Add suggestions

Azure-Samples · Jan 21, 2025 · 8c46067 · 8c46067
1 parent 6d19745
commit 8c46067
Show file tree

Hide file tree

Showing 7 changed files with 64 additions and 78 deletions.
diff --git a/app/backend/pom.xml b/app/backend/pom.xml
@@ -18,7 +18,7 @@
 
 		<spring-cloud-azure.version>5.14.0</spring-cloud-azure.version>
 		<azure-search.version>11.7.2</azure-search.version>
-		<semantic-kernel.version>1.4.1</semantic-kernel.version>
+		<semantic-kernel.version>1.4.2</semantic-kernel.version>
 		<mockito-inline.version>4.5.1</mockito-inline.version>
 		<maven.compiler-plugin.version>3.11.0</maven.compiler-plugin.version>
 

diff --git a/.../samples/rag/ask/approaches/semantickernel/JavaSemanticKernelWithVectorStoreApproach.java b/.../samples/rag/ask/approaches/semantickernel/JavaSemanticKernelWithVectorStoreApproach.java
@@ -2,16 +2,13 @@
 package com.microsoft.openai.samples.rag.ask.approaches.semantickernel;
 
 import com.azure.ai.openai.OpenAIAsyncClient;
-import com.azure.core.credential.TokenCredential;
-import com.azure.search.documents.SearchAsyncClient;
 import com.azure.search.documents.indexes.SearchIndexAsyncClient;
-import com.azure.search.documents.indexes.SearchIndexClientBuilder;
 import com.microsoft.openai.samples.rag.approaches.ContentSource;
 import com.microsoft.openai.samples.rag.approaches.RAGApproach;
 import com.microsoft.openai.samples.rag.approaches.RAGOptions;
 import com.microsoft.openai.samples.rag.approaches.RAGResponse;
 import com.microsoft.openai.samples.rag.chat.approaches.semantickernel.JavaSemanticKernelWithVectorStoreChatApproach;
-import com.microsoft.openai.samples.rag.retrieval.semantickernel.AzureAISearchVectorStoreApproach;
+import com.microsoft.openai.samples.rag.retrieval.semantickernel.AzureAISearchVectorStoreUtils;
 import com.microsoft.semantickernel.Kernel;
 
 import com.microsoft.semantickernel.aiservices.openai.chatcompletion.OpenAIChatCompletion;
@@ -38,7 +35,7 @@
 import org.springframework.beans.factory.annotation.Value;
 import org.springframework.stereotype.Component;
 
-import com.microsoft.openai.samples.rag.retrieval.semantickernel.AzureAISearchVectorStoreApproach.MemoryRecord;
+import com.microsoft.openai.samples.rag.retrieval.semantickernel.AzureAISearchVectorStoreUtils.DocumentRecord;
 
 /**
  * Use Java Semantic Kernel framework with built-in VectorStores for embeddings similarity search. A
@@ -51,15 +48,9 @@
 public class JavaSemanticKernelWithVectorStoreApproach implements RAGApproach<String, RAGResponse> {
     private static final Logger LOGGER =
             LoggerFactory.getLogger(JavaSemanticKernelWithVectorStoreApproach.class);
-    private final TokenCredential tokenCredential;
     private final OpenAIAsyncClient openAIAsyncClient;
+    private final SearchIndexAsyncClient searchAsyncClient;
 
-    private final SearchAsyncClient searchAsyncClient;
-
-    private final String EMBEDDING_FIELD_NAME = "embedding";
-
-    @Value("${cognitive.search.service}")
-    String searchServiceName;
 
     @Value("${cognitive.search.index}")
     String indexName;
@@ -71,10 +62,8 @@ public class JavaSemanticKernelWithVectorStoreApproach implements RAGApproach<St
     private String embeddingDeploymentModelId;
 
     public JavaSemanticKernelWithVectorStoreApproach(
-            TokenCredential tokenCredential,
             OpenAIAsyncClient openAIAsyncClient,
-            SearchAsyncClient searchAsyncClient) {
-        this.tokenCredential = tokenCredential;
+            SearchIndexAsyncClient searchAsyncClient) {
         this.openAIAsyncClient = openAIAsyncClient;
         this.searchAsyncClient = searchAsyncClient;
     }
@@ -90,29 +79,23 @@ public RAGResponse run(String question, RAGOptions options) {
         // skill is imported from src/main/resources/semantickernel/Plugins.
         Kernel semanticKernel = buildSemanticKernel(options);
 
-        // STEP 1: Create Azure AI Search client
-        SearchIndexAsyncClient client = new SearchIndexClientBuilder()
-                .endpoint("https://%s.search.windows.net".formatted(searchServiceName))
-                .credential(tokenCredential)
-                .buildAsyncClient();
-
-        // STEP 2: Build Vector Record Collection
-        AzureAISearchVectorStoreRecordCollection<MemoryRecord> recordCollection = new AzureAISearchVectorStoreRecordCollection<>(
-                client,
+        // STEP 1: Build Vector Record Collection
+        AzureAISearchVectorStoreRecordCollection<DocumentRecord> recordCollection = new AzureAISearchVectorStoreRecordCollection<>(
+                searchAsyncClient,
                 indexName,
-                AzureAISearchVectorStoreRecordCollectionOptions.<MemoryRecord>builder()
-                        .withRecordClass(MemoryRecord.class)
+                AzureAISearchVectorStoreRecordCollectionOptions.<DocumentRecord>builder()
+                        .withRecordClass(DocumentRecord.class)
                         .build()
         );
 
-        // STEP 3: Retrieve relevant documents using user question.
-        List<MemoryRecord> memoryResult = AzureAISearchVectorStoreApproach.searchAsync(
+        // STEP 2: Retrieve relevant documents using user question.
+        List<DocumentRecord> memoryResult = AzureAISearchVectorStoreUtils.searchAsync(
                 question, semanticKernel, recordCollection, options);
 
-        String sources = AzureAISearchVectorStoreApproach.buildSourcesText(memoryResult);
-        List<ContentSource> sourcesList = AzureAISearchVectorStoreApproach.buildSources(memoryResult);
+        String sources = AzureAISearchVectorStoreUtils.buildSourcesText(memoryResult);
+        List<ContentSource> sourcesList = AzureAISearchVectorStoreUtils.buildSources(memoryResult);
 
-        // STEP 4: Generate a contextual and content specific answer using the search results and question
+        // STEP 3: Generate a contextual and content specific answer using the search results and question
         KernelFunction<String> answerQuestion = semanticKernel.getFunction("RAG", "AnswerQuestion");
         KernelFunctionArguments arguments = KernelFunctionArguments.builder()
                 .withVariable("sources", sourcesList)

diff --git a/...les/rag/chat/approaches/semantickernel/JavaSemanticKernelWithVectorStoreChatApproach.java b/...les/rag/chat/approaches/semantickernel/JavaSemanticKernelWithVectorStoreChatApproach.java
@@ -1,16 +1,14 @@
 package com.microsoft.openai.samples.rag.chat.approaches.semantickernel;
 
 import com.azure.ai.openai.OpenAIAsyncClient;
-import com.azure.core.credential.TokenCredential;
 import com.azure.search.documents.indexes.SearchIndexAsyncClient;
-import com.azure.search.documents.indexes.SearchIndexClientBuilder;
 import com.microsoft.openai.samples.rag.approaches.ContentSource;
 import com.microsoft.openai.samples.rag.approaches.RAGApproach;
 import com.microsoft.openai.samples.rag.approaches.RAGOptions;
 import com.microsoft.openai.samples.rag.approaches.RAGResponse;
 import com.microsoft.openai.samples.rag.common.ChatGPTConversation;
 import com.microsoft.openai.samples.rag.common.ChatGPTUtils;
-import com.microsoft.openai.samples.rag.retrieval.semantickernel.AzureAISearchVectorStoreApproach;
+import com.microsoft.openai.samples.rag.retrieval.semantickernel.AzureAISearchVectorStoreUtils;
 import com.microsoft.semantickernel.Kernel;
 import com.microsoft.semantickernel.aiservices.openai.chatcompletion.OpenAIChatCompletion;
 import com.microsoft.semantickernel.aiservices.openai.textembedding.OpenAITextEmbeddingGenerationService;
@@ -38,7 +36,7 @@
 import java.util.ArrayList;
 import java.util.List;
 
-import com.microsoft.openai.samples.rag.retrieval.semantickernel.AzureAISearchVectorStoreApproach.MemoryRecord;
+import com.microsoft.openai.samples.rag.retrieval.semantickernel.AzureAISearchVectorStoreUtils.DocumentRecord;
 
 /**
  * Use Java Semantic Kernel framework with built-in VectorStores for embeddings similarity search. A
@@ -50,22 +48,20 @@
 @Component
 public class JavaSemanticKernelWithVectorStoreChatApproach implements RAGApproach<ChatGPTConversation, RAGResponse> {
     private static final Logger LOGGER = LoggerFactory.getLogger(JavaSemanticKernelWithVectorStoreChatApproach.class);
-    private final TokenCredential tokenCredential;
     private final OpenAIAsyncClient openAIAsyncClient;
+    private final SearchIndexAsyncClient searchAsyncClient;
     private String renderedConversation;
 
-    @Value("${cognitive.search.service}")
-    String searchServiceName;
     @Value("${cognitive.search.index}")
     String indexName;
     @Value("${openai.chatgpt.deployment}")
     private String gptChatDeploymentModelId;
     @Value("${openai.embedding.deployment}")
     private String embeddingDeploymentModelId;
 
-    public JavaSemanticKernelWithVectorStoreChatApproach(TokenCredential tokenCredential, OpenAIAsyncClient openAIAsyncClient) {
-        this.tokenCredential = tokenCredential;
+    public JavaSemanticKernelWithVectorStoreChatApproach(OpenAIAsyncClient openAIAsyncClient, SearchIndexAsyncClient searchAsyncClient) {
         this.openAIAsyncClient = openAIAsyncClient;
+        this.searchAsyncClient = searchAsyncClient;
     }
 
     @Override
@@ -76,29 +72,23 @@ public RAGResponse run(ChatGPTConversation questionOrConversation, RAGOptions op
         // Build semantic kernel context with AnswerConversation and ExtractKeywords plugins, EmbeddingGenerationService and ChatCompletionService.
         Kernel semanticKernel = buildSemanticKernel();
 
-        // STEP 1: Create Azure AI Search client
-        SearchIndexAsyncClient client = new SearchIndexClientBuilder()
-                .endpoint("https://%s.search.windows.net".formatted(searchServiceName))
-                .credential(tokenCredential)
-                .buildAsyncClient();
-
-        // STEP 2: Build Vector Record Collection
-        AzureAISearchVectorStoreRecordCollection<AzureAISearchVectorStoreApproach.MemoryRecord> recordCollection = new AzureAISearchVectorStoreRecordCollection<>(
-                client,
+        // STEP 1: Build Vector Record Collection
+        AzureAISearchVectorStoreRecordCollection<DocumentRecord> recordCollection = new AzureAISearchVectorStoreRecordCollection<>(
+                searchAsyncClient,
                 indexName,
-                AzureAISearchVectorStoreRecordCollectionOptions.<MemoryRecord>builder()
-                        .withRecordClass(AzureAISearchVectorStoreApproach.MemoryRecord.class)
+                AzureAISearchVectorStoreRecordCollectionOptions.<DocumentRecord>builder()
+                        .withRecordClass(DocumentRecord.class)
                         .build()
         );
 
-        // STEP 3: Retrieve relevant documents using keywords extracted from the chat history
+        // STEP 2: Retrieve relevant documents using keywords extracted from the chat history
         String conversationString = ChatGPTUtils.formatAsChatML(questionOrConversation.toOpenAIChatMessages());
-        List<AzureAISearchVectorStoreApproach.MemoryRecord> sourcesResult = getSourcesFromConversation(conversationString, semanticKernel, recordCollection, options);
+        List<DocumentRecord> sourcesResult = getSourcesFromConversation(conversationString, semanticKernel, recordCollection, options);
 
         LOGGER.info("Total {} sources found in cognitive vector store for search query[{}]", sourcesResult.size(), question);
 
-        String sources = AzureAISearchVectorStoreApproach.buildSourcesText(sourcesResult);
-        List<ContentSource> sourcesList = AzureAISearchVectorStoreApproach.buildSources(sourcesResult);
+        String sources = AzureAISearchVectorStoreUtils.buildSourcesText(sourcesResult);
+        List<ContentSource> sourcesList = AzureAISearchVectorStoreUtils.buildSources(sourcesResult);
 
         // STEP 3: Generate a contextual and content specific answer using the search results and chat history
         KernelFunction<String> answerConversation = semanticKernel.getFunction("RAG", "AnswerConversation");
@@ -133,10 +123,10 @@ private ChatHistory removeLastMessage(ChatHistory conversation) {
         return new ChatHistory(messages);
     }
 
-    private List<MemoryRecord> getSourcesFromConversation(String conversation,
-                                                          Kernel kernel,
-                                                          AzureAISearchVectorStoreRecordCollection<MemoryRecord> recordCollection,
-                                                          RAGOptions ragOptions) {
+    private List<DocumentRecord> getSourcesFromConversation(String conversation,
+                                                            Kernel kernel,
+                                                            AzureAISearchVectorStoreRecordCollection<DocumentRecord> recordCollection,
+                                                            RAGOptions ragOptions) {
         KernelFunction<String> extractKeywords = kernel
                 .getPlugin("RAG")
                 .get("ExtractKeywords");
@@ -151,7 +141,7 @@ private List<MemoryRecord> getSourcesFromConversation(String conversation,
                 .block();
         String searchQuery = result.getResult();
 
-        return AzureAISearchVectorStoreApproach.searchAsync(
+        return AzureAISearchVectorStoreUtils.searchAsync(
                 searchQuery,
                 kernel,
                 recordCollection,

diff --git a/...end/src/main/java/com/microsoft/openai/samples/rag/config/AzureAISearchConfiguration.java b/...end/src/main/java/com/microsoft/openai/samples/rag/config/AzureAISearchConfiguration.java
@@ -7,6 +7,8 @@
 import com.azure.search.documents.SearchAsyncClient;
 import com.azure.search.documents.SearchClient;
 import com.azure.search.documents.SearchClientBuilder;
+import com.azure.search.documents.indexes.SearchIndexAsyncClient;
+import com.azure.search.documents.indexes.SearchIndexClientBuilder;
 import org.springframework.beans.factory.annotation.Value;
 import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
 import org.springframework.context.annotation.Bean;
@@ -82,4 +84,15 @@ public SearchAsyncClient asyncSearchDefaultClient() {
                 .indexName(indexName)
                 .buildAsyncClient();
     }
+
+    @Bean
+    @ConditionalOnProperty(name = "cognitive.tracing.enabled", havingValue = "true")
+    public SearchIndexAsyncClient asyncSearchIndexDefaultClient() {
+        String endpoint = "https://%s.search.windows.net".formatted(searchServiceName);
+
+        return new SearchIndexClientBuilder()
+                .endpoint(endpoint)
+                .credential(tokenCredential)
+                .buildAsyncClient();
+    }
 }
diff --git a/...nel/AzureAISearchVectorStoreApproach.java → ...kernel/AzureAISearchVectorStoreUtils.java b/...nel/AzureAISearchVectorStoreApproach.java → ...kernel/AzureAISearchVectorStoreUtils.java
@@ -27,13 +27,13 @@
 import java.util.Optional;
 import java.util.stream.Collectors;
 
-public class AzureAISearchVectorStoreApproach {
+public class AzureAISearchVectorStoreUtils {
 
-    private static final Logger LOGGER = LoggerFactory.getLogger(AzureAISearchVectorStoreApproach.class);
+    private static final Logger LOGGER = LoggerFactory.getLogger(AzureAISearchVectorStoreUtils.class);
 
     private static final String EMBEDDING_FIELD_NAME = "embedding";
 
-    public static class MemoryRecord {
+    public static class DocumentRecord {
         @VectorStoreRecordKey
         private final String id;
         @VectorStoreRecordData
@@ -49,7 +49,7 @@ public static class MemoryRecord {
         @VectorStoreRecordData
         private final String sourceFile;
 
-        public MemoryRecord(
+        public DocumentRecord(
                 @JsonProperty("id") String id,
                 @JsonProperty("content") String content,
                 @JsonProperty("embedding") List<Float> embedding,
@@ -90,10 +90,10 @@ public String getSourceFile() {
     }
 
 
-    public static List<MemoryRecord> searchAsync(String searchQuery,
-                                                 Kernel kernel,
-                                                 AzureAISearchVectorStoreRecordCollection<MemoryRecord> recordCollection,
-                                                 RAGOptions ragOptions) {
+    public static List<DocumentRecord> searchAsync(String searchQuery,
+                                                   Kernel kernel,
+                                                   AzureAISearchVectorStoreRecordCollection<DocumentRecord> recordCollection,
+                                                   RAGOptions ragOptions) {
         // Create VectorSearch options
         VectorSearchOptions vectorSearchOptions = VectorSearchOptions.builder()
                 .withTop(ragOptions.getTop())
@@ -126,15 +126,15 @@ public static List<MemoryRecord> searchAsync(String searchQuery,
         }
 
         // Search the vector store for the relevant documents with the generated embeddings
-        VectorSearchResults<MemoryRecord> memoryResult = recordCollection.hybridSearchAsync(searchQuery, questionVector, vectorSearchOptions, searchOptions)
+        VectorSearchResults<DocumentRecord> memoryResult = recordCollection.hybridSearchAsync(searchQuery, questionVector, vectorSearchOptions, searchOptions)
                 .block();
 
         // Remove the score from the result
         return memoryResult.getResults().stream().map(VectorSearchResult::getRecord).collect(Collectors.toList());
     }
 
 
-    public static List<ContentSource> buildSources(List<MemoryRecord> memoryResult) {
+    public static List<ContentSource> buildSources(List<DocumentRecord> memoryResult) {
         return memoryResult
                 .stream()
                 .map(result -> {
@@ -147,7 +147,7 @@ public static List<ContentSource> buildSources(List<MemoryRecord> memoryResult)
     }
 
 
-    public static String buildSourcesText(List<MemoryRecord> memoryResult) {
+    public static String buildSourcesText(List<DocumentRecord> memoryResult) {
         StringBuilder sourcesContentBuffer = new StringBuilder();
         memoryResult.stream().forEach(memory -> {
             sourcesContentBuffer.append(memory.getSourceFile())

diff --git a/app/frontend/src/pages/chat/Chat.tsx b/app/frontend/src/pages/chat/Chat.tsx
@@ -252,12 +252,12 @@ const Chat = () => {
             text: "Java Azure Open AI SDK"
         },
         {
-            key: Approaches.JAVA_SEMANTIC_KERNEL_PLANNER,
+            key: Approaches.JAVA_SEMANTIC_KERNEL,
             text: "Java Semantic Kernel"
         },
         {
-            key: Approaches.JAVA_SEMANTIC_KERNEL,
-            text: "Java Semantic Kernel - Vector Store"
+            key: Approaches.JAVA_SEMANTIC_KERNEL_PLANNER,
+            text: "Java Semantic Kernel - Chains"
         },
     ];
 

diff --git a/app/frontend/src/pages/oneshot/OneShot.tsx b/app/frontend/src/pages/oneshot/OneShot.tsx
@@ -160,12 +160,12 @@ export function Component(): JSX.Element {
             text: "Java Azure Open AI SDK"
         },
         {
-            key: Approaches.JAVA_SEMANTIC_KERNEL_PLANNER,
+            key: Approaches.JAVA_SEMANTIC_KERNEL,
             text: "Java Semantic Kernel"
         },
         {
-            key: Approaches.JAVA_SEMANTIC_KERNEL,
-            text: "Java Semantic Kernel - Vector Store"
+            key: Approaches.JAVA_SEMANTIC_KERNEL_PLANNER,
+            text: "Java Semantic Kernel - Chains"
         },
     ];