Talend Component Kit Developer Reference Guide

Dynamic schema

Since the 1.1.25 release, the dynamic column feature is supported in Studio with component-runtime components.

Dynamic column is available with Enterprise versions of Talend Studio only.

Accessing columns metadata

In Studio, we can define for each component a schema with associated metadata.

To access those informations in your component, you’ve to do a few things:

Using the @Structure annotation

API: @org.talend.sdk.component.api.configuration.ui.widget.Structure

According the specified field type, you will acess to

the column names list with List<String>
a subset or all wanted metadata with List<MySchemaMeta> (see below)

@Data
@GridLayout({ @GridLayout.Row({ "dataset" }),
              @GridLayout.Row({ "incomingSchema" }) }) (5)
public class OutputConfig implements Serializable {

    @Option
    @Documentation("My dataset.")
    private Dataset dataset;

    @Option (1)
    @Documentation("Incoming metadata.")
    @Structure(type = Structure.Type.IN) (2) (3)
    private List<SchemaInfo> incomingSchema; (4)

1	`@Option`: mark class’s attributes as being a configuration entry.
2	`@Structure`: mark this configuration entry as a schema container.
3	`Structure.Type.IN`: marks the schema for an incoming flow (Output). Use `Structure.Type.OUT` for outgoing flow (Input).
4	`List<SchemaInfo>`: is a custom class for holding metadata.
5	`@GridLayout`: option should be defined in the UI layout.

Then, we should have a class SchemaInfo as following:

Defining a specific class for holding metadata If you don’t want just only column names (using List<String>), you’ll have to define a custom class.

@Data
@GridLayout({ @GridLayout.Row({ "label", "key", "talendType", "nullable", "pattern" }) })
@Documentation("Schema definition.")
public class SchemaInfo implements Serializable {

    @Option
    @Documentation("Column name.")
    private String label;

    @Option
    @Documentation("Is it a Key column.")
    private boolean key;

    @Option
    @Documentation("Talend type such as id_String.")
    private String talendType;

    @Option
    @Documentation("Is it a Nullable column.")
    private boolean nullable;

    @Option
    @Documentation("Pattern used for datetime processing.")
    private String pattern = "yyyy-MM-dd HH:mm";
}

Available Studio metadata informations

Field name	Type	Name in Studio
`label`	String	Column
`originalDbColumnName`	String	Db Column
`key`	Boolean	Key
`type`	String	DB Type
`talendType`	String	Type
`nullable`	Boolean	Nullable
`pattern`	String	Date Pattern
`length`	int	Length
`precision`	int	Precision
`defaultValue`	String	Default
`comment`	String	Comment

Notes when designing an output connector

Available since 1.43.x release

As Talend Component Kit Schema's types aren’t matching all Studio types, we wrap those types in wider types (like Character or char wrapped into String, Short to Integer, and so on…).

Anyway, the original type coming from Studio’s IPersistableRow is stored in record’s schema properties under the property name talend.studio.type.

Studio managed types are: id_BigDecimal, id_Boolean, id_Byte, id_byte[], id_Character, id_Date, id_Double, id_Document, id_Dynamic, id_Float, id_Integer, id_List, id_Long, id_Object, id_Short, id_String.

When handling an output connector designed for Studio, you should have to check for this property to get an accurate type for output.

For instance, java.math.BigDecimal is handled in framework as a Type.STRING, so when an output connector will receive a record, in studio context, you’ll need to check for the property and cast it correctly.

Here is a simple processor before writing to backend destination:

@ElementListener
public void process(final Record input) {
    final String value = input.getString("myBigDecimal");
    final Boolean isBigDec = "id_BigDecimal".equals(input.getSchema().getEntry("myBigDecimal").getProp("talend.studio.type"));
    queueWriter.write(isBigDec ? new BigDecimal(value) : value);
}

This usage of properties is cumbersome but may fix some potential issues for now. We plan to widen managed types in Record and Schema in a few iterations (No ETA defined yet).

Studio schema

Dynamic schema

Accessing columns metadata

Notes when designing an output connector

Related articles