Mercurial Hosting > luan
annotate src/luan/modules/lucene/LuceneIndex.java @ 1392:002152af497a
hosted postgres
author | Franklin Schmidt <fschmidt@gmail.com> |
---|---|
date | Fri, 06 Sep 2019 00:19:47 -0600 |
parents | 94f48cc76de8 |
children | cc0dbca576dc |
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 | 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 | 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 | 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 | 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 | 81 import luan.modules.parsers.LuanToString; |
1392 | 82 import luan.modules.logging.LuanLogger; |
1337 | 83 import luan.lib.logging.Logger; |
84 import luan.lib.logging.LoggerFactory; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
85 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
86 |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
87 public final class LuceneIndex { |
521
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
88 private static final Logger logger = LoggerFactory.getLogger(LuceneIndex.class); |
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
89 |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
90 private static final class Closer implements Closeable { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
91 final LuceneIndex li; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
92 boolean isClosed = false; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
93 private final Exception created = new Exception("created"); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
94 |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
95 Closer(Luan luan,LuceneIndex li) { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
96 this.li = li; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
97 luan.onClose(this); |
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 |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
100 public void close() throws IOException { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
101 if( !isClosed ) { |
1392 | 102 try { |
103 li.close(); | |
104 } catch(SQLException e) { | |
105 throw new RuntimeException(e); | |
106 } | |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
107 isClosed = true; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
108 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
109 } |
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 protected void finalize() throws Throwable { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
112 if( !isClosed ) { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
113 logger.error("not closed",created); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
114 close(); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
115 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
116 super.finalize(); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
117 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
118 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
119 |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
120 private static Map<String,LuceneIndex> indexes = new HashMap<String,LuceneIndex>(); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
121 |
1392 | 122 public static Object[] getLuceneIndex(Luan luan,File indexDir,FieldParser defaultFieldParser,String[] defaultFields,LuanTable postgresSpec) |
123 throws LuanException, IOException, ClassNotFoundException, SQLException | |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
124 { |
1369
709f7498a363
change Lucene.index() and add Lucene.recover()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1347
diff
changeset
|
125 String key = indexDir.getCanonicalPath(); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
126 synchronized(indexes) { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
127 LuceneIndex li = indexes.get(key); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
128 if( li == null ) { |
1392 | 129 li = new LuceneIndex(luan,indexDir,defaultFieldParser,defaultFields,key,postgresSpec); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
130 li.openCount = 1; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
131 indexes.put(key,li); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
132 } else { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
133 if( defaultFieldParser != li.defaultFieldParser ) |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
134 throw new LuanException("default_type doesn't match previous use"); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
135 if( !Arrays.equals(defaultFields,li.defaultFields) ) |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
136 throw new LuanException("default_fields don't match previous use"); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
137 li.openCount++; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
138 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
139 return new Object[]{li,new Closer(luan,li)}; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
140 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
141 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
142 |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
143 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
|
144 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
|
145 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
|
146 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
|
147 |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
148 private final ReentrantLock writeLock = new ReentrantLock(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
149 private final File indexDir; |
754 | 150 private SnapshotDeletionPolicy snapshotDeletionPolicy; |
151 private IndexWriter writer; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
152 private DirectoryReader reader; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
153 private IndexSearcher searcher; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
154 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
|
155 private final MultiFieldParser mfp; |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
156 private final Analyzer analyzer; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
157 |
1391 | 158 private FSDirectory fsDir; |
704 | 159 private int writeCount; |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
160 private AtomicInteger writeCounter = new AtomicInteger(); |
704 | 161 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
162 private Set<String> indexOnly = new HashSet<String>(); |
1271
48f302bdc187
fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents:
1267
diff
changeset
|
163 |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
164 private int openCount; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
165 private final String key; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
166 private final FieldParser defaultFieldParser; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
167 private final String[] defaultFields; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
168 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
169 private final PostgresBackup postgresBackup; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
170 |
1392 | 171 private LuceneIndex(Luan luan,File indexDir,FieldParser defaultFieldParser,String[] defaultFields,String key,LuanTable postgresSpec) |
172 throws LuanException, IOException, ClassNotFoundException, SQLException | |
1267 | 173 { |
1392 | 174 final Logger logger = LuanLogger.getLogger(luan,LuceneIndex.class); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
175 this.key = key; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
176 this.defaultFieldParser = defaultFieldParser; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
177 this.defaultFields = defaultFields; |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
178 mfp = defaultFieldParser==null ? new MultiFieldParser() : new MultiFieldParser(defaultFieldParser,defaultFields); |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
179 mfp.fields.put( "type", STRING_FIELD_PARSER ); |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
180 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
|
181 this.indexDir = indexDir; |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
182 Analyzer analyzer = STRING_FIELD_PARSER.analyzer; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
183 if( defaultFieldParser instanceof StringFieldParser ) { |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
184 StringFieldParser sfp = (StringFieldParser)defaultFieldParser; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
185 analyzer = sfp.analyzer; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
186 } |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
187 this.analyzer = analyzer; |
1390 | 188 boolean wasCreated = reopen(); |
1392 | 189 if( postgresSpec == null ) { |
190 postgresBackup = null; | |
191 } else { | |
192 Map spec = postgresSpec.asMap(); | |
193 LuanFunction completer = Utils.removeRequiredFunction(spec,"completer"); | |
194 postgresBackup = new PostgresBackup(spec); | |
195 if( postgresBackup != null ) { | |
196 if( !wasCreated && postgresBackup.wasCreated ) { | |
197 logger.error("rebuilding postgres backup"); | |
198 rebuild_postgres_backup(completer); | |
199 } else if( wasCreated && !postgresBackup.wasCreated ) { | |
200 logger.error("restoring from postgres"); | |
201 restore_from_postgres(luan); | |
202 } | |
1390 | 203 } |
204 } | |
754 | 205 } |
206 | |
1390 | 207 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
|
208 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
|
209 snapshotDeletionPolicy = new SnapshotDeletionPolicy(conf.getIndexDeletionPolicy()); |
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
210 conf.setIndexDeletionPolicy(snapshotDeletionPolicy); |
1391 | 211 fsDir = FSDirectory.open(indexDir); |
212 boolean wasCreated = !fsDir.getDirectory().exists(); | |
213 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
|
214 writer.commit(); // commit index creation |
1391 | 215 reader = DirectoryReader.open(fsDir); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
216 searcher = new IndexSearcher(reader); |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
217 initId(); |
1390 | 218 return wasCreated; |
233
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
219 } |
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
220 |
704 | 221 private void wrote() { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
222 writeCounter.incrementAndGet(); |
704 | 223 } |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
224 |
1392 | 225 public void delete_all() throws IOException, SQLException { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
226 boolean commit = !writeLock.isHeldByCurrentThread(); |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
227 writeLock.lock(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
228 try { |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
229 writer.deleteAll(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
230 id = idLim = 0; |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
231 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
232 postgresBackup.deleteAll(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
233 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
234 } finally { |
704 | 235 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
236 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
237 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
238 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
239 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
240 private static Term term(String key,long value) { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
241 BytesRef br = new BytesRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
242 NumericUtils.longToPrefixCoded(value,0,br); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
243 return new Term(key,br); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
244 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
245 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
246 private void backupDelete(Query query) |
1392 | 247 throws IOException, SQLException, LuanException |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
248 { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
249 if( postgresBackup != null ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
250 final List<Long> ids = new ArrayList<Long>(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
251 IndexSearcher searcher = openSearcher(); |
1391 | 252 try { |
253 MyCollector col = new MyCollector() { | |
254 @Override public void collect(int iDoc) throws IOException { | |
255 Document doc = searcher.doc( docBase + iDoc ); | |
256 Long id = (Long)doc.getField("id").numericValue(); | |
257 ids.add(id); | |
258 } | |
259 }; | |
260 searcher.search(query,col); | |
261 } finally { | |
262 close(searcher); | |
263 } | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
264 postgresBackup.begin(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
265 for( Long id : ids ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
266 postgresBackup.delete(id); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
267 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
268 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
269 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
270 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
271 |
1346 | 272 public void delete(String queryStr) |
1392 | 273 throws IOException, ParseException, SQLException, LuanException |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
274 { |
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
|
275 Query query = SaneQueryParser.parseQuery(mfp,queryStr); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
276 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
277 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
278 writeLock.lock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
279 try { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
280 backupDelete(query); |
547
0be287ab0309
add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents:
546
diff
changeset
|
281 writer.deleteDocuments(query); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
282 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
283 } finally { |
704 | 284 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
285 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
286 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
287 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
288 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
289 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
|
290 indexOnly.addAll(fields); |
1271
48f302bdc187
fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents:
1267
diff
changeset
|
291 } |
48f302bdc187
fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents:
1267
diff
changeset
|
292 |
1346 | 293 public void save(LuanTable doc,LuanTable boosts) |
1392 | 294 throws LuanException, IOException, SQLException |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
295 { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
296 if( boosts!=null && postgresBackup!=null ) |
1392 | 297 throw new LuanException("boosts are not saved to postgres backup"); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
298 |
1267 | 299 Object obj = doc.get("id"); |
601 | 300 Long id; |
301 try { | |
302 id = (Long)obj; | |
303 } catch(ClassCastException e) { | |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
304 throw new LuanException("id should be Long but is "+obj.getClass().getSimpleName()); |
601 | 305 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
306 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
307 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
308 writeLock.lock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
309 try { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
310 if( id == null ) { |
1346 | 311 id = nextId(); |
1267 | 312 doc.put("id",id); |
1391 | 313 if( postgresBackup != null ) |
314 postgresBackup.add(doc); | |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
315 writer.addDocument(toLucene(doc,boosts)); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
316 } else { |
1391 | 317 if( postgresBackup != null ) |
318 postgresBackup.update(doc); | |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
319 writer.updateDocument( term("id",id), toLucene(doc,boosts) ); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
320 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
321 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
322 } finally { |
704 | 323 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
324 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
325 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
326 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
327 |
1392 | 328 public Object run_in_transaction(LuanFunction fn) |
329 throws IOException, LuanException, SQLException | |
330 { | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
331 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
332 writeLock.lock(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
333 boolean ok = false; |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
334 try { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
335 if( commit && postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
336 postgresBackup.begin(); |
1379 | 337 Object rtn = fn.call(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
338 ok = true; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
339 if(commit) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
340 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
341 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
342 writer.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
343 } |
1379 | 344 return rtn; |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
345 } finally { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
346 if( !ok && commit ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
347 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
348 postgresBackup.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
349 writer.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
350 reopen(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
351 } |
704 | 352 wrote(); |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
353 writeLock.unlock(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
354 } |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
355 } |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
356 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
357 // ??? |
1379 | 358 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
|
359 if( writeLock.isHeldByCurrentThread() ) |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
360 throw new RuntimeException(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
361 writeLock.lock(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
362 try { |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
363 synchronized(this) { |
1379 | 364 return fn.call(); |
756
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
365 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
366 } finally { |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
367 wrote(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
368 writeLock.unlock(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
369 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
370 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
371 |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
372 |
754 | 373 private long id; |
374 private long idLim; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
375 private final int idBatch = 10; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
376 |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
377 private void initId() throws IOException { |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
378 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
|
379 switch(td.totalHits) { |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
380 case 0: |
754 | 381 id = 0; |
382 idLim = 0; | |
383 break; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
384 case 1: |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
385 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
|
386 id = idLim; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
387 break; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
388 default: |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
389 throw new RuntimeException(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
390 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
391 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
392 |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
393 private void saveNextId(long nextId) throws LuanException, IOException { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
394 Map doc = new HashMap(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
395 doc.put( "type", "next_id" ); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
396 doc.put( FLD_NEXT_ID, idLim ); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
397 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
|
398 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
399 |
1346 | 400 public synchronized long nextId() throws LuanException, IOException { |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
401 if( ++id > idLim ) { |
233
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
402 idLim += idBatch; |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
403 saveNextId(idLim); |
704 | 404 wrote(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
405 } |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
406 return id; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
407 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
408 |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
409 /* |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
410 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
|
411 if( !zipFile.endsWith(".zip") ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
412 throw new LuanException("file "+zipFile+" doesn't end with '.zip'"); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
413 IndexCommit ic = snapshotDeletionPolicy.snapshot(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
414 try { |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
415 ZipOutputStream out = new ZipOutputStream(new FileOutputStream(zipFile)); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
416 for( String fileName : ic.getFileNames() ) { |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
417 out.putNextEntry(new ZipEntry(fileName)); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
418 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
|
419 Utils.copyAll(in,out); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
420 in.close(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
421 out.closeEntry(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
422 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
423 out.close(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
424 } finally { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
425 snapshotDeletionPolicy.release(ic); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
426 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
427 } |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
428 */ |
1129
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
429 public SnapshotDeletionPolicy snapshotDeletionPolicy() { |
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
430 return snapshotDeletionPolicy; |
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
431 } |
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
432 |
1346 | 433 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
|
434 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
|
435 try { |
1391 | 436 String dir = fsDir.getDirectory().toString(); |
1346 | 437 LuanTable fileNames = new LuanTable(fn.luan(),new ArrayList(ic.getFileNames())); |
1335 | 438 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
|
439 } finally { |
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
440 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
|
441 } |
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
442 } |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
443 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
444 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
445 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
446 public String to_string() { |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
447 return writer.getDirectory().toString(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
448 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
449 |
1392 | 450 private synchronized void close() throws IOException, SQLException { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
451 if( openCount > 0 ) { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
452 if( --openCount == 0 ) { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
453 doClose(); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
454 synchronized(indexes) { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
455 indexes.remove(key); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
456 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
457 } |
521
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
458 } |
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
459 } |
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
460 |
1392 | 461 public void doClose() throws IOException, SQLException { |
462 writer.close(); | |
463 reader.close(); | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
464 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
465 postgresBackup.close(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
466 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
467 |
312
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
468 |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
469 private static class DocFn extends LuanFunction { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
470 final IndexSearcher searcher; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
471 final Query query; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
472 int docID; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
473 |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
474 DocFn(Luan luan,IndexSearcher searcher,Query query) { |
1335 | 475 super(luan); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
476 this.searcher = searcher; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
477 this.query = query; |
545
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 |
1335 | 480 @Override public Object call(Object[] args) throws LuanException { |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
481 try { |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
482 LuanTable doc = toTable(luan(),searcher.doc(docID)); |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
483 if( args.length > 0 && "explain".equals(args[0]) ) { |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
484 Explanation explanation = searcher.explain(query,docID); |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
485 return new Object[]{doc,explanation}; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
486 } else { |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
487 return doc; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
488 } |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
489 } catch(IOException e) { |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
490 throw new LuanException(e); |
545
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 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
494 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
495 private static abstract class MyCollector extends Collector { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
496 int docBase; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
497 int i = 0; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
498 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
499 @Override public void setScorer(Scorer scorer) {} |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
500 @Override public void setNextReader(AtomicReaderContext context) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
501 this.docBase = context.docBase; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
502 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
503 @Override public boolean acceptsDocsOutOfOrder() { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
504 return true; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
505 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
506 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
507 |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
508 private synchronized IndexSearcher openSearcher() throws IOException { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
509 int gwc = writeCounter.get(); |
704 | 510 if( writeCount != gwc ) { |
511 writeCount = gwc; | |
512 DirectoryReader newReader = DirectoryReader.openIfChanged(reader); | |
513 if( newReader != null ) { | |
514 reader.decRef(); | |
515 reader = newReader; | |
516 searcher = new IndexSearcher(reader); | |
517 } | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
518 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
519 reader.incRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
520 return searcher; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
521 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
522 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
523 // call in finally block |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
524 private static void close(IndexSearcher searcher) throws IOException { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
525 searcher.getIndexReader().decRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
526 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
527 |
591
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
528 public void ensure_open() throws IOException { |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
529 close(openSearcher()); |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
530 } |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
531 |
1346 | 532 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
|
533 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
|
534 { |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
535 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
|
536 Query query = SaneQueryParser.parseQuery(mfp,queryStr); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
537 IndexSearcher searcher = threadLocalSearcher.get(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
538 boolean inTransaction = searcher != null; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
539 if( !inTransaction ) |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
540 searcher = openSearcher(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
541 try { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
542 if( fn!=null && n==null ) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
543 if( sortStr != null ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
544 throw new LuanException("sort must be nil when n is nil"); |
1346 | 545 final DocFn docFn = new DocFn(fn.luan(),searcher,query); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
546 MyCollector col = new MyCollector() { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
547 @Override public void collect(int doc) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
548 try { |
547
0be287ab0309
add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents:
546
diff
changeset
|
549 docFn.docID = docBase + doc; |
1335 | 550 fn.call(++i,docFn); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
551 } catch(LuanException e) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
552 throw new LuanRuntimeException(e); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
553 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
554 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
555 }; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
556 try { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
557 searcher.search(query,col); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
558 } catch(LuanRuntimeException e) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
559 throw (LuanException)e.getCause(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
560 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
561 return col.i; |
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 if( fn==null || n==0 ) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
564 TotalHitCountCollector thcc = new TotalHitCountCollector(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
565 searcher.search(query,thcc); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
566 return thcc.getTotalHits(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
567 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
568 Sort sort = sortStr==null ? null : SaneQueryParser.parseSort(mfp,sortStr); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
569 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
|
570 final ScoreDoc[] scoreDocs = td.scoreDocs; |
1346 | 571 DocFn docFn = new DocFn(fn.luan(),searcher,query); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
572 for( int i=0; i<scoreDocs.length; i++ ) { |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
573 ScoreDoc scoreDoc = scoreDocs[i]; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
574 docFn.docID = scoreDoc.doc; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
575 fn.call(i+1,docFn,scoreDoc.score); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
576 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
577 return td.totalHits; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
578 } finally { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
579 if( !inTransaction ) |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
580 close(searcher); |
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 |
1335 | 584 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
|
585 if( threadLocalSearcher.get() != null ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
586 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
|
587 IndexSearcher searcher = openSearcher(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
588 threadLocalSearcher.set(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
589 try { |
1335 | 590 return fn.call(); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
591 } finally { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
592 threadLocalSearcher.set(null); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
593 close(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
594 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
595 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
596 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
597 |
796 | 598 public FieldParser getIndexedFieldParser(String field) { |
599 return mfp.fields.get(field); | |
600 } | |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
601 |
796 | 602 public void setIndexedFieldParser(String field,FieldParser fp) { |
603 if( fp==null ) { // delete | |
604 mfp.fields.remove(field); | |
605 return; | |
606 } | |
607 mfp.fields.put( field, fp ); | |
608 } | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
609 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
610 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
611 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
|
612 throws LuanException |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
613 { |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
614 boolean hasBoost = boost!=null; |
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
615 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
|
616 if( hasBoost ) |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
617 ((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
|
618 return fld; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
619 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
620 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
621 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
|
622 throws LuanException |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
623 { |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
624 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
|
625 if( value instanceof String ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
626 String s = (String)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
627 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
|
628 if( fp != null ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
629 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
|
630 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
|
631 } else if (hasBoost) { |
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
632 // fuck you modern lucene developers |
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
633 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
|
634 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
635 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
|
636 } |
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, s); |
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 Integer ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
641 int i = (Integer)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 IntField(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 Long ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
648 long i = (Long)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
649 if( indexed.contains(name) ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
650 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
|
651 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
652 return new StoredField(name, i); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
653 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
654 } else if( value instanceof Double ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
655 double i = (Double)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
656 if( indexed.contains(name) ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
657 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
|
658 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
659 return new StoredField(name, i); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
660 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
661 } else if( value instanceof byte[] ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
662 byte[] b = (byte[])value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
663 return new StoredField(name, b); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
664 } else |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
665 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
|
666 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
667 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
668 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
|
669 return toLucene(table.iterable(),boosts); |
1346 | 670 } |
671 | |
672 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
|
673 Set<String> indexed = mfp.fields.keySet(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
674 Document doc = new Document(); |
1346 | 675 for( Map.Entry<Object,Object> entry : iterable ) { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
676 Object key = entry.getKey(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
677 if( !(key instanceof String) ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
678 throw new LuanException("key must be string"); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
679 String name = (String)key; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
680 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
|
681 Float boost = null; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
682 if( boosts != null ) { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
683 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
|
684 if( obj != null ) { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
685 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
|
686 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
|
687 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
|
688 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
689 } |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
690 if( !(value instanceof LuanTable) ) { |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
691 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
|
692 } else { // list |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
693 LuanTable list = (LuanTable)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
694 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
|
695 doc.add(newField( name, el, indexed, boost )); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
696 } |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
697 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
698 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
699 return doc; |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
700 } |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
701 |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
702 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
|
703 BytesRef br = ifld.binaryValue(); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
704 if( br != null ) |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
705 return br.bytes; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
706 Number n = ifld.numericValue(); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
707 if( n != null ) |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
708 return n; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
709 String s = ifld.stringValue(); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
710 if( s != null ) |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
711 return s; |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
712 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
|
713 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
714 |
1333
25746915a241
merge Luan and LuanState
Franklin Schmidt <fschmidt@gmail.com>
parents:
1271
diff
changeset
|
715 private static LuanTable toTable(Luan luan,Document doc) throws LuanException { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
716 if( doc==null ) |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
717 return null; |
1267 | 718 LuanTable table = new LuanTable(luan); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
719 for( IndexableField ifld : doc ) { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
720 String name = ifld.name(); |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
721 Object value = getValue(ifld); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
722 Object old = table.rawGet(name); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
723 if( old == null ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
724 table.rawPut(name,value); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
725 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
726 LuanTable list; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
727 if( old instanceof LuanTable ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
728 list = (LuanTable)old; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
729 } else { |
1267 | 730 list = new LuanTable(luan); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
731 list.rawPut(1,old); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
732 table.rawPut(name,list); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
733 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
734 list.rawPut(list.rawLength()+1,value); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
735 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
736 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
737 return table; |
312
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
738 } |
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
739 |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
740 |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
741 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
|
742 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
|
743 return originalText; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
744 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
745 }; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
746 |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
747 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
|
748 throws ParseException |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
749 { |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
750 Query query = SaneQueryParser.parseQuery(mfp,queryStr); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
751 Formatter fmt = new Formatter() { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
752 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
|
753 if( tokenGroup.getTotalScore() <= 0 ) |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
754 return originalText; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
755 try { |
1335 | 756 return (String)Luan.first(formatter.call(originalText)); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
757 } catch(LuanException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
758 throw new LuanRuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
759 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
760 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
761 }; |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
762 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
|
763 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
|
764 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
|
765 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
|
766 final Highlighter hl = new Highlighter(fmt,queryScorer); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
767 hl.setTextFragmenter( new NullFragmenter() ); |
1335 | 768 return new LuanFunction(false) { // ??? |
769 @Override public String call(Object[] args) throws LuanException { | |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
770 String text = (String)args[0]; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
771 try { |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
772 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
|
773 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
|
774 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
|
775 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
|
776 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
|
777 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
|
778 if( !atStart ) |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
779 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
|
780 if( !atEnd ) |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
781 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
|
782 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
783 text = s; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
784 } 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
|
785 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
|
786 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
|
787 text += "..."; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
788 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
789 } |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
790 String s = hl.getBestFragment(analyzer,null,text); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
791 return s!=null ? s : text; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
792 } catch(LuanRuntimeException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
793 throw (LuanException)e.getCause(); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
794 } catch(IOException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
795 throw new RuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
796 } catch(InvalidTokenOffsetsException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
797 throw new RuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
798 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
799 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
800 }; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
801 } |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
802 |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
803 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
|
804 throws IOException |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
805 { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
806 int n = 0; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
807 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
|
808 ts.reset(); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
809 while( ts.incrementToken() ) { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
810 n++; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
811 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
812 ts.close(); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
813 return n; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
814 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
815 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
816 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
817 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
818 public boolean hasPostgresBackup() { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
819 return postgresBackup != null; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
820 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
821 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
822 public void rebuild_postgres_backup(LuanFunction completer) |
1392 | 823 throws IOException, LuanException, SQLException |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
824 { |
1392 | 825 final Logger logger = LuanLogger.getLogger(completer.luan(),LuceneIndex.class); |
826 logger.info("start rebuild_postgres_backup"); | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
827 writeLock.lock(); |
1391 | 828 IndexSearcher searcher = openSearcher(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
829 boolean ok = false; |
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 postgresBackup.begin(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
832 postgresBackup.deleteAll(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
833 Query query = new PrefixQuery(new Term("id")); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
834 MyCollector col = new MyCollector() { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
835 @Override public void collect(int iDoc) throws IOException { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
836 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
837 Document doc = searcher.doc( docBase + iDoc ); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
838 LuanTable tbl = toTable(completer.luan(),doc); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
839 tbl = (LuanTable)completer.call(tbl); |
1391 | 840 postgresBackup.add(tbl); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
841 } catch(LuanException e) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
842 throw new LuanRuntimeException(e); |
1392 | 843 } catch(SQLException e) { |
844 throw new RuntimeException(e); | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
845 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
846 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
847 }; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
848 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
849 searcher.search(query,col); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
850 } catch(LuanRuntimeException e) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
851 throw (LuanException)e.getCause(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
852 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
853 ok = true; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
854 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
855 } finally { |
1391 | 856 close(searcher); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
857 if( !ok ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
858 postgresBackup.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
859 writeLock.unlock(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
860 } |
1392 | 861 logger.info("end rebuild_postgres_backup"); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
862 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
863 |
1392 | 864 public void restore_from_postgres(Luan luan) |
865 throws IOException, LuanException, SQLException | |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
866 { |
1392 | 867 final Logger logger = LuanLogger.getLogger(luan,LuceneIndex.class); |
868 logger.warn("start restore_from_postgres"); | |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
869 if( postgresBackup==null ) |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
870 throw new NullPointerException(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
871 if( writeLock.isHeldByCurrentThread() ) |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
872 throw new RuntimeException(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
873 writeLock.lock(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
874 boolean ok = false; |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
875 try { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
876 writer.deleteAll(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
877 long nextId = postgresBackup.maxId() + 1; |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
878 postgresBackup.restoreLucene(this); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
879 id = idLim = nextId; |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
880 ok = true; |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
881 writer.commit(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
882 } finally { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
883 if( !ok ) { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
884 writer.rollback(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
885 reopen(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
886 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
887 wrote(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
888 writeLock.unlock(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
889 } |
1392 | 890 logger.warn("end restore_from_postgres"); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
891 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
892 |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
893 void restore(LuanTable doc) |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
894 throws LuanException, IOException |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
895 { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
896 writer.addDocument(toLucene(doc,null)); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
897 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
898 |
1391 | 899 public void check(LuanFunction completer) throws IOException, SQLException, LuanException { |
1392 | 900 final Logger logger = LuanLogger.getLogger(completer.luan(),LuceneIndex.class); |
1391 | 901 logger.info("start check"); |
902 CheckIndex.Status status = new CheckIndex(fsDir).checkIndex(); | |
903 if( !status.clean ) | |
904 logger.error("index not clean"); | |
905 if( postgresBackup != null ) | |
906 checkPostgres(completer); | |
907 logger.info("end check"); | |
908 } | |
909 | |
910 private void checkPostgres(LuanFunction completer) throws IOException, SQLException, LuanException { | |
1392 | 911 final Logger logger = LuanLogger.getLogger(completer.luan(),LuceneIndex.class); |
1391 | 912 final PostgresBackup.Checker postgresChecker; |
913 final IndexSearcher searcher; | |
914 writeLock.lock(); | |
915 try { | |
916 postgresChecker = postgresBackup.newChecker(); | |
917 searcher = openSearcher(); | |
918 } finally { | |
919 writeLock.unlock(); | |
920 } | |
921 try { | |
922 final List<Long> idsLucene = new ArrayList<Long>(); | |
923 Query query = new PrefixQuery(new Term("id")); | |
924 MyCollector col = new MyCollector() { | |
925 @Override public void collect(int iDoc) throws IOException { | |
926 Document doc = searcher.doc( docBase + iDoc ); | |
927 Long id = (Long)doc.getField("id").numericValue(); | |
928 idsLucene.add(id); | |
929 } | |
930 }; | |
931 searcher.search(query,col); | |
932 Collections.sort(idsLucene); | |
933 final List<Long> idsPostgres = postgresChecker.getIds(); | |
934 final int nLucene = idsLucene.size(); | |
935 final int nPostgres = idsPostgres.size(); | |
936 int iLucene = 0; | |
937 int iPostgres = 0; | |
938 LuanToString lts = new LuanToString(); | |
939 lts.strict = true; | |
940 lts.numberTypes = true; | |
941 while( iLucene < nLucene && iPostgres < nPostgres ) { | |
942 long idLucene = idsLucene.get(iLucene); | |
943 long idPostgres = idsPostgres.get(iPostgres); | |
944 if( idLucene < idPostgres ) { | |
945 iLucene++; | |
946 logger.error("id "+idLucene+" found in lucene but not postgres"); | |
947 } else if( idLucene > idPostgres ) { | |
948 iPostgres++; | |
949 logger.error("id "+idPostgres+" found in postgres but not lucene"); | |
950 } else { // == | |
951 LuanTable docPostgres = postgresChecker.getDoc(idPostgres); | |
952 TopDocs td = searcher.search(new TermQuery(term("id",idLucene)),1); | |
953 if( td.totalHits != 1 ) throw new RuntimeException(); | |
954 Document doc = searcher.doc( td.scoreDocs[0].doc ); | |
955 LuanTable docLucene = toTable(completer.luan(),doc); | |
956 docLucene = (LuanTable)completer.call(docLucene); | |
957 if( !equal(docPostgres,docLucene) ) { | |
958 logger.error("id "+idLucene+" not equal"); | |
959 logger.error("lucene = "+lts.toString(docLucene)); | |
960 logger.error("postgres = "+lts.toString(docPostgres)); | |
961 } | |
962 iLucene++; | |
963 iPostgres++; | |
964 } | |
965 } | |
966 while( iLucene < nLucene ) { | |
967 long idLucene = idsLucene.get(iLucene++); | |
968 logger.error("id "+idLucene+" found in lucene but not postgres"); | |
969 } | |
970 while( iPostgres < nPostgres ) { | |
971 long idPostgres = idsPostgres.get(iPostgres++); | |
972 logger.error("id "+idPostgres+" found in postgres but not lucene"); | |
973 } | |
974 } finally { | |
975 close(searcher); | |
976 postgresChecker.close(); | |
977 } | |
978 } | |
979 | |
980 private boolean equal(LuanTable t1,LuanTable t2) throws LuanException { | |
981 return t1.asMap().equals(t2.asMap()); | |
982 } | |
983 | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
984 } |