Package com.lucidworks.indexing.api

Interface IndexStage<T extends IndexStageConfig>

  • Type Parameters:
    T - index stage configuration class
    All Known Implementing Classes:
    IndexStageBase
    public interface IndexStage<T extends IndexStageConfig>
    Fusion index pipeline stage. Custom index stages must implement this interface to be discovered and called by Fusion. Example of custom index stage class: 
     @Stage(type = "myStage", configClass = MyStageConfig.class)
 public class MyStage implements IndexStage<MyStageConfig> {

   @Override
   public void init(MyStageConfig config, Fusion fusion) {
     // stage initialization logic
   }

   @Override
   public Document process(Document document, Context context) {
     // document processing logic
   }
    Implementations of this class must be stateless. Fusion can create and use multiple index stage instances at the same time.

    • Method Detail

      • init

        void init​(T config,
          Fusion fusion)
        Stage initialization callback. This method will be called by Fusion when index stage instance is created and before 'process' method is called. Stage configuration will be passed by Fusion as IndexStageConfig instance. Additionally Fusion interface instance will be passed to allow calling Fusion API from the index stage.
        Parameters:
        config - index pipeline stage configuration
        fusion - Fusion API instance

      • process

        default Document process​(Document document,
                         Context context)
        Process single document. This method is called for each document passing through index pipeline. Implement this method to perform processing of single Document instance that results in either 1 or 0 documents being emitted. Return null to drop document from the pipeline. This method is a convenience and will not be called if process(Document, Context, Consumer) implementation is overridden. Default implementation is NOOP if not implemented. Note that this method implementation must be thread-safe as it can be invoked concurrently by multiple threads.
        Parameters:
        document - document going through index pipeline
        context - index pipeline context
        Returns:
        processed document or null to drop document from the pipeline

      • process

        default void process​(Document document,
                     Context context,
                     java.util.function.Consumer<Document> output)
        Process single document. This method is called for each document passing through index pipeline. Implement this method to perform processing of single Document instance that results in arbitrary number of documents being emitted. Call output.accept(document) for each document you want to emit as a result of processing. Note that after sending a document instance to the output, its state may be changed by subsequent stages, therefore it is strongly advised to discard the document instance immediately after emitting it. Passing null to output consumer will cause IllegalArgumentException. Overriding the default implementation of this method will result in process(Document, Context) to not be called. Default implementation is to call process(Document, Context) method. Note that this method implementation must be thread-safe as it can be invoked concurrently by multiple threads.
        Parameters:
        document - document going through index pipeline
        context - index pipeline context
        output - consumer for documents emitted as the result of processing