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 +1,880 @@
|
|
1
|
+
#ifndef FRT_INDEX_H
|
2
|
+
#define FRT_INDEX_H
|
3
|
+
|
4
|
+
#include "frt_global.h"
|
5
|
+
#include "frt_document.h"
|
6
|
+
#include "frt_analysis.h"
|
7
|
+
#include "frt_hash.h"
|
8
|
+
#include "frt_hashset.h"
|
9
|
+
#include "frt_store.h"
|
10
|
+
#include "frt_mempool.h"
|
11
|
+
#include "frt_similarity.h"
|
12
|
+
#include "frt_bitvector.h"
|
13
|
+
#include "frt_priorityqueue.h"
|
14
|
+
|
15
|
+
typedef struct FrtIndexReader FrtIndexReader;
|
16
|
+
typedef struct FrtMultiReader FrtMultiReader;
|
17
|
+
typedef struct FrtDeleter FrtDeleter;
|
18
|
+
|
19
|
+
/****************************************************************************
|
20
|
+
*
|
21
|
+
* FrtConfig
|
22
|
+
*
|
23
|
+
****************************************************************************/
|
24
|
+
|
25
|
+
typedef struct FrtConfig
|
26
|
+
{
|
27
|
+
int chunk_size;
|
28
|
+
int max_buffer_memory;
|
29
|
+
int index_interval;
|
30
|
+
int skip_interval;
|
31
|
+
int merge_factor;
|
32
|
+
int max_buffered_docs;
|
33
|
+
int max_merge_docs;
|
34
|
+
int max_field_length;
|
35
|
+
bool use_compound_file;
|
36
|
+
} FrtConfig;
|
37
|
+
|
38
|
+
extern const FrtConfig frt_default_config;
|
39
|
+
|
40
|
+
/***************************************************************************
|
41
|
+
*
|
42
|
+
* FrtCacheObject
|
43
|
+
*
|
44
|
+
***************************************************************************/
|
45
|
+
|
46
|
+
typedef struct FrtCacheObject {
|
47
|
+
FrtHash *ref_tab1;
|
48
|
+
FrtHash *ref_tab2;
|
49
|
+
void *ref1;
|
50
|
+
void *ref2;
|
51
|
+
void *obj;
|
52
|
+
void (*destroy)(void *p);
|
53
|
+
} FrtCacheObject;
|
54
|
+
|
55
|
+
extern FrtCacheObject *frt_co_create(FrtHash *ref_tab1,
|
56
|
+
FrtHash *ref_tab2,
|
57
|
+
void *ref1, void *ref2, frt_free_ft destroy, void *obj);
|
58
|
+
extern FrtHash *frt_co_hash_create();
|
59
|
+
|
60
|
+
/****************************************************************************
|
61
|
+
*
|
62
|
+
* FrtFieldInfo
|
63
|
+
*
|
64
|
+
****************************************************************************/
|
65
|
+
|
66
|
+
typedef enum
|
67
|
+
{
|
68
|
+
FRT_STORE_NO = 0,
|
69
|
+
FRT_STORE_YES = 1
|
70
|
+
} FrtStoreValue;
|
71
|
+
|
72
|
+
typedef enum
|
73
|
+
{
|
74
|
+
FRT_INDEX_NO = 0,
|
75
|
+
FRT_INDEX_UNTOKENIZED = 1,
|
76
|
+
FRT_INDEX_YES = 3,
|
77
|
+
FRT_INDEX_UNTOKENIZED_OMIT_NORMS = 5,
|
78
|
+
FRT_INDEX_YES_OMIT_NORMS = 7
|
79
|
+
} FrtIndexValue;
|
80
|
+
|
81
|
+
typedef enum
|
82
|
+
{
|
83
|
+
FRT_TERM_VECTOR_NO = 0,
|
84
|
+
FRT_TERM_VECTOR_YES = 1,
|
85
|
+
FRT_TERM_VECTOR_WITH_POSITIONS = 3,
|
86
|
+
FRT_TERM_VECTOR_WITH_OFFSETS = 5,
|
87
|
+
FRT_TERM_VECTOR_WITH_POSITIONS_OFFSETS = 7
|
88
|
+
} FrtTermVectorValue;
|
89
|
+
|
90
|
+
#define FRT_FI_IS_STORED_BM 0x001
|
91
|
+
#define FRT_FI_IS_INDEXED_BM 0x004
|
92
|
+
#define FRT_FI_IS_TOKENIZED_BM 0x008
|
93
|
+
#define FRT_FI_OMIT_NORMS_BM 0x010
|
94
|
+
#define FRT_FI_STORE_TERM_VECTOR_BM 0x020
|
95
|
+
#define FRT_FI_STORE_POSITIONS_BM 0x040
|
96
|
+
#define FRT_FI_STORE_OFFSETS_BM 0x080
|
97
|
+
|
98
|
+
typedef struct FrtFieldInfo
|
99
|
+
{
|
100
|
+
FrtSymbol name;
|
101
|
+
float boost;
|
102
|
+
unsigned int bits;
|
103
|
+
int number;
|
104
|
+
int ref_cnt;
|
105
|
+
} FrtFieldInfo;
|
106
|
+
|
107
|
+
extern FrtFieldInfo *frt_fi_new(FrtSymbol name,
|
108
|
+
FrtStoreValue store,
|
109
|
+
FrtIndexValue index,
|
110
|
+
FrtTermVectorValue term_vector);
|
111
|
+
extern char *frt_fi_to_s(FrtFieldInfo *fi);
|
112
|
+
extern void frt_fi_deref(FrtFieldInfo *fi);
|
113
|
+
|
114
|
+
#define fi_is_stored(fi) (((fi)->bits & FRT_FI_IS_STORED_BM) != 0)
|
115
|
+
#define fi_is_indexed(fi) (((fi)->bits & FRT_FI_IS_INDEXED_BM) != 0)
|
116
|
+
#define fi_is_tokenized(fi) (((fi)->bits & FRT_FI_IS_TOKENIZED_BM) != 0)
|
117
|
+
#define fi_omit_norms(fi) (((fi)->bits & FRT_FI_OMIT_NORMS_BM) != 0)
|
118
|
+
#define fi_store_term_vector(fi) (((fi)->bits & FRT_FI_STORE_TERM_VECTOR_BM) != 0)
|
119
|
+
#define fi_store_positions(fi) (((fi)->bits & FRT_FI_STORE_POSITIONS_BM) != 0)
|
120
|
+
#define fi_store_offsets(fi) (((fi)->bits & FRT_FI_STORE_OFFSETS_BM) != 0)
|
121
|
+
#define fi_has_norms(fi)\
|
122
|
+
(((fi)->bits & (FRT_FI_OMIT_NORMS_BM|FRT_FI_IS_INDEXED_BM)) == FRT_FI_IS_INDEXED_BM)
|
123
|
+
|
124
|
+
/****************************************************************************
|
125
|
+
*
|
126
|
+
* FrtFieldInfos
|
127
|
+
*
|
128
|
+
****************************************************************************/
|
129
|
+
|
130
|
+
#define FIELD_INFOS_INIT_CAPA 4
|
131
|
+
/* carry changes over to dummy_fis in test/test_segments.c */
|
132
|
+
typedef struct FrtFieldInfos
|
133
|
+
{
|
134
|
+
FrtStoreValue store;
|
135
|
+
FrtIndexValue index;
|
136
|
+
FrtTermVectorValue term_vector;
|
137
|
+
int size;
|
138
|
+
int capa;
|
139
|
+
FrtFieldInfo **fields;
|
140
|
+
FrtHash *field_dict;
|
141
|
+
int ref_cnt;
|
142
|
+
} FrtFieldInfos;
|
143
|
+
|
144
|
+
FrtFieldInfos *frt_fis_new(FrtStoreValue store, FrtIndexValue index,
|
145
|
+
FrtTermVectorValue term_vector);
|
146
|
+
extern FrtFieldInfo *frt_fis_add_field(FrtFieldInfos *fis, FrtFieldInfo *fi);
|
147
|
+
extern FrtFieldInfo *frt_fis_get_field(FrtFieldInfos *fis, FrtSymbol name);
|
148
|
+
extern int frt_fis_get_field_num(FrtFieldInfos *fis, FrtSymbol name);
|
149
|
+
extern FrtFieldInfo *frt_fis_get_or_add_field(FrtFieldInfos *fis,
|
150
|
+
FrtSymbol name);
|
151
|
+
extern void frt_fis_write(FrtFieldInfos *fis, FrtOutStream *os);
|
152
|
+
extern FrtFieldInfos *frt_fis_read(FrtInStream *is);
|
153
|
+
extern char *frt_fis_to_s(FrtFieldInfos *fis);
|
154
|
+
extern void frt_fis_deref(FrtFieldInfos *fis);
|
155
|
+
|
156
|
+
/****************************************************************************
|
157
|
+
*
|
158
|
+
* FrtSegmentInfo
|
159
|
+
*
|
160
|
+
****************************************************************************/
|
161
|
+
|
162
|
+
#define FRT_SEGMENT_NAME_MAX_LENGTH 100
|
163
|
+
#define FRT_SEGMENTS_FILE_NAME "segments"
|
164
|
+
|
165
|
+
typedef struct FrtSegmentInfo
|
166
|
+
{
|
167
|
+
int ref_cnt;
|
168
|
+
char *name;
|
169
|
+
FrtStore *store;
|
170
|
+
int doc_cnt;
|
171
|
+
int del_gen;
|
172
|
+
int *norm_gens;
|
173
|
+
int norm_gens_size;
|
174
|
+
bool use_compound_file;
|
175
|
+
} FrtSegmentInfo;
|
176
|
+
|
177
|
+
extern FrtSegmentInfo *frt_si_new(char *name, int doc_cnt, FrtStore *store);
|
178
|
+
extern void frt_si_deref(FrtSegmentInfo *si);
|
179
|
+
extern bool frt_si_has_deletions(FrtSegmentInfo *si);
|
180
|
+
extern bool frt_si_has_separate_norms(FrtSegmentInfo *si);
|
181
|
+
extern void frt_si_advance_norm_gen(FrtSegmentInfo *si, int field_num);
|
182
|
+
|
183
|
+
/****************************************************************************
|
184
|
+
*
|
185
|
+
* FrtSegmentInfos
|
186
|
+
*
|
187
|
+
****************************************************************************/
|
188
|
+
|
189
|
+
typedef struct FrtSegmentInfos
|
190
|
+
{
|
191
|
+
FrtFieldInfos *fis;
|
192
|
+
frt_u64 counter;
|
193
|
+
frt_u64 version;
|
194
|
+
frt_i64 generation;
|
195
|
+
frt_i32 format;
|
196
|
+
FrtStore *store;
|
197
|
+
FrtSegmentInfo **segs;
|
198
|
+
int size;
|
199
|
+
int capa;
|
200
|
+
} FrtSegmentInfos;
|
201
|
+
|
202
|
+
extern char *frt_fn_for_generation(char *buf,
|
203
|
+
const char *base,
|
204
|
+
const char *ext,
|
205
|
+
frt_i64 gen);
|
206
|
+
|
207
|
+
extern FrtSegmentInfos *frt_sis_new(FrtFieldInfos *fis);
|
208
|
+
extern FrtSegmentInfo *frt_sis_new_segment(FrtSegmentInfos *sis, int dcnt, FrtStore *store);
|
209
|
+
extern FrtSegmentInfo *frt_sis_add_si(FrtSegmentInfos *sis, FrtSegmentInfo *si);
|
210
|
+
extern void frt_sis_del_at(FrtSegmentInfos *sis, int at);
|
211
|
+
extern void frt_sis_del_from_to(FrtSegmentInfos *sis, int from, int to);
|
212
|
+
extern FrtSegmentInfos *frt_sis_read(FrtStore *store);
|
213
|
+
extern void frt_sis_write(FrtSegmentInfos *sis, FrtStore *store, FrtDeleter *deleter);
|
214
|
+
extern frt_u64 frt_sis_read_current_version(FrtStore *store);
|
215
|
+
extern void frt_sis_destroy(FrtSegmentInfos *sis);
|
216
|
+
extern frt_i64 frt_sis_current_segment_generation(FrtStore *store);
|
217
|
+
extern char *frt_sis_curr_seg_file_name(char *buf, FrtStore *store);
|
218
|
+
extern void frt_sis_put(FrtSegmentInfos *sis, FILE *stream);
|
219
|
+
|
220
|
+
/****************************************************************************
|
221
|
+
*
|
222
|
+
* FrtTermInfo
|
223
|
+
*
|
224
|
+
****************************************************************************/
|
225
|
+
|
226
|
+
typedef struct FrtTermInfo
|
227
|
+
{
|
228
|
+
int doc_freq;
|
229
|
+
off_t frq_ptr;
|
230
|
+
off_t prx_ptr;
|
231
|
+
off_t skip_offset;
|
232
|
+
} FrtTermInfo;
|
233
|
+
|
234
|
+
#define frt_ti_set(ti, mdf, mfp, mpp, mso) do {\
|
235
|
+
(ti).doc_freq = mdf;\
|
236
|
+
(ti).frq_ptr = mfp;\
|
237
|
+
(ti).prx_ptr = mpp;\
|
238
|
+
(ti).skip_offset = mso;\
|
239
|
+
} while (0)
|
240
|
+
|
241
|
+
/****************************************************************************
|
242
|
+
*
|
243
|
+
* FrtTermEnum
|
244
|
+
*
|
245
|
+
****************************************************************************/
|
246
|
+
|
247
|
+
typedef struct FrtTermEnum FrtTermEnum;
|
248
|
+
|
249
|
+
struct FrtTermEnum
|
250
|
+
{
|
251
|
+
char curr_term[FRT_MAX_WORD_SIZE];
|
252
|
+
char prev_term[FRT_MAX_WORD_SIZE];
|
253
|
+
FrtTermInfo curr_ti;
|
254
|
+
int curr_term_len;
|
255
|
+
int field_num;
|
256
|
+
FrtTermEnum *(*set_field)(FrtTermEnum *te, int field_num);
|
257
|
+
char *(*next)(FrtTermEnum *te);
|
258
|
+
char *(*skip_to)(FrtTermEnum *te, const char *term);
|
259
|
+
void (*close)(FrtTermEnum *te);
|
260
|
+
FrtTermEnum *(*clone)(FrtTermEnum *te);
|
261
|
+
};
|
262
|
+
|
263
|
+
char *frt_te_get_term(struct FrtTermEnum *te);
|
264
|
+
FrtTermInfo *frt_te_get_ti(struct FrtTermEnum *te);
|
265
|
+
|
266
|
+
/****************************************************************************
|
267
|
+
*
|
268
|
+
* FrtSegmentTermEnum
|
269
|
+
*
|
270
|
+
****************************************************************************/
|
271
|
+
|
272
|
+
/* * FrtSegmentTermIndex * */
|
273
|
+
|
274
|
+
typedef struct FrtSegmentTermIndex
|
275
|
+
{
|
276
|
+
off_t index_ptr;
|
277
|
+
off_t ptr;
|
278
|
+
int index_cnt;
|
279
|
+
int size;
|
280
|
+
char **index_terms;
|
281
|
+
int *index_term_lens;
|
282
|
+
FrtTermInfo *index_term_infos;
|
283
|
+
off_t *index_ptrs;
|
284
|
+
} FrtSegmentTermIndex;
|
285
|
+
|
286
|
+
/* * FrtSegmentFieldIndex * */
|
287
|
+
|
288
|
+
typedef struct FrtSegmentTermEnum FrtSegmentTermEnum;
|
289
|
+
|
290
|
+
typedef struct FrtSegmentFieldIndex
|
291
|
+
{
|
292
|
+
frt_mutex_t mutex;
|
293
|
+
int skip_interval;
|
294
|
+
int index_interval;
|
295
|
+
off_t index_ptr;
|
296
|
+
FrtTermEnum *index_te;
|
297
|
+
FrtHash *field_dict;
|
298
|
+
} FrtSegmentFieldIndex;
|
299
|
+
|
300
|
+
extern FrtSegmentFieldIndex *frt_sfi_open(FrtStore *store, const char *segment);
|
301
|
+
extern void frt_sfi_close(FrtSegmentFieldIndex *sfi);
|
302
|
+
|
303
|
+
|
304
|
+
/* * FrtSegmentTermEnum * */
|
305
|
+
struct FrtSegmentTermEnum
|
306
|
+
{
|
307
|
+
FrtTermEnum te;
|
308
|
+
FrtInStream *is;
|
309
|
+
int size;
|
310
|
+
int pos;
|
311
|
+
int skip_interval;
|
312
|
+
FrtSegmentFieldIndex *sfi;
|
313
|
+
};
|
314
|
+
|
315
|
+
extern void frt_ste_close(FrtTermEnum *te);
|
316
|
+
extern FrtTermEnum *frt_ste_clone(FrtTermEnum *te);
|
317
|
+
extern FrtTermEnum *frt_ste_new(FrtInStream *is, FrtSegmentFieldIndex *sfi);
|
318
|
+
|
319
|
+
/* * MultiTermEnum * */
|
320
|
+
|
321
|
+
extern FrtTermEnum *frt_mte_new(FrtMultiReader *mr, int field_num, const char *term);
|
322
|
+
|
323
|
+
/****************************************************************************
|
324
|
+
*
|
325
|
+
* FrtTermInfosReader
|
326
|
+
*
|
327
|
+
****************************************************************************/
|
328
|
+
|
329
|
+
typedef struct FrtTermInfosReader
|
330
|
+
{
|
331
|
+
frt_thread_key_t thread_te;
|
332
|
+
void **te_bucket;
|
333
|
+
FrtTermEnum *orig_te;
|
334
|
+
int field_num;
|
335
|
+
} FrtTermInfosReader;
|
336
|
+
|
337
|
+
extern FrtTermInfosReader *frt_tir_open(FrtStore *store,
|
338
|
+
FrtSegmentFieldIndex *sfi,
|
339
|
+
const char *segment);
|
340
|
+
extern FrtTermInfosReader *frt_tir_set_field(FrtTermInfosReader *tir, int field_num);
|
341
|
+
extern FrtTermInfo *frt_tir_get_ti(FrtTermInfosReader *tir, const char *term);
|
342
|
+
extern char *frt_tir_get_term(FrtTermInfosReader *tir, int pos);
|
343
|
+
extern void frt_tir_close(FrtTermInfosReader *tir);
|
344
|
+
|
345
|
+
/****************************************************************************
|
346
|
+
*
|
347
|
+
* FrtTermInfosWriter
|
348
|
+
*
|
349
|
+
****************************************************************************/
|
350
|
+
|
351
|
+
#define FRT_INDEX_INTERVAL 128
|
352
|
+
#define FRT_SKIP_INTERVAL 16
|
353
|
+
|
354
|
+
typedef struct FrtTermWriter
|
355
|
+
{
|
356
|
+
int counter;
|
357
|
+
const char *last_term;
|
358
|
+
FrtTermInfo last_term_info;
|
359
|
+
FrtOutStream *os;
|
360
|
+
} FrtTermWriter;
|
361
|
+
|
362
|
+
typedef struct FrtTermInfosWriter
|
363
|
+
{
|
364
|
+
int field_count;
|
365
|
+
int index_interval;
|
366
|
+
int skip_interval;
|
367
|
+
off_t last_index_ptr;
|
368
|
+
FrtOutStream *tfx_out;
|
369
|
+
FrtTermWriter *tix_writer;
|
370
|
+
FrtTermWriter *tis_writer;
|
371
|
+
} FrtTermInfosWriter;
|
372
|
+
|
373
|
+
extern FrtTermInfosWriter *frt_tiw_open(FrtStore *store,
|
374
|
+
const char *segment,
|
375
|
+
int index_interval,
|
376
|
+
int skip_interval);
|
377
|
+
extern void frt_tiw_start_field(FrtTermInfosWriter *tiw, int field_num);
|
378
|
+
extern void frt_tiw_add(FrtTermInfosWriter *tiw,
|
379
|
+
const char *term,
|
380
|
+
int t_len,
|
381
|
+
FrtTermInfo *ti);
|
382
|
+
extern void frt_tiw_close(FrtTermInfosWriter *tiw);
|
383
|
+
|
384
|
+
/****************************************************************************
|
385
|
+
*
|
386
|
+
* FrtTermDocEnum
|
387
|
+
*
|
388
|
+
****************************************************************************/
|
389
|
+
|
390
|
+
typedef struct FrtTermDocEnum FrtTermDocEnum;
|
391
|
+
struct FrtTermDocEnum
|
392
|
+
{
|
393
|
+
void (*seek)(FrtTermDocEnum *tde, int field_num, const char *term);
|
394
|
+
void (*seek_te)(FrtTermDocEnum *tde, FrtTermEnum *te);
|
395
|
+
void (*seek_ti)(FrtTermDocEnum *tde, FrtTermInfo *ti);
|
396
|
+
int (*doc_num)(FrtTermDocEnum *tde);
|
397
|
+
int (*freq)(FrtTermDocEnum *tde);
|
398
|
+
bool (*next)(FrtTermDocEnum *tde);
|
399
|
+
int (*read)(FrtTermDocEnum *tde, int *docs, int *freqs, int req_num);
|
400
|
+
bool (*skip_to)(FrtTermDocEnum *tde, int target);
|
401
|
+
int (*next_position)(FrtTermDocEnum *tde);
|
402
|
+
void (*close)(FrtTermDocEnum *tde);
|
403
|
+
};
|
404
|
+
|
405
|
+
/* * FrtSegmentTermDocEnum * */
|
406
|
+
|
407
|
+
typedef struct FrtSegmentTermDocEnum FrtSegmentTermDocEnum;
|
408
|
+
struct FrtSegmentTermDocEnum
|
409
|
+
{
|
410
|
+
FrtTermDocEnum tde;
|
411
|
+
void (*seek_prox)(FrtSegmentTermDocEnum *stde, off_t prx_ptr);
|
412
|
+
void (*skip_prox)(FrtSegmentTermDocEnum *stde);
|
413
|
+
FrtTermInfosReader *tir;
|
414
|
+
FrtInStream *frq_in;
|
415
|
+
FrtInStream *prx_in;
|
416
|
+
FrtInStream *skip_in;
|
417
|
+
FrtBitVector *deleted_docs;
|
418
|
+
int count; /* number of docs for this term skipped */
|
419
|
+
int doc_freq; /* number of doc this term appears in */
|
420
|
+
int doc_num;
|
421
|
+
int freq;
|
422
|
+
int num_skips;
|
423
|
+
int skip_interval;
|
424
|
+
int skip_count;
|
425
|
+
int skip_doc;
|
426
|
+
int prx_cnt;
|
427
|
+
int position;
|
428
|
+
off_t frq_ptr;
|
429
|
+
off_t prx_ptr;
|
430
|
+
off_t skip_ptr;
|
431
|
+
bool have_skipped : 1;
|
432
|
+
};
|
433
|
+
|
434
|
+
extern FrtTermDocEnum *frt_stde_new(FrtTermInfosReader *tir, FrtInStream *frq_in,
|
435
|
+
FrtBitVector *deleted_docs, int skip_interval);
|
436
|
+
|
437
|
+
/* * FrtSegmentTermDocEnum * */
|
438
|
+
extern FrtTermDocEnum *frt_stpe_new(FrtTermInfosReader *tir, FrtInStream *frq_in,
|
439
|
+
FrtInStream *prx_in, FrtBitVector *deleted_docs,
|
440
|
+
int skip_interval);
|
441
|
+
|
442
|
+
/****************************************************************************
|
443
|
+
* MultipleTermDocPosEnum
|
444
|
+
****************************************************************************/
|
445
|
+
|
446
|
+
extern FrtTermDocEnum *frt_mtdpe_new(FrtIndexReader *ir, int field_num, char **terms,
|
447
|
+
int t_cnt);
|
448
|
+
|
449
|
+
/****************************************************************************
|
450
|
+
*
|
451
|
+
* FrtOffset
|
452
|
+
*
|
453
|
+
****************************************************************************/
|
454
|
+
|
455
|
+
typedef struct FrtOffset
|
456
|
+
{
|
457
|
+
off_t start;
|
458
|
+
off_t end;
|
459
|
+
} FrtOffset;
|
460
|
+
|
461
|
+
/****************************************************************************
|
462
|
+
*
|
463
|
+
* FrtOccurence
|
464
|
+
*
|
465
|
+
****************************************************************************/
|
466
|
+
|
467
|
+
typedef struct FrtOccurence
|
468
|
+
{
|
469
|
+
struct FrtOccurence *next;
|
470
|
+
int pos;
|
471
|
+
} FrtOccurence;
|
472
|
+
|
473
|
+
/****************************************************************************
|
474
|
+
*
|
475
|
+
* FrtPosting
|
476
|
+
*
|
477
|
+
****************************************************************************/
|
478
|
+
|
479
|
+
typedef struct FrtPosting
|
480
|
+
{
|
481
|
+
int freq;
|
482
|
+
int doc_num;
|
483
|
+
FrtOccurence *first_occ;
|
484
|
+
struct FrtPosting *next;
|
485
|
+
} FrtPosting;
|
486
|
+
|
487
|
+
extern FrtPosting *frt_p_new(FrtMemoryPool *mp, int doc_num, int pos);
|
488
|
+
|
489
|
+
/****************************************************************************
|
490
|
+
*
|
491
|
+
* FrtPostingList
|
492
|
+
*
|
493
|
+
****************************************************************************/
|
494
|
+
|
495
|
+
typedef struct FrtPostingList
|
496
|
+
{
|
497
|
+
const char *term;
|
498
|
+
int term_len;
|
499
|
+
FrtPosting *first;
|
500
|
+
FrtPosting *last;
|
501
|
+
FrtOccurence *last_occ;
|
502
|
+
} FrtPostingList;
|
503
|
+
|
504
|
+
extern FrtPostingList *frt_pl_new(FrtMemoryPool *mp, const char *term,
|
505
|
+
int term_len, FrtPosting *p);
|
506
|
+
extern void frt_pl_add_occ(FrtMemoryPool *mp, FrtPostingList *pl, int pos);
|
507
|
+
extern int frt_pl_cmp(const FrtPostingList **pl1, const FrtPostingList **pl2);
|
508
|
+
|
509
|
+
/****************************************************************************
|
510
|
+
*
|
511
|
+
* FrtTVField
|
512
|
+
*
|
513
|
+
****************************************************************************/
|
514
|
+
|
515
|
+
typedef struct FrtTVField
|
516
|
+
{
|
517
|
+
int field_num;
|
518
|
+
int size;
|
519
|
+
} FrtTVField;
|
520
|
+
|
521
|
+
/****************************************************************************
|
522
|
+
*
|
523
|
+
* FrtTVTerm
|
524
|
+
*
|
525
|
+
****************************************************************************/
|
526
|
+
|
527
|
+
typedef struct FrtTVTerm
|
528
|
+
{
|
529
|
+
char *text;
|
530
|
+
int freq;
|
531
|
+
int *positions;
|
532
|
+
} FrtTVTerm;
|
533
|
+
|
534
|
+
/****************************************************************************
|
535
|
+
*
|
536
|
+
* FrtTermVector
|
537
|
+
*
|
538
|
+
****************************************************************************/
|
539
|
+
|
540
|
+
#define FRT_TV_FIELD_INIT_CAPA 8
|
541
|
+
typedef struct FrtTermVector
|
542
|
+
{
|
543
|
+
int field_num;
|
544
|
+
FrtSymbol field;
|
545
|
+
int term_cnt;
|
546
|
+
FrtTVTerm *terms;
|
547
|
+
int offset_cnt;
|
548
|
+
FrtOffset *offsets;
|
549
|
+
} FrtTermVector;
|
550
|
+
|
551
|
+
extern void frt_tv_destroy(FrtTermVector *tv);
|
552
|
+
extern int frt_tv_get_term_index(FrtTermVector *tv, const char *term);
|
553
|
+
extern int frt_tv_scan_to_term_index(FrtTermVector *tv, const char *term);
|
554
|
+
extern FrtTVTerm *frt_tv_get_tv_term(FrtTermVector *tv, const char *term);
|
555
|
+
|
556
|
+
/****************************************************************************
|
557
|
+
*
|
558
|
+
* FrtLazyDoc
|
559
|
+
*
|
560
|
+
****************************************************************************/
|
561
|
+
|
562
|
+
/* * * FrtLazyDocField * * */
|
563
|
+
typedef struct FrtLazyDocFieldData
|
564
|
+
{
|
565
|
+
off_t start;
|
566
|
+
int length;
|
567
|
+
char *text;
|
568
|
+
} FrtLazyDocFieldData;
|
569
|
+
|
570
|
+
typedef struct FrtLazyDoc FrtLazyDoc;
|
571
|
+
typedef struct FrtLazyDocField
|
572
|
+
{
|
573
|
+
FrtSymbol name;
|
574
|
+
FrtLazyDocFieldData *data;
|
575
|
+
FrtLazyDoc *doc;
|
576
|
+
int size; /* number of data elements */
|
577
|
+
int len; /* length of data elements concatenated */
|
578
|
+
} FrtLazyDocField;
|
579
|
+
|
580
|
+
extern char *frt_lazy_df_get_data(FrtLazyDocField *self, int i);
|
581
|
+
extern void frt_lazy_df_get_bytes(FrtLazyDocField *self, char *buf,
|
582
|
+
int start, int len);
|
583
|
+
|
584
|
+
/* * * FrtLazyDoc * * */
|
585
|
+
struct FrtLazyDoc
|
586
|
+
{
|
587
|
+
FrtHash *field_dictionary;
|
588
|
+
int size;
|
589
|
+
FrtLazyDocField **fields;
|
590
|
+
FrtInStream *fields_in;
|
591
|
+
};
|
592
|
+
|
593
|
+
extern void frt_lazy_doc_close(FrtLazyDoc *self);
|
594
|
+
extern FrtLazyDocField *frt_lazy_doc_get(FrtLazyDoc *self, FrtSymbol field);
|
595
|
+
|
596
|
+
/****************************************************************************
|
597
|
+
*
|
598
|
+
* FrtFieldsReader
|
599
|
+
*
|
600
|
+
****************************************************************************/
|
601
|
+
|
602
|
+
typedef struct FrtFieldsReader
|
603
|
+
{
|
604
|
+
int size;
|
605
|
+
FrtFieldInfos *fis;
|
606
|
+
FrtStore *store;
|
607
|
+
FrtInStream *fdx_in;
|
608
|
+
FrtInStream *fdt_in;
|
609
|
+
} FrtFieldsReader;
|
610
|
+
|
611
|
+
extern FrtFieldsReader *frt_fr_open(FrtStore *store,
|
612
|
+
const char *segment, FrtFieldInfos *fis);
|
613
|
+
extern FrtFieldsReader *frt_fr_clone(FrtFieldsReader *orig);
|
614
|
+
extern void frt_fr_close(FrtFieldsReader *fr);
|
615
|
+
extern FrtDocument *frt_fr_get_doc(FrtFieldsReader *fr, int doc_num);
|
616
|
+
extern FrtLazyDoc *frt_fr_get_lazy_doc(FrtFieldsReader *fr, int doc_num);
|
617
|
+
extern FrtHash *frt_fr_get_tv(FrtFieldsReader *fr, int doc_num);
|
618
|
+
extern FrtTermVector *frt_fr_get_field_tv(FrtFieldsReader *fr, int doc_num,
|
619
|
+
int field_num);
|
620
|
+
|
621
|
+
/****************************************************************************
|
622
|
+
*
|
623
|
+
* FrtFieldsWriter
|
624
|
+
*
|
625
|
+
****************************************************************************/
|
626
|
+
|
627
|
+
typedef struct FrtFieldsWriter
|
628
|
+
{
|
629
|
+
FrtFieldInfos *fis;
|
630
|
+
FrtOutStream *fdt_out;
|
631
|
+
FrtOutStream *fdx_out;
|
632
|
+
FrtOutStream *buffer;
|
633
|
+
FrtTVField *tv_fields;
|
634
|
+
off_t start_ptr;
|
635
|
+
} FrtFieldsWriter;
|
636
|
+
|
637
|
+
extern FrtFieldsWriter *frt_fw_open(FrtStore *store,
|
638
|
+
const char *segment, FrtFieldInfos *fis);
|
639
|
+
extern void frt_fw_close(FrtFieldsWriter *fw);
|
640
|
+
extern void frt_fw_add_doc(FrtFieldsWriter *fw, FrtDocument *doc);
|
641
|
+
extern void frt_fw_add_postings(FrtFieldsWriter *fw,
|
642
|
+
int field_num,
|
643
|
+
FrtPostingList **plists,
|
644
|
+
int posting_count,
|
645
|
+
FrtOffset *offsets,
|
646
|
+
int offset_count);
|
647
|
+
extern void frt_fw_write_tv_index(FrtFieldsWriter *fw);
|
648
|
+
|
649
|
+
/****************************************************************************
|
650
|
+
*
|
651
|
+
* FrtDeleter
|
652
|
+
*
|
653
|
+
* A utility class (used by both FrtIndexReader and FrtIndexWriter) to keep track of
|
654
|
+
* files that need to be deleted because they are no longer referenced by the
|
655
|
+
* index.
|
656
|
+
*
|
657
|
+
****************************************************************************/
|
658
|
+
|
659
|
+
struct FrtDeleter
|
660
|
+
{
|
661
|
+
FrtStore *store;
|
662
|
+
FrtSegmentInfos *sis;
|
663
|
+
FrtHashSet *pending;
|
664
|
+
};
|
665
|
+
|
666
|
+
extern FrtDeleter *frt_deleter_new(FrtSegmentInfos *sis, FrtStore *store);
|
667
|
+
extern void frt_deleter_destroy(FrtDeleter *dlr);
|
668
|
+
extern void frt_deleter_delete_file(FrtDeleter *dlr, char *file_name);
|
669
|
+
extern void frt_deleter_find_deletable_files(FrtDeleter *dlr);
|
670
|
+
extern void frt_deleter_delete_files(FrtDeleter *dlr, char **files, int file_cnt);
|
671
|
+
|
672
|
+
/****************************************************************************
|
673
|
+
*
|
674
|
+
* FrtIndexReader
|
675
|
+
*
|
676
|
+
****************************************************************************/
|
677
|
+
|
678
|
+
#define FRT_WRITE_LOCK_NAME "write"
|
679
|
+
#define FRT_COMMIT_LOCK_NAME "commit"
|
680
|
+
|
681
|
+
struct FrtIndexReader
|
682
|
+
{
|
683
|
+
int (*num_docs)(FrtIndexReader *ir);
|
684
|
+
int (*max_doc)(FrtIndexReader *ir);
|
685
|
+
FrtDocument *(*get_doc)(FrtIndexReader *ir, int doc_num);
|
686
|
+
FrtLazyDoc *(*get_lazy_doc)(FrtIndexReader *ir, int doc_num);
|
687
|
+
frt_uchar *(*get_norms)(FrtIndexReader *ir, int field_num);
|
688
|
+
frt_uchar *(*get_norms_into)(FrtIndexReader *ir, int field_num,
|
689
|
+
frt_uchar *buf);
|
690
|
+
FrtTermEnum *(*terms)(FrtIndexReader *ir, int field_num);
|
691
|
+
FrtTermEnum *(*terms_from)(FrtIndexReader *ir, int field_num,
|
692
|
+
const char *term);
|
693
|
+
int (*doc_freq)(FrtIndexReader *ir, int field_num,
|
694
|
+
const char *term);
|
695
|
+
FrtTermDocEnum *(*term_docs)(FrtIndexReader *ir);
|
696
|
+
FrtTermDocEnum *(*term_positions)(FrtIndexReader *ir);
|
697
|
+
FrtTermVector *(*term_vector)(FrtIndexReader *ir, int doc_num,
|
698
|
+
FrtSymbol field);
|
699
|
+
FrtHash *(*term_vectors)(FrtIndexReader *ir, int doc_num);
|
700
|
+
bool (*is_deleted)(FrtIndexReader *ir, int doc_num);
|
701
|
+
bool (*has_deletions)(FrtIndexReader *ir);
|
702
|
+
void (*acquire_write_lock)(FrtIndexReader *ir);
|
703
|
+
void (*set_norm_i)(FrtIndexReader *ir, int doc_num,
|
704
|
+
int field_num, frt_uchar val);
|
705
|
+
void (*delete_doc_i)(FrtIndexReader *ir, int doc_num);
|
706
|
+
void (*undelete_all_i)(FrtIndexReader *ir);
|
707
|
+
void (*set_deleter_i)(FrtIndexReader *ir, FrtDeleter *dlr);
|
708
|
+
bool (*is_latest_i)(FrtIndexReader *ir);
|
709
|
+
void (*commit_i)(FrtIndexReader *ir);
|
710
|
+
void (*close_i)(FrtIndexReader *ir);
|
711
|
+
int ref_cnt;
|
712
|
+
FrtDeleter *deleter;
|
713
|
+
FrtStore *store;
|
714
|
+
FrtLock *write_lock;
|
715
|
+
FrtSegmentInfos *sis;
|
716
|
+
FrtFieldInfos *fis;
|
717
|
+
FrtHash *cache;
|
718
|
+
FrtHash *field_index_cache;
|
719
|
+
frt_mutex_t field_index_mutex;
|
720
|
+
frt_uchar *fake_norms;
|
721
|
+
frt_mutex_t mutex;
|
722
|
+
bool has_changes : 1;
|
723
|
+
bool is_stale : 1;
|
724
|
+
bool is_owner : 1;
|
725
|
+
};
|
726
|
+
|
727
|
+
extern FrtIndexReader *frt_ir_open(FrtStore *store);
|
728
|
+
extern void frt_ir_close(FrtIndexReader *ir);
|
729
|
+
extern void frt_ir_commit(FrtIndexReader *ir);
|
730
|
+
extern void frt_ir_delete_doc(FrtIndexReader *ir, int doc_num);
|
731
|
+
extern void frt_ir_undelete_all(FrtIndexReader *ir);
|
732
|
+
extern int frt_ir_doc_freq(FrtIndexReader *ir, FrtSymbol field, const char *term);
|
733
|
+
extern void frt_ir_set_norm(FrtIndexReader *ir, int doc_num, FrtSymbol field, frt_uchar val);
|
734
|
+
extern frt_uchar *frt_ir_get_norms_i(FrtIndexReader *ir, int field_num);
|
735
|
+
extern frt_uchar *frt_ir_get_norms(FrtIndexReader *ir, FrtSymbol field);
|
736
|
+
extern frt_uchar *frt_ir_get_norms_into(FrtIndexReader *ir, FrtSymbol field, frt_uchar *buf);
|
737
|
+
extern FrtDocument *frt_ir_get_doc_with_term(FrtIndexReader *ir, FrtSymbol field, const char *term);
|
738
|
+
extern FrtTermEnum *frt_ir_terms(FrtIndexReader *ir, FrtSymbol field);
|
739
|
+
extern FrtTermEnum *frt_ir_terms_from(FrtIndexReader *ir, FrtSymbol field, const char *t);
|
740
|
+
extern FrtTermDocEnum *ir_term_docs_for(FrtIndexReader *ir, FrtSymbol field, const char *term);
|
741
|
+
extern FrtTermDocEnum *frt_ir_term_positions_for(FrtIndexReader *ir, FrtSymbol field, const char *t);
|
742
|
+
extern void frt_ir_add_cache(FrtIndexReader *ir);
|
743
|
+
extern bool frt_ir_is_latest(FrtIndexReader *ir);
|
744
|
+
|
745
|
+
/****************************************************************************
|
746
|
+
* FrtMultiReader
|
747
|
+
****************************************************************************/
|
748
|
+
|
749
|
+
struct FrtMultiReader {
|
750
|
+
FrtIndexReader ir;
|
751
|
+
int max_doc;
|
752
|
+
int num_docs_cache;
|
753
|
+
int r_cnt;
|
754
|
+
int *starts;
|
755
|
+
FrtIndexReader **sub_readers;
|
756
|
+
FrtHash *norms_cache;
|
757
|
+
bool has_deletions : 1;
|
758
|
+
int **field_num_map;
|
759
|
+
};
|
760
|
+
|
761
|
+
extern int frt_mr_get_field_num(FrtMultiReader *mr, int ir_num, int f_num);
|
762
|
+
extern FrtIndexReader *frt_mr_open(FrtIndexReader **sub_readers, const int r_cnt);
|
763
|
+
|
764
|
+
/****************************************************************************
|
765
|
+
*
|
766
|
+
* FrtFieldInverter
|
767
|
+
*
|
768
|
+
****************************************************************************/
|
769
|
+
|
770
|
+
typedef struct FrtFieldInverter
|
771
|
+
{
|
772
|
+
FrtHash *plists;
|
773
|
+
frt_uchar *norms;
|
774
|
+
FrtFieldInfo *fi;
|
775
|
+
int length;
|
776
|
+
bool is_tokenized : 1;
|
777
|
+
bool store_term_vector : 1;
|
778
|
+
bool store_offsets : 1;
|
779
|
+
bool has_norms : 1;
|
780
|
+
} FrtFieldInverter;
|
781
|
+
|
782
|
+
/****************************************************************************
|
783
|
+
*
|
784
|
+
* FrtDocWriter
|
785
|
+
*
|
786
|
+
****************************************************************************/
|
787
|
+
|
788
|
+
#define DW_OFFSET_INIT_CAPA 512
|
789
|
+
typedef struct FrtIndexWriter FrtIndexWriter;
|
790
|
+
|
791
|
+
typedef struct FrtDocWriter
|
792
|
+
{
|
793
|
+
FrtStore *store;
|
794
|
+
FrtSegmentInfo *si;
|
795
|
+
FrtFieldInfos *fis;
|
796
|
+
FrtFieldsWriter *fw;
|
797
|
+
FrtMemoryPool *mp;
|
798
|
+
FrtAnalyzer *analyzer;
|
799
|
+
FrtHash *curr_plists;
|
800
|
+
FrtHash *fields;
|
801
|
+
FrtSimilarity *similarity;
|
802
|
+
FrtOffset *offsets;
|
803
|
+
int offsets_size;
|
804
|
+
int offsets_capa;
|
805
|
+
int doc_num;
|
806
|
+
int index_interval;
|
807
|
+
int skip_interval;
|
808
|
+
int max_field_length;
|
809
|
+
int max_buffered_docs;
|
810
|
+
} FrtDocWriter;
|
811
|
+
|
812
|
+
extern FrtDocWriter *frt_dw_open(FrtIndexWriter *is, FrtSegmentInfo *si);
|
813
|
+
extern void frt_dw_close(FrtDocWriter *dw);
|
814
|
+
extern void frt_dw_add_doc(FrtDocWriter *dw, FrtDocument *doc);
|
815
|
+
extern void frt_dw_new_segment(FrtDocWriter *dw, FrtSegmentInfo *si);
|
816
|
+
/* For testing. need to remove somehow. FIXME */
|
817
|
+
extern FrtHash *frt_dw_invert_field(FrtDocWriter *dw,
|
818
|
+
FrtFieldInverter *fld_inv,
|
819
|
+
FrtDocField *df);
|
820
|
+
extern FrtFieldInverter *frt_dw_get_fld_inv(FrtDocWriter *dw, FrtFieldInfo *fi);
|
821
|
+
extern void frt_dw_reset_postings(FrtHash *postings);
|
822
|
+
|
823
|
+
/****************************************************************************
|
824
|
+
*
|
825
|
+
* FrtIndexWriter
|
826
|
+
*
|
827
|
+
****************************************************************************/
|
828
|
+
|
829
|
+
struct FrtIndexWriter
|
830
|
+
{
|
831
|
+
FrtConfig config;
|
832
|
+
frt_mutex_t mutex;
|
833
|
+
FrtStore *store;
|
834
|
+
FrtAnalyzer *analyzer;
|
835
|
+
FrtSegmentInfos *sis;
|
836
|
+
FrtFieldInfos *fis;
|
837
|
+
FrtDocWriter *dw;
|
838
|
+
FrtSimilarity *similarity;
|
839
|
+
FrtLock *write_lock;
|
840
|
+
FrtDeleter *deleter;
|
841
|
+
};
|
842
|
+
|
843
|
+
extern void frt_index_create(FrtStore *store, FrtFieldInfos *fis);
|
844
|
+
extern bool frt_index_is_locked(FrtStore *store);
|
845
|
+
extern FrtIndexWriter *frt_iw_open(FrtStore *store, FrtAnalyzer *analyzer, const FrtConfig *config);
|
846
|
+
extern void frt_iw_delete_term(FrtIndexWriter *iw, FrtSymbol field, const char *term);
|
847
|
+
extern void frt_iw_delete_terms(FrtIndexWriter *iw, FrtSymbol field, char **terms, const int term_cnt);
|
848
|
+
extern void frt_iw_close(FrtIndexWriter *iw);
|
849
|
+
extern void frt_iw_add_doc(FrtIndexWriter *iw, FrtDocument *doc);
|
850
|
+
extern int frt_iw_doc_count(FrtIndexWriter *iw);
|
851
|
+
extern void frt_iw_commit(FrtIndexWriter *iw);
|
852
|
+
extern void frt_iw_optimize(FrtIndexWriter *iw);
|
853
|
+
extern void frt_iw_add_readers(FrtIndexWriter *iw, FrtIndexReader **readers, const int r_cnt);
|
854
|
+
|
855
|
+
/****************************************************************************
|
856
|
+
*
|
857
|
+
* FrtCompoundWriter
|
858
|
+
*
|
859
|
+
****************************************************************************/
|
860
|
+
|
861
|
+
#define FRT_CW_INIT_CAPA 16
|
862
|
+
typedef struct FrtCWFileEntry
|
863
|
+
{
|
864
|
+
char *name;
|
865
|
+
off_t dir_offset;
|
866
|
+
off_t data_offset;
|
867
|
+
} FrtCWFileEntry;
|
868
|
+
|
869
|
+
typedef struct FrtCompoundWriter {
|
870
|
+
FrtStore *store;
|
871
|
+
const char *name;
|
872
|
+
FrtHashSet *ids;
|
873
|
+
FrtCWFileEntry *file_entries;
|
874
|
+
} FrtCompoundWriter;
|
875
|
+
|
876
|
+
extern FrtCompoundWriter *frt_open_cw(FrtStore *store, char *name);
|
877
|
+
extern void frt_cw_add_file(FrtCompoundWriter *cw, char *id);
|
878
|
+
extern void frt_cw_close(FrtCompoundWriter *cw);
|
879
|
+
|
880
|
+
#endif
|