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 +1,247 @@
|
|
1
|
+
#ifndef FRT_ANALYSIS_H
|
2
|
+
#define FRT_ANALYSIS_H
|
3
|
+
|
4
|
+
#include <wchar.h>
|
5
|
+
#include "frt_global.h"
|
6
|
+
#include "frt_hash.h"
|
7
|
+
#include "frt_multimapper.h"
|
8
|
+
|
9
|
+
/****************************************************************************
|
10
|
+
*
|
11
|
+
* FrtToken
|
12
|
+
*
|
13
|
+
****************************************************************************/
|
14
|
+
|
15
|
+
typedef struct FrtToken
|
16
|
+
{
|
17
|
+
char text[FRT_MAX_WORD_SIZE];
|
18
|
+
int len;
|
19
|
+
off_t start;
|
20
|
+
off_t end;
|
21
|
+
int pos_inc;
|
22
|
+
} FrtToken;
|
23
|
+
|
24
|
+
extern FrtToken *frt_tk_new();
|
25
|
+
extern void frt_tk_destroy(void *p);
|
26
|
+
extern FrtToken *frt_tk_set(FrtToken *tk, char *text, int tlen, off_t start, off_t end, int pos_inc);
|
27
|
+
extern FrtToken *frt_tk_set_no_len(FrtToken *tk, char *text, off_t start, off_t end, int pos_inc);
|
28
|
+
extern int frt_tk_eq(FrtToken *tk1, FrtToken *tk2);
|
29
|
+
extern int frt_tk_cmp(FrtToken *tk1, FrtToken *tk2);
|
30
|
+
|
31
|
+
/****************************************************************************
|
32
|
+
*
|
33
|
+
* FrtTokenStream
|
34
|
+
*
|
35
|
+
****************************************************************************/
|
36
|
+
|
37
|
+
|
38
|
+
typedef struct FrtTokenStream FrtTokenStream;
|
39
|
+
struct FrtTokenStream
|
40
|
+
{
|
41
|
+
char *t; /* ptr used to scan text */
|
42
|
+
char *text;
|
43
|
+
FrtToken *(*next)(FrtTokenStream *ts);
|
44
|
+
FrtTokenStream *(*reset)(FrtTokenStream *ts, char *text);
|
45
|
+
FrtTokenStream *(*clone_i)(FrtTokenStream *ts);
|
46
|
+
void (*destroy_i)(FrtTokenStream *ts);
|
47
|
+
int ref_cnt;
|
48
|
+
};
|
49
|
+
|
50
|
+
#define frt_ts_new(type) frt_ts_new_i(sizeof(type))
|
51
|
+
extern FrtTokenStream *frt_ts_new_i(size_t size);
|
52
|
+
extern FrtTokenStream *frt_ts_clone_size(FrtTokenStream *orig_ts, size_t size);
|
53
|
+
|
54
|
+
typedef struct FrtCachedTokenStream
|
55
|
+
{
|
56
|
+
FrtTokenStream super;
|
57
|
+
FrtToken token;
|
58
|
+
} FrtCachedTokenStream;
|
59
|
+
|
60
|
+
typedef struct FrtMultiByteTokenStream
|
61
|
+
{
|
62
|
+
FrtCachedTokenStream super;
|
63
|
+
mbstate_t state;
|
64
|
+
} FrtMultiByteTokenStream;
|
65
|
+
|
66
|
+
typedef enum
|
67
|
+
{
|
68
|
+
FRT_STT_ASCII,
|
69
|
+
FRT_STT_MB,
|
70
|
+
FRT_STT_UTF8
|
71
|
+
} FrtStandardTokenizerType;
|
72
|
+
|
73
|
+
typedef struct FrtStandardTokenizer
|
74
|
+
{
|
75
|
+
FrtCachedTokenStream super;
|
76
|
+
FrtStandardTokenizerType type;
|
77
|
+
} FrtStandardTokenizer;
|
78
|
+
|
79
|
+
typedef struct FrtLegacyStandardTokenizer
|
80
|
+
{
|
81
|
+
FrtCachedTokenStream super;
|
82
|
+
bool (*advance_to_start)(FrtTokenStream *ts);
|
83
|
+
bool (*is_tok_char)(char *c);
|
84
|
+
int (*get_alpha)(FrtTokenStream *ts, char *token);
|
85
|
+
int (*get_apostrophe)(char *input);
|
86
|
+
} FrtLegacyStandardTokenizer;
|
87
|
+
|
88
|
+
typedef struct FrtTokenFilter
|
89
|
+
{
|
90
|
+
FrtTokenStream super;
|
91
|
+
FrtTokenStream *sub_ts;
|
92
|
+
} FrtTokenFilter;
|
93
|
+
|
94
|
+
extern FrtTokenStream *frt_filter_clone_size(FrtTokenStream *ts, size_t size);
|
95
|
+
#define tf_new(type, sub) frt_tf_new_i(sizeof(type), sub)
|
96
|
+
extern FrtTokenStream *frt_tf_new_i(size_t size, FrtTokenStream *sub_ts);
|
97
|
+
|
98
|
+
typedef struct FrtStopFilter
|
99
|
+
{
|
100
|
+
FrtTokenFilter super;
|
101
|
+
FrtHash *words;
|
102
|
+
} FrtStopFilter;
|
103
|
+
|
104
|
+
typedef struct FrtMappingFilter
|
105
|
+
{
|
106
|
+
FrtTokenFilter super;
|
107
|
+
FrtMultiMapper *mapper;
|
108
|
+
} FrtMappingFilter;
|
109
|
+
|
110
|
+
typedef struct FrtHyphenFilter
|
111
|
+
{
|
112
|
+
FrtTokenFilter super;
|
113
|
+
char text[FRT_MAX_WORD_SIZE];
|
114
|
+
int start;
|
115
|
+
int pos;
|
116
|
+
int len;
|
117
|
+
FrtToken *tk;
|
118
|
+
} FrtHyphenFilter;
|
119
|
+
|
120
|
+
typedef struct FrtStemFilter
|
121
|
+
{
|
122
|
+
FrtTokenFilter super;
|
123
|
+
struct sb_stemmer *stemmer;
|
124
|
+
char *algorithm;
|
125
|
+
char *charenc;
|
126
|
+
} FrtStemFilter;
|
127
|
+
|
128
|
+
#define frt_ts_next(mts) mts->next(mts)
|
129
|
+
#define frt_ts_clone(mts) mts->clone_i(mts)
|
130
|
+
|
131
|
+
extern void frt_ts_deref(FrtTokenStream *ts);
|
132
|
+
|
133
|
+
extern FrtTokenStream *frt_non_tokenizer_new();
|
134
|
+
|
135
|
+
extern FrtTokenStream *frt_whitespace_tokenizer_new();
|
136
|
+
extern FrtTokenStream *frt_mb_whitespace_tokenizer_new(bool lowercase);
|
137
|
+
|
138
|
+
extern FrtTokenStream *frt_letter_tokenizer_new();
|
139
|
+
extern FrtTokenStream *frt_mb_letter_tokenizer_new(bool lowercase);
|
140
|
+
|
141
|
+
extern FrtTokenStream *frt_standard_tokenizer_new();
|
142
|
+
extern FrtTokenStream *frt_mb_standard_tokenizer_new();
|
143
|
+
extern FrtTokenStream *frt_utf8_standard_tokenizer_new();
|
144
|
+
|
145
|
+
extern FrtTokenStream *frt_legacy_standard_tokenizer_new();
|
146
|
+
extern FrtTokenStream *frt_mb_legacy_standard_tokenizer_new();
|
147
|
+
|
148
|
+
extern FrtTokenStream *frt_hyphen_filter_new(FrtTokenStream *ts);
|
149
|
+
extern FrtTokenStream *frt_lowercase_filter_new(FrtTokenStream *ts);
|
150
|
+
extern FrtTokenStream *frt_mb_lowercase_filter_new(FrtTokenStream *ts);
|
151
|
+
|
152
|
+
extern const char *FRT_ENGLISH_STOP_WORDS[];
|
153
|
+
extern const char *FRT_FULL_ENGLISH_STOP_WORDS[];
|
154
|
+
extern const char *FRT_EXTENDED_ENGLISH_STOP_WORDS[];
|
155
|
+
extern const char *FRT_FULL_FRENCH_STOP_WORDS[];
|
156
|
+
extern const char *FRT_FULL_SPANISH_STOP_WORDS[];
|
157
|
+
extern const char *FRT_FULL_PORTUGUESE_STOP_WORDS[];
|
158
|
+
extern const char *FRT_FULL_ITALIAN_STOP_WORDS[];
|
159
|
+
extern const char *FRT_FULL_GERMAN_STOP_WORDS[];
|
160
|
+
extern const char *FRT_FULL_DUTCH_STOP_WORDS[];
|
161
|
+
extern const char *FRT_FULL_SWEDISH_STOP_WORDS[];
|
162
|
+
extern const char *FRT_FULL_NORWEGIAN_STOP_WORDS[];
|
163
|
+
extern const char *FRT_FULL_DANISH_STOP_WORDS[];
|
164
|
+
extern const char *FRT_FULL_RUSSIAN_STOP_WORDS[];
|
165
|
+
extern const char *FRT_FULL_FINNISH_STOP_WORDS[];
|
166
|
+
extern const char *FRT_FULL_HUNGARIAN_STOP_WORDS[];
|
167
|
+
|
168
|
+
extern FrtTokenStream *frt_stop_filter_new_with_words_len(FrtTokenStream *ts,
|
169
|
+
const char **words, int len);
|
170
|
+
extern FrtTokenStream *frt_stop_filter_new_with_words(FrtTokenStream *ts,
|
171
|
+
const char **words);
|
172
|
+
extern FrtTokenStream *frt_stop_filter_new(FrtTokenStream *ts);
|
173
|
+
extern FrtTokenStream *frt_stem_filter_new(FrtTokenStream *ts, const char *algorithm,
|
174
|
+
const char *charenc);
|
175
|
+
|
176
|
+
extern FrtTokenStream *frt_mapping_filter_new(FrtTokenStream *ts);
|
177
|
+
extern FrtTokenStream *frt_mapping_filter_add(FrtTokenStream *ts, const char *pattern,
|
178
|
+
const char *replacement);
|
179
|
+
|
180
|
+
/****************************************************************************
|
181
|
+
*
|
182
|
+
* FrtAnalyzer
|
183
|
+
*
|
184
|
+
****************************************************************************/
|
185
|
+
|
186
|
+
typedef struct FrtAnalyzer
|
187
|
+
{
|
188
|
+
FrtTokenStream *current_ts;
|
189
|
+
FrtTokenStream *(*get_ts)(struct FrtAnalyzer *a, FrtSymbol field, char *text);
|
190
|
+
void (*destroy_i)(struct FrtAnalyzer *a);
|
191
|
+
int ref_cnt;
|
192
|
+
} FrtAnalyzer;
|
193
|
+
|
194
|
+
extern void frt_a_deref(FrtAnalyzer *a);
|
195
|
+
|
196
|
+
#define frt_a_get_ts(ma, field, text) ma->get_ts(ma, field, text)
|
197
|
+
|
198
|
+
extern FrtAnalyzer *frt_analyzer_new(FrtTokenStream *ts,
|
199
|
+
void (*destroy)(FrtAnalyzer *a),
|
200
|
+
FrtTokenStream *(*get_ts)(FrtAnalyzer *a,
|
201
|
+
FrtSymbol field,
|
202
|
+
char *text));
|
203
|
+
extern FrtAnalyzer *frt_non_analyzer_new();
|
204
|
+
|
205
|
+
extern void frt_a_standard_destroy(FrtAnalyzer *a);
|
206
|
+
|
207
|
+
extern FrtAnalyzer *frt_whitespace_analyzer_new(bool lowercase);
|
208
|
+
extern FrtAnalyzer *frt_mb_whitespace_analyzer_new(bool lowercase);
|
209
|
+
|
210
|
+
extern FrtAnalyzer *frt_letter_analyzer_new(bool lowercase);
|
211
|
+
extern FrtAnalyzer *frt_mb_letter_analyzer_new(bool lowercase);
|
212
|
+
|
213
|
+
extern FrtAnalyzer *frt_standard_analyzer_new(bool lowercase);
|
214
|
+
extern FrtAnalyzer *frt_mb_standard_analyzer_new(bool lowercase);
|
215
|
+
extern FrtAnalyzer *frt_utf8_standard_analyzer_new(bool lowercase);
|
216
|
+
|
217
|
+
extern FrtAnalyzer *frt_standard_analyzer_new_with_words(
|
218
|
+
const char **words, bool lowercase);
|
219
|
+
extern FrtAnalyzer *frt_standard_analyzer_new_with_words_len(
|
220
|
+
const char **words, int len, bool lowercase);
|
221
|
+
extern FrtAnalyzer *frt_mb_standard_analyzer_new_with_words(
|
222
|
+
const char **words, bool lowercase);
|
223
|
+
extern FrtAnalyzer *frt_utf8_standard_analyzer_new_with_words(
|
224
|
+
const char **words, bool lowercase);
|
225
|
+
|
226
|
+
extern FrtAnalyzer *frt_legacy_standard_analyzer_new(bool lowercase);
|
227
|
+
extern FrtAnalyzer *frt_mb_legacy_standard_analyzer_new(bool lowercase);
|
228
|
+
|
229
|
+
extern FrtAnalyzer *frt_legacy_standard_analyzer_new_with_words(
|
230
|
+
const char **words, bool lowercase);
|
231
|
+
extern FrtAnalyzer *frt_mb_legacy_standard_analyzer_new_with_words(
|
232
|
+
const char **words, bool lowercase);
|
233
|
+
|
234
|
+
#define PFA(analyzer) ((FrtPerFieldAnalyzer *)(analyzer))
|
235
|
+
typedef struct FrtPerFieldAnalyzer
|
236
|
+
{
|
237
|
+
FrtAnalyzer super;
|
238
|
+
FrtHash *dict;
|
239
|
+
FrtAnalyzer *default_a;
|
240
|
+
} FrtPerFieldAnalyzer;
|
241
|
+
|
242
|
+
extern FrtAnalyzer *frt_per_field_analyzer_new(FrtAnalyzer *a);
|
243
|
+
extern void frt_pfa_add_field(FrtAnalyzer *self,
|
244
|
+
FrtSymbol field,
|
245
|
+
FrtAnalyzer *analyzer);
|
246
|
+
|
247
|
+
#endif
|
@@ -0,0 +1,124 @@
|
|
1
|
+
#include "frt_array.h"
|
2
|
+
#include <string.h>
|
3
|
+
|
4
|
+
#define META_CNT FRT_ARY_META_CNT
|
5
|
+
#define DATA_SZ sizeof(int) * META_CNT
|
6
|
+
|
7
|
+
void **frt_ary_new_i(int type_size, int init_capa)
|
8
|
+
{
|
9
|
+
void **ary;
|
10
|
+
if (init_capa <= 0) {
|
11
|
+
init_capa = FRT_ARY_INIT_CAPA;
|
12
|
+
}
|
13
|
+
ary = (void **)&(((int *)frt_ecalloc(DATA_SZ +
|
14
|
+
init_capa * type_size))[META_CNT]);
|
15
|
+
frt_ary_type_size(ary) = type_size;
|
16
|
+
frt_ary_capa(ary) = init_capa;
|
17
|
+
return ary;
|
18
|
+
}
|
19
|
+
|
20
|
+
void frt_ary_resize_i(void ***ary, int size)
|
21
|
+
{
|
22
|
+
size++;
|
23
|
+
if (size > frt_ary_sz(*ary)) {
|
24
|
+
int capa = frt_ary_capa(*ary);
|
25
|
+
if (size >= capa) {
|
26
|
+
int *frt_ary_start = &((int *)*ary)[-META_CNT];
|
27
|
+
while (size >= capa) {
|
28
|
+
capa <<= 1;
|
29
|
+
}
|
30
|
+
|
31
|
+
frt_ary_start = (int *)frt_erealloc(frt_ary_start,
|
32
|
+
DATA_SZ + capa * frt_ary_type_size(*ary));
|
33
|
+
*ary = (void **)&(frt_ary_start[META_CNT]);
|
34
|
+
memset(((char *)*ary) + frt_ary_type_size(*ary) * frt_ary_sz(*ary), 0,
|
35
|
+
(capa - frt_ary_sz(*ary)) * frt_ary_type_size(*ary));
|
36
|
+
frt_ary_capa(*ary) = capa;
|
37
|
+
}
|
38
|
+
frt_ary_sz(*ary) = size;
|
39
|
+
}
|
40
|
+
}
|
41
|
+
|
42
|
+
void frt_ary_set_i(void ***ary, int index, void *value)
|
43
|
+
{
|
44
|
+
if (index < 0) {
|
45
|
+
index += frt_ary_sz(*ary);
|
46
|
+
if (index < 0) {
|
47
|
+
FRT_RAISE(FRT_INDEX_ERROR, "index %d out array", index);
|
48
|
+
}
|
49
|
+
}
|
50
|
+
frt_ary_resize_i(ary, index);
|
51
|
+
(*ary)[index] = value;
|
52
|
+
}
|
53
|
+
|
54
|
+
void *frt_ary_get_i(void **ary, int index)
|
55
|
+
{
|
56
|
+
if (index < 0) {
|
57
|
+
index += frt_ary_sz(ary);
|
58
|
+
}
|
59
|
+
if (index >= 0 && index < frt_ary_sz(ary)) {
|
60
|
+
return ary[index];
|
61
|
+
}
|
62
|
+
else {
|
63
|
+
return NULL;
|
64
|
+
}
|
65
|
+
}
|
66
|
+
|
67
|
+
void frt_ary_push_i(void ***ary, void *value)
|
68
|
+
{
|
69
|
+
int size = frt_ary_sz(*ary);
|
70
|
+
frt_ary_resize_i(ary, size);
|
71
|
+
(*ary)[size] = value;
|
72
|
+
}
|
73
|
+
|
74
|
+
void *frt_ary_pop_i(void **ary)
|
75
|
+
{
|
76
|
+
void *val = ary[--frt_ary_sz(ary)];
|
77
|
+
ary[frt_ary_sz(ary)] = NULL;
|
78
|
+
return val;
|
79
|
+
}
|
80
|
+
|
81
|
+
void frt_ary_unshift_i(void ***ary, void *value)
|
82
|
+
{
|
83
|
+
int size = frt_ary_sz(*ary);
|
84
|
+
frt_ary_resize_i(ary, size);
|
85
|
+
memmove(*ary + 1, *ary, size * sizeof(void *));
|
86
|
+
(*ary)[0] = value;
|
87
|
+
}
|
88
|
+
|
89
|
+
void *frt_ary_shift_i(void **ary)
|
90
|
+
{
|
91
|
+
void *val = ary[0];
|
92
|
+
int size = --frt_ary_sz(ary);
|
93
|
+
memmove(ary, ary + 1, size * sizeof(void *));
|
94
|
+
ary[size] = NULL;
|
95
|
+
return val;
|
96
|
+
}
|
97
|
+
|
98
|
+
void *frt_ary_remove_i(void **ary, int index)
|
99
|
+
{
|
100
|
+
if (index >= 0 && index < frt_ary_sz(ary)) {
|
101
|
+
void *val = ary[index];
|
102
|
+
memmove(ary + index, ary + index + 1,
|
103
|
+
(frt_ary_sz(ary) - index + 1) * sizeof(void *));
|
104
|
+
frt_ary_sz(ary)--;
|
105
|
+
return val;
|
106
|
+
}
|
107
|
+
else {
|
108
|
+
return NULL;
|
109
|
+
}
|
110
|
+
}
|
111
|
+
|
112
|
+
void frt_ary_delete_i(void **ary, int index, void (*free_elem)(void *p))
|
113
|
+
{
|
114
|
+
free_elem(frt_ary_remove(ary, index));
|
115
|
+
}
|
116
|
+
|
117
|
+
void frt_ary_destroy_i(void **ary, void (*free_elem)(void *p))
|
118
|
+
{
|
119
|
+
int i;
|
120
|
+
for (i = frt_ary_sz(ary) - 1; i >= 0; i--) {
|
121
|
+
free_elem(ary[i]);
|
122
|
+
}
|
123
|
+
frt_ary_free(ary);
|
124
|
+
}
|
@@ -0,0 +1,54 @@
|
|
1
|
+
#ifndef FRT_ARRAY_H
|
2
|
+
#define FRT_ARRAY_H
|
3
|
+
|
4
|
+
#include "frt_global.h"
|
5
|
+
|
6
|
+
#if defined POSH_OS_SOLARIS || defined POSH_OS_SUNOS
|
7
|
+
# define FRT_ARY_META_CNT 4
|
8
|
+
#else
|
9
|
+
# define FRT_ARY_META_CNT 3
|
10
|
+
#endif
|
11
|
+
|
12
|
+
#define FRT_ARY_INIT_CAPA 8
|
13
|
+
#define frt_ary_size(ary) frt_ary_sz(ary)
|
14
|
+
#define frt_ary_sz(ary) (((int *)ary)[-1])
|
15
|
+
#define frt_ary_capa(ary) (((int *)ary)[-2])
|
16
|
+
#define frt_ary_type_size(ary) (((int *)ary)[-3])
|
17
|
+
#define frt_ary_start(ary) ((void **)&(((int *)ary)[-FRT_ARY_META_CNT]))
|
18
|
+
#define frt_ary_free(ary) free(frt_ary_start(ary))
|
19
|
+
|
20
|
+
#define frt_ary_new_type_capa(type, init_capa)\
|
21
|
+
(type *)frt_ary_new_i(sizeof(type), init_capa)
|
22
|
+
#define frt_ary_new_type(type) (type *)frt_ary_new_i(sizeof(type), 0)
|
23
|
+
#define frt_ary_new_capa(init_capa) frt_ary_new_i(sizeof(void *), init_capa)
|
24
|
+
#define frt_ary_new() frt_ary_new_i(sizeof(void *), 0)
|
25
|
+
#define frt_ary_resize(ary, size) frt_ary_resize_i(((void ***)(void *)&ary), size)
|
26
|
+
#define frt_ary_set(ary, i, val) frt_ary_set_i(((void ***)(void *)&ary), i, val)
|
27
|
+
#define frt_ary_get(ary, i) frt_ary_get_i(((void **)ary), i)
|
28
|
+
#define frt_ary_push(ary, val) frt_ary_push_i(((void ***)(void *)&ary), val)
|
29
|
+
#define frt_ary_pop(ary) frt_ary_pop_i(((void **)ary))
|
30
|
+
#define frt_ary_unshift(ary, val) frt_ary_unshift_i(((void ***)(void *)&ary), val)
|
31
|
+
#define frt_ary_shift(ary) frt_ary_shift_i(((void **)ary))
|
32
|
+
#define frt_ary_remove(ary, i) frt_ary_remove_i(((void **)ary), i)
|
33
|
+
#define frt_ary_delete(ary, i, f) frt_ary_delete_i(((void **)ary), i, (frt_free_ft)f)
|
34
|
+
#define frt_ary_destroy(ary, f) frt_ary_destroy_i(((void **)ary), (frt_free_ft)f)
|
35
|
+
#define frt_ary_rsz(ary, size) frt_ary_resize(ary, size)
|
36
|
+
#define frt_ary_grow(ary) frt_ary_resize(ary, frt_ary_sz(ary))
|
37
|
+
#define frt_ary_last(ary) ary[frt_ary_sz(ary) - 1]
|
38
|
+
#define frt_ary_sort(ary, cmp) qsort(ary, frt_ary_size(ary), frt_ary_type_size(ary), cmp)
|
39
|
+
#define frt_ary_each_rev(ary, i) for (i = frt_ary_size(ary) - 1; i >= 0; i--)
|
40
|
+
#define frt_ary_each(ary, i) for (i = 0; i < frt_ary_size(ary); i++)
|
41
|
+
|
42
|
+
extern void frt_ary_resize_i(void ***ary, int size);
|
43
|
+
extern void **frt_ary_new_i(int type_size, int init_capa);
|
44
|
+
extern void frt_ary_set_i(void ***ary, int index, void *value);
|
45
|
+
extern void *frt_ary_get_i(void **ary, int index);
|
46
|
+
extern void frt_ary_push_i(void ***ary, void *value);
|
47
|
+
extern void *frt_ary_pop_i(void **ary);
|
48
|
+
extern void frt_ary_unshift_i(void ***ary, void *value);
|
49
|
+
extern void *frt_ary_shift_i(void **ary);
|
50
|
+
extern void *frt_ary_remove_i(void **ary, int index);
|
51
|
+
extern void frt_ary_delete_i(void **ary, int index, frt_free_ft p);
|
52
|
+
extern void frt_ary_destroy_i(void **ary, frt_free_ft p);
|
53
|
+
|
54
|
+
#endif
|
@@ -0,0 +1,95 @@
|
|
1
|
+
#include "frt_bitvector.h"
|
2
|
+
#include <string.h>
|
3
|
+
|
4
|
+
FrtBitVector *frt_bv_new_capa(int capa)
|
5
|
+
{
|
6
|
+
FrtBitVector *bv = FRT_ALLOC_AND_ZERO(FrtBitVector);
|
7
|
+
|
8
|
+
/* The capacity passed by the user is number of bits allowed, however we
|
9
|
+
* store capacity as the number of words (U32) allocated. */
|
10
|
+
bv->capa = FRT_MAX(FRT_TO_WORD(capa), 4);
|
11
|
+
bv->bits = FRT_ALLOC_AND_ZERO_N(frt_u32, bv->capa);
|
12
|
+
bv->curr_bit = -1;
|
13
|
+
bv->ref_cnt = 1;
|
14
|
+
return bv;
|
15
|
+
}
|
16
|
+
|
17
|
+
FrtBitVector *frt_bv_new()
|
18
|
+
{
|
19
|
+
return frt_bv_new_capa(FRT_BV_INIT_CAPA);
|
20
|
+
}
|
21
|
+
|
22
|
+
void frt_bv_destroy(FrtBitVector *bv)
|
23
|
+
{
|
24
|
+
if (--(bv->ref_cnt) == 0) {
|
25
|
+
free(bv->bits);
|
26
|
+
free(bv);
|
27
|
+
}
|
28
|
+
}
|
29
|
+
|
30
|
+
void frt_bv_clear(FrtBitVector *bv)
|
31
|
+
{
|
32
|
+
memset(bv->bits, 0, bv->capa * sizeof(frt_u32));
|
33
|
+
bv->extends_as_ones = 0;
|
34
|
+
bv->count = 0;
|
35
|
+
bv->size = 0;
|
36
|
+
}
|
37
|
+
|
38
|
+
void frt_bv_scan_reset(FrtBitVector *bv)
|
39
|
+
{
|
40
|
+
bv->curr_bit = -1;
|
41
|
+
}
|
42
|
+
|
43
|
+
int frt_bv_eq(FrtBitVector *bv1, FrtBitVector *bv2)
|
44
|
+
{
|
45
|
+
frt_u32 *bits, *bits2;
|
46
|
+
int min_size, word_size, ext_word_size = 0, i;
|
47
|
+
if (bv1 == bv2) {
|
48
|
+
return true;
|
49
|
+
}
|
50
|
+
|
51
|
+
if (bv1->extends_as_ones != bv2->extends_as_ones) {
|
52
|
+
return false;
|
53
|
+
}
|
54
|
+
|
55
|
+
bits = bv1->bits;
|
56
|
+
bits2 = bv2->bits;
|
57
|
+
min_size = FRT_MIN(bv1->size, bv2->size);
|
58
|
+
word_size = FRT_TO_WORD(min_size);
|
59
|
+
|
60
|
+
for (i = 0; i < word_size; i++) {
|
61
|
+
if (bits[i] != bits2[i]) {
|
62
|
+
return false;
|
63
|
+
}
|
64
|
+
}
|
65
|
+
if (bv1->size > min_size) {
|
66
|
+
bits = bv1->bits;
|
67
|
+
ext_word_size = FRT_TO_WORD(bv1->size);
|
68
|
+
}
|
69
|
+
else if (bv2->size > min_size) {
|
70
|
+
bits = bv2->bits;
|
71
|
+
ext_word_size = FRT_TO_WORD(bv2->size);
|
72
|
+
}
|
73
|
+
if (ext_word_size) {
|
74
|
+
const frt_u32 expected = (bv1->extends_as_ones ? 0xFFFFFFFF : 0);
|
75
|
+
for (i = word_size; i < ext_word_size; i++) {
|
76
|
+
if (bits[i] != expected) {
|
77
|
+
return false;
|
78
|
+
}
|
79
|
+
}
|
80
|
+
}
|
81
|
+
return true;
|
82
|
+
}
|
83
|
+
|
84
|
+
unsigned long long frt_bv_hash(FrtBitVector *bv)
|
85
|
+
{
|
86
|
+
unsigned long long hash = 0;
|
87
|
+
const frt_u32 empty_word = bv->extends_as_ones ? 0xFFFFFFFF : 0;
|
88
|
+
int i;
|
89
|
+
for (i = FRT_TO_WORD(bv->size) - 1; i >= 0; i--) {
|
90
|
+
const frt_u32 word = bv->bits[i];
|
91
|
+
if (word != empty_word)
|
92
|
+
hash = (hash << 1) ^ word;
|
93
|
+
}
|
94
|
+
return (hash << 1) | bv->extends_as_ones;
|
95
|
+
}
|