Mercurial Hosting > luan
annotate src/luan/modules/lucene/LuceneIndex.java @ 1541:dc23c96f5021
minor
author | Franklin Schmidt <fschmidt@gmail.com> |
---|---|
date | Sun, 13 Sep 2020 15:50:52 -0600 |
parents | 79f67662d3e7 |
children | 1db694d98003 |
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; |
1402
27efb1fcbcb5
move luan.lib to goodjava
Franklin Schmidt <fschmidt@gmail.com>
parents:
1399
diff
changeset
|
83 import goodjava.parser.ParseException; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
84 import luan.modules.Utils; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
85 import luan.Luan; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
86 import luan.LuanTable; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
87 import luan.LuanFunction; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
88 import luan.LuanException; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
89 import luan.LuanRuntimeException; |
1391 | 90 import luan.modules.parsers.LuanToString; |
1402
27efb1fcbcb5
move luan.lib to goodjava
Franklin Schmidt <fschmidt@gmail.com>
parents:
1399
diff
changeset
|
91 import goodjava.logging.Logger; |
27efb1fcbcb5
move luan.lib to goodjava
Franklin Schmidt <fschmidt@gmail.com>
parents:
1399
diff
changeset
|
92 import goodjava.logging.LoggerFactory; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
93 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
94 |
1398 | 95 public final class LuceneIndex { |
1454 | 96 private static final Logger logger = LoggerFactory.getLogger(LuceneIndex.class); |
521
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
97 |
1397 | 98 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
|
99 |
1397 | 100 public static LuceneIndex getLuceneIndex(Luan luan,File indexDir,LuanTable options) |
1392 | 101 throws LuanException, IOException, ClassNotFoundException, SQLException |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
102 { |
1369
709f7498a363
change Lucene.index() and add Lucene.recover()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1347
diff
changeset
|
103 String key = indexDir.getCanonicalPath(); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
104 synchronized(indexes) { |
1397 | 105 Reference<LuceneIndex> ref = indexes.get(key); |
106 if( ref != null ) { | |
107 LuceneIndex li = ref.get(); | |
1398 | 108 if( li != null ) { |
109 Object version = options.get("version"); | |
110 if( version==null || version.equals(li.version) ) | |
111 return li; | |
1397 | 112 li.closeWriter(); |
1398 | 113 } |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
114 } |
1397 | 115 LuceneIndex li = new LuceneIndex(luan,indexDir,options); |
116 indexes.put(key, new WeakReference<LuceneIndex>(li)); | |
117 return li; | |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
118 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
119 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
120 |
1398 | 121 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
|
122 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
|
123 public static final StringFieldParser LOWERCASE_FIELD_PARSER = new StringFieldParser(new LowercaseAnalyzer(luceneVersion)); |
1398 | 124 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
|
125 private static final SortField ID_SORT = new SortField("id",SortField.Type.LONG); |
1539 | 126 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
|
127 |
1398 | 128 private final Object version; |
1393 | 129 |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
130 private final ReentrantLock writeLock = new ReentrantLock(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
131 private final File indexDir; |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
132 private GoodIndexWriter writer; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
133 private DirectoryReader reader; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
134 private IndexSearcher searcher; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
135 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
|
136 private final MultiFieldParser mfp; |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
137 private final Analyzer analyzer; // ??? |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
138 |
1391 | 139 private FSDirectory fsDir; |
704 | 140 private int writeCount; |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
141 private AtomicInteger writeCounter = new AtomicInteger(); |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
142 private final GoodIndexWriterConfig config; |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
143 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
144 private final PostgresBackup postgresBackup; |
1397 | 145 private boolean wasCreated; |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
146 private final File logDir; |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
147 |
1397 | 148 private LuceneIndex(Luan luan,File indexDir,LuanTable options) |
1392 | 149 throws LuanException, IOException, ClassNotFoundException, SQLException |
1267 | 150 { |
1420 | 151 options = new LuanTable(options); |
152 this.version = options.remove("version"); | |
153 FieldParser defaultFieldParser = (FieldParser)options.remove("default_type"); | |
154 LuanTable defaultFieldsTbl = Utils.removeTable(options,"default_fields"); | |
1397 | 155 String[] defaultFields = defaultFieldsTbl==null ? null : (String[])defaultFieldsTbl.asList().toArray(new String[0]); |
1420 | 156 LuanTable postgresSpec = Utils.removeTable(options,"postgres_spec"); |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
157 LuanFunction supplementer = Utils.removeFunction(options,"supplementer"); |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
158 logDir = (File)options.remove("log_dir"); |
1420 | 159 Utils.checkEmpty(options); |
1397 | 160 |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
161 mfp = defaultFieldParser==null ? new MultiFieldParser() : new MultiFieldParser(defaultFieldParser,defaultFields); |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
162 mfp.fields.put( "type", STRING_FIELD_PARSER ); |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
163 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
|
164 this.indexDir = indexDir; |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
165 Analyzer analyzer = STRING_FIELD_PARSER.analyzer; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
166 if( defaultFieldParser instanceof StringFieldParser ) { |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
167 StringFieldParser sfp = (StringFieldParser)defaultFieldParser; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
168 analyzer = sfp.analyzer; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
169 } |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
170 this.analyzer = analyzer; |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
171 this.config = new SupplementingConfig(luceneVersion,mfp,supplementer); |
1397 | 172 wasCreated = reopen(); |
1392 | 173 if( postgresSpec == null ) { |
174 postgresBackup = null; | |
175 } else { | |
1420 | 176 postgresBackup = new PostgresBackup(luan,postgresSpec); |
1397 | 177 if( !wasCreated && postgresBackup.wasCreated ) { |
1454 | 178 logger.error("rebuilding postgres backup"); |
1453 | 179 rebuild_postgres_backup(luan); |
1397 | 180 /* |
181 } else if( wasCreated && !postgresBackup.wasCreated ) { | |
1454 | 182 logger.error("restoring from postgres"); |
1397 | 183 restore_from_postgres(); |
184 */ | |
1390 | 185 } |
186 } | |
754 | 187 } |
188 | |
1390 | 189 public boolean reopen() throws IOException { |
1391 | 190 fsDir = FSDirectory.open(indexDir); |
191 boolean wasCreated = !fsDir.getDirectory().exists(); | |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
192 writer = new LuceneIndexWriter(fsDir,config); |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
193 if( logDir != null ) |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
194 writer = new LoggingIndexWriter((LuceneIndexWriter)writer,logDir); |
1391 | 195 reader = DirectoryReader.open(fsDir); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
196 searcher = new IndexSearcher(reader); |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
197 initId(); |
1390 | 198 return wasCreated; |
233
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
199 } |
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
200 |
704 | 201 private void wrote() { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
202 writeCounter.incrementAndGet(); |
704 | 203 } |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
204 |
1392 | 205 public void delete_all() throws IOException, SQLException { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
206 boolean commit = !writeLock.isHeldByCurrentThread(); |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
207 writeLock.lock(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
208 try { |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
209 writer.deleteAll(); |
1539 | 210 id = 0; |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
211 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
212 postgresBackup.deleteAll(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
213 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
214 } finally { |
704 | 215 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
216 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
217 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
218 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
219 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
220 private static Term term(String key,long value) { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
221 BytesRef br = new BytesRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
222 NumericUtils.longToPrefixCoded(value,0,br); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
223 return new Term(key,br); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
224 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
225 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
226 private void backupDelete(Query query) |
1392 | 227 throws IOException, SQLException, LuanException |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
228 { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
229 if( postgresBackup != null ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
230 final List<Long> ids = new ArrayList<Long>(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
231 IndexSearcher searcher = openSearcher(); |
1391 | 232 try { |
233 MyCollector col = new MyCollector() { | |
234 @Override public void collect(int iDoc) throws IOException { | |
235 Document doc = searcher.doc( docBase + iDoc ); | |
236 Long id = (Long)doc.getField("id").numericValue(); | |
237 ids.add(id); | |
238 } | |
239 }; | |
240 searcher.search(query,col); | |
241 } finally { | |
242 close(searcher); | |
243 } | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
244 postgresBackup.begin(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
245 for( Long id : ids ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
246 postgresBackup.delete(id); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
247 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
248 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
249 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
250 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
251 |
1346 | 252 public void delete(String queryStr) |
1392 | 253 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
|
254 { |
1459 | 255 Query query = GoodQueryParser.parseQuery(mfp,queryStr); |
546
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 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
258 writeLock.lock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
259 try { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
260 backupDelete(query); |
547
0be287ab0309
add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents:
546
diff
changeset
|
261 writer.deleteDocuments(query); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
262 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
263 } finally { |
704 | 264 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
265 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
266 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
267 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
268 |
1535 | 269 public void reindex(String queryStr) |
270 throws IOException, ParseException | |
271 { | |
272 Query query = GoodQueryParser.parseQuery(mfp,queryStr); | |
273 | |
274 boolean commit = !writeLock.isHeldByCurrentThread(); | |
275 writeLock.lock(); | |
276 try { | |
277 writer.reindexDocuments("id",query); | |
278 if(commit) writer.commit(); | |
279 } finally { | |
280 wrote(); | |
281 writeLock.unlock(); | |
282 } | |
283 } | |
284 | |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
285 public void save(LuanTable doc) |
1392 | 286 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
|
287 { |
1267 | 288 Object obj = doc.get("id"); |
601 | 289 Long id; |
290 try { | |
291 id = (Long)obj; | |
292 } catch(ClassCastException e) { | |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
293 throw new LuanException("id should be Long but is "+obj.getClass().getSimpleName()); |
601 | 294 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
295 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
296 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
297 writeLock.lock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
298 try { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
299 if( id == null ) { |
1539 | 300 id = ++this.id; |
1267 | 301 doc.put("id",id); |
1391 | 302 if( postgresBackup != null ) |
303 postgresBackup.add(doc); | |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
304 writer.addDocument(toLucene(doc)); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
305 } else { |
1391 | 306 if( postgresBackup != null ) |
307 postgresBackup.update(doc); | |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
308 writer.updateDocument( "id", toLucene(doc) ); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
309 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
310 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
311 } finally { |
704 | 312 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
313 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
314 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
315 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
316 |
1533
28f06ac6ce39
add not_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents:
1532
diff
changeset
|
317 public boolean is_in_transaction() { |
28f06ac6ce39
add not_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents:
1532
diff
changeset
|
318 return writeLock.isHeldByCurrentThread(); |
28f06ac6ce39
add not_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents:
1532
diff
changeset
|
319 } |
28f06ac6ce39
add not_in_transaction
Franklin Schmidt <fschmidt@gmail.com>
parents:
1532
diff
changeset
|
320 |
1392 | 321 public Object run_in_transaction(LuanFunction fn) |
322 throws IOException, LuanException, SQLException | |
323 { | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
324 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
325 writeLock.lock(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
326 boolean ok = false; |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
327 try { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
328 if( commit && postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
329 postgresBackup.begin(); |
1379 | 330 Object rtn = fn.call(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
331 ok = true; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
332 if(commit) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
333 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
334 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
335 writer.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
336 } |
1379 | 337 return rtn; |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
338 } finally { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
339 if( !ok && commit ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
340 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
341 postgresBackup.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
342 writer.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
343 reopen(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
344 } |
704 | 345 wrote(); |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
346 writeLock.unlock(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
347 } |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
348 } |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
349 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
350 // ??? |
1379 | 351 public Object run_in_lock(LuanFunction fn) throws IOException, LuanException { |
756
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
352 if( writeLock.isHeldByCurrentThread() ) |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
353 throw new RuntimeException(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
354 writeLock.lock(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
355 try { |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
356 synchronized(this) { |
1379 | 357 return fn.call(); |
756
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
358 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
359 } finally { |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
360 wrote(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
361 writeLock.unlock(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
362 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
363 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
364 |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
365 |
754 | 366 private long id; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
367 |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
368 private void initId() throws IOException { |
1539 | 369 TopDocs td = searcher.search(new MatchAllDocsQuery(),1,new Sort(ID_DESC_SORT)); |
370 switch(td.scoreDocs.length) { | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
371 case 0: |
754 | 372 id = 0; |
373 break; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
374 case 1: |
1539 | 375 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
|
376 break; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
377 default: |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
378 throw new RuntimeException(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
379 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
380 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
381 |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
382 /* |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
383 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
|
384 if( !zipFile.endsWith(".zip") ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
385 throw new LuanException("file "+zipFile+" doesn't end with '.zip'"); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
386 IndexCommit ic = snapshotDeletionPolicy.snapshot(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
387 try { |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
388 ZipOutputStream out = new ZipOutputStream(new FileOutputStream(zipFile)); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
389 for( String fileName : ic.getFileNames() ) { |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
390 out.putNextEntry(new ZipEntry(fileName)); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
391 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
|
392 Utils.copyAll(in,out); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
393 in.close(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
394 out.closeEntry(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
395 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
396 out.close(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
397 } finally { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
398 snapshotDeletionPolicy.release(ic); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
399 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
400 } |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
401 */ |
1129
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
402 public SnapshotDeletionPolicy snapshotDeletionPolicy() { |
1539 | 403 return (SnapshotDeletionPolicy)writer.getLuceneIndexWriter().getConfig().getIndexDeletionPolicy(); |
1129
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
404 } |
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
405 |
1346 | 406 public Object snapshot(LuanFunction fn) throws LuanException, IOException { |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
407 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
|
408 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
|
409 try { |
1391 | 410 String dir = fsDir.getDirectory().toString(); |
1346 | 411 LuanTable fileNames = new LuanTable(fn.luan(),new ArrayList(ic.getFileNames())); |
1335 | 412 return fn.call(dir,fileNames); |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
413 } finally { |
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
414 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
|
415 } |
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
416 } |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
417 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
418 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
419 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
420 public String to_string() { |
1539 | 421 return writer.getLuceneIndexWriter().getDirectory().toString(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
422 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
423 |
1398 | 424 protected void finalize() throws Throwable { |
425 close(); | |
426 super.finalize(); | |
521
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
427 } |
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
428 |
1398 | 429 public void close() throws IOException, SQLException { |
430 closeWriter(); | |
1392 | 431 reader.close(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
432 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
433 |
1398 | 434 private void closeWriter() throws IOException, SQLException { |
1397 | 435 writeLock.lock(); |
436 try { | |
437 writer.close(); | |
1398 | 438 if( postgresBackup != null ) |
439 postgresBackup.close(); | |
1397 | 440 } finally { |
441 writeLock.unlock(); | |
442 } | |
443 } | |
444 | |
312
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
445 |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
446 private static class DocFn extends LuanFunction { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
447 final IndexSearcher searcher; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
448 final Query query; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
449 int docID; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
450 |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
451 DocFn(Luan luan,IndexSearcher searcher,Query query) { |
1335 | 452 super(luan); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
453 this.searcher = searcher; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
454 this.query = query; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
455 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
456 |
1335 | 457 @Override public Object call(Object[] args) throws LuanException { |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
458 try { |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
459 LuanTable doc = toTable(luan(),searcher.doc(docID)); |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
460 if( args.length > 0 && "explain".equals(args[0]) ) { |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
461 Explanation explanation = searcher.explain(query,docID); |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
462 return new Object[]{doc,explanation}; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
463 } else { |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
464 return doc; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
465 } |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
466 } catch(IOException e) { |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
467 throw new LuanException(e); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
468 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
469 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
470 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
471 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
472 private static abstract class MyCollector extends Collector { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
473 int docBase; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
474 int i = 0; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
475 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
476 @Override public void setScorer(Scorer scorer) {} |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
477 @Override public void setNextReader(AtomicReaderContext context) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
478 this.docBase = context.docBase; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
479 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
480 @Override public boolean acceptsDocsOutOfOrder() { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
481 return true; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
482 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
483 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
484 |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
485 private synchronized IndexSearcher openSearcher() throws IOException { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
486 int gwc = writeCounter.get(); |
704 | 487 if( writeCount != gwc ) { |
488 writeCount = gwc; | |
489 DirectoryReader newReader = DirectoryReader.openIfChanged(reader); | |
1539 | 490 // DirectoryReader newReader = DirectoryReader.openIfChanged(reader,writer.getLuceneIndexWriter(),true); |
704 | 491 if( newReader != null ) { |
492 reader.decRef(); | |
493 reader = newReader; | |
494 searcher = new IndexSearcher(reader); | |
495 } | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
496 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
497 reader.incRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
498 return searcher; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
499 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
500 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
501 // call in finally block |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
502 private static void close(IndexSearcher searcher) throws IOException { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
503 searcher.getIndexReader().decRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
504 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
505 |
591
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
506 public void ensure_open() throws IOException { |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
507 close(openSearcher()); |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
508 } |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
509 |
1346 | 510 public int advanced_search( String queryStr, LuanFunction fn, Integer n, String sortStr ) |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
511 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
|
512 { |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
513 Utils.checkNotNull(queryStr); |
1459 | 514 Query query = GoodQueryParser.parseQuery(mfp,queryStr); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
515 IndexSearcher searcher = threadLocalSearcher.get(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
516 boolean inTransaction = searcher != null; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
517 if( !inTransaction ) |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
518 searcher = openSearcher(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
519 try { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
520 if( fn!=null && n==null ) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
521 if( sortStr != null ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
522 throw new LuanException("sort must be nil when n is nil"); |
1346 | 523 final DocFn docFn = new DocFn(fn.luan(),searcher,query); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
524 MyCollector col = new MyCollector() { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
525 @Override public void collect(int doc) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
526 try { |
547
0be287ab0309
add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents:
546
diff
changeset
|
527 docFn.docID = docBase + doc; |
1335 | 528 fn.call(++i,docFn); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
529 } catch(LuanException e) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
530 throw new LuanRuntimeException(e); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
531 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
532 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
533 }; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
534 try { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
535 searcher.search(query,col); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
536 } catch(LuanRuntimeException e) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
537 throw (LuanException)e.getCause(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
538 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
539 return col.i; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
540 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
541 if( fn==null || n==0 ) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
542 TotalHitCountCollector thcc = new TotalHitCountCollector(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
543 searcher.search(query,thcc); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
544 return thcc.getTotalHits(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
545 } |
1459 | 546 Sort sort = sortStr==null ? null : GoodQueryParser.parseSort(mfp,sortStr); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
547 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
|
548 final ScoreDoc[] scoreDocs = td.scoreDocs; |
1346 | 549 DocFn docFn = new DocFn(fn.luan(),searcher,query); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
550 for( int i=0; i<scoreDocs.length; i++ ) { |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
551 ScoreDoc scoreDoc = scoreDocs[i]; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
552 docFn.docID = scoreDoc.doc; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
553 fn.call(i+1,docFn,scoreDoc.score); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
554 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
555 return td.totalHits; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
556 } finally { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
557 if( !inTransaction ) |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
558 close(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
559 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
560 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
561 |
1335 | 562 public Object search_in_transaction(LuanFunction fn) throws LuanException, IOException { |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
563 if( threadLocalSearcher.get() != null ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
564 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
|
565 IndexSearcher searcher = openSearcher(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
566 threadLocalSearcher.set(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
567 try { |
1335 | 568 return fn.call(); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
569 } finally { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
570 threadLocalSearcher.set(null); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
571 close(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
572 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
573 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
574 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
575 |
796 | 576 public FieldParser getIndexedFieldParser(String field) { |
577 return mfp.fields.get(field); | |
578 } | |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
579 |
796 | 580 public void setIndexedFieldParser(String field,FieldParser fp) { |
581 if( fp==null ) { // delete | |
582 mfp.fields.remove(field); | |
583 return; | |
584 } | |
585 mfp.fields.put( field, fp ); | |
586 } | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
587 |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
588 static Map<String,Object> toLucene(LuanTable table) throws LuanException { |
1529 | 589 return SupplementingConfig.toLucene(table); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
590 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
591 |
1333
25746915a241
merge Luan and LuanState
Franklin Schmidt <fschmidt@gmail.com>
parents:
1271
diff
changeset
|
592 private static LuanTable toTable(Luan luan,Document doc) throws LuanException { |
1529 | 593 return doc==null ? null : SupplementingConfig.toTable(luan,LuceneUtils.toMap(doc)); |
312
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
594 } |
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
595 |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
596 |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
597 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
|
598 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
|
599 return originalText; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
600 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
601 }; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
602 |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
603 public LuanFunction highlighter(String queryStr,final LuanFunction formatter,final Integer fragmentSize,String dotdotdot) |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
604 throws ParseException |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
605 { |
1459 | 606 Query query = GoodQueryParser.parseQuery(mfp,queryStr); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
607 Formatter fmt = new Formatter() { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
608 public String highlightTerm(String originalText,TokenGroup tokenGroup) { |
1435 | 609 if( tokenGroup.getTotalScore() <= 0 ) |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
610 return originalText; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
611 try { |
1335 | 612 return (String)Luan.first(formatter.call(originalText)); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
613 } catch(LuanException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
614 throw new LuanRuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
615 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
616 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
617 }; |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
618 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
|
619 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
|
620 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
|
621 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
|
622 final Highlighter hl = new Highlighter(fmt,queryScorer); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
623 hl.setTextFragmenter( new NullFragmenter() ); |
1335 | 624 return new LuanFunction(false) { // ??? |
625 @Override public String call(Object[] args) throws LuanException { | |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
626 String text = (String)args[0]; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
627 try { |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
628 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
|
629 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
|
630 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
|
631 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
|
632 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
|
633 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
|
634 if( !atStart ) |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
635 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
|
636 if( !atEnd ) |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
637 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
|
638 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
639 text = s; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
640 } 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
|
641 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
|
642 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
|
643 text += "..."; |
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 } |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
646 String s = hl.getBestFragment(analyzer,null,text); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
647 return s!=null ? s : text; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
648 } catch(LuanRuntimeException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
649 throw (LuanException)e.getCause(); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
650 } catch(IOException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
651 throw new RuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
652 } catch(InvalidTokenOffsetsException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
653 throw new RuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
654 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
655 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
656 }; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
657 } |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
658 |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
659 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
|
660 throws IOException |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
661 { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
662 int n = 0; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
663 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
|
664 ts.reset(); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
665 while( ts.incrementToken() ) { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
666 n++; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
667 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
668 ts.close(); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
669 return n; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
670 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
671 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
672 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
673 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
674 public boolean hasPostgresBackup() { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
675 return postgresBackup != null; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
676 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
677 |
1453 | 678 public void rebuild_postgres_backup(Luan luan) |
1392 | 679 throws IOException, LuanException, SQLException |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
680 { |
1454 | 681 logger.info("start rebuild_postgres_backup"); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
682 writeLock.lock(); |
1391 | 683 IndexSearcher searcher = openSearcher(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
684 boolean ok = false; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
685 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
686 postgresBackup.begin(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
687 postgresBackup.deleteAll(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
688 Query query = new PrefixQuery(new Term("id")); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
689 MyCollector col = new MyCollector() { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
690 @Override public void collect(int iDoc) throws IOException { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
691 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
692 Document doc = searcher.doc( docBase + iDoc ); |
1453 | 693 LuanTable tbl = toTable(luan,doc); |
1391 | 694 postgresBackup.add(tbl); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
695 } catch(LuanException e) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
696 throw new LuanRuntimeException(e); |
1392 | 697 } catch(SQLException e) { |
698 throw new RuntimeException(e); | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
699 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
700 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
701 }; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
702 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
703 searcher.search(query,col); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
704 } catch(LuanRuntimeException e) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
705 throw (LuanException)e.getCause(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
706 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
707 ok = true; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
708 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
709 } finally { |
1391 | 710 close(searcher); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
711 if( !ok ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
712 postgresBackup.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
713 writeLock.unlock(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
714 } |
1454 | 715 logger.info("end rebuild_postgres_backup"); |
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 |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
718 public void restore_from_postgres() |
1447 | 719 throws IOException, LuanException, SQLException, ParseException |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
720 { |
1397 | 721 if( postgresBackup!=null && wasCreated && !postgresBackup.wasCreated ) { |
1454 | 722 logger.error("restoring from postgres"); |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
723 force_restore_from_postgres(); |
1397 | 724 } |
725 } | |
726 | |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
727 public void force_restore_from_postgres() |
1447 | 728 throws IOException, LuanException, SQLException, ParseException |
1397 | 729 { |
1454 | 730 logger.warn("start restore_from_postgres"); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
731 if( postgresBackup==null ) |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
732 throw new NullPointerException(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
733 if( writeLock.isHeldByCurrentThread() ) |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
734 throw new RuntimeException(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
735 writeLock.lock(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
736 boolean ok = false; |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
737 try { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
738 writer.deleteAll(); |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
739 postgresBackup.restoreLucene(this); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
740 ok = true; |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
741 writer.commit(); |
1539 | 742 wrote(); |
743 ensure_open(); // refresh searcher | |
744 initId(); | |
1398 | 745 wasCreated = false; |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
746 } finally { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
747 if( !ok ) { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
748 writer.rollback(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
749 reopen(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
750 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
751 wrote(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
752 writeLock.unlock(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
753 } |
1454 | 754 logger.warn("end restore_from_postgres"); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
755 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
756 |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
757 void restore(LuanTable doc) |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
758 throws LuanException, IOException |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
759 { |
1528
3bd4d7963456
use goodjava/lucene/api
Franklin Schmidt <fschmidt@gmail.com>
parents:
1526
diff
changeset
|
760 writer.addDocument(toLucene(doc)); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
761 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
762 |
1453 | 763 public void check(Luan luan) throws IOException, SQLException, LuanException, ParseException { |
1406 | 764 boolean hasPostgres = postgresBackup != null; |
765 String msg = "start check"; | |
766 if( hasPostgres ) | |
767 msg += " with postgres"; | |
1454 | 768 logger.info(msg); |
1391 | 769 CheckIndex.Status status = new CheckIndex(fsDir).checkIndex(); |
770 if( !status.clean ) | |
1454 | 771 logger.error("index not clean"); |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
772 if( writer instanceof LoggingIndexWriter ) { |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
773 LoggingIndexWriter loggingWriter = (LoggingIndexWriter)writer; |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
774 logger.info("log check"); |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
775 boolean ok = loggingWriter.check(ID_SORT); |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
776 } |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
777 if( hasPostgres ) { |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
778 logger.info("postgres check"); |
1453 | 779 checkPostgres(luan); |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
780 } |
1454 | 781 logger.info("end check"); |
1391 | 782 } |
783 | |
1538
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
784 public void rebuild_log() throws IOException { |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
785 logger.info("start rebuild_log"); |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
786 LoggingIndexWriter loggingWriter = (LoggingIndexWriter)writer; |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
787 loggingWriter.newLogs(); |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
788 logger.info("end rebuild_log"); |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
789 } |
634f6765830e
use goodjava/lucene/logging
Franklin Schmidt <fschmidt@gmail.com>
parents:
1535
diff
changeset
|
790 |
1453 | 791 private void checkPostgres(Luan luan) |
1447 | 792 throws IOException, SQLException, LuanException, ParseException |
1430 | 793 { |
794 final PostgresBackup.Checker postgresChecker = postgresBackup.newChecker(); | |
795 final IndexSearcher searcher = openSearcher(); | |
1391 | 796 try { |
797 final List<Long> idsLucene = new ArrayList<Long>(); | |
798 Query query = new PrefixQuery(new Term("id")); | |
799 MyCollector col = new MyCollector() { | |
800 @Override public void collect(int iDoc) throws IOException { | |
801 Document doc = searcher.doc( docBase + iDoc ); | |
802 Long id = (Long)doc.getField("id").numericValue(); | |
803 idsLucene.add(id); | |
804 } | |
805 }; | |
806 searcher.search(query,col); | |
807 Collections.sort(idsLucene); | |
808 final List<Long> idsPostgres = postgresChecker.getIds(); | |
809 final int nLucene = idsLucene.size(); | |
810 final int nPostgres = idsPostgres.size(); | |
811 int iLucene = 0; | |
812 int iPostgres = 0; | |
813 LuanToString lts = new LuanToString(); | |
814 lts.strict = true; | |
815 lts.numberTypes = true; | |
816 while( iLucene < nLucene && iPostgres < nPostgres ) { | |
817 long idLucene = idsLucene.get(iLucene); | |
818 long idPostgres = idsPostgres.get(iPostgres); | |
819 if( idLucene < idPostgres ) { | |
820 iLucene++; | |
1453 | 821 checkPostgres(luan,postgresChecker,lts,idLucene); |
1391 | 822 } else if( idLucene > idPostgres ) { |
823 iPostgres++; | |
1453 | 824 checkPostgres(luan,postgresChecker,lts,idPostgres); |
1391 | 825 } else { // == |
826 LuanTable docPostgres = postgresChecker.getDoc(idPostgres); | |
827 TopDocs td = searcher.search(new TermQuery(term("id",idLucene)),1); | |
828 if( td.totalHits != 1 ) throw new RuntimeException(); | |
829 Document doc = searcher.doc( td.scoreDocs[0].doc ); | |
1453 | 830 LuanTable docLucene = toTable(luan,doc); |
1391 | 831 if( !equal(docPostgres,docLucene) ) { |
1453 | 832 checkPostgres(luan,postgresChecker,lts,idPostgres); |
1391 | 833 } |
834 iLucene++; | |
835 iPostgres++; | |
836 } | |
837 } | |
838 while( iLucene < nLucene ) { | |
839 long idLucene = idsLucene.get(iLucene++); | |
1453 | 840 checkPostgres(luan,postgresChecker,lts,idLucene); |
1391 | 841 } |
842 while( iPostgres < nPostgres ) { | |
843 long idPostgres = idsPostgres.get(iPostgres++); | |
1453 | 844 checkPostgres(luan,postgresChecker,lts,idPostgres); |
1391 | 845 } |
846 } finally { | |
847 close(searcher); | |
848 postgresChecker.close(); | |
849 } | |
850 } | |
851 | |
1453 | 852 private void checkPostgres(Luan luan,PostgresBackup.Checker postgresChecker,LuanToString lts,long id) |
1447 | 853 throws IOException, SQLException, LuanException, ParseException |
1430 | 854 { |
1454 | 855 //logger.info("check id "+id); |
1430 | 856 writeLock.lock(); |
857 try { | |
858 final IndexSearcher searcher = openSearcher(); | |
859 try { | |
860 LuanTable docPostgres = postgresChecker.getDoc(id); | |
861 TopDocs td = searcher.search(new TermQuery(term("id",id)),1); | |
862 LuanTable docLucene; | |
863 if( td.totalHits == 0 ) { | |
864 docLucene = null; | |
865 } else if( td.totalHits == 1 ) { | |
866 Document doc = searcher.doc( td.scoreDocs[0].doc ); | |
1453 | 867 docLucene = toTable(luan,doc); |
1430 | 868 } else |
869 throw new RuntimeException(); | |
870 if( docPostgres == null ) { | |
871 if( docLucene != null ) | |
1454 | 872 logger.error("id "+id+" found in lucene but not postgres"); |
1430 | 873 return; |
874 } | |
875 if( docLucene == null ) { | |
1454 | 876 logger.error("id "+id+" found in postgres but not lucene"); |
1430 | 877 return; |
878 } | |
879 if( !equal(docPostgres,docLucene) ) { | |
1454 | 880 logger.error("id "+id+" not equal"); |
881 logger.error("lucene = "+lts.toString(docLucene)); | |
882 logger.error("postgres = "+lts.toString(docPostgres)); | |
1430 | 883 } |
884 } finally { | |
885 close(searcher); | |
886 } | |
887 } finally { | |
888 writeLock.unlock(); | |
889 } | |
890 } | |
891 | |
1453 | 892 private static boolean equal(LuanTable t1,LuanTable t2) throws LuanException { |
893 return t1!=null && t2!=null && toJava(t1).equals(toJava(t2)); | |
1391 | 894 } |
895 | |
1453 | 896 private static Map toJava(LuanTable t) throws LuanException { |
897 Map map = t.asMap(); | |
1526 | 898 for( Iterator iter = map.entrySet().iterator(); iter.hasNext(); ) { |
899 Map.Entry entry = (Map.Entry)iter.next(); | |
1453 | 900 Object value = entry.getValue(); |
901 if( value instanceof LuanTable ) { | |
902 LuanTable v = (LuanTable)value; | |
903 if( !v.isList() ) | |
1454 | 904 logger.error("not list"); |
1530 | 905 List list = v.asList(); |
906 if( list.isEmpty() ) { | |
1526 | 907 iter.remove(); |
1530 | 908 } else if( list.size() == 1 ) { |
909 entry.setValue(list.get(0)); | |
1526 | 910 } else { |
1530 | 911 entry.setValue(list); |
1526 | 912 } |
1453 | 913 } |
914 } | |
915 return map; | |
916 } | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
917 } |