Mercurial Hosting > luan
annotate src/luan/modules/lucene/LuceneIndex.java @ 1397:0dc9837c16be
fix lucene issues
author | Franklin Schmidt <fschmidt@gmail.com> |
---|---|
date | Wed, 11 Sep 2019 01:31:21 -0600 |
parents | 9dfff82dfc59 |
children | 67c0e47b5be3 |
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; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
54 import org.apache.lucene.search.TopDocs; |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
55 import org.apache.lucene.search.Sort; |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
56 import org.apache.lucene.search.SortField; |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
57 import org.apache.lucene.search.IndexSearcher; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
58 import org.apache.lucene.search.TotalHitCountCollector; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
59 import org.apache.lucene.search.ScoreDoc; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
60 import org.apache.lucene.search.Collector; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
61 import org.apache.lucene.search.Scorer; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
62 import org.apache.lucene.search.Explanation; |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
63 import org.apache.lucene.search.highlight.Formatter; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
64 import org.apache.lucene.search.highlight.Highlighter; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
65 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
|
66 import org.apache.lucene.search.highlight.Fragmenter; |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
67 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
|
68 import org.apache.lucene.search.highlight.SimpleSpanFragmenter; |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
69 import org.apache.lucene.search.highlight.QueryScorer; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
70 import org.apache.lucene.search.highlight.TokenGroup; |
1344
dc2af9d5463b
move queryparser to lib
Franklin Schmidt <fschmidt@gmail.com>
parents:
1343
diff
changeset
|
71 import luan.lib.queryparser.SaneQueryParser; |
dc2af9d5463b
move queryparser to lib
Franklin Schmidt <fschmidt@gmail.com>
parents:
1343
diff
changeset
|
72 import luan.lib.queryparser.FieldParser; |
dc2af9d5463b
move queryparser to lib
Franklin Schmidt <fschmidt@gmail.com>
parents:
1343
diff
changeset
|
73 import luan.lib.queryparser.MultiFieldParser; |
dc2af9d5463b
move queryparser to lib
Franklin Schmidt <fschmidt@gmail.com>
parents:
1343
diff
changeset
|
74 import luan.lib.queryparser.StringFieldParser; |
dc2af9d5463b
move queryparser to lib
Franklin Schmidt <fschmidt@gmail.com>
parents:
1343
diff
changeset
|
75 import luan.lib.queryparser.NumberFieldParser; |
1111
88b5b81cad4a
move Parser to luan.lib.parser
Franklin Schmidt <fschmidt@gmail.com>
parents:
1110
diff
changeset
|
76 import luan.lib.parser.ParseException; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
77 import luan.modules.Utils; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
78 import luan.Luan; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
79 import luan.LuanTable; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
80 import luan.LuanFunction; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
81 import luan.LuanException; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
82 import luan.LuanRuntimeException; |
1391 | 83 import luan.modules.parsers.LuanToString; |
1337 | 84 import luan.lib.logging.Logger; |
85 import luan.lib.logging.LoggerFactory; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
86 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
87 |
1397 | 88 public final class LuceneIndex implements Closeable { |
1393 | 89 private static final Logger sysLogger = LoggerFactory.getLogger(LuceneIndex.class); |
521
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
90 |
1397 | 91 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
|
92 |
1397 | 93 public static LuceneIndex getLuceneIndex(Luan luan,File indexDir,LuanTable options) |
1392 | 94 throws LuanException, IOException, ClassNotFoundException, SQLException |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
95 { |
1369
709f7498a363
change Lucene.index() and add Lucene.recover()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1347
diff
changeset
|
96 String key = indexDir.getCanonicalPath(); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
97 synchronized(indexes) { |
1397 | 98 Reference<LuceneIndex> ref = indexes.get(key); |
99 if( ref != null ) { | |
100 LuceneIndex li = ref.get(); | |
101 if( li != null ) | |
102 li.closeWriter(); | |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
103 } |
1397 | 104 LuceneIndex li = new LuceneIndex(luan,indexDir,options); |
105 indexes.put(key, new WeakReference<LuceneIndex>(li)); | |
106 return li; | |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
107 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
108 } |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
109 |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
110 private static final Version version = Version.LUCENE_4_9; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
111 private static final String FLD_NEXT_ID = "nextId"; |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
112 public static final StringFieldParser STRING_FIELD_PARSER = new StringFieldParser(new KeywordAnalyzer()); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
113 public static final StringFieldParser ENGLISH_FIELD_PARSER = new StringFieldParser(new EnglishAnalyzer(version)); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
114 |
1393 | 115 private final Logger luanLogger; |
116 | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
117 private final ReentrantLock writeLock = new ReentrantLock(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
118 private final File indexDir; |
754 | 119 private SnapshotDeletionPolicy snapshotDeletionPolicy; |
120 private IndexWriter writer; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
121 private DirectoryReader reader; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
122 private IndexSearcher searcher; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
123 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
|
124 private final MultiFieldParser mfp; |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
125 private final Analyzer analyzer; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
126 |
1391 | 127 private FSDirectory fsDir; |
704 | 128 private int writeCount; |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
129 private AtomicInteger writeCounter = new AtomicInteger(); |
704 | 130 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
131 private Set<String> indexOnly = new HashSet<String>(); |
1271
48f302bdc187
fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents:
1267
diff
changeset
|
132 |
1397 | 133 private boolean isClosed = false; |
134 private final Exception created = new Exception("created"); | |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
135 private final FieldParser defaultFieldParser; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
136 private final String[] defaultFields; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
137 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
138 private final PostgresBackup postgresBackup; |
1397 | 139 private boolean wasCreated; |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
140 |
1397 | 141 private LuceneIndex(Luan luan,File indexDir,LuanTable options) |
1392 | 142 throws LuanException, IOException, ClassNotFoundException, SQLException |
1267 | 143 { |
1397 | 144 Map map = options.asMap(); |
145 FieldParser defaultFieldParser = (FieldParser)map.remove("default_type"); | |
146 LuanTable defaultFieldsTbl = Utils.removeTable(map,"default_fields"); | |
147 String[] defaultFields = defaultFieldsTbl==null ? null : (String[])defaultFieldsTbl.asList().toArray(new String[0]); | |
148 LuanFunction completer = Utils.removeFunction(map,"completer"); | |
149 LuanTable postgresSpec = Utils.removeTable(map,"postgres_spec"); | |
150 Utils.checkEmpty(map); | |
151 | |
1393 | 152 this.luanLogger = luan.getLogger(LuceneIndex.class); |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
153 this.defaultFieldParser = defaultFieldParser; |
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
154 this.defaultFields = defaultFields; |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
155 mfp = defaultFieldParser==null ? new MultiFieldParser() : new MultiFieldParser(defaultFieldParser,defaultFields); |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
156 mfp.fields.put( "type", STRING_FIELD_PARSER ); |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
157 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
|
158 this.indexDir = indexDir; |
599
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
159 Analyzer analyzer = STRING_FIELD_PARSER.analyzer; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
160 if( defaultFieldParser instanceof StringFieldParser ) { |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
161 StringFieldParser sfp = (StringFieldParser)defaultFieldParser; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
162 analyzer = sfp.analyzer; |
50540f0813e2
support default search fields in lucene;
Franklin Schmidt <fschmidt@gmail.com>
parents:
591
diff
changeset
|
163 } |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
164 this.analyzer = analyzer; |
1397 | 165 wasCreated = reopen(); |
1392 | 166 if( postgresSpec == null ) { |
167 postgresBackup = null; | |
168 } else { | |
1395
9dfff82dfc59
finish postgres work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1393
diff
changeset
|
169 if( completer == null ) |
9dfff82dfc59
finish postgres work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1393
diff
changeset
|
170 throw new LuanException("completer is required for postgres_spec"); |
1392 | 171 Map spec = postgresSpec.asMap(); |
1393 | 172 postgresBackup = new PostgresBackup(luan,spec); |
1397 | 173 if( !wasCreated && postgresBackup.wasCreated ) { |
174 luanLogger.error("rebuilding postgres backup"); | |
175 rebuild_postgres_backup(completer); | |
176 /* | |
177 } else if( wasCreated && !postgresBackup.wasCreated ) { | |
178 luanLogger.error("restoring from postgres"); | |
179 restore_from_postgres(); | |
180 */ | |
1390 | 181 } |
182 } | |
1397 | 183 luan.onClose(this); |
184 } | |
185 | |
186 protected void finalize() throws Throwable { | |
187 if( !isClosed ) { | |
188 sysLogger.error("not closed",created); | |
189 close(); | |
190 } | |
191 super.finalize(); | |
754 | 192 } |
193 | |
1390 | 194 public boolean reopen() throws IOException { |
233
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
195 IndexWriterConfig conf = new IndexWriterConfig(version,analyzer); |
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
196 snapshotDeletionPolicy = new SnapshotDeletionPolicy(conf.getIndexDeletionPolicy()); |
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
197 conf.setIndexDeletionPolicy(snapshotDeletionPolicy); |
1391 | 198 fsDir = FSDirectory.open(indexDir); |
199 boolean wasCreated = !fsDir.getDirectory().exists(); | |
200 writer = new IndexWriter(fsDir,conf); | |
233
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
201 writer.commit(); // commit index creation |
1391 | 202 reader = DirectoryReader.open(fsDir); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
203 searcher = new IndexSearcher(reader); |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
204 initId(); |
1390 | 205 return wasCreated; |
233
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
206 } |
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
207 |
704 | 208 private void wrote() { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
209 writeCounter.incrementAndGet(); |
704 | 210 } |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
211 |
1392 | 212 public void delete_all() throws IOException, SQLException { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
213 boolean commit = !writeLock.isHeldByCurrentThread(); |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
214 writeLock.lock(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
215 try { |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
216 writer.deleteAll(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
217 id = idLim = 0; |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
218 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
219 postgresBackup.deleteAll(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
220 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
221 } finally { |
704 | 222 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
223 writeLock.unlock(); |
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 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
226 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
227 private static Term term(String key,long value) { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
228 BytesRef br = new BytesRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
229 NumericUtils.longToPrefixCoded(value,0,br); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
230 return new Term(key,br); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
231 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
232 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
233 private void backupDelete(Query query) |
1392 | 234 throws IOException, SQLException, LuanException |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
235 { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
236 if( postgresBackup != null ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
237 final List<Long> ids = new ArrayList<Long>(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
238 IndexSearcher searcher = openSearcher(); |
1391 | 239 try { |
240 MyCollector col = new MyCollector() { | |
241 @Override public void collect(int iDoc) throws IOException { | |
242 Document doc = searcher.doc( docBase + iDoc ); | |
243 Long id = (Long)doc.getField("id").numericValue(); | |
244 ids.add(id); | |
245 } | |
246 }; | |
247 searcher.search(query,col); | |
248 } finally { | |
249 close(searcher); | |
250 } | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
251 postgresBackup.begin(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
252 for( Long id : ids ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
253 postgresBackup.delete(id); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
254 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
255 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
256 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
257 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
258 |
1346 | 259 public void delete(String queryStr) |
1392 | 260 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
|
261 { |
622
1a53333eb4d5
remove Lucene all_search_terms_must_match() since now query suffix "~a" handles this
Franklin Schmidt <fschmidt@gmail.com>
parents:
621
diff
changeset
|
262 Query query = SaneQueryParser.parseQuery(mfp,queryStr); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
263 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
264 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
265 writeLock.lock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
266 try { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
267 backupDelete(query); |
547
0be287ab0309
add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents:
546
diff
changeset
|
268 writer.deleteDocuments(query); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
269 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
270 } finally { |
704 | 271 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
272 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
273 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
274 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
275 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
276 public void indexed_only_fields(List<String> fields) { |
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
277 indexOnly.addAll(fields); |
1271
48f302bdc187
fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents:
1267
diff
changeset
|
278 } |
48f302bdc187
fix indexed_only_field
Franklin Schmidt <fschmidt@gmail.com>
parents:
1267
diff
changeset
|
279 |
1346 | 280 public void save(LuanTable doc,LuanTable boosts) |
1392 | 281 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
|
282 { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
283 if( boosts!=null && postgresBackup!=null ) |
1392 | 284 throw new LuanException("boosts are not saved to postgres backup"); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
285 |
1267 | 286 Object obj = doc.get("id"); |
601 | 287 Long id; |
288 try { | |
289 id = (Long)obj; | |
290 } catch(ClassCastException e) { | |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
291 throw new LuanException("id should be Long but is "+obj.getClass().getSimpleName()); |
601 | 292 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
293 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
294 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
295 writeLock.lock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
296 try { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
297 if( id == null ) { |
1346 | 298 id = nextId(); |
1267 | 299 doc.put("id",id); |
1391 | 300 if( postgresBackup != null ) |
301 postgresBackup.add(doc); | |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
302 writer.addDocument(toLucene(doc,boosts)); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
303 } else { |
1391 | 304 if( postgresBackup != null ) |
305 postgresBackup.update(doc); | |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
306 writer.updateDocument( term("id",id), toLucene(doc,boosts) ); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
307 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
308 if(commit) writer.commit(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
309 } finally { |
704 | 310 wrote(); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
311 writeLock.unlock(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
312 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
313 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
314 |
1392 | 315 public Object run_in_transaction(LuanFunction fn) |
316 throws IOException, LuanException, SQLException | |
317 { | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
318 boolean commit = !writeLock.isHeldByCurrentThread(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
319 writeLock.lock(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
320 boolean ok = false; |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
321 try { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
322 if( commit && postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
323 postgresBackup.begin(); |
1379 | 324 Object rtn = fn.call(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
325 ok = true; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
326 if(commit) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
327 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
328 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
329 writer.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
330 } |
1379 | 331 return rtn; |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
332 } finally { |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
333 if( !ok && commit ) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
334 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
335 postgresBackup.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
336 writer.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
337 reopen(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
338 } |
704 | 339 wrote(); |
252
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
340 writeLock.unlock(); |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
341 } |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
342 } |
3896138955b1
web testing...
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
233
diff
changeset
|
343 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
344 // ??? |
1379 | 345 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
|
346 if( writeLock.isHeldByCurrentThread() ) |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
347 throw new RuntimeException(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
348 writeLock.lock(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
349 try { |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
350 synchronized(this) { |
1379 | 351 return fn.call(); |
756
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
352 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
353 } finally { |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
354 wrote(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
355 writeLock.unlock(); |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
356 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
357 } |
9092e52f94eb
better synchronization for lucene restore
Franklin Schmidt <fschmidt@gmail.com>
parents:
754
diff
changeset
|
358 |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
359 |
754 | 360 private long id; |
361 private long idLim; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
362 private final int idBatch = 10; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
363 |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
364 private void initId() throws IOException { |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
365 TopDocs td = searcher.search(new TermQuery(new Term("type","next_id")),1); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
366 switch(td.totalHits) { |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
367 case 0: |
754 | 368 id = 0; |
369 idLim = 0; | |
370 break; | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
371 case 1: |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
372 idLim = (Long)searcher.doc(td.scoreDocs[0].doc).getField(FLD_NEXT_ID).numericValue(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
373 id = idLim; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
374 break; |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
375 default: |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
376 throw new RuntimeException(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
377 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
378 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
379 |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
380 private void saveNextId(long nextId) throws LuanException, IOException { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
381 Map doc = new HashMap(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
382 doc.put( "type", "next_id" ); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
383 doc.put( FLD_NEXT_ID, idLim ); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
384 writer.updateDocument(new Term("type","next_id"),toLucene(doc.entrySet(),null)); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
385 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
386 |
1346 | 387 public synchronized long nextId() throws LuanException, IOException { |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
388 if( ++id > idLim ) { |
233
ef39bc4d3f70
basic lucene works
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
232
diff
changeset
|
389 idLim += idBatch; |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
390 saveNextId(idLim); |
704 | 391 wrote(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
392 } |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
393 return id; |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
394 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
395 |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
396 /* |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
397 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
|
398 if( !zipFile.endsWith(".zip") ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
399 throw new LuanException("file "+zipFile+" doesn't end with '.zip'"); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
400 IndexCommit ic = snapshotDeletionPolicy.snapshot(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
401 try { |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
402 ZipOutputStream out = new ZipOutputStream(new FileOutputStream(zipFile)); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
403 for( String fileName : ic.getFileNames() ) { |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
404 out.putNextEntry(new ZipEntry(fileName)); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
405 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
|
406 Utils.copyAll(in,out); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
407 in.close(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
408 out.closeEntry(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
409 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
410 out.close(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
411 } finally { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
412 snapshotDeletionPolicy.release(ic); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
413 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
414 } |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
415 */ |
1129
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
416 public SnapshotDeletionPolicy snapshotDeletionPolicy() { |
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
417 return snapshotDeletionPolicy; |
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
418 } |
3234a14bb1f8
minor lucene changes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1111
diff
changeset
|
419 |
1346 | 420 public Object snapshot(LuanFunction fn) throws LuanException, IOException { |
707
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
421 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
|
422 try { |
1391 | 423 String dir = fsDir.getDirectory().toString(); |
1346 | 424 LuanTable fileNames = new LuanTable(fn.luan(),new ArrayList(ic.getFileNames())); |
1335 | 425 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
|
426 } finally { |
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
427 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
|
428 } |
1ed9e55f0be8
replace lucene.backup with lucene.zip implemented in luan, and add lucene.snapshot
Franklin Schmidt <fschmidt@gmail.com>
parents:
704
diff
changeset
|
429 } |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
430 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
431 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
432 |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
433 public String to_string() { |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
434 return writer.getDirectory().toString(); |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
435 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
436 |
1397 | 437 public synchronized void close() throws IOException { |
438 try { | |
439 doClose(); | |
440 } catch(SQLException e) { | |
441 throw new RuntimeException(e); | |
521
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
442 } |
1397 | 443 isClosed = true; |
521
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
444 } |
8a217fe5b4f3
cleaner LuanState.onClose()
Franklin Schmidt <fschmidt@gmail.com>
parents:
427
diff
changeset
|
445 |
1392 | 446 public void doClose() throws IOException, SQLException { |
447 writer.close(); | |
448 reader.close(); | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
449 if( postgresBackup != null ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
450 postgresBackup.close(); |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
451 } |
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
452 |
1397 | 453 private void closeWriter() throws IOException { |
454 writeLock.lock(); | |
455 try { | |
456 writer.close(); | |
457 } finally { | |
458 writeLock.unlock(); | |
459 } | |
460 } | |
461 | |
312
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
462 |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
463 private static class DocFn extends LuanFunction { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
464 final IndexSearcher searcher; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
465 final Query query; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
466 int docID; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
467 |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
468 DocFn(Luan luan,IndexSearcher searcher,Query query) { |
1335 | 469 super(luan); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
470 this.searcher = searcher; |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
471 this.query = query; |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
472 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
473 |
1335 | 474 @Override public Object call(Object[] args) throws LuanException { |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
475 try { |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
476 LuanTable doc = toTable(luan(),searcher.doc(docID)); |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
477 if( args.length > 0 && "explain".equals(args[0]) ) { |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
478 Explanation explanation = searcher.explain(query,docID); |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
479 return new Object[]{doc,explanation}; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
480 } else { |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
481 return doc; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
482 } |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
483 } catch(IOException e) { |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
484 throw new LuanException(e); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
485 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
486 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
487 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
488 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
489 private static abstract class MyCollector extends Collector { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
490 int docBase; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
491 int i = 0; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
492 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
493 @Override public void setScorer(Scorer scorer) {} |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
494 @Override public void setNextReader(AtomicReaderContext context) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
495 this.docBase = context.docBase; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
496 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
497 @Override public boolean acceptsDocsOutOfOrder() { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
498 return true; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
499 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
500 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
501 |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
502 private synchronized IndexSearcher openSearcher() throws IOException { |
1345
6f8988830098
unique LuceneIndex per dir
Franklin Schmidt <fschmidt@gmail.com>
parents:
1344
diff
changeset
|
503 int gwc = writeCounter.get(); |
704 | 504 if( writeCount != gwc ) { |
505 writeCount = gwc; | |
506 DirectoryReader newReader = DirectoryReader.openIfChanged(reader); | |
507 if( newReader != null ) { | |
508 reader.decRef(); | |
509 reader = newReader; | |
510 searcher = new IndexSearcher(reader); | |
511 } | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
512 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
513 reader.incRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
514 return searcher; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
515 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
516 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
517 // call in finally block |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
518 private static void close(IndexSearcher searcher) throws IOException { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
519 searcher.getIndexReader().decRef(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
520 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
521 |
591
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
522 public void ensure_open() throws IOException { |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
523 close(openSearcher()); |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
524 } |
790d5de23042
add "strict" param to Io.repr();
Franklin Schmidt <fschmidt@gmail.com>
parents:
578
diff
changeset
|
525 |
1346 | 526 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
|
527 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
|
528 { |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
529 Utils.checkNotNull(queryStr); |
622
1a53333eb4d5
remove Lucene all_search_terms_must_match() since now query suffix "~a" handles this
Franklin Schmidt <fschmidt@gmail.com>
parents:
621
diff
changeset
|
530 Query query = SaneQueryParser.parseQuery(mfp,queryStr); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
531 IndexSearcher searcher = threadLocalSearcher.get(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
532 boolean inTransaction = searcher != null; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
533 if( !inTransaction ) |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
534 searcher = openSearcher(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
535 try { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
536 if( fn!=null && n==null ) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
537 if( sortStr != null ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
538 throw new LuanException("sort must be nil when n is nil"); |
1346 | 539 final DocFn docFn = new DocFn(fn.luan(),searcher,query); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
540 MyCollector col = new MyCollector() { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
541 @Override public void collect(int doc) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
542 try { |
547
0be287ab0309
add lucene/Versioning and simplify Lucene fn names
Franklin Schmidt <fschmidt@gmail.com>
parents:
546
diff
changeset
|
543 docFn.docID = docBase + doc; |
1335 | 544 fn.call(++i,docFn); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
545 } catch(LuanException e) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
546 throw new LuanRuntimeException(e); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
547 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
548 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
549 }; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
550 try { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
551 searcher.search(query,col); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
552 } catch(LuanRuntimeException e) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
553 throw (LuanException)e.getCause(); |
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 col.i; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
556 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
557 if( fn==null || n==0 ) { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
558 TotalHitCountCollector thcc = new TotalHitCountCollector(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
559 searcher.search(query,thcc); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
560 return thcc.getTotalHits(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
561 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
562 Sort sort = sortStr==null ? null : SaneQueryParser.parseSort(mfp,sortStr); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
563 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
|
564 final ScoreDoc[] scoreDocs = td.scoreDocs; |
1346 | 565 DocFn docFn = new DocFn(fn.luan(),searcher,query); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
566 for( int i=0; i<scoreDocs.length; i++ ) { |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
567 ScoreDoc scoreDoc = scoreDocs[i]; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
568 docFn.docID = scoreDoc.doc; |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
569 fn.call(i+1,docFn,scoreDoc.score); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
570 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
571 return td.totalHits; |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
572 } finally { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
573 if( !inTransaction ) |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
574 close(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
575 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
576 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
577 |
1335 | 578 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
|
579 if( threadLocalSearcher.get() != null ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
580 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
|
581 IndexSearcher searcher = openSearcher(); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
582 threadLocalSearcher.set(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
583 try { |
1335 | 584 return fn.call(); |
545
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
585 } finally { |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
586 threadLocalSearcher.set(null); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
587 close(searcher); |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
588 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
589 } |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
590 |
ddcd4296107a
clean up lucene search
Franklin Schmidt <fschmidt@gmail.com>
parents:
544
diff
changeset
|
591 |
796 | 592 public FieldParser getIndexedFieldParser(String field) { |
593 return mfp.fields.get(field); | |
594 } | |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
595 |
796 | 596 public void setIndexedFieldParser(String field,FieldParser fp) { |
597 if( fp==null ) { // delete | |
598 mfp.fields.remove(field); | |
599 return; | |
600 } | |
601 mfp.fields.put( field, fp ); | |
602 } | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
603 |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
604 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
605 private IndexableField newField(String name,Object value,Set<String> indexed,Float boost) |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
606 throws LuanException |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
607 { |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
608 boolean hasBoost = boost!=null; |
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
609 IndexableField fld = newField2(name,value,indexed,hasBoost); |
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
610 if( hasBoost ) |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
611 ((Field)fld).setBoost(boost); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
612 return fld; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
613 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
614 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
615 private IndexableField newField2(String name,Object value,Set<String> indexed,boolean hasBoost) |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
616 throws LuanException |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
617 { |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
618 Field.Store store = indexOnly.contains(name) ? Field.Store.NO : Field.Store.YES; |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
619 if( value instanceof String ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
620 String s = (String)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
621 FieldParser fp = mfp.fields.get(name); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
622 if( fp != null ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
623 if( fp instanceof StringFieldParser && fp != STRING_FIELD_PARSER ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
624 return new TextField(name, s, store); |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
625 } else if (hasBoost) { |
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
626 // fuck you modern lucene developers |
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
627 return new Field(name, s, store, Field.Index.NOT_ANALYZED); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
628 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
629 return new StringField(name, s, store); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
630 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
631 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
632 return new StoredField(name, s); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
633 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
634 } else if( value instanceof Integer ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
635 int i = (Integer)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
636 if( indexed.contains(name) ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
637 return new IntField(name, i, store); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
638 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
639 return new StoredField(name, i); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
640 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
641 } else if( value instanceof Long ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
642 long i = (Long)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
643 if( indexed.contains(name) ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
644 return new LongField(name, i, store); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
645 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
646 return new StoredField(name, i); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
647 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
648 } else if( value instanceof Double ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
649 double i = (Double)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
650 if( indexed.contains(name) ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
651 return new DoubleField(name, i, store); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
652 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
653 return new StoredField(name, i); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
654 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
655 } else if( value instanceof byte[] ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
656 byte[] b = (byte[])value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
657 return new StoredField(name, b); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
658 } else |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
659 throw new LuanException("invalid value type "+value.getClass()+"' for '"+name+"'"); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
660 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
661 |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
662 private Document toLucene(LuanTable table,LuanTable boosts) throws LuanException { |
1347
643cf1c37723
move webserver to lib and bug fixes
Franklin Schmidt <fschmidt@gmail.com>
parents:
1346
diff
changeset
|
663 return toLucene(table.iterable(),boosts); |
1346 | 664 } |
665 | |
666 private Document toLucene(Iterable<Map.Entry> iterable,LuanTable boosts) throws LuanException { | |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
667 Set<String> indexed = mfp.fields.keySet(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
668 Document doc = new Document(); |
1346 | 669 for( Map.Entry<Object,Object> entry : iterable ) { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
670 Object key = entry.getKey(); |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
671 if( !(key instanceof String) ) |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
672 throw new LuanException("key must be string"); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
673 String name = (String)key; |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
674 Object value = entry.getValue(); |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
675 Float boost = null; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
676 if( boosts != null ) { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
677 Object obj = boosts.get(name); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
678 if( obj != null ) { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
679 if( !(obj instanceof Number) ) |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
680 throw new LuanException("boost '"+name+"' must be number"); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
681 boost = ((Number)obj).floatValue(); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
682 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
683 } |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
684 if( !(value instanceof LuanTable) ) { |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
685 doc.add(newField( name, value, indexed, boost )); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
686 } else { // list |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
687 LuanTable list = (LuanTable)value; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
688 for( Object el : list.asList() ) { |
1343
7d9a1f8894b0
lucene change indexed_only_field() to indexed_only_fields()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1342
diff
changeset
|
689 doc.add(newField( name, el, indexed, boost )); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
690 } |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
691 } |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
692 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
693 return doc; |
544
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
694 } |
c5a93767cc5c
lucene overhaul, untested
Franklin Schmidt <fschmidt@gmail.com>
parents:
527
diff
changeset
|
695 |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
696 private static Object getValue(IndexableField ifld) throws LuanException { |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
697 BytesRef br = ifld.binaryValue(); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
698 if( br != null ) |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
699 return br.bytes; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
700 Number n = ifld.numericValue(); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
701 if( n != null ) |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
702 return n; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
703 String s = ifld.stringValue(); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
704 if( s != null ) |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
705 return s; |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
706 throw new LuanException("invalid field type for "+ifld); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
707 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
708 |
1333
25746915a241
merge Luan and LuanState
Franklin Schmidt <fschmidt@gmail.com>
parents:
1271
diff
changeset
|
709 private static LuanTable toTable(Luan luan,Document doc) throws LuanException { |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
710 if( doc==null ) |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
711 return null; |
1267 | 712 LuanTable table = new LuanTable(luan); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
713 for( IndexableField ifld : doc ) { |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
714 String name = ifld.name(); |
646
cdc70de628b5
simplify LuanException
Franklin Schmidt <fschmidt@gmail.com>
parents:
624
diff
changeset
|
715 Object value = getValue(ifld); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
716 Object old = table.rawGet(name); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
717 if( old == null ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
718 table.rawPut(name,value); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
719 } else { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
720 LuanTable list; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
721 if( old instanceof LuanTable ) { |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
722 list = (LuanTable)old; |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
723 } else { |
1267 | 724 list = new LuanTable(luan); |
621
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
725 list.rawPut(1,old); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
726 table.rawPut(name,list); |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
727 } |
fd15da41afca
allow list of values to be stored in lucene
Franklin Schmidt <fschmidt@gmail.com>
parents:
620
diff
changeset
|
728 list.rawPut(list.rawLength()+1,value); |
546
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
729 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
730 } |
eaef1005ab87
general lucene cleanup
Franklin Schmidt <fschmidt@gmail.com>
parents:
545
diff
changeset
|
731 return table; |
312
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
732 } |
d34be4588556
add lucene query parsing
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
303
diff
changeset
|
733 |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
734 |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
735 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
|
736 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
|
737 return originalText; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
738 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
739 }; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
740 |
1342
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
741 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
|
742 throws ParseException |
60599adc27b8
add lucene search options
Franklin Schmidt <fschmidt@gmail.com>
parents:
1341
diff
changeset
|
743 { |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
744 Query query = SaneQueryParser.parseQuery(mfp,queryStr); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
745 Formatter fmt = new Formatter() { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
746 public String highlightTerm(String originalText,TokenGroup tokenGroup) { |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
747 if( tokenGroup.getTotalScore() <= 0 ) |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
748 return originalText; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
749 try { |
1335 | 750 return (String)Luan.first(formatter.call(originalText)); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
751 } catch(LuanException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
752 throw new LuanRuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
753 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
754 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
755 }; |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
756 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
|
757 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
|
758 if( chooser != null ) |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
759 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
|
760 final Highlighter hl = new Highlighter(fmt,queryScorer); |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
761 hl.setTextFragmenter( new NullFragmenter() ); |
1335 | 762 return new LuanFunction(false) { // ??? |
763 @Override public String call(Object[] args) throws LuanException { | |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
764 String text = (String)args[0]; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
765 try { |
1248
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
766 if( chooser != null ) { |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
767 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
|
768 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
|
769 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
|
770 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
|
771 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
|
772 if( !atStart ) |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
773 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
|
774 if( !atEnd ) |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
775 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
|
776 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
777 text = s; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
778 } 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
|
779 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
|
780 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
|
781 text += "..."; |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
782 } |
475905984870
improve lucene highlighter and allow bbcode_to_text quoter to be nil
Franklin Schmidt <fschmidt@gmail.com>
parents:
1227
diff
changeset
|
783 } |
624
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
784 String s = hl.getBestFragment(analyzer,null,text); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
785 return s!=null ? s : text; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
786 } catch(LuanRuntimeException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
787 throw (LuanException)e.getCause(); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
788 } catch(IOException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
789 throw new RuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
790 } catch(InvalidTokenOffsetsException e) { |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
791 throw new RuntimeException(e); |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
792 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
793 } |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
794 }; |
8281a248c47e
add lucene highlighter
Franklin Schmidt <fschmidt@gmail.com>
parents:
622
diff
changeset
|
795 } |
1341
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
796 |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
797 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
|
798 throws IOException |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
799 { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
800 int n = 0; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
801 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
|
802 ts.reset(); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
803 while( ts.incrementToken() ) { |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
804 n++; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
805 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
806 ts.close(); |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
807 return n; |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
808 } |
a015a0b5c388
add Html.decode(), Lucene.count_tokens(), lucene boosts, Sql.database.set()
Franklin Schmidt <fschmidt@gmail.com>
parents:
1337
diff
changeset
|
809 |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
810 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
811 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
812 public boolean hasPostgresBackup() { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
813 return postgresBackup != null; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
814 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
815 |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
816 public void rebuild_postgres_backup(LuanFunction completer) |
1392 | 817 throws IOException, LuanException, SQLException |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
818 { |
1393 | 819 luanLogger.info("start rebuild_postgres_backup"); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
820 writeLock.lock(); |
1391 | 821 IndexSearcher searcher = openSearcher(); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
822 boolean ok = false; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
823 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
824 postgresBackup.begin(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
825 postgresBackup.deleteAll(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
826 Query query = new PrefixQuery(new Term("id")); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
827 MyCollector col = new MyCollector() { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
828 @Override public void collect(int iDoc) throws IOException { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
829 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
830 Document doc = searcher.doc( docBase + iDoc ); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
831 LuanTable tbl = toTable(completer.luan(),doc); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
832 tbl = (LuanTable)completer.call(tbl); |
1391 | 833 postgresBackup.add(tbl); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
834 } catch(LuanException e) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
835 throw new LuanRuntimeException(e); |
1392 | 836 } catch(SQLException e) { |
837 throw new RuntimeException(e); | |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
838 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
839 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
840 }; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
841 try { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
842 searcher.search(query,col); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
843 } catch(LuanRuntimeException e) { |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
844 throw (LuanException)e.getCause(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
845 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
846 ok = true; |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
847 postgresBackup.commit(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
848 } finally { |
1391 | 849 close(searcher); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
850 if( !ok ) |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
851 postgresBackup.rollback(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
852 writeLock.unlock(); |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
853 } |
1393 | 854 luanLogger.info("end rebuild_postgres_backup"); |
1387
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
855 } |
bc40bc9aab3a
start postgres backup
Franklin Schmidt <fschmidt@gmail.com>
parents:
1379
diff
changeset
|
856 |
1393 | 857 public void restore_from_postgres() |
1392 | 858 throws IOException, LuanException, SQLException |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
859 { |
1397 | 860 if( postgresBackup!=null && wasCreated && !postgresBackup.wasCreated ) { |
861 luanLogger.error("restoring from postgres"); | |
862 force_restore_from_postgres(); | |
863 wasCreated = false; | |
864 } | |
865 } | |
866 | |
867 public void force_restore_from_postgres() | |
868 throws IOException, LuanException, SQLException | |
869 { | |
1393 | 870 luanLogger.warn("start restore_from_postgres"); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
871 if( postgresBackup==null ) |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
872 throw new NullPointerException(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
873 if( writeLock.isHeldByCurrentThread() ) |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
874 throw new RuntimeException(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
875 writeLock.lock(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
876 boolean ok = false; |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
877 try { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
878 writer.deleteAll(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
879 long nextId = postgresBackup.maxId() + 1; |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
880 postgresBackup.restoreLucene(this); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
881 id = idLim = nextId; |
1395
9dfff82dfc59
finish postgres work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1393
diff
changeset
|
882 saveNextId(nextId); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
883 ok = true; |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
884 writer.commit(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
885 } finally { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
886 if( !ok ) { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
887 writer.rollback(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
888 reopen(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
889 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
890 wrote(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
891 writeLock.unlock(); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
892 } |
1393 | 893 luanLogger.warn("end restore_from_postgres"); |
1388
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
894 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
895 |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
896 void restore(LuanTable doc) |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
897 throws LuanException, IOException |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
898 { |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
899 writer.addDocument(toLucene(doc,null)); |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
900 } |
2024d23ddd64
add restore_from_postgres
Franklin Schmidt <fschmidt@gmail.com>
parents:
1387
diff
changeset
|
901 |
1391 | 902 public void check(LuanFunction completer) throws IOException, SQLException, LuanException { |
1393 | 903 luanLogger.info("start check"); |
1391 | 904 CheckIndex.Status status = new CheckIndex(fsDir).checkIndex(); |
905 if( !status.clean ) | |
1393 | 906 luanLogger.error("index not clean"); |
1391 | 907 if( postgresBackup != null ) |
908 checkPostgres(completer); | |
1393 | 909 luanLogger.info("end check"); |
1391 | 910 } |
911 | |
912 private void checkPostgres(LuanFunction completer) throws IOException, SQLException, LuanException { | |
1395
9dfff82dfc59
finish postgres work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1393
diff
changeset
|
913 luanLogger.info("start postgres check"); |
1391 | 914 final PostgresBackup.Checker postgresChecker; |
915 final IndexSearcher searcher; | |
916 writeLock.lock(); | |
917 try { | |
918 postgresChecker = postgresBackup.newChecker(); | |
919 searcher = openSearcher(); | |
920 } finally { | |
921 writeLock.unlock(); | |
922 } | |
923 try { | |
924 final List<Long> idsLucene = new ArrayList<Long>(); | |
925 Query query = new PrefixQuery(new Term("id")); | |
926 MyCollector col = new MyCollector() { | |
927 @Override public void collect(int iDoc) throws IOException { | |
928 Document doc = searcher.doc( docBase + iDoc ); | |
929 Long id = (Long)doc.getField("id").numericValue(); | |
930 idsLucene.add(id); | |
931 } | |
932 }; | |
933 searcher.search(query,col); | |
934 Collections.sort(idsLucene); | |
935 final List<Long> idsPostgres = postgresChecker.getIds(); | |
936 final int nLucene = idsLucene.size(); | |
937 final int nPostgres = idsPostgres.size(); | |
938 int iLucene = 0; | |
939 int iPostgres = 0; | |
940 LuanToString lts = new LuanToString(); | |
941 lts.strict = true; | |
942 lts.numberTypes = true; | |
943 while( iLucene < nLucene && iPostgres < nPostgres ) { | |
944 long idLucene = idsLucene.get(iLucene); | |
945 long idPostgres = idsPostgres.get(iPostgres); | |
946 if( idLucene < idPostgres ) { | |
947 iLucene++; | |
1393 | 948 luanLogger.error("id "+idLucene+" found in lucene but not postgres"); |
1391 | 949 } else if( idLucene > idPostgres ) { |
950 iPostgres++; | |
1393 | 951 luanLogger.error("id "+idPostgres+" found in postgres but not lucene"); |
1391 | 952 } else { // == |
953 LuanTable docPostgres = postgresChecker.getDoc(idPostgres); | |
954 TopDocs td = searcher.search(new TermQuery(term("id",idLucene)),1); | |
955 if( td.totalHits != 1 ) throw new RuntimeException(); | |
956 Document doc = searcher.doc( td.scoreDocs[0].doc ); | |
957 LuanTable docLucene = toTable(completer.luan(),doc); | |
958 docLucene = (LuanTable)completer.call(docLucene); | |
959 if( !equal(docPostgres,docLucene) ) { | |
1393 | 960 luanLogger.error("id "+idLucene+" not equal"); |
961 luanLogger.error("lucene = "+lts.toString(docLucene)); | |
962 luanLogger.error("postgres = "+lts.toString(docPostgres)); | |
1391 | 963 } |
964 iLucene++; | |
965 iPostgres++; | |
966 } | |
967 } | |
968 while( iLucene < nLucene ) { | |
969 long idLucene = idsLucene.get(iLucene++); | |
1393 | 970 luanLogger.error("id "+idLucene+" found in lucene but not postgres"); |
1391 | 971 } |
972 while( iPostgres < nPostgres ) { | |
973 long idPostgres = idsPostgres.get(iPostgres++); | |
1393 | 974 luanLogger.error("id "+idPostgres+" found in postgres but not lucene"); |
1391 | 975 } |
976 } finally { | |
977 close(searcher); | |
978 postgresChecker.close(); | |
979 } | |
980 } | |
981 | |
982 private boolean equal(LuanTable t1,LuanTable t2) throws LuanException { | |
983 return t1.asMap().equals(t2.asMap()); | |
984 } | |
985 | |
230
4438cb2e04d0
start lucene
fschmidt@gmail.com <fschmidt@gmail.com@21e917c8-12df-6dd8-5cb6-c86387c605b9>
parents:
diff
changeset
|
986 } |