ferret 0.11.6 → 0.11.8.4
Sign up to get free protection for your applications and to get access to all the features.
- data/README +10 -22
- data/RELEASE_CHANGES +137 -0
- data/RELEASE_NOTES +60 -0
- data/Rakefile +379 -274
- data/TODO +100 -8
- data/bin/ferret-browser +0 -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/{api.c → STEMMER_api.c} +7 -10
- data/ext/{libstemmer.c → STEMMER_libstemmer.c} +3 -2
- data/ext/{stem_ISO_8859_1_danish.c → STEMMER_stem_ISO_8859_1_danish.c} +123 -124
- data/ext/{stem_ISO_8859_1_dutch.c → STEMMER_stem_ISO_8859_1_dutch.c} +177 -188
- data/ext/STEMMER_stem_ISO_8859_1_english.c +1117 -0
- data/ext/{stem_ISO_8859_1_finnish.c → STEMMER_stem_ISO_8859_1_finnish.c} +276 -306
- data/ext/STEMMER_stem_ISO_8859_1_french.c +1246 -0
- data/ext/{stem_ISO_8859_1_german.c → STEMMER_stem_ISO_8859_1_german.c} +161 -170
- 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/{stem_ISO_8859_1_porter.c → STEMMER_stem_ISO_8859_1_porter.c} +263 -290
- data/ext/{stem_ISO_8859_1_portuguese.c → STEMMER_stem_ISO_8859_1_portuguese.c} +362 -380
- 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/{stem_KOI8_R_russian.c → STEMMER_stem_KOI8_R_russian.c} +244 -245
- data/ext/STEMMER_stem_UTF_8_danish.c +339 -0
- data/ext/{stem_UTF_8_dutch.c → STEMMER_stem_UTF_8_dutch.c} +192 -211
- data/ext/STEMMER_stem_UTF_8_english.c +1125 -0
- data/ext/{stem_UTF_8_finnish.c → STEMMER_stem_UTF_8_finnish.c} +284 -324
- data/ext/STEMMER_stem_UTF_8_french.c +1256 -0
- data/ext/{stem_UTF_8_german.c → STEMMER_stem_UTF_8_german.c} +170 -187
- 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/{stem_UTF_8_porter.c → STEMMER_stem_UTF_8_porter.c} +271 -310
- 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/{utilities.c → STEMMER_utilities.c} +100 -68
- data/ext/analysis.c +276 -121
- data/ext/analysis.h +190 -143
- data/ext/api.h +3 -4
- data/ext/array.c +5 -3
- data/ext/array.h +52 -43
- data/ext/bitvector.c +38 -482
- data/ext/bitvector.h +446 -124
- data/ext/bzlib.h +282 -0
- data/ext/bzlib_private.h +503 -0
- data/ext/compound_io.c +23 -22
- data/ext/config.h +21 -11
- data/ext/document.c +43 -40
- data/ext/document.h +31 -21
- data/ext/except.c +20 -38
- data/ext/except.h +89 -76
- data/ext/extconf.rb +3 -2
- data/ext/ferret.c +49 -35
- data/ext/ferret.h +14 -11
- data/ext/field_index.c +262 -0
- data/ext/field_index.h +52 -0
- data/ext/filter.c +11 -10
- data/ext/fs_store.c +65 -47
- data/ext/global.c +245 -165
- data/ext/global.h +252 -54
- data/ext/hash.c +200 -243
- data/ext/hash.h +205 -163
- data/ext/hashset.c +118 -96
- data/ext/hashset.h +110 -82
- data/ext/header.h +19 -19
- data/ext/helper.c +11 -10
- data/ext/helper.h +14 -6
- data/ext/index.c +745 -366
- data/ext/index.h +503 -529
- data/ext/internal.h +1020 -0
- data/ext/lang.c +10 -0
- data/ext/lang.h +35 -15
- data/ext/mempool.c +5 -4
- data/ext/mempool.h +30 -22
- data/ext/modules.h +35 -7
- data/ext/multimapper.c +43 -2
- data/ext/multimapper.h +32 -23
- data/ext/posh.c +0 -0
- data/ext/posh.h +4 -38
- data/ext/priorityqueue.c +10 -12
- data/ext/priorityqueue.h +33 -21
- data/ext/q_boolean.c +22 -9
- data/ext/q_const_score.c +3 -2
- data/ext/q_filtered_query.c +15 -12
- data/ext/q_fuzzy.c +147 -135
- data/ext/q_match_all.c +3 -2
- data/ext/q_multi_term.c +28 -32
- data/ext/q_parser.c +451 -173
- data/ext/q_phrase.c +158 -79
- data/ext/q_prefix.c +16 -18
- data/ext/q_range.c +363 -31
- data/ext/q_span.c +130 -141
- data/ext/q_term.c +21 -21
- data/ext/q_wildcard.c +19 -23
- data/ext/r_analysis.c +369 -242
- data/ext/r_index.c +421 -434
- data/ext/r_qparser.c +142 -92
- data/ext/r_search.c +790 -407
- data/ext/r_store.c +44 -44
- data/ext/r_utils.c +264 -96
- data/ext/ram_store.c +29 -23
- 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 +210 -87
- data/ext/search.h +556 -488
- data/ext/similarity.c +17 -16
- data/ext/similarity.h +51 -44
- data/ext/sort.c +157 -354
- data/ext/stem_ISO_8859_1_hungarian.h +16 -0
- data/ext/stem_ISO_8859_2_romanian.h +16 -0
- data/ext/stem_UTF_8_hungarian.h +16 -0
- data/ext/stem_UTF_8_romanian.h +16 -0
- data/ext/stem_UTF_8_turkish.h +16 -0
- data/ext/stopwords.c +287 -278
- data/ext/store.c +57 -51
- data/ext/store.h +308 -286
- data/ext/symbol.c +10 -0
- data/ext/symbol.h +23 -0
- data/ext/term_vectors.c +14 -293
- data/ext/threading.h +22 -22
- data/ext/win32.h +12 -4
- data/lib/ferret.rb +2 -1
- data/lib/ferret/browser.rb +1 -1
- data/lib/ferret/field_symbol.rb +94 -0
- data/lib/ferret/index.rb +221 -34
- data/lib/ferret/number_tools.rb +6 -6
- data/lib/ferret/version.rb +3 -0
- data/test/{unit → long_running}/largefile/tc_largefile.rb +1 -1
- data/test/test_helper.rb +7 -2
- data/test/test_installed.rb +1 -0
- data/test/threading/thread_safety_index_test.rb +10 -1
- data/test/threading/thread_safety_read_write_test.rb +4 -7
- data/test/threading/thread_safety_test.rb +0 -0
- data/test/unit/analysis/tc_analyzer.rb +29 -27
- data/test/unit/analysis/tc_token_stream.rb +23 -16
- data/test/unit/index/tc_index.rb +116 -11
- data/test/unit/index/tc_index_reader.rb +27 -27
- data/test/unit/index/tc_index_writer.rb +10 -0
- data/test/unit/index/th_doc.rb +38 -21
- data/test/unit/search/tc_filter.rb +31 -10
- data/test/unit/search/tc_index_searcher.rb +6 -0
- data/test/unit/search/tm_searcher.rb +53 -1
- data/test/unit/store/tc_fs_store.rb +40 -2
- data/test/unit/store/tc_ram_store.rb +0 -0
- data/test/unit/store/tm_store.rb +0 -0
- data/test/unit/store/tm_store_lock.rb +7 -6
- data/test/unit/tc_field_symbol.rb +26 -0
- data/test/unit/ts_analysis.rb +0 -0
- data/test/unit/ts_index.rb +0 -0
- data/test/unit/ts_store.rb +0 -0
- data/test/unit/ts_utils.rb +0 -0
- data/test/unit/utils/tc_number_tools.rb +0 -0
- data/test/utils/content_generator.rb +226 -0
- metadata +262 -221
- data/ext/inc/lang.h +0 -48
- data/ext/inc/threading.h +0 -31
- data/ext/stem_ISO_8859_1_english.c +0 -1156
- data/ext/stem_ISO_8859_1_french.c +0 -1276
- data/ext/stem_ISO_8859_1_italian.c +0 -1091
- data/ext/stem_ISO_8859_1_norwegian.c +0 -296
- data/ext/stem_ISO_8859_1_spanish.c +0 -1119
- data/ext/stem_ISO_8859_1_swedish.c +0 -307
- data/ext/stem_UTF_8_danish.c +0 -344
- data/ext/stem_UTF_8_english.c +0 -1176
- data/ext/stem_UTF_8_french.c +0 -1296
- data/ext/stem_UTF_8_italian.c +0 -1113
- data/ext/stem_UTF_8_norwegian.c +0 -302
- data/ext/stem_UTF_8_portuguese.c +0 -1055
- data/ext/stem_UTF_8_russian.c +0 -709
- data/ext/stem_UTF_8_spanish.c +0 -1137
- data/ext/stem_UTF_8_swedish.c +0 -313
- data/lib/ferret_version.rb +0 -3
data/ext/r_index.c
CHANGED
@@ -1,6 +1,10 @@
|
|
1
1
|
#include "ferret.h"
|
2
2
|
#include "index.h"
|
3
|
-
#
|
3
|
+
#ifdef FRT_RUBY_VERSION_1_9
|
4
|
+
# include <ruby/st.h>
|
5
|
+
#else
|
6
|
+
# include <st.h>
|
7
|
+
#endif
|
4
8
|
|
5
9
|
VALUE mIndex;
|
6
10
|
|
@@ -52,15 +56,17 @@ static VALUE sym_with_positions;
|
|
52
56
|
static VALUE sym_with_offsets;
|
53
57
|
static VALUE sym_with_positions_offsets;
|
54
58
|
|
59
|
+
static Symbol fsym_content;
|
60
|
+
|
55
61
|
static ID id_term;
|
56
62
|
static ID id_fields;
|
57
63
|
static ID id_fld_num_map;
|
58
64
|
static ID id_field_num;
|
59
65
|
static ID id_boost;
|
60
66
|
|
61
|
-
extern void
|
62
|
-
extern Analyzer *
|
63
|
-
extern VALUE
|
67
|
+
extern void frb_set_term(VALUE rterm, Term *t);
|
68
|
+
extern Analyzer *frb_get_cwrapped_analyzer(VALUE ranalyzer);
|
69
|
+
extern VALUE frb_get_analyzer(Analyzer *a);
|
64
70
|
|
65
71
|
/****************************************************************************
|
66
72
|
*
|
@@ -69,17 +75,17 @@ extern VALUE frt_get_analyzer(Analyzer *a);
|
|
69
75
|
****************************************************************************/
|
70
76
|
|
71
77
|
static void
|
72
|
-
|
78
|
+
frb_fi_free(void *p)
|
73
79
|
{
|
74
80
|
object_del(p);
|
75
81
|
fi_deref((FieldInfo *)p);
|
76
82
|
}
|
77
83
|
|
78
84
|
static void
|
79
|
-
|
80
|
-
|
81
|
-
|
82
|
-
|
85
|
+
frb_fi_get_params(VALUE roptions,
|
86
|
+
StoreValue *store,
|
87
|
+
IndexValue *index,
|
88
|
+
TermVectorValue *term_vector,
|
83
89
|
float *boost)
|
84
90
|
{
|
85
91
|
VALUE v;
|
@@ -151,14 +157,14 @@ frt_fi_get_params(VALUE roptions,
|
|
151
157
|
}
|
152
158
|
|
153
159
|
static VALUE
|
154
|
-
|
160
|
+
frb_get_field_info(FieldInfo *fi)
|
155
161
|
{
|
156
162
|
|
157
163
|
VALUE rfi = Qnil;
|
158
164
|
if (fi) {
|
159
165
|
rfi = object_get(fi);
|
160
166
|
if (rfi == Qnil) {
|
161
|
-
rfi = Data_Wrap_Struct(cFieldInfo, NULL, &
|
167
|
+
rfi = Data_Wrap_Struct(cFieldInfo, NULL, &frb_fi_free, fi);
|
162
168
|
REF(fi);
|
163
169
|
object_add(fi, rfi);
|
164
170
|
}
|
@@ -176,22 +182,22 @@ frt_get_field_info(FieldInfo *fi)
|
|
176
182
|
* information on these properties.
|
177
183
|
*/
|
178
184
|
static VALUE
|
179
|
-
|
185
|
+
frb_fi_init(int argc, VALUE *argv, VALUE self)
|
180
186
|
{
|
181
187
|
VALUE roptions, rname;
|
182
188
|
FieldInfo *fi;
|
183
|
-
|
184
|
-
|
185
|
-
|
189
|
+
StoreValue store = STORE_YES;
|
190
|
+
IndexValue index = INDEX_YES;
|
191
|
+
TermVectorValue term_vector = TERM_VECTOR_WITH_POSITIONS_OFFSETS;
|
186
192
|
float boost = 1.0f;
|
187
193
|
|
188
194
|
rb_scan_args(argc, argv, "11", &rname, &roptions);
|
189
195
|
if (argc > 1) {
|
190
|
-
|
196
|
+
frb_fi_get_params(roptions, &store, &index, &term_vector, &boost);
|
191
197
|
}
|
192
|
-
fi = fi_new(
|
198
|
+
fi = fi_new(frb_field(rname), store, index, term_vector);
|
193
199
|
fi->boost = boost;
|
194
|
-
Frt_Wrap_Struct(self, NULL, &
|
200
|
+
Frt_Wrap_Struct(self, NULL, &frb_fi_free, fi);
|
195
201
|
object_add(fi, self);
|
196
202
|
return self;
|
197
203
|
}
|
@@ -203,10 +209,10 @@ frt_fi_init(int argc, VALUE *argv, VALUE self)
|
|
203
209
|
* Return the name of the field
|
204
210
|
*/
|
205
211
|
static VALUE
|
206
|
-
|
212
|
+
frb_fi_name(VALUE self)
|
207
213
|
{
|
208
214
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
209
|
-
return ID2SYM(
|
215
|
+
return ID2SYM(fi->name);
|
210
216
|
}
|
211
217
|
|
212
218
|
/*
|
@@ -216,7 +222,7 @@ frt_fi_name(VALUE self)
|
|
216
222
|
* Return true if the field is stored in the index.
|
217
223
|
*/
|
218
224
|
static VALUE
|
219
|
-
|
225
|
+
frb_fi_is_stored(VALUE self)
|
220
226
|
{
|
221
227
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
222
228
|
return fi_is_stored(fi) ? Qtrue : Qfalse;
|
@@ -229,7 +235,7 @@ frt_fi_is_stored(VALUE self)
|
|
229
235
|
* Return true if the field is stored in the index in compressed format.
|
230
236
|
*/
|
231
237
|
static VALUE
|
232
|
-
|
238
|
+
frb_fi_is_compressed(VALUE self)
|
233
239
|
{
|
234
240
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
235
241
|
return fi_is_compressed(fi) ? Qtrue : Qfalse;
|
@@ -242,7 +248,7 @@ frt_fi_is_compressed(VALUE self)
|
|
242
248
|
* Return true if the field is indexed, ie searchable in the index.
|
243
249
|
*/
|
244
250
|
static VALUE
|
245
|
-
|
251
|
+
frb_fi_is_indexed(VALUE self)
|
246
252
|
{
|
247
253
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
248
254
|
return fi_is_indexed(fi) ? Qtrue : Qfalse;
|
@@ -260,7 +266,7 @@ frt_fi_is_indexed(VALUE self)
|
|
260
266
|
* A field can only be tokenized if it is indexed.
|
261
267
|
*/
|
262
268
|
static VALUE
|
263
|
-
|
269
|
+
frb_fi_is_tokenized(VALUE self)
|
264
270
|
{
|
265
271
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
266
272
|
return fi_is_tokenized(fi) ? Qtrue : Qfalse;
|
@@ -278,7 +284,7 @@ frt_fi_is_tokenized(VALUE self)
|
|
278
284
|
* large number of documents.
|
279
285
|
*/
|
280
286
|
static VALUE
|
281
|
-
|
287
|
+
frb_fi_omit_norms(VALUE self)
|
282
288
|
{
|
283
289
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
284
290
|
return fi_omit_norms(fi) ? Qtrue : Qfalse;
|
@@ -291,7 +297,7 @@ frt_fi_omit_norms(VALUE self)
|
|
291
297
|
* Return true if the term-vectors are stored for this field.
|
292
298
|
*/
|
293
299
|
static VALUE
|
294
|
-
|
300
|
+
frb_fi_store_term_vector(VALUE self)
|
295
301
|
{
|
296
302
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
297
303
|
return fi_store_term_vector(fi) ? Qtrue : Qfalse;
|
@@ -304,7 +310,7 @@ frt_fi_store_term_vector(VALUE self)
|
|
304
310
|
* Return true if positions are stored with the term-vectors for this field.
|
305
311
|
*/
|
306
312
|
static VALUE
|
307
|
-
|
313
|
+
frb_fi_store_positions(VALUE self)
|
308
314
|
{
|
309
315
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
310
316
|
return fi_store_positions(fi) ? Qtrue : Qfalse;
|
@@ -317,7 +323,7 @@ frt_fi_store_positions(VALUE self)
|
|
317
323
|
* Return true if offsets are stored with the term-vectors for this field.
|
318
324
|
*/
|
319
325
|
static VALUE
|
320
|
-
|
326
|
+
frb_fi_store_offsets(VALUE self)
|
321
327
|
{
|
322
328
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
323
329
|
return fi_store_offsets(fi) ? Qtrue : Qfalse;
|
@@ -332,7 +338,7 @@ frt_fi_store_offsets(VALUE self)
|
|
332
338
|
* fi.indexed? and not fi.omit_norms?
|
333
339
|
*/
|
334
340
|
static VALUE
|
335
|
-
|
341
|
+
frb_fi_has_norms(VALUE self)
|
336
342
|
{
|
337
343
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
338
344
|
return fi_has_norms(fi) ? Qtrue : Qfalse;
|
@@ -345,7 +351,7 @@ frt_fi_has_norms(VALUE self)
|
|
345
351
|
* Return the default boost for this field
|
346
352
|
*/
|
347
353
|
static VALUE
|
348
|
-
|
354
|
+
frb_fi_boost(VALUE self)
|
349
355
|
{
|
350
356
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
351
357
|
return rb_float_new((double)fi->boost);
|
@@ -358,7 +364,7 @@ frt_fi_boost(VALUE self)
|
|
358
364
|
* Return a string representation of the FieldInfo object.
|
359
365
|
*/
|
360
366
|
static VALUE
|
361
|
-
|
367
|
+
frb_fi_to_s(VALUE self)
|
362
368
|
{
|
363
369
|
FieldInfo *fi = (FieldInfo *)DATA_PTR(self);
|
364
370
|
char *fi_s = fi_to_s(fi);
|
@@ -374,32 +380,32 @@ frt_fi_to_s(VALUE self)
|
|
374
380
|
****************************************************************************/
|
375
381
|
|
376
382
|
static void
|
377
|
-
|
383
|
+
frb_fis_free(void *p)
|
378
384
|
{
|
379
385
|
object_del(p);
|
380
386
|
fis_deref((FieldInfos *)p);
|
381
387
|
}
|
382
388
|
|
383
389
|
static void
|
384
|
-
|
390
|
+
frb_fis_mark(void *p)
|
385
391
|
{
|
386
392
|
int i;
|
387
393
|
FieldInfos *fis = (FieldInfos *)p;
|
388
394
|
|
389
395
|
for (i = 0; i < fis->size; i++) {
|
390
|
-
|
396
|
+
frb_gc_mark(fis->fields[i]);
|
391
397
|
}
|
392
398
|
}
|
393
399
|
|
394
400
|
static VALUE
|
395
|
-
|
401
|
+
frb_get_field_infos(FieldInfos *fis)
|
396
402
|
{
|
397
403
|
|
398
404
|
VALUE rfis = Qnil;
|
399
405
|
if (fis) {
|
400
406
|
rfis = object_get(fis);
|
401
407
|
if (rfis == Qnil) {
|
402
|
-
rfis = Data_Wrap_Struct(cFieldInfos, &
|
408
|
+
rfis = Data_Wrap_Struct(cFieldInfos, &frb_fis_mark, &frb_fis_free,
|
403
409
|
fis);
|
404
410
|
REF(fis);
|
405
411
|
object_add(fis, rfis);
|
@@ -417,21 +423,21 @@ frt_get_field_infos(FieldInfos *fis)
|
|
417
423
|
* property values.
|
418
424
|
*/
|
419
425
|
static VALUE
|
420
|
-
|
426
|
+
frb_fis_init(int argc, VALUE *argv, VALUE self)
|
421
427
|
{
|
422
428
|
VALUE roptions;
|
423
429
|
FieldInfos *fis;
|
424
|
-
|
425
|
-
|
426
|
-
|
430
|
+
StoreValue store = STORE_YES;
|
431
|
+
IndexValue index = INDEX_YES;
|
432
|
+
TermVectorValue term_vector = TERM_VECTOR_WITH_POSITIONS_OFFSETS;
|
427
433
|
float boost;
|
428
434
|
|
429
435
|
rb_scan_args(argc, argv, "01", &roptions);
|
430
436
|
if (argc > 0) {
|
431
|
-
|
437
|
+
frb_fi_get_params(roptions, &store, &index, &term_vector, &boost);
|
432
438
|
}
|
433
439
|
fis = fis_new(store, index, term_vector);
|
434
|
-
Frt_Wrap_Struct(self, &
|
440
|
+
Frt_Wrap_Struct(self, &frb_fis_mark, &frb_fis_free, fis);
|
435
441
|
object_add(fis, self);
|
436
442
|
return self;
|
437
443
|
}
|
@@ -444,14 +450,14 @@ frt_fis_init(int argc, VALUE *argv, VALUE self)
|
|
444
450
|
* object.
|
445
451
|
*/
|
446
452
|
static VALUE
|
447
|
-
|
453
|
+
frb_fis_to_a(VALUE self)
|
448
454
|
{
|
449
455
|
FieldInfos *fis = (FieldInfos *)DATA_PTR(self);
|
450
456
|
VALUE rary = rb_ary_new();
|
451
457
|
int i;
|
452
458
|
|
453
459
|
for (i = 0; i < fis->size; i++) {
|
454
|
-
rb_ary_push(rary,
|
460
|
+
rb_ary_push(rary, frb_get_field_info(fis->fields[i]));
|
455
461
|
}
|
456
462
|
return rary;
|
457
463
|
}
|
@@ -469,7 +475,7 @@ frt_fis_to_a(VALUE self)
|
|
469
475
|
* fi = fis[2]
|
470
476
|
*/
|
471
477
|
static VALUE
|
472
|
-
|
478
|
+
frb_fis_get(VALUE self, VALUE ridx)
|
473
479
|
{
|
474
480
|
FieldInfos *fis = (FieldInfos *)DATA_PTR(self);
|
475
481
|
VALUE rfi = Qnil;
|
@@ -479,17 +485,20 @@ frt_fis_get(VALUE self, VALUE ridx)
|
|
479
485
|
if (index < 0) index += fis->size;
|
480
486
|
if (index < 0 || index >= fis->size) {
|
481
487
|
rb_raise(rb_eArgError, "index of %d is out of range (0..%d)\n",
|
482
|
-
index, fis->size);
|
488
|
+
index, fis->size - 1);
|
483
489
|
}
|
484
|
-
rfi =
|
490
|
+
rfi = frb_get_field_info(fis->fields[index]);
|
485
491
|
break;
|
486
492
|
}
|
487
493
|
case T_SYMBOL:
|
488
|
-
|
494
|
+
case T_STRING:
|
495
|
+
rfi = frb_get_field_info(fis_get_field(fis, frb_field(ridx)));
|
489
496
|
break;
|
497
|
+
/*
|
490
498
|
case T_STRING:
|
491
|
-
rfi =
|
499
|
+
rfi = frb_get_field_info(fis_get_field(fis, StringValuePtr(ridx)));
|
492
500
|
break;
|
501
|
+
*/
|
493
502
|
default:
|
494
503
|
rb_raise(rb_eArgError, "Can't index FieldInfos with %s",
|
495
504
|
rs2s(rb_obj_as_string(ridx)));
|
@@ -507,10 +516,10 @@ frt_fis_get(VALUE self, VALUE ridx)
|
|
507
516
|
* possible.
|
508
517
|
*/
|
509
518
|
static VALUE
|
510
|
-
|
519
|
+
frb_fis_add(VALUE self, VALUE rfi)
|
511
520
|
{
|
512
521
|
FieldInfos *fis = (FieldInfos *)DATA_PTR(self);
|
513
|
-
FieldInfo *fi = (FieldInfo *)
|
522
|
+
FieldInfo *fi = (FieldInfo *)frb_rb_data_ptr(rfi);
|
514
523
|
fis_add_field(fis, fi);
|
515
524
|
REF(fi);
|
516
525
|
return self;
|
@@ -524,21 +533,21 @@ frt_fis_add(VALUE self, VALUE rfi)
|
|
524
533
|
* of the available properties.
|
525
534
|
*/
|
526
535
|
static VALUE
|
527
|
-
|
536
|
+
frb_fis_add_field(int argc, VALUE *argv, VALUE self)
|
528
537
|
{
|
529
538
|
FieldInfos *fis = (FieldInfos *)DATA_PTR(self);
|
530
539
|
FieldInfo *fi;
|
531
|
-
|
532
|
-
|
533
|
-
|
540
|
+
StoreValue store = fis->store;
|
541
|
+
IndexValue index = fis->index;
|
542
|
+
TermVectorValue term_vector = fis->term_vector;
|
534
543
|
float boost = 1.0f;
|
535
544
|
VALUE rname, roptions;
|
536
545
|
|
537
546
|
rb_scan_args(argc, argv, "11", &rname, &roptions);
|
538
547
|
if (argc > 1) {
|
539
|
-
|
548
|
+
frb_fi_get_params(roptions, &store, &index, &term_vector, &boost);
|
540
549
|
}
|
541
|
-
fi = fi_new(
|
550
|
+
fi = fi_new(frb_field(rname), store, index, term_vector);
|
542
551
|
fi->boost = boost;
|
543
552
|
fis_add_field(fis, fi);
|
544
553
|
return self;
|
@@ -551,13 +560,13 @@ frt_fis_add_field(int argc, VALUE *argv, VALUE self)
|
|
551
560
|
* Iterate through the FieldInfo objects.
|
552
561
|
*/
|
553
562
|
static VALUE
|
554
|
-
|
563
|
+
frb_fis_each(VALUE self)
|
555
564
|
{
|
556
565
|
int i;
|
557
566
|
FieldInfos *fis = (FieldInfos *)DATA_PTR(self);
|
558
567
|
|
559
568
|
for (i = 0; i < fis->size; i++) {
|
560
|
-
rb_yield(
|
569
|
+
rb_yield(frb_get_field_info(fis->fields[i]));
|
561
570
|
}
|
562
571
|
return self;
|
563
572
|
}
|
@@ -569,7 +578,7 @@ frt_fis_each(VALUE self)
|
|
569
578
|
* Return a string representation of the FieldInfos object.
|
570
579
|
*/
|
571
580
|
static VALUE
|
572
|
-
|
581
|
+
frb_fis_to_s(VALUE self)
|
573
582
|
{
|
574
583
|
FieldInfos *fis = (FieldInfos *)DATA_PTR(self);
|
575
584
|
char *fis_s = fis_to_s(fis);
|
@@ -585,7 +594,7 @@ frt_fis_to_s(VALUE self)
|
|
585
594
|
* Return the number of fields in the FieldInfos object.
|
586
595
|
*/
|
587
596
|
static VALUE
|
588
|
-
|
597
|
+
frb_fis_size(VALUE self)
|
589
598
|
{
|
590
599
|
FieldInfos *fis = (FieldInfos *)DATA_PTR(self);
|
591
600
|
return INT2FIX(fis->size);
|
@@ -602,7 +611,7 @@ frt_fis_size(VALUE self)
|
|
602
611
|
* directory and overwritten by the new index.
|
603
612
|
*/
|
604
613
|
static VALUE
|
605
|
-
|
614
|
+
frb_fis_create_index(VALUE self, VALUE rdir)
|
606
615
|
{
|
607
616
|
FieldInfos *fis = (FieldInfos *)DATA_PTR(self);
|
608
617
|
Store *store = NULL;
|
@@ -611,7 +620,7 @@ frt_fis_create_index(VALUE self, VALUE rdir)
|
|
611
620
|
REF(store);
|
612
621
|
} else {
|
613
622
|
StringValue(rdir);
|
614
|
-
|
623
|
+
frb_create_dir(rdir);
|
615
624
|
store = open_fs_store(rs2s(rdir));
|
616
625
|
}
|
617
626
|
index_create(store, fis);
|
@@ -622,18 +631,19 @@ frt_fis_create_index(VALUE self, VALUE rdir)
|
|
622
631
|
/*
|
623
632
|
* call-seq:
|
624
633
|
* fis.fields -> symbol array
|
634
|
+
* fis.field_names -> symbol array
|
625
635
|
*
|
626
636
|
* Return a list of the field names (as symbols) of all the fields in the
|
627
637
|
* index.
|
628
638
|
*/
|
629
639
|
static VALUE
|
630
|
-
|
640
|
+
frb_fis_get_fields(VALUE self)
|
631
641
|
{
|
632
642
|
FieldInfos *fis = (FieldInfos *)DATA_PTR(self);
|
633
643
|
VALUE rfield_names = rb_ary_new();
|
634
644
|
int i;
|
635
645
|
for (i = 0; i < fis->size; i++) {
|
636
|
-
rb_ary_push(rfield_names,
|
646
|
+
rb_ary_push(rfield_names, FSYM2SYM(fis->fields[i]->name));
|
637
647
|
}
|
638
648
|
return rfield_names;
|
639
649
|
}
|
@@ -646,14 +656,14 @@ frt_fis_get_fields(VALUE self)
|
|
646
656
|
* in the index.
|
647
657
|
*/
|
648
658
|
static VALUE
|
649
|
-
|
659
|
+
frb_fis_get_tk_fields(VALUE self)
|
650
660
|
{
|
651
661
|
FieldInfos *fis = (FieldInfos *)DATA_PTR(self);
|
652
662
|
VALUE rfield_names = rb_ary_new();
|
653
663
|
int i;
|
654
664
|
for (i = 0; i < fis->size; i++) {
|
655
665
|
if (!fi_is_tokenized(fis->fields[i])) continue;
|
656
|
-
rb_ary_push(rfield_names,
|
666
|
+
rb_ary_push(rfield_names, FSYM2SYM(fis->fields[i]->name));
|
657
667
|
}
|
658
668
|
return rfield_names;
|
659
669
|
}
|
@@ -665,14 +675,14 @@ frt_fis_get_tk_fields(VALUE self)
|
|
665
675
|
****************************************************************************/
|
666
676
|
|
667
677
|
static void
|
668
|
-
|
678
|
+
frb_te_free(void *p)
|
669
679
|
{
|
670
680
|
TermEnum *te = (TermEnum *)p;
|
671
681
|
te->close(te);
|
672
682
|
}
|
673
683
|
|
674
684
|
static VALUE
|
675
|
-
|
685
|
+
frb_te_get_set_term(VALUE self, const char *term)
|
676
686
|
{
|
677
687
|
TermEnum *te = (TermEnum *)DATA_PTR(self);
|
678
688
|
VALUE str = term ? rb_str_new(term, te->curr_term_len) : Qnil;
|
@@ -681,12 +691,12 @@ frt_te_get_set_term(VALUE self, const char *term)
|
|
681
691
|
}
|
682
692
|
|
683
693
|
static VALUE
|
684
|
-
|
694
|
+
frb_get_te(VALUE rir, TermEnum *te)
|
685
695
|
{
|
686
696
|
VALUE self = Qnil;
|
687
697
|
if (te != NULL) {
|
688
|
-
self = Data_Wrap_Struct(cTermEnum, NULL, &
|
689
|
-
|
698
|
+
self = Data_Wrap_Struct(cTermEnum, NULL, &frb_te_free, te);
|
699
|
+
frb_te_get_set_term(self, te->curr_term);
|
690
700
|
rb_ivar_set(self, id_fld_num_map, rb_ivar_get(rir, id_fld_num_map));
|
691
701
|
}
|
692
702
|
return self;
|
@@ -699,10 +709,10 @@ frt_get_te(VALUE rir, TermEnum *te)
|
|
699
709
|
* Returns the next term in the enumeration or nil otherwise.
|
700
710
|
*/
|
701
711
|
static VALUE
|
702
|
-
|
712
|
+
frb_te_next(VALUE self)
|
703
713
|
{
|
704
714
|
TermEnum *te = (TermEnum *)DATA_PTR(self);
|
705
|
-
return
|
715
|
+
return frb_te_get_set_term(self, te->next(te));
|
706
716
|
}
|
707
717
|
|
708
718
|
/*
|
@@ -713,7 +723,7 @@ frt_te_next(VALUE self)
|
|
713
723
|
* be called after a successful call to TermEnum#next.
|
714
724
|
*/
|
715
725
|
static VALUE
|
716
|
-
|
726
|
+
frb_te_term(VALUE self)
|
717
727
|
{
|
718
728
|
return rb_ivar_get(self, id_term);
|
719
729
|
}
|
@@ -727,7 +737,7 @@ frt_te_term(VALUE self)
|
|
727
737
|
* should only be called after a successful call to TermEnum#next.
|
728
738
|
*/
|
729
739
|
static VALUE
|
730
|
-
|
740
|
+
frb_te_doc_freq(VALUE self)
|
731
741
|
{
|
732
742
|
TermEnum *te = (TermEnum *)DATA_PTR(self);
|
733
743
|
return INT2FIX(te->curr_ti.doc_freq);
|
@@ -745,10 +755,10 @@ frt_te_doc_freq(VALUE self)
|
|
745
755
|
* Returns the first term greater than or equal to +target+
|
746
756
|
*/
|
747
757
|
static VALUE
|
748
|
-
|
758
|
+
frb_te_skip_to(VALUE self, VALUE rterm)
|
749
759
|
{
|
750
760
|
TermEnum *te = (TermEnum *)DATA_PTR(self);
|
751
|
-
return
|
761
|
+
return frb_te_get_set_term(self, te->skip_to(te, rs2s(rterm)));
|
752
762
|
}
|
753
763
|
|
754
764
|
/*
|
@@ -759,15 +769,14 @@ frt_te_skip_to(VALUE self, VALUE rterm)
|
|
759
769
|
* document frequency.
|
760
770
|
*/
|
761
771
|
static VALUE
|
762
|
-
|
772
|
+
frb_te_each(VALUE self)
|
763
773
|
{
|
764
774
|
TermEnum *te = (TermEnum *)DATA_PTR(self);
|
765
775
|
char *term;
|
766
776
|
int term_cnt = 0;
|
767
777
|
VALUE vals = rb_ary_new2(2);
|
768
|
-
|
769
|
-
|
770
|
-
|
778
|
+
rb_ary_store(vals, 0, Qnil);
|
779
|
+
rb_ary_store(vals, 1, Qnil);
|
771
780
|
|
772
781
|
/* each is being called so there will be no current term */
|
773
782
|
rb_ivar_set(self, id_term, Qnil);
|
@@ -775,8 +784,8 @@ frt_te_each(VALUE self)
|
|
775
784
|
|
776
785
|
while (NULL != (term = te->next(te))) {
|
777
786
|
term_cnt++;
|
778
|
-
|
779
|
-
|
787
|
+
RARRAY_PTR(vals)[0] = rb_str_new(term, te->curr_term_len);
|
788
|
+
RARRAY_PTR(vals)[1] = INT2FIX(te->curr_ti.doc_freq);
|
780
789
|
rb_yield(vals);
|
781
790
|
}
|
782
791
|
return INT2FIX(term_cnt);
|
@@ -794,7 +803,7 @@ frt_te_each(VALUE self)
|
|
794
803
|
* end
|
795
804
|
*/
|
796
805
|
static VALUE
|
797
|
-
|
806
|
+
frb_te_set_field(VALUE self, VALUE rfield)
|
798
807
|
{
|
799
808
|
TermEnum *te = (TermEnum *)DATA_PTR(self);
|
800
809
|
int field_num = 0;
|
@@ -806,7 +815,7 @@ frt_te_set_field(VALUE self, VALUE rfield)
|
|
806
815
|
} else {
|
807
816
|
Check_Type(rfield, T_SYMBOL);
|
808
817
|
rb_raise(rb_eArgError, "field %s doesn't exist in the index",
|
809
|
-
|
818
|
+
(char *)frb_field(rfield));
|
810
819
|
}
|
811
820
|
te->set_field(te, field_num);
|
812
821
|
|
@@ -836,7 +845,7 @@ frt_te_set_field(VALUE self, VALUE rfield)
|
|
836
845
|
* # ]
|
837
846
|
*/
|
838
847
|
static VALUE
|
839
|
-
|
848
|
+
frb_te_to_json(int argc, VALUE *argv, VALUE self)
|
840
849
|
{
|
841
850
|
TermEnum *te = (TermEnum *)DATA_PTR(self);
|
842
851
|
VALUE rjson;
|
@@ -900,16 +909,16 @@ frt_te_to_json(int argc, VALUE *argv, VALUE self)
|
|
900
909
|
****************************************************************************/
|
901
910
|
|
902
911
|
static void
|
903
|
-
|
912
|
+
frb_tde_free(void *p)
|
904
913
|
{
|
905
914
|
TermDocEnum *tde = (TermDocEnum *)p;
|
906
915
|
tde->close(tde);
|
907
916
|
}
|
908
917
|
|
909
918
|
static VALUE
|
910
|
-
|
919
|
+
frb_get_tde(VALUE rir, TermDocEnum *tde)
|
911
920
|
{
|
912
|
-
VALUE self = Data_Wrap_Struct(cTermDocEnum, NULL, &
|
921
|
+
VALUE self = Data_Wrap_Struct(cTermDocEnum, NULL, &frb_tde_free, tde);
|
913
922
|
rb_ivar_set(self, id_fld_num_map, rb_ivar_get(rir, id_fld_num_map));
|
914
923
|
return self;
|
915
924
|
}
|
@@ -923,7 +932,7 @@ frt_get_tde(VALUE rir, TermDocEnum *tde)
|
|
923
932
|
* this particular term.
|
924
933
|
*/
|
925
934
|
static VALUE
|
926
|
-
|
935
|
+
frb_tde_seek(VALUE self, VALUE rfield, VALUE rterm)
|
927
936
|
{
|
928
937
|
TermDocEnum *tde = (TermDocEnum *)DATA_PTR(self);
|
929
938
|
char *term;
|
@@ -935,7 +944,7 @@ frt_tde_seek(VALUE self, VALUE rfield, VALUE rterm)
|
|
935
944
|
field_num = FIX2INT(rfnum);
|
936
945
|
} else {
|
937
946
|
rb_raise(rb_eArgError, "field %s doesn't exist in the index",
|
938
|
-
|
947
|
+
(char *)frb_field(rfield));
|
939
948
|
}
|
940
949
|
tde->seek(tde, field_num, term);
|
941
950
|
return self;
|
@@ -954,10 +963,10 @@ frt_tde_seek(VALUE self, VALUE rfield, VALUE rterm)
|
|
954
963
|
* a large performance improvement.
|
955
964
|
*/
|
956
965
|
static VALUE
|
957
|
-
|
966
|
+
frb_tde_seek_te(VALUE self, VALUE rterm_enum)
|
958
967
|
{
|
959
968
|
TermDocEnum *tde = (TermDocEnum *)DATA_PTR(self);
|
960
|
-
TermEnum *te = (TermEnum *)
|
969
|
+
TermEnum *te = (TermEnum *)frb_rb_data_ptr(rterm_enum);
|
961
970
|
tde->seek_te(tde, te);
|
962
971
|
return self;
|
963
972
|
}
|
@@ -969,7 +978,7 @@ frt_tde_seek_te(VALUE self, VALUE rterm_enum)
|
|
969
978
|
* Returns the current document number pointed to by the +term_doc_enum+.
|
970
979
|
*/
|
971
980
|
static VALUE
|
972
|
-
|
981
|
+
frb_tde_doc(VALUE self)
|
973
982
|
{
|
974
983
|
TermDocEnum *tde = (TermDocEnum *)DATA_PTR(self);
|
975
984
|
return INT2FIX(tde->doc_num(tde));
|
@@ -983,7 +992,7 @@ frt_tde_doc(VALUE self)
|
|
983
992
|
* +term_doc_enum+.
|
984
993
|
*/
|
985
994
|
static VALUE
|
986
|
-
|
995
|
+
frb_tde_freq(VALUE self)
|
987
996
|
{
|
988
997
|
TermDocEnum *tde = (TermDocEnum *)DATA_PTR(self);
|
989
998
|
return INT2FIX(tde->freq(tde));
|
@@ -997,7 +1006,7 @@ frt_tde_freq(VALUE self)
|
|
997
1006
|
* there is another document or +false+ otherwise.
|
998
1007
|
*/
|
999
1008
|
static VALUE
|
1000
|
-
|
1009
|
+
frb_tde_next(VALUE self)
|
1001
1010
|
{
|
1002
1011
|
TermDocEnum *tde = (TermDocEnum *)DATA_PTR(self);
|
1003
1012
|
return tde->next(tde) ? Qtrue : Qfalse;
|
@@ -1011,7 +1020,7 @@ frt_tde_next(VALUE self)
|
|
1011
1020
|
* there is another document or +false+ otherwise.
|
1012
1021
|
*/
|
1013
1022
|
static VALUE
|
1014
|
-
|
1023
|
+
frb_tde_next_position(VALUE self)
|
1015
1024
|
{
|
1016
1025
|
TermDocEnum *tde = (TermDocEnum *)DATA_PTR(self);
|
1017
1026
|
int pos;
|
@@ -1035,18 +1044,18 @@ frt_tde_next_position(VALUE self)
|
|
1035
1044
|
* call +#each+ again then you should call +#seek+ again too.
|
1036
1045
|
*/
|
1037
1046
|
static VALUE
|
1038
|
-
|
1047
|
+
frb_tde_each(VALUE self)
|
1039
1048
|
{
|
1040
1049
|
int doc_cnt = 0;
|
1041
1050
|
TermDocEnum *tde = (TermDocEnum *)DATA_PTR(self);
|
1042
1051
|
VALUE vals = rb_ary_new2(2);
|
1043
|
-
|
1044
|
-
|
1052
|
+
rb_ary_store(vals, 0, Qnil);
|
1053
|
+
rb_ary_store(vals, 1, Qnil);
|
1045
1054
|
|
1046
1055
|
while (tde->next(tde)) {
|
1047
1056
|
doc_cnt++;
|
1048
|
-
|
1049
|
-
|
1057
|
+
RARRAY_PTR(vals)[0] = INT2FIX(tde->doc_num(tde));
|
1058
|
+
RARRAY_PTR(vals)[1] = INT2FIX(tde->freq(tde));
|
1050
1059
|
rb_yield(vals);
|
1051
1060
|
|
1052
1061
|
}
|
@@ -1079,7 +1088,7 @@ frt_tde_each(VALUE self)
|
|
1079
1088
|
* # ]
|
1080
1089
|
*/
|
1081
1090
|
static VALUE
|
1082
|
-
|
1091
|
+
frb_tde_to_json(int argc, VALUE *argv, VALUE self)
|
1083
1092
|
{
|
1084
1093
|
TermDocEnum *tde = (TermDocEnum *)DATA_PTR(self);
|
1085
1094
|
VALUE rjson;
|
@@ -1153,7 +1162,7 @@ frt_tde_to_json(int argc, VALUE *argv, VALUE self)
|
|
1153
1162
|
* end
|
1154
1163
|
*/
|
1155
1164
|
static VALUE
|
1156
|
-
|
1165
|
+
frb_tde_each_position(VALUE self)
|
1157
1166
|
{
|
1158
1167
|
TermDocEnum *tde = (TermDocEnum *)DATA_PTR(self);
|
1159
1168
|
int pos;
|
@@ -1176,7 +1185,7 @@ frt_tde_each_position(VALUE self)
|
|
1176
1185
|
* a document >= +target+.
|
1177
1186
|
*/
|
1178
1187
|
static VALUE
|
1179
|
-
|
1188
|
+
frb_tde_skip_to(VALUE self, VALUE rtarget)
|
1180
1189
|
{
|
1181
1190
|
TermDocEnum *tde = (TermDocEnum *)DATA_PTR(self);
|
1182
1191
|
return tde->skip_to(tde, FIX2INT(rtarget)) ? Qtrue : Qfalse;
|
@@ -1189,11 +1198,11 @@ frt_tde_skip_to(VALUE self, VALUE rtarget)
|
|
1189
1198
|
****************************************************************************/
|
1190
1199
|
|
1191
1200
|
static VALUE
|
1192
|
-
|
1201
|
+
frb_get_tv_offsets(Offset *offset)
|
1193
1202
|
{
|
1194
1203
|
return rb_struct_new(cTVOffsets,
|
1195
|
-
ULL2NUM((
|
1196
|
-
ULL2NUM((
|
1204
|
+
ULL2NUM((u64)offset->start),
|
1205
|
+
ULL2NUM((u64)offset->end),
|
1197
1206
|
NULL);
|
1198
1207
|
}
|
1199
1208
|
|
@@ -1204,7 +1213,7 @@ frt_get_tv_offsets(Offset *offset)
|
|
1204
1213
|
****************************************************************************/
|
1205
1214
|
|
1206
1215
|
static VALUE
|
1207
|
-
|
1216
|
+
frb_get_tv_term(TVTerm *tv_term)
|
1208
1217
|
{
|
1209
1218
|
int i;
|
1210
1219
|
const int freq = tv_term->freq;
|
@@ -1212,16 +1221,13 @@ frt_get_tv_term(TVTerm *tv_term)
|
|
1212
1221
|
VALUE rpositions = Qnil;
|
1213
1222
|
rtext = rb_str_new2(tv_term->text);
|
1214
1223
|
if (tv_term->positions) {
|
1215
|
-
VALUE *rpos;
|
1216
1224
|
int *positions = tv_term->positions;
|
1217
1225
|
rpositions = rb_ary_new2(freq);
|
1218
|
-
rpos = RARRAY(rpositions)->ptr;
|
1219
1226
|
for (i = 0; i < freq; i++) {
|
1220
|
-
|
1227
|
+
rb_ary_store(rpositions, i, INT2FIX(positions[i]));
|
1221
1228
|
}
|
1222
|
-
RARRAY(rpositions)->len = freq;
|
1223
1229
|
}
|
1224
|
-
return rb_struct_new(cTVTerm, rtext, rpositions, NULL);
|
1230
|
+
return rb_struct_new(cTVTerm, rtext, INT2FIX(freq), rpositions, NULL);
|
1225
1231
|
}
|
1226
1232
|
|
1227
1233
|
/****************************************************************************
|
@@ -1231,31 +1237,26 @@ frt_get_tv_term(TVTerm *tv_term)
|
|
1231
1237
|
****************************************************************************/
|
1232
1238
|
|
1233
1239
|
static VALUE
|
1234
|
-
|
1240
|
+
frb_get_tv(TermVector *tv)
|
1235
1241
|
{
|
1236
1242
|
int i;
|
1237
1243
|
TVTerm *terms = tv->terms;
|
1238
1244
|
const int t_cnt = tv->term_cnt;
|
1239
1245
|
const int o_cnt = tv->offset_cnt;
|
1240
|
-
VALUE rfield, rterms
|
1246
|
+
VALUE rfield, rterms;
|
1241
1247
|
VALUE roffsets = Qnil;
|
1242
|
-
rfield =
|
1248
|
+
rfield = FSYM2SYM(tv->field);
|
1243
1249
|
|
1244
1250
|
rterms = rb_ary_new2(t_cnt);
|
1245
|
-
rts = RARRAY(rterms)->ptr;
|
1246
1251
|
for (i = 0; i < t_cnt; i++) {
|
1247
|
-
|
1248
|
-
RARRAY(rterms)->len++;
|
1252
|
+
rb_ary_store(rterms, i, frb_get_tv_term(&terms[i]));
|
1249
1253
|
}
|
1250
1254
|
|
1251
1255
|
if (tv->offsets) {
|
1252
|
-
VALUE *ros;
|
1253
1256
|
Offset *offsets = tv->offsets;
|
1254
1257
|
roffsets = rb_ary_new2(o_cnt);
|
1255
|
-
ros = RARRAY(roffsets)->ptr;
|
1256
1258
|
for (i = 0; i < o_cnt; i++) {
|
1257
|
-
|
1258
|
-
RARRAY(roffsets)->len++;
|
1259
|
+
rb_ary_store(roffsets, i, frb_get_tv_offsets(&offsets[i]));
|
1259
1260
|
}
|
1260
1261
|
}
|
1261
1262
|
|
@@ -1269,18 +1270,18 @@ frt_get_tv(TermVector *tv)
|
|
1269
1270
|
****************************************************************************/
|
1270
1271
|
|
1271
1272
|
void
|
1272
|
-
|
1273
|
+
frb_iw_free(void *p)
|
1273
1274
|
{
|
1274
1275
|
iw_close((IndexWriter *)p);
|
1275
1276
|
}
|
1276
1277
|
|
1277
1278
|
void
|
1278
|
-
|
1279
|
+
frb_iw_mark(void *p)
|
1279
1280
|
{
|
1280
1281
|
IndexWriter *iw = (IndexWriter *)p;
|
1281
|
-
|
1282
|
-
|
1283
|
-
|
1282
|
+
frb_gc_mark(iw->analyzer);
|
1283
|
+
frb_gc_mark(iw->store);
|
1284
|
+
frb_gc_mark(iw->fis);
|
1284
1285
|
}
|
1285
1286
|
|
1286
1287
|
/*
|
@@ -1292,7 +1293,7 @@ frt_iw_mark(void *p)
|
|
1292
1293
|
* automatically if not called explicitly.
|
1293
1294
|
*/
|
1294
1295
|
static VALUE
|
1295
|
-
|
1296
|
+
frb_iw_close(VALUE self)
|
1296
1297
|
{
|
1297
1298
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1298
1299
|
Frt_Unwrap_Struct(self);
|
@@ -1325,7 +1326,7 @@ frt_iw_close(VALUE self)
|
|
1325
1326
|
* See IndexWriter for more options.
|
1326
1327
|
*/
|
1327
1328
|
static VALUE
|
1328
|
-
|
1329
|
+
frb_iw_init(int argc, VALUE *argv, VALUE self)
|
1329
1330
|
{
|
1330
1331
|
VALUE roptions, rval;
|
1331
1332
|
bool create = false;
|
@@ -1344,7 +1345,7 @@ frt_iw_init(int argc, VALUE *argv, VALUE self)
|
|
1344
1345
|
store = DATA_PTR(rval);
|
1345
1346
|
} else if ((rval = rb_hash_aref(roptions, sym_path)) != Qnil) {
|
1346
1347
|
StringValue(rval);
|
1347
|
-
|
1348
|
+
frb_create_dir(rval);
|
1348
1349
|
store = open_fs_store(rs2s(rval));
|
1349
1350
|
DEREF(store);
|
1350
1351
|
}
|
@@ -1361,7 +1362,7 @@ frt_iw_init(int argc, VALUE *argv, VALUE self)
|
|
1361
1362
|
: true;
|
1362
1363
|
|
1363
1364
|
if ((rval = rb_hash_aref(roptions, sym_analyzer)) != Qnil) {
|
1364
|
-
analyzer =
|
1365
|
+
analyzer = frb_get_cwrapped_analyzer(rval);
|
1365
1366
|
}
|
1366
1367
|
|
1367
1368
|
create = RTEST(rb_hash_aref(roptions, sym_create));
|
@@ -1399,11 +1400,11 @@ frt_iw_init(int argc, VALUE *argv, VALUE self)
|
|
1399
1400
|
|
1400
1401
|
iw = iw_open(store, analyzer, &config);
|
1401
1402
|
|
1402
|
-
Frt_Wrap_Struct(self, &
|
1403
|
+
Frt_Wrap_Struct(self, &frb_iw_mark, &frb_iw_free, iw);
|
1403
1404
|
|
1404
1405
|
if (rb_block_given_p()) {
|
1405
1406
|
rb_yield(self);
|
1406
|
-
|
1407
|
+
frb_iw_close(self);
|
1407
1408
|
return Qnil;
|
1408
1409
|
} else {
|
1409
1410
|
return self;
|
@@ -1418,35 +1419,22 @@ frt_iw_init(int argc, VALUE *argv, VALUE self)
|
|
1418
1419
|
* taken into account until the IndexWriter has been committed.
|
1419
1420
|
*/
|
1420
1421
|
static VALUE
|
1421
|
-
|
1422
|
+
frb_iw_get_doc_count(VALUE self)
|
1422
1423
|
{
|
1423
1424
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1424
1425
|
return INT2FIX(iw_doc_count(iw));
|
1425
1426
|
}
|
1426
1427
|
|
1427
1428
|
static int
|
1428
|
-
|
1429
|
+
frb_hash_to_doc_i(VALUE key, VALUE value, VALUE arg)
|
1429
1430
|
{
|
1430
1431
|
if (key == Qundef) {
|
1431
1432
|
return ST_CONTINUE;
|
1432
1433
|
} else {
|
1433
1434
|
Document *doc = (Document *)arg;
|
1434
|
-
|
1435
|
+
Symbol field = frb_field(key);
|
1435
1436
|
VALUE val;
|
1436
1437
|
DocField *df;
|
1437
|
-
switch (TYPE(key)) {
|
1438
|
-
case T_STRING:
|
1439
|
-
field = rs2s(key);
|
1440
|
-
break;
|
1441
|
-
case T_SYMBOL:
|
1442
|
-
field = rb_id2name(SYM2ID(key));
|
1443
|
-
break;
|
1444
|
-
default:
|
1445
|
-
rb_raise(rb_eArgError,
|
1446
|
-
"%s cannot be a key to a field. Field keys must "
|
1447
|
-
" be symbols.", rs2s(rb_obj_as_string(key)));
|
1448
|
-
break;
|
1449
|
-
}
|
1450
1438
|
if (NULL == (df = doc_get_field(doc, field))) {
|
1451
1439
|
df = df_new(field);
|
1452
1440
|
}
|
@@ -1458,19 +1446,19 @@ frt_hash_to_doc_i(VALUE key, VALUE value, VALUE arg)
|
|
1458
1446
|
{
|
1459
1447
|
int i;
|
1460
1448
|
df->destroy_data = true;
|
1461
|
-
for (i = 0; i <
|
1462
|
-
val = rb_obj_as_string(
|
1463
|
-
df_add_data_len(df,
|
1449
|
+
for (i = 0; i < RARRAY_LEN(value); i++) {
|
1450
|
+
val = rb_obj_as_string(RARRAY_PTR(value)[i]);
|
1451
|
+
df_add_data_len(df, rstrdup(val), RSTRING_LEN(val));
|
1464
1452
|
}
|
1465
1453
|
}
|
1466
1454
|
break;
|
1467
1455
|
case T_STRING:
|
1468
|
-
df_add_data_len(df, rs2s(value),
|
1456
|
+
df_add_data_len(df, rs2s(value), RSTRING_LEN(value));
|
1469
1457
|
break;
|
1470
1458
|
default:
|
1471
1459
|
val = rb_obj_as_string(value);
|
1472
1460
|
df->destroy_data = true;
|
1473
|
-
df_add_data_len(df,
|
1461
|
+
df_add_data_len(df, rstrdup(val), RSTRING_LEN(val));
|
1474
1462
|
break;
|
1475
1463
|
}
|
1476
1464
|
doc_add_field(doc, df);
|
@@ -1479,7 +1467,7 @@ frt_hash_to_doc_i(VALUE key, VALUE value, VALUE arg)
|
|
1479
1467
|
}
|
1480
1468
|
|
1481
1469
|
static Document *
|
1482
|
-
|
1470
|
+
frb_get_doc(VALUE rdoc)
|
1483
1471
|
{
|
1484
1472
|
VALUE val;
|
1485
1473
|
Document *doc = doc_new();
|
@@ -1491,33 +1479,34 @@ frt_get_doc(VALUE rdoc)
|
|
1491
1479
|
|
1492
1480
|
switch (TYPE(rdoc)) {
|
1493
1481
|
case T_HASH:
|
1494
|
-
rb_hash_foreach(rdoc,
|
1482
|
+
rb_hash_foreach(rdoc, frb_hash_to_doc_i, (VALUE)doc);
|
1495
1483
|
break;
|
1496
1484
|
case T_ARRAY:
|
1497
1485
|
{
|
1498
1486
|
int i;
|
1499
|
-
df = df_new(
|
1487
|
+
df = df_new(fsym_content);
|
1500
1488
|
df->destroy_data = true;
|
1501
|
-
for (i = 0; i <
|
1502
|
-
val = rb_obj_as_string(
|
1503
|
-
df_add_data_len(df,
|
1489
|
+
for (i = 0; i < RARRAY_LEN(rdoc); i++) {
|
1490
|
+
val = rb_obj_as_string(RARRAY_PTR(rdoc)[i]);
|
1491
|
+
df_add_data_len(df, rstrdup(val), RSTRING_LEN(val));
|
1504
1492
|
}
|
1505
1493
|
doc_add_field(doc, df);
|
1506
1494
|
}
|
1507
1495
|
break;
|
1508
1496
|
case T_SYMBOL:
|
1509
|
-
|
1497
|
+
/* TODO: clean up this ugly cast */
|
1498
|
+
df = df_add_data(df_new(fsym_content), (char *)rb_id2name(SYM2ID(rdoc)));
|
1510
1499
|
doc_add_field(doc, df);
|
1511
1500
|
break;
|
1512
1501
|
case T_STRING:
|
1513
|
-
df = df_add_data_len(df_new(
|
1514
|
-
|
1502
|
+
df = df_add_data_len(df_new(fsym_content), rs2s(rdoc),
|
1503
|
+
RSTRING_LEN(rdoc));
|
1515
1504
|
doc_add_field(doc, df);
|
1516
1505
|
break;
|
1517
1506
|
default:
|
1518
1507
|
val = rb_obj_as_string(rdoc);
|
1519
|
-
df = df_add_data_len(df_new(
|
1520
|
-
|
1508
|
+
df = df_add_data_len(df_new(fsym_content), rstrdup(val),
|
1509
|
+
RSTRING_LEN(val));
|
1521
1510
|
df->destroy_data = true;
|
1522
1511
|
doc_add_field(doc, df);
|
1523
1512
|
break;
|
@@ -1534,10 +1523,10 @@ frt_get_doc(VALUE rdoc)
|
|
1534
1523
|
* hash object.
|
1535
1524
|
*/
|
1536
1525
|
static VALUE
|
1537
|
-
|
1526
|
+
frb_iw_add_doc(VALUE self, VALUE rdoc)
|
1538
1527
|
{
|
1539
1528
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1540
|
-
Document *doc =
|
1529
|
+
Document *doc = frb_get_doc(rdoc);
|
1541
1530
|
iw_add_doc(iw, doc);
|
1542
1531
|
doc_destroy(doc);
|
1543
1532
|
return self;
|
@@ -1556,7 +1545,7 @@ frt_iw_add_doc(VALUE self, VALUE rdoc)
|
|
1556
1545
|
* process).
|
1557
1546
|
*/
|
1558
1547
|
static VALUE
|
1559
|
-
|
1548
|
+
frb_iw_optimize(VALUE self)
|
1560
1549
|
{
|
1561
1550
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1562
1551
|
iw_optimize(iw);
|
@@ -1572,7 +1561,7 @@ frt_iw_optimize(VALUE self)
|
|
1572
1561
|
* with an IndexWriter.
|
1573
1562
|
*/
|
1574
1563
|
static VALUE
|
1575
|
-
|
1564
|
+
frb_iw_commit(VALUE self)
|
1576
1565
|
{
|
1577
1566
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1578
1567
|
iw_commit(iw);
|
@@ -1590,39 +1579,52 @@ frt_iw_commit(VALUE self)
|
|
1590
1579
|
* index.
|
1591
1580
|
*/
|
1592
1581
|
static VALUE
|
1593
|
-
|
1582
|
+
frb_iw_add_readers(VALUE self, VALUE rreaders)
|
1594
1583
|
{
|
1595
1584
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1596
1585
|
int i;
|
1597
1586
|
IndexReader **irs;
|
1598
1587
|
Check_Type(rreaders, T_ARRAY);
|
1599
1588
|
|
1600
|
-
irs = ALLOC_N(IndexReader *,
|
1601
|
-
i =
|
1589
|
+
irs = ALLOC_N(IndexReader *, RARRAY_LEN(rreaders));
|
1590
|
+
i = RARRAY_LEN(rreaders);
|
1602
1591
|
while (i-- > 0) {
|
1603
1592
|
IndexReader *ir;
|
1604
|
-
Data_Get_Struct(
|
1593
|
+
Data_Get_Struct(RARRAY_PTR(rreaders)[i], IndexReader, ir);
|
1605
1594
|
irs[i] = ir;
|
1606
1595
|
}
|
1607
|
-
iw_add_readers(iw, irs,
|
1596
|
+
iw_add_readers(iw, irs, RARRAY_LEN(rreaders));
|
1608
1597
|
free(irs);
|
1609
1598
|
return self;
|
1610
1599
|
}
|
1611
1600
|
|
1612
1601
|
/*
|
1613
1602
|
* call-seq:
|
1614
|
-
* iw.delete(field, term)
|
1603
|
+
* iw.delete(field, term) -> iw
|
1604
|
+
* iw.delete(field, terms) -> iw
|
1615
1605
|
*
|
1616
|
-
* Delete all documents in the index with the
|
1617
|
-
* +field+. You should usually have a unique document id which you use
|
1618
|
-
* this method, rather then deleting all documents with the word "the"
|
1619
|
-
* them.
|
1606
|
+
* Delete all documents in the index with the given +term+ or +terms+ in the
|
1607
|
+
* field +field+. You should usually have a unique document id which you use
|
1608
|
+
* with this method, rather then deleting all documents with the word "the"
|
1609
|
+
* in them. There are of course exceptions to this rule. For example, you may
|
1610
|
+
* want to delete all documents with the term "viagra" when deleting spam.
|
1620
1611
|
*/
|
1621
1612
|
static VALUE
|
1622
|
-
|
1613
|
+
frb_iw_delete(VALUE self, VALUE rfield, VALUE rterm)
|
1623
1614
|
{
|
1624
1615
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1625
|
-
|
1616
|
+
if (TYPE(rterm) == T_ARRAY) {
|
1617
|
+
const int term_cnt = RARRAY_LEN(rterm);
|
1618
|
+
int i;
|
1619
|
+
char **terms = ALLOC_N(char *, term_cnt);
|
1620
|
+
for (i = 0; i < term_cnt; i++) {
|
1621
|
+
terms[i] = StringValuePtr(RARRAY_PTR(rterm)[i]);
|
1622
|
+
}
|
1623
|
+
iw_delete_terms(iw, frb_field(rfield), terms, term_cnt);
|
1624
|
+
free(terms);
|
1625
|
+
} else {
|
1626
|
+
iw_delete_term(iw, frb_field(rfield), StringValuePtr(rterm));
|
1627
|
+
}
|
1626
1628
|
return self;
|
1627
1629
|
}
|
1628
1630
|
|
@@ -1634,10 +1636,10 @@ frt_iw_delete(VALUE self, VALUE rfield, VALUE rterm)
|
|
1634
1636
|
* to dynamically add new fields to the index with specific properties.
|
1635
1637
|
*/
|
1636
1638
|
static VALUE
|
1637
|
-
|
1639
|
+
frb_iw_field_infos(VALUE self)
|
1638
1640
|
{
|
1639
1641
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1640
|
-
return
|
1642
|
+
return frb_get_field_infos(iw->fis);
|
1641
1643
|
}
|
1642
1644
|
|
1643
1645
|
/*
|
@@ -1648,10 +1650,10 @@ frt_iw_field_infos(VALUE self)
|
|
1648
1650
|
* to use the same analyzer in a QueryParser.
|
1649
1651
|
*/
|
1650
1652
|
static VALUE
|
1651
|
-
|
1653
|
+
frb_iw_get_analyzer(VALUE self)
|
1652
1654
|
{
|
1653
1655
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1654
|
-
return
|
1656
|
+
return frb_get_analyzer(iw->analyzer);
|
1655
1657
|
}
|
1656
1658
|
|
1657
1659
|
/*
|
@@ -1663,12 +1665,12 @@ frt_iw_get_analyzer(VALUE self)
|
|
1663
1665
|
* same analyzer will be used for all documents during search.
|
1664
1666
|
*/
|
1665
1667
|
static VALUE
|
1666
|
-
|
1668
|
+
frb_iw_set_analyzer(VALUE self, VALUE ranalyzer)
|
1667
1669
|
{
|
1668
1670
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1669
1671
|
|
1670
1672
|
a_deref(iw->analyzer);
|
1671
|
-
iw->analyzer =
|
1673
|
+
iw->analyzer = frb_get_cwrapped_analyzer(ranalyzer);
|
1672
1674
|
return ranalyzer;
|
1673
1675
|
}
|
1674
1676
|
|
@@ -1679,7 +1681,7 @@ frt_iw_set_analyzer(VALUE self, VALUE ranalyzer)
|
|
1679
1681
|
* Returns the current version of the index writer.
|
1680
1682
|
*/
|
1681
1683
|
static VALUE
|
1682
|
-
|
1684
|
+
frb_iw_version(VALUE self)
|
1683
1685
|
{
|
1684
1686
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1685
1687
|
return ULL2NUM(iw->sis->version);
|
@@ -1692,7 +1694,7 @@ frt_iw_version(VALUE self)
|
|
1692
1694
|
* Return the current value of chunk_size
|
1693
1695
|
*/
|
1694
1696
|
static VALUE
|
1695
|
-
|
1697
|
+
frb_iw_get_chunk_size(VALUE self)
|
1696
1698
|
{
|
1697
1699
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1698
1700
|
return INT2FIX(iw->config.chunk_size);
|
@@ -1705,7 +1707,7 @@ frt_iw_get_chunk_size(VALUE self)
|
|
1705
1707
|
* Set the chunk_size parameter
|
1706
1708
|
*/
|
1707
1709
|
static VALUE
|
1708
|
-
|
1710
|
+
frb_iw_set_chunk_size(VALUE self, VALUE rval)
|
1709
1711
|
{
|
1710
1712
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1711
1713
|
iw->config.chunk_size = FIX2INT(rval);
|
@@ -1719,7 +1721,7 @@ frt_iw_set_chunk_size(VALUE self, VALUE rval)
|
|
1719
1721
|
* Return the current value of max_buffer_memory
|
1720
1722
|
*/
|
1721
1723
|
static VALUE
|
1722
|
-
|
1724
|
+
frb_iw_get_max_buffer_memory(VALUE self)
|
1723
1725
|
{
|
1724
1726
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1725
1727
|
return INT2FIX(iw->config.max_buffer_memory);
|
@@ -1732,7 +1734,7 @@ frt_iw_get_max_buffer_memory(VALUE self)
|
|
1732
1734
|
* Set the max_buffer_memory parameter
|
1733
1735
|
*/
|
1734
1736
|
static VALUE
|
1735
|
-
|
1737
|
+
frb_iw_set_max_buffer_memory(VALUE self, VALUE rval)
|
1736
1738
|
{
|
1737
1739
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1738
1740
|
iw->config.max_buffer_memory = FIX2INT(rval);
|
@@ -1746,7 +1748,7 @@ frt_iw_set_max_buffer_memory(VALUE self, VALUE rval)
|
|
1746
1748
|
* Return the current value of term_index_interval
|
1747
1749
|
*/
|
1748
1750
|
static VALUE
|
1749
|
-
|
1751
|
+
frb_iw_get_index_interval(VALUE self)
|
1750
1752
|
{
|
1751
1753
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1752
1754
|
return INT2FIX(iw->config.index_interval);
|
@@ -1759,7 +1761,7 @@ frt_iw_get_index_interval(VALUE self)
|
|
1759
1761
|
* Set the term_index_interval parameter
|
1760
1762
|
*/
|
1761
1763
|
static VALUE
|
1762
|
-
|
1764
|
+
frb_iw_set_index_interval(VALUE self, VALUE rval)
|
1763
1765
|
{
|
1764
1766
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1765
1767
|
iw->config.index_interval = FIX2INT(rval);
|
@@ -1773,7 +1775,7 @@ frt_iw_set_index_interval(VALUE self, VALUE rval)
|
|
1773
1775
|
* Return the current value of doc_skip_interval
|
1774
1776
|
*/
|
1775
1777
|
static VALUE
|
1776
|
-
|
1778
|
+
frb_iw_get_skip_interval(VALUE self)
|
1777
1779
|
{
|
1778
1780
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1779
1781
|
return INT2FIX(iw->config.skip_interval);
|
@@ -1786,7 +1788,7 @@ frt_iw_get_skip_interval(VALUE self)
|
|
1786
1788
|
* Set the doc_skip_interval parameter
|
1787
1789
|
*/
|
1788
1790
|
static VALUE
|
1789
|
-
|
1791
|
+
frb_iw_set_skip_interval(VALUE self, VALUE rval)
|
1790
1792
|
{
|
1791
1793
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1792
1794
|
iw->config.skip_interval = FIX2INT(rval);
|
@@ -1800,7 +1802,7 @@ frt_iw_set_skip_interval(VALUE self, VALUE rval)
|
|
1800
1802
|
* Return the current value of merge_factor
|
1801
1803
|
*/
|
1802
1804
|
static VALUE
|
1803
|
-
|
1805
|
+
frb_iw_get_merge_factor(VALUE self)
|
1804
1806
|
{
|
1805
1807
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1806
1808
|
return INT2FIX(iw->config.merge_factor);
|
@@ -1813,7 +1815,7 @@ frt_iw_get_merge_factor(VALUE self)
|
|
1813
1815
|
* Set the merge_factor parameter
|
1814
1816
|
*/
|
1815
1817
|
static VALUE
|
1816
|
-
|
1818
|
+
frb_iw_set_merge_factor(VALUE self, VALUE rval)
|
1817
1819
|
{
|
1818
1820
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1819
1821
|
iw->config.merge_factor = FIX2INT(rval);
|
@@ -1827,7 +1829,7 @@ frt_iw_set_merge_factor(VALUE self, VALUE rval)
|
|
1827
1829
|
* Return the current value of max_buffered_docs
|
1828
1830
|
*/
|
1829
1831
|
static VALUE
|
1830
|
-
|
1832
|
+
frb_iw_get_max_buffered_docs(VALUE self)
|
1831
1833
|
{
|
1832
1834
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1833
1835
|
return INT2FIX(iw->config.max_buffered_docs);
|
@@ -1840,7 +1842,7 @@ frt_iw_get_max_buffered_docs(VALUE self)
|
|
1840
1842
|
* Set the max_buffered_docs parameter
|
1841
1843
|
*/
|
1842
1844
|
static VALUE
|
1843
|
-
|
1845
|
+
frb_iw_set_max_buffered_docs(VALUE self, VALUE rval)
|
1844
1846
|
{
|
1845
1847
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1846
1848
|
iw->config.max_buffered_docs = FIX2INT(rval);
|
@@ -1854,7 +1856,7 @@ frt_iw_set_max_buffered_docs(VALUE self, VALUE rval)
|
|
1854
1856
|
* Return the current value of max_merge_docs
|
1855
1857
|
*/
|
1856
1858
|
static VALUE
|
1857
|
-
|
1859
|
+
frb_iw_get_max_merge_docs(VALUE self)
|
1858
1860
|
{
|
1859
1861
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1860
1862
|
return INT2FIX(iw->config.max_merge_docs);
|
@@ -1867,7 +1869,7 @@ frt_iw_get_max_merge_docs(VALUE self)
|
|
1867
1869
|
* Set the max_merge_docs parameter
|
1868
1870
|
*/
|
1869
1871
|
static VALUE
|
1870
|
-
|
1872
|
+
frb_iw_set_max_merge_docs(VALUE self, VALUE rval)
|
1871
1873
|
{
|
1872
1874
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1873
1875
|
iw->config.max_merge_docs = FIX2INT(rval);
|
@@ -1881,7 +1883,7 @@ frt_iw_set_max_merge_docs(VALUE self, VALUE rval)
|
|
1881
1883
|
* Return the current value of max_field_length
|
1882
1884
|
*/
|
1883
1885
|
static VALUE
|
1884
|
-
|
1886
|
+
frb_iw_get_max_field_length(VALUE self)
|
1885
1887
|
{
|
1886
1888
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1887
1889
|
return INT2FIX(iw->config.max_field_length);
|
@@ -1894,7 +1896,7 @@ frt_iw_get_max_field_length(VALUE self)
|
|
1894
1896
|
* Set the max_field_length parameter
|
1895
1897
|
*/
|
1896
1898
|
static VALUE
|
1897
|
-
|
1899
|
+
frb_iw_set_max_field_length(VALUE self, VALUE rval)
|
1898
1900
|
{
|
1899
1901
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1900
1902
|
iw->config.max_field_length = FIX2INT(rval);
|
@@ -1908,7 +1910,7 @@ frt_iw_set_max_field_length(VALUE self, VALUE rval)
|
|
1908
1910
|
* Return the current value of use_compound_file
|
1909
1911
|
*/
|
1910
1912
|
static VALUE
|
1911
|
-
|
1913
|
+
frb_iw_get_use_compound_file(VALUE self)
|
1912
1914
|
{
|
1913
1915
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1914
1916
|
return iw->config.use_compound_file ? Qtrue : Qfalse;
|
@@ -1921,7 +1923,7 @@ frt_iw_get_use_compound_file(VALUE self)
|
|
1921
1923
|
* Set the use_compound_file parameter
|
1922
1924
|
*/
|
1923
1925
|
static VALUE
|
1924
|
-
|
1926
|
+
frb_iw_set_use_compound_file(VALUE self, VALUE rval)
|
1925
1927
|
{
|
1926
1928
|
IndexWriter *iw = (IndexWriter *)DATA_PTR(self);
|
1927
1929
|
iw->config.use_compound_file = RTEST(rval);
|
@@ -1935,13 +1937,13 @@ frt_iw_set_use_compound_file(VALUE self, VALUE rval)
|
|
1935
1937
|
****************************************************************************/
|
1936
1938
|
|
1937
1939
|
static void
|
1938
|
-
|
1940
|
+
frb_lzd_date_free(void *p)
|
1939
1941
|
{
|
1940
1942
|
lazy_doc_close((LazyDoc *)p);
|
1941
1943
|
}
|
1942
1944
|
|
1943
1945
|
static VALUE
|
1944
|
-
|
1946
|
+
frb_lazy_df_load(VALUE self, VALUE rkey, LazyDocField *lazy_df)
|
1945
1947
|
{
|
1946
1948
|
VALUE rdata = Qnil;
|
1947
1949
|
if (lazy_df) {
|
@@ -1953,9 +1955,7 @@ frt_lazy_df_load(VALUE self, VALUE rkey, LazyDocField *lazy_df)
|
|
1953
1955
|
rdata = rb_ary_new2(lazy_df->size);
|
1954
1956
|
for (i = 0; i < lazy_df->size; i++) {
|
1955
1957
|
char *data = lazy_df_get_data(lazy_df, i);
|
1956
|
-
|
1957
|
-
rb_str_new(data, lazy_df->data[i].length);
|
1958
|
-
RARRAY(rdata)->len++;
|
1958
|
+
rb_ary_store(rdata, i, rb_str_new(data, lazy_df->data[i].length));
|
1959
1959
|
}
|
1960
1960
|
}
|
1961
1961
|
rb_hash_aset(self, rkey, rdata);
|
@@ -1971,25 +1971,13 @@ frt_lazy_df_load(VALUE self, VALUE rkey, LazyDocField *lazy_df)
|
|
1971
1971
|
* really need to call it yourself.
|
1972
1972
|
*/
|
1973
1973
|
static VALUE
|
1974
|
-
|
1974
|
+
frb_lzd_default(VALUE self, VALUE rkey)
|
1975
1975
|
{
|
1976
1976
|
LazyDoc *lazy_doc = (LazyDoc *)DATA_PTR(rb_ivar_get(self, id_data));
|
1977
|
-
|
1978
|
-
|
1979
|
-
|
1980
|
-
|
1981
|
-
rkey = ID2SYM(rb_intern(field));
|
1982
|
-
break;
|
1983
|
-
case T_SYMBOL:
|
1984
|
-
field = frt_field(rkey);
|
1985
|
-
break;
|
1986
|
-
default:
|
1987
|
-
rb_raise(rb_eArgError,
|
1988
|
-
"%s cannot be a key to a field. Field keys must "
|
1989
|
-
" be symbols.", rs2s(rb_obj_as_string(rkey)));
|
1990
|
-
break;
|
1991
|
-
}
|
1992
|
-
return frt_lazy_df_load(self, rkey, h_get(lazy_doc->field_dict, field));
|
1977
|
+
Symbol field = frb_field(rkey);
|
1978
|
+
VALUE rfield = FSYM2SYM(field);
|
1979
|
+
|
1980
|
+
return frb_lazy_df_load(self, rfield, lazy_doc_get(lazy_doc, field));
|
1993
1981
|
}
|
1994
1982
|
|
1995
1983
|
/*
|
@@ -2001,7 +1989,7 @@ frt_lzd_default(VALUE self, VALUE rkey)
|
|
2001
1989
|
* Try to access any other field an nil will be returned.
|
2002
1990
|
*/
|
2003
1991
|
static VALUE
|
2004
|
-
|
1992
|
+
frb_lzd_fields(VALUE self)
|
2005
1993
|
{
|
2006
1994
|
return rb_ivar_get(self, id_fields);
|
2007
1995
|
}
|
@@ -2013,19 +2001,19 @@ frt_lzd_fields(VALUE self)
|
|
2013
2001
|
* Load all unloaded fields in the document from the index.
|
2014
2002
|
*/
|
2015
2003
|
static VALUE
|
2016
|
-
|
2004
|
+
frb_lzd_load(VALUE self)
|
2017
2005
|
{
|
2018
2006
|
LazyDoc *lazy_doc = (LazyDoc *)DATA_PTR(rb_ivar_get(self, id_data));
|
2019
2007
|
int i;
|
2020
2008
|
for (i = 0; i < lazy_doc->size; i++) {
|
2021
2009
|
LazyDocField *lazy_df = lazy_doc->fields[i];
|
2022
|
-
|
2010
|
+
frb_lazy_df_load(self, FSYM2SYM(lazy_df->name), lazy_df);
|
2023
2011
|
}
|
2024
2012
|
return self;
|
2025
2013
|
}
|
2026
2014
|
|
2027
2015
|
VALUE
|
2028
|
-
|
2016
|
+
frb_get_lazy_doc(LazyDoc *lazy_doc)
|
2029
2017
|
{
|
2030
2018
|
int i;
|
2031
2019
|
VALUE rfields = rb_ary_new2(lazy_doc->size);
|
@@ -2034,12 +2022,11 @@ frt_get_lazy_doc(LazyDoc *lazy_doc)
|
|
2034
2022
|
self = rb_hash_new();
|
2035
2023
|
OBJSETUP(self, cLazyDoc, T_HASH);
|
2036
2024
|
|
2037
|
-
rdata = Data_Wrap_Struct(cLazyDocData, NULL, &
|
2025
|
+
rdata = Data_Wrap_Struct(cLazyDocData, NULL, &frb_lzd_date_free, lazy_doc);
|
2038
2026
|
rb_ivar_set(self, id_data, rdata);
|
2039
2027
|
|
2040
2028
|
for (i = 0; i < lazy_doc->size; i++) {
|
2041
|
-
|
2042
|
-
RARRAY(rfields)->len++;
|
2029
|
+
rb_ary_store(rfields, i, FSYM2SYM(lazy_doc->fields[i]->name));
|
2043
2030
|
}
|
2044
2031
|
rb_ivar_set(self, id_fields, rfields);
|
2045
2032
|
|
@@ -2053,28 +2040,28 @@ frt_get_lazy_doc(LazyDoc *lazy_doc)
|
|
2053
2040
|
****************************************************************************/
|
2054
2041
|
|
2055
2042
|
void
|
2056
|
-
|
2043
|
+
frb_ir_free(void *p)
|
2057
2044
|
{
|
2058
2045
|
object_del(p);
|
2059
2046
|
ir_close((IndexReader *)p);
|
2060
2047
|
}
|
2061
2048
|
|
2062
2049
|
void
|
2063
|
-
|
2050
|
+
frb_ir_mark(void *p)
|
2064
2051
|
{
|
2065
2052
|
IndexReader *ir = (IndexReader *)p;
|
2066
|
-
|
2053
|
+
frb_gc_mark(ir->store);
|
2067
2054
|
}
|
2068
2055
|
|
2069
|
-
static VALUE
|
2056
|
+
static VALUE frb_ir_close(VALUE self);
|
2070
2057
|
|
2071
2058
|
void
|
2072
|
-
|
2059
|
+
frb_mr_mark(void *p)
|
2073
2060
|
{
|
2074
2061
|
MultiReader *mr = (MultiReader *)p;
|
2075
2062
|
int i;
|
2076
2063
|
for (i = 0; i < mr->r_cnt; i++) {
|
2077
|
-
|
2064
|
+
frb_gc_mark(mr->sub_readers[i]);
|
2078
2065
|
}
|
2079
2066
|
}
|
2080
2067
|
|
@@ -2105,7 +2092,7 @@ frt_mr_mark(void *p)
|
|
2105
2092
|
* iw = IndexReader.new(["/path/to/index1", "/path/to/index2"])
|
2106
2093
|
*/
|
2107
2094
|
static VALUE
|
2108
|
-
|
2095
|
+
frb_ir_init(VALUE self, VALUE rdir)
|
2109
2096
|
{
|
2110
2097
|
Store *store = NULL;
|
2111
2098
|
IndexReader *ir;
|
@@ -2115,11 +2102,11 @@ frt_ir_init(VALUE self, VALUE rdir)
|
|
2115
2102
|
|
2116
2103
|
if (TYPE(rdir) == T_ARRAY) {
|
2117
2104
|
VALUE rdirs = rdir;
|
2118
|
-
const int reader_cnt =
|
2105
|
+
const int reader_cnt = RARRAY_LEN(rdir);
|
2119
2106
|
IndexReader **sub_readers = ALLOC_N(IndexReader *, reader_cnt);
|
2120
2107
|
int i;
|
2121
2108
|
for (i = 0; i < reader_cnt; i++) {
|
2122
|
-
rdir =
|
2109
|
+
rdir = RARRAY_PTR(rdirs)[i];
|
2123
2110
|
switch (TYPE(rdir)) {
|
2124
2111
|
case T_DATA:
|
2125
2112
|
if (CLASS_OF(rdir) == cIndexReader) {
|
@@ -2137,7 +2124,7 @@ frt_ir_init(VALUE self, VALUE rdir)
|
|
2137
2124
|
}
|
2138
2125
|
break;
|
2139
2126
|
case T_STRING:
|
2140
|
-
|
2127
|
+
frb_create_dir(rdir);
|
2141
2128
|
store = open_fs_store(rs2s(rdir));
|
2142
2129
|
DEREF(store);
|
2143
2130
|
break;
|
@@ -2151,14 +2138,14 @@ frt_ir_init(VALUE self, VALUE rdir)
|
|
2151
2138
|
sub_readers[i] = ir_open(store);
|
2152
2139
|
}
|
2153
2140
|
ir = mr_open(sub_readers, reader_cnt);
|
2154
|
-
Frt_Wrap_Struct(self, &
|
2141
|
+
Frt_Wrap_Struct(self, &frb_mr_mark, &frb_ir_free, ir);
|
2155
2142
|
} else {
|
2156
2143
|
switch (TYPE(rdir)) {
|
2157
2144
|
case T_DATA:
|
2158
2145
|
store = DATA_PTR(rdir);
|
2159
2146
|
break;
|
2160
2147
|
case T_STRING:
|
2161
|
-
|
2148
|
+
frb_create_dir(rdir);
|
2162
2149
|
store = open_fs_store(rs2s(rdir));
|
2163
2150
|
DEREF(store);
|
2164
2151
|
break;
|
@@ -2169,7 +2156,7 @@ frt_ir_init(VALUE self, VALUE rdir)
|
|
2169
2156
|
break;
|
2170
2157
|
}
|
2171
2158
|
ir = ir_open(store);
|
2172
|
-
Frt_Wrap_Struct(self, &
|
2159
|
+
Frt_Wrap_Struct(self, &frb_ir_mark, &frb_ir_free, ir);
|
2173
2160
|
}
|
2174
2161
|
object_add(ir, self);
|
2175
2162
|
|
@@ -2177,7 +2164,7 @@ frt_ir_init(VALUE self, VALUE rdir)
|
|
2177
2164
|
for (i = 0; i < fis->size; i++) {
|
2178
2165
|
FieldInfo *fi = fis->fields[i];
|
2179
2166
|
rb_hash_aset(rfield_num_map,
|
2180
|
-
|
2167
|
+
FSYM2SYM(fi->name),
|
2181
2168
|
INT2FIX(fi->number));
|
2182
2169
|
}
|
2183
2170
|
rb_ivar_set(self, id_fld_num_map, rfield_num_map);
|
@@ -2194,10 +2181,10 @@ frt_ir_init(VALUE self, VALUE rdir)
|
|
2194
2181
|
* encoded float value.
|
2195
2182
|
*/
|
2196
2183
|
static VALUE
|
2197
|
-
|
2184
|
+
frb_ir_set_norm(VALUE self, VALUE rdoc_id, VALUE rfield, VALUE rval)
|
2198
2185
|
{
|
2199
2186
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2200
|
-
ir_set_norm(ir, FIX2INT(rdoc_id),
|
2187
|
+
ir_set_norm(ir, FIX2INT(rdoc_id), frb_field(rfield), (uchar)NUM2CHR(rval));
|
2201
2188
|
return self;
|
2202
2189
|
}
|
2203
2190
|
|
@@ -2210,11 +2197,11 @@ frt_ir_set_norm(VALUE self, VALUE rdoc_id, VALUE rfield, VALUE rval)
|
|
2210
2197
|
* could have null bytes.
|
2211
2198
|
*/
|
2212
2199
|
static VALUE
|
2213
|
-
|
2200
|
+
frb_ir_norms(VALUE self, VALUE rfield)
|
2214
2201
|
{
|
2215
2202
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2216
2203
|
uchar *norms;
|
2217
|
-
norms = ir_get_norms(ir,
|
2204
|
+
norms = ir_get_norms(ir, frb_field(rfield));
|
2218
2205
|
if (norms) {
|
2219
2206
|
return rb_str_new((char *)norms, ir->max_doc(ir));
|
2220
2207
|
} else {
|
@@ -2229,20 +2216,20 @@ frt_ir_norms(VALUE self, VALUE rfield)
|
|
2229
2216
|
* Expert: Get the norm values into a string +buffer+ starting at +offset+.
|
2230
2217
|
*/
|
2231
2218
|
static VALUE
|
2232
|
-
|
2219
|
+
frb_ir_get_norms_into(VALUE self, VALUE rfield, VALUE rnorms, VALUE roffset)
|
2233
2220
|
{
|
2234
2221
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2235
2222
|
int offset;
|
2236
2223
|
offset = FIX2INT(roffset);
|
2237
2224
|
Check_Type(rnorms, T_STRING);
|
2238
|
-
if (
|
2239
|
-
rb_raise(rb_eArgError, "supplied a string of length:%
|
2225
|
+
if (RSTRING_LEN(rnorms) < offset + ir->max_doc(ir)) {
|
2226
|
+
rb_raise(rb_eArgError, "supplied a string of length:%ld to "
|
2240
2227
|
"IndexReader#get_norms_into but needed a string of length "
|
2241
2228
|
"offset:%d + maxdoc:%d",
|
2242
|
-
|
2229
|
+
RSTRING_LEN(rnorms), offset, ir->max_doc(ir));
|
2243
2230
|
}
|
2244
2231
|
|
2245
|
-
ir_get_norms_into(ir,
|
2232
|
+
ir_get_norms_into(ir, frb_field(rfield),
|
2246
2233
|
(uchar *)rs2s(rnorms) + offset);
|
2247
2234
|
return rnorms;
|
2248
2235
|
}
|
@@ -2255,7 +2242,7 @@ frt_ir_get_norms_into(VALUE self, VALUE rfield, VALUE rnorms, VALUE roffset)
|
|
2255
2242
|
* will use open a Commit lock.
|
2256
2243
|
*/
|
2257
2244
|
static VALUE
|
2258
|
-
|
2245
|
+
frb_ir_commit(VALUE self)
|
2259
2246
|
{
|
2260
2247
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2261
2248
|
ir_commit(ir);
|
@@ -2273,7 +2260,7 @@ frt_ir_commit(VALUE self)
|
|
2273
2260
|
* errors.
|
2274
2261
|
*/
|
2275
2262
|
static VALUE
|
2276
|
-
|
2263
|
+
frb_ir_close(VALUE self)
|
2277
2264
|
{
|
2278
2265
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2279
2266
|
object_del(ir);
|
@@ -2290,7 +2277,7 @@ frt_ir_close(VALUE self)
|
|
2290
2277
|
* IndexReader or committed by any other IndexReader.
|
2291
2278
|
*/
|
2292
2279
|
static VALUE
|
2293
|
-
|
2280
|
+
frb_ir_has_deletions(VALUE self)
|
2294
2281
|
{
|
2295
2282
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2296
2283
|
return ir->has_deletions(ir) ? Qtrue : Qfalse;
|
@@ -2305,7 +2292,7 @@ frt_ir_has_deletions(VALUE self)
|
|
2305
2292
|
* returned by search methods.
|
2306
2293
|
*/
|
2307
2294
|
static VALUE
|
2308
|
-
|
2295
|
+
frb_ir_delete(VALUE self, VALUE rdoc_id)
|
2309
2296
|
{
|
2310
2297
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2311
2298
|
ir_delete_doc(ir, FIX2INT(rdoc_id));
|
@@ -2319,7 +2306,7 @@ frt_ir_delete(VALUE self, VALUE rdoc_id)
|
|
2319
2306
|
* Returns true if the document at +doc_id+ has been deleted.
|
2320
2307
|
*/
|
2321
2308
|
static VALUE
|
2322
|
-
|
2309
|
+
frb_ir_is_deleted(VALUE self, VALUE rdoc_id)
|
2323
2310
|
{
|
2324
2311
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2325
2312
|
return ir->is_deleted(ir, FIX2INT(rdoc_id)) ? Qtrue : Qfalse;
|
@@ -2335,7 +2322,7 @@ frt_ir_is_deleted(VALUE self, VALUE rdoc_id)
|
|
2335
2322
|
* in the index.
|
2336
2323
|
*/
|
2337
2324
|
static VALUE
|
2338
|
-
|
2325
|
+
frb_ir_max_doc(VALUE self)
|
2339
2326
|
{
|
2340
2327
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2341
2328
|
return INT2FIX(ir->max_doc(ir));
|
@@ -2350,7 +2337,7 @@ frt_ir_max_doc(VALUE self)
|
|
2350
2337
|
* deleted from the index.
|
2351
2338
|
*/
|
2352
2339
|
static VALUE
|
2353
|
-
|
2340
|
+
frb_ir_num_docs(VALUE self)
|
2354
2341
|
{
|
2355
2342
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2356
2343
|
return INT2FIX(ir->num_docs(ir));
|
@@ -2366,7 +2353,7 @@ frt_ir_num_docs(VALUE self)
|
|
2366
2353
|
* effect on these documents.
|
2367
2354
|
*/
|
2368
2355
|
static VALUE
|
2369
|
-
|
2356
|
+
frb_ir_undelete_all(VALUE self)
|
2370
2357
|
{
|
2371
2358
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2372
2359
|
ir_undelete_all(ir);
|
@@ -2374,7 +2361,7 @@ frt_ir_undelete_all(VALUE self)
|
|
2374
2361
|
}
|
2375
2362
|
|
2376
2363
|
static VALUE
|
2377
|
-
|
2364
|
+
frb_get_doc_range(IndexReader *ir, int pos, int len, int max)
|
2378
2365
|
{
|
2379
2366
|
VALUE ary;
|
2380
2367
|
int i;
|
@@ -2382,8 +2369,7 @@ frt_get_doc_range(IndexReader *ir, int pos, int len, int max)
|
|
2382
2369
|
len = max - pos;
|
2383
2370
|
ary = rb_ary_new2(len);
|
2384
2371
|
for (i = 0; i < len; i++) {
|
2385
|
-
|
2386
|
-
RARRAY(ary)->len++;
|
2372
|
+
rb_ary_store(ary, i, frb_get_lazy_doc(ir->get_lazy_doc(ir, i + pos)));
|
2387
2373
|
}
|
2388
2374
|
return ary;
|
2389
2375
|
}
|
@@ -2398,7 +2384,7 @@ frt_get_doc_range(IndexReader *ir, int pos, int len, int max)
|
|
2398
2384
|
* which are returned by the Searchers search methods.
|
2399
2385
|
*/
|
2400
2386
|
static VALUE
|
2401
|
-
|
2387
|
+
frb_ir_get_doc(int argc, VALUE *argv, VALUE self)
|
2402
2388
|
{
|
2403
2389
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2404
2390
|
VALUE arg1, arg2;
|
@@ -2410,29 +2396,29 @@ frt_ir_get_doc(int argc, VALUE *argv, VALUE self)
|
|
2410
2396
|
pos = FIX2INT(arg1);
|
2411
2397
|
pos = (pos < 0) ? (max + pos) : pos;
|
2412
2398
|
if (pos < 0 || pos >= max) {
|
2413
|
-
rb_raise(rb_eArgError, "
|
2414
|
-
"IndexReader#[]", pos, 0, max
|
2415
|
-
rb_id2name(SYM2ID(argv)));
|
2399
|
+
rb_raise(rb_eArgError, "index %ld is out of range [%d..%ld] for "
|
2400
|
+
"IndexReader#[]", pos, 0, max);
|
2416
2401
|
}
|
2417
|
-
return
|
2402
|
+
return frb_get_lazy_doc(ir->get_lazy_doc(ir, pos));
|
2418
2403
|
}
|
2419
2404
|
|
2420
2405
|
/* check if idx is Range */
|
2406
|
+
/* FIXME: test this with dodgy values */
|
2421
2407
|
switch (rb_range_beg_len(arg1, &pos, &len, max, 0)) {
|
2422
2408
|
case Qfalse:
|
2423
2409
|
rb_raise(rb_eArgError, ":%s isn't a valid argument for "
|
2424
2410
|
"IndexReader.get_document(index)",
|
2425
|
-
rb_id2name(SYM2ID(
|
2411
|
+
rb_id2name(SYM2ID(arg1)));
|
2426
2412
|
case Qnil:
|
2427
2413
|
return Qnil;
|
2428
2414
|
default:
|
2429
|
-
return
|
2415
|
+
return frb_get_doc_range(ir, pos, len, max);
|
2430
2416
|
}
|
2431
2417
|
}
|
2432
2418
|
else {
|
2433
2419
|
pos = FIX2LONG(arg1);
|
2434
2420
|
len = FIX2LONG(arg2);
|
2435
|
-
return
|
2421
|
+
return frb_get_doc_range(ir, pos, len, max);
|
2436
2422
|
}
|
2437
2423
|
}
|
2438
2424
|
|
@@ -2445,7 +2431,7 @@ frt_ir_get_doc(int argc, VALUE *argv, VALUE self)
|
|
2445
2431
|
* index to search the latest documents added to the index.
|
2446
2432
|
*/
|
2447
2433
|
static VALUE
|
2448
|
-
|
2434
|
+
frb_ir_is_latest(VALUE self)
|
2449
2435
|
{
|
2450
2436
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2451
2437
|
return ir_is_latest(ir) ? Qtrue : Qfalse;
|
@@ -2456,17 +2442,17 @@ frt_ir_is_latest(VALUE self)
|
|
2456
2442
|
* index_reader.term_vector(doc_id, field) -> TermVector
|
2457
2443
|
*
|
2458
2444
|
* Return the TermVector for the field +field+ in the document at +doc_id+ in
|
2459
|
-
* the index. Return nil
|
2445
|
+
* the index. Return nil if no such term_vector exists. See TermVector.
|
2460
2446
|
*/
|
2461
2447
|
static VALUE
|
2462
|
-
|
2448
|
+
frb_ir_term_vector(VALUE self, VALUE rdoc_id, VALUE rfield)
|
2463
2449
|
{
|
2464
2450
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2465
2451
|
TermVector *tv;
|
2466
2452
|
VALUE rtv;
|
2467
|
-
tv = ir->term_vector(ir, FIX2INT(rdoc_id),
|
2453
|
+
tv = ir->term_vector(ir, FIX2INT(rdoc_id), frb_field(rfield));
|
2468
2454
|
if (tv) {
|
2469
|
-
rtv =
|
2455
|
+
rtv = frb_get_tv(tv);
|
2470
2456
|
tv_destroy(tv);
|
2471
2457
|
return rtv;
|
2472
2458
|
}
|
@@ -2476,9 +2462,9 @@ frt_ir_term_vector(VALUE self, VALUE rdoc_id, VALUE rfield)
|
|
2476
2462
|
}
|
2477
2463
|
|
2478
2464
|
static void
|
2479
|
-
|
2465
|
+
frb_add_each_tv(void *key, void *value, void *rtvs)
|
2480
2466
|
{
|
2481
|
-
rb_hash_aset((VALUE)rtvs, ID2SYM(
|
2467
|
+
rb_hash_aset((VALUE)rtvs, ID2SYM((ID)key), frb_get_tv(value));
|
2482
2468
|
}
|
2483
2469
|
|
2484
2470
|
/*
|
@@ -2490,12 +2476,12 @@ frt_add_each_tv(void *key, void *value, void *rtvs)
|
|
2490
2476
|
* and they are referenced by field names (as symbols).
|
2491
2477
|
*/
|
2492
2478
|
static VALUE
|
2493
|
-
|
2479
|
+
frb_ir_term_vectors(VALUE self, VALUE rdoc_id)
|
2494
2480
|
{
|
2495
2481
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2496
|
-
|
2482
|
+
Hash *tvs = ir->term_vectors(ir, FIX2INT(rdoc_id));
|
2497
2483
|
VALUE rtvs = rb_hash_new();
|
2498
|
-
h_each(tvs, &
|
2484
|
+
h_each(tvs, &frb_add_each_tv, (void *)rtvs);
|
2499
2485
|
h_destroy(tvs);
|
2500
2486
|
|
2501
2487
|
return rtvs;
|
@@ -2510,10 +2496,10 @@ frt_ir_term_vectors(VALUE self, VALUE rdoc_id)
|
|
2510
2496
|
* See TermDocEnum for more info.
|
2511
2497
|
*/
|
2512
2498
|
static VALUE
|
2513
|
-
|
2499
|
+
frb_ir_term_docs(VALUE self)
|
2514
2500
|
{
|
2515
2501
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2516
|
-
return
|
2502
|
+
return frb_get_tde(self, ir->term_docs(ir));
|
2517
2503
|
}
|
2518
2504
|
|
2519
2505
|
/*
|
@@ -2524,11 +2510,11 @@ frt_ir_term_docs(VALUE self)
|
|
2524
2510
|
* term +term+ in the field +field+. See TermDocEnum for more info.
|
2525
2511
|
*/
|
2526
2512
|
static VALUE
|
2527
|
-
|
2513
|
+
frb_ir_term_docs_for(VALUE self, VALUE rfield, VALUE rterm)
|
2528
2514
|
{
|
2529
2515
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2530
|
-
return
|
2531
|
-
|
2516
|
+
return frb_get_tde(self, ir_term_docs_for(ir,
|
2517
|
+
frb_field(rfield),
|
2532
2518
|
StringValuePtr(rterm)));
|
2533
2519
|
}
|
2534
2520
|
|
@@ -2541,10 +2527,10 @@ frt_ir_term_docs_for(VALUE self, VALUE rfield, VALUE rterm)
|
|
2541
2527
|
* more info.
|
2542
2528
|
*/
|
2543
2529
|
static VALUE
|
2544
|
-
|
2530
|
+
frb_ir_term_positions(VALUE self)
|
2545
2531
|
{
|
2546
2532
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2547
|
-
return
|
2533
|
+
return frb_get_tde(self, ir->term_positions(ir));
|
2548
2534
|
}
|
2549
2535
|
|
2550
2536
|
/*
|
@@ -2556,11 +2542,11 @@ frt_ir_term_positions(VALUE self)
|
|
2556
2542
|
* TermDocEnum for more info.
|
2557
2543
|
*/
|
2558
2544
|
static VALUE
|
2559
|
-
|
2545
|
+
frb_ir_t_pos_for(VALUE self, VALUE rfield, VALUE rterm)
|
2560
2546
|
{
|
2561
2547
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2562
|
-
return
|
2563
|
-
|
2548
|
+
return frb_get_tde(self, ir_term_positions_for(ir,
|
2549
|
+
frb_field(rfield),
|
2564
2550
|
StringValuePtr(rterm)));
|
2565
2551
|
}
|
2566
2552
|
|
@@ -2572,11 +2558,11 @@ frt_ir_t_pos_for(VALUE self, VALUE rfield, VALUE rterm)
|
|
2572
2558
|
* field +field+.
|
2573
2559
|
*/
|
2574
2560
|
static VALUE
|
2575
|
-
|
2561
|
+
frb_ir_doc_freq(VALUE self, VALUE rfield, VALUE rterm)
|
2576
2562
|
{
|
2577
2563
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2578
2564
|
return INT2FIX(ir_doc_freq(ir,
|
2579
|
-
|
2565
|
+
frb_field(rfield),
|
2580
2566
|
StringValuePtr(rterm)));
|
2581
2567
|
}
|
2582
2568
|
|
@@ -2588,10 +2574,10 @@ frt_ir_doc_freq(VALUE self, VALUE rfield, VALUE rterm)
|
|
2588
2574
|
* terms in the field +field+ in the index.
|
2589
2575
|
*/
|
2590
2576
|
static VALUE
|
2591
|
-
|
2577
|
+
frb_ir_terms(VALUE self, VALUE rfield)
|
2592
2578
|
{
|
2593
2579
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2594
|
-
return
|
2580
|
+
return frb_get_te(self, ir_terms(ir, frb_field(rfield)));
|
2595
2581
|
}
|
2596
2582
|
|
2597
2583
|
/*
|
@@ -2602,11 +2588,11 @@ frt_ir_terms(VALUE self, VALUE rfield)
|
|
2602
2588
|
* at term +term+.
|
2603
2589
|
*/
|
2604
2590
|
static VALUE
|
2605
|
-
|
2591
|
+
frb_ir_terms_from(VALUE self, VALUE rfield, VALUE rterm)
|
2606
2592
|
{
|
2607
2593
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2608
|
-
return
|
2609
|
-
|
2594
|
+
return frb_get_te(self, ir_terms_from(ir,
|
2595
|
+
frb_field(rfield),
|
2610
2596
|
StringValuePtr(rterm)));
|
2611
2597
|
}
|
2612
2598
|
|
@@ -2617,10 +2603,10 @@ frt_ir_terms_from(VALUE self, VALUE rfield, VALUE rterm)
|
|
2617
2603
|
* Same return a count of the number of terms in the field
|
2618
2604
|
*/
|
2619
2605
|
static VALUE
|
2620
|
-
|
2606
|
+
frb_ir_term_count(VALUE self, VALUE rfield)
|
2621
2607
|
{
|
2622
2608
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2623
|
-
TermEnum *te = ir_terms(ir,
|
2609
|
+
TermEnum *te = ir_terms(ir, frb_field(rfield));
|
2624
2610
|
int count = 0;
|
2625
2611
|
while (te->next(te)) {
|
2626
2612
|
count++;
|
@@ -2639,14 +2625,14 @@ frt_ir_term_count(VALUE self, VALUE rfield)
|
|
2639
2625
|
* gathered from the FieldInfos object.
|
2640
2626
|
*/
|
2641
2627
|
static VALUE
|
2642
|
-
|
2628
|
+
frb_ir_fields(VALUE self)
|
2643
2629
|
{
|
2644
2630
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2645
2631
|
FieldInfos *fis = ir->fis;
|
2646
2632
|
VALUE rfield_names = rb_ary_new();
|
2647
2633
|
int i;
|
2648
2634
|
for (i = 0; i < fis->size; i++) {
|
2649
|
-
rb_ary_push(rfield_names,
|
2635
|
+
rb_ary_push(rfield_names, FSYM2SYM(fis->fields[i]->name));
|
2650
2636
|
}
|
2651
2637
|
return rfield_names;
|
2652
2638
|
}
|
@@ -2658,10 +2644,10 @@ frt_ir_fields(VALUE self)
|
|
2658
2644
|
* Get the FieldInfos object for this IndexReader.
|
2659
2645
|
*/
|
2660
2646
|
static VALUE
|
2661
|
-
|
2647
|
+
frb_ir_field_infos(VALUE self)
|
2662
2648
|
{
|
2663
2649
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2664
|
-
return
|
2650
|
+
return frb_get_field_infos(ir->fis);
|
2665
2651
|
}
|
2666
2652
|
|
2667
2653
|
/*
|
@@ -2674,7 +2660,7 @@ frt_ir_field_infos(VALUE self)
|
|
2674
2660
|
* of field names can also be gathered from the FieldInfos object.
|
2675
2661
|
*/
|
2676
2662
|
static VALUE
|
2677
|
-
|
2663
|
+
frb_ir_tk_fields(VALUE self)
|
2678
2664
|
{
|
2679
2665
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2680
2666
|
FieldInfos *fis = ir->fis;
|
@@ -2682,7 +2668,7 @@ frt_ir_tk_fields(VALUE self)
|
|
2682
2668
|
int i;
|
2683
2669
|
for (i = 0; i < fis->size; i++) {
|
2684
2670
|
if (!fi_is_tokenized(fis->fields[i])) continue;
|
2685
|
-
rb_ary_push(rfield_names,
|
2671
|
+
rb_ary_push(rfield_names, FSYM2SYM(fis->fields[i]->name));
|
2686
2672
|
}
|
2687
2673
|
return rfield_names;
|
2688
2674
|
}
|
@@ -2694,7 +2680,7 @@ frt_ir_tk_fields(VALUE self)
|
|
2694
2680
|
* Returns the current version of the index reader.
|
2695
2681
|
*/
|
2696
2682
|
static VALUE
|
2697
|
-
|
2683
|
+
frb_ir_version(VALUE self)
|
2698
2684
|
{
|
2699
2685
|
IndexReader *ir = (IndexReader *)DATA_PTR(self);
|
2700
2686
|
return ULL2NUM(ir->sis->version);
|
@@ -2850,24 +2836,24 @@ Init_FieldInfo(void)
|
|
2850
2836
|
sym_with_positions_offsets = ID2SYM(rb_intern("with_positions_offsets"));
|
2851
2837
|
|
2852
2838
|
cFieldInfo = rb_define_class_under(mIndex, "FieldInfo", rb_cObject);
|
2853
|
-
rb_define_alloc_func(cFieldInfo,
|
2854
|
-
|
2855
|
-
rb_define_method(cFieldInfo, "initialize",
|
2856
|
-
rb_define_method(cFieldInfo, "name",
|
2857
|
-
rb_define_method(cFieldInfo, "stored?",
|
2858
|
-
rb_define_method(cFieldInfo, "compressed?",
|
2859
|
-
rb_define_method(cFieldInfo, "indexed?",
|
2860
|
-
rb_define_method(cFieldInfo, "tokenized?",
|
2861
|
-
rb_define_method(cFieldInfo, "omit_norms?",
|
2839
|
+
rb_define_alloc_func(cFieldInfo, frb_data_alloc);
|
2840
|
+
|
2841
|
+
rb_define_method(cFieldInfo, "initialize", frb_fi_init, -1);
|
2842
|
+
rb_define_method(cFieldInfo, "name", frb_fi_name, 0);
|
2843
|
+
rb_define_method(cFieldInfo, "stored?", frb_fi_is_stored, 0);
|
2844
|
+
rb_define_method(cFieldInfo, "compressed?", frb_fi_is_compressed, 0);
|
2845
|
+
rb_define_method(cFieldInfo, "indexed?", frb_fi_is_indexed, 0);
|
2846
|
+
rb_define_method(cFieldInfo, "tokenized?", frb_fi_is_tokenized, 0);
|
2847
|
+
rb_define_method(cFieldInfo, "omit_norms?", frb_fi_omit_norms, 0);
|
2862
2848
|
rb_define_method(cFieldInfo, "store_term_vector?",
|
2863
|
-
|
2849
|
+
frb_fi_store_term_vector, 0);
|
2864
2850
|
rb_define_method(cFieldInfo, "store_positions?",
|
2865
|
-
|
2851
|
+
frb_fi_store_positions, 0);
|
2866
2852
|
rb_define_method(cFieldInfo, "store_offsets?",
|
2867
|
-
|
2868
|
-
rb_define_method(cFieldInfo, "has_norms?",
|
2869
|
-
rb_define_method(cFieldInfo, "boost",
|
2870
|
-
rb_define_method(cFieldInfo, "to_s",
|
2853
|
+
frb_fi_store_offsets, 0);
|
2854
|
+
rb_define_method(cFieldInfo, "has_norms?", frb_fi_has_norms, 0);
|
2855
|
+
rb_define_method(cFieldInfo, "boost", frb_fi_boost, 0);
|
2856
|
+
rb_define_method(cFieldInfo, "to_s", frb_fi_to_s, 0);
|
2871
2857
|
}
|
2872
2858
|
|
2873
2859
|
/*
|
@@ -2916,21 +2902,21 @@ Init_FieldInfos(void)
|
|
2916
2902
|
Init_FieldInfo();
|
2917
2903
|
|
2918
2904
|
cFieldInfos = rb_define_class_under(mIndex, "FieldInfos", rb_cObject);
|
2919
|
-
rb_define_alloc_func(cFieldInfos,
|
2920
|
-
|
2921
|
-
rb_define_method(cFieldInfos, "initialize",
|
2922
|
-
rb_define_method(cFieldInfos, "to_a",
|
2923
|
-
rb_define_method(cFieldInfos, "[]",
|
2924
|
-
rb_define_method(cFieldInfos, "add",
|
2925
|
-
rb_define_method(cFieldInfos, "<<",
|
2926
|
-
rb_define_method(cFieldInfos, "add_field",
|
2927
|
-
rb_define_method(cFieldInfos, "each",
|
2928
|
-
rb_define_method(cFieldInfos, "to_s",
|
2929
|
-
rb_define_method(cFieldInfos, "size",
|
2905
|
+
rb_define_alloc_func(cFieldInfos, frb_data_alloc);
|
2906
|
+
|
2907
|
+
rb_define_method(cFieldInfos, "initialize", frb_fis_init, -1);
|
2908
|
+
rb_define_method(cFieldInfos, "to_a", frb_fis_to_a, 0);
|
2909
|
+
rb_define_method(cFieldInfos, "[]", frb_fis_get, 1);
|
2910
|
+
rb_define_method(cFieldInfos, "add", frb_fis_add, 1);
|
2911
|
+
rb_define_method(cFieldInfos, "<<", frb_fis_add, 1);
|
2912
|
+
rb_define_method(cFieldInfos, "add_field", frb_fis_add_field, -1);
|
2913
|
+
rb_define_method(cFieldInfos, "each", frb_fis_each, 0);
|
2914
|
+
rb_define_method(cFieldInfos, "to_s", frb_fis_to_s, 0);
|
2915
|
+
rb_define_method(cFieldInfos, "size", frb_fis_size, 0);
|
2930
2916
|
rb_define_method(cFieldInfos, "create_index",
|
2931
|
-
|
2932
|
-
rb_define_method(cFieldInfos, "fields",
|
2933
|
-
rb_define_method(cFieldInfos, "tokenized_fields",
|
2917
|
+
frb_fis_create_index, 1);
|
2918
|
+
rb_define_method(cFieldInfos, "fields", frb_fis_get_fields, 0);
|
2919
|
+
rb_define_method(cFieldInfos, "tokenized_fields", frb_fis_get_tk_fields, 0);
|
2934
2920
|
}
|
2935
2921
|
|
2936
2922
|
/*
|
@@ -2960,16 +2946,16 @@ Init_TermEnum(void)
|
|
2960
2946
|
id_term = rb_intern("@term");
|
2961
2947
|
|
2962
2948
|
cTermEnum = rb_define_class_under(mIndex, "TermEnum", rb_cObject);
|
2963
|
-
rb_define_alloc_func(cTermEnum,
|
2949
|
+
rb_define_alloc_func(cTermEnum, frb_data_alloc);
|
2964
2950
|
|
2965
|
-
rb_define_method(cTermEnum, "next?",
|
2966
|
-
rb_define_method(cTermEnum, "term",
|
2967
|
-
rb_define_method(cTermEnum, "doc_freq",
|
2968
|
-
rb_define_method(cTermEnum, "skip_to",
|
2969
|
-
rb_define_method(cTermEnum, "each",
|
2970
|
-
rb_define_method(cTermEnum, "field=",
|
2971
|
-
rb_define_method(cTermEnum, "set_field",
|
2972
|
-
rb_define_method(cTermEnum, "to_json",
|
2951
|
+
rb_define_method(cTermEnum, "next?", frb_te_next, 0);
|
2952
|
+
rb_define_method(cTermEnum, "term", frb_te_term, 0);
|
2953
|
+
rb_define_method(cTermEnum, "doc_freq", frb_te_doc_freq, 0);
|
2954
|
+
rb_define_method(cTermEnum, "skip_to", frb_te_skip_to, 1);
|
2955
|
+
rb_define_method(cTermEnum, "each", frb_te_each, 0);
|
2956
|
+
rb_define_method(cTermEnum, "field=", frb_te_set_field, 1);
|
2957
|
+
rb_define_method(cTermEnum, "set_field",frb_te_set_field, 1);
|
2958
|
+
rb_define_method(cTermEnum, "to_json", frb_te_to_json, -1);
|
2973
2959
|
}
|
2974
2960
|
|
2975
2961
|
/*
|
@@ -3011,17 +2997,17 @@ Init_TermDocEnum(void)
|
|
3011
2997
|
id_field_num = rb_intern("@field_num");
|
3012
2998
|
|
3013
2999
|
cTermDocEnum = rb_define_class_under(mIndex, "TermDocEnum", rb_cObject);
|
3014
|
-
rb_define_alloc_func(cTermDocEnum,
|
3015
|
-
rb_define_method(cTermDocEnum, "seek",
|
3016
|
-
rb_define_method(cTermDocEnum, "seek_term_enum",
|
3017
|
-
rb_define_method(cTermDocEnum, "doc",
|
3018
|
-
rb_define_method(cTermDocEnum, "freq",
|
3019
|
-
rb_define_method(cTermDocEnum, "next?",
|
3020
|
-
rb_define_method(cTermDocEnum, "next_position",
|
3021
|
-
rb_define_method(cTermDocEnum, "each",
|
3022
|
-
rb_define_method(cTermDocEnum, "each_position",
|
3023
|
-
rb_define_method(cTermDocEnum, "skip_to",
|
3024
|
-
rb_define_method(cTermDocEnum, "to_json",
|
3000
|
+
rb_define_alloc_func(cTermDocEnum, frb_data_alloc);
|
3001
|
+
rb_define_method(cTermDocEnum, "seek", frb_tde_seek, 2);
|
3002
|
+
rb_define_method(cTermDocEnum, "seek_term_enum", frb_tde_seek_te, 1);
|
3003
|
+
rb_define_method(cTermDocEnum, "doc", frb_tde_doc, 0);
|
3004
|
+
rb_define_method(cTermDocEnum, "freq", frb_tde_freq, 0);
|
3005
|
+
rb_define_method(cTermDocEnum, "next?", frb_tde_next, 0);
|
3006
|
+
rb_define_method(cTermDocEnum, "next_position", frb_tde_next_position, 0);
|
3007
|
+
rb_define_method(cTermDocEnum, "each", frb_tde_each, 0);
|
3008
|
+
rb_define_method(cTermDocEnum, "each_position", frb_tde_each_position, 0);
|
3009
|
+
rb_define_method(cTermDocEnum, "skip_to", frb_tde_skip_to, 1);
|
3010
|
+
rb_define_method(cTermDocEnum, "to_json", frb_tde_to_json, -1);
|
3025
3011
|
}
|
3026
3012
|
|
3027
3013
|
/* rdochack
|
@@ -3078,7 +3064,7 @@ Init_TVTerm(void)
|
|
3078
3064
|
/* rdochack
|
3079
3065
|
cTVTerm = rb_define_class_under(cTermVector, "TVTerm", rb_cObject);
|
3080
3066
|
*/
|
3081
|
-
cTVTerm = rb_struct_define(tv_term_class, "text", "positions", NULL);
|
3067
|
+
cTVTerm = rb_struct_define(tv_term_class, "text", "freq", "positions", NULL);
|
3082
3068
|
rb_set_class_path(cTVTerm, cTermVector, tv_term_class);
|
3083
3069
|
rb_const_set(mIndex, rb_intern(tv_term_class), cTVTerm);
|
3084
3070
|
}
|
@@ -3097,7 +3083,7 @@ Init_TVTerm(void)
|
|
3097
3083
|
*
|
3098
3084
|
* == Example
|
3099
3085
|
*
|
3100
|
-
* tv = index_reader.term_vector(:content)
|
3086
|
+
* tv = index_reader.term_vector(doc_id, :content)
|
3101
3087
|
* tv_term = tv.find {|tvt| tvt.term = "fox"}
|
3102
3088
|
*
|
3103
3089
|
* # get the term frequency
|
@@ -3262,7 +3248,7 @@ Init_IndexWriter(void)
|
|
3262
3248
|
sym_use_compound_file = ID2SYM(rb_intern("use_compound_file"));
|
3263
3249
|
|
3264
3250
|
cIndexWriter = rb_define_class_under(mIndex, "IndexWriter", rb_cObject);
|
3265
|
-
rb_define_alloc_func(cIndexWriter,
|
3251
|
+
rb_define_alloc_func(cIndexWriter, frb_data_alloc);
|
3266
3252
|
|
3267
3253
|
rb_define_const(cIndexWriter, "WRITE_LOCK_TIMEOUT", INT2FIX(1));
|
3268
3254
|
rb_define_const(cIndexWriter, "COMMIT_LOCK_TIMEOUT", INT2FIX(10));
|
@@ -3289,64 +3275,64 @@ Init_IndexWriter(void)
|
|
3289
3275
|
rb_define_const(cIndexWriter, "DEFAULT_USE_COMPOUND_FILE",
|
3290
3276
|
default_config.use_compound_file ? Qtrue : Qfalse);
|
3291
3277
|
|
3292
|
-
rb_define_method(cIndexWriter, "initialize",
|
3293
|
-
rb_define_method(cIndexWriter, "doc_count",
|
3294
|
-
rb_define_method(cIndexWriter, "close",
|
3295
|
-
rb_define_method(cIndexWriter, "add_document",
|
3296
|
-
rb_define_method(cIndexWriter, "<<",
|
3297
|
-
rb_define_method(cIndexWriter, "optimize",
|
3298
|
-
rb_define_method(cIndexWriter, "commit",
|
3299
|
-
rb_define_method(cIndexWriter, "add_readers",
|
3300
|
-
rb_define_method(cIndexWriter, "delete",
|
3301
|
-
rb_define_method(cIndexWriter, "field_infos",
|
3302
|
-
rb_define_method(cIndexWriter, "analyzer",
|
3303
|
-
rb_define_method(cIndexWriter, "analyzer=",
|
3304
|
-
rb_define_method(cIndexWriter, "version",
|
3278
|
+
rb_define_method(cIndexWriter, "initialize", frb_iw_init, -1);
|
3279
|
+
rb_define_method(cIndexWriter, "doc_count", frb_iw_get_doc_count, 0);
|
3280
|
+
rb_define_method(cIndexWriter, "close", frb_iw_close, 0);
|
3281
|
+
rb_define_method(cIndexWriter, "add_document", frb_iw_add_doc, 1);
|
3282
|
+
rb_define_method(cIndexWriter, "<<", frb_iw_add_doc, 1);
|
3283
|
+
rb_define_method(cIndexWriter, "optimize", frb_iw_optimize, 0);
|
3284
|
+
rb_define_method(cIndexWriter, "commit", frb_iw_commit, 0);
|
3285
|
+
rb_define_method(cIndexWriter, "add_readers", frb_iw_add_readers, 1);
|
3286
|
+
rb_define_method(cIndexWriter, "delete", frb_iw_delete, 2);
|
3287
|
+
rb_define_method(cIndexWriter, "field_infos", frb_iw_field_infos, 0);
|
3288
|
+
rb_define_method(cIndexWriter, "analyzer", frb_iw_get_analyzer, 0);
|
3289
|
+
rb_define_method(cIndexWriter, "analyzer=", frb_iw_set_analyzer, 1);
|
3290
|
+
rb_define_method(cIndexWriter, "version", frb_iw_version, 0);
|
3305
3291
|
|
3306
3292
|
rb_define_method(cIndexWriter, "chunk_size",
|
3307
|
-
|
3293
|
+
frb_iw_get_chunk_size, 0);
|
3308
3294
|
rb_define_method(cIndexWriter, "chunk_size=",
|
3309
|
-
|
3295
|
+
frb_iw_set_chunk_size, 1);
|
3310
3296
|
|
3311
3297
|
rb_define_method(cIndexWriter, "max_buffer_memory",
|
3312
|
-
|
3298
|
+
frb_iw_get_max_buffer_memory, 0);
|
3313
3299
|
rb_define_method(cIndexWriter, "max_buffer_memory=",
|
3314
|
-
|
3300
|
+
frb_iw_set_max_buffer_memory, 1);
|
3315
3301
|
|
3316
3302
|
rb_define_method(cIndexWriter, "term_index_interval",
|
3317
|
-
|
3303
|
+
frb_iw_get_index_interval, 0);
|
3318
3304
|
rb_define_method(cIndexWriter, "term_index_interval=",
|
3319
|
-
|
3305
|
+
frb_iw_set_index_interval, 1);
|
3320
3306
|
|
3321
3307
|
rb_define_method(cIndexWriter, "doc_skip_interval",
|
3322
|
-
|
3308
|
+
frb_iw_get_skip_interval, 0);
|
3323
3309
|
rb_define_method(cIndexWriter, "doc_skip_interval=",
|
3324
|
-
|
3310
|
+
frb_iw_set_skip_interval, 1);
|
3325
3311
|
|
3326
3312
|
rb_define_method(cIndexWriter, "merge_factor",
|
3327
|
-
|
3313
|
+
frb_iw_get_merge_factor, 0);
|
3328
3314
|
rb_define_method(cIndexWriter, "merge_factor=",
|
3329
|
-
|
3315
|
+
frb_iw_set_merge_factor, 1);
|
3330
3316
|
|
3331
3317
|
rb_define_method(cIndexWriter, "max_buffered_docs",
|
3332
|
-
|
3318
|
+
frb_iw_get_max_buffered_docs, 0);
|
3333
3319
|
rb_define_method(cIndexWriter, "max_buffered_docs=",
|
3334
|
-
|
3320
|
+
frb_iw_set_max_buffered_docs, 1);
|
3335
3321
|
|
3336
3322
|
rb_define_method(cIndexWriter, "max_merge_docs",
|
3337
|
-
|
3323
|
+
frb_iw_get_max_merge_docs, 0);
|
3338
3324
|
rb_define_method(cIndexWriter, "max_merge_docs=",
|
3339
|
-
|
3325
|
+
frb_iw_set_max_merge_docs, 1);
|
3340
3326
|
|
3341
3327
|
rb_define_method(cIndexWriter, "max_field_length",
|
3342
|
-
|
3328
|
+
frb_iw_get_max_field_length, 0);
|
3343
3329
|
rb_define_method(cIndexWriter, "max_field_length=",
|
3344
|
-
|
3330
|
+
frb_iw_set_max_field_length, 1);
|
3345
3331
|
|
3346
3332
|
rb_define_method(cIndexWriter, "use_compound_file",
|
3347
|
-
|
3333
|
+
frb_iw_get_use_compound_file, 0);
|
3348
3334
|
rb_define_method(cIndexWriter, "use_compound_file=",
|
3349
|
-
|
3335
|
+
frb_iw_set_use_compound_file, 1);
|
3350
3336
|
|
3351
3337
|
}
|
3352
3338
|
|
@@ -3387,12 +3373,12 @@ Init_LazyDoc(void)
|
|
3387
3373
|
|
3388
3374
|
|
3389
3375
|
cLazyDoc = rb_define_class_under(mIndex, "LazyDoc", rb_cHash);
|
3390
|
-
rb_define_method(cLazyDoc, "default",
|
3391
|
-
rb_define_method(cLazyDoc, "load",
|
3392
|
-
rb_define_method(cLazyDoc, "fields",
|
3376
|
+
rb_define_method(cLazyDoc, "default", frb_lzd_default, 1);
|
3377
|
+
rb_define_method(cLazyDoc, "load", frb_lzd_load, 0);
|
3378
|
+
rb_define_method(cLazyDoc, "fields", frb_lzd_fields, 0);
|
3393
3379
|
|
3394
3380
|
cLazyDocData = rb_define_class_under(cLazyDoc, "LazyDocData", rb_cObject);
|
3395
|
-
rb_define_alloc_func(cLazyDocData,
|
3381
|
+
rb_define_alloc_func(cLazyDocData, frb_data_alloc);
|
3396
3382
|
}
|
3397
3383
|
|
3398
3384
|
/*
|
@@ -3409,37 +3395,37 @@ void
|
|
3409
3395
|
Init_IndexReader(void)
|
3410
3396
|
{
|
3411
3397
|
cIndexReader = rb_define_class_under(mIndex, "IndexReader", rb_cObject);
|
3412
|
-
rb_define_alloc_func(cIndexReader,
|
3413
|
-
rb_define_method(cIndexReader, "initialize",
|
3414
|
-
rb_define_method(cIndexReader, "set_norm",
|
3415
|
-
rb_define_method(cIndexReader, "norms",
|
3416
|
-
rb_define_method(cIndexReader, "get_norms_into",
|
3417
|
-
rb_define_method(cIndexReader, "commit",
|
3418
|
-
rb_define_method(cIndexReader, "close",
|
3419
|
-
rb_define_method(cIndexReader, "has_deletions?",
|
3420
|
-
rb_define_method(cIndexReader, "delete",
|
3421
|
-
rb_define_method(cIndexReader, "deleted?",
|
3422
|
-
rb_define_method(cIndexReader, "max_doc",
|
3423
|
-
rb_define_method(cIndexReader, "num_docs",
|
3424
|
-
rb_define_method(cIndexReader, "undelete_all",
|
3425
|
-
rb_define_method(cIndexReader, "latest?",
|
3426
|
-
rb_define_method(cIndexReader, "get_document",
|
3427
|
-
rb_define_method(cIndexReader, "[]",
|
3428
|
-
rb_define_method(cIndexReader, "term_vector",
|
3429
|
-
rb_define_method(cIndexReader, "term_vectors",
|
3430
|
-
rb_define_method(cIndexReader, "term_docs",
|
3431
|
-
rb_define_method(cIndexReader, "term_positions",
|
3432
|
-
rb_define_method(cIndexReader, "term_docs_for",
|
3433
|
-
rb_define_method(cIndexReader, "term_positions_for",
|
3434
|
-
rb_define_method(cIndexReader, "doc_freq",
|
3435
|
-
rb_define_method(cIndexReader, "terms",
|
3436
|
-
rb_define_method(cIndexReader, "terms_from",
|
3437
|
-
rb_define_method(cIndexReader, "term_count",
|
3438
|
-
rb_define_method(cIndexReader, "fields",
|
3439
|
-
rb_define_method(cIndexReader, "field_names",
|
3440
|
-
rb_define_method(cIndexReader, "field_infos",
|
3441
|
-
rb_define_method(cIndexReader, "tokenized_fields",
|
3442
|
-
rb_define_method(cIndexReader, "version",
|
3398
|
+
rb_define_alloc_func(cIndexReader, frb_data_alloc);
|
3399
|
+
rb_define_method(cIndexReader, "initialize", frb_ir_init, 1);
|
3400
|
+
rb_define_method(cIndexReader, "set_norm", frb_ir_set_norm, 3);
|
3401
|
+
rb_define_method(cIndexReader, "norms", frb_ir_norms, 1);
|
3402
|
+
rb_define_method(cIndexReader, "get_norms_into",frb_ir_get_norms_into, 3);
|
3403
|
+
rb_define_method(cIndexReader, "commit", frb_ir_commit, 0);
|
3404
|
+
rb_define_method(cIndexReader, "close", frb_ir_close, 0);
|
3405
|
+
rb_define_method(cIndexReader, "has_deletions?",frb_ir_has_deletions, 0);
|
3406
|
+
rb_define_method(cIndexReader, "delete", frb_ir_delete, 1);
|
3407
|
+
rb_define_method(cIndexReader, "deleted?", frb_ir_is_deleted, 1);
|
3408
|
+
rb_define_method(cIndexReader, "max_doc", frb_ir_max_doc, 0);
|
3409
|
+
rb_define_method(cIndexReader, "num_docs", frb_ir_num_docs, 0);
|
3410
|
+
rb_define_method(cIndexReader, "undelete_all", frb_ir_undelete_all, 0);
|
3411
|
+
rb_define_method(cIndexReader, "latest?", frb_ir_is_latest, 0);
|
3412
|
+
rb_define_method(cIndexReader, "get_document", frb_ir_get_doc, -1);
|
3413
|
+
rb_define_method(cIndexReader, "[]", frb_ir_get_doc, -1);
|
3414
|
+
rb_define_method(cIndexReader, "term_vector", frb_ir_term_vector, 2);
|
3415
|
+
rb_define_method(cIndexReader, "term_vectors", frb_ir_term_vectors, 1);
|
3416
|
+
rb_define_method(cIndexReader, "term_docs", frb_ir_term_docs, 0);
|
3417
|
+
rb_define_method(cIndexReader, "term_positions",frb_ir_term_positions, 0);
|
3418
|
+
rb_define_method(cIndexReader, "term_docs_for", frb_ir_term_docs_for, 2);
|
3419
|
+
rb_define_method(cIndexReader, "term_positions_for", frb_ir_t_pos_for, 2);
|
3420
|
+
rb_define_method(cIndexReader, "doc_freq", frb_ir_doc_freq, 2);
|
3421
|
+
rb_define_method(cIndexReader, "terms", frb_ir_terms, 1);
|
3422
|
+
rb_define_method(cIndexReader, "terms_from", frb_ir_terms_from, 2);
|
3423
|
+
rb_define_method(cIndexReader, "term_count", frb_ir_term_count, 1);
|
3424
|
+
rb_define_method(cIndexReader, "fields", frb_ir_fields, 0);
|
3425
|
+
rb_define_method(cIndexReader, "field_names", frb_ir_fields, 0);
|
3426
|
+
rb_define_method(cIndexReader, "field_infos", frb_ir_field_infos, 0);
|
3427
|
+
rb_define_method(cIndexReader, "tokenized_fields", frb_ir_tk_fields, 0);
|
3428
|
+
rb_define_method(cIndexReader, "version", frb_ir_version, 0);
|
3443
3429
|
}
|
3444
3430
|
|
3445
3431
|
/* rdoc hack
|
@@ -3472,6 +3458,7 @@ Init_Index(void)
|
|
3472
3458
|
sym_boost = ID2SYM(rb_intern("boost"));
|
3473
3459
|
sym_analyzer = ID2SYM(rb_intern("analyzer"));
|
3474
3460
|
sym_close_dir = ID2SYM(rb_intern("close_dir"));
|
3461
|
+
fsym_content = I("content");
|
3475
3462
|
|
3476
3463
|
Init_TermVector();
|
3477
3464
|
Init_TermEnum();
|