isomorfeus-ferret 0.12.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/LICENSE +612 -0
- data/README.md +44 -0
- data/ext/isomorfeus_ferret_ext/benchmark.c +223 -0
- data/ext/isomorfeus_ferret_ext/benchmark.h +45 -0
- data/ext/isomorfeus_ferret_ext/benchmarks_all.h +25 -0
- data/ext/isomorfeus_ferret_ext/bm_bitvector.c +123 -0
- data/ext/isomorfeus_ferret_ext/bm_hash.c +118 -0
- data/ext/isomorfeus_ferret_ext/bm_micro_string.c +40 -0
- data/ext/isomorfeus_ferret_ext/bm_store.c +93 -0
- data/ext/isomorfeus_ferret_ext/email.rl +21 -0
- data/ext/isomorfeus_ferret_ext/extconf.rb +5 -0
- data/ext/isomorfeus_ferret_ext/fio_tmpfile.h +53 -0
- data/ext/isomorfeus_ferret_ext/frb_analysis.c +2577 -0
- data/ext/isomorfeus_ferret_ext/frb_index.c +3457 -0
- data/ext/isomorfeus_ferret_ext/frb_lang.c +9 -0
- data/ext/isomorfeus_ferret_ext/frb_lang.h +17 -0
- data/ext/isomorfeus_ferret_ext/frb_qparser.c +629 -0
- data/ext/isomorfeus_ferret_ext/frb_search.c +4460 -0
- data/ext/isomorfeus_ferret_ext/frb_store.c +515 -0
- data/ext/isomorfeus_ferret_ext/frb_threading.h +30 -0
- data/ext/isomorfeus_ferret_ext/frb_utils.c +1127 -0
- data/ext/isomorfeus_ferret_ext/frt_analysis.c +1644 -0
- data/ext/isomorfeus_ferret_ext/frt_analysis.h +247 -0
- data/ext/isomorfeus_ferret_ext/frt_array.c +124 -0
- data/ext/isomorfeus_ferret_ext/frt_array.h +54 -0
- data/ext/isomorfeus_ferret_ext/frt_bitvector.c +95 -0
- data/ext/isomorfeus_ferret_ext/frt_bitvector.h +586 -0
- data/ext/isomorfeus_ferret_ext/frt_compound_io.c +374 -0
- data/ext/isomorfeus_ferret_ext/frt_config.h +44 -0
- data/ext/isomorfeus_ferret_ext/frt_document.c +134 -0
- data/ext/isomorfeus_ferret_ext/frt_document.h +52 -0
- data/ext/isomorfeus_ferret_ext/frt_except.c +95 -0
- data/ext/isomorfeus_ferret_ext/frt_except.h +188 -0
- data/ext/isomorfeus_ferret_ext/frt_field_index.c +233 -0
- data/ext/isomorfeus_ferret_ext/frt_field_index.h +42 -0
- data/ext/isomorfeus_ferret_ext/frt_filter.c +157 -0
- data/ext/isomorfeus_ferret_ext/frt_fs_store.c +502 -0
- data/ext/isomorfeus_ferret_ext/frt_global.c +427 -0
- data/ext/isomorfeus_ferret_ext/frt_global.h +290 -0
- data/ext/isomorfeus_ferret_ext/frt_hash.c +518 -0
- data/ext/isomorfeus_ferret_ext/frt_hash.h +466 -0
- data/ext/isomorfeus_ferret_ext/frt_hashset.c +191 -0
- data/ext/isomorfeus_ferret_ext/frt_hashset.h +206 -0
- data/ext/isomorfeus_ferret_ext/frt_helper.c +62 -0
- data/ext/isomorfeus_ferret_ext/frt_helper.h +13 -0
- data/ext/isomorfeus_ferret_ext/frt_ind.c +353 -0
- data/ext/isomorfeus_ferret_ext/frt_ind.h +54 -0
- data/ext/isomorfeus_ferret_ext/frt_index.c +6377 -0
- data/ext/isomorfeus_ferret_ext/frt_index.h +880 -0
- data/ext/isomorfeus_ferret_ext/frt_lang.c +104 -0
- data/ext/isomorfeus_ferret_ext/frt_lang.h +44 -0
- data/ext/isomorfeus_ferret_ext/frt_mempool.c +87 -0
- data/ext/isomorfeus_ferret_ext/frt_mempool.h +33 -0
- data/ext/isomorfeus_ferret_ext/frt_multimapper.c +349 -0
- data/ext/isomorfeus_ferret_ext/frt_multimapper.h +52 -0
- data/ext/isomorfeus_ferret_ext/frt_posh.c +1006 -0
- data/ext/isomorfeus_ferret_ext/frt_posh.h +973 -0
- data/ext/isomorfeus_ferret_ext/frt_priorityqueue.c +147 -0
- data/ext/isomorfeus_ferret_ext/frt_priorityqueue.h +147 -0
- data/ext/isomorfeus_ferret_ext/frt_q_boolean.c +1612 -0
- data/ext/isomorfeus_ferret_ext/frt_q_const_score.c +157 -0
- data/ext/isomorfeus_ferret_ext/frt_q_filtered_query.c +209 -0
- data/ext/isomorfeus_ferret_ext/frt_q_fuzzy.c +281 -0
- data/ext/isomorfeus_ferret_ext/frt_q_match_all.c +147 -0
- data/ext/isomorfeus_ferret_ext/frt_q_multi_term.c +672 -0
- data/ext/isomorfeus_ferret_ext/frt_q_parser.c +3084 -0
- data/ext/isomorfeus_ferret_ext/frt_q_phrase.c +1182 -0
- data/ext/isomorfeus_ferret_ext/frt_q_prefix.c +98 -0
- data/ext/isomorfeus_ferret_ext/frt_q_range.c +665 -0
- data/ext/isomorfeus_ferret_ext/frt_q_span.c +2386 -0
- data/ext/isomorfeus_ferret_ext/frt_q_term.c +311 -0
- data/ext/isomorfeus_ferret_ext/frt_q_wildcard.c +166 -0
- data/ext/isomorfeus_ferret_ext/frt_ram_store.c +460 -0
- data/ext/isomorfeus_ferret_ext/frt_scanner.c +899 -0
- data/ext/isomorfeus_ferret_ext/frt_scanner.h +28 -0
- data/ext/isomorfeus_ferret_ext/frt_scanner_mb.c +6705 -0
- data/ext/isomorfeus_ferret_ext/frt_scanner_utf8.c +4419 -0
- data/ext/isomorfeus_ferret_ext/frt_search.c +1824 -0
- data/ext/isomorfeus_ferret_ext/frt_search.h +924 -0
- data/ext/isomorfeus_ferret_ext/frt_similarity.c +150 -0
- data/ext/isomorfeus_ferret_ext/frt_similarity.h +79 -0
- data/ext/isomorfeus_ferret_ext/frt_sort.c +796 -0
- data/ext/isomorfeus_ferret_ext/frt_stopwords.c +395 -0
- data/ext/isomorfeus_ferret_ext/frt_store.c +680 -0
- data/ext/isomorfeus_ferret_ext/frt_store.h +789 -0
- data/ext/isomorfeus_ferret_ext/frt_term_vectors.c +72 -0
- data/ext/isomorfeus_ferret_ext/frt_threading.h +23 -0
- data/ext/isomorfeus_ferret_ext/frt_win32.h +54 -0
- data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.c +409 -0
- data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.h +95 -0
- data/ext/isomorfeus_ferret_ext/libstemmer.c +93 -0
- data/ext/isomorfeus_ferret_ext/libstemmer.h +73 -0
- data/ext/isomorfeus_ferret_ext/q_parser.y +1366 -0
- data/ext/isomorfeus_ferret_ext/scanner.h +28 -0
- data/ext/isomorfeus_ferret_ext/scanner.in +43 -0
- data/ext/isomorfeus_ferret_ext/scanner.rl +84 -0
- data/ext/isomorfeus_ferret_ext/scanner_mb.rl +200 -0
- data/ext/isomorfeus_ferret_ext/scanner_utf8.rl +85 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_danish.c +324 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_danish.h +7 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_dutch.c +610 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_dutch.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_english.c +1104 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_english.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_finnish.c +749 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_finnish.h +7 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_french.c +1233 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_french.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_german.c +490 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_german.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_hungarian.c +1217 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_hungarian.h +7 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_italian.c +1052 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_italian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_norwegian.c +283 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_norwegian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_porter.c +735 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_porter.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_portuguese.c +1003 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_portuguese.h +7 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_spanish.c +1079 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_spanish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_swedish.c +293 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_1_swedish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_romanian.c +984 -0
- data/ext/isomorfeus_ferret_ext/stem_ISO_8859_2_romanian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_KOI8_R_russian.c +686 -0
- data/ext/isomorfeus_ferret_ext/stem_KOI8_R_russian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_danish.c +325 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_danish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_dutch.c +620 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_dutch.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_english.c +1111 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_english.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_finnish.c +754 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_finnish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_french.c +1242 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_french.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_german.c +495 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_german.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_hungarian.c +1220 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_hungarian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_italian.c +1059 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_italian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_norwegian.c +285 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_norwegian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_porter.c +741 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_porter.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_portuguese.c +1009 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_portuguese.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_romanian.c +990 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_romanian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_russian.c +680 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_russian.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_spanish.c +1083 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_spanish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_swedish.c +294 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_swedish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_turkish.c +2191 -0
- data/ext/isomorfeus_ferret_ext/stem_UTF_8_turkish.h +6 -0
- data/ext/isomorfeus_ferret_ext/stem_api.c +66 -0
- data/ext/isomorfeus_ferret_ext/stem_api.h +26 -0
- data/ext/isomorfeus_ferret_ext/stem_header.h +57 -0
- data/ext/isomorfeus_ferret_ext/stem_modules.h +190 -0
- data/ext/isomorfeus_ferret_ext/stem_modules.txt +50 -0
- data/ext/isomorfeus_ferret_ext/stem_utilities.c +478 -0
- data/ext/isomorfeus_ferret_ext/test.c +850 -0
- data/ext/isomorfeus_ferret_ext/test.h +416 -0
- data/ext/isomorfeus_ferret_ext/test_1710.c +63 -0
- data/ext/isomorfeus_ferret_ext/test_analysis.c +1221 -0
- data/ext/isomorfeus_ferret_ext/test_array.c +272 -0
- data/ext/isomorfeus_ferret_ext/test_bitvector.c +600 -0
- data/ext/isomorfeus_ferret_ext/test_compound_io.c +170 -0
- data/ext/isomorfeus_ferret_ext/test_document.c +156 -0
- data/ext/isomorfeus_ferret_ext/test_except.c +244 -0
- data/ext/isomorfeus_ferret_ext/test_fields.c +522 -0
- data/ext/isomorfeus_ferret_ext/test_file_deleter.c +185 -0
- data/ext/isomorfeus_ferret_ext/test_filter.c +331 -0
- data/ext/isomorfeus_ferret_ext/test_fs_store.c +25 -0
- data/ext/isomorfeus_ferret_ext/test_global.c +299 -0
- data/ext/isomorfeus_ferret_ext/test_hash.c +485 -0
- data/ext/isomorfeus_ferret_ext/test_hashset.c +288 -0
- data/ext/isomorfeus_ferret_ext/test_helper.c +47 -0
- data/ext/isomorfeus_ferret_ext/test_highlighter.c +548 -0
- data/ext/isomorfeus_ferret_ext/test_index.c +2323 -0
- data/ext/isomorfeus_ferret_ext/test_lang.c +74 -0
- data/ext/isomorfeus_ferret_ext/test_mempool.c +102 -0
- data/ext/isomorfeus_ferret_ext/test_multimapper.c +64 -0
- data/ext/isomorfeus_ferret_ext/test_priorityqueue.c +213 -0
- data/ext/isomorfeus_ferret_ext/test_q_const_score.c +84 -0
- data/ext/isomorfeus_ferret_ext/test_q_filtered.c +61 -0
- data/ext/isomorfeus_ferret_ext/test_q_fuzzy.c +241 -0
- data/ext/isomorfeus_ferret_ext/test_q_parser.c +464 -0
- data/ext/isomorfeus_ferret_ext/test_q_span.c +575 -0
- data/ext/isomorfeus_ferret_ext/test_ram_store.c +77 -0
- data/ext/isomorfeus_ferret_ext/test_search.c +1874 -0
- data/ext/isomorfeus_ferret_ext/test_segments.c +167 -0
- data/ext/isomorfeus_ferret_ext/test_similarity.c +25 -0
- data/ext/isomorfeus_ferret_ext/test_sort.c +333 -0
- data/ext/isomorfeus_ferret_ext/test_store.c +591 -0
- data/ext/isomorfeus_ferret_ext/test_store.h +3 -0
- data/ext/isomorfeus_ferret_ext/test_term.c +351 -0
- data/ext/isomorfeus_ferret_ext/test_term_vectors.c +373 -0
- data/ext/isomorfeus_ferret_ext/test_test.c +83 -0
- data/ext/isomorfeus_ferret_ext/test_threading.c +188 -0
- data/ext/isomorfeus_ferret_ext/testhelper.c +561 -0
- data/ext/isomorfeus_ferret_ext/testhelper.h +25 -0
- data/ext/isomorfeus_ferret_ext/tests_all.h +87 -0
- data/ext/isomorfeus_ferret_ext/uchar-ucs4.rl +1854 -0
- data/ext/isomorfeus_ferret_ext/uchar-utf8.rl +1999 -0
- data/ext/isomorfeus_ferret_ext/url.rl +27 -0
- data/ext/isomorfeus_ferret_ext/word_list.h +15156 -0
- data/lib/isomorfeus/ferret/document.rb +132 -0
- data/lib/isomorfeus/ferret/field_symbol.rb +85 -0
- data/lib/isomorfeus/ferret/index/field_infos.rb +48 -0
- data/lib/isomorfeus/ferret/index/index.rb +970 -0
- data/lib/isomorfeus/ferret/monitor.rb +323 -0
- data/lib/isomorfeus/ferret/stdlib_patches.rb +151 -0
- data/lib/isomorfeus/ferret/version.rb +5 -0
- data/lib/isomorfeus-ferret.rb +8 -0
- metadata +307 -0
@@ -0,0 +1,796 @@
|
|
1
|
+
#include <string.h>
|
2
|
+
#include "frt_search.h"
|
3
|
+
#include "frt_index.h"
|
4
|
+
#include "frt_field_index.h"
|
5
|
+
|
6
|
+
/***************************************************************************
|
7
|
+
*
|
8
|
+
* FrtSortField
|
9
|
+
*
|
10
|
+
***************************************************************************/
|
11
|
+
|
12
|
+
static FrtSortField *sort_field_alloc(FrtSymbol field,
|
13
|
+
SortType type,
|
14
|
+
bool reverse,
|
15
|
+
int (*compare)(void *index_ptr, FrtHit *hit1, FrtHit *hit2),
|
16
|
+
void (*get_val)(void *index_ptr, FrtHit *hit1, FrtComparable *comparable),
|
17
|
+
const FrtFieldIndexClass *field_index_class)
|
18
|
+
{
|
19
|
+
FrtSortField *self = FRT_ALLOC(FrtSortField);
|
20
|
+
self->field = field;
|
21
|
+
self->type = type;
|
22
|
+
self->reverse = reverse;
|
23
|
+
self->field_index_class = field_index_class;
|
24
|
+
self->compare = compare;
|
25
|
+
self->get_val = get_val;
|
26
|
+
return self;
|
27
|
+
}
|
28
|
+
|
29
|
+
FrtSortField *frt_sort_field_new(FrtSymbol field, SortType type, bool reverse)
|
30
|
+
{
|
31
|
+
FrtSortField *sf = NULL;
|
32
|
+
switch (type) {
|
33
|
+
case FRT_SORT_TYPE_SCORE:
|
34
|
+
sf = frt_sort_field_score_new(reverse);
|
35
|
+
break;
|
36
|
+
case FRT_SORT_TYPE_DOC:
|
37
|
+
sf = frt_sort_field_doc_new(reverse);
|
38
|
+
break;
|
39
|
+
case FRT_SORT_TYPE_BYTE:
|
40
|
+
sf = frt_sort_field_byte_new(field, reverse);
|
41
|
+
break;
|
42
|
+
case FRT_SORT_TYPE_INTEGER:
|
43
|
+
sf = frt_sort_field_int_new(field, reverse);
|
44
|
+
break;
|
45
|
+
case FRT_SORT_TYPE_FLOAT:
|
46
|
+
sf = frt_sort_field_float_new(field, reverse);
|
47
|
+
break;
|
48
|
+
case FRT_SORT_TYPE_STRING:
|
49
|
+
sf = frt_sort_field_string_new(field, reverse);
|
50
|
+
break;
|
51
|
+
case FRT_SORT_TYPE_AUTO:
|
52
|
+
sf = frt_sort_field_auto_new(field, reverse);
|
53
|
+
break;
|
54
|
+
}
|
55
|
+
return sf;
|
56
|
+
}
|
57
|
+
|
58
|
+
void frt_sort_field_destroy(void *p)
|
59
|
+
{
|
60
|
+
free(p);
|
61
|
+
}
|
62
|
+
|
63
|
+
/*
|
64
|
+
* field:<type>!
|
65
|
+
*/
|
66
|
+
char *frt_sort_field_to_s(FrtSortField *self)
|
67
|
+
{
|
68
|
+
char *str;
|
69
|
+
const char *type = NULL;
|
70
|
+
switch (self->type) {
|
71
|
+
case FRT_SORT_TYPE_SCORE:
|
72
|
+
type = "<SCORE>";
|
73
|
+
break;
|
74
|
+
case FRT_SORT_TYPE_DOC:
|
75
|
+
type = "<DOC>";
|
76
|
+
break;
|
77
|
+
case FRT_SORT_TYPE_BYTE:
|
78
|
+
type = "<byte>";
|
79
|
+
break;
|
80
|
+
case FRT_SORT_TYPE_INTEGER:
|
81
|
+
type = "<integer>";
|
82
|
+
break;
|
83
|
+
case FRT_SORT_TYPE_FLOAT:
|
84
|
+
type = "<float>";
|
85
|
+
break;
|
86
|
+
case FRT_SORT_TYPE_STRING:
|
87
|
+
type = "<string>";
|
88
|
+
break;
|
89
|
+
case FRT_SORT_TYPE_AUTO:
|
90
|
+
type = "<auto>";
|
91
|
+
break;
|
92
|
+
}
|
93
|
+
if (self->field) {
|
94
|
+
const char *field_name = rb_id2name(self->field);
|
95
|
+
str = FRT_ALLOC_N(char, 3 + strlen(field_name) + strlen(type));
|
96
|
+
sprintf(str, "%s:%s%s", field_name, type, (self->reverse ? "!" : ""));
|
97
|
+
}
|
98
|
+
else {
|
99
|
+
str = FRT_ALLOC_N(char, 2 + strlen(type));
|
100
|
+
sprintf(str, "%s%s", type, (self->reverse ? "!" : ""));
|
101
|
+
}
|
102
|
+
return str;
|
103
|
+
}
|
104
|
+
|
105
|
+
/***************************************************************************
|
106
|
+
* ScoreSortField
|
107
|
+
***************************************************************************/
|
108
|
+
|
109
|
+
static void sf_score_get_val(void *index, FrtHit *hit, FrtComparable *comparable)
|
110
|
+
{
|
111
|
+
(void)index;
|
112
|
+
comparable->val.f = hit->score;
|
113
|
+
}
|
114
|
+
|
115
|
+
static int sf_score_compare(void *index_ptr, FrtHit *hit2, FrtHit *hit1)
|
116
|
+
{
|
117
|
+
float val1 = hit1->score;
|
118
|
+
float val2 = hit2->score;
|
119
|
+
(void)index_ptr;
|
120
|
+
|
121
|
+
if (val1 > val2) return 1;
|
122
|
+
else if (val1 < val2) return -1;
|
123
|
+
else return 0;
|
124
|
+
}
|
125
|
+
|
126
|
+
FrtSortField *frt_sort_field_score_new(bool reverse)
|
127
|
+
{
|
128
|
+
return sort_field_alloc((ID)NULL, FRT_SORT_TYPE_SCORE, reverse, &sf_score_compare, &sf_score_get_val, NULL);
|
129
|
+
}
|
130
|
+
|
131
|
+
const FrtSortField FRT_SORT_FIELD_SCORE = {
|
132
|
+
NULL, /* field_index_class */
|
133
|
+
(ID)NULL, /* field */
|
134
|
+
FRT_SORT_TYPE_SCORE, /* type */
|
135
|
+
false, /* reverse */
|
136
|
+
&sf_score_compare, /* compare */
|
137
|
+
&sf_score_get_val, /* get_val */
|
138
|
+
};
|
139
|
+
|
140
|
+
const FrtSortField FRT_SORT_FIELD_SCORE_REV = {
|
141
|
+
NULL, /* field_index_class */
|
142
|
+
(ID)NULL, /* field */
|
143
|
+
FRT_SORT_TYPE_SCORE, /* type */
|
144
|
+
true, /* reverse */
|
145
|
+
&sf_score_compare, /* compare */
|
146
|
+
&sf_score_get_val, /* get_val */
|
147
|
+
};
|
148
|
+
|
149
|
+
/**************************************************************************
|
150
|
+
* DocSortField
|
151
|
+
***************************************************************************/
|
152
|
+
|
153
|
+
static void sf_doc_get_val(void *index, FrtHit *hit, FrtComparable *comparable)
|
154
|
+
{
|
155
|
+
(void)index;
|
156
|
+
comparable->val.l = hit->doc;
|
157
|
+
}
|
158
|
+
|
159
|
+
static int sf_doc_compare(void *index_ptr, FrtHit *hit1, FrtHit *hit2)
|
160
|
+
{
|
161
|
+
int val1 = hit1->doc;
|
162
|
+
int val2 = hit2->doc;
|
163
|
+
(void)index_ptr;
|
164
|
+
|
165
|
+
if (val1 > val2) return 1;
|
166
|
+
else if (val1 < val2) return -1;
|
167
|
+
else return 0;
|
168
|
+
}
|
169
|
+
|
170
|
+
FrtSortField *frt_sort_field_doc_new(bool reverse)
|
171
|
+
{
|
172
|
+
return sort_field_alloc((ID)NULL, FRT_SORT_TYPE_DOC, reverse,
|
173
|
+
&sf_doc_compare, &sf_doc_get_val, NULL);
|
174
|
+
}
|
175
|
+
|
176
|
+
const FrtSortField FRT_SORT_FIELD_DOC = {
|
177
|
+
NULL, /* field_index_class */
|
178
|
+
(ID)NULL, /* field */
|
179
|
+
FRT_SORT_TYPE_DOC, /* type */
|
180
|
+
false, /* reverse */
|
181
|
+
&sf_doc_compare, /* compare */
|
182
|
+
&sf_doc_get_val, /* get_val */
|
183
|
+
};
|
184
|
+
|
185
|
+
const FrtSortField FRT_SORT_FIELD_DOC_REV = {
|
186
|
+
NULL, /* field_index_class */
|
187
|
+
(ID)NULL, /* field */
|
188
|
+
FRT_SORT_TYPE_DOC, /* type */
|
189
|
+
true, /* reverse */
|
190
|
+
&sf_doc_compare, /* compare */
|
191
|
+
&sf_doc_get_val, /* get_val */
|
192
|
+
};
|
193
|
+
|
194
|
+
/***************************************************************************
|
195
|
+
* ByteSortField
|
196
|
+
***************************************************************************/
|
197
|
+
|
198
|
+
static void sf_byte_get_val(void *index, FrtHit *hit, FrtComparable *comparable)
|
199
|
+
{
|
200
|
+
comparable->val.l = ((long *)index)[hit->doc];
|
201
|
+
}
|
202
|
+
|
203
|
+
static int sf_byte_compare(void *index, FrtHit *hit1, FrtHit *hit2)
|
204
|
+
{
|
205
|
+
long val1 = ((long *)index)[hit1->doc];
|
206
|
+
long val2 = ((long *)index)[hit2->doc];
|
207
|
+
if (val1 > val2) return 1;
|
208
|
+
else if (val1 < val2) return -1;
|
209
|
+
else return 0;
|
210
|
+
}
|
211
|
+
|
212
|
+
FrtSortField *frt_sort_field_byte_new(FrtSymbol field, bool reverse)
|
213
|
+
{
|
214
|
+
return sort_field_alloc(field, FRT_SORT_TYPE_BYTE, reverse,
|
215
|
+
&sf_byte_compare, &sf_byte_get_val,
|
216
|
+
&FRT_BYTE_FIELD_INDEX_CLASS);
|
217
|
+
}
|
218
|
+
|
219
|
+
/***************************************************************************
|
220
|
+
* IntegerSortField
|
221
|
+
***************************************************************************/
|
222
|
+
|
223
|
+
static void sf_int_get_val(void *index, FrtHit *hit, FrtComparable *comparable)
|
224
|
+
{
|
225
|
+
comparable->val.l = ((long *)index)[hit->doc];
|
226
|
+
}
|
227
|
+
|
228
|
+
static int sf_int_compare(void *index, FrtHit *hit1, FrtHit *hit2)
|
229
|
+
{
|
230
|
+
long val1 = ((long *)index)[hit1->doc];
|
231
|
+
long val2 = ((long *)index)[hit2->doc];
|
232
|
+
if (val1 > val2) return 1;
|
233
|
+
else if (val1 < val2) return -1;
|
234
|
+
else return 0;
|
235
|
+
}
|
236
|
+
|
237
|
+
FrtSortField *frt_sort_field_int_new(FrtSymbol field, bool reverse)
|
238
|
+
{
|
239
|
+
return sort_field_alloc(field, FRT_SORT_TYPE_INTEGER, reverse,
|
240
|
+
&sf_int_compare, &sf_int_get_val,
|
241
|
+
&FRT_INTEGER_FIELD_INDEX_CLASS);
|
242
|
+
}
|
243
|
+
|
244
|
+
/***************************************************************************
|
245
|
+
* FloatSortField
|
246
|
+
***************************************************************************/
|
247
|
+
|
248
|
+
static void sf_float_get_val(void *index, FrtHit *hit, FrtComparable *comparable)
|
249
|
+
{
|
250
|
+
comparable->val.f = ((float *)index)[hit->doc];
|
251
|
+
}
|
252
|
+
|
253
|
+
static int sf_float_compare(void *index, FrtHit *hit1, FrtHit *hit2)
|
254
|
+
{
|
255
|
+
float val1 = ((float *)index)[hit1->doc];
|
256
|
+
float val2 = ((float *)index)[hit2->doc];
|
257
|
+
if (val1 > val2) return 1;
|
258
|
+
else if (val1 < val2) return -1;
|
259
|
+
else return 0;
|
260
|
+
}
|
261
|
+
|
262
|
+
FrtSortField *frt_sort_field_float_new(FrtSymbol field, bool reverse)
|
263
|
+
{
|
264
|
+
return sort_field_alloc(field, FRT_SORT_TYPE_FLOAT, reverse,
|
265
|
+
&sf_float_compare, &sf_float_get_val,
|
266
|
+
&FRT_FLOAT_FIELD_INDEX_CLASS);
|
267
|
+
}
|
268
|
+
|
269
|
+
/***************************************************************************
|
270
|
+
* StringSortField
|
271
|
+
***************************************************************************/
|
272
|
+
|
273
|
+
static void sf_string_get_val(void *index, FrtHit *hit, FrtComparable *comparable)
|
274
|
+
{
|
275
|
+
comparable->val.s
|
276
|
+
= ((FrtStringIndex *)index)->values[
|
277
|
+
((FrtStringIndex *)index)->index[hit->doc]];
|
278
|
+
}
|
279
|
+
|
280
|
+
static int sf_string_compare(void *index, FrtHit *hit1, FrtHit *hit2)
|
281
|
+
{
|
282
|
+
char *s1 = ((FrtStringIndex *)index)->values[
|
283
|
+
((FrtStringIndex *)index)->index[hit1->doc]];
|
284
|
+
char *s2 = ((FrtStringIndex *)index)->values[
|
285
|
+
((FrtStringIndex *)index)->index[hit2->doc]];
|
286
|
+
|
287
|
+
if (s1 == NULL) return s2 ? 1 : 0;
|
288
|
+
if (s2 == NULL) return -1;
|
289
|
+
|
290
|
+
#if defined POSH_OS_WIN32 || defined POSH_OS_WIN64
|
291
|
+
return strcmp(s1, s2);
|
292
|
+
#else
|
293
|
+
return strcoll(s1, s2);
|
294
|
+
#endif
|
295
|
+
|
296
|
+
/*
|
297
|
+
* TODO: investigate whether it would be a good idea to presort strings.
|
298
|
+
*
|
299
|
+
int val1 = index->index[hit1->doc];
|
300
|
+
int val2 = index->index[hit2->doc];
|
301
|
+
if (val1 > val2) return 1;
|
302
|
+
else if (val1 < val2) return -1;
|
303
|
+
else return 0;
|
304
|
+
*/
|
305
|
+
}
|
306
|
+
|
307
|
+
FrtSortField *frt_sort_field_string_new(FrtSymbol field, bool reverse)
|
308
|
+
{
|
309
|
+
return sort_field_alloc(field, FRT_SORT_TYPE_STRING, reverse,
|
310
|
+
&sf_string_compare, &sf_string_get_val,
|
311
|
+
&FRT_STRING_FIELD_INDEX_CLASS);
|
312
|
+
}
|
313
|
+
|
314
|
+
/***************************************************************************
|
315
|
+
* AutoSortField
|
316
|
+
***************************************************************************/
|
317
|
+
|
318
|
+
FrtSortField *frt_sort_field_auto_new(FrtSymbol field, bool reverse)
|
319
|
+
{
|
320
|
+
return sort_field_alloc(field, FRT_SORT_TYPE_AUTO, reverse, NULL, NULL, NULL);
|
321
|
+
}
|
322
|
+
|
323
|
+
/***************************************************************************
|
324
|
+
*
|
325
|
+
* FieldSortedHitQueue
|
326
|
+
*
|
327
|
+
***************************************************************************/
|
328
|
+
|
329
|
+
/***************************************************************************
|
330
|
+
* Comparator
|
331
|
+
***************************************************************************/
|
332
|
+
|
333
|
+
typedef struct Comparator {
|
334
|
+
void *index;
|
335
|
+
bool reverse : 1;
|
336
|
+
int (*compare)(void *index_ptr, FrtHit *hit1, FrtHit *hit2);
|
337
|
+
} Comparator;
|
338
|
+
|
339
|
+
static Comparator *comparator_new(void *index, bool reverse,
|
340
|
+
int (*compare)(void *index_ptr, FrtHit *hit1, FrtHit *hit2))
|
341
|
+
{
|
342
|
+
Comparator *self = FRT_ALLOC(Comparator);
|
343
|
+
self->index = index;
|
344
|
+
self->reverse = reverse;
|
345
|
+
self->compare = compare;
|
346
|
+
return self;
|
347
|
+
}
|
348
|
+
|
349
|
+
/***************************************************************************
|
350
|
+
* Sorter
|
351
|
+
***************************************************************************/
|
352
|
+
|
353
|
+
typedef struct Sorter {
|
354
|
+
Comparator **comparators;
|
355
|
+
int c_cnt;
|
356
|
+
FrtSort *sort;
|
357
|
+
} Sorter;
|
358
|
+
|
359
|
+
#define SET_AUTO(upper_type, lower_type) \
|
360
|
+
sf->type = FRT_SORT_TYPE_ ## upper_type;\
|
361
|
+
sf->field_index_class = &FRT_ ## upper_type ## _FIELD_INDEX_CLASS;\
|
362
|
+
sf->compare = sf_ ## lower_type ## _compare;\
|
363
|
+
sf->get_val = sf_ ## lower_type ## _get_val
|
364
|
+
|
365
|
+
static void sort_field_auto_evaluate(FrtSortField *sf, char *text)
|
366
|
+
{
|
367
|
+
int int_val;
|
368
|
+
float float_val;
|
369
|
+
int text_len = 0, scan_len = 0;
|
370
|
+
|
371
|
+
text_len = (int)strlen(text);
|
372
|
+
sscanf(text, "%d%n", &int_val, &scan_len);
|
373
|
+
if (scan_len == text_len) {
|
374
|
+
SET_AUTO(INTEGER, int);
|
375
|
+
} else {
|
376
|
+
sscanf(text, "%f%n", &float_val, &scan_len);
|
377
|
+
if (scan_len == text_len) {
|
378
|
+
SET_AUTO(FLOAT, float);
|
379
|
+
} else {
|
380
|
+
SET_AUTO(STRING, string);
|
381
|
+
}
|
382
|
+
}
|
383
|
+
}
|
384
|
+
|
385
|
+
static Comparator *sorter_get_comparator(FrtSortField *sf, FrtIndexReader *ir)
|
386
|
+
{
|
387
|
+
void *index = NULL;
|
388
|
+
if (sf->type > FRT_SORT_TYPE_DOC) {
|
389
|
+
FrtFieldIndex *field_index = NULL;
|
390
|
+
if (sf->type == FRT_SORT_TYPE_AUTO) {
|
391
|
+
FrtTermEnum *te = frt_ir_terms(ir, sf->field);
|
392
|
+
if (te) {
|
393
|
+
if (!te->next(te) && (ir->num_docs(ir) > 0)) {
|
394
|
+
FRT_RAISE(FRT_ARG_ERROR,
|
395
|
+
"Cannot sort by field \"%s\" as there are no terms "
|
396
|
+
"in that field in the index.", rb_id2name(sf->field));
|
397
|
+
}
|
398
|
+
sort_field_auto_evaluate(sf, te->curr_term);
|
399
|
+
te->close(te);
|
400
|
+
}
|
401
|
+
}
|
402
|
+
frt_mutex_lock(&ir->field_index_mutex);
|
403
|
+
field_index = frt_field_index_get(ir, sf->field, sf->field_index_class);
|
404
|
+
frt_mutex_unlock(&ir->field_index_mutex);
|
405
|
+
index = field_index->index;
|
406
|
+
}
|
407
|
+
return comparator_new(index, sf->reverse, sf->compare);
|
408
|
+
}
|
409
|
+
|
410
|
+
static void sorter_destroy(Sorter *self)
|
411
|
+
{
|
412
|
+
int i;
|
413
|
+
|
414
|
+
for (i = 0; i < self->c_cnt; i++) {
|
415
|
+
free(self->comparators[i]);
|
416
|
+
}
|
417
|
+
free(self->comparators);
|
418
|
+
free(self);
|
419
|
+
}
|
420
|
+
|
421
|
+
static Sorter *sorter_new(FrtSort *sort)
|
422
|
+
{
|
423
|
+
Sorter *self = FRT_ALLOC(Sorter);
|
424
|
+
self->c_cnt = sort->size;
|
425
|
+
self->comparators = FRT_ALLOC_AND_ZERO_N(Comparator *, self->c_cnt);
|
426
|
+
self->sort = sort;
|
427
|
+
return self;
|
428
|
+
}
|
429
|
+
|
430
|
+
/***************************************************************************
|
431
|
+
* FieldSortedHitQueue
|
432
|
+
***************************************************************************/
|
433
|
+
|
434
|
+
static bool fshq_less_than(const void *hit1, const void *hit2)
|
435
|
+
{
|
436
|
+
int cmp = 0;
|
437
|
+
printf("Whoops, shouldn't call this.\n");
|
438
|
+
if (cmp != 0) {
|
439
|
+
return cmp;
|
440
|
+
} else {
|
441
|
+
return ((FrtHit *)hit1)->score < ((FrtHit *)hit2)->score;
|
442
|
+
}
|
443
|
+
}
|
444
|
+
|
445
|
+
static bool fshq_lt(Sorter *sorter, FrtHit *hit1, FrtHit *hit2)
|
446
|
+
{
|
447
|
+
Comparator *comp;
|
448
|
+
int diff = 0, i;
|
449
|
+
for (i = 0; i < sorter->c_cnt && diff == 0; i++) {
|
450
|
+
comp = sorter->comparators[i];
|
451
|
+
if (comp->reverse) {
|
452
|
+
diff = comp->compare(comp->index, hit2, hit1);
|
453
|
+
} else {
|
454
|
+
diff = comp->compare(comp->index, hit1, hit2);
|
455
|
+
}
|
456
|
+
}
|
457
|
+
|
458
|
+
if (diff != 0) {
|
459
|
+
return diff > 0;
|
460
|
+
} else {
|
461
|
+
return hit1->doc > hit2->doc;
|
462
|
+
}
|
463
|
+
}
|
464
|
+
|
465
|
+
void fshq_pq_down(FrtPriorityQueue *pq)
|
466
|
+
{
|
467
|
+
register int i = 1;
|
468
|
+
register int j = 2; /* i << 1; */
|
469
|
+
register int k = 3; /* j + 1; */
|
470
|
+
FrtHit **heap = (FrtHit **)pq->heap;
|
471
|
+
FrtHit *node = heap[i]; /* save top node */
|
472
|
+
Sorter *sorter = (Sorter *)heap[0];
|
473
|
+
|
474
|
+
if ((k <= pq->size) && fshq_lt(sorter, heap[k], heap[j])) {
|
475
|
+
j = k;
|
476
|
+
}
|
477
|
+
|
478
|
+
while ((j <= pq->size) && fshq_lt(sorter, heap[j], node)) {
|
479
|
+
heap[i] = heap[j]; /* shift up child */
|
480
|
+
i = j;
|
481
|
+
j = i << 1;
|
482
|
+
k = j + 1;
|
483
|
+
if ((k <= pq->size) && fshq_lt(sorter, heap[k], heap[j])) {
|
484
|
+
j = k;
|
485
|
+
}
|
486
|
+
}
|
487
|
+
heap[i] = node;
|
488
|
+
}
|
489
|
+
|
490
|
+
FrtHit *frt_fshq_pq_pop(FrtPriorityQueue *pq)
|
491
|
+
{
|
492
|
+
if (pq->size > 0) {
|
493
|
+
FrtHit *hit = (FrtHit *)pq->heap[1]; /* save first value */
|
494
|
+
pq->heap[1] = pq->heap[pq->size]; /* move last to first */
|
495
|
+
pq->heap[pq->size] = NULL;
|
496
|
+
pq->size--;
|
497
|
+
fshq_pq_down(pq); /* adjust heap */
|
498
|
+
return hit;
|
499
|
+
} else {
|
500
|
+
return NULL;
|
501
|
+
}
|
502
|
+
}
|
503
|
+
|
504
|
+
static void fshq_pq_up(FrtPriorityQueue *pq)
|
505
|
+
{
|
506
|
+
FrtHit **heap = (FrtHit **)pq->heap;
|
507
|
+
FrtHit *node;
|
508
|
+
int i = pq->size;
|
509
|
+
int j = i >> 1;
|
510
|
+
Sorter *sorter = (Sorter *)heap[0];
|
511
|
+
node = heap[i];
|
512
|
+
|
513
|
+
while ((j > 0) && fshq_lt(sorter, node, heap[j])) {
|
514
|
+
heap[i] = heap[j];
|
515
|
+
i = j;
|
516
|
+
j = j >> 1;
|
517
|
+
}
|
518
|
+
heap[i] = node;
|
519
|
+
}
|
520
|
+
|
521
|
+
void frt_fshq_pq_insert(FrtPriorityQueue *pq, FrtHit *hit)
|
522
|
+
{
|
523
|
+
if (pq->size < pq->capa) {
|
524
|
+
FrtHit *new_hit = FRT_ALLOC(FrtHit);
|
525
|
+
memcpy(new_hit, hit, sizeof(FrtHit));
|
526
|
+
pq->size++;
|
527
|
+
if (pq->size >= pq->mem_capa) {
|
528
|
+
pq->mem_capa <<= 1;
|
529
|
+
FRT_REALLOC_N(pq->heap, void *, pq->mem_capa);
|
530
|
+
}
|
531
|
+
pq->heap[pq->size] = new_hit;
|
532
|
+
fshq_pq_up(pq);
|
533
|
+
} else if (pq->size > 0
|
534
|
+
&& fshq_lt((Sorter *)pq->heap[0], (FrtHit *)pq->heap[1], hit)) {
|
535
|
+
memcpy(pq->heap[1], hit, sizeof(FrtHit));
|
536
|
+
fshq_pq_down(pq);
|
537
|
+
}
|
538
|
+
}
|
539
|
+
|
540
|
+
void frt_fshq_pq_destroy(FrtPriorityQueue *self)
|
541
|
+
{
|
542
|
+
sorter_destroy((Sorter *)self->heap[0]);
|
543
|
+
frt_pq_destroy(self);
|
544
|
+
}
|
545
|
+
|
546
|
+
FrtPriorityQueue *frt_fshq_pq_new(int size, FrtSort *sort, FrtIndexReader *ir)
|
547
|
+
{
|
548
|
+
FrtPriorityQueue *self = frt_pq_new(size, &fshq_less_than, &free);
|
549
|
+
int i;
|
550
|
+
Sorter *sorter = sorter_new(sort);
|
551
|
+
FrtSortField *sf;
|
552
|
+
|
553
|
+
for (i = 0; i < sort->size; i++) {
|
554
|
+
sf = sort->sort_fields[i];
|
555
|
+
sorter->comparators[i] = sorter_get_comparator(sf, ir);
|
556
|
+
}
|
557
|
+
self->heap[0] = sorter;
|
558
|
+
|
559
|
+
return self;
|
560
|
+
}
|
561
|
+
|
562
|
+
FrtHit *frt_fshq_pq_pop_fd(FrtPriorityQueue *pq)
|
563
|
+
{
|
564
|
+
if (pq->size <= 0) {
|
565
|
+
return NULL;
|
566
|
+
}
|
567
|
+
else {
|
568
|
+
int j;
|
569
|
+
Sorter *sorter = (Sorter *)pq->heap[0];
|
570
|
+
const int cmp_cnt = sorter->c_cnt;
|
571
|
+
FrtSortField **sort_fields = sorter->sort->sort_fields;
|
572
|
+
FrtHit *hit = (FrtHit *)pq->heap[1]; /* save first value */
|
573
|
+
FrtFieldDoc *field_doc;
|
574
|
+
FrtComparable *comparables;
|
575
|
+
Comparator **comparators = sorter->comparators;
|
576
|
+
pq->heap[1] = pq->heap[pq->size]; /* move last to first */
|
577
|
+
pq->heap[pq->size] = NULL;
|
578
|
+
pq->size--;
|
579
|
+
fshq_pq_down(pq); /* adjust heap */
|
580
|
+
|
581
|
+
field_doc = (FrtFieldDoc *)frt_emalloc(sizeof(FrtFieldDoc) + sizeof(FrtComparable) * cmp_cnt);
|
582
|
+
comparables = field_doc->comparables;
|
583
|
+
memcpy(field_doc, hit, sizeof(FrtHit));
|
584
|
+
field_doc->size = cmp_cnt;
|
585
|
+
|
586
|
+
for (j = 0; j < cmp_cnt; j++) {
|
587
|
+
FrtSortField *sf = sort_fields[j];
|
588
|
+
Comparator *comparator = comparators[j];
|
589
|
+
sf->get_val(comparator->index, hit, &(comparables[j]));
|
590
|
+
comparables[j].type = sf->type;
|
591
|
+
comparables[j].reverse = comparator->reverse;
|
592
|
+
}
|
593
|
+
free(hit);
|
594
|
+
return (FrtHit *)field_doc;
|
595
|
+
}
|
596
|
+
}
|
597
|
+
|
598
|
+
/***************************************************************************
|
599
|
+
* FieldDocSortedHitQueue
|
600
|
+
***************************************************************************/
|
601
|
+
|
602
|
+
bool frt_fdshq_lt(FrtFieldDoc *fd1, FrtFieldDoc *fd2)
|
603
|
+
{
|
604
|
+
int i;
|
605
|
+
bool c = false;
|
606
|
+
bool all_equal = false;
|
607
|
+
int sc;
|
608
|
+
FrtComparable *cmps1 = fd1->comparables;
|
609
|
+
FrtComparable *cmps2 = fd2->comparables;
|
610
|
+
|
611
|
+
for (i = 0; (i < fd1->size) && (!c); i++) {
|
612
|
+
int type = cmps1[i].type;
|
613
|
+
if (cmps1[i].reverse) {
|
614
|
+
switch (type) {
|
615
|
+
case FRT_SORT_TYPE_SCORE:
|
616
|
+
if (cmps1[i].val.f > cmps2[i].val.f) { all_equal = false; c = true; }
|
617
|
+
else if (cmps1[i].val.f == cmps2[i].val.f) { if (!c) all_equal = true; }
|
618
|
+
else { all_equal = false; }
|
619
|
+
break;
|
620
|
+
case FRT_SORT_TYPE_FLOAT:
|
621
|
+
if (cmps1[i].val.f < cmps2[i].val.f) { all_equal = false; c = true; }
|
622
|
+
else if (cmps1[i].val.f == cmps2[i].val.f) { if (!c) all_equal = true; }
|
623
|
+
else { all_equal = false; }
|
624
|
+
break;
|
625
|
+
case FRT_SORT_TYPE_DOC:
|
626
|
+
if (fd1->hit.doc < fd2->hit.doc) { all_equal = false; c = true; }
|
627
|
+
break;
|
628
|
+
case FRT_SORT_TYPE_INTEGER:
|
629
|
+
if (cmps1[i].val.l < cmps2[i].val.l) { all_equal = false; c = true; }
|
630
|
+
else if (cmps1[i].val.l == cmps2[i].val.l) { if (!c) all_equal = true; }
|
631
|
+
else { all_equal = false; }
|
632
|
+
break;
|
633
|
+
case FRT_SORT_TYPE_BYTE:
|
634
|
+
if (cmps1[i].val.l > cmps2[i].val.l) { all_equal = false; c = true; }
|
635
|
+
else if (cmps1[i].val.l == cmps2[i].val.l) { if (!c) all_equal = true; }
|
636
|
+
else { all_equal = false; }
|
637
|
+
break;
|
638
|
+
case FRT_SORT_TYPE_STRING:
|
639
|
+
do {
|
640
|
+
char *s1 = cmps1[i].val.s;
|
641
|
+
char *s2 = cmps2[i].val.s;
|
642
|
+
if ((s1 == NULL) && s2) { all_equal = false; }
|
643
|
+
else if (s1 && (s2 == NULL)) { all_equal = false; c = true; }
|
644
|
+
else if (s1 && s2) {
|
645
|
+
#if defined POSH_OS_WIN32 || defined POSH_OS_WIN64
|
646
|
+
sc = strcmp(s1, s2);
|
647
|
+
#else
|
648
|
+
sc = strcoll(s1, s2);
|
649
|
+
#endif
|
650
|
+
if (sc < 0) { all_equal = false; c = true; }
|
651
|
+
else if (sc == 0) { if (!c) all_equal = true; }
|
652
|
+
else { all_equal = false; }
|
653
|
+
} else { all_equal = false; }
|
654
|
+
} while (0);
|
655
|
+
break;
|
656
|
+
default:
|
657
|
+
FRT_RAISE(FRT_ARG_ERROR, "Unknown sort type: %d.", type);
|
658
|
+
break;
|
659
|
+
}
|
660
|
+
} else {
|
661
|
+
switch (type) {
|
662
|
+
case FRT_SORT_TYPE_SCORE:
|
663
|
+
if (cmps1[i].val.f < cmps2[i].val.f) { all_equal = false; c = true; }
|
664
|
+
else if (cmps1[i].val.f == cmps2[i].val.f) { if (!c) all_equal = true; }
|
665
|
+
else { all_equal = false; }
|
666
|
+
break;
|
667
|
+
case FRT_SORT_TYPE_FLOAT:
|
668
|
+
if (cmps1[i].val.f > cmps2[i].val.f) { all_equal = false; c = true; }
|
669
|
+
else if (cmps1[i].val.f == cmps2[i].val.f) { if (!c) all_equal = true; }
|
670
|
+
else { all_equal = false; }
|
671
|
+
break;
|
672
|
+
case FRT_SORT_TYPE_DOC:
|
673
|
+
if (fd1->hit.doc > fd2->hit.doc) { all_equal = false; c = true; }
|
674
|
+
break;
|
675
|
+
case FRT_SORT_TYPE_INTEGER:
|
676
|
+
if (cmps1[i].val.l > cmps2[i].val.l) { all_equal = false; c = true; }
|
677
|
+
else if (cmps1[i].val.l == cmps2[i].val.l) { if (!c) all_equal = true; }
|
678
|
+
else { all_equal = false; }
|
679
|
+
break;
|
680
|
+
case FRT_SORT_TYPE_BYTE:
|
681
|
+
if (cmps1[i].val.l < cmps2[i].val.l) { all_equal = false; c = true; }
|
682
|
+
else if (cmps1[i].val.l == cmps2[i].val.l) { if (!c) all_equal = true; }
|
683
|
+
else { all_equal = false; }
|
684
|
+
break;
|
685
|
+
case FRT_SORT_TYPE_STRING:
|
686
|
+
do {
|
687
|
+
char *s1 = cmps1[i].val.s;
|
688
|
+
char *s2 = cmps2[i].val.s;
|
689
|
+
if (s1 && (s2 == NULL)) { if (!c) all_equal = false; }
|
690
|
+
else if ((s1 == NULL) && s2) { all_equal = false; c = true; }
|
691
|
+
else if (s1 && s2) {
|
692
|
+
#if defined POSH_OS_WIN32 || defined POSH_OS_WIN64
|
693
|
+
sc = strcmp(s1, s2);
|
694
|
+
#else
|
695
|
+
sc = strcoll(s1, s2);
|
696
|
+
#endif
|
697
|
+
if (sc > 0) { all_equal = false; c = true; }
|
698
|
+
else if (sc == 0) { if (!c) all_equal = true; }
|
699
|
+
else { all_equal = false; }
|
700
|
+
} else { all_equal = false; }
|
701
|
+
} while (0);
|
702
|
+
break;
|
703
|
+
default:
|
704
|
+
FRT_RAISE(FRT_ARG_ERROR, "Unknown sort type: %d.", type);
|
705
|
+
break;
|
706
|
+
}
|
707
|
+
}
|
708
|
+
if (!all_equal) break;
|
709
|
+
}
|
710
|
+
if (all_equal) {
|
711
|
+
if (cmps1[0].reverse) {
|
712
|
+
if (fd1->hit.doc > fd2->hit.doc) c = true;
|
713
|
+
} else {
|
714
|
+
if (fd1->hit.doc > fd2->hit.doc) c = true;
|
715
|
+
}
|
716
|
+
}
|
717
|
+
return c;
|
718
|
+
}
|
719
|
+
|
720
|
+
/***************************************************************************
|
721
|
+
*
|
722
|
+
* Sort
|
723
|
+
*
|
724
|
+
***************************************************************************/
|
725
|
+
|
726
|
+
#define SORT_INIT_SIZE 4
|
727
|
+
|
728
|
+
FrtSort *frt_sort_new()
|
729
|
+
{
|
730
|
+
FrtSort *self = FRT_ALLOC(FrtSort);
|
731
|
+
self->size = 0;
|
732
|
+
self->capa = SORT_INIT_SIZE;
|
733
|
+
self->sort_fields = FRT_ALLOC_N(FrtSortField *, SORT_INIT_SIZE);
|
734
|
+
self->destroy_all = true;
|
735
|
+
self->start = 0;
|
736
|
+
|
737
|
+
return self;
|
738
|
+
}
|
739
|
+
|
740
|
+
void frt_sort_clear(FrtSort *self)
|
741
|
+
{
|
742
|
+
int i;
|
743
|
+
if (self->destroy_all) {
|
744
|
+
for (i = 0; i < self->size; i++) {
|
745
|
+
frt_sort_field_destroy(self->sort_fields[i]);
|
746
|
+
}
|
747
|
+
}
|
748
|
+
self->size = 0;
|
749
|
+
}
|
750
|
+
|
751
|
+
void frt_sort_destroy(void *p)
|
752
|
+
{
|
753
|
+
FrtSort *self = (FrtSort *)p;
|
754
|
+
frt_sort_clear(self);
|
755
|
+
free(self->sort_fields);
|
756
|
+
free(self);
|
757
|
+
}
|
758
|
+
|
759
|
+
void frt_sort_add_sort_field(FrtSort *self, FrtSortField *sf)
|
760
|
+
{
|
761
|
+
if (self->size == self->capa) {
|
762
|
+
self->capa <<= 1;
|
763
|
+
FRT_REALLOC_N(self->sort_fields, FrtSortField *, self->capa);
|
764
|
+
}
|
765
|
+
|
766
|
+
self->sort_fields[self->size] = sf;
|
767
|
+
self->size++;
|
768
|
+
}
|
769
|
+
|
770
|
+
char *frt_sort_to_s(FrtSort *self)
|
771
|
+
{
|
772
|
+
int i, len = 20;
|
773
|
+
char *s;
|
774
|
+
char *str;
|
775
|
+
char **sf_strs = FRT_ALLOC_N(char *, self->size);
|
776
|
+
|
777
|
+
for (i = 0; i < self->size; i++) {
|
778
|
+
sf_strs[i] = s = frt_sort_field_to_s(self->sort_fields[i]);
|
779
|
+
len += (int)strlen(s) + 2;
|
780
|
+
}
|
781
|
+
|
782
|
+
str = FRT_ALLOC_N(char, len);
|
783
|
+
s = frt_strapp(str, "Sort[");
|
784
|
+
|
785
|
+
for (i = 0; i < self->size; i++) {
|
786
|
+
s += sprintf(s, "%s, ", sf_strs[i]);
|
787
|
+
free(sf_strs[i]);
|
788
|
+
}
|
789
|
+
free(sf_strs);
|
790
|
+
|
791
|
+
if (self->size > 0) {
|
792
|
+
s -= 2;
|
793
|
+
}
|
794
|
+
sprintf(s, "]");
|
795
|
+
return str;
|
796
|
+
}
|