forked from D-Net/dnet-hadoop
[orcid propagation] skip empty directories
This commit is contained in:
parent
9cf0a98699
commit
6aa91204a5
|
@ -228,10 +228,15 @@ public class PropagationConstant {
|
||||||
|
|
||||||
public static <R> Dataset<R> readPath(
|
public static <R> Dataset<R> readPath(
|
||||||
SparkSession spark, String inputPath, Class<R> clazz) {
|
SparkSession spark, String inputPath, Class<R> clazz) {
|
||||||
return spark
|
|
||||||
.read()
|
if (HdfsSupport.exists(inputPath, spark.sparkContext().hadoopConfiguration())) {
|
||||||
.textFile(inputPath)
|
return spark
|
||||||
.map((MapFunction<String, R>) value -> OBJECT_MAPPER.readValue(value, clazz), Encoders.bean(clazz));
|
.read()
|
||||||
|
.textFile(inputPath)
|
||||||
|
.map((MapFunction<String, R>) value -> OBJECT_MAPPER.readValue(value, clazz), Encoders.bean(clazz));
|
||||||
|
} else {
|
||||||
|
return spark.emptyDataset(Encoders.bean(clazz));
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
}
|
}
|
||||||
|
|
Loading…
Reference in New Issue