Project

General

Profile

Download (4.5 KB) Statistics
| Branch: | Tag: | Revision:
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.dwca.in;
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.dwca.TermUri;
19
import eu.etaxonomy.cdm.io.stream.StreamImportBase;
20
import eu.etaxonomy.cdm.io.stream.StreamImportConfiguratorBase;
21
import eu.etaxonomy.cdm.io.stream.StreamImportStateBase;
22
import eu.etaxonomy.cdm.model.common.CdmBase;
23
import eu.etaxonomy.cdm.model.common.DefinedTermBase;
24
import eu.etaxonomy.cdm.model.common.IdentifiableEntity;
25
import eu.etaxonomy.cdm.model.reference.Reference;
26

    
27

    
28
/**
29
 * @author a.mueller
30
 *
31
 */
32
public class StreamPartitioner<ITEM extends IConverterInput>
33
            implements INamespaceReader<IReader<MappedCdmBase<? extends CdmBase>>>{
34
	private static final Logger logger = Logger.getLogger(StreamPartitioner.class);
35

    
36
	private final int partitionSize;
37
	private final LookAheadStream<ITEM> inStream;
38
	private final IPartitionableConverter converter;
39
	private final StreamImportStateBase<StreamImportConfiguratorBase, StreamImportBase> state;
40
	private ConcatenatingReader<MappedCdmBase<? extends CdmBase>> outStream;
41

    
42
	public StreamPartitioner(INamespaceReader<ITEM> input, IPartitionableConverter converter,
43
	        StreamImportStateBase<StreamImportConfiguratorBase, StreamImportBase> state, Integer size){
44
		 this.inStream = new LookAheadStream<>(input);
45
		 this.converter = converter;
46
		 this.partitionSize = size;
47
		 this.state = state;
48
		 initNewOutStream();
49
	}
50

    
51

    
52
	private void initNewOutStream(){
53
		outStream = new ConcatenatingReader<>();
54
	}
55

    
56
	@Override
57
    public boolean hasNext() {
58
		if (this.outStream.hasNext()){
59
			return true;
60
		}else{
61
			return inStream.hasNext();  //TODO what, if converter returns no ouput for inStream.hasNext() ??
62
			//but be aware that requesting the next object from the next partition crosses the transactional borders
63
		}
64
	}
65

    
66
	@Override
67
	public IReader<MappedCdmBase<? extends CdmBase>> read() {
68
		logger.debug("Start partitioner read");
69
		handleNextPartition();
70
		IReader<MappedCdmBase<? extends CdmBase>> result = this.outStream;
71

    
72
		initNewOutStream();
73
		logger.debug("End partitioner read");
74
		return result;
75
	}
76

    
77
	private void handleNextPartition(){
78
	    List<ITEM> lookaheadArray = new ArrayList<>();
79
		while (this.inStream.hasNextLookAhead(partitionSize)){
80
			lookaheadArray.add(this.inStream.readLookAhead());
81
		}
82

    
83
		IReader<ITEM> lookaheadStream = new ListReader<>(lookaheadArray);
84

    
85
		Map<String, Set<String>> foreignKeys = converter.getPartitionForeignKeys(lookaheadStream);
86
		IImportMapping mapping = state.getMapping();
87
		InMemoryMapping partialMapping = mapping.getPartialMapping(foreignKeys);
88
		Reference sourceRef = state.getCurrentIO().getReferenceService().find(state.getConfig().getSourceRefUuid());
89
		partialMapping.putMapping(TermUri.CDM_SOURCE_REFERENCE.toString(), state.getConfig().getSourceRefUuid().toString(), sourceRef);
90

    
91
		state.loadRelatedObjects(partialMapping);
92

    
93
		while (inStream.isLookingAhead() && inStream.hasNext()){
94
			IReader<MappedCdmBase<? extends CdmBase>> resultReader = converter.map(inStream.read());
95
			List<MappedCdmBase<? extends CdmBase>> resultList = new ArrayList<>();  //maybe better let converter return list from the beginning
96
			while (resultReader.hasNext()){
97
				MappedCdmBase<? extends CdmBase> item = resultReader.read();
98
				resultList.add(item);
99
				addItemToRelatedObjects(item);
100
			}
101
			outStream.add(new ListReader<>(resultList));
102
		}
103

    
104
		return;
105

    
106
	}
107

    
108

    
109
	/**
110
	 * Add new items to the local mapping
111
	 * @param item
112
	 */
113
	private void addItemToRelatedObjects(MappedCdmBase<? extends CdmBase> item) {
114
		CdmBase cdmBase = item.getCdmBase();
115
		if (cdmBase.getId() == 0 || cdmBase.isInstanceOf(DefinedTermBase.class)){
116
			if (cdmBase.isInstanceOf(IdentifiableEntity.class)){
117
			    IdentifiableEntity<?> identifiableEntity = CdmBase.deproxy(cdmBase, IdentifiableEntity.class);
118
			    Set<String> requiredSourceNamespaces = converter.requiredSourceNamespaces();
119
				if (requiredSourceNamespaces.contains(item.getNamespace())){
120
					state.addRelatedObject(item.getNamespace(), item.getSourceId(), identifiableEntity);
121
				}
122
			}else{
123
			    if (logger.isTraceEnabled()){logger.trace("Non identifiable are not added to related objects");}
124
			}
125
		}
126
	}
127

    
128

    
129
	@Override
130
	public TermUri getTerm() {
131
		return inStream.getTerm();
132
	}
133

    
134

    
135
}
(37-37/37)