Mercurial Hosting > luan
annotate src/luan/modules/lucene/LuceneIndex.java @ 1395:9dfff82dfc59
finish postgres work
author | Franklin Schmidt <fschmidt@gmail.com> |
---|---|
date | Mon, 09 Sep 2019 01:22:23 -0600 |
parents | cc0dbca576dc |
children | 0dc9837c16be |
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; |
1337 | 82 import luan.lib.logging.Logger; |
83 import luan.lib.logging.LoggerFactory; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
84 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
85 |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
86 public final class LuceneIndex { |
1393 | 87 private static final Logger sysLogger = LoggerFactory.getLogger(LuceneIndex.class); |
521
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
88 |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
89 private static final class Closer implements Closeable { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
90 final LuceneIndex li; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
91 boolean isClosed = false; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
92 private final Exception created = new Exception("created"); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
93 |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
94 Closer(Luan luan,LuceneIndex li) { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
95 this.li = li; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
96 luan.onClose(this); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
97 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
98 |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
99 public void close() throws IOException { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
100 if( !isClosed ) { |
1392 | 101 try { |
102 li.close(); | |
103 } catch(SQLException e) { | |
104 throw new RuntimeException(e); | |
105 } | |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
106 isClosed = true; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
107 } |
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 protected void finalize() throws Throwable { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
111 if( !isClosed ) { |
1393 | 112 sysLogger.error("not closed",created); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
113 close(); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
114 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
115 super.finalize(); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
116 } |
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 private static Map<String,LuceneIndex> indexes = new HashMap<String,LuceneIndex>(); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
120 |
1395
9dfff82dfc59
finish postgres work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1393
diff
changeset
|
121 public static Object[] getLuceneIndex(Luan luan,File indexDir,FieldParser defaultFieldParser,String[] defaultFields,LuanFunction completer,LuanTable postgresSpec) |
1392 | 122 throws LuanException, IOException, ClassNotFoundException, SQLException |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
123 { |
1369
709f7498a363
change Lucene.index() and add Lucene.recover()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1347
diff
changeset
|
124 String key = indexDir.getCanonicalPath(); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
125 synchronized(indexes) { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
126 LuceneIndex li = indexes.get(key); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
127 if( li == null ) { |
1395
9dfff82dfc59
finish postgres work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1393
diff
changeset
|
128 li = new LuceneIndex(luan,indexDir,defaultFieldParser,defaultFields,key,completer,postgresSpec); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
129 li.openCount = 1; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
130 indexes.put(key,li); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
131 } else { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
132 if( defaultFieldParser != li.defaultFieldParser ) |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
133 throw new LuanException("default_type doesn't match previous use"); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
134 if( !Arrays.equals(defaultFields,li.defaultFields) ) |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
135 throw new LuanException("default_fields don't match previous use"); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
136 li.openCount++; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
137 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
138 return new Object[]{li,new Closer(luan,li)}; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
139 } |
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 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
|
143 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
|
144 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
|
145 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
|
146 |
1393 | 147 private final Logger luanLogger; |
148 | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
149 private final ReentrantLock writeLock = new ReentrantLock(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
150 private final File indexDir; |
754 | 151 private SnapshotDeletionPolicy snapshotDeletionPolicy; |
152 private IndexWriter writer; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
153 private DirectoryReader reader; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
154 private IndexSearcher searcher; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
155 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
|
156 private final MultiFieldParser mfp; |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
157 private final Analyzer analyzer; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
158 |
1391 | 159 private FSDirectory fsDir; |
704 | 160 private int writeCount; |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
161 private AtomicInteger writeCounter = new AtomicInteger(); |
704 | 162 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
163 private Set<String> indexOnly = new HashSet<String>(); |
1271
48f302bdc187
fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents:
1267
diff
changeset
|
164 |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
165 private int openCount; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
166 private final String key; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
167 private final FieldParser defaultFieldParser; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
168 private final String[] defaultFields; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
169 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
170 private final PostgresBackup postgresBackup; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
171 |
1395
9dfff82dfc59
finish postgres work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1393
diff
changeset
|
172 private LuceneIndex(Luan luan,File indexDir,FieldParser defaultFieldParser,String[] defaultFields,String key,LuanFunction completer,LuanTable postgresSpec) |
1392 | 173 throws LuanException, IOException, ClassNotFoundException, SQLException |
1267 | 174 { |
1393 | 175 this.luanLogger = luan.getLogger(LuceneIndex.class); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
176 this.key = key; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
177 this.defaultFieldParser = defaultFieldParser; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
178 this.defaultFields = defaultFields; |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
179 mfp = defaultFieldParser==null ? new MultiFieldParser() : new MultiFieldParser(defaultFieldParser,defaultFields); |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
180 mfp.fields.put( "type", STRING_FIELD_PARSER ); |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
181 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
|
182 this.indexDir = indexDir; |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
183 Analyzer analyzer = STRING_FIELD_PARSER.analyzer; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
184 if( defaultFieldParser instanceof StringFieldParser ) { |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
185 StringFieldParser sfp = (StringFieldParser)defaultFieldParser; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
186 analyzer = sfp.analyzer; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
187 } |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
188 this.analyzer = analyzer; |
1390 | 189 boolean wasCreated = reopen(); |
1392 | 190 if( postgresSpec == null ) { |
191 postgresBackup = null; | |
192 } else { | |
1395
9dfff82dfc59
finish postgres work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1393
diff
changeset
|
193 if( completer == null ) |
9dfff82dfc59
finish postgres work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1393
diff
changeset
|
194 throw new LuanException("completer is required for postgres_spec"); |
1392 | 195 Map spec = postgresSpec.asMap(); |
1393 | 196 postgresBackup = new PostgresBackup(luan,spec); |
1392 | 197 if( postgresBackup != null ) { |
198 if( !wasCreated && postgresBackup.wasCreated ) { | |
1393 | 199 luanLogger.error("rebuilding postgres backup"); |
1392 | 200 rebuild_postgres_backup(completer); |
201 } else if( wasCreated && !postgresBackup.wasCreated ) { | |
1393 | 202 luanLogger.error("restoring from postgres"); |
203 restore_from_postgres(); | |
1392 | 204 } |
1390 | 205 } |
206 } | |
754 | 207 } |
208 | |
1390 | 209 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
|
210 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
|
211 snapshotDeletionPolicy = new SnapshotDeletionPolicy(conf.getIndexDeletionPolicy()); |
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
212 conf.setIndexDeletionPolicy(snapshotDeletionPolicy); |
1391 | 213 fsDir = FSDirectory.open(indexDir); |
214 boolean wasCreated = !fsDir.getDirectory().exists(); | |
215 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
|
216 writer.commit(); // commit index creation |
1391 | 217 reader = DirectoryReader.open(fsDir); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
218 searcher = new IndexSearcher(reader); |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
219 initId(); |
1390 | 220 return wasCreated; |
233
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
221 } |
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
222 |
704 | 223 private void wrote() { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
224 writeCounter.incrementAndGet(); |
704 | 225 } |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
226 |
1392 | 227 public void delete_all() throws IOException, SQLException { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
228 boolean commit = !writeLock.isHeldByCurrentThread(); |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
229 writeLock.lock(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
230 try { |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
231 writer.deleteAll(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
232 id = idLim = 0; |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
233 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
234 postgresBackup.deleteAll(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
235 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
236 } finally { |
704 | 237 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
238 writeLock.unlock(); |
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 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
241 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
242 private static Term term(String key,long value) { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
243 BytesRef br = new BytesRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
244 NumericUtils.longToPrefixCoded(value,0,br); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
245 return new Term(key,br); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
246 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
247 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
248 private void backupDelete(Query query) |
1392 | 249 throws IOException, SQLException, LuanException |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
250 { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
251 if( postgresBackup != null ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
252 final List<Long> ids = new ArrayList<Long>(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
253 IndexSearcher searcher = openSearcher(); |
1391 | 254 try { |
255 MyCollector col = new MyCollector() { | |
256 @Override public void collect(int iDoc) throws IOException { | |
257 Document doc = searcher.doc( docBase + iDoc ); | |
258 Long id = (Long)doc.getField("id").numericValue(); | |
259 ids.add(id); | |
260 } | |
261 }; | |
262 searcher.search(query,col); | |
263 } finally { | |
264 close(searcher); | |
265 } | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
266 postgresBackup.begin(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
267 for( Long id : ids ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
268 postgresBackup.delete(id); |
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 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
271 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
272 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
273 |
1346 | 274 public void delete(String queryStr) |
1392 | 275 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
|
276 { |
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
|
277 Query query = SaneQueryParser.parseQuery(mfp,queryStr); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
278 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
279 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
280 writeLock.lock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
281 try { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
282 backupDelete(query); |
547
0be287ab0309
add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents:
546
diff
changeset
|
283 writer.deleteDocuments(query); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
284 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
285 } finally { |
704 | 286 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
287 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
288 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
289 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
290 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
291 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
|
292 indexOnly.addAll(fields); |
1271
48f302bdc187
fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents:
1267
diff
changeset
|
293 } |
48f302bdc187
fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents:
1267
diff
changeset
|
294 |
1346 | 295 public void save(LuanTable doc,LuanTable boosts) |
1392 | 296 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
|
297 { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
298 if( boosts!=null && postgresBackup!=null ) |
1392 | 299 throw new LuanException("boosts are not saved to postgres backup"); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
300 |
1267 | 301 Object obj = doc.get("id"); |
601 | 302 Long id; |
303 try { | |
304 id = (Long)obj; | |
305 } catch(ClassCastException e) { | |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
306 throw new LuanException("id should be Long but is "+obj.getClass().getSimpleName()); |
601 | 307 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
308 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
309 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
310 writeLock.lock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
311 try { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
312 if( id == null ) { |
1346 | 313 id = nextId(); |
1267 | 314 doc.put("id",id); |
1391 | 315 if( postgresBackup != null ) |
316 postgresBackup.add(doc); | |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
317 writer.addDocument(toLucene(doc,boosts)); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
318 } else { |
1391 | 319 if( postgresBackup != null ) |
320 postgresBackup.update(doc); | |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
321 writer.updateDocument( term("id",id), toLucene(doc,boosts) ); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
322 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
323 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
324 } finally { |
704 | 325 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
326 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
327 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
328 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
329 |
1392 | 330 public Object run_in_transaction(LuanFunction fn) |
331 throws IOException, LuanException, SQLException | |
332 { | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
333 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
334 writeLock.lock(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
335 boolean ok = false; |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
336 try { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
337 if( commit && postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
338 postgresBackup.begin(); |
1379 | 339 Object rtn = fn.call(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
340 ok = true; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
341 if(commit) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
342 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
343 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
344 writer.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
345 } |
1379 | 346 return rtn; |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
347 } finally { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
348 if( !ok && commit ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
349 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
350 postgresBackup.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
351 writer.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
352 reopen(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
353 } |
704 | 354 wrote(); |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
355 writeLock.unlock(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
356 } |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
357 } |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
358 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
359 // ??? |
1379 | 360 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
|
361 if( writeLock.isHeldByCurrentThread() ) |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
362 throw new RuntimeException(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
363 writeLock.lock(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
364 try { |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
365 synchronized(this) { |
1379 | 366 return fn.call(); |
756
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
367 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
368 } finally { |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
369 wrote(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
370 writeLock.unlock(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
371 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
372 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
373 |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
374 |
754 | 375 private long id; |
376 private long idLim; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
377 private final int idBatch = 10; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
378 |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
379 private void initId() throws IOException { |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
380 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
|
381 switch(td.totalHits) { |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
382 case 0: |
754 | 383 id = 0; |
384 idLim = 0; | |
385 break; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
386 case 1: |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
387 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
|
388 id = idLim; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
389 break; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
390 default: |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
391 throw new RuntimeException(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
392 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
393 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
394 |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
395 private void saveNextId(long nextId) throws LuanException, IOException { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
396 Map doc = new HashMap(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
397 doc.put( "type", "next_id" ); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
398 doc.put( FLD_NEXT_ID, idLim ); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
399 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
|
400 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
401 |
1346 | 402 public synchronized long nextId() throws LuanException, IOException { |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
403 if( ++id > idLim ) { |
233
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
404 idLim += idBatch; |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
405 saveNextId(idLim); |
704 | 406 wrote(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
407 } |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
408 return id; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
409 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
410 |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
411 /* |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
412 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
|
413 if( !zipFile.endsWith(".zip") ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
414 throw new LuanException("file "+zipFile+" doesn't end with '.zip'"); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
415 IndexCommit ic = snapshotDeletionPolicy.snapshot(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
416 try { |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
417 ZipOutputStream out = new ZipOutputStream(new FileOutputStream(zipFile)); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
418 for( String fileName : ic.getFileNames() ) { |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
419 out.putNextEntry(new ZipEntry(fileName)); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
420 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
|
421 Utils.copyAll(in,out); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
422 in.close(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
423 out.closeEntry(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
424 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
425 out.close(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
426 } finally { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
427 snapshotDeletionPolicy.release(ic); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
428 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
429 } |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
430 */ |
1129
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
431 public SnapshotDeletionPolicy snapshotDeletionPolicy() { |
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
432 return snapshotDeletionPolicy; |
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
433 } |
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
434 |
1346 | 435 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
|
436 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
|
437 try { |
1391 | 438 String dir = fsDir.getDirectory().toString(); |
1346 | 439 LuanTable fileNames = new LuanTable(fn.luan(),new ArrayList(ic.getFileNames())); |
1335 | 440 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
|
441 } finally { |
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
442 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
|
443 } |
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
444 } |
230
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 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
447 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
448 public String to_string() { |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
449 return writer.getDirectory().toString(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
450 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
451 |
1392 | 452 private synchronized void close() throws IOException, SQLException { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
453 if( openCount > 0 ) { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
454 if( --openCount == 0 ) { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
455 doClose(); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
456 synchronized(indexes) { |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
457 indexes.remove(key); |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
458 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
459 } |
521
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
460 } |
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
461 } |
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
462 |
1392 | 463 public void doClose() throws IOException, SQLException { |
464 writer.close(); | |
465 reader.close(); | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
466 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
467 postgresBackup.close(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
468 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
469 |
312
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
470 |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
471 private static class DocFn extends LuanFunction { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
472 final IndexSearcher searcher; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
473 final Query query; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
474 int docID; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
475 |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
476 DocFn(Luan luan,IndexSearcher searcher,Query query) { |
1335 | 477 super(luan); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
478 this.searcher = searcher; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
479 this.query = query; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
480 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
481 |
1335 | 482 @Override public Object call(Object[] args) throws LuanException { |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
483 try { |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
484 LuanTable doc = toTable(luan(),searcher.doc(docID)); |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
485 if( args.length > 0 && "explain".equals(args[0]) ) { |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
486 Explanation explanation = searcher.explain(query,docID); |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
487 return new Object[]{doc,explanation}; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
488 } else { |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
489 return doc; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
490 } |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
491 } catch(IOException e) { |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
492 throw new LuanException(e); |
545
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 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
496 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
497 private static abstract class MyCollector extends Collector { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
498 int docBase; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
499 int i = 0; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
500 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
501 @Override public void setScorer(Scorer scorer) {} |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
502 @Override public void setNextReader(AtomicReaderContext context) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
503 this.docBase = context.docBase; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
504 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
505 @Override public boolean acceptsDocsOutOfOrder() { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
506 return true; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
507 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
508 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
509 |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
510 private synchronized IndexSearcher openSearcher() throws IOException { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
511 int gwc = writeCounter.get(); |
704 | 512 if( writeCount != gwc ) { |
513 writeCount = gwc; | |
514 DirectoryReader newReader = DirectoryReader.openIfChanged(reader); | |
515 if( newReader != null ) { | |
516 reader.decRef(); | |
517 reader = newReader; | |
518 searcher = new IndexSearcher(reader); | |
519 } | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
520 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
521 reader.incRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
522 return searcher; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
523 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
524 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
525 // call in finally block |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
526 private static void close(IndexSearcher searcher) throws IOException { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
527 searcher.getIndexReader().decRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
528 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
529 |
591
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
530 public void ensure_open() throws IOException { |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
531 close(openSearcher()); |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
532 } |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
533 |
1346 | 534 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
|
535 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
|
536 { |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
537 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
|
538 Query query = SaneQueryParser.parseQuery(mfp,queryStr); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
539 IndexSearcher searcher = threadLocalSearcher.get(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
540 boolean inTransaction = searcher != null; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
541 if( !inTransaction ) |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
542 searcher = openSearcher(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
543 try { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
544 if( fn!=null && n==null ) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
545 if( sortStr != null ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
546 throw new LuanException("sort must be nil when n is nil"); |
1346 | 547 final DocFn docFn = new DocFn(fn.luan(),searcher,query); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
548 MyCollector col = new MyCollector() { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
549 @Override public void collect(int doc) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
550 try { |
547
0be287ab0309
add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents:
546
diff
changeset
|
551 docFn.docID = docBase + doc; |
1335 | 552 fn.call(++i,docFn); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
553 } catch(LuanException e) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
554 throw new LuanRuntimeException(e); |
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 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
557 }; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
558 try { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
559 searcher.search(query,col); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
560 } catch(LuanRuntimeException e) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
561 throw (LuanException)e.getCause(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
562 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
563 return col.i; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
564 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
565 if( fn==null || n==0 ) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
566 TotalHitCountCollector thcc = new TotalHitCountCollector(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
567 searcher.search(query,thcc); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
568 return thcc.getTotalHits(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
569 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
570 Sort sort = sortStr==null ? null : SaneQueryParser.parseSort(mfp,sortStr); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
571 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
|
572 final ScoreDoc[] scoreDocs = td.scoreDocs; |
1346 | 573 DocFn docFn = new DocFn(fn.luan(),searcher,query); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
574 for( int i=0; i<scoreDocs.length; i++ ) { |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
575 ScoreDoc scoreDoc = scoreDocs[i]; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
576 docFn.docID = scoreDoc.doc; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
577 fn.call(i+1,docFn,scoreDoc.score); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
578 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
579 return td.totalHits; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
580 } finally { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
581 if( !inTransaction ) |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
582 close(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
583 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
584 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
585 |
1335 | 586 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
|
587 if( threadLocalSearcher.get() != null ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
588 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
|
589 IndexSearcher searcher = openSearcher(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
590 threadLocalSearcher.set(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
591 try { |
1335 | 592 return fn.call(); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
593 } finally { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
594 threadLocalSearcher.set(null); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
595 close(searcher); |
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 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
598 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
599 |
796 | 600 public FieldParser getIndexedFieldParser(String field) { |
601 return mfp.fields.get(field); | |
602 } | |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
603 |
796 | 604 public void setIndexedFieldParser(String field,FieldParser fp) { |
605 if( fp==null ) { // delete | |
606 mfp.fields.remove(field); | |
607 return; | |
608 } | |
609 mfp.fields.put( field, fp ); | |
610 } | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
611 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
612 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
613 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
|
614 throws LuanException |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
615 { |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
616 boolean hasBoost = boost!=null; |
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
617 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
|
618 if( hasBoost ) |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
619 ((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
|
620 return fld; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
621 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
622 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
623 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
|
624 throws LuanException |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
625 { |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
626 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
|
627 if( value instanceof String ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
628 String s = (String)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
629 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
|
630 if( fp != null ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
631 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
|
632 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
|
633 } else if (hasBoost) { |
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
634 // fuck you modern lucene developers |
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
635 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
|
636 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
637 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
|
638 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
639 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
640 return new StoredField(name, s); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
641 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
642 } else if( value instanceof Integer ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
643 int i = (Integer)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
644 if( indexed.contains(name) ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
645 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
|
646 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
647 return new StoredField(name, i); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
648 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
649 } else if( value instanceof Long ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
650 long i = (Long)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
651 if( indexed.contains(name) ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
652 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
|
653 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
654 return new StoredField(name, i); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
655 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
656 } else if( value instanceof Double ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
657 double i = (Double)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
658 if( indexed.contains(name) ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
659 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
|
660 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
661 return new StoredField(name, i); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
662 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
663 } else if( value instanceof byte[] ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
664 byte[] b = (byte[])value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
665 return new StoredField(name, b); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
666 } else |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
667 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
|
668 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
669 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
670 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
|
671 return toLucene(table.iterable(),boosts); |
1346 | 672 } |
673 | |
674 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
|
675 Set<String> indexed = mfp.fields.keySet(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
676 Document doc = new Document(); |
1346 | 677 for( Map.Entry<Object,Object> entry : iterable ) { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
678 Object key = entry.getKey(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
679 if( !(key instanceof String) ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
680 throw new LuanException("key must be string"); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
681 String name = (String)key; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
682 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
|
683 Float boost = null; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
684 if( boosts != null ) { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
685 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
|
686 if( obj != null ) { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
687 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
|
688 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
|
689 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
|
690 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
691 } |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
692 if( !(value instanceof LuanTable) ) { |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
693 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
|
694 } else { // list |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
695 LuanTable list = (LuanTable)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
696 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
|
697 doc.add(newField( name, el, indexed, boost )); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
698 } |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
699 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
700 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
701 return doc; |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
702 } |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
703 |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
704 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
|
705 BytesRef br = ifld.binaryValue(); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
706 if( br != null ) |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
707 return br.bytes; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
708 Number n = ifld.numericValue(); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
709 if( n != null ) |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
710 return n; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
711 String s = ifld.stringValue(); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
712 if( s != null ) |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
713 return s; |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
714 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
|
715 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
716 |
1333
25746915a241
merge Luan and LuanState
Franklin Schmidt <fschmidt@gmail.com>
parents:
1271
diff
changeset
|
717 private static LuanTable toTable(Luan luan,Document doc) throws LuanException { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
718 if( doc==null ) |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
719 return null; |
1267 | 720 LuanTable table = new LuanTable(luan); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
721 for( IndexableField ifld : doc ) { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
722 String name = ifld.name(); |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
723 Object value = getValue(ifld); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
724 Object old = table.rawGet(name); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
725 if( old == null ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
726 table.rawPut(name,value); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
727 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
728 LuanTable list; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
729 if( old instanceof LuanTable ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
730 list = (LuanTable)old; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
731 } else { |
1267 | 732 list = new LuanTable(luan); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
733 list.rawPut(1,old); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
734 table.rawPut(name,list); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
735 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
736 list.rawPut(list.rawLength()+1,value); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
737 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
738 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
739 return table; |
312
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
740 } |
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
741 |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
742 |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
743 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
|
744 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
|
745 return originalText; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
746 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
747 }; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
748 |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
749 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
|
750 throws ParseException |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
751 { |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
752 Query query = SaneQueryParser.parseQuery(mfp,queryStr); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
753 Formatter fmt = new Formatter() { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
754 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
|
755 if( tokenGroup.getTotalScore() <= 0 ) |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
756 return originalText; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
757 try { |
1335 | 758 return (String)Luan.first(formatter.call(originalText)); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
759 } catch(LuanException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
760 throw new LuanRuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
761 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
762 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
763 }; |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
764 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
|
765 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
|
766 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
|
767 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
|
768 final Highlighter hl = new Highlighter(fmt,queryScorer); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
769 hl.setTextFragmenter( new NullFragmenter() ); |
1335 | 770 return new LuanFunction(false) { // ??? |
771 @Override public String call(Object[] args) throws LuanException { | |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
772 String text = (String)args[0]; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
773 try { |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
774 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
|
775 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
|
776 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
|
777 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
|
778 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
|
779 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
|
780 if( !atStart ) |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
781 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
|
782 if( !atEnd ) |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
783 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
|
784 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
785 text = s; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
786 } 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
|
787 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
|
788 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
|
789 text += "..."; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
790 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
791 } |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
792 String s = hl.getBestFragment(analyzer,null,text); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
793 return s!=null ? s : text; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
794 } catch(LuanRuntimeException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
795 throw (LuanException)e.getCause(); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
796 } catch(IOException 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 } catch(InvalidTokenOffsetsException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
799 throw new RuntimeException(e); |
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 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
802 }; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
803 } |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
804 |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
805 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
|
806 throws IOException |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
807 { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
808 int n = 0; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
809 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
|
810 ts.reset(); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
811 while( ts.incrementToken() ) { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
812 n++; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
813 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
814 ts.close(); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
815 return n; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
816 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
817 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
818 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
819 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
820 public boolean hasPostgresBackup() { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
821 return postgresBackup != null; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
822 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
823 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
824 public void rebuild_postgres_backup(LuanFunction completer) |
1392 | 825 throws IOException, LuanException, SQLException |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
826 { |
1393 | 827 luanLogger.info("start rebuild_postgres_backup"); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
828 writeLock.lock(); |
1391 | 829 IndexSearcher searcher = openSearcher(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
830 boolean ok = false; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
831 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
832 postgresBackup.begin(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
833 postgresBackup.deleteAll(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
834 Query query = new PrefixQuery(new Term("id")); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
835 MyCollector col = new MyCollector() { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
836 @Override public void collect(int iDoc) throws IOException { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
837 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
838 Document doc = searcher.doc( docBase + iDoc ); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
839 LuanTable tbl = toTable(completer.luan(),doc); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
840 tbl = (LuanTable)completer.call(tbl); |
1391 | 841 postgresBackup.add(tbl); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
842 } catch(LuanException e) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
843 throw new LuanRuntimeException(e); |
1392 | 844 } catch(SQLException e) { |
845 throw new RuntimeException(e); | |
1387
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 }; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
849 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
850 searcher.search(query,col); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
851 } catch(LuanRuntimeException e) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
852 throw (LuanException)e.getCause(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
853 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
854 ok = true; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
855 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
856 } finally { |
1391 | 857 close(searcher); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
858 if( !ok ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
859 postgresBackup.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
860 writeLock.unlock(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
861 } |
1393 | 862 luanLogger.info("end rebuild_postgres_backup"); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
863 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
864 |
1393 | 865 public void restore_from_postgres() |
1392 | 866 throws IOException, LuanException, SQLException |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
867 { |
1393 | 868 luanLogger.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; |
1395
9dfff82dfc59
finish postgres work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1393
diff
changeset
|
880 saveNextId(nextId); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
881 ok = true; |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
882 writer.commit(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
883 } finally { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
884 if( !ok ) { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
885 writer.rollback(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
886 reopen(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
887 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
888 wrote(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
889 writeLock.unlock(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
890 } |
1393 | 891 luanLogger.warn("end restore_from_postgres"); |
1388
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 |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
894 void restore(LuanTable doc) |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
895 throws LuanException, IOException |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
896 { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
897 writer.addDocument(toLucene(doc,null)); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
898 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
899 |
1391 | 900 public void check(LuanFunction completer) throws IOException, SQLException, LuanException { |
1393 | 901 luanLogger.info("start check"); |
1391 | 902 CheckIndex.Status status = new CheckIndex(fsDir).checkIndex(); |
903 if( !status.clean ) | |
1393 | 904 luanLogger.error("index not clean"); |
1391 | 905 if( postgresBackup != null ) |
906 checkPostgres(completer); | |
1393 | 907 luanLogger.info("end check"); |
1391 | 908 } |
909 | |
910 private void checkPostgres(LuanFunction completer) throws IOException, SQLException, LuanException { | |
1395
9dfff82dfc59
finish postgres work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1393
diff
changeset
|
911 luanLogger.info("start postgres check"); |
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++; | |
1393 | 946 luanLogger.error("id "+idLucene+" found in lucene but not postgres"); |
1391 | 947 } else if( idLucene > idPostgres ) { |
948 iPostgres++; | |
1393 | 949 luanLogger.error("id "+idPostgres+" found in postgres but not lucene"); |
1391 | 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) ) { | |
1393 | 958 luanLogger.error("id "+idLucene+" not equal"); |
959 luanLogger.error("lucene = "+lts.toString(docLucene)); | |
960 luanLogger.error("postgres = "+lts.toString(docPostgres)); | |
1391 | 961 } |
962 iLucene++; | |
963 iPostgres++; | |
964 } | |
965 } | |
966 while( iLucene < nLucene ) { | |
967 long idLucene = idsLucene.get(iLucene++); | |
1393 | 968 luanLogger.error("id "+idLucene+" found in lucene but not postgres"); |
1391 | 969 } |
970 while( iPostgres < nPostgres ) { | |
971 long idPostgres = idsPostgres.get(iPostgres++); | |
1393 | 972 luanLogger.error("id "+idPostgres+" found in postgres but not lucene"); |
1391 | 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 } |