I am trying to traverse a Dataset to do some string similarity calculations like Jaro winkler or Cosine Similarity. I convert my Dataset to list of rows and then traverse with for statement which is not efficient spark way to do it. So I am looking forward for a better approach in Spark.
public class sample {
public static void main(String[] args) {
JavaSparkContext sc = new JavaSparkContext(new SparkConf().setAppName("Example").setMaster("local[*]"));
SQLContext sqlContext = new SQLContext(sc);
SparkSession spark = SparkSession.builder().appName("JavaTokenizerExample").getOrCreate();
List<Row> data = Arrays.asList(RowFactory.create("Mysore","Mysuru"),
RowFactory.create("Name","FirstName"));
StructType schema = new StructType(
new StructField[] { new StructField("Word1", DataTypes.StringType, true, Metadata.empty()),
new StructField("Word2", DataTypes.StringType, true, Metadata.empty()) });
Dataset<Row> oldDF = spark.createDataFrame(data, schema);
oldDF.show();
List<Row> rowslist = oldDF.collectAsList();
}
}
I have found many JavaRDD examples which I am not clear. An Example for Dataset will help me a lot.
You can use org.apache.spark.api.java.function.ForeachFunction like below.
oldDF.foreach((ForeachFunction<Row>) row -> System.out.println(row));
For old java jdks that don't support lambda expressions, you can use the following after importing:
import org.apache.spark.api.java.function.VoidFunction;
yourDataSet.toJavaRDD().foreach(new VoidFunction<Row>() {
public void call(Row r) throws Exception {
System.out.println(r.getAs("your column name here"));
}
});