From ed1bf13977eabbf4803c7e5017f62a2c56cc55df Mon Sep 17 00:00:00 2001 From: Claus Stadler Date: Fri, 8 Oct 2021 21:37:47 +0200 Subject: [PATCH] Moved java operators for jena to its own lightweight package --- pom.xml | 7 ++ sansa-rdf/sansa-rdf-spark/pom.xml | 5 + sansa-spark-jena-java/pom.xml | 98 +++++++++++++++++++ .../rdf/spark/io/OutputFormatRdfRegistry.java | 0 .../sansa_stack/rdf/spark/io/RddRdfSaver.java | 0 .../rdf/spark/io/input/api/RddRdfLoader.java | 0 .../io/input/api/RddRdfLoaderRegistry.java | 0 .../rdf/spark/io/input/api/RdfSource.java | 0 .../spark/io/input/api/RdfSourceFactory.java | 0 .../input/impl/RddRdfLoaderRegistryImpl.java | 0 .../io/input/impl/RdfSourceFactoryImpl.java | 0 .../spark/io/input/impl/RdfSourceImpl.java | 0 .../spark/rdd/op/JavaRddOfDatasetsOps.java | 0 .../spark/rdd/op/JavaRddOfNamedModelsOps.java | 3 - .../rdf/spark/rdd/op/JavaRddOfQuadsOps.java | 0 .../spark/rdd/op/JavaRddOfResourcesOps.java | 0 .../rdf/spark/rdd/op/JavaRddOfTriplesOps.java | 0 .../rdf/spark/rdd/op/JavaRddOps.java | 2 - .../rdf/spark/util/JavaSparkContextUtils.java | 0 .../rdd/function/JavaPairRddFunction.java | 2 +- .../spark/rdd/function/JavaRddFunction.java | 2 +- .../rdd/function/ToJavaPairRddFunction.java | 2 +- .../spark/rdd/function/ToJavaRddFunction.java | 2 +- 23 files changed, 114 insertions(+), 9 deletions(-) create mode 100644 sansa-spark-jena-java/pom.xml rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/io/OutputFormatRdfRegistry.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/io/RddRdfSaver.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RddRdfLoader.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RddRdfLoaderRegistry.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSource.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSourceFactory.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RddRdfLoaderRegistryImpl.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceFactoryImpl.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceImpl.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfDatasetsOps.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfNamedModelsOps.java (96%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfQuadsOps.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfResourcesOps.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfTriplesOps.java (100%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOps.java (95%) rename {sansa-rdf/sansa-rdf-spark => sansa-spark-jena-java}/src/main/java/net/sansa_stack/rdf/spark/util/JavaSparkContextUtils.java (100%) rename {sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf => sansa-spark-jena-java/src/main/java/net/sansa_stack}/spark/rdd/function/JavaPairRddFunction.java (95%) rename {sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf => sansa-spark-jena-java/src/main/java/net/sansa_stack}/spark/rdd/function/JavaRddFunction.java (96%) rename {sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf => sansa-spark-jena-java/src/main/java/net/sansa_stack}/spark/rdd/function/ToJavaPairRddFunction.java (94%) rename {sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf => sansa-spark-jena-java/src/main/java/net/sansa_stack}/spark/rdd/function/ToJavaRddFunction.java (94%) diff --git a/pom.xml b/pom.xml index eb99cc141..975f361d2 100644 --- a/pom.xml +++ b/pom.xml @@ -116,6 +116,7 @@ sansa-hadoop-jena sansa-kryo-jena + sansa-spark-jena-java sansa-test-resources sansa-resource-metadata @@ -172,6 +173,12 @@ 4.0.2 + + net.sansa-stack + sansa-spark-jena-java_${scala.binary.version} + ${project.version} + + net.sansa-stack sansa-test-resources_${scala.binary.version} diff --git a/sansa-rdf/sansa-rdf-spark/pom.xml b/sansa-rdf/sansa-rdf-spark/pom.xml index daf85b1b1..2201deb58 100644 --- a/sansa-rdf/sansa-rdf-spark/pom.xml +++ b/sansa-rdf/sansa-rdf-spark/pom.xml @@ -11,6 +11,11 @@ + + ${project.groupId} + sansa-spark-jena-java_${scala.binary.version} + + ${project.groupId} diff --git a/sansa-spark-jena-java/pom.xml b/sansa-spark-jena-java/pom.xml new file mode 100644 index 000000000..184075c73 --- /dev/null +++ b/sansa-spark-jena-java/pom.xml @@ -0,0 +1,98 @@ + + + 4.0.0 + + + net.sansa-stack + sansa-parent_2.12 + 0.8.0-RC2-SNAPSHOT + + + sansa-spark-jena-java_2.12 + jar + + + + + org.scalastyle + scalastyle-maven-plugin + + + net.alchim31.maven + scala-maven-plugin + + + + + + + + + + net.sansa-stack + sansa-hadoop-jena + + + + org.apache.spark + spark-core_${scala.binary.version} + ${spark.version} + + + + + org.apache.spark + spark-sql_${scala.binary.version} + + + + junit + junit + test + + + + net.sansa-stack + sansa-test-resources_2.12 + test + + + + + + + diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/OutputFormatRdfRegistry.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/OutputFormatRdfRegistry.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/OutputFormatRdfRegistry.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/OutputFormatRdfRegistry.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/RddRdfSaver.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/RddRdfSaver.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/RddRdfSaver.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/RddRdfSaver.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RddRdfLoader.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RddRdfLoader.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RddRdfLoader.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RddRdfLoader.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RddRdfLoaderRegistry.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RddRdfLoaderRegistry.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RddRdfLoaderRegistry.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RddRdfLoaderRegistry.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSource.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSource.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSource.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSource.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSourceFactory.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSourceFactory.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSourceFactory.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/api/RdfSourceFactory.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RddRdfLoaderRegistryImpl.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RddRdfLoaderRegistryImpl.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RddRdfLoaderRegistryImpl.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RddRdfLoaderRegistryImpl.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceFactoryImpl.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceFactoryImpl.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceFactoryImpl.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceFactoryImpl.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceImpl.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceImpl.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceImpl.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/io/input/impl/RdfSourceImpl.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfDatasetsOps.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfDatasetsOps.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfDatasetsOps.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfDatasetsOps.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfNamedModelsOps.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfNamedModelsOps.java similarity index 96% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfNamedModelsOps.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfNamedModelsOps.java index cf2c73be0..2d743d67a 100644 --- a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfNamedModelsOps.java +++ b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfNamedModelsOps.java @@ -1,7 +1,6 @@ package net.sansa_stack.rdf.spark.rdd.op; import org.aksw.jena_sparql_api.rx.DatasetFactoryEx; -import org.aksw.jena_sparql_api.rx.ModelFactoryEx; import org.aksw.jena_sparql_api.utils.model.ResourceInDataset; import org.aksw.jena_sparql_api.utils.model.ResourceInDatasetImpl; import org.apache.jena.graph.Node; @@ -9,9 +8,7 @@ import org.apache.jena.query.Dataset; import org.apache.jena.query.DatasetFactory; import org.apache.jena.rdf.model.Model; -import org.apache.jena.rdf.model.ModelFactory; import org.apache.jena.rdf.model.Resource; -import org.apache.spark.HashPartitioner; import org.apache.spark.api.java.JavaPairRDD; import org.apache.spark.api.java.JavaRDD; diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfQuadsOps.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfQuadsOps.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfQuadsOps.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfQuadsOps.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfResourcesOps.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfResourcesOps.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfResourcesOps.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfResourcesOps.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfTriplesOps.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfTriplesOps.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfTriplesOps.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOfTriplesOps.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOps.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOps.java similarity index 95% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOps.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOps.java index c7d9517b4..83eac9858 100644 --- a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOps.java +++ b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/rdd/op/JavaRddOps.java @@ -1,6 +1,5 @@ package net.sansa_stack.rdf.spark.rdd.op; -import org.aksw.commons.lambda.serializable.SerializableBinaryOperator; import org.apache.spark.HashPartitioner; import org.apache.spark.api.java.JavaPairRDD; import org.apache.spark.api.java.JavaRDD; @@ -8,7 +7,6 @@ import java.util.Collections; import java.util.function.BiConsumer; -import java.util.function.BinaryOperator; import java.util.stream.Collector; public class JavaRddOps { diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/util/JavaSparkContextUtils.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/util/JavaSparkContextUtils.java similarity index 100% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/util/JavaSparkContextUtils.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/rdf/spark/util/JavaSparkContextUtils.java diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/JavaPairRddFunction.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/JavaPairRddFunction.java similarity index 95% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/JavaPairRddFunction.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/JavaPairRddFunction.java index bac2b82f6..32300f6bb 100644 --- a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/JavaPairRddFunction.java +++ b/sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/JavaPairRddFunction.java @@ -1,4 +1,4 @@ -package net.sansa_stack.rdf.spark.rdd.function; +package net.sansa_stack.spark.rdd.function; import org.aksw.commons.lambda.serializable.SerializableFunction; import org.apache.spark.api.java.JavaPairRDD; diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/JavaRddFunction.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/JavaRddFunction.java similarity index 96% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/JavaRddFunction.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/JavaRddFunction.java index 4c7bb9a8e..76105a58d 100644 --- a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/JavaRddFunction.java +++ b/sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/JavaRddFunction.java @@ -1,4 +1,4 @@ -package net.sansa_stack.rdf.spark.rdd.function; +package net.sansa_stack.spark.rdd.function; import org.aksw.commons.lambda.serializable.SerializableFunction; import org.apache.spark.api.java.JavaRDD; diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/ToJavaPairRddFunction.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/ToJavaPairRddFunction.java similarity index 94% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/ToJavaPairRddFunction.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/ToJavaPairRddFunction.java index 871dd898f..1f9fea6a1 100644 --- a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/ToJavaPairRddFunction.java +++ b/sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/ToJavaPairRddFunction.java @@ -1,4 +1,4 @@ -package net.sansa_stack.rdf.spark.rdd.function; +package net.sansa_stack.spark.rdd.function; import org.aksw.commons.lambda.serializable.SerializableFunction; import org.apache.spark.api.java.JavaPairRDD; diff --git a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/ToJavaRddFunction.java b/sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/ToJavaRddFunction.java similarity index 94% rename from sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/ToJavaRddFunction.java rename to sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/ToJavaRddFunction.java index b9819f16e..a4880d5e3 100644 --- a/sansa-rdf/sansa-rdf-spark/src/main/java/net/sansa_stack/rdf/spark/rdd/function/ToJavaRddFunction.java +++ b/sansa-spark-jena-java/src/main/java/net/sansa_stack/spark/rdd/function/ToJavaRddFunction.java @@ -1,4 +1,4 @@ -package net.sansa_stack.rdf.spark.rdd.function; +package net.sansa_stack.spark.rdd.function; import org.aksw.commons.lambda.serializable.SerializableFunction; import org.apache.spark.api.java.JavaPairRDD;