annotate src/luan/modules/lucene/LuceneIndex.java @ 1391:94f48cc76de8

add lucene check
author Franklin Schmidt <fschmidt@gmail.com>
date Thu, 05 Sep 2019 01:29:57 -0600
parents 179c4882c6b6
children 002152af497a
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
1 package luan.modules.lucene;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
2
527
Franklin Schmidt <fschmidt@gmail.com>
parents: 524
diff changeset
3 import java.io.Closeable;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
4 import java.io.File;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
5 import java.io.FileOutputStream;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
6 import java.io.FileInputStream;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
7 import java.io.IOException;
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
8 import java.sql.SQLException;
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
9 import java.util.Arrays;
544
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
10 import java.util.Iterator;
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
11 import java.util.Map;
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
12 import java.util.HashMap;
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
13 import java.util.List;
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
14 import java.util.ArrayList;
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
15 import java.util.Set;
618
5e495e4e560b add lucene indexed_only_fields
Franklin Schmidt <fschmidt@gmail.com>
parents: 617
diff changeset
16 import java.util.HashSet;
5e495e4e560b add lucene indexed_only_fields
Franklin Schmidt <fschmidt@gmail.com>
parents: 617
diff changeset
17 import java.util.Collections;
704
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
18 import java.util.concurrent.atomic.AtomicInteger;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
19 import java.util.concurrent.locks.Lock;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
20 import java.util.concurrent.locks.ReentrantLock;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
21 import java.util.zip.ZipOutputStream;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
22 import java.util.zip.ZipEntry;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
23 import org.apache.lucene.analysis.Analyzer;
1341
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
24 import org.apache.lucene.analysis.TokenStream;
544
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
25 import org.apache.lucene.analysis.core.KeywordAnalyzer;
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
26 import org.apache.lucene.analysis.en.EnglishAnalyzer;
233
ef39bc4d3f70 basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 232
diff changeset
27 import org.apache.lucene.document.Document;
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
28 import org.apache.lucene.document.Field;
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
29 import org.apache.lucene.document.StoredField;
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
30 import org.apache.lucene.document.StringField;
599
50540f0813e2 support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents: 591
diff changeset
31 import org.apache.lucene.document.TextField;
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
32 import org.apache.lucene.document.IntField;
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
33 import org.apache.lucene.document.LongField;
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
34 import org.apache.lucene.document.DoubleField;
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
35 import org.apache.lucene.index.IndexableField;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
36 import org.apache.lucene.index.IndexWriter;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
37 import org.apache.lucene.index.IndexWriterConfig;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
38 import org.apache.lucene.index.DirectoryReader;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
39 import org.apache.lucene.index.Term;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
40 import org.apache.lucene.index.SnapshotDeletionPolicy;
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
41 import org.apache.lucene.index.IndexCommit;
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
42 import org.apache.lucene.index.AtomicReaderContext;
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
43 import org.apache.lucene.index.CheckIndex;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
44 import org.apache.lucene.store.Directory;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
45 import org.apache.lucene.store.FSDirectory;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
46 import org.apache.lucene.util.Version;
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
47 import org.apache.lucene.util.BytesRef;
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
48 import org.apache.lucene.util.NumericUtils;
312
d34be4588556 add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 303
diff changeset
49 import org.apache.lucene.search.Query;
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
50 import org.apache.lucene.search.PrefixQuery;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
51 import org.apache.lucene.search.TermQuery;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
52 import org.apache.lucene.search.TopDocs;
544
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
53 import org.apache.lucene.search.Sort;
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
54 import org.apache.lucene.search.SortField;
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
55 import org.apache.lucene.search.IndexSearcher;
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
56 import org.apache.lucene.search.TotalHitCountCollector;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
57 import org.apache.lucene.search.ScoreDoc;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
58 import org.apache.lucene.search.Collector;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
59 import org.apache.lucene.search.Scorer;
1342
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
60 import org.apache.lucene.search.Explanation;
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
61 import org.apache.lucene.search.highlight.Formatter;
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
62 import org.apache.lucene.search.highlight.Highlighter;
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
63 import org.apache.lucene.search.highlight.InvalidTokenOffsetsException;
1248
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
64 import org.apache.lucene.search.highlight.Fragmenter;
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
65 import org.apache.lucene.search.highlight.NullFragmenter;
1248
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
66 import org.apache.lucene.search.highlight.SimpleSpanFragmenter;
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
67 import org.apache.lucene.search.highlight.QueryScorer;
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
68 import org.apache.lucene.search.highlight.TokenGroup;
1344
dc2af9d5463b move queryparser to lib
Franklin Schmidt <fschmidt@gmail.com>
parents: 1343
diff changeset
69 import luan.lib.queryparser.SaneQueryParser;
dc2af9d5463b move queryparser to lib
Franklin Schmidt <fschmidt@gmail.com>
parents: 1343
diff changeset
70 import luan.lib.queryparser.FieldParser;
dc2af9d5463b move queryparser to lib
Franklin Schmidt <fschmidt@gmail.com>
parents: 1343
diff changeset
71 import luan.lib.queryparser.MultiFieldParser;
dc2af9d5463b move queryparser to lib
Franklin Schmidt <fschmidt@gmail.com>
parents: 1343
diff changeset
72 import luan.lib.queryparser.StringFieldParser;
dc2af9d5463b move queryparser to lib
Franklin Schmidt <fschmidt@gmail.com>
parents: 1343
diff changeset
73 import luan.lib.queryparser.NumberFieldParser;
1111
88b5b81cad4a move Parser to luan.lib.parser
Franklin Schmidt <fschmidt@gmail.com>
parents: 1110
diff changeset
74 import luan.lib.parser.ParseException;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
75 import luan.modules.Utils;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
76 import luan.Luan;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
77 import luan.LuanTable;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
78 import luan.LuanFunction;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
79 import luan.LuanException;
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
80 import luan.LuanRuntimeException;
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
81 import luan.modules.parsers.LuanToString;
1337
8b61c8c4e07a remove slf4j
Franklin Schmidt <fschmidt@gmail.com>
parents: 1335
diff changeset
82 import luan.lib.logging.Logger;
8b61c8c4e07a remove slf4j
Franklin Schmidt <fschmidt@gmail.com>
parents: 1335
diff changeset
83 import luan.lib.logging.LoggerFactory;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
84
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
85
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
86 public final class LuceneIndex {
521
8a217fe5b4f3 cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents: 427
diff changeset
87 private static final Logger logger = LoggerFactory.getLogger(LuceneIndex.class);
8a217fe5b4f3 cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents: 427
diff changeset
88
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
89 private static final class Closer implements Closeable {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
90 final LuceneIndex li;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
91 boolean isClosed = false;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
92 private final Exception created = new Exception("created");
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
93
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
94 Closer(Luan luan,LuceneIndex li) {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
95 this.li = li;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
96 luan.onClose(this);
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
97 }
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
98
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
99 public void close() throws IOException {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
100 if( !isClosed ) {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
101 li.close();
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
102 isClosed = true;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
103 }
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
104 }
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
105
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
106 protected void finalize() throws Throwable {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
107 if( !isClosed ) {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
108 logger.error("not closed",created);
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
109 close();
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
110 }
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
111 super.finalize();
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
112 }
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
113 }
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
114
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
115 private static Map<String,LuceneIndex> indexes = new HashMap<String,LuceneIndex>();
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
116
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
117 public static Object[] getLuceneIndex(Luan luan,File indexDir,FieldParser defaultFieldParser,String[] defaultFields,LuanFunction completer)
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
118 throws LuanException, IOException
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
119 {
1369
709f7498a363 change Lucene.index() and add Lucene.recover()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1347
diff changeset
120 String key = indexDir.getCanonicalPath();
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
121 synchronized(indexes) {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
122 LuceneIndex li = indexes.get(key);
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
123 if( li == null ) {
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
124 li = new LuceneIndex(indexDir,defaultFieldParser,defaultFields,key,completer);
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
125 li.openCount = 1;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
126 indexes.put(key,li);
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
127 } else {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
128 if( defaultFieldParser != li.defaultFieldParser )
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
129 throw new LuanException("default_type doesn't match previous use");
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
130 if( !Arrays.equals(defaultFields,li.defaultFields) )
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
131 throw new LuanException("default_fields don't match previous use");
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
132 li.openCount++;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
133 }
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
134 return new Object[]{li,new Closer(luan,li)};
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
135 }
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
136 }
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
137
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
138 private static final Version version = Version.LUCENE_4_9;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
139 private static final String FLD_NEXT_ID = "nextId";
599
50540f0813e2 support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents: 591
diff changeset
140 public static final StringFieldParser STRING_FIELD_PARSER = new StringFieldParser(new KeywordAnalyzer());
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
141 public static final StringFieldParser ENGLISH_FIELD_PARSER = new StringFieldParser(new EnglishAnalyzer(version));
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
142
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
143 private final ReentrantLock writeLock = new ReentrantLock();
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
144 private final File indexDir;
754
1a101ac9ea46 add lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 730
diff changeset
145 private SnapshotDeletionPolicy snapshotDeletionPolicy;
1a101ac9ea46 add lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 730
diff changeset
146 private IndexWriter writer;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
147 private DirectoryReader reader;
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
148 private IndexSearcher searcher;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
149 private final ThreadLocal<IndexSearcher> threadLocalSearcher = new ThreadLocal<IndexSearcher>();
599
50540f0813e2 support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents: 591
diff changeset
150 private final MultiFieldParser mfp;
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
151 private final Analyzer analyzer;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
152
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
153 private FSDirectory fsDir;
704
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
154 private int writeCount;
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
155 private AtomicInteger writeCounter = new AtomicInteger();
704
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
156
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
157 private Set<String> indexOnly = new HashSet<String>();
1271
48f302bdc187 fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents: 1267
diff changeset
158
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
159 private int openCount;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
160 private final String key;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
161 private final FieldParser defaultFieldParser;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
162 private final String[] defaultFields;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
163
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
164 private final PostgresBackup postgresBackup;
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
165
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
166 private LuceneIndex(File indexDir,FieldParser defaultFieldParser,String[] defaultFields,String key,LuanFunction completer)
1267
9fa8b8389578 add LuanTable.luan;
Franklin Schmidt <fschmidt@gmail.com>
parents: 1248
diff changeset
167 throws LuanException, IOException
9fa8b8389578 add LuanTable.luan;
Franklin Schmidt <fschmidt@gmail.com>
parents: 1248
diff changeset
168 {
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
169 this.key = key;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
170 this.defaultFieldParser = defaultFieldParser;
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
171 this.defaultFields = defaultFields;
599
50540f0813e2 support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents: 591
diff changeset
172 mfp = defaultFieldParser==null ? new MultiFieldParser() : new MultiFieldParser(defaultFieldParser,defaultFields);
544
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
173 mfp.fields.put( "type", STRING_FIELD_PARSER );
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
174 mfp.fields.put( "id", NumberFieldParser.LONG );
233
ef39bc4d3f70 basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 232
diff changeset
175 this.indexDir = indexDir;
599
50540f0813e2 support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents: 591
diff changeset
176 Analyzer analyzer = STRING_FIELD_PARSER.analyzer;
50540f0813e2 support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents: 591
diff changeset
177 if( defaultFieldParser instanceof StringFieldParser ) {
50540f0813e2 support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents: 591
diff changeset
178 StringFieldParser sfp = (StringFieldParser)defaultFieldParser;
50540f0813e2 support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents: 591
diff changeset
179 analyzer = sfp.analyzer;
50540f0813e2 support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents: 591
diff changeset
180 }
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
181 this.analyzer = analyzer;
1390
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
182 boolean wasCreated = reopen();
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
183 postgresBackup = completer!=null ? PostgresBackup.newInstance() : null;
1390
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
184 if( postgresBackup != null ) {
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
185 if( !wasCreated && postgresBackup.wasCreated ) {
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
186 logger.error("rebuilding postgres backup");
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
187 rebuild_postgres_backup(completer);
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
188 } else if( wasCreated && !postgresBackup.wasCreated ) {
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
189 logger.error("restoring from postgres");
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
190 restore_from_postgres();
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
191 }
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
192 }
754
1a101ac9ea46 add lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 730
diff changeset
193 }
1a101ac9ea46 add lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 730
diff changeset
194
1390
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
195 public boolean reopen() throws IOException {
233
ef39bc4d3f70 basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 232
diff changeset
196 IndexWriterConfig conf = new IndexWriterConfig(version,analyzer);
ef39bc4d3f70 basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 232
diff changeset
197 snapshotDeletionPolicy = new SnapshotDeletionPolicy(conf.getIndexDeletionPolicy());
ef39bc4d3f70 basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 232
diff changeset
198 conf.setIndexDeletionPolicy(snapshotDeletionPolicy);
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
199 fsDir = FSDirectory.open(indexDir);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
200 boolean wasCreated = !fsDir.getDirectory().exists();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
201 writer = new IndexWriter(fsDir,conf);
233
ef39bc4d3f70 basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 232
diff changeset
202 writer.commit(); // commit index creation
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
203 reader = DirectoryReader.open(fsDir);
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
204 searcher = new IndexSearcher(reader);
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
205 initId();
1390
179c4882c6b6 backup work
Franklin Schmidt <fschmidt@gmail.com>
parents: 1388
diff changeset
206 return wasCreated;
233
ef39bc4d3f70 basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 232
diff changeset
207 }
ef39bc4d3f70 basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 232
diff changeset
208
704
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
209 private void wrote() {
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
210 writeCounter.incrementAndGet();
704
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
211 }
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
212
252
3896138955b1 web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 233
diff changeset
213 public void delete_all() throws IOException {
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
214 boolean commit = !writeLock.isHeldByCurrentThread();
252
3896138955b1 web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 233
diff changeset
215 writeLock.lock();
3896138955b1 web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 233
diff changeset
216 try {
3896138955b1 web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 233
diff changeset
217 writer.deleteAll();
3896138955b1 web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 233
diff changeset
218 id = idLim = 0;
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
219 if( postgresBackup != null )
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
220 postgresBackup.deleteAll();
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
221 if(commit) writer.commit();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
222 } finally {
704
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
223 wrote();
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
224 writeLock.unlock();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
225 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
226 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
227
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
228 private static Term term(String key,long value) {
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
229 BytesRef br = new BytesRef();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
230 NumericUtils.longToPrefixCoded(value,0,br);
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
231 return new Term(key,br);
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
232 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
233
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
234 private void backupDelete(Query query)
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
235 throws IOException
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
236 {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
237 if( postgresBackup != null ) {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
238 final List<Long> ids = new ArrayList<Long>();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
239 IndexSearcher searcher = openSearcher();
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
240 try {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
241 MyCollector col = new MyCollector() {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
242 @Override public void collect(int iDoc) throws IOException {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
243 Document doc = searcher.doc( docBase + iDoc );
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
244 Long id = (Long)doc.getField("id").numericValue();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
245 ids.add(id);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
246 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
247 };
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
248 searcher.search(query,col);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
249 } finally {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
250 close(searcher);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
251 }
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
252 postgresBackup.begin();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
253 for( Long id : ids ) {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
254 postgresBackup.delete(id);
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
255 }
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
256 postgresBackup.commit();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
257 }
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
258 }
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
259
1346
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
260 public void delete(String queryStr)
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
261 throws IOException, ParseException
1341
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
262 {
622
1a53333eb4d5 remove Lucene all_search_terms_must_match() since now query suffix "~a" handles this
Franklin Schmidt <fschmidt@gmail.com>
parents: 621
diff changeset
263 Query query = SaneQueryParser.parseQuery(mfp,queryStr);
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
264
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
265 boolean commit = !writeLock.isHeldByCurrentThread();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
266 writeLock.lock();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
267 try {
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
268 backupDelete(query);
547
0be287ab0309 add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents: 546
diff changeset
269 writer.deleteDocuments(query);
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
270 if(commit) writer.commit();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
271 } finally {
704
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
272 wrote();
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
273 writeLock.unlock();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
274 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
275 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
276
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
277 public void indexed_only_fields(List<String> fields) {
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
278 indexOnly.addAll(fields);
1271
48f302bdc187 fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents: 1267
diff changeset
279 }
48f302bdc187 fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents: 1267
diff changeset
280
1346
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
281 public void save(LuanTable doc,LuanTable boosts)
1341
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
282 throws LuanException, IOException
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
283 {
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
284 if( boosts!=null && postgresBackup!=null )
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
285 logger.error("boosts are not saved to postgres backup");
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
286
1267
9fa8b8389578 add LuanTable.luan;
Franklin Schmidt <fschmidt@gmail.com>
parents: 1248
diff changeset
287 Object obj = doc.get("id");
601
d36027b41570 better error message
Franklin Schmidt <fschmidt@gmail.com>
parents: 599
diff changeset
288 Long id;
d36027b41570 better error message
Franklin Schmidt <fschmidt@gmail.com>
parents: 599
diff changeset
289 try {
d36027b41570 better error message
Franklin Schmidt <fschmidt@gmail.com>
parents: 599
diff changeset
290 id = (Long)obj;
d36027b41570 better error message
Franklin Schmidt <fschmidt@gmail.com>
parents: 599
diff changeset
291 } catch(ClassCastException e) {
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
292 throw new LuanException("id should be Long but is "+obj.getClass().getSimpleName());
601
d36027b41570 better error message
Franklin Schmidt <fschmidt@gmail.com>
parents: 599
diff changeset
293 }
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
294
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
295 boolean commit = !writeLock.isHeldByCurrentThread();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
296 writeLock.lock();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
297 try {
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
298 if( id == null ) {
1346
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
299 id = nextId();
1267
9fa8b8389578 add LuanTable.luan;
Franklin Schmidt <fschmidt@gmail.com>
parents: 1248
diff changeset
300 doc.put("id",id);
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
301 if( postgresBackup != null )
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
302 postgresBackup.add(doc);
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
303 writer.addDocument(toLucene(doc,boosts));
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
304 } else {
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
305 if( postgresBackup != null )
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
306 postgresBackup.update(doc);
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
307 writer.updateDocument( term("id",id), toLucene(doc,boosts) );
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
308 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
309 if(commit) writer.commit();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
310 } finally {
704
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
311 wrote();
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
312 writeLock.unlock();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
313 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
314 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
315
1379
87a3738d7cc5 run_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents: 1369
diff changeset
316 public Object run_in_transaction(LuanFunction fn) throws IOException, LuanException {
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
317 boolean commit = !writeLock.isHeldByCurrentThread();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
318 writeLock.lock();
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
319 boolean ok = false;
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
320 try {
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
321 if( commit && postgresBackup != null )
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
322 postgresBackup.begin();
1379
87a3738d7cc5 run_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents: 1369
diff changeset
323 Object rtn = fn.call();
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
324 ok = true;
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
325 if(commit) {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
326 if( postgresBackup != null )
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
327 postgresBackup.commit();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
328 writer.commit();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
329 }
1379
87a3738d7cc5 run_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents: 1369
diff changeset
330 return rtn;
252
3896138955b1 web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 233
diff changeset
331 } finally {
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
332 if( !ok && commit ) {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
333 if( postgresBackup != null )
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
334 postgresBackup.rollback();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
335 writer.rollback();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
336 reopen();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
337 }
704
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
338 wrote();
252
3896138955b1 web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 233
diff changeset
339 writeLock.unlock();
3896138955b1 web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 233
diff changeset
340 }
3896138955b1 web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 233
diff changeset
341 }
3896138955b1 web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 233
diff changeset
342
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
343 // ???
1379
87a3738d7cc5 run_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents: 1369
diff changeset
344 public Object run_in_lock(LuanFunction fn) throws IOException, LuanException {
756
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
345 if( writeLock.isHeldByCurrentThread() )
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
346 throw new RuntimeException();
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
347 writeLock.lock();
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
348 try {
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
349 synchronized(this) {
1379
87a3738d7cc5 run_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents: 1369
diff changeset
350 return fn.call();
756
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
351 }
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
352 } finally {
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
353 wrote();
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
354 writeLock.unlock();
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
355 }
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
356 }
9092e52f94eb better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 754
diff changeset
357
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
358
754
1a101ac9ea46 add lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 730
diff changeset
359 private long id;
1a101ac9ea46 add lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 730
diff changeset
360 private long idLim;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
361 private final int idBatch = 10;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
362
1388
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
363 private void initId() throws IOException {
544
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
364 TopDocs td = searcher.search(new TermQuery(new Term("type","next_id")),1);
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
365 switch(td.totalHits) {
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
366 case 0:
754
1a101ac9ea46 add lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 730
diff changeset
367 id = 0;
1a101ac9ea46 add lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 730
diff changeset
368 idLim = 0;
1a101ac9ea46 add lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents: 730
diff changeset
369 break;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
370 case 1:
544
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
371 idLim = (Long)searcher.doc(td.scoreDocs[0].doc).getField(FLD_NEXT_ID).numericValue();
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
372 id = idLim;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
373 break;
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
374 default:
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
375 throw new RuntimeException();
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
376 }
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
377 }
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
378
1388
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
379 private void saveNextId(long nextId) throws LuanException, IOException {
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
380 Map doc = new HashMap();
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
381 doc.put( "type", "next_id" );
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
382 doc.put( FLD_NEXT_ID, idLim );
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
383 writer.updateDocument(new Term("type","next_id"),toLucene(doc.entrySet(),null));
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
384 }
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
385
1346
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
386 public synchronized long nextId() throws LuanException, IOException {
544
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
387 if( ++id > idLim ) {
233
ef39bc4d3f70 basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 232
diff changeset
388 idLim += idBatch;
1388
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
389 saveNextId(idLim);
704
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
390 wrote();
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
391 }
544
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
392 return id;
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
393 }
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
394
707
1ed9e55f0be8 replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents: 704
diff changeset
395 /*
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
396 public void backup(String zipFile) throws LuanException, IOException {
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
397 if( !zipFile.endsWith(".zip") )
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
398 throw new LuanException("file "+zipFile+" doesn't end with '.zip'");
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
399 IndexCommit ic = snapshotDeletionPolicy.snapshot();
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
400 try {
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
401 ZipOutputStream out = new ZipOutputStream(new FileOutputStream(zipFile));
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
402 for( String fileName : ic.getFileNames() ) {
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
403 out.putNextEntry(new ZipEntry(fileName));
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
404 FileInputStream in = new FileInputStream(new File(indexDir,fileName));
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
405 Utils.copyAll(in,out);
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
406 in.close();
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
407 out.closeEntry();
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
408 }
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
409 out.close();
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
410 } finally {
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
411 snapshotDeletionPolicy.release(ic);
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
412 }
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
413 }
707
1ed9e55f0be8 replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents: 704
diff changeset
414 */
1129
3234a14bb1f8 minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents: 1111
diff changeset
415 public SnapshotDeletionPolicy snapshotDeletionPolicy() {
3234a14bb1f8 minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents: 1111
diff changeset
416 return snapshotDeletionPolicy;
3234a14bb1f8 minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents: 1111
diff changeset
417 }
3234a14bb1f8 minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents: 1111
diff changeset
418
1346
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
419 public Object snapshot(LuanFunction fn) throws LuanException, IOException {
707
1ed9e55f0be8 replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents: 704
diff changeset
420 IndexCommit ic = snapshotDeletionPolicy.snapshot();
1ed9e55f0be8 replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents: 704
diff changeset
421 try {
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
422 String dir = fsDir.getDirectory().toString();
1346
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
423 LuanTable fileNames = new LuanTable(fn.luan(),new ArrayList(ic.getFileNames()));
1335
e0cf0d108a77 major cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1333
diff changeset
424 return fn.call(dir,fileNames);
707
1ed9e55f0be8 replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents: 704
diff changeset
425 } finally {
1ed9e55f0be8 replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents: 704
diff changeset
426 snapshotDeletionPolicy.release(ic);
1ed9e55f0be8 replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents: 704
diff changeset
427 }
1ed9e55f0be8 replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents: 704
diff changeset
428 }
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
429
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
430
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
431
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
432 public String to_string() {
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
433 return writer.getDirectory().toString();
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
434 }
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
435
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
436 private synchronized void close() throws IOException {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
437 if( openCount > 0 ) {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
438 if( --openCount == 0 ) {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
439 doClose();
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
440 synchronized(indexes) {
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
441 indexes.remove(key);
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
442 }
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
443 }
521
8a217fe5b4f3 cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents: 427
diff changeset
444 }
8a217fe5b4f3 cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents: 427
diff changeset
445 }
8a217fe5b4f3 cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents: 427
diff changeset
446
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
447 public void doClose() throws IOException {
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
448 if( postgresBackup != null )
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
449 postgresBackup.close();
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
450 writer.close();
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
451 reader.close();
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
452 }
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
453
312
d34be4588556 add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 303
diff changeset
454
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
455 private static class DocFn extends LuanFunction {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
456 final IndexSearcher searcher;
1342
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
457 final Query query;
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
458 int docID;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
459
1342
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
460 DocFn(Luan luan,IndexSearcher searcher,Query query) {
1335
e0cf0d108a77 major cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1333
diff changeset
461 super(luan);
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
462 this.searcher = searcher;
1342
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
463 this.query = query;
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
464 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
465
1335
e0cf0d108a77 major cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1333
diff changeset
466 @Override public Object call(Object[] args) throws LuanException {
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
467 try {
1342
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
468 LuanTable doc = toTable(luan(),searcher.doc(docID));
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
469 if( args.length > 0 && "explain".equals(args[0]) ) {
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
470 Explanation explanation = searcher.explain(query,docID);
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
471 return new Object[]{doc,explanation};
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
472 } else {
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
473 return doc;
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
474 }
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
475 } catch(IOException e) {
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
476 throw new LuanException(e);
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
477 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
478 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
479 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
480
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
481 private static abstract class MyCollector extends Collector {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
482 int docBase;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
483 int i = 0;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
484
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
485 @Override public void setScorer(Scorer scorer) {}
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
486 @Override public void setNextReader(AtomicReaderContext context) {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
487 this.docBase = context.docBase;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
488 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
489 @Override public boolean acceptsDocsOutOfOrder() {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
490 return true;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
491 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
492 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
493
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
494 private synchronized IndexSearcher openSearcher() throws IOException {
1345
6f8988830098 unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents: 1344
diff changeset
495 int gwc = writeCounter.get();
704
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
496 if( writeCount != gwc ) {
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
497 writeCount = gwc;
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
498 DirectoryReader newReader = DirectoryReader.openIfChanged(reader);
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
499 if( newReader != null ) {
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
500 reader.decRef();
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
501 reader = newReader;
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
502 searcher = new IndexSearcher(reader);
90c89138c234 optimize lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 696
diff changeset
503 }
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
504 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
505 reader.incRef();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
506 return searcher;
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
507 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
508
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
509 // call in finally block
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
510 private static void close(IndexSearcher searcher) throws IOException {
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
511 searcher.getIndexReader().decRef();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
512 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
513
591
790d5de23042 add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents: 578
diff changeset
514 public void ensure_open() throws IOException {
790d5de23042 add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents: 578
diff changeset
515 close(openSearcher());
790d5de23042 add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents: 578
diff changeset
516 }
790d5de23042 add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents: 578
diff changeset
517
1346
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
518 public int advanced_search( String queryStr, LuanFunction fn, Integer n, String sortStr )
1341
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
519 throws LuanException, IOException, ParseException
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
520 {
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
521 Utils.checkNotNull(queryStr);
622
1a53333eb4d5 remove Lucene all_search_terms_must_match() since now query suffix "~a" handles this
Franklin Schmidt <fschmidt@gmail.com>
parents: 621
diff changeset
522 Query query = SaneQueryParser.parseQuery(mfp,queryStr);
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
523 IndexSearcher searcher = threadLocalSearcher.get();
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
524 boolean inTransaction = searcher != null;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
525 if( !inTransaction )
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
526 searcher = openSearcher();
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
527 try {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
528 if( fn!=null && n==null ) {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
529 if( sortStr != null )
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
530 throw new LuanException("sort must be nil when n is nil");
1346
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
531 final DocFn docFn = new DocFn(fn.luan(),searcher,query);
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
532 MyCollector col = new MyCollector() {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
533 @Override public void collect(int doc) {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
534 try {
547
0be287ab0309 add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents: 546
diff changeset
535 docFn.docID = docBase + doc;
1335
e0cf0d108a77 major cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1333
diff changeset
536 fn.call(++i,docFn);
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
537 } catch(LuanException e) {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
538 throw new LuanRuntimeException(e);
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
539 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
540 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
541 };
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
542 try {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
543 searcher.search(query,col);
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
544 } catch(LuanRuntimeException e) {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
545 throw (LuanException)e.getCause();
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
546 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
547 return col.i;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
548 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
549 if( fn==null || n==0 ) {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
550 TotalHitCountCollector thcc = new TotalHitCountCollector();
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
551 searcher.search(query,thcc);
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
552 return thcc.getTotalHits();
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
553 }
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
554 Sort sort = sortStr==null ? null : SaneQueryParser.parseSort(mfp,sortStr);
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
555 TopDocs td = sort==null ? searcher.search(query,n) : searcher.search(query,n,sort);
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
556 final ScoreDoc[] scoreDocs = td.scoreDocs;
1346
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
557 DocFn docFn = new DocFn(fn.luan(),searcher,query);
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
558 for( int i=0; i<scoreDocs.length; i++ ) {
1342
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
559 ScoreDoc scoreDoc = scoreDocs[i];
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
560 docFn.docID = scoreDoc.doc;
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
561 fn.call(i+1,docFn,scoreDoc.score);
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
562 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
563 return td.totalHits;
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
564 } finally {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
565 if( !inTransaction )
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
566 close(searcher);
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
567 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
568 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
569
1335
e0cf0d108a77 major cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1333
diff changeset
570 public Object search_in_transaction(LuanFunction fn) throws LuanException, IOException {
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
571 if( threadLocalSearcher.get() != null )
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
572 throw new LuanException("can't nest search_in_transaction calls");
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
573 IndexSearcher searcher = openSearcher();
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
574 threadLocalSearcher.set(searcher);
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
575 try {
1335
e0cf0d108a77 major cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1333
diff changeset
576 return fn.call();
545
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
577 } finally {
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
578 threadLocalSearcher.set(null);
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
579 close(searcher);
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
580 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
581 }
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
582
ddcd4296107a clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents: 544
diff changeset
583
796
6b8ea0a9b7c9 remove LuanMeta
Franklin Schmidt <fschmidt@gmail.com>
parents: 775
diff changeset
584 public FieldParser getIndexedFieldParser(String field) {
6b8ea0a9b7c9 remove LuanMeta
Franklin Schmidt <fschmidt@gmail.com>
parents: 775
diff changeset
585 return mfp.fields.get(field);
6b8ea0a9b7c9 remove LuanMeta
Franklin Schmidt <fschmidt@gmail.com>
parents: 775
diff changeset
586 }
544
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
587
796
6b8ea0a9b7c9 remove LuanMeta
Franklin Schmidt <fschmidt@gmail.com>
parents: 775
diff changeset
588 public void setIndexedFieldParser(String field,FieldParser fp) {
6b8ea0a9b7c9 remove LuanMeta
Franklin Schmidt <fschmidt@gmail.com>
parents: 775
diff changeset
589 if( fp==null ) { // delete
6b8ea0a9b7c9 remove LuanMeta
Franklin Schmidt <fschmidt@gmail.com>
parents: 775
diff changeset
590 mfp.fields.remove(field);
6b8ea0a9b7c9 remove LuanMeta
Franklin Schmidt <fschmidt@gmail.com>
parents: 775
diff changeset
591 return;
6b8ea0a9b7c9 remove LuanMeta
Franklin Schmidt <fschmidt@gmail.com>
parents: 775
diff changeset
592 }
6b8ea0a9b7c9 remove LuanMeta
Franklin Schmidt <fschmidt@gmail.com>
parents: 775
diff changeset
593 mfp.fields.put( field, fp );
6b8ea0a9b7c9 remove LuanMeta
Franklin Schmidt <fschmidt@gmail.com>
parents: 775
diff changeset
594 }
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
595
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
596
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
597 private IndexableField newField(String name,Object value,Set<String> indexed,Float boost)
1341
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
598 throws LuanException
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
599 {
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
600 boolean hasBoost = boost!=null;
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
601 IndexableField fld = newField2(name,value,indexed,hasBoost);
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
602 if( hasBoost )
1341
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
603 ((Field)fld).setBoost(boost);
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
604 return fld;
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
605 }
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
606
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
607 private IndexableField newField2(String name,Object value,Set<String> indexed,boolean hasBoost)
621
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
608 throws LuanException
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
609 {
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
610 Field.Store store = indexOnly.contains(name) ? Field.Store.NO : Field.Store.YES;
621
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
611 if( value instanceof String ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
612 String s = (String)value;
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
613 FieldParser fp = mfp.fields.get(name);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
614 if( fp != null ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
615 if( fp instanceof StringFieldParser && fp != STRING_FIELD_PARSER ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
616 return new TextField(name, s, store);
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
617 } else if (hasBoost) {
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
618 // fuck you modern lucene developers
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
619 return new Field(name, s, store, Field.Index.NOT_ANALYZED);
621
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
620 } else {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
621 return new StringField(name, s, store);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
622 }
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
623 } else {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
624 return new StoredField(name, s);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
625 }
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
626 } else if( value instanceof Integer ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
627 int i = (Integer)value;
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
628 if( indexed.contains(name) ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
629 return new IntField(name, i, store);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
630 } else {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
631 return new StoredField(name, i);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
632 }
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
633 } else if( value instanceof Long ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
634 long i = (Long)value;
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
635 if( indexed.contains(name) ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
636 return new LongField(name, i, store);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
637 } else {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
638 return new StoredField(name, i);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
639 }
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
640 } else if( value instanceof Double ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
641 double i = (Double)value;
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
642 if( indexed.contains(name) ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
643 return new DoubleField(name, i, store);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
644 } else {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
645 return new StoredField(name, i);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
646 }
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
647 } else if( value instanceof byte[] ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
648 byte[] b = (byte[])value;
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
649 return new StoredField(name, b);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
650 } else
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
651 throw new LuanException("invalid value type "+value.getClass()+"' for '"+name+"'");
621
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
652 }
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
653
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
654 private Document toLucene(LuanTable table,LuanTable boosts) throws LuanException {
1347
643cf1c37723 move webserver to lib and bug fixes
Franklin Schmidt <fschmidt@gmail.com>
parents: 1346
diff changeset
655 return toLucene(table.iterable(),boosts);
1346
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
656 }
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
657
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
658 private Document toLucene(Iterable<Map.Entry> iterable,LuanTable boosts) throws LuanException {
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
659 Set<String> indexed = mfp.fields.keySet();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
660 Document doc = new Document();
1346
Franklin Schmidt <fschmidt@gmail.com>
parents: 1345
diff changeset
661 for( Map.Entry<Object,Object> entry : iterable ) {
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
662 Object key = entry.getKey();
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
663 if( !(key instanceof String) )
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
664 throw new LuanException("key must be string");
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
665 String name = (String)key;
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
666 Object value = entry.getValue();
1341
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
667 Float boost = null;
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
668 if( boosts != null ) {
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
669 Object obj = boosts.get(name);
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
670 if( obj != null ) {
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
671 if( !(obj instanceof Number) )
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
672 throw new LuanException("boost '"+name+"' must be number");
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
673 boost = ((Number)obj).floatValue();
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
674 }
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
675 }
621
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
676 if( !(value instanceof LuanTable) ) {
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
677 doc.add(newField( name, value, indexed, boost ));
621
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
678 } else { // list
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
679 LuanTable list = (LuanTable)value;
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
680 for( Object el : list.asList() ) {
1343
7d9a1f8894b0 lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1342
diff changeset
681 doc.add(newField( name, el, indexed, boost ));
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
682 }
621
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
683 }
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
684 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
685 return doc;
544
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
686 }
c5a93767cc5c lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents: 527
diff changeset
687
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
688 private static Object getValue(IndexableField ifld) throws LuanException {
621
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
689 BytesRef br = ifld.binaryValue();
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
690 if( br != null )
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
691 return br.bytes;
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
692 Number n = ifld.numericValue();
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
693 if( n != null )
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
694 return n;
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
695 String s = ifld.stringValue();
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
696 if( s != null )
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
697 return s;
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
698 throw new LuanException("invalid field type for "+ifld);
621
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
699 }
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
700
1333
25746915a241 merge Luan and LuanState
Franklin Schmidt <fschmidt@gmail.com>
parents: 1271
diff changeset
701 private static LuanTable toTable(Luan luan,Document doc) throws LuanException {
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
702 if( doc==null )
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
703 return null;
1267
9fa8b8389578 add LuanTable.luan;
Franklin Schmidt <fschmidt@gmail.com>
parents: 1248
diff changeset
704 LuanTable table = new LuanTable(luan);
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
705 for( IndexableField ifld : doc ) {
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
706 String name = ifld.name();
646
cdc70de628b5 simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents: 624
diff changeset
707 Object value = getValue(ifld);
621
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
708 Object old = table.rawGet(name);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
709 if( old == null ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
710 table.rawPut(name,value);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
711 } else {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
712 LuanTable list;
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
713 if( old instanceof LuanTable ) {
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
714 list = (LuanTable)old;
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
715 } else {
1267
9fa8b8389578 add LuanTable.luan;
Franklin Schmidt <fschmidt@gmail.com>
parents: 1248
diff changeset
716 list = new LuanTable(luan);
621
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
717 list.rawPut(1,old);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
718 table.rawPut(name,list);
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
719 }
fd15da41afca allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents: 620
diff changeset
720 list.rawPut(list.rawLength()+1,value);
546
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
721 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
722 }
eaef1005ab87 general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 545
diff changeset
723 return table;
312
d34be4588556 add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 303
diff changeset
724 }
d34be4588556 add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents: 303
diff changeset
725
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
726
1248
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
727 private static final Formatter nullFormatter = new Formatter() {
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
728 public String highlightTerm(String originalText,TokenGroup tokenGroup) {
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
729 return originalText;
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
730 }
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
731 };
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
732
1342
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
733 public LuanFunction highlighter(String queryStr,final LuanFunction formatter,final Integer fragmentSize,String dotdotdot)
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
734 throws ParseException
60599adc27b8 add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents: 1341
diff changeset
735 {
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
736 Query query = SaneQueryParser.parseQuery(mfp,queryStr);
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
737 Formatter fmt = new Formatter() {
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
738 public String highlightTerm(String originalText,TokenGroup tokenGroup) {
1248
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
739 if( tokenGroup.getTotalScore() <= 0 )
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
740 return originalText;
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
741 try {
1335
e0cf0d108a77 major cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1333
diff changeset
742 return (String)Luan.first(formatter.call(originalText));
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
743 } catch(LuanException e) {
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
744 throw new LuanRuntimeException(e);
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
745 }
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
746 }
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
747 };
1248
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
748 QueryScorer queryScorer = new QueryScorer(query);
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
749 final Highlighter chooser = fragmentSize==null ? null : new Highlighter(nullFormatter,queryScorer);
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
750 if( chooser != null )
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
751 chooser.setTextFragmenter( new SimpleSpanFragmenter(queryScorer,fragmentSize) );
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
752 final Highlighter hl = new Highlighter(fmt,queryScorer);
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
753 hl.setTextFragmenter( new NullFragmenter() );
1335
e0cf0d108a77 major cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1333
diff changeset
754 return new LuanFunction(false) { // ???
e0cf0d108a77 major cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1333
diff changeset
755 @Override public String call(Object[] args) throws LuanException {
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
756 String text = (String)args[0];
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
757 try {
1248
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
758 if( chooser != null ) {
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
759 String s = chooser.getBestFragment(analyzer,null,text);
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
760 if( s != null ) {
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
761 if( dotdotdot != null ) {
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
762 boolean atStart = text.startsWith(s);
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
763 boolean atEnd = text.endsWith(s);
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
764 if( !atStart )
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
765 s = dotdotdot + s;
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
766 if( !atEnd )
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
767 s = s + dotdotdot;
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
768 }
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
769 text = s;
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
770 } else if( text.length() > fragmentSize ) {
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
771 text = text.substring(0,fragmentSize);
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
772 if( dotdotdot != null )
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
773 text += "...";
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
774 }
475905984870 improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents: 1227
diff changeset
775 }
624
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
776 String s = hl.getBestFragment(analyzer,null,text);
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
777 return s!=null ? s : text;
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
778 } catch(LuanRuntimeException e) {
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
779 throw (LuanException)e.getCause();
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
780 } catch(IOException e) {
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
781 throw new RuntimeException(e);
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
782 } catch(InvalidTokenOffsetsException e) {
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
783 throw new RuntimeException(e);
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
784 }
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
785 }
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
786 };
8281a248c47e add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents: 622
diff changeset
787 }
1341
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
788
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
789 public int count_tokens(String text)
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
790 throws IOException
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
791 {
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
792 int n = 0;
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
793 TokenStream ts = analyzer.tokenStream(null,text);
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
794 ts.reset();
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
795 while( ts.incrementToken() ) {
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
796 n++;
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
797 }
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
798 ts.close();
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
799 return n;
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
800 }
a015a0b5c388 add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents: 1337
diff changeset
801
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
802
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
803
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
804 public boolean hasPostgresBackup() {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
805 return postgresBackup != null;
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
806 }
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
807
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
808 public void rebuild_postgres_backup(LuanFunction completer)
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
809 throws IOException, LuanException
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
810 {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
811 writeLock.lock();
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
812 IndexSearcher searcher = openSearcher();
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
813 boolean ok = false;
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
814 try {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
815 postgresBackup.begin();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
816 postgresBackup.deleteAll();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
817 Query query = new PrefixQuery(new Term("id"));
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
818 MyCollector col = new MyCollector() {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
819 @Override public void collect(int iDoc) throws IOException {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
820 try {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
821 Document doc = searcher.doc( docBase + iDoc );
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
822 LuanTable tbl = toTable(completer.luan(),doc);
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
823 tbl = (LuanTable)completer.call(tbl);
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
824 postgresBackup.add(tbl);
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
825 } catch(LuanException e) {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
826 throw new LuanRuntimeException(e);
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
827 }
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
828 }
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
829 };
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
830 try {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
831 searcher.search(query,col);
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
832 } catch(LuanRuntimeException e) {
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
833 throw (LuanException)e.getCause();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
834 }
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
835 ok = true;
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
836 postgresBackup.commit();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
837 } finally {
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
838 close(searcher);
1387
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
839 if( !ok )
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
840 postgresBackup.rollback();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
841 writeLock.unlock();
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
842 }
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
843 }
bc40bc9aab3a start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents: 1379
diff changeset
844
1388
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
845 public void restore_from_postgres()
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
846 throws IOException, LuanException
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
847 {
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
848 if( postgresBackup==null )
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
849 throw new NullPointerException();
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
850 if( writeLock.isHeldByCurrentThread() )
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
851 throw new RuntimeException();
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
852 writeLock.lock();
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
853 boolean ok = false;
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
854 try {
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
855 writer.deleteAll();
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
856 long nextId = postgresBackup.maxId() + 1;
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
857 postgresBackup.restoreLucene(this);
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
858 id = idLim = nextId;
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
859 ok = true;
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
860 writer.commit();
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
861 } finally {
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
862 if( !ok ) {
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
863 writer.rollback();
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
864 reopen();
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
865 }
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
866 wrote();
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
867 writeLock.unlock();
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
868 }
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
869 }
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
870
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
871 void restore(LuanTable doc)
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
872 throws LuanException, IOException
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
873 {
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
874 writer.addDocument(toLucene(doc,null));
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
875 }
2024d23ddd64 add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents: 1387
diff changeset
876
1391
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
877 public void check(LuanFunction completer) throws IOException, SQLException, LuanException {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
878 logger.info("start check");
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
879 CheckIndex.Status status = new CheckIndex(fsDir).checkIndex();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
880 if( !status.clean )
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
881 logger.error("index not clean");
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
882 if( postgresBackup != null )
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
883 checkPostgres(completer);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
884 logger.info("end check");
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
885 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
886
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
887 private void checkPostgres(LuanFunction completer) throws IOException, SQLException, LuanException {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
888 final PostgresBackup.Checker postgresChecker;
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
889 final IndexSearcher searcher;
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
890 writeLock.lock();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
891 try {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
892 postgresChecker = postgresBackup.newChecker();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
893 searcher = openSearcher();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
894 } finally {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
895 writeLock.unlock();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
896 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
897 try {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
898 final List<Long> idsLucene = new ArrayList<Long>();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
899 Query query = new PrefixQuery(new Term("id"));
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
900 MyCollector col = new MyCollector() {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
901 @Override public void collect(int iDoc) throws IOException {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
902 Document doc = searcher.doc( docBase + iDoc );
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
903 Long id = (Long)doc.getField("id").numericValue();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
904 idsLucene.add(id);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
905 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
906 };
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
907 searcher.search(query,col);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
908 Collections.sort(idsLucene);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
909 final List<Long> idsPostgres = postgresChecker.getIds();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
910 final int nLucene = idsLucene.size();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
911 final int nPostgres = idsPostgres.size();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
912 int iLucene = 0;
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
913 int iPostgres = 0;
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
914 LuanToString lts = new LuanToString();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
915 lts.strict = true;
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
916 lts.numberTypes = true;
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
917 while( iLucene < nLucene && iPostgres < nPostgres ) {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
918 long idLucene = idsLucene.get(iLucene);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
919 long idPostgres = idsPostgres.get(iPostgres);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
920 if( idLucene < idPostgres ) {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
921 iLucene++;
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
922 logger.error("id "+idLucene+" found in lucene but not postgres");
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
923 } else if( idLucene > idPostgres ) {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
924 iPostgres++;
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
925 logger.error("id "+idPostgres+" found in postgres but not lucene");
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
926 } else { // ==
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
927 LuanTable docPostgres = postgresChecker.getDoc(idPostgres);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
928 TopDocs td = searcher.search(new TermQuery(term("id",idLucene)),1);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
929 if( td.totalHits != 1 ) throw new RuntimeException();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
930 Document doc = searcher.doc( td.scoreDocs[0].doc );
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
931 LuanTable docLucene = toTable(completer.luan(),doc);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
932 docLucene = (LuanTable)completer.call(docLucene);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
933 if( !equal(docPostgres,docLucene) ) {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
934 logger.error("id "+idLucene+" not equal");
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
935 logger.error("lucene = "+lts.toString(docLucene));
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
936 logger.error("postgres = "+lts.toString(docPostgres));
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
937 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
938 iLucene++;
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
939 iPostgres++;
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
940 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
941 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
942 while( iLucene < nLucene ) {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
943 long idLucene = idsLucene.get(iLucene++);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
944 logger.error("id "+idLucene+" found in lucene but not postgres");
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
945 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
946 while( iPostgres < nPostgres ) {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
947 long idPostgres = idsPostgres.get(iPostgres++);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
948 logger.error("id "+idPostgres+" found in postgres but not lucene");
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
949 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
950 } finally {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
951 close(searcher);
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
952 postgresChecker.close();
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
953 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
954 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
955
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
956 private boolean equal(LuanTable t1,LuanTable t2) throws LuanException {
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
957 return t1.asMap().equals(t2.asMap());
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
958 }
94f48cc76de8 add lucene check
Franklin Schmidt <fschmidt@gmail.com>
parents: 1390
diff changeset
959
230
4438cb2e04d0 start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff changeset
960 }