|
2361
|
+
XENDTRY
|
2362
|
+
}
|
2363
|
+
|
2364
|
+
return q;
|
2365
|
+
}
|
2366
|
+
|
2367
|
+
static void spanprq_destroy(Query *self)
|
2368
|
+
{
|
2369
|
+
free(SpQ(self)->field);
|
2370
|
+
free(SpPfxQ(self)->prefix);
|
2371
|
+
spanq_destroy_i(self);
|
2372
|
+
}
|
2373
|
+
|
2374
|
+
static unsigned long spanprq_hash(Query *self)
|
2375
|
+
{
|
2376
|
+
return str_hash(SpQ(self)->field) ^ str_hash(SpPfxQ(self)->prefix);
|
2377
|
+
}
|
2378
|
+
|
2379
|
+
static int spanprq_eq(Query *self, Query *o)
|
2380
|
+
{
|
2381
|
+
return (strcmp(SpPfxQ(self)->prefix, SpPfxQ(o)->prefix) == 0)
|
2382
|
+
&& (strcmp(SpQ(self)->field, SpQ(o)->field) == 0);
|
2383
|
+
}
|
2384
|
+
|
2385
|
+
Query *spanprq_new(const char *field, const char *prefix)
|
2386
|
+
{
|
2387
|
+
Query *self = q_new(SpanPrefixQuery);
|
2388
|
+
|
2389
|
+
SpQ(self)->field = estrdup(field);
|
2390
|
+
SpPfxQ(self)->prefix = estrdup(prefix);
|
2391
|
+
SpPfxQ(self)->max_terms = SPAN_PREFIX_QUERY_MAX_TERMS;
|
2392
|
+
|
2393
|
+
self->type = SPAN_PREFIX_QUERY;
|
2394
|
+
self->rewrite = &spanprq_rewrite;
|
2395
|
+
self->to_s = &spanprq_to_s;
|
2396
|
+
self->hash = &spanprq_hash;
|
2397
|
+
self->eq = &spanprq_eq;
|
2398
|
+
self->destroy_i = &spanprq_destroy;
|
2399
|
+
self->create_weight_i = &q_create_weight_unsup;
|
2400
|
+
|
2401
|
+
return self;
|
2402
|
+
}
|