Mercurial Hosting > luan
annotate src/luan/modules/lucene/LuceneIndex.java @ 1672:8dd8c556c449
backup work
author | Franklin Schmidt <fschmidt@gmail.com> |
---|---|
date | Thu, 26 May 2022 21:10:02 -0600 |
parents | 582384548a69 |
children | 1b9f9fdb3b41 |
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; |
1397 | 8 import java.lang.ref.Reference; |
9 import java.lang.ref.WeakReference; | |
1391 | 10 import java.sql.SQLException; |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
11 import java.util.Arrays; |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
12 import java.util.Iterator; |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
13 import java.util.Map; |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
14 import java.util.HashMap; |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
15 import java.util.List; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
16 import java.util.ArrayList; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
17 import java.util.Set; |
618
5e495e4e560b
add lucene indexed_only_fields
Franklin Schmidt <fschmidt@gmail.com>
parents:
617
diff
changeset
|
18 import java.util.HashSet; |
5e495e4e560b
add lucene indexed_only_fields
Franklin Schmidt <fschmidt@gmail.com>
parents:
617
diff
changeset
|
19 import java.util.Collections; |
704 | 20 import java.util.concurrent.atomic.AtomicInteger; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
21 import java.util.concurrent.locks.Lock; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
22 import java.util.concurrent.locks.ReentrantLock; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
23 import java.util.zip.ZipOutputStream; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
24 import java.util.zip.ZipEntry; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
25 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
|
26 import org.apache.lucene.analysis.TokenStream; |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
27 import org.apache.lucene.analysis.core.KeywordAnalyzer; |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
28 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
|
29 import org.apache.lucene.document.Document; |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
30 import org.apache.lucene.document.Field; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
31 import org.apache.lucene.document.StoredField; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
32 import org.apache.lucene.document.StringField; |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
33 import org.apache.lucene.document.TextField; |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
34 import org.apache.lucene.document.IntField; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
35 import org.apache.lucene.document.LongField; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
36 import org.apache.lucene.document.DoubleField; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
37 import org.apache.lucene.index.IndexableField; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
38 import org.apache.lucene.index.IndexWriter; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
39 import org.apache.lucene.index.IndexWriterConfig; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
40 import org.apache.lucene.index.DirectoryReader; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
41 import org.apache.lucene.index.Term; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
42 import org.apache.lucene.index.SnapshotDeletionPolicy; |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
43 import org.apache.lucene.index.IndexCommit; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
44 import org.apache.lucene.index.AtomicReaderContext; |
1391 | 45 import org.apache.lucene.index.CheckIndex; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
46 import org.apache.lucene.store.Directory; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
47 import org.apache.lucene.store.FSDirectory; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
48 import org.apache.lucene.util.Version; |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
49 import org.apache.lucene.util.BytesRef; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
50 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
|
51 import org.apache.lucene.search.Query; |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
52 import org.apache.lucene.search.PrefixQuery; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
53 import org.apache.lucene.search.TermQuery; |
1539 | 54 import org.apache.lucene.search.MatchAllDocsQuery; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
55 import org.apache.lucene.search.TopDocs; |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
56 import org.apache.lucene.search.Sort; |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
57 import org.apache.lucene.search.SortField; |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
58 import org.apache.lucene.search.IndexSearcher; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
59 import org.apache.lucene.search.TotalHitCountCollector; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
60 import org.apache.lucene.search.ScoreDoc; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
61 import org.apache.lucene.search.Collector; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
62 import org.apache.lucene.search.Scorer; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
63 import org.apache.lucene.search.Explanation; |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
64 import org.apache.lucene.search.highlight.Formatter; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
65 import org.apache.lucene.search.highlight.Highlighter; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
66 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
|
67 import org.apache.lucene.search.highlight.Fragmenter; |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
68 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
|
69 import org.apache.lucene.search.highlight.SimpleSpanFragmenter; |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
70 import org.apache.lucene.search.highlight.QueryScorer; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
71 import org.apache.lucene.search.highlight.TokenGroup; |
1474 | 72 import goodjava.lucene.analysis.LowercaseAnalyzer; |
1459 | 73 import goodjava.lucene.queryparser.GoodQueryParser; |
1458 | 74 import goodjava.lucene.queryparser.FieldParser; |
75 import goodjava.lucene.queryparser.MultiFieldParser; | |
76 import goodjava.lucene.queryparser.StringFieldParser; | |
77 import goodjava.lucene.queryparser.NumberFieldParser; | |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
78 import goodjava.lucene.api.GoodIndexWriter; |
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
79 import goodjava.lucene.api.LuceneIndexWriter; |
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
80 import goodjava.lucene.api.GoodIndexWriterConfig; |
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
81 import goodjava.lucene.api.LuceneUtils; |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
82 import goodjava.lucene.logging.LoggingIndexWriter; |
1551 | 83 import goodjava.lucene.logging.OpDoer; |
1672 | 84 import goodjava.lucene.backup.BackupIndexWriter; |
1402
27efb1fcbcb5
move luan.lib to goodjava
Franklin Schmidt <fschmidt@gmail.com>
parents:
1399
diff
changeset
|
85 import goodjava.parser.ParseException; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
86 import luan.modules.Utils; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
87 import luan.Luan; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
88 import luan.LuanTable; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
89 import luan.LuanFunction; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
90 import luan.LuanException; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
91 import luan.LuanRuntimeException; |
1391 | 92 import luan.modules.parsers.LuanToString; |
1402
27efb1fcbcb5
move luan.lib to goodjava
Franklin Schmidt <fschmidt@gmail.com>
parents:
1399
diff
changeset
|
93 import goodjava.logging.Logger; |
27efb1fcbcb5
move luan.lib to goodjava
Franklin Schmidt <fschmidt@gmail.com>
parents:
1399
diff
changeset
|
94 import goodjava.logging.LoggerFactory; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
95 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
96 |
1398 | 97 public final class LuceneIndex { |
1454 | 98 private static final Logger logger = LoggerFactory.getLogger(LuceneIndex.class); |
521
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
99 |
1397 | 100 private static Map<String,Reference<LuceneIndex>> indexes = new HashMap<String,Reference<LuceneIndex>>(); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
101 |
1397 | 102 public static LuceneIndex getLuceneIndex(Luan luan,File indexDir,LuanTable options) |
1392 | 103 throws LuanException, IOException, ClassNotFoundException, SQLException |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
104 { |
1369
709f7498a363
change Lucene.index() and add Lucene.recover()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1347
diff
changeset
|
105 String key = indexDir.getCanonicalPath(); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
106 synchronized(indexes) { |
1397 | 107 Reference<LuceneIndex> ref = indexes.get(key); |
108 if( ref != null ) { | |
109 LuceneIndex li = ref.get(); | |
1398 | 110 if( li != null ) { |
1562 | 111 Object version = options.get(luan,"version"); |
1398 | 112 if( version==null || version.equals(li.version) ) |
113 return li; | |
1397 | 114 li.closeWriter(); |
1398 | 115 } |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
116 } |
1397 | 117 LuceneIndex li = new LuceneIndex(luan,indexDir,options); |
118 indexes.put(key, new WeakReference<LuceneIndex>(li)); | |
119 return li; | |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
120 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
121 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
122 |
1398 | 123 private static final Version luceneVersion = Version.LUCENE_4_9; |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
124 public static final StringFieldParser STRING_FIELD_PARSER = new StringFieldParser(new KeywordAnalyzer()); |
1532
060ff7695317
Lucene.type.lowercase
Franklin Schmidt <fschmidt@gmail.com>
parents:
1530
diff
changeset
|
125 public static final StringFieldParser LOWERCASE_FIELD_PARSER = new StringFieldParser(new LowercaseAnalyzer(luceneVersion)); |
1398 | 126 public static final StringFieldParser ENGLISH_FIELD_PARSER = new StringFieldParser(new EnglishAnalyzer(luceneVersion)); |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
127 private static final SortField ID_SORT = new SortField("id",SortField.Type.LONG); |
1539 | 128 private static final SortField ID_DESC_SORT = new SortField("id",SortField.Type.LONG,true); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
129 |
1398 | 130 private final Object version; |
1393 | 131 |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
132 private final ReentrantLock writeLock = new ReentrantLock(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
133 private final File indexDir; |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
134 private GoodIndexWriter writer; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
135 private DirectoryReader reader; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
136 private IndexSearcher searcher; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
137 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
|
138 private final MultiFieldParser mfp; |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
139 private final Analyzer analyzer; // ??? |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
140 |
1391 | 141 private FSDirectory fsDir; |
704 | 142 private int writeCount; |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
143 private AtomicInteger writeCounter = new AtomicInteger(); |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
144 private final GoodIndexWriterConfig config; |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
145 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
146 private final PostgresBackup postgresBackup; |
1397 | 147 private boolean wasCreated; |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
148 private final File logDir; |
1548 | 149 private final long logTime; |
1672 | 150 private final String name; |
151 private final String domain; | |
152 private final String password; | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
153 |
1397 | 154 private LuceneIndex(Luan luan,File indexDir,LuanTable options) |
1392 | 155 throws LuanException, IOException, ClassNotFoundException, SQLException |
1267 | 156 { |
1420 | 157 options = new LuanTable(options); |
158 this.version = options.remove("version"); | |
159 FieldParser defaultFieldParser = (FieldParser)options.remove("default_type"); | |
160 LuanTable defaultFieldsTbl = Utils.removeTable(options,"default_fields"); | |
1397 | 161 String[] defaultFields = defaultFieldsTbl==null ? null : (String[])defaultFieldsTbl.asList().toArray(new String[0]); |
1420 | 162 LuanTable postgresSpec = Utils.removeTable(options,"postgres_spec"); |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
163 LuanFunction supplementer = Utils.removeFunction(options,"supplementer"); |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
164 logDir = (File)options.remove("log_dir"); |
1548 | 165 logTime = (Long)options.remove("log_time"); |
1672 | 166 name = (String)options.remove("name"); |
1420 | 167 Utils.checkEmpty(options); |
1397 | 168 |
1672 | 169 { |
170 LuanTable module = (LuanTable)luan.require("luan:http/Http.luan"); | |
171 String domain = (String)module.get(luan,"domain"); | |
172 if( domain == null ) | |
173 domain = "localhost"; | |
174 this.domain = domain; | |
175 } | |
176 { | |
177 LuanTable module = (LuanTable)luan.require("luan:Io.luan"); | |
178 String password = (String)module.get(luan,"password"); | |
179 if( password == null ) | |
180 password = "password"; | |
181 this.password = password; | |
182 } | |
183 | |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
184 mfp = defaultFieldParser==null ? new MultiFieldParser() : new MultiFieldParser(defaultFieldParser,defaultFields); |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
185 mfp.fields.put( "type", STRING_FIELD_PARSER ); |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
186 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
|
187 this.indexDir = indexDir; |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
188 Analyzer analyzer = STRING_FIELD_PARSER.analyzer; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
189 if( defaultFieldParser instanceof StringFieldParser ) { |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
190 StringFieldParser sfp = (StringFieldParser)defaultFieldParser; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
191 analyzer = sfp.analyzer; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
192 } |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
193 this.analyzer = analyzer; |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
194 this.config = new SupplementingConfig(luceneVersion,mfp,luan,supplementer); |
1397 | 195 wasCreated = reopen(); |
1392 | 196 if( postgresSpec == null ) { |
197 postgresBackup = null; | |
198 } else { | |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
199 postgresBackup = new PostgresBackup(postgresSpec); |
1397 | 200 if( !wasCreated && postgresBackup.wasCreated ) { |
1454 | 201 logger.error("rebuilding postgres backup"); |
1453 | 202 rebuild_postgres_backup(luan); |
1397 | 203 /* |
204 } else if( wasCreated && !postgresBackup.wasCreated ) { | |
1454 | 205 logger.error("restoring from postgres"); |
1397 | 206 restore_from_postgres(); |
207 */ | |
1390 | 208 } |
209 } | |
754 | 210 } |
211 | |
1390 | 212 public boolean reopen() throws IOException { |
1391 | 213 fsDir = FSDirectory.open(indexDir); |
214 boolean wasCreated = !fsDir.getDirectory().exists(); | |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
215 writer = new LuceneIndexWriter(fsDir,config); |
1672 | 216 if( logDir != null ) { |
217 if( BackupIndexWriter.backupDomains == null ) { | |
218 writer = new LoggingIndexWriter((LuceneIndexWriter)writer,logDir,logTime); | |
219 } else { | |
220 String name = this.domain; | |
221 if( this.name != null ) | |
222 name += "~" + this.name; | |
223 writer = new BackupIndexWriter((LuceneIndexWriter)writer,logDir,logTime,name,password); | |
224 //qqq | |
225 } | |
226 } | |
1391 | 227 reader = DirectoryReader.open(fsDir); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
228 searcher = new IndexSearcher(reader); |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
229 initId(); |
1390 | 230 return wasCreated; |
233
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
231 } |
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
232 |
704 | 233 private void wrote() { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
234 writeCounter.incrementAndGet(); |
704 | 235 } |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
236 |
1392 | 237 public void delete_all() throws IOException, SQLException { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
238 boolean commit = !writeLock.isHeldByCurrentThread(); |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
239 writeLock.lock(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
240 try { |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
241 writer.deleteAll(); |
1539 | 242 id = 0; |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
243 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
244 postgresBackup.deleteAll(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
245 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
246 } finally { |
704 | 247 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
248 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
249 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
250 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
251 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
252 private static Term term(String key,long value) { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
253 BytesRef br = new BytesRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
254 NumericUtils.longToPrefixCoded(value,0,br); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
255 return new Term(key,br); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
256 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
257 |
1543 | 258 private static final Set<String> ID_SET = Collections.singleton("id"); |
259 | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
260 private void backupDelete(Query query) |
1392 | 261 throws IOException, SQLException, LuanException |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
262 { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
263 if( postgresBackup != null ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
264 final List<Long> ids = new ArrayList<Long>(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
265 IndexSearcher searcher = openSearcher(); |
1391 | 266 try { |
267 MyCollector col = new MyCollector() { | |
268 @Override public void collect(int iDoc) throws IOException { | |
1543 | 269 Document doc = searcher.doc( docBase + iDoc, ID_SET ); |
1391 | 270 Long id = (Long)doc.getField("id").numericValue(); |
271 ids.add(id); | |
272 } | |
273 }; | |
274 searcher.search(query,col); | |
275 } finally { | |
276 close(searcher); | |
277 } | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
278 postgresBackup.begin(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
279 for( Long id : ids ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
280 postgresBackup.delete(id); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
281 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
282 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
283 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
284 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
285 |
1346 | 286 public void delete(String queryStr) |
1392 | 287 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
|
288 { |
1459 | 289 Query query = GoodQueryParser.parseQuery(mfp,queryStr); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
290 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
291 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
292 writeLock.lock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
293 try { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
294 backupDelete(query); |
547
0be287ab0309
add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents:
546
diff
changeset
|
295 writer.deleteDocuments(query); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
296 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
297 } finally { |
704 | 298 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
299 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
300 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
301 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
302 |
1535 | 303 public void reindex(String queryStr) |
304 throws IOException, ParseException | |
305 { | |
306 Query query = GoodQueryParser.parseQuery(mfp,queryStr); | |
307 | |
308 boolean commit = !writeLock.isHeldByCurrentThread(); | |
309 writeLock.lock(); | |
310 try { | |
311 writer.reindexDocuments("id",query); | |
312 if(commit) writer.commit(); | |
313 } finally { | |
314 wrote(); | |
315 writeLock.unlock(); | |
316 } | |
317 } | |
318 | |
1562 | 319 public void save(Luan luan,LuanTable doc) |
1392 | 320 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
|
321 { |
1562 | 322 Object obj = doc.get(luan,"id"); |
601 | 323 Long id; |
324 try { | |
325 id = (Long)obj; | |
326 } catch(ClassCastException e) { | |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
327 throw new LuanException("id should be Long but is "+obj.getClass().getSimpleName()); |
601 | 328 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
329 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
330 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
331 writeLock.lock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
332 try { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
333 if( id == null ) { |
1539 | 334 id = ++this.id; |
1562 | 335 doc.put(luan,"id",id); |
1391 | 336 if( postgresBackup != null ) |
1562 | 337 postgresBackup.add(luan,doc); |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
338 writer.addDocument(toLucene(doc)); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
339 } else { |
1391 | 340 if( postgresBackup != null ) |
1562 | 341 postgresBackup.update(luan,doc); |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
342 writer.updateDocument( "id", toLucene(doc) ); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
343 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
344 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
345 } finally { |
704 | 346 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
347 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
348 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
349 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
350 |
1533
28f06ac6ce39
add not_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents:
1532
diff
changeset
|
351 public boolean is_in_transaction() { |
28f06ac6ce39
add not_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents:
1532
diff
changeset
|
352 return writeLock.isHeldByCurrentThread(); |
28f06ac6ce39
add not_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents:
1532
diff
changeset
|
353 } |
28f06ac6ce39
add not_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents:
1532
diff
changeset
|
354 |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
355 public Object run_in_transaction(Luan luan,LuanFunction fn) |
1392 | 356 throws IOException, LuanException, SQLException |
357 { | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
358 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
359 writeLock.lock(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
360 boolean ok = false; |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
361 try { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
362 if( commit && postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
363 postgresBackup.begin(); |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
364 Object rtn = fn.call(luan); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
365 ok = true; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
366 if(commit) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
367 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
368 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
369 writer.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
370 } |
1379 | 371 return rtn; |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
372 } finally { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
373 if( !ok && commit ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
374 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
375 postgresBackup.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
376 writer.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
377 reopen(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
378 } |
704 | 379 wrote(); |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
380 writeLock.unlock(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
381 } |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
382 } |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
383 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
384 // ??? |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
385 public Object run_in_lock(Luan luan,LuanFunction fn) throws IOException, LuanException { |
756
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
386 if( writeLock.isHeldByCurrentThread() ) |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
387 throw new RuntimeException(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
388 writeLock.lock(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
389 try { |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
390 synchronized(this) { |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
391 return fn.call(luan); |
756
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
392 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
393 } finally { |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
394 wrote(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
395 writeLock.unlock(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
396 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
397 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
398 |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
399 |
754 | 400 private long id; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
401 |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
402 private void initId() throws IOException { |
1539 | 403 TopDocs td = searcher.search(new MatchAllDocsQuery(),1,new Sort(ID_DESC_SORT)); |
404 switch(td.scoreDocs.length) { | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
405 case 0: |
754 | 406 id = 0; |
407 break; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
408 case 1: |
1539 | 409 id = (Long)searcher.doc(td.scoreDocs[0].doc).getField("id").numericValue(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
410 break; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
411 default: |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
412 throw new RuntimeException(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
413 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
414 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
415 |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
416 /* |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
417 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
|
418 if( !zipFile.endsWith(".zip") ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
419 throw new LuanException("file "+zipFile+" doesn't end with '.zip'"); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
420 IndexCommit ic = snapshotDeletionPolicy.snapshot(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
421 try { |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
422 ZipOutputStream out = new ZipOutputStream(new FileOutputStream(zipFile)); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
423 for( String fileName : ic.getFileNames() ) { |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
424 out.putNextEntry(new ZipEntry(fileName)); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
425 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
|
426 Utils.copyAll(in,out); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
427 in.close(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
428 out.closeEntry(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
429 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
430 out.close(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
431 } finally { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
432 snapshotDeletionPolicy.release(ic); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
433 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
434 } |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
435 */ |
1129
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
436 public SnapshotDeletionPolicy snapshotDeletionPolicy() { |
1539 | 437 return (SnapshotDeletionPolicy)writer.getLuceneIndexWriter().getConfig().getIndexDeletionPolicy(); |
1129
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
438 } |
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
439 |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
440 public Object snapshot(Luan luan,LuanFunction fn) throws LuanException, IOException { |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
441 SnapshotDeletionPolicy snapshotDeletionPolicy = snapshotDeletionPolicy(); |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
442 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
|
443 try { |
1391 | 444 String dir = fsDir.getDirectory().toString(); |
1562 | 445 LuanTable fileNames = new LuanTable(new ArrayList(ic.getFileNames())); |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
446 return fn.call(luan,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
|
447 } finally { |
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
448 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
|
449 } |
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
450 } |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
451 |
1544
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
452 public void tag(String tag) throws IOException { |
1556 | 453 boolean commit = !writeLock.isHeldByCurrentThread(); |
454 writeLock.lock(); | |
455 try { | |
456 writer.tag(tag); | |
457 if(commit) writer.commit(); | |
458 } finally { | |
459 writeLock.unlock(); | |
460 } | |
1544
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
461 } |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
462 |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
463 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
464 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
465 public String to_string() { |
1539 | 466 return writer.getLuceneIndexWriter().getDirectory().toString(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
467 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
468 |
1398 | 469 protected void finalize() throws Throwable { |
470 close(); | |
471 super.finalize(); | |
521
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
472 } |
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
473 |
1398 | 474 public void close() throws IOException, SQLException { |
475 closeWriter(); | |
1392 | 476 reader.close(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
477 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
478 |
1398 | 479 private void closeWriter() throws IOException, SQLException { |
1397 | 480 writeLock.lock(); |
481 try { | |
482 writer.close(); | |
1398 | 483 if( postgresBackup != null ) |
484 postgresBackup.close(); | |
1397 | 485 } finally { |
486 writeLock.unlock(); | |
487 } | |
488 } | |
489 | |
312
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
490 |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
491 private static class DocFn extends LuanFunction { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
492 final IndexSearcher searcher; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
493 final Query query; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
494 int docID; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
495 |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
496 DocFn(IndexSearcher searcher,Query query) { |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
497 this.searcher = searcher; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
498 this.query = query; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
499 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
500 |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
501 @Override public Object call(Luan luan,Object[] args) throws LuanException { |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
502 try { |
1562 | 503 LuanTable doc = toTable(searcher.doc(docID)); |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
504 if( args.length > 0 && "explain".equals(args[0]) ) { |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
505 Explanation explanation = searcher.explain(query,docID); |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
506 return new Object[]{doc,explanation}; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
507 } else { |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
508 return doc; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
509 } |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
510 } catch(IOException e) { |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
511 throw new LuanException(e); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
512 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
513 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
514 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
515 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
516 private static abstract class MyCollector extends Collector { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
517 int docBase; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
518 int i = 0; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
519 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
520 @Override public void setScorer(Scorer scorer) {} |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
521 @Override public void setNextReader(AtomicReaderContext context) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
522 this.docBase = context.docBase; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
523 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
524 @Override public boolean acceptsDocsOutOfOrder() { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
525 return true; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
526 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
527 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
528 |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
529 private synchronized IndexSearcher openSearcher() throws IOException { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
530 int gwc = writeCounter.get(); |
704 | 531 if( writeCount != gwc ) { |
532 writeCount = gwc; | |
533 DirectoryReader newReader = DirectoryReader.openIfChanged(reader); | |
1539 | 534 // DirectoryReader newReader = DirectoryReader.openIfChanged(reader,writer.getLuceneIndexWriter(),true); |
704 | 535 if( newReader != null ) { |
536 reader.decRef(); | |
537 reader = newReader; | |
538 searcher = new IndexSearcher(reader); | |
539 } | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
540 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
541 reader.incRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
542 return searcher; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
543 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
544 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
545 // call in finally block |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
546 private static void close(IndexSearcher searcher) throws IOException { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
547 searcher.getIndexReader().decRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
548 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
549 |
591
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
550 public void ensure_open() throws IOException { |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
551 close(openSearcher()); |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
552 } |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
553 |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
554 public int advanced_search( final Luan luan, 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
|
555 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
|
556 { |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
557 Utils.checkNotNull(queryStr); |
1459 | 558 Query query = GoodQueryParser.parseQuery(mfp,queryStr); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
559 IndexSearcher searcher = threadLocalSearcher.get(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
560 boolean inTransaction = searcher != null; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
561 if( !inTransaction ) |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
562 searcher = openSearcher(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
563 try { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
564 if( fn!=null && n==null ) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
565 if( sortStr != null ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
566 throw new LuanException("sort must be nil when n is nil"); |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
567 final DocFn docFn = new DocFn(searcher,query); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
568 MyCollector col = new MyCollector() { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
569 @Override public void collect(int doc) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
570 try { |
547
0be287ab0309
add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents:
546
diff
changeset
|
571 docFn.docID = docBase + doc; |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
572 fn.call(luan,++i,docFn); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
573 } catch(LuanException e) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
574 throw new LuanRuntimeException(e); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
575 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
576 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
577 }; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
578 try { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
579 searcher.search(query,col); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
580 } catch(LuanRuntimeException e) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
581 throw (LuanException)e.getCause(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
582 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
583 return col.i; |
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 if( fn==null || n==0 ) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
586 TotalHitCountCollector thcc = new TotalHitCountCollector(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
587 searcher.search(query,thcc); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
588 return thcc.getTotalHits(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
589 } |
1459 | 590 Sort sort = sortStr==null ? null : GoodQueryParser.parseSort(mfp,sortStr); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
591 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
|
592 final ScoreDoc[] scoreDocs = td.scoreDocs; |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
593 DocFn docFn = new DocFn(searcher,query); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
594 for( int i=0; i<scoreDocs.length; i++ ) { |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
595 ScoreDoc scoreDoc = scoreDocs[i]; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
596 docFn.docID = scoreDoc.doc; |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
597 fn.call(luan,i+1,docFn,scoreDoc.score); |
545
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 return td.totalHits; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
600 } finally { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
601 if( !inTransaction ) |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
602 close(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
603 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
604 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
605 |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
606 public Object search_in_transaction(Luan luan,LuanFunction fn) throws LuanException, IOException { |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
607 if( threadLocalSearcher.get() != null ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
608 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
|
609 IndexSearcher searcher = openSearcher(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
610 threadLocalSearcher.set(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
611 try { |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
612 return fn.call(luan); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
613 } finally { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
614 threadLocalSearcher.set(null); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
615 close(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
616 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
617 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
618 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
619 |
796 | 620 public FieldParser getIndexedFieldParser(String field) { |
621 return mfp.fields.get(field); | |
622 } | |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
623 |
796 | 624 public void setIndexedFieldParser(String field,FieldParser fp) { |
625 if( fp==null ) { // delete | |
626 mfp.fields.remove(field); | |
627 return; | |
628 } | |
629 mfp.fields.put( field, fp ); | |
630 } | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
631 |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
632 static Map<String,Object> toLucene(LuanTable table) throws LuanException { |
1529 | 633 return SupplementingConfig.toLucene(table); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
634 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
635 |
1562 | 636 private static LuanTable toTable(Document doc) throws LuanException { |
637 return doc==null ? null : SupplementingConfig.toTable(LuceneUtils.toMap(doc)); | |
312
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
638 } |
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
639 |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
640 |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
641 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
|
642 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
|
643 return originalText; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
644 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
645 }; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
646 |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
647 public LuanFunction highlighter(final Luan luan,String queryStr,final LuanFunction formatter,final Integer fragmentSize,String dotdotdot) |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
648 throws ParseException |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
649 { |
1459 | 650 Query query = GoodQueryParser.parseQuery(mfp,queryStr); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
651 Formatter fmt = new Formatter() { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
652 public String highlightTerm(String originalText,TokenGroup tokenGroup) { |
1435 | 653 if( tokenGroup.getTotalScore() <= 0 ) |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
654 return originalText; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
655 try { |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
656 return (String)Luan.first(formatter.call(luan,originalText)); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
657 } catch(LuanException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
658 throw new LuanRuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
659 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
660 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
661 }; |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
662 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
|
663 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
|
664 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
|
665 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
|
666 final Highlighter hl = new Highlighter(fmt,queryScorer); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
667 hl.setTextFragmenter( new NullFragmenter() ); |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
668 return new LuanFunction() { |
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
669 @Override public String call(Luan luan,Object[] args) throws LuanException { |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
670 String text = (String)args[0]; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
671 try { |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
672 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
|
673 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
|
674 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
|
675 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
|
676 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
|
677 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
|
678 if( !atStart ) |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
679 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
|
680 if( !atEnd ) |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
681 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
|
682 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
683 text = s; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
684 } 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
|
685 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
|
686 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
|
687 text += "..."; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
688 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
689 } |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
690 String s = hl.getBestFragment(analyzer,null,text); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
691 return s!=null ? s : text; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
692 } catch(LuanRuntimeException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
693 throw (LuanException)e.getCause(); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
694 } catch(IOException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
695 throw new RuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
696 } catch(InvalidTokenOffsetsException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
697 throw new RuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
698 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
699 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
700 }; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
701 } |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
702 |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
703 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
|
704 throws IOException |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
705 { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
706 int n = 0; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
707 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
|
708 ts.reset(); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
709 while( ts.incrementToken() ) { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
710 n++; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
711 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
712 ts.close(); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
713 return n; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
714 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
715 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
716 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
717 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
718 public boolean hasPostgresBackup() { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
719 return postgresBackup != null; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
720 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
721 |
1453 | 722 public void rebuild_postgres_backup(Luan luan) |
1392 | 723 throws IOException, LuanException, SQLException |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
724 { |
1454 | 725 logger.info("start rebuild_postgres_backup"); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
726 writeLock.lock(); |
1391 | 727 IndexSearcher searcher = openSearcher(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
728 boolean ok = false; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
729 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
730 postgresBackup.begin(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
731 postgresBackup.deleteAll(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
732 Query query = new PrefixQuery(new Term("id")); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
733 MyCollector col = new MyCollector() { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
734 @Override public void collect(int iDoc) throws IOException { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
735 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
736 Document doc = searcher.doc( docBase + iDoc ); |
1562 | 737 LuanTable tbl = toTable(doc); |
738 postgresBackup.add(luan,tbl); | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
739 } catch(LuanException e) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
740 throw new LuanRuntimeException(e); |
1392 | 741 } catch(SQLException e) { |
742 throw new RuntimeException(e); | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
743 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
744 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
745 }; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
746 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
747 searcher.search(query,col); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
748 } catch(LuanRuntimeException e) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
749 throw (LuanException)e.getCause(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
750 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
751 ok = true; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
752 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
753 } finally { |
1391 | 754 close(searcher); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
755 if( !ok ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
756 postgresBackup.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
757 writeLock.unlock(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
758 } |
1454 | 759 logger.info("end rebuild_postgres_backup"); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
760 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
761 |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
762 public void restore_from_postgres() |
1447 | 763 throws IOException, LuanException, SQLException, ParseException |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
764 { |
1397 | 765 if( postgresBackup!=null && wasCreated && !postgresBackup.wasCreated ) { |
1454 | 766 logger.error("restoring from postgres"); |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
767 force_restore_from_postgres(); |
1397 | 768 } |
769 } | |
770 | |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
771 public void force_restore_from_postgres() |
1447 | 772 throws IOException, LuanException, SQLException, ParseException |
1397 | 773 { |
1454 | 774 logger.warn("start restore_from_postgres"); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
775 if( postgresBackup==null ) |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
776 throw new NullPointerException(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
777 if( writeLock.isHeldByCurrentThread() ) |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
778 throw new RuntimeException(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
779 writeLock.lock(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
780 boolean ok = false; |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
781 try { |
1548 | 782 writer.tag("restore_from_postgres"); |
783 writer.deleteAll(); | |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
784 postgresBackup.restoreLucene(this); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
785 ok = true; |
1548 | 786 writer.commit(); |
1539 | 787 wrote(); |
788 ensure_open(); // refresh searcher | |
789 initId(); | |
1398 | 790 wasCreated = false; |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
791 } finally { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
792 if( !ok ) { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
793 writer.rollback(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
794 reopen(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
795 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
796 wrote(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
797 writeLock.unlock(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
798 } |
1454 | 799 logger.warn("end restore_from_postgres"); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
800 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
801 |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
802 void restore(LuanTable doc) |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
803 throws LuanException, IOException |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
804 { |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
805 writer.addDocument(toLucene(doc)); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
806 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
807 |
1556 | 808 public void relog() |
809 throws IOException, LuanException | |
810 { | |
811 logger.info("start relog"); | |
812 writeLock.lock(); | |
813 try { | |
814 LoggingIndexWriter loggingWriter = (LoggingIndexWriter)writer; | |
815 loggingWriter.logLucene(); | |
816 } finally { | |
817 writeLock.unlock(); | |
818 } | |
819 logger.info("end relog"); | |
820 } | |
821 | |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
822 public void restore_from_log(Luan luan,LuanFunction handler) |
1544
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
823 throws IOException, LuanException, SQLException, ParseException |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
824 { |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
825 LoggingIndexWriter loggingWriter = (LoggingIndexWriter)writer; |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
826 if( wasCreated && !loggingWriter.wasCreated ) { |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
827 logger.error("restoring from log"); |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
828 force_restore_from_log(luan,handler); |
1544
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
829 } |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
830 } |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
831 |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
832 public void force_restore_from_log(Luan luan,LuanFunction handler) |
1544
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
833 throws IOException |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
834 { |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
835 logger.warn("start force_restore_from_log"); |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
836 if( writeLock.isHeldByCurrentThread() ) |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
837 throw new RuntimeException(); |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
838 OpDoer opDoer = handler==null ? null : new LuanOpDoer(writer,luan,handler); |
1544
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
839 writeLock.lock(); |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
840 boolean ok = false; |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
841 try { |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
842 LoggingIndexWriter loggingWriter = (LoggingIndexWriter)writer; |
1551 | 843 loggingWriter.playLogs(opDoer); |
1544
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
844 ok = true; |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
845 wrote(); |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
846 ensure_open(); // refresh searcher |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
847 initId(); |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
848 wasCreated = false; |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
849 } finally { |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
850 if( !ok ) { |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
851 writer.rollback(); |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
852 reopen(); |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
853 } |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
854 wrote(); |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
855 writeLock.unlock(); |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
856 } |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
857 logger.warn("end force_restore_from_log"); |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
858 } |
35601f15ecc3
add lucene log tag and restore_from_log
Franklin Schmidt <fschmidt@gmail.com>
parents:
1543
diff
changeset
|
859 |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
860 public void check() throws IOException, SQLException, LuanException, ParseException { |
1406 | 861 boolean hasPostgres = postgresBackup != null; |
862 String msg = "start check"; | |
863 if( hasPostgres ) | |
864 msg += " with postgres"; | |
1454 | 865 logger.info(msg); |
1391 | 866 CheckIndex.Status status = new CheckIndex(fsDir).checkIndex(); |
867 if( !status.clean ) | |
1454 | 868 logger.error("index not clean"); |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
869 if( writer instanceof LoggingIndexWriter ) { |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
870 LoggingIndexWriter loggingWriter = (LoggingIndexWriter)writer; |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
871 logger.info("log check"); |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
872 boolean ok = loggingWriter.check(ID_SORT); |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
873 } |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
874 if( hasPostgres ) { |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
875 logger.info("postgres check"); |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
876 checkPostgres(); |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
877 } |
1454 | 878 logger.info("end check"); |
1391 | 879 } |
880 | |
1563
8fbcc4747091
remove LuanFunction.luan
Franklin Schmidt <fschmidt@gmail.com>
parents:
1562
diff
changeset
|
881 private void checkPostgres() |
1447 | 882 throws IOException, SQLException, LuanException, ParseException |
1430 | 883 { |
884 final PostgresBackup.Checker postgresChecker = postgresBackup.newChecker(); | |
885 final IndexSearcher searcher = openSearcher(); | |
1391 | 886 try { |
887 final List<Long> idsLucene = new ArrayList<Long>(); | |
888 Query query = new PrefixQuery(new Term("id")); | |
889 MyCollector col = new MyCollector() { | |
890 @Override public void collect(int iDoc) throws IOException { | |
891 Document doc = searcher.doc( docBase + iDoc ); | |
892 Long id = (Long)doc.getField("id").numericValue(); | |
893 idsLucene.add(id); | |
894 } | |
895 }; | |
896 searcher.search(query,col); | |
897 Collections.sort(idsLucene); | |
898 final List<Long> idsPostgres = postgresChecker.getIds(); | |
899 final int nLucene = idsLucene.size(); | |
900 final int nPostgres = idsPostgres.size(); | |
901 int iLucene = 0; | |
902 int iPostgres = 0; | |
1623
582384548a69
LuanToString takes function
Franklin Schmidt <fschmidt@gmail.com>
parents:
1563
diff
changeset
|
903 LuanToString lts = new LuanToString(); |
1553 | 904 lts.settingsInit.strict = true; |
905 lts.settingsInit.numberTypes = true; | |
1391 | 906 while( iLucene < nLucene && iPostgres < nPostgres ) { |
907 long idLucene = idsLucene.get(iLucene); | |
908 long idPostgres = idsPostgres.get(iPostgres); | |
909 if( idLucene < idPostgres ) { | |
910 iLucene++; | |
1562 | 911 checkPostgres(postgresChecker,lts,idLucene); |
1391 | 912 } else if( idLucene > idPostgres ) { |
913 iPostgres++; | |
1562 | 914 checkPostgres(postgresChecker,lts,idPostgres); |
1391 | 915 } else { // == |
916 LuanTable docPostgres = postgresChecker.getDoc(idPostgres); | |
917 TopDocs td = searcher.search(new TermQuery(term("id",idLucene)),1); | |
918 if( td.totalHits != 1 ) throw new RuntimeException(); | |
919 Document doc = searcher.doc( td.scoreDocs[0].doc ); | |
1562 | 920 LuanTable docLucene = toTable(doc); |
1391 | 921 if( !equal(docPostgres,docLucene) ) { |
1562 | 922 checkPostgres(postgresChecker,lts,idPostgres); |
1391 | 923 } |
924 iLucene++; | |
925 iPostgres++; | |
926 } | |
927 } | |
928 while( iLucene < nLucene ) { | |
929 long idLucene = idsLucene.get(iLucene++); | |
1562 | 930 checkPostgres(postgresChecker,lts,idLucene); |
1391 | 931 } |
932 while( iPostgres < nPostgres ) { | |
933 long idPostgres = idsPostgres.get(iPostgres++); | |
1562 | 934 checkPostgres(postgresChecker,lts,idPostgres); |
1391 | 935 } |
936 } finally { | |
937 close(searcher); | |
938 postgresChecker.close(); | |
939 } | |
940 } | |
941 | |
1562 | 942 private void checkPostgres(PostgresBackup.Checker postgresChecker,LuanToString lts,long id) |
1447 | 943 throws IOException, SQLException, LuanException, ParseException |
1430 | 944 { |
1454 | 945 //logger.info("check id "+id); |
1430 | 946 writeLock.lock(); |
947 try { | |
948 final IndexSearcher searcher = openSearcher(); | |
949 try { | |
950 LuanTable docPostgres = postgresChecker.getDoc(id); | |
951 TopDocs td = searcher.search(new TermQuery(term("id",id)),1); | |
952 LuanTable docLucene; | |
953 if( td.totalHits == 0 ) { | |
954 docLucene = null; | |
955 } else if( td.totalHits == 1 ) { | |
956 Document doc = searcher.doc( td.scoreDocs[0].doc ); | |
1562 | 957 docLucene = toTable(doc); |
1430 | 958 } else |
959 throw new RuntimeException(); | |
960 if( docPostgres == null ) { | |
961 if( docLucene != null ) | |
1454 | 962 logger.error("id "+id+" found in lucene but not postgres"); |
1430 | 963 return; |
964 } | |
965 if( docLucene == null ) { | |
1454 | 966 logger.error("id "+id+" found in postgres but not lucene"); |
1430 | 967 return; |
968 } | |
969 if( !equal(docPostgres,docLucene) ) { | |
1454 | 970 logger.error("id "+id+" not equal"); |
971 logger.error("lucene = "+lts.toString(docLucene)); | |
972 logger.error("postgres = "+lts.toString(docPostgres)); | |
1430 | 973 } |
974 } finally { | |
975 close(searcher); | |
976 } | |
977 } finally { | |
978 writeLock.unlock(); | |
979 } | |
980 } | |
981 | |
1453 | 982 private static boolean equal(LuanTable t1,LuanTable t2) throws LuanException { |
983 return t1!=null && t2!=null && toJava(t1).equals(toJava(t2)); | |
1391 | 984 } |
985 | |
1453 | 986 private static Map toJava(LuanTable t) throws LuanException { |
987 Map map = t.asMap(); | |
1526 | 988 for( Iterator iter = map.entrySet().iterator(); iter.hasNext(); ) { |
989 Map.Entry entry = (Map.Entry)iter.next(); | |
1453 | 990 Object value = entry.getValue(); |
991 if( value instanceof LuanTable ) { | |
992 LuanTable v = (LuanTable)value; | |
993 if( !v.isList() ) | |
1454 | 994 logger.error("not list"); |
1530 | 995 List list = v.asList(); |
996 if( list.isEmpty() ) { | |
1526 | 997 iter.remove(); |
1530 | 998 } else if( list.size() == 1 ) { |
999 entry.setValue(list.get(0)); | |
1526 | 1000 } else { |
1530 | 1001 entry.setValue(list); |
1526 | 1002 } |
1453 | 1003 } |
1004 } | |
1005 return map; | |
1006 } | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
1007 } |