org.apache.beam.sdk.schemas.transforms

Class Select

java.lang.Object

org.apache.beam.sdk.schemas.transforms.Select

@Experimental(value=SCHEMAS)
public class Select
extends java.lang.Object

A PTransform for selecting a subset of fields from a schema type.

This transforms allows projecting out a subset of fields from a schema type. The output of this transform is of type Row, though that can be converted into any other type with matching schema using the Convert transform.

For example, consider the following POJO type:

 @DefaultSchema(JavaFieldSchema.class)
 public class UserEvent {
   public String userId;
   public String eventId;
   public int eventType;
   public Location location;
 }

 @DefaultSchema(JavaFieldSchema.class)
 public class Location {
   public double latitude;
   public double longtitude;
 }

Say you want to select just the set of userId, eventId pairs from each element, you would write the following:

 PCollection<UserEvent> events = readUserEvents();
 PCollection<Row> rows = event.apply(Select.fieldNames("userId", "eventId"));
 

It's possible to select a nested field as well. For example, if you want just the location information from each element:

 PCollection<UserEvent> events = readUserEvents();
 PCollection<Location> rows = event.apply(Select.fieldNames("location")
                              .apply(Convert.to(Location.class));
 

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	Select.Fields<T>
static class	Select.Flattened<T> A PTransform representing a flattened schema.

Constructor Summary

Constructors

Constructor and Description
Select()

Method Summary

Modifier and Type	Method and Description
static <T> Select.Fields<T>	create()
static <T> Select.Fields<T>	fieldAccess(FieldAccessDescriptor fieldAccessDescriptor) Select a set of fields described in a FieldAccessDescriptor.
static <T> Select.Fields<T>	fieldIds(java.lang.Integer... ids) Select a set of top-level field ids from the row.
static <T> Select.Fields<T>	fieldNames(java.lang.String... names) Select a set of top-level field names from the row.
static <T> Select.Flattened<T>	flattenedSchema() Selects every leaf-level field.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

Select

public Select()

Method Detail

create

public static <T> Select.Fields<T> create()

fieldIds

public static <T> Select.Fields<T> fieldIds(java.lang.Integer... ids)

Select a set of top-level field ids from the row.

fieldNames

public static <T> Select.Fields<T> fieldNames(java.lang.String... names)

Select a set of top-level field names from the row.

fieldAccess

public static <T> Select.Fields<T> fieldAccess(FieldAccessDescriptor fieldAccessDescriptor)

Select a set of fields described in a FieldAccessDescriptor.

This allows for nested fields to be selected as well.

flattenedSchema

public static <T> Select.Flattened<T> flattenedSchema()

Selects every leaf-level field. This results in a a nested schema being flattened into a single top-level schema. By default nested field names will be concatenated with _ characters, though this can be overridden using Select.Flattened.keepMostNestedFieldName() and Select.Flattened.withFieldNameAs(java.lang.String, java.lang.String).