fix issue with flatMap - the return type must not be null

This commit is contained in:
Miriam Baglioni 2020-05-28 23:50:32 +02:00
parent 773735f870
commit 01f7876595
2 changed files with 30 additions and 47 deletions

View File

@ -75,17 +75,27 @@ public class PrepareProjects {
HdfsSupport.remove(path, spark.sparkContext().hadoopConfiguration());
}
private static void exec(SparkSession spark, String progjectPath, String dbProjectPath, String outputPath) {
Dataset<CSVProject> project = readPath(spark, progjectPath, CSVProject.class);
private static void exec(SparkSession spark, String projectPath, String dbProjectPath, String outputPath) {
Dataset<CSVProject> project = readPath(spark, projectPath, CSVProject.class);
Dataset<ProjectSubset> dbProjects = readPath(spark, dbProjectPath, ProjectSubset.class);
dbProjects.joinWith(project, dbProjects.col("code").equalTo(project.col("id")), "left")
.flatMap((FlatMapFunction<Tuple2<ProjectSubset, CSVProject>, CSVProject>) value -> {
Optional<CSVProject> csvProject = Optional.ofNullable(value._2());
if(! csvProject.isPresent()){
return null;
dbProjects
.joinWith(project, dbProjects.col("code").equalTo(project.col("id")), "left")
.flatMap(getTuple2CSVProjectFlatMapFunction(), Encoders.bean(CSVProject.class))
.filter(Objects::nonNull)
.write()
.mode(SaveMode.Overwrite)
.option("compression", "gzip")
.json(outputPath);
}
private static FlatMapFunction<Tuple2<ProjectSubset, CSVProject>, CSVProject> getTuple2CSVProjectFlatMapFunction() {
return (FlatMapFunction<Tuple2<ProjectSubset, CSVProject>, CSVProject>) value -> {
Optional<CSVProject> csvProject = Optional.ofNullable(value._2());
List<CSVProject> csvProjectList = new ArrayList<>();
if (csvProject.isPresent()) {
String[] programme = csvProject.get().getProgramme().split(";");
Arrays
.stream(programme)
@ -95,36 +105,9 @@ public class PrepareProjects {
proj.setId(csvProject.get().getId());
csvProjectList.add(proj);
});
}
return csvProjectList.iterator();
}, Encoders.bean(CSVProject.class))
.write()
.mode(SaveMode.Overwrite)
.option("compression", "gzip")
.json(outputPath);
//
// .map(value -> {
// Optional<CSVProject> csvProject = Optional.ofNullable(value._2());
// }, Encoders.bean(CSVProject.class))
// .filter(Objects::nonNull)
// .toJavaRDD()
// .flatMap(p -> {
// List<CSVProject> csvProjectList = new ArrayList<>();
// String[] programme = p.getProgramme().split(";");
// Arrays
// .stream(programme)
// .forEach(value -> {
// CSVProject csvProject = new CSVProject();
// csvProject.setProgramme(value);
// csvProject.setId(p.getId());
// csvProjectList.add(csvProject);
// });
//
// return csvProjectList.iterator();
// })
// .map(p -> OBJECT_MAPPER.writeValueAsString(p))
// .saveAsTextFile(outputPath);
};
}
public static <R> Dataset<R> readPath(