1
|
package eu.dnetlib.data.mapreduce.util;
|
2
|
|
3
|
import java.util.List;
|
4
|
import java.util.Set;
|
5
|
|
6
|
import com.google.common.base.Splitter;
|
7
|
import com.google.common.collect.Iterables;
|
8
|
import com.google.common.collect.Lists;
|
9
|
import com.google.common.collect.Sets;
|
10
|
import eu.dnetlib.data.proto.FieldTypeProtos.*;
|
11
|
import org.apache.hadoop.hbase.io.ImmutableBytesWritable;
|
12
|
|
13
|
import com.google.common.base.Function;
|
14
|
import com.google.protobuf.InvalidProtocolBufferException;
|
15
|
|
16
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
17
|
import eu.dnetlib.data.transform.OafUtils;
|
18
|
import org.apache.hadoop.mapreduce.Mapper;
|
19
|
import org.apache.hadoop.mapreduce.Reducer;
|
20
|
|
21
|
public class OafHbaseUtils extends OafUtils {
|
22
|
|
23
|
public static OafDecoder decode(final ImmutableBytesWritable oaf) {
|
24
|
return new OafDecoder(oaf.copyBytes());
|
25
|
}
|
26
|
|
27
|
public static Function<ImmutableBytesWritable, OafDecoder> decoder() {
|
28
|
return input -> OafDecoder.decode(input.copyBytes());
|
29
|
}
|
30
|
|
31
|
public static Iterable<Oaf> asOaf(final Iterable<ImmutableBytesWritable> in) {
|
32
|
return Iterables.transform(in, oafDecoder());
|
33
|
}
|
34
|
|
35
|
public static Function<ImmutableBytesWritable, Oaf> oafDecoder() {
|
36
|
return input -> parse(input);
|
37
|
}
|
38
|
|
39
|
public static Oaf parse(final ImmutableBytesWritable input) {
|
40
|
try {
|
41
|
return Oaf.parseFrom(input.copyBytes());
|
42
|
} catch (final InvalidProtocolBufferException e) {
|
43
|
throw new IllegalArgumentException(e);
|
44
|
}
|
45
|
}
|
46
|
|
47
|
public static <T> String getValue(T t) {
|
48
|
return mapValue(t);
|
49
|
}
|
50
|
|
51
|
public static <T> String getKey(T t) {
|
52
|
return mapKey(t);
|
53
|
}
|
54
|
|
55
|
public static <T> String getValue(Iterable<T> ts) {
|
56
|
return Iterables.getFirst(listValues(ts), "");
|
57
|
}
|
58
|
|
59
|
public static <T> Set<String> hashSetValues(Iterable<T> ts) {
|
60
|
return Sets.newHashSet(Iterables.transform(ts, t -> mapValue(t)));
|
61
|
}
|
62
|
|
63
|
public static <T> List<String> listValues(Iterable<T> ts) {
|
64
|
return Lists.newArrayList(Iterables.transform(ts, t -> mapValue(t)));
|
65
|
}
|
66
|
|
67
|
public static <T> String getKey(Iterable<T> ts) {
|
68
|
return Iterables.getFirst(listKeys(ts), "");
|
69
|
}
|
70
|
|
71
|
public static <T> List<String> listKeys(Iterable<T> ts) {
|
72
|
return Lists.newArrayList(Iterables.transform(ts, t -> mapKey(t)));
|
73
|
}
|
74
|
|
75
|
public static <T> Set<String> hashSetKeys(Iterable<T> ts) {
|
76
|
return Sets.newHashSet(Iterables.transform(ts, t -> mapKey(t)));
|
77
|
}
|
78
|
|
79
|
private static <T> String mapKey(final T t) {
|
80
|
if (t instanceof KeyValue) return ((KeyValue) t).getKey();
|
81
|
if (t instanceof String) return (String) t;
|
82
|
if (t instanceof Qualifier) return ((Qualifier) t).getClassid();
|
83
|
|
84
|
throw new IllegalArgumentException(String.format("type %s not mapped", t.getClass()));
|
85
|
}
|
86
|
|
87
|
private static <T> String mapValue(final T t) {
|
88
|
if (t instanceof StructuredProperty) return ((StructuredProperty) t).getValue();
|
89
|
if (t instanceof KeyValue) return ((KeyValue) t).getValue();
|
90
|
if (t instanceof String) return (String) t;
|
91
|
if (t instanceof StringField) return ((StringField) t).getValue();
|
92
|
if (t instanceof Qualifier) return ((Qualifier) t).getClassname();
|
93
|
if (t instanceof Author) return ((Author) t).getFullname();
|
94
|
|
95
|
throw new IllegalArgumentException(String.format("type %s not mapped", t.getClass()));
|
96
|
}
|
97
|
|
98
|
public static List<String> getPropertyValues(final Reducer.Context context, final String name) {
|
99
|
return doGetPropertyValues(context.getConfiguration().get(name, ""));
|
100
|
}
|
101
|
|
102
|
public static List<String> getPropertyValues(final Mapper.Context context, final String name) {
|
103
|
return doGetPropertyValues(context.getConfiguration().get(name, ""));
|
104
|
}
|
105
|
|
106
|
private static List<String> doGetPropertyValues(final String s) {
|
107
|
return Lists.newArrayList(Splitter.on(",").omitEmptyStrings().split(s));
|
108
|
}
|
109
|
|
110
|
}
|