jk-ferret 0.11.8.2
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 +90 -0
- data/RELEASE_CHANGES +137 -0
- data/RELEASE_NOTES +60 -0
- data/Rakefile +443 -0
- data/TODO +109 -0
- data/TUTORIAL +231 -0
- data/bin/ferret-browser +79 -0
- data/ext/BZLIB_blocksort.c +1094 -0
- data/ext/BZLIB_bzlib.c +1578 -0
- data/ext/BZLIB_compress.c +672 -0
- data/ext/BZLIB_crctable.c +104 -0
- data/ext/BZLIB_decompress.c +626 -0
- data/ext/BZLIB_huffman.c +205 -0
- data/ext/BZLIB_randtable.c +84 -0
- data/ext/STEMMER_api.c +66 -0
- data/ext/STEMMER_libstemmer.c +93 -0
- data/ext/STEMMER_stem_ISO_8859_1_danish.c +337 -0
- data/ext/STEMMER_stem_ISO_8859_1_dutch.c +624 -0
- data/ext/STEMMER_stem_ISO_8859_1_english.c +1117 -0
- data/ext/STEMMER_stem_ISO_8859_1_finnish.c +762 -0
- data/ext/STEMMER_stem_ISO_8859_1_french.c +1246 -0
- data/ext/STEMMER_stem_ISO_8859_1_german.c +503 -0
- data/ext/STEMMER_stem_ISO_8859_1_hungarian.c +1230 -0
- data/ext/STEMMER_stem_ISO_8859_1_italian.c +1065 -0
- data/ext/STEMMER_stem_ISO_8859_1_norwegian.c +297 -0
- data/ext/STEMMER_stem_ISO_8859_1_porter.c +749 -0
- data/ext/STEMMER_stem_ISO_8859_1_portuguese.c +1017 -0
- data/ext/STEMMER_stem_ISO_8859_1_spanish.c +1093 -0
- data/ext/STEMMER_stem_ISO_8859_1_swedish.c +307 -0
- data/ext/STEMMER_stem_ISO_8859_2_romanian.c +998 -0
- data/ext/STEMMER_stem_KOI8_R_russian.c +700 -0
- data/ext/STEMMER_stem_UTF_8_danish.c +339 -0
- data/ext/STEMMER_stem_UTF_8_dutch.c +634 -0
- data/ext/STEMMER_stem_UTF_8_english.c +1125 -0
- data/ext/STEMMER_stem_UTF_8_finnish.c +768 -0
- data/ext/STEMMER_stem_UTF_8_french.c +1256 -0
- data/ext/STEMMER_stem_UTF_8_german.c +509 -0
- data/ext/STEMMER_stem_UTF_8_hungarian.c +1234 -0
- data/ext/STEMMER_stem_UTF_8_italian.c +1073 -0
- data/ext/STEMMER_stem_UTF_8_norwegian.c +299 -0
- data/ext/STEMMER_stem_UTF_8_porter.c +755 -0
- data/ext/STEMMER_stem_UTF_8_portuguese.c +1023 -0
- data/ext/STEMMER_stem_UTF_8_romanian.c +1004 -0
- data/ext/STEMMER_stem_UTF_8_russian.c +694 -0
- data/ext/STEMMER_stem_UTF_8_spanish.c +1097 -0
- data/ext/STEMMER_stem_UTF_8_swedish.c +309 -0
- data/ext/STEMMER_stem_UTF_8_turkish.c +2205 -0
- data/ext/STEMMER_utilities.c +478 -0
- data/ext/analysis.c +1710 -0
- data/ext/analysis.h +266 -0
- data/ext/api.h +26 -0
- data/ext/array.c +125 -0
- data/ext/array.h +62 -0
- data/ext/bitvector.c +96 -0
- data/ext/bitvector.h +594 -0
- data/ext/bzlib.h +282 -0
- data/ext/bzlib_private.h +503 -0
- data/ext/compound_io.c +384 -0
- data/ext/config.h +52 -0
- data/ext/document.c +159 -0
- data/ext/document.h +63 -0
- data/ext/except.c +102 -0
- data/ext/except.h +176 -0
- data/ext/extconf.rb +15 -0
- data/ext/ferret.c +416 -0
- data/ext/ferret.h +94 -0
- data/ext/field_index.c +262 -0
- data/ext/field_index.h +52 -0
- data/ext/filter.c +157 -0
- data/ext/fs_store.c +493 -0
- data/ext/global.c +458 -0
- data/ext/global.h +302 -0
- data/ext/hash.c +524 -0
- data/ext/hash.h +515 -0
- data/ext/hashset.c +192 -0
- data/ext/hashset.h +215 -0
- data/ext/header.h +58 -0
- data/ext/helper.c +63 -0
- data/ext/helper.h +21 -0
- data/ext/index.c +6804 -0
- data/ext/index.h +935 -0
- data/ext/internal.h +1019 -0
- data/ext/lang.c +10 -0
- data/ext/lang.h +68 -0
- data/ext/libstemmer.h +79 -0
- data/ext/mempool.c +88 -0
- data/ext/mempool.h +43 -0
- data/ext/modules.h +190 -0
- data/ext/multimapper.c +351 -0
- data/ext/multimapper.h +60 -0
- data/ext/posh.c +1006 -0
- data/ext/posh.h +973 -0
- data/ext/priorityqueue.c +149 -0
- data/ext/priorityqueue.h +155 -0
- data/ext/q_boolean.c +1621 -0
- data/ext/q_const_score.c +162 -0
- data/ext/q_filtered_query.c +212 -0
- data/ext/q_fuzzy.c +280 -0
- data/ext/q_match_all.c +149 -0
- data/ext/q_multi_term.c +673 -0
- data/ext/q_parser.c +3103 -0
- data/ext/q_phrase.c +1206 -0
- data/ext/q_prefix.c +98 -0
- data/ext/q_range.c +682 -0
- data/ext/q_span.c +2390 -0
- data/ext/q_term.c +337 -0
- data/ext/q_wildcard.c +167 -0
- data/ext/r_analysis.c +2626 -0
- data/ext/r_index.c +3468 -0
- data/ext/r_qparser.c +635 -0
- data/ext/r_search.c +4490 -0
- data/ext/r_store.c +513 -0
- data/ext/r_utils.c +1131 -0
- data/ext/ram_store.c +476 -0
- data/ext/scanner.c +895 -0
- data/ext/scanner.h +36 -0
- data/ext/scanner_mb.c +6701 -0
- data/ext/scanner_utf8.c +4415 -0
- data/ext/search.c +1864 -0
- data/ext/search.h +953 -0
- data/ext/similarity.c +151 -0
- data/ext/similarity.h +89 -0
- data/ext/sort.c +786 -0
- data/ext/stem_ISO_8859_1_danish.h +16 -0
- data/ext/stem_ISO_8859_1_dutch.h +16 -0
- data/ext/stem_ISO_8859_1_english.h +16 -0
- data/ext/stem_ISO_8859_1_finnish.h +16 -0
- data/ext/stem_ISO_8859_1_french.h +16 -0
- data/ext/stem_ISO_8859_1_german.h +16 -0
- data/ext/stem_ISO_8859_1_hungarian.h +16 -0
- data/ext/stem_ISO_8859_1_italian.h +16 -0
- data/ext/stem_ISO_8859_1_norwegian.h +16 -0
- data/ext/stem_ISO_8859_1_porter.h +16 -0
- data/ext/stem_ISO_8859_1_portuguese.h +16 -0
- data/ext/stem_ISO_8859_1_spanish.h +16 -0
- data/ext/stem_ISO_8859_1_swedish.h +16 -0
- data/ext/stem_ISO_8859_2_romanian.h +16 -0
- data/ext/stem_KOI8_R_russian.h +16 -0
- data/ext/stem_UTF_8_danish.h +16 -0
- data/ext/stem_UTF_8_dutch.h +16 -0
- data/ext/stem_UTF_8_english.h +16 -0
- data/ext/stem_UTF_8_finnish.h +16 -0
- data/ext/stem_UTF_8_french.h +16 -0
- data/ext/stem_UTF_8_german.h +16 -0
- data/ext/stem_UTF_8_hungarian.h +16 -0
- data/ext/stem_UTF_8_italian.h +16 -0
- data/ext/stem_UTF_8_norwegian.h +16 -0
- data/ext/stem_UTF_8_porter.h +16 -0
- data/ext/stem_UTF_8_portuguese.h +16 -0
- data/ext/stem_UTF_8_romanian.h +16 -0
- data/ext/stem_UTF_8_russian.h +16 -0
- data/ext/stem_UTF_8_spanish.h +16 -0
- data/ext/stem_UTF_8_swedish.h +16 -0
- data/ext/stem_UTF_8_turkish.h +16 -0
- data/ext/stopwords.c +410 -0
- data/ext/store.c +698 -0
- data/ext/store.h +799 -0
- data/ext/symbol.c +10 -0
- data/ext/symbol.h +23 -0
- data/ext/term_vectors.c +73 -0
- data/ext/threading.h +31 -0
- data/ext/win32.h +62 -0
- data/lib/ferret.rb +30 -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/field_symbol.rb +87 -0
- data/lib/ferret/index.rb +973 -0
- data/lib/ferret/number_tools.rb +157 -0
- data/lib/ferret/version.rb +3 -0
- data/setup.rb +1555 -0
- data/test/long_running/largefile/tc_largefile.rb +46 -0
- data/test/test_all.rb +5 -0
- data/test/test_helper.rb +29 -0
- data/test/test_installed.rb +1 -0
- data/test/threading/number_to_spoken.rb +132 -0
- data/test/threading/thread_safety_index_test.rb +88 -0
- data/test/threading/thread_safety_read_write_test.rb +73 -0
- data/test/threading/thread_safety_test.rb +133 -0
- data/test/unit/analysis/tc_analyzer.rb +550 -0
- data/test/unit/analysis/tc_token_stream.rb +653 -0
- data/test/unit/index/tc_index.rb +867 -0
- data/test/unit/index/tc_index_reader.rb +699 -0
- data/test/unit/index/tc_index_writer.rb +447 -0
- data/test/unit/index/th_doc.rb +332 -0
- data/test/unit/query_parser/tc_query_parser.rb +238 -0
- data/test/unit/search/tc_filter.rb +156 -0
- data/test/unit/search/tc_fuzzy_query.rb +147 -0
- data/test/unit/search/tc_index_searcher.rb +67 -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 +436 -0
- data/test/unit/store/tc_fs_store.rb +115 -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/tc_field_symbol.rb +26 -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
- data/test/utils/content_generator.rb +226 -0
- metadata +319 -0
data/ext/index.h
ADDED
@@ -0,0 +1,935 @@
|
|
1
|
+
#ifndef FRT_INDEX_H
|
2
|
+
#define FRT_INDEX_H
|
3
|
+
|
4
|
+
#ifdef __cplusplus
|
5
|
+
extern "C" {
|
6
|
+
#endif
|
7
|
+
|
8
|
+
#include "global.h"
|
9
|
+
#include "document.h"
|
10
|
+
#include "analysis.h"
|
11
|
+
#include "hash.h"
|
12
|
+
#include "hashset.h"
|
13
|
+
#include "store.h"
|
14
|
+
#include "mempool.h"
|
15
|
+
#include "similarity.h"
|
16
|
+
#include "bitvector.h"
|
17
|
+
#include "priorityqueue.h"
|
18
|
+
|
19
|
+
typedef struct FrtIndexReader FrtIndexReader;
|
20
|
+
typedef struct FrtMultiReader FrtMultiReader;
|
21
|
+
typedef struct FrtDeleter FrtDeleter;
|
22
|
+
|
23
|
+
extern bool frt_file_name_filter_is_index_file(const char *file_name, bool include_locks);
|
24
|
+
|
25
|
+
/****************************************************************************
|
26
|
+
*
|
27
|
+
* FrtConfig
|
28
|
+
*
|
29
|
+
****************************************************************************/
|
30
|
+
|
31
|
+
typedef struct FrtConfig
|
32
|
+
{
|
33
|
+
int chunk_size;
|
34
|
+
int max_buffer_memory;
|
35
|
+
int index_interval;
|
36
|
+
int skip_interval;
|
37
|
+
int merge_factor;
|
38
|
+
int max_buffered_docs;
|
39
|
+
int max_merge_docs;
|
40
|
+
int max_field_length;
|
41
|
+
bool use_compound_file;
|
42
|
+
} FrtConfig;
|
43
|
+
|
44
|
+
extern const FrtConfig frt_default_config;
|
45
|
+
|
46
|
+
/***************************************************************************
|
47
|
+
*
|
48
|
+
* FrtCacheObject
|
49
|
+
*
|
50
|
+
***************************************************************************/
|
51
|
+
|
52
|
+
typedef struct FrtCacheObject {
|
53
|
+
FrtHash *ref_tab1;
|
54
|
+
FrtHash *ref_tab2;
|
55
|
+
void *ref1;
|
56
|
+
void *ref2;
|
57
|
+
void *obj;
|
58
|
+
void (*destroy)(void *p);
|
59
|
+
} FrtCacheObject;
|
60
|
+
|
61
|
+
extern void frt_cache_destroy(FrtCacheObject *co);
|
62
|
+
extern FrtCacheObject *frt_co_create(FrtHash *ref_tab1,
|
63
|
+
FrtHash *ref_tab2,
|
64
|
+
void *ref1, void *ref2, frt_free_ft destroy, void *obj);
|
65
|
+
extern FrtHash *frt_co_hash_create();
|
66
|
+
|
67
|
+
/****************************************************************************
|
68
|
+
*
|
69
|
+
* FrtFieldInfo
|
70
|
+
*
|
71
|
+
****************************************************************************/
|
72
|
+
|
73
|
+
typedef enum
|
74
|
+
{
|
75
|
+
FRT_STORE_NO = 0,
|
76
|
+
FRT_STORE_YES = 1,
|
77
|
+
FRT_STORE_COMPRESS = 2
|
78
|
+
} FrtStoreValue;
|
79
|
+
|
80
|
+
typedef enum
|
81
|
+
{
|
82
|
+
FRT_INDEX_NO = 0,
|
83
|
+
FRT_INDEX_UNTOKENIZED = 1,
|
84
|
+
FRT_INDEX_YES = 3,
|
85
|
+
FRT_INDEX_UNTOKENIZED_OMIT_NORMS = 5,
|
86
|
+
FRT_INDEX_YES_OMIT_NORMS = 7
|
87
|
+
} FrtIndexValue;
|
88
|
+
|
89
|
+
typedef enum
|
90
|
+
{
|
91
|
+
FRT_TERM_VECTOR_NO = 0,
|
92
|
+
FRT_TERM_VECTOR_YES = 1,
|
93
|
+
FRT_TERM_VECTOR_WITH_POSITIONS = 3,
|
94
|
+
FRT_TERM_VECTOR_WITH_OFFSETS = 5,
|
95
|
+
FRT_TERM_VECTOR_WITH_POSITIONS_OFFSETS = 7
|
96
|
+
} FrtTermVectorValue;
|
97
|
+
|
98
|
+
#define FRT_FI_IS_STORED_BM 0x001
|
99
|
+
#define FRT_FI_IS_COMPRESSED_BM 0x002
|
100
|
+
#define FRT_FI_IS_INDEXED_BM 0x004
|
101
|
+
#define FRT_FI_IS_TOKENIZED_BM 0x008
|
102
|
+
#define FRT_FI_OMIT_NORMS_BM 0x010
|
103
|
+
#define FRT_FI_STORE_TERM_VECTOR_BM 0x020
|
104
|
+
#define FRT_FI_STORE_POSITIONS_BM 0x040
|
105
|
+
#define FRT_FI_STORE_OFFSETS_BM 0x080
|
106
|
+
|
107
|
+
typedef struct FrtFieldInfo
|
108
|
+
{
|
109
|
+
FrtSymbol name;
|
110
|
+
float boost;
|
111
|
+
unsigned int bits;
|
112
|
+
int number;
|
113
|
+
int ref_cnt;
|
114
|
+
} FrtFieldInfo;
|
115
|
+
|
116
|
+
extern FrtFieldInfo *frt_fi_new(FrtSymbol name,
|
117
|
+
FrtStoreValue store,
|
118
|
+
FrtIndexValue index,
|
119
|
+
FrtTermVectorValue term_vector);
|
120
|
+
extern char *frt_fi_to_s(FrtFieldInfo *fi);
|
121
|
+
extern void frt_fi_deref(FrtFieldInfo *fi);
|
122
|
+
|
123
|
+
#define fi_is_stored(fi) (((fi)->bits & FRT_FI_IS_STORED_BM) != 0)
|
124
|
+
#define fi_is_compressed(fi) (((fi)->bits & FRT_FI_IS_COMPRESSED_BM) != 0)
|
125
|
+
#define fi_is_indexed(fi) (((fi)->bits & FRT_FI_IS_INDEXED_BM) != 0)
|
126
|
+
#define fi_is_tokenized(fi) (((fi)->bits & FRT_FI_IS_TOKENIZED_BM) != 0)
|
127
|
+
#define fi_omit_norms(fi) (((fi)->bits & FRT_FI_OMIT_NORMS_BM) != 0)
|
128
|
+
#define fi_store_term_vector(fi) (((fi)->bits & FRT_FI_STORE_TERM_VECTOR_BM) != 0)
|
129
|
+
#define fi_store_positions(fi) (((fi)->bits & FRT_FI_STORE_POSITIONS_BM) != 0)
|
130
|
+
#define fi_store_offsets(fi) (((fi)->bits & FRT_FI_STORE_OFFSETS_BM) != 0)
|
131
|
+
#define fi_has_norms(fi)\
|
132
|
+
(((fi)->bits & (FRT_FI_OMIT_NORMS_BM|FRT_FI_IS_INDEXED_BM)) == FRT_FI_IS_INDEXED_BM)
|
133
|
+
|
134
|
+
/****************************************************************************
|
135
|
+
*
|
136
|
+
* FrtFieldInfos
|
137
|
+
*
|
138
|
+
****************************************************************************/
|
139
|
+
|
140
|
+
#define FIELD_INFOS_INIT_CAPA 4
|
141
|
+
/* carry changes over to dummy_fis in test/test_segments.c */
|
142
|
+
typedef struct FrtFieldInfos
|
143
|
+
{
|
144
|
+
FrtStoreValue store;
|
145
|
+
FrtIndexValue index;
|
146
|
+
FrtTermVectorValue term_vector;
|
147
|
+
int size;
|
148
|
+
int capa;
|
149
|
+
FrtFieldInfo **fields;
|
150
|
+
FrtHash *field_dict;
|
151
|
+
int ref_cnt;
|
152
|
+
} FrtFieldInfos;
|
153
|
+
|
154
|
+
FrtFieldInfos *frt_fis_new(FrtStoreValue store, FrtIndexValue index,
|
155
|
+
FrtTermVectorValue term_vector);
|
156
|
+
extern FrtFieldInfo *frt_fis_add_field(FrtFieldInfos *fis, FrtFieldInfo *fi);
|
157
|
+
extern FrtFieldInfo *frt_fis_get_field(FrtFieldInfos *fis, FrtSymbol name);
|
158
|
+
extern int frt_fis_get_field_num(FrtFieldInfos *fis, FrtSymbol name);
|
159
|
+
extern FrtFieldInfo *frt_fis_by_number(FrtFieldInfos *fis, int num);
|
160
|
+
extern FrtFieldInfo *frt_fis_get_or_add_field(FrtFieldInfos *fis,
|
161
|
+
FrtSymbol name);
|
162
|
+
extern void frt_fis_write(FrtFieldInfos *fis, FrtOutStream *os);
|
163
|
+
extern FrtFieldInfos *frt_fis_read(FrtInStream *is);
|
164
|
+
extern char *frt_fis_to_s(FrtFieldInfos *fis);
|
165
|
+
extern void frt_fis_deref(FrtFieldInfos *fis);
|
166
|
+
|
167
|
+
/****************************************************************************
|
168
|
+
*
|
169
|
+
* FrtSegmentInfo
|
170
|
+
*
|
171
|
+
****************************************************************************/
|
172
|
+
|
173
|
+
#define FRT_SEGMENT_NAME_MAX_LENGTH 100
|
174
|
+
#define FRT_SEGMENTS_FILE_NAME "segments"
|
175
|
+
|
176
|
+
typedef struct FrtSegmentInfo
|
177
|
+
{
|
178
|
+
int ref_cnt;
|
179
|
+
char *name;
|
180
|
+
FrtStore *store;
|
181
|
+
int doc_cnt;
|
182
|
+
int del_gen;
|
183
|
+
int *norm_gens;
|
184
|
+
int norm_gens_size;
|
185
|
+
bool use_compound_file;
|
186
|
+
} FrtSegmentInfo;
|
187
|
+
|
188
|
+
extern FrtSegmentInfo *frt_si_new(char *name, int doc_cnt, FrtStore *store);
|
189
|
+
extern void frt_si_deref(FrtSegmentInfo *si);
|
190
|
+
extern bool frt_si_has_deletions(FrtSegmentInfo *si);
|
191
|
+
extern bool frt_si_uses_compound_file(FrtSegmentInfo *si);
|
192
|
+
extern bool frt_si_has_separate_norms(FrtSegmentInfo *si);
|
193
|
+
extern void frt_si_advance_norm_gen(FrtSegmentInfo *si, int field_num);
|
194
|
+
|
195
|
+
/****************************************************************************
|
196
|
+
*
|
197
|
+
* FrtSegmentInfos
|
198
|
+
*
|
199
|
+
****************************************************************************/
|
200
|
+
|
201
|
+
typedef struct FrtSegmentInfos
|
202
|
+
{
|
203
|
+
FrtFieldInfos *fis;
|
204
|
+
frt_u64 counter;
|
205
|
+
frt_u64 version;
|
206
|
+
frt_i64 generation;
|
207
|
+
frt_i32 format;
|
208
|
+
FrtStore *store;
|
209
|
+
FrtSegmentInfo **segs;
|
210
|
+
int size;
|
211
|
+
int capa;
|
212
|
+
} FrtSegmentInfos;
|
213
|
+
|
214
|
+
extern char *frt_fn_for_generation(char *buf, char *base, char *ext, frt_i64 gen);
|
215
|
+
|
216
|
+
extern FrtSegmentInfos *frt_sis_new(FrtFieldInfos *fis);
|
217
|
+
extern FrtSegmentInfo *frt_sis_new_segment(FrtSegmentInfos *sis, int dcnt, FrtStore *store);
|
218
|
+
extern FrtSegmentInfo *frt_sis_add_si(FrtSegmentInfos *sis, FrtSegmentInfo *si);
|
219
|
+
extern void frt_sis_del_at(FrtSegmentInfos *sis, int at);
|
220
|
+
extern void frt_sis_del_from_to(FrtSegmentInfos *sis, int from, int to);
|
221
|
+
extern void frt_sis_clear(FrtSegmentInfos *sis);
|
222
|
+
extern FrtSegmentInfos *frt_sis_read(FrtStore *store);
|
223
|
+
extern void frt_sis_write(FrtSegmentInfos *sis, FrtStore *store, FrtDeleter *deleter);
|
224
|
+
extern frt_u64 frt_sis_read_current_version(FrtStore *store);
|
225
|
+
extern void frt_sis_destroy(FrtSegmentInfos *sis);
|
226
|
+
extern frt_i64 frt_sis_current_segment_generation(FrtStore *store);
|
227
|
+
extern char *frt_sis_curr_seg_file_name(char *buf, FrtStore *store);
|
228
|
+
extern void frt_sis_put(FrtSegmentInfos *sis, FILE *stream);
|
229
|
+
|
230
|
+
/****************************************************************************
|
231
|
+
*
|
232
|
+
* FrtTermInfo
|
233
|
+
*
|
234
|
+
****************************************************************************/
|
235
|
+
|
236
|
+
typedef struct FrtTermInfo
|
237
|
+
{
|
238
|
+
int doc_freq;
|
239
|
+
off_t frq_ptr;
|
240
|
+
off_t prx_ptr;
|
241
|
+
off_t skip_offset;
|
242
|
+
} FrtTermInfo;
|
243
|
+
|
244
|
+
#define frt_ti_set(ti, mdf, mfp, mpp, mso) do {\
|
245
|
+
(ti).doc_freq = mdf;\
|
246
|
+
(ti).frq_ptr = mfp;\
|
247
|
+
(ti).prx_ptr = mpp;\
|
248
|
+
(ti).skip_offset = mso;\
|
249
|
+
} while (0)
|
250
|
+
|
251
|
+
/****************************************************************************
|
252
|
+
*
|
253
|
+
* FrtTermEnum
|
254
|
+
*
|
255
|
+
****************************************************************************/
|
256
|
+
|
257
|
+
typedef struct FrtTermEnum FrtTermEnum;
|
258
|
+
|
259
|
+
struct FrtTermEnum
|
260
|
+
{
|
261
|
+
char curr_term[FRT_MAX_WORD_SIZE];
|
262
|
+
char prev_term[FRT_MAX_WORD_SIZE];
|
263
|
+
FrtTermInfo curr_ti;
|
264
|
+
int curr_term_len;
|
265
|
+
int field_num;
|
266
|
+
FrtTermEnum *(*set_field)(FrtTermEnum *te, int field_num);
|
267
|
+
char *(*next)(FrtTermEnum *te);
|
268
|
+
char *(*skip_to)(FrtTermEnum *te, const char *term);
|
269
|
+
void (*close)(FrtTermEnum *te);
|
270
|
+
FrtTermEnum *(*clone)(FrtTermEnum *te);
|
271
|
+
};
|
272
|
+
|
273
|
+
char *frt_te_get_term(struct FrtTermEnum *te);
|
274
|
+
FrtTermInfo *frt_te_get_ti(struct FrtTermEnum *te);
|
275
|
+
|
276
|
+
/****************************************************************************
|
277
|
+
*
|
278
|
+
* FrtSegmentTermEnum
|
279
|
+
*
|
280
|
+
****************************************************************************/
|
281
|
+
|
282
|
+
/* * FrtSegmentTermIndex * */
|
283
|
+
|
284
|
+
typedef struct FrtSegmentTermIndex
|
285
|
+
{
|
286
|
+
off_t index_ptr;
|
287
|
+
off_t ptr;
|
288
|
+
int index_cnt;
|
289
|
+
int size;
|
290
|
+
char **index_terms;
|
291
|
+
int *index_term_lens;
|
292
|
+
FrtTermInfo *index_term_infos;
|
293
|
+
off_t *index_ptrs;
|
294
|
+
} FrtSegmentTermIndex;
|
295
|
+
|
296
|
+
/* * FrtSegmentFieldIndex * */
|
297
|
+
|
298
|
+
typedef struct FrtSegmentTermEnum FrtSegmentTermEnum;
|
299
|
+
|
300
|
+
typedef struct FrtSegmentFieldIndex
|
301
|
+
{
|
302
|
+
frt_mutex_t mutex;
|
303
|
+
int skip_interval;
|
304
|
+
int index_interval;
|
305
|
+
off_t index_ptr;
|
306
|
+
FrtTermEnum *index_te;
|
307
|
+
FrtHash *field_dict;
|
308
|
+
} FrtSegmentFieldIndex;
|
309
|
+
|
310
|
+
extern FrtSegmentFieldIndex *frt_sfi_open(FrtStore *store, const char *segment);
|
311
|
+
extern void frt_sfi_close(FrtSegmentFieldIndex *sfi);
|
312
|
+
|
313
|
+
|
314
|
+
/* * FrtSegmentTermEnum * */
|
315
|
+
struct FrtSegmentTermEnum
|
316
|
+
{
|
317
|
+
FrtTermEnum te;
|
318
|
+
FrtInStream *is;
|
319
|
+
int size;
|
320
|
+
int pos;
|
321
|
+
int skip_interval;
|
322
|
+
FrtSegmentFieldIndex *sfi;
|
323
|
+
};
|
324
|
+
|
325
|
+
extern void frt_ste_close(FrtTermEnum *te);
|
326
|
+
extern FrtTermEnum *frt_ste_clone(FrtTermEnum *te);
|
327
|
+
extern FrtTermEnum *frt_ste_new(FrtInStream *is, FrtSegmentFieldIndex *sfi);
|
328
|
+
|
329
|
+
/* * MultiTermEnum * */
|
330
|
+
|
331
|
+
extern FrtTermEnum *frt_mte_new(FrtMultiReader *mr, int field_num, const char *term);
|
332
|
+
|
333
|
+
/****************************************************************************
|
334
|
+
*
|
335
|
+
* FrtTermInfosReader
|
336
|
+
*
|
337
|
+
****************************************************************************/
|
338
|
+
|
339
|
+
#define FRT_TE_BUCKET_INIT_CAPA 1
|
340
|
+
|
341
|
+
typedef struct FrtTermInfosReader
|
342
|
+
{
|
343
|
+
frt_thread_key_t thread_te;
|
344
|
+
void **te_bucket;
|
345
|
+
FrtTermEnum *orig_te;
|
346
|
+
int field_num;
|
347
|
+
} FrtTermInfosReader;
|
348
|
+
|
349
|
+
extern FrtTermInfosReader *frt_tir_open(FrtStore *store,
|
350
|
+
FrtSegmentFieldIndex *sfi,
|
351
|
+
const char *segment);
|
352
|
+
extern FrtTermInfosReader *frt_tir_set_field(FrtTermInfosReader *tir, int field_num);
|
353
|
+
extern FrtTermInfo *frt_tir_get_ti(FrtTermInfosReader *tir, const char *term);
|
354
|
+
extern char *frt_tir_get_term(FrtTermInfosReader *tir, int pos);
|
355
|
+
extern void frt_tir_close(FrtTermInfosReader *tir);
|
356
|
+
|
357
|
+
/****************************************************************************
|
358
|
+
*
|
359
|
+
* FrtTermInfosWriter
|
360
|
+
*
|
361
|
+
****************************************************************************/
|
362
|
+
|
363
|
+
#define FRT_INDEX_INTERVAL 128
|
364
|
+
#define FRT_SKIP_INTERVAL 16
|
365
|
+
|
366
|
+
typedef struct FrtTermWriter
|
367
|
+
{
|
368
|
+
int counter;
|
369
|
+
const char *last_term;
|
370
|
+
FrtTermInfo last_term_info;
|
371
|
+
FrtOutStream *os;
|
372
|
+
} FrtTermWriter;
|
373
|
+
|
374
|
+
typedef struct FrtTermInfosWriter
|
375
|
+
{
|
376
|
+
int field_count;
|
377
|
+
int index_interval;
|
378
|
+
int skip_interval;
|
379
|
+
off_t last_index_ptr;
|
380
|
+
FrtOutStream *tfx_out;
|
381
|
+
FrtTermWriter *tix_writer;
|
382
|
+
FrtTermWriter *tis_writer;
|
383
|
+
} FrtTermInfosWriter;
|
384
|
+
|
385
|
+
extern FrtTermInfosWriter *frt_tiw_open(FrtStore *store,
|
386
|
+
const char *segment,
|
387
|
+
int index_interval,
|
388
|
+
int skip_interval);
|
389
|
+
extern void frt_tiw_start_field(FrtTermInfosWriter *tiw, int field_num);
|
390
|
+
extern void frt_tiw_add(FrtTermInfosWriter *tiw,
|
391
|
+
const char *term,
|
392
|
+
int t_len,
|
393
|
+
FrtTermInfo *ti);
|
394
|
+
extern void frt_tiw_close(FrtTermInfosWriter *tiw);
|
395
|
+
|
396
|
+
/****************************************************************************
|
397
|
+
*
|
398
|
+
* FrtTermDocEnum
|
399
|
+
*
|
400
|
+
****************************************************************************/
|
401
|
+
|
402
|
+
typedef struct FrtTermDocEnum FrtTermDocEnum;
|
403
|
+
struct FrtTermDocEnum
|
404
|
+
{
|
405
|
+
void (*seek)(FrtTermDocEnum *tde, int field_num, const char *term);
|
406
|
+
void (*seek_te)(FrtTermDocEnum *tde, FrtTermEnum *te);
|
407
|
+
void (*seek_ti)(FrtTermDocEnum *tde, FrtTermInfo *ti);
|
408
|
+
int (*doc_num)(FrtTermDocEnum *tde);
|
409
|
+
int (*freq)(FrtTermDocEnum *tde);
|
410
|
+
bool (*next)(FrtTermDocEnum *tde);
|
411
|
+
int (*read)(FrtTermDocEnum *tde, int *docs, int *freqs, int req_num);
|
412
|
+
bool (*skip_to)(FrtTermDocEnum *tde, int target);
|
413
|
+
int (*next_position)(FrtTermDocEnum *tde);
|
414
|
+
void (*close)(FrtTermDocEnum *tde);
|
415
|
+
};
|
416
|
+
|
417
|
+
/* * FrtSegmentTermDocEnum * */
|
418
|
+
|
419
|
+
typedef struct FrtSegmentTermDocEnum FrtSegmentTermDocEnum;
|
420
|
+
struct FrtSegmentTermDocEnum
|
421
|
+
{
|
422
|
+
FrtTermDocEnum tde;
|
423
|
+
void (*seek_prox)(FrtSegmentTermDocEnum *stde, off_t prx_ptr);
|
424
|
+
void (*skip_prox)(FrtSegmentTermDocEnum *stde);
|
425
|
+
FrtTermInfosReader *tir;
|
426
|
+
FrtInStream *frq_in;
|
427
|
+
FrtInStream *prx_in;
|
428
|
+
FrtInStream *skip_in;
|
429
|
+
FrtBitVector *deleted_docs;
|
430
|
+
int count; /* number of docs for this term skipped */
|
431
|
+
int doc_freq; /* number of doc this term appears in */
|
432
|
+
int doc_num;
|
433
|
+
int freq;
|
434
|
+
int num_skips;
|
435
|
+
int skip_interval;
|
436
|
+
int skip_count;
|
437
|
+
int skip_doc;
|
438
|
+
int prx_cnt;
|
439
|
+
int position;
|
440
|
+
off_t frq_ptr;
|
441
|
+
off_t prx_ptr;
|
442
|
+
off_t skip_ptr;
|
443
|
+
bool have_skipped : 1;
|
444
|
+
};
|
445
|
+
|
446
|
+
extern FrtTermDocEnum *frt_stde_new(FrtTermInfosReader *tir, FrtInStream *frq_in,
|
447
|
+
FrtBitVector *deleted_docs, int skip_interval);
|
448
|
+
|
449
|
+
/* * FrtSegmentTermDocEnum * */
|
450
|
+
extern FrtTermDocEnum *frt_stpe_new(FrtTermInfosReader *tir, FrtInStream *frq_in,
|
451
|
+
FrtInStream *prx_in, FrtBitVector *deleted_docs,
|
452
|
+
int skip_interval);
|
453
|
+
|
454
|
+
/****************************************************************************
|
455
|
+
* MultipleTermDocPosEnum
|
456
|
+
****************************************************************************/
|
457
|
+
|
458
|
+
extern FrtTermDocEnum *frt_mtdpe_new(FrtIndexReader *ir, int field_num, char **terms,
|
459
|
+
int t_cnt);
|
460
|
+
|
461
|
+
/****************************************************************************
|
462
|
+
*
|
463
|
+
* FrtOffset
|
464
|
+
*
|
465
|
+
****************************************************************************/
|
466
|
+
|
467
|
+
typedef struct FrtOffset
|
468
|
+
{
|
469
|
+
off_t start;
|
470
|
+
off_t end;
|
471
|
+
} FrtOffset;
|
472
|
+
|
473
|
+
extern FrtOffset *frt_offset_new(off_t start, off_t end);
|
474
|
+
|
475
|
+
/****************************************************************************
|
476
|
+
*
|
477
|
+
* FrtOccurence
|
478
|
+
*
|
479
|
+
****************************************************************************/
|
480
|
+
|
481
|
+
typedef struct FrtOccurence
|
482
|
+
{
|
483
|
+
struct FrtOccurence *next;
|
484
|
+
int pos;
|
485
|
+
} FrtOccurence;
|
486
|
+
|
487
|
+
/****************************************************************************
|
488
|
+
*
|
489
|
+
* FrtPosting
|
490
|
+
*
|
491
|
+
****************************************************************************/
|
492
|
+
|
493
|
+
typedef struct FrtPosting
|
494
|
+
{
|
495
|
+
int freq;
|
496
|
+
int doc_num;
|
497
|
+
FrtOccurence *first_occ;
|
498
|
+
struct FrtPosting *next;
|
499
|
+
} FrtPosting;
|
500
|
+
|
501
|
+
extern FrtPosting *frt_p_new(FrtMemoryPool *mp, int doc_num, int pos);
|
502
|
+
|
503
|
+
/****************************************************************************
|
504
|
+
*
|
505
|
+
* FrtPostingList
|
506
|
+
*
|
507
|
+
****************************************************************************/
|
508
|
+
|
509
|
+
typedef struct FrtPostingList
|
510
|
+
{
|
511
|
+
const char *term;
|
512
|
+
int term_len;
|
513
|
+
FrtPosting *first;
|
514
|
+
FrtPosting *last;
|
515
|
+
FrtOccurence *last_occ;
|
516
|
+
} FrtPostingList;
|
517
|
+
|
518
|
+
extern FrtPostingList *frt_pl_new(FrtMemoryPool *mp, const char *term,
|
519
|
+
int term_len, FrtPosting *p);
|
520
|
+
extern void frt_pl_add_occ(FrtMemoryPool *mp, FrtPostingList *pl, int pos);
|
521
|
+
extern int frt_pl_cmp(const FrtPostingList **pl1, const FrtPostingList **pl2);
|
522
|
+
|
523
|
+
/****************************************************************************
|
524
|
+
*
|
525
|
+
* FrtTVField
|
526
|
+
*
|
527
|
+
****************************************************************************/
|
528
|
+
|
529
|
+
typedef struct FrtTVField
|
530
|
+
{
|
531
|
+
int field_num;
|
532
|
+
int size;
|
533
|
+
} FrtTVField;
|
534
|
+
|
535
|
+
/****************************************************************************
|
536
|
+
*
|
537
|
+
* FrtTVTerm
|
538
|
+
*
|
539
|
+
****************************************************************************/
|
540
|
+
|
541
|
+
typedef struct FrtTVTerm
|
542
|
+
{
|
543
|
+
char *text;
|
544
|
+
int freq;
|
545
|
+
int *positions;
|
546
|
+
} FrtTVTerm;
|
547
|
+
|
548
|
+
/****************************************************************************
|
549
|
+
*
|
550
|
+
* FrtTermVector
|
551
|
+
*
|
552
|
+
****************************************************************************/
|
553
|
+
|
554
|
+
#define FRT_TV_FIELD_INIT_CAPA 8
|
555
|
+
typedef struct FrtTermVector
|
556
|
+
{
|
557
|
+
int field_num;
|
558
|
+
FrtSymbol field;
|
559
|
+
int term_cnt;
|
560
|
+
FrtTVTerm *terms;
|
561
|
+
int offset_cnt;
|
562
|
+
FrtOffset *offsets;
|
563
|
+
} FrtTermVector;
|
564
|
+
|
565
|
+
extern void frt_tv_destroy(FrtTermVector *tv);
|
566
|
+
extern int frt_tv_get_term_index(FrtTermVector *tv, const char *term);
|
567
|
+
extern int frt_tv_scan_to_term_index(FrtTermVector *tv, const char *term);
|
568
|
+
extern FrtTVTerm *frt_tv_get_tv_term(FrtTermVector *tv, const char *term);
|
569
|
+
|
570
|
+
/****************************************************************************
|
571
|
+
*
|
572
|
+
* FrtLazyDoc
|
573
|
+
*
|
574
|
+
****************************************************************************/
|
575
|
+
|
576
|
+
/* * * FrtLazyDocField * * */
|
577
|
+
typedef struct FrtLazyDocFieldData
|
578
|
+
{
|
579
|
+
off_t start;
|
580
|
+
int length;
|
581
|
+
char *text;
|
582
|
+
} FrtLazyDocFieldData;
|
583
|
+
|
584
|
+
typedef struct FrtLazyDoc FrtLazyDoc;
|
585
|
+
typedef struct FrtLazyDocField
|
586
|
+
{
|
587
|
+
FrtSymbol name;
|
588
|
+
FrtLazyDocFieldData *data;
|
589
|
+
FrtLazyDoc *doc;
|
590
|
+
int size; /* number of data elements */
|
591
|
+
int len; /* length of data elements concatenated */
|
592
|
+
bool is_compressed : 2; /* set to 2 after all data is loaded */
|
593
|
+
} FrtLazyDocField;
|
594
|
+
|
595
|
+
extern char *frt_lazy_df_get_data(FrtLazyDocField *self, int i);
|
596
|
+
extern void frt_lazy_df_get_bytes(FrtLazyDocField *self, char *buf,
|
597
|
+
int start, int len);
|
598
|
+
|
599
|
+
/* * * FrtLazyDoc * * */
|
600
|
+
struct FrtLazyDoc
|
601
|
+
{
|
602
|
+
FrtHash *field_dictionary;
|
603
|
+
int size;
|
604
|
+
FrtLazyDocField **fields;
|
605
|
+
FrtInStream *fields_in;
|
606
|
+
};
|
607
|
+
|
608
|
+
extern void frt_lazy_doc_close(FrtLazyDoc *self);
|
609
|
+
extern FrtLazyDocField *frt_lazy_doc_get(FrtLazyDoc *self, FrtSymbol field);
|
610
|
+
|
611
|
+
/****************************************************************************
|
612
|
+
*
|
613
|
+
* FrtFieldsReader
|
614
|
+
*
|
615
|
+
****************************************************************************/
|
616
|
+
|
617
|
+
typedef struct FrtFieldsReader
|
618
|
+
{
|
619
|
+
int size;
|
620
|
+
FrtFieldInfos *fis;
|
621
|
+
FrtStore *store;
|
622
|
+
FrtInStream *fdx_in;
|
623
|
+
FrtInStream *fdt_in;
|
624
|
+
} FrtFieldsReader;
|
625
|
+
|
626
|
+
extern FrtFieldsReader *frt_fr_open(FrtStore *store,
|
627
|
+
const char *segment, FrtFieldInfos *fis);
|
628
|
+
extern FrtFieldsReader *frt_fr_clone(FrtFieldsReader *orig);
|
629
|
+
extern void frt_fr_close(FrtFieldsReader *fr);
|
630
|
+
extern FrtDocument *frt_fr_get_doc(FrtFieldsReader *fr, int doc_num);
|
631
|
+
extern FrtLazyDoc *frt_fr_get_lazy_doc(FrtFieldsReader *fr, int doc_num);
|
632
|
+
extern FrtHash *frt_fr_get_tv(FrtFieldsReader *fr, int doc_num);
|
633
|
+
extern FrtTermVector *frt_fr_get_field_tv(FrtFieldsReader *fr, int doc_num,
|
634
|
+
int field_num);
|
635
|
+
|
636
|
+
/****************************************************************************
|
637
|
+
*
|
638
|
+
* FrtFieldsWriter
|
639
|
+
*
|
640
|
+
****************************************************************************/
|
641
|
+
|
642
|
+
typedef struct FrtFieldsWriter
|
643
|
+
{
|
644
|
+
FrtFieldInfos *fis;
|
645
|
+
FrtOutStream *fdt_out;
|
646
|
+
FrtOutStream *fdx_out;
|
647
|
+
FrtOutStream *buffer;
|
648
|
+
FrtTVField *tv_fields;
|
649
|
+
off_t start_ptr;
|
650
|
+
} FrtFieldsWriter;
|
651
|
+
|
652
|
+
extern FrtFieldsWriter *frt_fw_open(FrtStore *store,
|
653
|
+
const char *segment, FrtFieldInfos *fis);
|
654
|
+
extern void frt_fw_close(FrtFieldsWriter *fw);
|
655
|
+
extern void frt_fw_add_doc(FrtFieldsWriter *fw, FrtDocument *doc);
|
656
|
+
extern void frt_fw_add_postings(FrtFieldsWriter *fw,
|
657
|
+
int field_num,
|
658
|
+
FrtPostingList **plists,
|
659
|
+
int posting_count,
|
660
|
+
FrtOffset *offsets,
|
661
|
+
int offset_count);
|
662
|
+
extern void frt_fw_write_tv_index(FrtFieldsWriter *fw);
|
663
|
+
|
664
|
+
/****************************************************************************
|
665
|
+
*
|
666
|
+
* FrtDeleter
|
667
|
+
*
|
668
|
+
* A utility class (used by both FrtIndexReader and FrtIndexWriter) to keep track of
|
669
|
+
* files that need to be deleted because they are no longer referenced by the
|
670
|
+
* index.
|
671
|
+
*
|
672
|
+
****************************************************************************/
|
673
|
+
|
674
|
+
struct FrtDeleter
|
675
|
+
{
|
676
|
+
FrtStore *store;
|
677
|
+
FrtSegmentInfos *sis;
|
678
|
+
FrtHashSet *pending;
|
679
|
+
};
|
680
|
+
|
681
|
+
extern FrtDeleter *frt_deleter_new(FrtSegmentInfos *sis, FrtStore *store);
|
682
|
+
extern void frt_deleter_destroy(FrtDeleter *dlr);
|
683
|
+
extern void frt_deleter_clear_pending_files(FrtDeleter *dlr);
|
684
|
+
extern void frt_deleter_delete_file(FrtDeleter *dlr, char *file_name);
|
685
|
+
extern void frt_deleter_find_deletable_files(FrtDeleter *dlr);
|
686
|
+
extern void frt_deleter_commit_pending_files(FrtDeleter *dlr);
|
687
|
+
extern void frt_deleter_delete_files(FrtDeleter *dlr, char **files, int file_cnt);
|
688
|
+
|
689
|
+
/****************************************************************************
|
690
|
+
*
|
691
|
+
* FrtIndexReader
|
692
|
+
*
|
693
|
+
****************************************************************************/
|
694
|
+
|
695
|
+
#define FRT_WRITE_LOCK_NAME "write"
|
696
|
+
#define FRT_COMMIT_LOCK_NAME "commit"
|
697
|
+
|
698
|
+
struct FrtIndexReader
|
699
|
+
{
|
700
|
+
int (*num_docs)(FrtIndexReader *ir);
|
701
|
+
int (*max_doc)(FrtIndexReader *ir);
|
702
|
+
FrtDocument *(*get_doc)(FrtIndexReader *ir, int doc_num);
|
703
|
+
FrtLazyDoc *(*get_lazy_doc)(FrtIndexReader *ir, int doc_num);
|
704
|
+
frt_uchar *(*get_norms)(FrtIndexReader *ir, int field_num);
|
705
|
+
frt_uchar *(*get_norms_into)(FrtIndexReader *ir, int field_num,
|
706
|
+
frt_uchar *buf);
|
707
|
+
FrtTermEnum *(*terms)(FrtIndexReader *ir, int field_num);
|
708
|
+
FrtTermEnum *(*terms_from)(FrtIndexReader *ir, int field_num,
|
709
|
+
const char *term);
|
710
|
+
int (*doc_freq)(FrtIndexReader *ir, int field_num,
|
711
|
+
const char *term);
|
712
|
+
FrtTermDocEnum *(*term_docs)(FrtIndexReader *ir);
|
713
|
+
FrtTermDocEnum *(*term_positions)(FrtIndexReader *ir);
|
714
|
+
FrtTermVector *(*term_vector)(FrtIndexReader *ir, int doc_num,
|
715
|
+
FrtSymbol field);
|
716
|
+
FrtHash *(*term_vectors)(FrtIndexReader *ir, int doc_num);
|
717
|
+
bool (*is_deleted)(FrtIndexReader *ir, int doc_num);
|
718
|
+
bool (*has_deletions)(FrtIndexReader *ir);
|
719
|
+
void (*acquire_write_lock)(FrtIndexReader *ir);
|
720
|
+
void (*set_norm_i)(FrtIndexReader *ir, int doc_num,
|
721
|
+
int field_num, frt_uchar val);
|
722
|
+
void (*delete_doc_i)(FrtIndexReader *ir, int doc_num);
|
723
|
+
void (*undelete_all_i)(FrtIndexReader *ir);
|
724
|
+
void (*set_deleter_i)(FrtIndexReader *ir, FrtDeleter *dlr);
|
725
|
+
bool (*is_latest_i)(FrtIndexReader *ir);
|
726
|
+
void (*commit_i)(FrtIndexReader *ir);
|
727
|
+
void (*close_i)(FrtIndexReader *ir);
|
728
|
+
int ref_cnt;
|
729
|
+
FrtDeleter *deleter;
|
730
|
+
FrtStore *store;
|
731
|
+
FrtLock *write_lock;
|
732
|
+
FrtSegmentInfos *sis;
|
733
|
+
FrtFieldInfos *fis;
|
734
|
+
FrtHash *cache;
|
735
|
+
FrtHash *field_index_cache;
|
736
|
+
frt_mutex_t field_index_mutex;
|
737
|
+
frt_uchar *fake_norms;
|
738
|
+
frt_mutex_t mutex;
|
739
|
+
bool has_changes : 1;
|
740
|
+
bool is_stale : 1;
|
741
|
+
bool is_owner : 1;
|
742
|
+
};
|
743
|
+
|
744
|
+
extern FrtIndexReader *frt_ir_create(FrtStore *store, FrtSegmentInfos *sis, int is_owner);
|
745
|
+
extern FrtIndexReader *frt_ir_open(FrtStore *store);
|
746
|
+
extern int frt_ir_get_field_num(FrtIndexReader *ir, FrtSymbol field);
|
747
|
+
extern bool frt_ir_index_exists(FrtStore *store);
|
748
|
+
extern void frt_ir_close(FrtIndexReader *ir);
|
749
|
+
extern void frt_ir_commit(FrtIndexReader *ir);
|
750
|
+
extern void frt_ir_delete_doc(FrtIndexReader *ir, int doc_num);
|
751
|
+
extern void frt_ir_undelete_all(FrtIndexReader *ir);
|
752
|
+
extern int frt_ir_doc_freq(FrtIndexReader *ir, FrtSymbol field, const char *term);
|
753
|
+
extern void frt_ir_set_norm(FrtIndexReader *ir, int doc_num, FrtSymbol field,
|
754
|
+
frt_uchar val);
|
755
|
+
extern frt_uchar *frt_ir_get_norms_i(FrtIndexReader *ir, int field_num);
|
756
|
+
extern frt_uchar *frt_ir_get_norms(FrtIndexReader *ir, FrtSymbol field);
|
757
|
+
extern frt_uchar *frt_ir_get_norms_into(FrtIndexReader *ir, FrtSymbol field, frt_uchar *buf);
|
758
|
+
extern void frt_ir_destroy(FrtIndexReader *self);
|
759
|
+
extern FrtDocument *frt_ir_get_doc_with_term(FrtIndexReader *ir, FrtSymbol field,
|
760
|
+
const char *term);
|
761
|
+
extern FrtTermEnum *frt_ir_terms(FrtIndexReader *ir, FrtSymbol field);
|
762
|
+
extern FrtTermEnum *frt_ir_terms_from(FrtIndexReader *ir, FrtSymbol field,
|
763
|
+
const char *t);
|
764
|
+
extern FrtTermDocEnum *frt_ir_term_docs_for(FrtIndexReader *ir, FrtSymbol field,
|
765
|
+
const char *term);
|
766
|
+
extern FrtTermDocEnum *frt_ir_term_positions_for(FrtIndexReader *ir,
|
767
|
+
FrtSymbol field,
|
768
|
+
const char *t);
|
769
|
+
extern void frt_ir_add_cache(FrtIndexReader *ir);
|
770
|
+
extern bool frt_ir_is_latest(FrtIndexReader *ir);
|
771
|
+
|
772
|
+
/****************************************************************************
|
773
|
+
* FrtMultiReader
|
774
|
+
****************************************************************************/
|
775
|
+
|
776
|
+
struct FrtMultiReader {
|
777
|
+
FrtIndexReader ir;
|
778
|
+
int max_doc;
|
779
|
+
int num_docs_cache;
|
780
|
+
int r_cnt;
|
781
|
+
int *starts;
|
782
|
+
FrtIndexReader **sub_readers;
|
783
|
+
FrtHash *norms_cache;
|
784
|
+
bool has_deletions : 1;
|
785
|
+
int **field_num_map;
|
786
|
+
};
|
787
|
+
|
788
|
+
extern int frt_mr_get_field_num(FrtMultiReader *mr, int ir_num, int f_num);
|
789
|
+
extern FrtIndexReader *frt_mr_open(FrtIndexReader **sub_readers, const int r_cnt);
|
790
|
+
|
791
|
+
|
792
|
+
/****************************************************************************
|
793
|
+
*
|
794
|
+
* FrtBoost
|
795
|
+
*
|
796
|
+
****************************************************************************/
|
797
|
+
|
798
|
+
typedef struct FrtBoost
|
799
|
+
{
|
800
|
+
float val;
|
801
|
+
int doc_num;
|
802
|
+
struct FrtBoost *next;
|
803
|
+
} FrtBoost;
|
804
|
+
|
805
|
+
/****************************************************************************
|
806
|
+
*
|
807
|
+
* FrtFieldInverter
|
808
|
+
*
|
809
|
+
****************************************************************************/
|
810
|
+
|
811
|
+
typedef struct FrtFieldInverter
|
812
|
+
{
|
813
|
+
FrtHash *plists;
|
814
|
+
frt_uchar *norms;
|
815
|
+
FrtFieldInfo *fi;
|
816
|
+
int length;
|
817
|
+
bool is_tokenized : 1;
|
818
|
+
bool store_term_vector : 1;
|
819
|
+
bool store_offsets : 1;
|
820
|
+
bool has_norms : 1;
|
821
|
+
} FrtFieldInverter;
|
822
|
+
|
823
|
+
/****************************************************************************
|
824
|
+
*
|
825
|
+
* FrtDocWriter
|
826
|
+
*
|
827
|
+
****************************************************************************/
|
828
|
+
|
829
|
+
#define DW_OFFSET_INIT_CAPA 512
|
830
|
+
typedef struct FrtIndexWriter FrtIndexWriter;
|
831
|
+
|
832
|
+
typedef struct FrtDocWriter
|
833
|
+
{
|
834
|
+
FrtStore *store;
|
835
|
+
FrtSegmentInfo *si;
|
836
|
+
FrtFieldInfos *fis;
|
837
|
+
FrtFieldsWriter *fw;
|
838
|
+
FrtMemoryPool *mp;
|
839
|
+
FrtAnalyzer *analyzer;
|
840
|
+
FrtHash *curr_plists;
|
841
|
+
FrtHash *fields;
|
842
|
+
FrtSimilarity *similarity;
|
843
|
+
FrtOffset *offsets;
|
844
|
+
int offsets_size;
|
845
|
+
int offsets_capa;
|
846
|
+
int doc_num;
|
847
|
+
int index_interval;
|
848
|
+
int skip_interval;
|
849
|
+
int max_field_length;
|
850
|
+
int max_buffered_docs;
|
851
|
+
} FrtDocWriter;
|
852
|
+
|
853
|
+
extern FrtDocWriter *frt_dw_open(FrtIndexWriter *is, FrtSegmentInfo *si);
|
854
|
+
extern void frt_dw_close(FrtDocWriter *dw);
|
855
|
+
extern void frt_dw_add_doc(FrtDocWriter *dw, FrtDocument *doc);
|
856
|
+
extern void frt_dw_new_segment(FrtDocWriter *dw, FrtSegmentInfo *si);
|
857
|
+
/* For testing. need to remove somehow. FIXME */
|
858
|
+
extern FrtHash *frt_dw_invert_field(FrtDocWriter *dw,
|
859
|
+
FrtFieldInverter *fld_inv,
|
860
|
+
FrtDocField *df);
|
861
|
+
extern FrtFieldInverter *frt_dw_get_fld_inv(FrtDocWriter *dw, FrtFieldInfo *fi);
|
862
|
+
extern void frt_dw_reset_postings(FrtHash *postings);
|
863
|
+
|
864
|
+
/****************************************************************************
|
865
|
+
*
|
866
|
+
* FrtIndexWriter
|
867
|
+
*
|
868
|
+
****************************************************************************/
|
869
|
+
|
870
|
+
typedef struct FrtDelTerm
|
871
|
+
{
|
872
|
+
int field_num;
|
873
|
+
char *term;
|
874
|
+
} FrtDelTerm;
|
875
|
+
|
876
|
+
struct FrtIndexWriter
|
877
|
+
{
|
878
|
+
FrtConfig config;
|
879
|
+
frt_mutex_t mutex;
|
880
|
+
FrtStore *store;
|
881
|
+
FrtAnalyzer *analyzer;
|
882
|
+
FrtSegmentInfos *sis;
|
883
|
+
FrtFieldInfos *fis;
|
884
|
+
FrtDocWriter *dw;
|
885
|
+
FrtSimilarity *similarity;
|
886
|
+
FrtLock *write_lock;
|
887
|
+
FrtDeleter *deleter;
|
888
|
+
};
|
889
|
+
|
890
|
+
extern void frt_index_create(FrtStore *store, FrtFieldInfos *fis);
|
891
|
+
extern bool frt_index_is_locked(FrtStore *store);
|
892
|
+
extern FrtIndexWriter *frt_iw_open(FrtStore *store, FrtAnalyzer *analyzer,
|
893
|
+
const FrtConfig *config);
|
894
|
+
extern void frt_iw_delete_term(FrtIndexWriter *iw, FrtSymbol field,
|
895
|
+
const char *term);
|
896
|
+
extern void frt_iw_delete_terms(FrtIndexWriter *iw, FrtSymbol field,
|
897
|
+
char **terms, const int term_cnt);
|
898
|
+
extern void frt_iw_close(FrtIndexWriter *iw);
|
899
|
+
extern void frt_iw_add_doc(FrtIndexWriter *iw, FrtDocument *doc);
|
900
|
+
extern int frt_iw_doc_count(FrtIndexWriter *iw);
|
901
|
+
extern void frt_iw_commit(FrtIndexWriter *iw);
|
902
|
+
extern void frt_iw_optimize(FrtIndexWriter *iw);
|
903
|
+
extern void frt_iw_add_readers(FrtIndexWriter *iw, FrtIndexReader **readers,
|
904
|
+
const int r_cnt);
|
905
|
+
|
906
|
+
/****************************************************************************
|
907
|
+
*
|
908
|
+
* FrtCompoundWriter
|
909
|
+
*
|
910
|
+
****************************************************************************/
|
911
|
+
|
912
|
+
#define FRT_CW_INIT_CAPA 16
|
913
|
+
typedef struct FrtCWFileEntry
|
914
|
+
{
|
915
|
+
char *name;
|
916
|
+
off_t dir_offset;
|
917
|
+
off_t data_offset;
|
918
|
+
} FrtCWFileEntry;
|
919
|
+
|
920
|
+
typedef struct FrtCompoundWriter {
|
921
|
+
FrtStore *store;
|
922
|
+
const char *name;
|
923
|
+
FrtHashSet *ids;
|
924
|
+
FrtCWFileEntry *file_entries;
|
925
|
+
} FrtCompoundWriter;
|
926
|
+
|
927
|
+
extern FrtCompoundWriter *frt_open_cw(FrtStore *store, char *name);
|
928
|
+
extern void frt_cw_add_file(FrtCompoundWriter *cw, char *id);
|
929
|
+
extern void frt_cw_close(FrtCompoundWriter *cw);
|
930
|
+
|
931
|
+
#ifdef __cplusplus
|
932
|
+
} // extern "C"
|
933
|
+
#endif
|
934
|
+
|
935
|
+
#endif
|