isomorfeus-ferret 0.17.2 → 0.17.4
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/ext/isomorfeus_ferret_ext/benchmark.c +9 -20
- data/ext/isomorfeus_ferret_ext/benchmarks_all.h +1 -2
- data/ext/isomorfeus_ferret_ext/bm_hash.c +1 -2
- data/ext/isomorfeus_ferret_ext/brotli_dec_decode.c +4 -2
- data/ext/isomorfeus_ferret_ext/brotli_enc_encode.c +3 -2
- data/ext/isomorfeus_ferret_ext/frb_analysis.c +4 -5
- data/ext/isomorfeus_ferret_ext/frb_field_info.c +3 -4
- data/ext/isomorfeus_ferret_ext/frb_index.c +161 -187
- data/ext/isomorfeus_ferret_ext/frb_lazy_doc.c +14 -16
- data/ext/isomorfeus_ferret_ext/frb_search.c +77 -69
- data/ext/isomorfeus_ferret_ext/frb_store.c +27 -13
- data/ext/isomorfeus_ferret_ext/frb_utils.c +3 -6
- data/ext/isomorfeus_ferret_ext/frt_analysis.c +39 -46
- data/ext/isomorfeus_ferret_ext/frt_analysis.h +9 -9
- data/ext/isomorfeus_ferret_ext/frt_array.c +11 -22
- data/ext/isomorfeus_ferret_ext/frt_bitvector.h +3 -6
- data/ext/isomorfeus_ferret_ext/frt_doc_field.c +87 -0
- data/ext/isomorfeus_ferret_ext/frt_doc_field.h +26 -0
- data/ext/isomorfeus_ferret_ext/frt_document.c +4 -97
- data/ext/isomorfeus_ferret_ext/frt_document.h +5 -33
- data/ext/isomorfeus_ferret_ext/frt_except.c +8 -6
- data/ext/isomorfeus_ferret_ext/frt_except.h +1 -2
- data/ext/isomorfeus_ferret_ext/frt_field_index.c +14 -33
- data/ext/isomorfeus_ferret_ext/frt_field_index.h +0 -6
- data/ext/isomorfeus_ferret_ext/frt_field_info.c +69 -0
- data/ext/isomorfeus_ferret_ext/frt_field_info.h +49 -0
- data/ext/isomorfeus_ferret_ext/frt_field_infos.c +196 -0
- data/ext/isomorfeus_ferret_ext/frt_field_infos.h +35 -0
- data/ext/isomorfeus_ferret_ext/frt_filter.c +2 -2
- data/ext/isomorfeus_ferret_ext/frt_global.c +10 -4
- data/ext/isomorfeus_ferret_ext/frt_global.h +11 -15
- data/ext/isomorfeus_ferret_ext/frt_hash.c +8 -8
- data/ext/isomorfeus_ferret_ext/frt_hash.h +1 -2
- data/ext/isomorfeus_ferret_ext/frt_hashset.c +20 -40
- data/ext/isomorfeus_ferret_ext/frt_hashset.h +1 -2
- data/ext/isomorfeus_ferret_ext/frt_helper.c +7 -15
- data/ext/isomorfeus_ferret_ext/frt_in_stream.c +35 -45
- data/ext/isomorfeus_ferret_ext/frt_in_stream.h +3 -2
- data/ext/isomorfeus_ferret_ext/frt_ind.c +21 -39
- data/ext/isomorfeus_ferret_ext/frt_ind.h +1 -1
- data/ext/isomorfeus_ferret_ext/frt_index.c +334 -848
- data/ext/isomorfeus_ferret_ext/frt_index.h +4 -105
- data/ext/isomorfeus_ferret_ext/frt_lang.c +5 -10
- data/ext/isomorfeus_ferret_ext/frt_lazy_doc.c +3 -3
- data/ext/isomorfeus_ferret_ext/frt_lazy_doc.h +1 -1
- data/ext/isomorfeus_ferret_ext/frt_lazy_doc_field.c +18 -25
- data/ext/isomorfeus_ferret_ext/frt_lazy_doc_field.h +5 -5
- data/ext/isomorfeus_ferret_ext/frt_mdbx_store.c +102 -70
- data/ext/isomorfeus_ferret_ext/frt_mempool.c +8 -16
- data/ext/isomorfeus_ferret_ext/frt_multimapper.c +23 -46
- data/ext/isomorfeus_ferret_ext/frt_multimapper.h +4 -8
- data/ext/isomorfeus_ferret_ext/frt_out_stream.c +31 -43
- data/ext/isomorfeus_ferret_ext/frt_out_stream.h +2 -2
- data/ext/isomorfeus_ferret_ext/frt_posh.c +6 -819
- data/ext/isomorfeus_ferret_ext/frt_posh.h +0 -57
- data/ext/isomorfeus_ferret_ext/frt_priorityqueue.c +11 -22
- data/ext/isomorfeus_ferret_ext/frt_priorityqueue.h +1 -2
- data/ext/isomorfeus_ferret_ext/frt_q_boolean.c +131 -217
- data/ext/isomorfeus_ferret_ext/frt_q_const_score.c +2 -2
- data/ext/isomorfeus_ferret_ext/frt_q_filtered_query.c +4 -4
- data/ext/isomorfeus_ferret_ext/frt_q_match_all.c +18 -26
- data/ext/isomorfeus_ferret_ext/frt_q_multi_term.c +27 -28
- data/ext/isomorfeus_ferret_ext/frt_q_parser.c +49 -98
- data/ext/isomorfeus_ferret_ext/frt_q_phrase.c +64 -116
- data/ext/isomorfeus_ferret_ext/frt_q_range.c +8 -14
- data/ext/isomorfeus_ferret_ext/frt_q_span.c +251 -365
- data/ext/isomorfeus_ferret_ext/frt_q_term.c +9 -9
- data/ext/isomorfeus_ferret_ext/frt_q_wildcard.c +1 -2
- data/ext/isomorfeus_ferret_ext/frt_ram_store.c +134 -85
- data/ext/isomorfeus_ferret_ext/frt_search.c +109 -191
- data/ext/isomorfeus_ferret_ext/frt_search.h +6 -6
- data/ext/isomorfeus_ferret_ext/frt_similarity.c +12 -23
- data/ext/isomorfeus_ferret_ext/frt_similarity.h +1 -2
- data/ext/isomorfeus_ferret_ext/frt_sort.c +20 -20
- data/ext/isomorfeus_ferret_ext/frt_store.c +13 -25
- data/ext/isomorfeus_ferret_ext/frt_store.h +86 -52
- data/ext/isomorfeus_ferret_ext/frt_term_vectors.c +8 -16
- data/ext/isomorfeus_ferret_ext/frt_win32.h +5 -10
- data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.c +12 -11
- data/ext/isomorfeus_ferret_ext/isomorfeus_ferret.h +11 -13
- data/ext/isomorfeus_ferret_ext/lz4.c +422 -195
- data/ext/isomorfeus_ferret_ext/lz4.h +114 -46
- data/ext/isomorfeus_ferret_ext/lz4frame.c +421 -242
- data/ext/isomorfeus_ferret_ext/lz4frame.h +122 -53
- data/ext/isomorfeus_ferret_ext/lz4hc.c +127 -111
- data/ext/isomorfeus_ferret_ext/lz4hc.h +14 -14
- data/ext/isomorfeus_ferret_ext/lz4xxhash.h +1 -1
- data/ext/isomorfeus_ferret_ext/mdbx.c +3762 -2526
- data/ext/isomorfeus_ferret_ext/mdbx.h +115 -70
- data/ext/isomorfeus_ferret_ext/test.c +41 -88
- data/ext/isomorfeus_ferret_ext/test.h +3 -6
- data/ext/isomorfeus_ferret_ext/test_1710.c +11 -13
- data/ext/isomorfeus_ferret_ext/test_analysis.c +32 -64
- data/ext/isomorfeus_ferret_ext/test_array.c +6 -12
- data/ext/isomorfeus_ferret_ext/test_bitvector.c +12 -24
- data/ext/isomorfeus_ferret_ext/test_document.c +23 -33
- data/ext/isomorfeus_ferret_ext/test_except.c +10 -21
- data/ext/isomorfeus_ferret_ext/test_fields.c +62 -68
- data/ext/isomorfeus_ferret_ext/test_file_deleter.c +16 -25
- data/ext/isomorfeus_ferret_ext/test_filter.c +22 -33
- data/ext/isomorfeus_ferret_ext/test_global.c +14 -29
- data/ext/isomorfeus_ferret_ext/test_hash.c +19 -38
- data/ext/isomorfeus_ferret_ext/test_hashset.c +8 -16
- data/ext/isomorfeus_ferret_ext/test_helper.c +4 -8
- data/ext/isomorfeus_ferret_ext/test_highlighter.c +16 -28
- data/ext/isomorfeus_ferret_ext/test_index.c +307 -519
- data/ext/isomorfeus_ferret_ext/test_lang.c +7 -14
- data/ext/isomorfeus_ferret_ext/test_mdbx_store.c +2 -5
- data/ext/isomorfeus_ferret_ext/test_mempool.c +5 -10
- data/ext/isomorfeus_ferret_ext/test_multimapper.c +3 -6
- data/ext/isomorfeus_ferret_ext/test_priorityqueue.c +9 -18
- data/ext/isomorfeus_ferret_ext/test_q_const_score.c +4 -6
- data/ext/isomorfeus_ferret_ext/test_q_filtered.c +3 -4
- data/ext/isomorfeus_ferret_ext/test_q_fuzzy.c +9 -15
- data/ext/isomorfeus_ferret_ext/test_q_parser.c +8 -16
- data/ext/isomorfeus_ferret_ext/test_q_span.c +19 -35
- data/ext/isomorfeus_ferret_ext/test_ram_store.c +14 -13
- data/ext/isomorfeus_ferret_ext/test_search.c +66 -115
- data/ext/isomorfeus_ferret_ext/test_segments.c +8 -13
- data/ext/isomorfeus_ferret_ext/test_similarity.c +2 -4
- data/ext/isomorfeus_ferret_ext/test_sort.c +17 -27
- data/ext/isomorfeus_ferret_ext/test_store.c +96 -115
- data/ext/isomorfeus_ferret_ext/test_term.c +9 -15
- data/ext/isomorfeus_ferret_ext/test_term_vectors.c +9 -14
- data/ext/isomorfeus_ferret_ext/test_test.c +4 -8
- data/ext/isomorfeus_ferret_ext/test_threading.c +15 -21
- data/ext/isomorfeus_ferret_ext/testhelper.c +11 -21
- data/ext/isomorfeus_ferret_ext/testhelper.h +1 -1
- data/ext/isomorfeus_ferret_ext/tests_all.h +1 -2
- data/lib/isomorfeus/ferret/index/index.rb +8 -8
- data/lib/isomorfeus/ferret/version.rb +1 -1
- metadata +32 -6
@@ -208,8 +208,7 @@ static VALUE frb_fis_add(VALUE self, VALUE rfi) {
|
|
208
208
|
* of the available properties.
|
209
209
|
*/
|
210
210
|
static VALUE
|
211
|
-
frb_fis_add_field(int argc, VALUE *argv, VALUE self)
|
212
|
-
{
|
211
|
+
frb_fis_add_field(int argc, VALUE *argv, VALUE self) {
|
213
212
|
FrtFieldInfos *fis = (FrtFieldInfos *)DATA_PTR(self);
|
214
213
|
FrtFieldInfo *fi;
|
215
214
|
unsigned int bits = fis->bits;
|
@@ -233,8 +232,7 @@ frb_fis_add_field(int argc, VALUE *argv, VALUE self)
|
|
233
232
|
* Iterate through the FieldInfo objects.
|
234
233
|
*/
|
235
234
|
static VALUE
|
236
|
-
frb_fis_each(VALUE self)
|
237
|
-
{
|
235
|
+
frb_fis_each(VALUE self) {
|
238
236
|
int i;
|
239
237
|
FrtFieldInfos *fis = (FrtFieldInfos *)DATA_PTR(self);
|
240
238
|
|
@@ -251,8 +249,7 @@ frb_fis_each(VALUE self)
|
|
251
249
|
* Return a string representation of the FieldInfos object.
|
252
250
|
*/
|
253
251
|
static VALUE
|
254
|
-
frb_fis_to_s(VALUE self)
|
255
|
-
{
|
252
|
+
frb_fis_to_s(VALUE self) {
|
256
253
|
FrtFieldInfos *fis = (FrtFieldInfos *)DATA_PTR(self);
|
257
254
|
char *fis_s = frt_fis_to_s(fis);
|
258
255
|
VALUE rfis_s = rb_str_new2(fis_s);
|
@@ -291,6 +288,7 @@ static VALUE frb_fis_create_index(VALUE self, VALUE rdir) {
|
|
291
288
|
StringValue(rdir);
|
292
289
|
frb_create_dir(rdir);
|
293
290
|
store = frt_open_mdbx_store(rs2s(rdir));
|
291
|
+
store->create_folder(store, segm_idx_name);
|
294
292
|
frt_index_create(store, fis);
|
295
293
|
frt_store_close(store);
|
296
294
|
}
|
@@ -306,8 +304,7 @@ static VALUE frb_fis_create_index(VALUE self, VALUE rdir) {
|
|
306
304
|
* index.
|
307
305
|
*/
|
308
306
|
static VALUE
|
309
|
-
frb_fis_get_fields(VALUE self)
|
310
|
-
{
|
307
|
+
frb_fis_get_fields(VALUE self) {
|
311
308
|
FrtFieldInfos *fis = (FrtFieldInfos *)DATA_PTR(self);
|
312
309
|
VALUE rfield_names = rb_ary_new();
|
313
310
|
int i;
|
@@ -325,8 +322,7 @@ frb_fis_get_fields(VALUE self)
|
|
325
322
|
* in the index.
|
326
323
|
*/
|
327
324
|
static VALUE
|
328
|
-
frb_fis_get_tk_fields(VALUE self)
|
329
|
-
{
|
325
|
+
frb_fis_get_tk_fields(VALUE self) {
|
330
326
|
FrtFieldInfos *fis = (FrtFieldInfos *)DATA_PTR(self);
|
331
327
|
VALUE rfield_names = rb_ary_new();
|
332
328
|
int i;
|
@@ -657,21 +653,14 @@ static VALUE frb_tde_seek_te(VALUE self, VALUE rterm_enum) {
|
|
657
653
|
return self;
|
658
654
|
}
|
659
655
|
|
660
|
-
/*
|
661
|
-
* call-seq:
|
662
|
-
* term_doc_enum.doc -> doc_id
|
663
|
-
*
|
664
|
-
* Returns the current document number pointed to by the +term_doc_enum+.
|
656
|
+
/* Returns the current document number pointed to by the +term_doc_enum+.
|
665
657
|
*/
|
666
|
-
static VALUE
|
658
|
+
static VALUE frb_tde_doc_num(VALUE self) {
|
667
659
|
FrtTermDocEnum *tde = (FrtTermDocEnum *)DATA_PTR(self);
|
668
660
|
return INT2FIX(tde->doc_num(tde));
|
669
661
|
}
|
670
662
|
|
671
663
|
/*
|
672
|
-
* call-seq:
|
673
|
-
* term_doc_enum.doc -> doc_id
|
674
|
-
*
|
675
664
|
* Returns the frequency of the current document pointed to by the
|
676
665
|
* +term_doc_enum+.
|
677
666
|
*/
|
@@ -680,11 +669,7 @@ static VALUE frb_tde_freq(VALUE self) {
|
|
680
669
|
return INT2FIX(tde->freq(tde));
|
681
670
|
}
|
682
671
|
|
683
|
-
/*
|
684
|
-
* call-seq:
|
685
|
-
* term_doc_enum.doc -> doc_id
|
686
|
-
*
|
687
|
-
* Move forward to the next document in the enumeration. Returns +true+ if
|
672
|
+
/* Move forward to the next document in the enumeration. Returns +true+ if
|
688
673
|
* there is another document or +false+ otherwise.
|
689
674
|
*/
|
690
675
|
static VALUE frb_tde_next(VALUE self) {
|
@@ -692,11 +677,7 @@ static VALUE frb_tde_next(VALUE self) {
|
|
692
677
|
return tde->next(tde) ? Qtrue : Qfalse;
|
693
678
|
}
|
694
679
|
|
695
|
-
/*
|
696
|
-
* call-seq:
|
697
|
-
* term_doc_enum.doc -> doc_id
|
698
|
-
*
|
699
|
-
* Move forward to the next document in the enumeration. Returns +true+ if
|
680
|
+
/* Move forward to the next document in the enumeration. Returns +true+ if
|
700
681
|
* there is another document or +false+ otherwise.
|
701
682
|
*/
|
702
683
|
static VALUE frb_tde_next_position(VALUE self) {
|
@@ -713,7 +694,7 @@ static VALUE frb_tde_next_position(VALUE self) {
|
|
713
694
|
|
714
695
|
/*
|
715
696
|
* call-seq:
|
716
|
-
* term_doc_enum.each {|
|
697
|
+
* term_doc_enum.each {|doc_num, freq| do_something() } -> doc_count
|
717
698
|
*
|
718
699
|
* Iterate through the documents and document frequencies in the
|
719
700
|
* +term_doc_enum+.
|
@@ -777,16 +758,13 @@ static VALUE frb_tde_to_json(int argc, VALUE *argv, VALUE self) {
|
|
777
758
|
if (do_positions) {
|
778
759
|
if (argc == 0) {
|
779
760
|
format = "{\"document\":%d,\"frequency\":%d,\"positions\":[";
|
780
|
-
}
|
781
|
-
else {
|
761
|
+
} else {
|
782
762
|
format = "[%d,%d,[";
|
783
763
|
}
|
784
|
-
}
|
785
|
-
else {
|
764
|
+
} else {
|
786
765
|
if (argc == 0) {
|
787
766
|
format = "{\"document\":%d,\"frequency\":%d},";
|
788
|
-
}
|
789
|
-
else {
|
767
|
+
} else {
|
790
768
|
format = "[%d,%d],";
|
791
769
|
}
|
792
770
|
}
|
@@ -828,8 +806,8 @@ static VALUE frb_tde_to_json(int argc, VALUE *argv, VALUE self) {
|
|
828
806
|
* used within the each method. For example, to print the terms documents and
|
829
807
|
* positions;
|
830
808
|
*
|
831
|
-
* tde.each do |
|
832
|
-
* puts "term appeared #{freq} times in document #{
|
809
|
+
* tde.each do |doc_num, freq|
|
810
|
+
* puts "term appeared #{freq} times in document #{doc_num}:"
|
833
811
|
* positions = []
|
834
812
|
* tde.each_position {|pos| positions << pos}
|
835
813
|
* puts " #{positions.join(', ')}"
|
@@ -1039,6 +1017,7 @@ static VALUE frb_iw_init(int argc, VALUE *argv, VALUE self) {
|
|
1039
1017
|
StringValue(rval);
|
1040
1018
|
frb_create_dir(rval);
|
1041
1019
|
store = frt_open_mdbx_store(rs2s(rval));
|
1020
|
+
store->create_folder(store, segm_idx_name);
|
1042
1021
|
}
|
1043
1022
|
|
1044
1023
|
if ((rval = rb_hash_aref(roptions, sym_analyzer)) != Qnil) {
|
@@ -1060,8 +1039,9 @@ static VALUE frb_iw_init(int argc, VALUE *argv, VALUE self) {
|
|
1060
1039
|
}
|
1061
1040
|
if (NULL == store) {
|
1062
1041
|
store = frt_open_ram_store(NULL);
|
1042
|
+
store->create_folder(store, segm_idx_name);
|
1063
1043
|
}
|
1064
|
-
if (!create && create_if_missing && !store->exists(store, "segments")) {
|
1044
|
+
if (!create && create_if_missing && !store->exists(store, segm_idx_name, "segments")) {
|
1065
1045
|
create = true;
|
1066
1046
|
}
|
1067
1047
|
if (create) {
|
@@ -1130,8 +1110,7 @@ frb_iw_get_doc_count(VALUE self) {
|
|
1130
1110
|
}
|
1131
1111
|
|
1132
1112
|
static int
|
1133
|
-
frb_hash_to_doc_i(VALUE key, VALUE value, VALUE arg)
|
1134
|
-
{
|
1113
|
+
frb_hash_to_doc_i(VALUE key, VALUE value, VALUE arg) {
|
1135
1114
|
if (key == Qundef) {
|
1136
1115
|
return ST_CONTINUE;
|
1137
1116
|
} else {
|
@@ -1151,20 +1130,18 @@ frb_hash_to_doc_i(VALUE key, VALUE value, VALUE arg)
|
|
1151
1130
|
case T_ARRAY:
|
1152
1131
|
{
|
1153
1132
|
int i;
|
1154
|
-
df->destroy_data = true;
|
1155
1133
|
for (i = 0; i < RARRAY_LEN(value); i++) {
|
1156
1134
|
val = rb_obj_as_string(RARRAY_PTR(value)[i]);
|
1157
|
-
frt_df_add_data_len(df,
|
1135
|
+
frt_df_add_data_len(df, RSTRING_PTR(val), RSTRING_LEN(val), rb_enc_get(val));
|
1158
1136
|
}
|
1159
1137
|
}
|
1160
1138
|
break;
|
1161
1139
|
case T_STRING:
|
1162
|
-
frt_df_add_data_len(df,
|
1140
|
+
frt_df_add_data_len(df, RSTRING_PTR(value), RSTRING_LEN(value), rb_enc_get(value));
|
1163
1141
|
break;
|
1164
1142
|
default:
|
1165
1143
|
val = rb_obj_as_string(value);
|
1166
|
-
df
|
1167
|
-
frt_df_add_data_len(df, rstrdup(val), RSTRING_LEN(val), rb_enc_get(val));
|
1144
|
+
frt_df_add_data_len(df, RSTRING_PTR(val), RSTRING_LEN(val), rb_enc_get(val));
|
1168
1145
|
break;
|
1169
1146
|
}
|
1170
1147
|
FRT_TRY
|
@@ -1181,8 +1158,7 @@ frb_hash_to_doc_i(VALUE key, VALUE value, VALUE arg)
|
|
1181
1158
|
}
|
1182
1159
|
|
1183
1160
|
static FrtDocument *
|
1184
|
-
frb_get_doc(VALUE rdoc)
|
1185
|
-
{
|
1161
|
+
frb_get_doc(VALUE rdoc) {
|
1186
1162
|
VALUE val;
|
1187
1163
|
FrtDocument *doc = frt_doc_new();
|
1188
1164
|
FrtDocField *df;
|
@@ -1199,27 +1175,25 @@ frb_get_doc(VALUE rdoc)
|
|
1199
1175
|
{
|
1200
1176
|
int i;
|
1201
1177
|
df = frt_df_new(fsym_content);
|
1202
|
-
df->destroy_data = true;
|
1203
1178
|
for (i = 0; i < RARRAY_LEN(rdoc); i++) {
|
1204
1179
|
val = rb_obj_as_string(RARRAY_PTR(rdoc)[i]);
|
1205
|
-
frt_df_add_data_len(df,
|
1180
|
+
frt_df_add_data_len(df, RSTRING_PTR(val), RSTRING_LEN(val), rb_enc_get(val));
|
1206
1181
|
}
|
1207
1182
|
frt_doc_add_field(doc, df);
|
1208
1183
|
}
|
1209
1184
|
break;
|
1210
1185
|
case T_SYMBOL:
|
1211
1186
|
/* TODO: clean up this ugly cast */
|
1212
|
-
df = frt_df_add_data(frt_df_new(fsym_content),
|
1187
|
+
df = frt_df_add_data(frt_df_new(fsym_content), rb_id2name(SYM2ID(rdoc)), rb_enc_get(rdoc));
|
1213
1188
|
frt_doc_add_field(doc, df);
|
1214
1189
|
break;
|
1215
1190
|
case T_STRING:
|
1216
|
-
df = frt_df_add_data_len(frt_df_new(fsym_content),
|
1191
|
+
df = frt_df_add_data_len(frt_df_new(fsym_content), RSTRING_PTR(rdoc), RSTRING_LEN(rdoc), rb_enc_get(rdoc));
|
1217
1192
|
frt_doc_add_field(doc, df);
|
1218
1193
|
break;
|
1219
1194
|
default:
|
1220
1195
|
val = rb_obj_as_string(rdoc);
|
1221
|
-
df = frt_df_add_data_len(frt_df_new(fsym_content),
|
1222
|
-
df->destroy_data = true;
|
1196
|
+
df = frt_df_add_data_len(frt_df_new(fsym_content), RSTRING_PTR(val), RSTRING_LEN(val), rb_enc_get(val));
|
1223
1197
|
frt_doc_add_field(doc, df);
|
1224
1198
|
break;
|
1225
1199
|
}
|
@@ -1439,8 +1413,7 @@ static VALUE frb_iw_field_infos(VALUE self) {
|
|
1439
1413
|
* to use the same analyzer in a QueryParser.
|
1440
1414
|
*/
|
1441
1415
|
static VALUE
|
1442
|
-
frb_iw_get_analyzer(VALUE self)
|
1443
|
-
{
|
1416
|
+
frb_iw_get_analyzer(VALUE self) {
|
1444
1417
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1445
1418
|
return frb_get_analyzer(iw->analyzer);
|
1446
1419
|
}
|
@@ -1454,8 +1427,7 @@ frb_iw_get_analyzer(VALUE self)
|
|
1454
1427
|
* same analyzer will be used for all documents during search.
|
1455
1428
|
*/
|
1456
1429
|
static VALUE
|
1457
|
-
frb_iw_set_analyzer(VALUE self, VALUE ranalyzer)
|
1458
|
-
{
|
1430
|
+
frb_iw_set_analyzer(VALUE self, VALUE ranalyzer) {
|
1459
1431
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1460
1432
|
|
1461
1433
|
frt_a_deref(iw->analyzer);
|
@@ -1470,8 +1442,7 @@ frb_iw_set_analyzer(VALUE self, VALUE ranalyzer)
|
|
1470
1442
|
* Returns the current version of the index writer.
|
1471
1443
|
*/
|
1472
1444
|
static VALUE
|
1473
|
-
frb_iw_version(VALUE self)
|
1474
|
-
{
|
1445
|
+
frb_iw_version(VALUE self) {
|
1475
1446
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1476
1447
|
return ULL2NUM(iw->sis->version);
|
1477
1448
|
}
|
@@ -1483,8 +1454,7 @@ frb_iw_version(VALUE self)
|
|
1483
1454
|
* Return the current value of chunk_size
|
1484
1455
|
*/
|
1485
1456
|
static VALUE
|
1486
|
-
frb_iw_get_chunk_size(VALUE self)
|
1487
|
-
{
|
1457
|
+
frb_iw_get_chunk_size(VALUE self) {
|
1488
1458
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1489
1459
|
return INT2FIX(iw->config.chunk_size);
|
1490
1460
|
}
|
@@ -1496,8 +1466,7 @@ frb_iw_get_chunk_size(VALUE self)
|
|
1496
1466
|
* Set the chunk_size parameter
|
1497
1467
|
*/
|
1498
1468
|
static VALUE
|
1499
|
-
frb_iw_set_chunk_size(VALUE self, VALUE rval)
|
1500
|
-
{
|
1469
|
+
frb_iw_set_chunk_size(VALUE self, VALUE rval) {
|
1501
1470
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1502
1471
|
iw->config.chunk_size = FIX2INT(rval);
|
1503
1472
|
return rval;
|
@@ -1510,8 +1479,7 @@ frb_iw_set_chunk_size(VALUE self, VALUE rval)
|
|
1510
1479
|
* Return the current value of max_buffer_memory
|
1511
1480
|
*/
|
1512
1481
|
static VALUE
|
1513
|
-
frb_iw_get_max_buffer_memory(VALUE self)
|
1514
|
-
{
|
1482
|
+
frb_iw_get_max_buffer_memory(VALUE self) {
|
1515
1483
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1516
1484
|
return INT2FIX(iw->config.max_buffer_memory);
|
1517
1485
|
}
|
@@ -1523,8 +1491,7 @@ frb_iw_get_max_buffer_memory(VALUE self)
|
|
1523
1491
|
* Set the max_buffer_memory parameter
|
1524
1492
|
*/
|
1525
1493
|
static VALUE
|
1526
|
-
frb_iw_set_max_buffer_memory(VALUE self, VALUE rval)
|
1527
|
-
{
|
1494
|
+
frb_iw_set_max_buffer_memory(VALUE self, VALUE rval) {
|
1528
1495
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1529
1496
|
iw->config.max_buffer_memory = FIX2INT(rval);
|
1530
1497
|
return rval;
|
@@ -1537,8 +1504,7 @@ frb_iw_set_max_buffer_memory(VALUE self, VALUE rval)
|
|
1537
1504
|
* Return the current value of term_index_interval
|
1538
1505
|
*/
|
1539
1506
|
static VALUE
|
1540
|
-
frb_iw_get_index_interval(VALUE self)
|
1541
|
-
{
|
1507
|
+
frb_iw_get_index_interval(VALUE self) {
|
1542
1508
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1543
1509
|
return INT2FIX(iw->config.index_interval);
|
1544
1510
|
}
|
@@ -1550,8 +1516,7 @@ frb_iw_get_index_interval(VALUE self)
|
|
1550
1516
|
* Set the term_index_interval parameter
|
1551
1517
|
*/
|
1552
1518
|
static VALUE
|
1553
|
-
frb_iw_set_index_interval(VALUE self, VALUE rval)
|
1554
|
-
{
|
1519
|
+
frb_iw_set_index_interval(VALUE self, VALUE rval) {
|
1555
1520
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1556
1521
|
iw->config.index_interval = FIX2INT(rval);
|
1557
1522
|
return rval;
|
@@ -1564,8 +1529,7 @@ frb_iw_set_index_interval(VALUE self, VALUE rval)
|
|
1564
1529
|
* Return the current value of doc_skip_interval
|
1565
1530
|
*/
|
1566
1531
|
static VALUE
|
1567
|
-
frb_iw_get_skip_interval(VALUE self)
|
1568
|
-
{
|
1532
|
+
frb_iw_get_skip_interval(VALUE self) {
|
1569
1533
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1570
1534
|
return INT2FIX(iw->config.skip_interval);
|
1571
1535
|
}
|
@@ -1577,8 +1541,7 @@ frb_iw_get_skip_interval(VALUE self)
|
|
1577
1541
|
* Set the doc_skip_interval parameter
|
1578
1542
|
*/
|
1579
1543
|
static VALUE
|
1580
|
-
frb_iw_set_skip_interval(VALUE self, VALUE rval)
|
1581
|
-
{
|
1544
|
+
frb_iw_set_skip_interval(VALUE self, VALUE rval) {
|
1582
1545
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1583
1546
|
iw->config.skip_interval = FIX2INT(rval);
|
1584
1547
|
return rval;
|
@@ -1591,8 +1554,7 @@ frb_iw_set_skip_interval(VALUE self, VALUE rval)
|
|
1591
1554
|
* Return the current value of merge_factor
|
1592
1555
|
*/
|
1593
1556
|
static VALUE
|
1594
|
-
frb_iw_get_merge_factor(VALUE self)
|
1595
|
-
{
|
1557
|
+
frb_iw_get_merge_factor(VALUE self) {
|
1596
1558
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1597
1559
|
return INT2FIX(iw->config.merge_factor);
|
1598
1560
|
}
|
@@ -1604,8 +1566,7 @@ frb_iw_get_merge_factor(VALUE self)
|
|
1604
1566
|
* Set the merge_factor parameter
|
1605
1567
|
*/
|
1606
1568
|
static VALUE
|
1607
|
-
frb_iw_set_merge_factor(VALUE self, VALUE rval)
|
1608
|
-
{
|
1569
|
+
frb_iw_set_merge_factor(VALUE self, VALUE rval) {
|
1609
1570
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1610
1571
|
iw->config.merge_factor = FIX2INT(rval);
|
1611
1572
|
return rval;
|
@@ -1618,8 +1579,7 @@ frb_iw_set_merge_factor(VALUE self, VALUE rval)
|
|
1618
1579
|
* Return the current value of max_buffered_docs
|
1619
1580
|
*/
|
1620
1581
|
static VALUE
|
1621
|
-
frb_iw_get_max_buffered_docs(VALUE self)
|
1622
|
-
{
|
1582
|
+
frb_iw_get_max_buffered_docs(VALUE self) {
|
1623
1583
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1624
1584
|
return INT2FIX(iw->config.max_buffered_docs);
|
1625
1585
|
}
|
@@ -1631,8 +1591,7 @@ frb_iw_get_max_buffered_docs(VALUE self)
|
|
1631
1591
|
* Set the max_buffered_docs parameter
|
1632
1592
|
*/
|
1633
1593
|
static VALUE
|
1634
|
-
frb_iw_set_max_buffered_docs(VALUE self, VALUE rval)
|
1635
|
-
{
|
1594
|
+
frb_iw_set_max_buffered_docs(VALUE self, VALUE rval) {
|
1636
1595
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1637
1596
|
iw->config.max_buffered_docs = FIX2INT(rval);
|
1638
1597
|
return rval;
|
@@ -1645,8 +1604,7 @@ frb_iw_set_max_buffered_docs(VALUE self, VALUE rval)
|
|
1645
1604
|
* Return the current value of max_merge_docs
|
1646
1605
|
*/
|
1647
1606
|
static VALUE
|
1648
|
-
frb_iw_get_max_merge_docs(VALUE self)
|
1649
|
-
{
|
1607
|
+
frb_iw_get_max_merge_docs(VALUE self) {
|
1650
1608
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1651
1609
|
return INT2FIX(iw->config.max_merge_docs);
|
1652
1610
|
}
|
@@ -1658,8 +1616,7 @@ frb_iw_get_max_merge_docs(VALUE self)
|
|
1658
1616
|
* Set the max_merge_docs parameter
|
1659
1617
|
*/
|
1660
1618
|
static VALUE
|
1661
|
-
frb_iw_set_max_merge_docs(VALUE self, VALUE rval)
|
1662
|
-
{
|
1619
|
+
frb_iw_set_max_merge_docs(VALUE self, VALUE rval) {
|
1663
1620
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1664
1621
|
iw->config.max_merge_docs = FIX2INT(rval);
|
1665
1622
|
return rval;
|
@@ -1672,8 +1629,7 @@ frb_iw_set_max_merge_docs(VALUE self, VALUE rval)
|
|
1672
1629
|
* Return the current value of max_field_length
|
1673
1630
|
*/
|
1674
1631
|
static VALUE
|
1675
|
-
frb_iw_get_max_field_length(VALUE self)
|
1676
|
-
{
|
1632
|
+
frb_iw_get_max_field_length(VALUE self) {
|
1677
1633
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1678
1634
|
return INT2FIX(iw->config.max_field_length);
|
1679
1635
|
}
|
@@ -1685,8 +1641,7 @@ frb_iw_get_max_field_length(VALUE self)
|
|
1685
1641
|
* Set the max_field_length parameter
|
1686
1642
|
*/
|
1687
1643
|
static VALUE
|
1688
|
-
frb_iw_set_max_field_length(VALUE self, VALUE rval)
|
1689
|
-
{
|
1644
|
+
frb_iw_set_max_field_length(VALUE self, VALUE rval) {
|
1690
1645
|
FrtIndexWriter *iw = (FrtIndexWriter *)DATA_PTR(self);
|
1691
1646
|
iw->config.max_field_length = FIX2INT(rval);
|
1692
1647
|
return rval;
|
@@ -1766,6 +1721,7 @@ static VALUE frb_ir_init(VALUE self, VALUE rdir) {
|
|
1766
1721
|
case T_STRING:
|
1767
1722
|
frb_create_dir(rdir);
|
1768
1723
|
store = frt_open_mdbx_store(rs2s(rdir));
|
1724
|
+
store->create_folder(store, segm_idx_name);
|
1769
1725
|
break;
|
1770
1726
|
default:
|
1771
1727
|
FRT_RAISE(FRT_ARG_ERROR, "%s isn't a valid directory "
|
@@ -1787,6 +1743,7 @@ static VALUE frb_ir_init(VALUE self, VALUE rdir) {
|
|
1787
1743
|
case T_STRING:
|
1788
1744
|
frb_create_dir(rdir);
|
1789
1745
|
store = frt_open_mdbx_store(rs2s(rdir));
|
1746
|
+
store->create_folder(store, segm_idx_name);
|
1790
1747
|
break;
|
1791
1748
|
default:
|
1792
1749
|
FRT_RAISE(FRT_ARG_ERROR, "%s isn't a valid directory argument. "
|
@@ -1824,20 +1781,20 @@ static VALUE frb_ir_init(VALUE self, VALUE rdir) {
|
|
1824
1781
|
|
1825
1782
|
/*
|
1826
1783
|
* call-seq:
|
1827
|
-
* index_reader.set_norm(
|
1784
|
+
* index_reader.set_norm(doc_num, field, val)
|
1828
1785
|
*
|
1829
|
-
* Expert: change the boost value for a +field+ in document at +
|
1786
|
+
* Expert: change the boost value for a +field+ in document at +doc_num+.
|
1830
1787
|
* +val+ should be an integer in the range 0..255 which corresponds to an
|
1831
1788
|
* encoded float value.
|
1832
1789
|
*/
|
1833
1790
|
static VALUE
|
1834
|
-
frb_ir_set_norm(VALUE self, VALUE
|
1791
|
+
frb_ir_set_norm(VALUE self, VALUE rdoc_num, VALUE rfield, VALUE rval) {
|
1835
1792
|
int ex_code = 0;
|
1836
1793
|
const char *msg = NULL;
|
1837
1794
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
1838
1795
|
|
1839
1796
|
FRT_TRY
|
1840
|
-
frt_ir_set_norm(ir, FIX2INT(
|
1797
|
+
frt_ir_set_norm(ir, FIX2INT(rdoc_num), frb_field(rfield), (frt_uchar)NUM2CHR(rval));
|
1841
1798
|
FRT_XCATCHALL
|
1842
1799
|
ex_code = xcontext.excode;
|
1843
1800
|
msg = xcontext.msg;
|
@@ -1860,13 +1817,12 @@ frb_ir_set_norm(VALUE self, VALUE rdoc_id, VALUE rfield, VALUE rval) {
|
|
1860
1817
|
* could have null bytes.
|
1861
1818
|
*/
|
1862
1819
|
static VALUE
|
1863
|
-
frb_ir_norms(VALUE self, VALUE rfield)
|
1864
|
-
{
|
1820
|
+
frb_ir_norms(VALUE self, VALUE rfield) {
|
1865
1821
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
1866
1822
|
frt_uchar *norms;
|
1867
1823
|
norms = frt_ir_get_norms(ir, frb_field(rfield));
|
1868
1824
|
if (norms) {
|
1869
|
-
return rb_str_new((char *)norms, ir->
|
1825
|
+
return rb_str_new((char *)norms, ir->max_doc_num(ir));
|
1870
1826
|
} else {
|
1871
1827
|
return Qnil;
|
1872
1828
|
}
|
@@ -1879,17 +1835,16 @@ frb_ir_norms(VALUE self, VALUE rfield)
|
|
1879
1835
|
* Expert: Get the norm values into a string +buffer+ starting at +offset+.
|
1880
1836
|
*/
|
1881
1837
|
static VALUE
|
1882
|
-
frb_ir_get_norms_into(VALUE self, VALUE rfield, VALUE rnorms, VALUE roffset)
|
1883
|
-
{
|
1838
|
+
frb_ir_get_norms_into(VALUE self, VALUE rfield, VALUE rnorms, VALUE roffset) {
|
1884
1839
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
1885
1840
|
int offset;
|
1886
1841
|
offset = FIX2INT(roffset);
|
1887
1842
|
Check_Type(rnorms, T_STRING);
|
1888
|
-
if (RSTRING_LEN(rnorms) < offset + ir->
|
1843
|
+
if (RSTRING_LEN(rnorms) < offset + ir->max_doc_num(ir)) {
|
1889
1844
|
rb_raise(rb_eArgError, "supplied a string of length:%ld to "
|
1890
1845
|
"IndexReader#get_norms_into but needed a string of length "
|
1891
1846
|
"offset:%d + maxdoc:%d",
|
1892
|
-
RSTRING_LEN(rnorms), offset, ir->
|
1847
|
+
RSTRING_LEN(rnorms), offset, ir->max_doc_num(ir));
|
1893
1848
|
}
|
1894
1849
|
|
1895
1850
|
frt_ir_get_norms_into(ir, frb_field(rfield),
|
@@ -1935,8 +1890,7 @@ frb_ir_commit(VALUE self) {
|
|
1935
1890
|
* errors.
|
1936
1891
|
*/
|
1937
1892
|
static VALUE
|
1938
|
-
frb_ir_close(VALUE self)
|
1939
|
-
{
|
1893
|
+
frb_ir_close(VALUE self) {
|
1940
1894
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
1941
1895
|
((struct RData *)(self))->data = NULL;
|
1942
1896
|
((struct RData *)(self))->dmark = NULL;
|
@@ -1953,28 +1907,26 @@ frb_ir_close(VALUE self)
|
|
1953
1907
|
* IndexReader or committed by any other IndexReader.
|
1954
1908
|
*/
|
1955
1909
|
static VALUE
|
1956
|
-
frb_ir_has_deletions(VALUE self)
|
1957
|
-
{
|
1910
|
+
frb_ir_has_deletions(VALUE self) {
|
1958
1911
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
1959
1912
|
return ir->has_deletions(ir) ? Qtrue : Qfalse;
|
1960
1913
|
}
|
1961
1914
|
|
1962
|
-
/*
|
1963
|
-
*
|
1964
|
-
* index_reader.delete(doc_id) -> index_reader
|
1915
|
+
/* call-seq:
|
1916
|
+
* index_reader.delete(doc_num) -> index_reader
|
1965
1917
|
*
|
1966
|
-
* Delete document referenced internally by document id +
|
1918
|
+
* Delete document referenced internally by document id +doc_num+. The
|
1967
1919
|
* document_id is the number used to reference documents in the index and is
|
1968
1920
|
* returned by search methods.
|
1969
1921
|
*/
|
1970
1922
|
static VALUE
|
1971
|
-
frb_ir_delete(VALUE self, VALUE
|
1923
|
+
frb_ir_delete(VALUE self, VALUE rdoc_num) {
|
1972
1924
|
int ex_code = 0;
|
1973
1925
|
const char *msg = NULL;
|
1974
1926
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
1975
1927
|
|
1976
1928
|
FRT_TRY
|
1977
|
-
frt_ir_delete_doc(ir, FIX2INT(
|
1929
|
+
frt_ir_delete_doc(ir, FIX2INT(rdoc_num));
|
1978
1930
|
FRT_XCATCHALL
|
1979
1931
|
ex_code = xcontext.excode;
|
1980
1932
|
msg = xcontext.msg;
|
@@ -1988,33 +1940,30 @@ frb_ir_delete(VALUE self, VALUE rdoc_id) {
|
|
1988
1940
|
return self;
|
1989
1941
|
}
|
1990
1942
|
|
1991
|
-
/*
|
1992
|
-
*
|
1993
|
-
* index_reader.deleted?(doc_id) -> bool
|
1943
|
+
/* call-seq:
|
1944
|
+
* index_reader.deleted?(doc_num) -> bool
|
1994
1945
|
*
|
1995
|
-
* Returns true if the document at +
|
1946
|
+
* Returns true if the document at +doc_num+ has been deleted.
|
1996
1947
|
*/
|
1997
1948
|
static VALUE
|
1998
|
-
frb_ir_is_deleted(VALUE self, VALUE
|
1999
|
-
{
|
1949
|
+
frb_ir_is_deleted(VALUE self, VALUE rdoc_num) {
|
2000
1950
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2001
|
-
return ir->is_deleted(ir, FIX2INT(
|
1951
|
+
return ir->is_deleted(ir, FIX2INT(rdoc_num)) ? Qtrue : Qfalse;
|
2002
1952
|
}
|
2003
1953
|
|
2004
|
-
/*
|
2005
|
-
* call-seq:
|
1954
|
+
/* call-seq:
|
2006
1955
|
* index_reader.max_doc -> number
|
2007
1956
|
*
|
2008
1957
|
* Returns 1 + the maximum document id in the index. It is the
|
2009
1958
|
* document_id that will be used by the next document added to the index. If
|
2010
1959
|
* there are no deletions, this number also refers to the number of documents
|
2011
1960
|
* in the index.
|
1961
|
+
* TODO: Rename to next_doc_num?
|
2012
1962
|
*/
|
2013
1963
|
static VALUE
|
2014
|
-
|
2015
|
-
{
|
1964
|
+
frb_ir_max_doc_num(VALUE self) {
|
2016
1965
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2017
|
-
return INT2FIX(ir->
|
1966
|
+
return INT2FIX(ir->max_doc_num(ir));
|
2018
1967
|
}
|
2019
1968
|
|
2020
1969
|
/*
|
@@ -2026,8 +1975,7 @@ frb_ir_max_doc(VALUE self)
|
|
2026
1975
|
* deleted from the index.
|
2027
1976
|
*/
|
2028
1977
|
static VALUE
|
2029
|
-
frb_ir_num_docs(VALUE self)
|
2030
|
-
{
|
1978
|
+
frb_ir_num_docs(VALUE self) {
|
2031
1979
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2032
1980
|
return INT2FIX(ir->num_docs(ir));
|
2033
1981
|
}
|
@@ -2042,31 +1990,43 @@ frb_ir_num_docs(VALUE self)
|
|
2042
1990
|
* effect on these documents.
|
2043
1991
|
*/
|
2044
1992
|
static VALUE
|
2045
|
-
frb_ir_undelete_all(VALUE self)
|
2046
|
-
{
|
1993
|
+
frb_ir_undelete_all(VALUE self) {
|
2047
1994
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2048
1995
|
frt_ir_undelete_all(ir);
|
2049
1996
|
return self;
|
2050
1997
|
}
|
2051
1998
|
|
2052
1999
|
static VALUE
|
2053
|
-
frb_get_doc_range(FrtIndexReader *ir, int pos, int len, int max)
|
2054
|
-
|
2000
|
+
frb_get_doc_range(FrtIndexReader *ir, int pos, int len, int max) {
|
2001
|
+
int ex_code = 0;
|
2002
|
+
const char *msg = NULL;
|
2055
2003
|
VALUE ary;
|
2056
2004
|
int i;
|
2057
2005
|
max = FRT_MIN(max, pos+len);
|
2058
2006
|
len = max - pos;
|
2059
2007
|
ary = rb_ary_new2(len);
|
2060
|
-
|
2061
|
-
|
2008
|
+
|
2009
|
+
FRT_TRY
|
2010
|
+
for (i = 0; i < len; i++) {
|
2011
|
+
rb_ary_store(ary, i, frb_get_lazy_doc(ir->get_lazy_doc(ir, i + pos)));
|
2012
|
+
}
|
2013
|
+
FRT_XCATCHALL
|
2014
|
+
ex_code = xcontext.excode;
|
2015
|
+
msg = xcontext.msg;
|
2016
|
+
FRT_HANDLED();
|
2017
|
+
FRT_XENDTRY
|
2018
|
+
|
2019
|
+
if (ex_code && msg) {
|
2020
|
+
frb_raise(ex_code, msg);
|
2062
2021
|
}
|
2022
|
+
|
2063
2023
|
return ary;
|
2064
2024
|
}
|
2065
2025
|
|
2066
2026
|
/*
|
2067
2027
|
* call-seq:
|
2068
|
-
* index_reader.get_document(
|
2069
|
-
* index_reader[
|
2028
|
+
* index_reader.get_document(doc_num) -> LazyDoc
|
2029
|
+
* index_reader[doc_num] -> LazyDoc
|
2070
2030
|
*
|
2071
2031
|
* Retrieve a document from the index. See LazyDoc for more details on the
|
2072
2032
|
* document returned. Documents are referenced internally by document ids
|
@@ -2077,7 +2037,7 @@ frb_ir_get_doc(int argc, VALUE *argv, VALUE self) {
|
|
2077
2037
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2078
2038
|
VALUE arg1, arg2;
|
2079
2039
|
long pos, len;
|
2080
|
-
long max = ir->
|
2040
|
+
long max = ir->max_doc_num(ir);
|
2081
2041
|
rb_scan_args(argc, argv, "11", &arg1, &arg2);
|
2082
2042
|
if (argc == 1) {
|
2083
2043
|
if (FIXNUM_P(arg1)) {
|
@@ -2087,7 +2047,24 @@ frb_ir_get_doc(int argc, VALUE *argv, VALUE self) {
|
|
2087
2047
|
rb_raise(rb_eArgError, "index %ld is out of range [%d..%ld] for "
|
2088
2048
|
"IndexReader#[]", pos, 0, max);
|
2089
2049
|
}
|
2090
|
-
|
2050
|
+
|
2051
|
+
int ex_code = 0;
|
2052
|
+
const char *msg = NULL;
|
2053
|
+
VALUE ld = Qnil;
|
2054
|
+
|
2055
|
+
FRT_TRY
|
2056
|
+
ld = frb_get_lazy_doc(ir->get_lazy_doc(ir, pos));
|
2057
|
+
FRT_XCATCHALL
|
2058
|
+
ex_code = xcontext.excode;
|
2059
|
+
msg = xcontext.msg;
|
2060
|
+
FRT_HANDLED();
|
2061
|
+
FRT_XENDTRY
|
2062
|
+
|
2063
|
+
if (ex_code && msg) {
|
2064
|
+
frb_raise(ex_code, msg);
|
2065
|
+
}
|
2066
|
+
|
2067
|
+
return ld;
|
2091
2068
|
}
|
2092
2069
|
|
2093
2070
|
/* check if idx is Range */
|
@@ -2118,55 +2095,48 @@ frb_ir_get_doc(int argc, VALUE *argv, VALUE self) {
|
|
2118
2095
|
* index to search the latest documents added to the index.
|
2119
2096
|
*/
|
2120
2097
|
static VALUE
|
2121
|
-
frb_ir_is_latest(VALUE self)
|
2122
|
-
{
|
2098
|
+
frb_ir_is_latest(VALUE self) {
|
2123
2099
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2124
2100
|
return frt_ir_is_latest(ir) ? Qtrue : Qfalse;
|
2125
2101
|
}
|
2126
2102
|
|
2127
|
-
/*
|
2128
|
-
*
|
2129
|
-
* index_reader.term_vector(doc_id, field) -> TermVector
|
2103
|
+
/* call-seq:
|
2104
|
+
* index_reader.term_vector(doc_num, field) -> TermVector
|
2130
2105
|
*
|
2131
|
-
* Return the TermVector for the field +field+ in the document at +
|
2106
|
+
* Return the TermVector for the field +field+ in the document at +doc_num+ in
|
2132
2107
|
* the index. Return nil if no such term_vector exists. See TermVector.
|
2133
2108
|
*/
|
2134
2109
|
static VALUE
|
2135
|
-
frb_ir_term_vector(VALUE self, VALUE
|
2136
|
-
{
|
2110
|
+
frb_ir_term_vector(VALUE self, VALUE rdoc_num, VALUE rfield) {
|
2137
2111
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2138
2112
|
FrtTermVector *tv;
|
2139
2113
|
VALUE rtv;
|
2140
|
-
tv = ir->term_vector(ir, FIX2INT(
|
2114
|
+
tv = ir->term_vector(ir, FIX2INT(rdoc_num), frb_field(rfield));
|
2141
2115
|
if (tv) {
|
2142
2116
|
rtv = frb_get_tv(tv);
|
2143
2117
|
frt_tv_destroy(tv);
|
2144
2118
|
return rtv;
|
2145
|
-
}
|
2146
|
-
else {
|
2119
|
+
} else {
|
2147
2120
|
return Qnil;
|
2148
2121
|
}
|
2149
2122
|
}
|
2150
2123
|
|
2151
2124
|
static void
|
2152
|
-
frb_add_each_tv(void *key, void *value, void *rtvs)
|
2153
|
-
{
|
2125
|
+
frb_add_each_tv(void *key, void *value, void *rtvs) {
|
2154
2126
|
rb_hash_aset((VALUE)rtvs, ID2SYM((ID)key), frb_get_tv(value));
|
2155
2127
|
}
|
2156
2128
|
|
2157
|
-
/*
|
2158
|
-
*
|
2159
|
-
* index_reader.term_vectors(doc_id) -> hash of TermVector
|
2129
|
+
/* call-seq:
|
2130
|
+
* index_reader.term_vectors(doc_num) -> hash of TermVector
|
2160
2131
|
*
|
2161
|
-
* Return the TermVectors for the document at +
|
2132
|
+
* Return the TermVectors for the document at +doc_num+ in the index. The
|
2162
2133
|
* value returned is a hash of the TermVectors for each field in the document
|
2163
2134
|
* and they are referenced by field names (as symbols).
|
2164
2135
|
*/
|
2165
2136
|
static VALUE
|
2166
|
-
frb_ir_term_vectors(VALUE self, VALUE
|
2167
|
-
{
|
2137
|
+
frb_ir_term_vectors(VALUE self, VALUE rdoc_num) {
|
2168
2138
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2169
|
-
FrtHash *tvs = ir->term_vectors(ir, FIX2INT(
|
2139
|
+
FrtHash *tvs = ir->term_vectors(ir, FIX2INT(rdoc_num));
|
2170
2140
|
VALUE rtvs = rb_hash_new();
|
2171
2141
|
frt_h_each(tvs, &frb_add_each_tv, (void *)rtvs);
|
2172
2142
|
frt_h_destroy(tvs);
|
@@ -2182,8 +2152,7 @@ frb_ir_term_vectors(VALUE self, VALUE rdoc_id)
|
|
2182
2152
|
* See TermDocEnum for more info.
|
2183
2153
|
*/
|
2184
2154
|
static VALUE
|
2185
|
-
frb_ir_term_docs(VALUE self)
|
2186
|
-
{
|
2155
|
+
frb_ir_term_docs(VALUE self) {
|
2187
2156
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2188
2157
|
return frb_get_tde(self, ir->term_docs(ir));
|
2189
2158
|
}
|
@@ -2196,8 +2165,7 @@ frb_ir_term_docs(VALUE self)
|
|
2196
2165
|
* term +term+ in the field +field+. See TermDocEnum for more info.
|
2197
2166
|
*/
|
2198
2167
|
static VALUE
|
2199
|
-
frb_ir_term_docs_for(VALUE self, VALUE rfield, VALUE rterm)
|
2200
|
-
{
|
2168
|
+
frb_ir_term_docs_for(VALUE self, VALUE rfield, VALUE rterm) {
|
2201
2169
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2202
2170
|
return frb_get_tde(self, ir_term_docs_for(ir,
|
2203
2171
|
frb_field(rfield),
|
@@ -2213,8 +2181,7 @@ frb_ir_term_docs_for(VALUE self, VALUE rfield, VALUE rterm)
|
|
2213
2181
|
* more info.
|
2214
2182
|
*/
|
2215
2183
|
static VALUE
|
2216
|
-
frb_ir_term_positions(VALUE self)
|
2217
|
-
{
|
2184
|
+
frb_ir_term_positions(VALUE self) {
|
2218
2185
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2219
2186
|
return frb_get_tde(self, ir->term_positions(ir));
|
2220
2187
|
}
|
@@ -2228,8 +2195,7 @@ frb_ir_term_positions(VALUE self)
|
|
2228
2195
|
* TermDocEnum for more info.
|
2229
2196
|
*/
|
2230
2197
|
static VALUE
|
2231
|
-
frb_ir_t_pos_for(VALUE self, VALUE rfield, VALUE rterm)
|
2232
|
-
{
|
2198
|
+
frb_ir_t_pos_for(VALUE self, VALUE rfield, VALUE rterm) {
|
2233
2199
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2234
2200
|
return frb_get_tde(self, frt_ir_term_positions_for(ir,
|
2235
2201
|
frb_field(rfield),
|
@@ -2244,8 +2210,7 @@ frb_ir_t_pos_for(VALUE self, VALUE rfield, VALUE rterm)
|
|
2244
2210
|
* field +field+.
|
2245
2211
|
*/
|
2246
2212
|
static VALUE
|
2247
|
-
frb_ir_doc_freq(VALUE self, VALUE rfield, VALUE rterm)
|
2248
|
-
{
|
2213
|
+
frb_ir_doc_freq(VALUE self, VALUE rfield, VALUE rterm) {
|
2249
2214
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2250
2215
|
return INT2FIX(frt_ir_doc_freq(ir,
|
2251
2216
|
frb_field(rfield),
|
@@ -2260,8 +2225,7 @@ frb_ir_doc_freq(VALUE self, VALUE rfield, VALUE rterm)
|
|
2260
2225
|
* terms in the field +field+ in the index.
|
2261
2226
|
*/
|
2262
2227
|
static VALUE
|
2263
|
-
frb_ir_terms(VALUE self, VALUE rfield)
|
2264
|
-
{
|
2228
|
+
frb_ir_terms(VALUE self, VALUE rfield) {
|
2265
2229
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2266
2230
|
return frb_get_te(self, frt_ir_terms(ir, frb_field(rfield)));
|
2267
2231
|
}
|
@@ -2274,8 +2238,7 @@ frb_ir_terms(VALUE self, VALUE rfield)
|
|
2274
2238
|
* at term +term+.
|
2275
2239
|
*/
|
2276
2240
|
static VALUE
|
2277
|
-
frb_ir_terms_from(VALUE self, VALUE rfield, VALUE rterm)
|
2278
|
-
{
|
2241
|
+
frb_ir_terms_from(VALUE self, VALUE rfield, VALUE rterm) {
|
2279
2242
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2280
2243
|
return frb_get_te(self, frt_ir_terms_from(ir,
|
2281
2244
|
frb_field(rfield),
|
@@ -2289,8 +2252,7 @@ frb_ir_terms_from(VALUE self, VALUE rfield, VALUE rterm)
|
|
2289
2252
|
* Same return a count of the number of terms in the field
|
2290
2253
|
*/
|
2291
2254
|
static VALUE
|
2292
|
-
frb_ir_term_count(VALUE self, VALUE rfield)
|
2293
|
-
{
|
2255
|
+
frb_ir_term_count(VALUE self, VALUE rfield) {
|
2294
2256
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2295
2257
|
FrtTermEnum *te = frt_ir_terms(ir, frb_field(rfield));
|
2296
2258
|
int count = 0;
|
@@ -2311,8 +2273,7 @@ frb_ir_term_count(VALUE self, VALUE rfield)
|
|
2311
2273
|
* gathered from the FieldInfos object.
|
2312
2274
|
*/
|
2313
2275
|
static VALUE
|
2314
|
-
frb_ir_fields(VALUE self)
|
2315
|
-
{
|
2276
|
+
frb_ir_fields(VALUE self) {
|
2316
2277
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2317
2278
|
FrtFieldInfos *fis = ir->fis;
|
2318
2279
|
VALUE rfield_names = rb_ary_new();
|
@@ -2345,8 +2306,7 @@ static VALUE frb_ir_field_infos(VALUE self) {
|
|
2345
2306
|
* of field names can also be gathered from the FieldInfos object.
|
2346
2307
|
*/
|
2347
2308
|
static VALUE
|
2348
|
-
frb_ir_tk_fields(VALUE self)
|
2349
|
-
{
|
2309
|
+
frb_ir_tk_fields(VALUE self) {
|
2350
2310
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2351
2311
|
FrtFieldInfos *fis = ir->fis;
|
2352
2312
|
VALUE rfield_names = rb_ary_new();
|
@@ -2377,14 +2337,28 @@ static VALUE frb_ir_to_enum(VALUE self) {
|
|
2377
2337
|
static VALUE frb_ir_each(VALUE self) {
|
2378
2338
|
FrtIndexReader *ir = (FrtIndexReader *)DATA_PTR(self);
|
2379
2339
|
if (rb_block_given_p()) {
|
2340
|
+
int ex_code = 0;
|
2341
|
+
const char *msg = NULL;
|
2380
2342
|
long i;
|
2381
|
-
|
2343
|
+
int max_doc_num = ir->max_doc_num(ir);
|
2382
2344
|
VALUE rld;
|
2383
|
-
|
2384
|
-
|
2385
|
-
|
2386
|
-
|
2345
|
+
|
2346
|
+
FRT_TRY
|
2347
|
+
for (i = 0; i < max_doc_num; i++) {
|
2348
|
+
if (ir->is_deleted(ir, i)) continue;
|
2349
|
+
rld = frb_get_lazy_doc(ir->get_lazy_doc(ir, i));
|
2350
|
+
rb_yield(rld);
|
2351
|
+
}
|
2352
|
+
FRT_XCATCHALL
|
2353
|
+
ex_code = xcontext.excode;
|
2354
|
+
msg = xcontext.msg;
|
2355
|
+
FRT_HANDLED();
|
2356
|
+
FRT_XENDTRY
|
2357
|
+
|
2358
|
+
if (ex_code && msg) {
|
2359
|
+
frb_raise(ex_code, msg);
|
2387
2360
|
}
|
2361
|
+
|
2388
2362
|
return self;
|
2389
2363
|
} else {
|
2390
2364
|
return frb_ir_to_enum(self);
|
@@ -2511,8 +2485,8 @@ Init_TermEnum(void) {
|
|
2511
2485
|
*
|
2512
2486
|
* tde = index_reader.term_docs_for(:content, "fox")
|
2513
2487
|
*
|
2514
|
-
* tde.each do |
|
2515
|
-
* puts "fox appeared #{freq} times in document #{
|
2488
|
+
* tde.each do |doc_num, freq|
|
2489
|
+
* puts "fox appeared #{freq} times in document #{doc_num}:"
|
2516
2490
|
* positions = []
|
2517
2491
|
* tde.each_position {|pos| positions << pos}
|
2518
2492
|
* puts " #{positions.join(', ')}"
|
@@ -2537,7 +2511,7 @@ static void Init_TermDocEnum(void) {
|
|
2537
2511
|
rb_define_alloc_func(cTermDocEnum, frb_tde_alloc);
|
2538
2512
|
rb_define_method(cTermDocEnum, "seek", frb_tde_seek, 2);
|
2539
2513
|
rb_define_method(cTermDocEnum, "seek_term_enum", frb_tde_seek_te, 1);
|
2540
|
-
rb_define_method(cTermDocEnum, "
|
2514
|
+
rb_define_method(cTermDocEnum, "doc_num", frb_tde_doc_num, 0);
|
2541
2515
|
rb_define_method(cTermDocEnum, "freq", frb_tde_freq, 0);
|
2542
2516
|
rb_define_method(cTermDocEnum, "next?", frb_tde_next, 0);
|
2543
2517
|
rb_define_method(cTermDocEnum, "next_position", frb_tde_next_position, 0);
|
@@ -2606,7 +2580,7 @@ static void Init_TVTerm(void) {
|
|
2606
2580
|
*
|
2607
2581
|
* == Example
|
2608
2582
|
*
|
2609
|
-
* tv = index_reader.term_vector(
|
2583
|
+
* tv = index_reader.term_vector(doc_num, :content)
|
2610
2584
|
* tv_term = tv.find {|tvt| tvt.term == "fox"}
|
2611
2585
|
*
|
2612
2586
|
* # get the term frequency
|
@@ -2832,7 +2806,7 @@ void Init_IndexReader(void) {
|
|
2832
2806
|
rb_define_method(cIndexReader, "has_deletions?", frb_ir_has_deletions, 0);
|
2833
2807
|
rb_define_method(cIndexReader, "delete", frb_ir_delete, 1);
|
2834
2808
|
rb_define_method(cIndexReader, "deleted?", frb_ir_is_deleted, 1);
|
2835
|
-
rb_define_method(cIndexReader, "
|
2809
|
+
rb_define_method(cIndexReader, "max_doc_num", frb_ir_max_doc_num, 0);
|
2836
2810
|
rb_define_method(cIndexReader, "num_docs", frb_ir_num_docs, 0);
|
2837
2811
|
rb_define_method(cIndexReader, "undelete_all", frb_ir_undelete_all, 0);
|
2838
2812
|
rb_define_method(cIndexReader, "latest?", frb_ir_is_latest, 0);
|