660 : Support for onRetrieved for TokenStream (langchain4j#1527)

docs/docs/tutorials/7-rag.md

@@ -206,6 +206,20 @@ String answer = result.content();
 List<Content> sources = result.sources();
 ```
 
+When streaming, a `Consumer<List<Content>>` can be specified using the `onRetrieved()` method:
+```java
+interface Assistant {
+
+    TokenStream chat(String userMessage);
+}
+
+assistant.chat("How to do Easy RAG with LangChain4j?")
+    .onRetrieved(sources -> ...)
+    .onNext(token -> ...)
+    .onError(error -> ...)
+    .start();
+```
+
 ## RAG APIs
 LangChain4j offers a rich set of APIs to make it easy for you to build custom RAG pipelines,
 ranging from simple ones to advanced ones.

langchain4j/src/main/java/dev/langchain4j/service/TokenStream.java

@@ -2,6 +2,7 @@
 
 import dev.langchain4j.data.message.AiMessage;
 import dev.langchain4j.model.output.Response;
+import dev.langchain4j.rag.RetrievalAugmentor;
 import dev.langchain4j.rag.content.Content;
 
 import java.util.List;
@@ -16,10 +17,10 @@ public interface TokenStream {
 
     /**
      * The provided consumer will be invoked when/if contents have been retrieved using {@link RetrievalAugmentor}.
-     * 
-     * This method is invoked before any call is made to the language model.
+     * <p>
+     * The invocation happens before any call is made to the language model.
      *
-     * @param contentHandler lambda that consumes all matching contents
+     * @param contentHandler lambda that consumes all retrieved contents
      * @return token stream instance used to configure or start stream processing
      */
     TokenStream onRetrieved(Consumer<List<Content>> contentHandler);
@@ -58,7 +59,7 @@ public interface TokenStream {
 
     /**
      * Completes the current token stream building and starts processing.
-     * 
+     * <p>
      * Will send a request to LLM and start response streaming.
      */
     void start();