1
|
package eu.dnetlib.data.mapreduce.hbase.propagation.projecttoresult;
|
2
|
|
3
|
import com.google.common.collect.Sets;
|
4
|
import eu.dnetlib.data.mapreduce.hbase.propagation.NotValidResultSequenceException;
|
5
|
import eu.dnetlib.data.mapreduce.hbase.propagation.ResultIterator;
|
6
|
import eu.dnetlib.data.mapreduce.hbase.propagation.Value;
|
7
|
import eu.dnetlib.data.proto.KindProtos;
|
8
|
import eu.dnetlib.data.proto.OafProtos;
|
9
|
import eu.dnetlib.data.proto.RelMetadataProtos;
|
10
|
import eu.dnetlib.data.proto.ResultProjectProtos;
|
11
|
import org.apache.commons.lang3.StringUtils;
|
12
|
import org.apache.hadoop.io.Text;
|
13
|
|
14
|
import java.util.*;
|
15
|
|
16
|
import static eu.dnetlib.data.mapreduce.hbase.propagation.PropagationConstants.*;
|
17
|
import static eu.dnetlib.data.mapreduce.hbase.propagation.Utils.getDataInfo;
|
18
|
import static eu.dnetlib.data.mapreduce.hbase.propagation.Utils.getQualifier;
|
19
|
|
20
|
public class ResultProjectIterator extends ResultIterator {
|
21
|
|
22
|
private Iterator<String> pli ;
|
23
|
|
24
|
public ResultProjectIterator(final Iterable<Text> values, final String key) throws NotValidResultSequenceException{
|
25
|
super(values, key);
|
26
|
}
|
27
|
|
28
|
@Override
|
29
|
protected void checkSequence() throws NotValidResultSequenceException {
|
30
|
if(!it.hasNext()){
|
31
|
throw new NotValidResultSequenceException("Empty information for key");
|
32
|
}
|
33
|
|
34
|
pli = loadProjectList().iterator();
|
35
|
getNext();
|
36
|
}
|
37
|
|
38
|
private void getNext(){
|
39
|
if (pli.hasNext())
|
40
|
resultId = pli.next();
|
41
|
else
|
42
|
resultId = TERMINATOR;
|
43
|
}
|
44
|
|
45
|
private Set<String> loadProjectList() throws NotValidResultSequenceException {
|
46
|
final Set<String> fromResult = new HashSet<>();
|
47
|
final Set<String> fromSemRel = new HashSet<>();
|
48
|
while (it.hasNext()) {
|
49
|
final Value v = Value.fromJson(it.next().toString());
|
50
|
if(trust == null) {
|
51
|
trust = v.getTrust();
|
52
|
}
|
53
|
switch (v.getType()) {
|
54
|
case fromsemrel:
|
55
|
fromSemRel.addAll(Arrays.asList(StringUtils.split(v.getValue(), ",")));
|
56
|
break;
|
57
|
case fromresult:
|
58
|
if(StringUtils.isNotBlank(v.getValue()))
|
59
|
fromResult.addAll(Arrays.asList(StringUtils.split(v.getValue(), ",")));
|
60
|
break;
|
61
|
case valid:
|
62
|
case notvalid:
|
63
|
throw new NotValidResultSequenceException("invalid type of result: " + v.getType());
|
64
|
}
|
65
|
}
|
66
|
//takes the projects to be associated to the result that have not already been associated to it
|
67
|
|
68
|
return Sets.difference(fromSemRel, fromResult);
|
69
|
}
|
70
|
|
71
|
@Override
|
72
|
public List<OafProtos.Oaf> next() {
|
73
|
|
74
|
ArrayList<OafProtos.Oaf> ret = new ArrayList<OafProtos.Oaf> (Arrays.asList(
|
75
|
getOafRel(keyb, resultId, REL_RESULT_PROJECT),
|
76
|
getOafRel(resultId, keyb, REL_PROJECT_RESULT)
|
77
|
));
|
78
|
|
79
|
getNext();
|
80
|
|
81
|
return ret;
|
82
|
}
|
83
|
|
84
|
private OafProtos.Oaf getOafRel(String source, String target, String semantics){
|
85
|
final ResultProjectProtos.ResultProject.Builder rpb = ResultProjectProtos.ResultProject.newBuilder()
|
86
|
.setOutcome(
|
87
|
ResultProjectProtos.ResultProject.Outcome.newBuilder()
|
88
|
.setRelMetadata(
|
89
|
RelMetadataProtos.RelMetadata.newBuilder()
|
90
|
.setSemantics(
|
91
|
getQualifier(semantics, DNET_RELATION_SCHEMA_PROJECTS, semantics, DNET_RELATION_SCHEMA_PROJECTS)
|
92
|
)
|
93
|
)
|
94
|
);
|
95
|
|
96
|
final OafProtos.OafRel.Builder relation = OafProtos.OafRel.newBuilder()
|
97
|
.setChild(false)
|
98
|
.setSubRelType(SUBREL_TYPE_PROJECT)
|
99
|
.setRelType(REL_TYPE_PROJECT)
|
100
|
.setRelClass(semantics)
|
101
|
.setTarget(target)
|
102
|
.setSource(source)
|
103
|
.setResultProject(rpb);
|
104
|
|
105
|
return OafProtos.Oaf.newBuilder().setKind(KindProtos.Kind.relation)
|
106
|
|
107
|
.setRel(relation)
|
108
|
.setDataInfo(getDataInfo(trust, CLASS_PROJECT_ID,SCHEMA_ID,SCHEMA_NAME,DATA_INFO_TYPE,CLASS_PROJECT_NAME))
|
109
|
.build();
|
110
|
}
|
111
|
|
112
|
|
113
|
}
|