diff --git a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/CleanGraphSparkJob.java b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/CleanGraphSparkJob.java index 04a5ef38d..3067d8639 100644 --- a/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/CleanGraphSparkJob.java +++ b/dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/CleanGraphSparkJob.java @@ -103,15 +103,9 @@ public class CleanGraphSparkJob { return spark .read() .textFile(inputEntityPath) - .filter((FilterFunction) s -> isEntityType(s, clazz)) - .map((MapFunction) s -> StringUtils.substringAfter(s, "|"), Encoders.STRING()) .map( (MapFunction) value -> OBJECT_MAPPER.readValue(value, clazz), Encoders.bean(clazz)); } - private static boolean isEntityType(final String s, final Class clazz) { - return StringUtils.substringBefore(s, "|").equals(clazz.getName()); - } - }