|
1
|
package eu.dnetlib.oai.mongo;
|
|
2
|
|
|
3
|
import java.util.Iterator;
|
|
4
|
import java.util.function.Function;
|
|
5
|
|
|
6
|
import com.google.common.collect.Lists;
|
|
7
|
import com.mongodb.DBObject;
|
|
8
|
import com.mongodb.client.MongoCursor;
|
|
9
|
|
|
10
|
import eu.dnetlib.oai.Cursor;
|
|
11
|
import eu.dnetlib.oai.info.RecordInfo;
|
|
12
|
|
|
13
|
public class DNetOAIMongoCursor implements Cursor {
|
|
14
|
|
|
15
|
/**
|
|
16
|
* Underlying mongo cursor.
|
|
17
|
*/
|
|
18
|
private MongoCursor<DBObject> dbCursor;
|
|
19
|
private int size = 0;
|
|
20
|
/**
|
|
21
|
* Function to apply to records before delivering.
|
|
22
|
*/
|
|
23
|
private Function<String, String> function;
|
|
24
|
|
|
25
|
/**
|
|
26
|
* true if the RecordInfo returned by this Cursor must include the record body, false otherwise.
|
|
27
|
*/
|
|
28
|
private boolean bodyIncluded;
|
|
29
|
|
|
30
|
private RecordInfoGenerator recordInfoGenerator;
|
|
31
|
private MetadataExtractor metadataExtractor;
|
|
32
|
private ProvenanceExtractor provenanceExtractor;
|
|
33
|
|
|
34
|
public DNetOAIMongoCursor() {
|
|
35
|
super();
|
|
36
|
}
|
|
37
|
|
|
38
|
public DNetOAIMongoCursor(final MongoCursor<DBObject> dbCursor, final boolean bodyIncluded, final RecordInfoGenerator recordInfoGenerator,
|
|
39
|
final MetadataExtractor metadataExtractor) {
|
|
40
|
this(dbCursor, null, bodyIncluded, recordInfoGenerator, metadataExtractor);
|
|
41
|
}
|
|
42
|
|
|
43
|
public DNetOAIMongoCursor(final MongoCursor<DBObject> dbCursor, final Function<String, String> function, final boolean bodyIncluded,
|
|
44
|
final RecordInfoGenerator recordInfoGenerator, final MetadataExtractor metadataExtractor) {
|
|
45
|
super();
|
|
46
|
this.dbCursor = dbCursor;
|
|
47
|
this.size = 0;
|
|
48
|
this.function = function;
|
|
49
|
this.bodyIncluded = bodyIncluded;
|
|
50
|
this.recordInfoGenerator = recordInfoGenerator;
|
|
51
|
this.metadataExtractor = metadataExtractor;
|
|
52
|
}
|
|
53
|
|
|
54
|
/**
|
|
55
|
* {@inheritDoc}
|
|
56
|
*/
|
|
57
|
@Override
|
|
58
|
public int count() {
|
|
59
|
// I can do it because MongoCursor are always created from queries with "limit", so I do not expect the creation of the list to
|
|
60
|
// explode
|
|
61
|
// to not exagerate, I'll get the size only if the current size is 0
|
|
62
|
if (this.size == 0) {
|
|
63
|
this.size = Lists.newArrayList(this.dbCursor).size();
|
|
64
|
}
|
|
65
|
return this.size;
|
|
66
|
}
|
|
67
|
|
|
68
|
/**
|
|
69
|
* {@inheritDoc}
|
|
70
|
*
|
|
71
|
* @see Iterable#iterator()
|
|
72
|
*/
|
|
73
|
@Override
|
|
74
|
public Iterator<RecordInfo> iterator() {
|
|
75
|
|
|
76
|
return new Iterator<RecordInfo>() {
|
|
77
|
|
|
78
|
@Override
|
|
79
|
public boolean hasNext() {
|
|
80
|
return DNetOAIMongoCursor.this.dbCursor.hasNext();
|
|
81
|
}
|
|
82
|
|
|
83
|
@Override
|
|
84
|
public RecordInfo next() {
|
|
85
|
final DBObject res = DNetOAIMongoCursor.this.dbCursor.next();
|
|
86
|
final RecordInfo info = DNetOAIMongoCursor.this.recordInfoGenerator.transformDBObject(res, DNetOAIMongoCursor.this.bodyIncluded);
|
|
87
|
if ((DNetOAIMongoCursor.this.function != null) && DNetOAIMongoCursor.this.bodyIncluded && (info != null)) {
|
|
88
|
info.setMetadata(DNetOAIMongoCursor.this.function.apply(info.getMetadata()));
|
|
89
|
}
|
|
90
|
return info;
|
|
91
|
}
|
|
92
|
|
|
93
|
@Override
|
|
94
|
public void remove() {
|
|
95
|
throw new UnsupportedOperationException();
|
|
96
|
}
|
|
97
|
|
|
98
|
};
|
|
99
|
}
|
|
100
|
|
|
101
|
public Function<String, String> getFunction() {
|
|
102
|
return this.function;
|
|
103
|
}
|
|
104
|
|
|
105
|
public void setFunction(final Function<String, String> function) {
|
|
106
|
this.function = function;
|
|
107
|
}
|
|
108
|
|
|
109
|
public MongoCursor<DBObject> getDbCursor() {
|
|
110
|
return this.dbCursor;
|
|
111
|
}
|
|
112
|
|
|
113
|
public void setDbCursor(final MongoCursor<DBObject> dbCursor) {
|
|
114
|
this.dbCursor = dbCursor;
|
|
115
|
}
|
|
116
|
|
|
117
|
@Override
|
|
118
|
public boolean isBodyIncluded() {
|
|
119
|
return this.bodyIncluded;
|
|
120
|
}
|
|
121
|
|
|
122
|
@Override
|
|
123
|
public void setBodyIncluded(final boolean bodyIncluded) {
|
|
124
|
this.bodyIncluded = bodyIncluded;
|
|
125
|
}
|
|
126
|
|
|
127
|
public RecordInfoGenerator getRecordInfoGenerator() {
|
|
128
|
return this.recordInfoGenerator;
|
|
129
|
}
|
|
130
|
|
|
131
|
public void setRecordInfoGenerator(final RecordInfoGenerator recordInfoGenerator) {
|
|
132
|
this.recordInfoGenerator = recordInfoGenerator;
|
|
133
|
}
|
|
134
|
|
|
135
|
public MetadataExtractor getMetadataExtractor() {
|
|
136
|
return this.metadataExtractor;
|
|
137
|
}
|
|
138
|
|
|
139
|
public void setMetadataExtractor(final MetadataExtractor metadataExtractor) {
|
|
140
|
this.metadataExtractor = metadataExtractor;
|
|
141
|
}
|
|
142
|
|
|
143
|
public ProvenanceExtractor getProvenanceExtractor() {
|
|
144
|
return this.provenanceExtractor;
|
|
145
|
}
|
|
146
|
|
|
147
|
public void setProvenanceExtractor(final ProvenanceExtractor provenanceExtractor) {
|
|
148
|
this.provenanceExtractor = provenanceExtractor;
|
|
149
|
}
|
|
150
|
|
|
151
|
}
|