1
|
/**
|
2
|
* Copyright (C) 2007 EDIT
|
3
|
* European Distributed Institute of Taxonomy
|
4
|
* http://www.e-taxonomy.eu
|
5
|
*
|
6
|
* The contents of this file are subject to the Mozilla Public License Version 1.1
|
7
|
* See LICENSE.TXT at the top of this package for the full license terms.
|
8
|
*/
|
9
|
package eu.etaxonomy.cdm.io.stream;
|
10
|
|
11
|
import java.util.ArrayList;
|
12
|
import java.util.List;
|
13
|
import java.util.Map;
|
14
|
import java.util.Set;
|
15
|
|
16
|
import org.apache.log4j.Logger;
|
17
|
|
18
|
import eu.etaxonomy.cdm.io.stream.mapping.IImportMapping;
|
19
|
import eu.etaxonomy.cdm.io.stream.mapping.InMemoryMapping;
|
20
|
import eu.etaxonomy.cdm.io.stream.terms.TermUri;
|
21
|
import eu.etaxonomy.cdm.model.common.CdmBase;
|
22
|
import eu.etaxonomy.cdm.model.common.IdentifiableEntity;
|
23
|
import eu.etaxonomy.cdm.model.reference.Reference;
|
24
|
import eu.etaxonomy.cdm.model.term.DefinedTermBase;
|
25
|
|
26
|
|
27
|
/**
|
28
|
* @author a.mueller
|
29
|
*
|
30
|
*/
|
31
|
public class StreamPartitioner<ITEM extends IConverterInput>
|
32
|
implements INamespaceReader<IReader<MappedCdmBase<? extends CdmBase>>>{
|
33
|
private static final Logger logger = Logger.getLogger(StreamPartitioner.class);
|
34
|
|
35
|
private final int partitionSize;
|
36
|
private final LookAheadStream<ITEM> inStream;
|
37
|
private final IPartitionableConverter converter;
|
38
|
private final StreamImportStateBase<StreamImportConfiguratorBase, StreamImportBase> state;
|
39
|
private ConcatenatingReader<MappedCdmBase<? extends CdmBase>> outStream;
|
40
|
|
41
|
public StreamPartitioner(INamespaceReader<ITEM> input, IPartitionableConverter converter,
|
42
|
StreamImportStateBase<StreamImportConfiguratorBase, StreamImportBase> state, Integer size){
|
43
|
this.inStream = new LookAheadStream<>(input);
|
44
|
this.converter = converter;
|
45
|
this.partitionSize = size;
|
46
|
this.state = state;
|
47
|
initNewOutStream();
|
48
|
}
|
49
|
|
50
|
|
51
|
private void initNewOutStream(){
|
52
|
outStream = new ConcatenatingReader<>();
|
53
|
}
|
54
|
|
55
|
@Override
|
56
|
public boolean hasNext() {
|
57
|
if (this.outStream.hasNext()){
|
58
|
return true;
|
59
|
}else{
|
60
|
return inStream.hasNext(); //TODO what, if converter returns no ouput for inStream.hasNext() ??
|
61
|
//but be aware that requesting the next object from the next partition crosses the transactional borders
|
62
|
}
|
63
|
}
|
64
|
|
65
|
@Override
|
66
|
public IReader<MappedCdmBase<? extends CdmBase>> read() {
|
67
|
logger.debug("Start partitioner read");
|
68
|
handleNextPartition();
|
69
|
IReader<MappedCdmBase<? extends CdmBase>> result = this.outStream;
|
70
|
|
71
|
initNewOutStream();
|
72
|
logger.debug("End partitioner read");
|
73
|
return result;
|
74
|
}
|
75
|
|
76
|
private void handleNextPartition(){
|
77
|
List<ITEM> lookaheadArray = new ArrayList<>();
|
78
|
while (this.inStream.hasNextLookAhead(partitionSize)){
|
79
|
lookaheadArray.add(this.inStream.readLookAhead());
|
80
|
}
|
81
|
|
82
|
IReader<ITEM> lookaheadStream = new ListReader<>(lookaheadArray);
|
83
|
|
84
|
Map<String, Set<String>> foreignKeys = converter.getPartitionForeignKeys(lookaheadStream);
|
85
|
IImportMapping mapping = state.getMapping();
|
86
|
InMemoryMapping partialMapping = mapping.getPartialMapping(foreignKeys);
|
87
|
Reference sourceRef = state.getCurrentIO().getReferenceService().find(state.getConfig().getSourceRefUuid());
|
88
|
partialMapping.putMapping(TermUri.CDM_SOURCE_REFERENCE.toString(), state.getConfig().getSourceRefUuid().toString(), sourceRef);
|
89
|
|
90
|
state.loadRelatedObjects(partialMapping);
|
91
|
|
92
|
while (inStream.isLookingAhead() && inStream.hasNext()){
|
93
|
IReader<MappedCdmBase<? extends CdmBase>> resultReader = converter.map(inStream.read());
|
94
|
List<MappedCdmBase<? extends CdmBase>> resultList = new ArrayList<>(); //maybe better let converter return list from the beginning
|
95
|
while (resultReader.hasNext()){
|
96
|
MappedCdmBase<? extends CdmBase> item = resultReader.read();
|
97
|
resultList.add(item);
|
98
|
addItemToRelatedObjects(item);
|
99
|
}
|
100
|
outStream.add(new ListReader<>(resultList));
|
101
|
}
|
102
|
|
103
|
return;
|
104
|
|
105
|
}
|
106
|
|
107
|
|
108
|
/**
|
109
|
* Add new items to the local mapping
|
110
|
* @param item
|
111
|
*/
|
112
|
private void addItemToRelatedObjects(MappedCdmBase<? extends CdmBase> item) {
|
113
|
CdmBase cdmBase = item.getCdmBase();
|
114
|
if (cdmBase.getId() == 0 || cdmBase.isInstanceOf(DefinedTermBase.class)){
|
115
|
if (cdmBase.isInstanceOf(IdentifiableEntity.class)){
|
116
|
IdentifiableEntity<?> identifiableEntity = CdmBase.deproxy(cdmBase, IdentifiableEntity.class);
|
117
|
Set<String> requiredSourceNamespaces = converter.requiredSourceNamespaces();
|
118
|
if (requiredSourceNamespaces.contains(item.getNamespace())){
|
119
|
state.addRelatedObject(item.getNamespace(), item.getSourceId(), identifiableEntity);
|
120
|
}
|
121
|
}else{
|
122
|
if (logger.isTraceEnabled()){logger.trace("Non identifiable are not added to related objects");}
|
123
|
}
|
124
|
}
|
125
|
}
|
126
|
|
127
|
|
128
|
@Override
|
129
|
public TermUri getTerm() {
|
130
|
return inStream.getTerm();
|
131
|
}
|
132
|
|
133
|
|
134
|
}
|