修改适配VLLM

ask-data-ai/ask-data-ai-biz/pom.xml

@@ -59,13 +59,19 @@
                 </exclusion>
             </exclusions>
         </dependency>
-
+        <dependency>
+            <groupId>org.springframework.ai</groupId>
+            <artifactId>spring-ai-starter-model-ollama</artifactId>
+        </dependency>
         <!-- undertow容器 -->
         <dependency>
             <groupId>org.springframework.boot</groupId>
             <artifactId>spring-boot-starter-undertow</artifactId>
         </dependency>
-
+        <dependency>
+            <groupId>io.projectreactor.netty</groupId>
+            <artifactId>reactor-netty-http</artifactId>
+        </dependency>
         <!-- SpringBoot Actuator -->
         <dependency>
             <groupId>org.springframework.boot</groupId>

ask-data-ai/ask-data-ai-biz/src/main/java/com/ask/config/CommonConfiguration.java

@@ -11,13 +11,18 @@ import org.springframework.ai.chat.memory.repository.jdbc.JdbcChatMemoryReposito
 import org.springframework.ai.chat.prompt.PromptTemplate;
 import org.springframework.ai.deepseek.DeepSeekChatModel;
 import org.springframework.ai.openai.OpenAiChatModel;
+import org.springframework.ai.openai.api.OpenAiApi;
 import org.springframework.ai.rag.advisor.RetrievalAugmentationAdvisor;
 import org.springframework.ai.rag.generation.augmentation.ContextualQueryAugmenter;
 import org.springframework.ai.rag.retrieval.search.VectorStoreDocumentRetriever;
 import org.springframework.ai.vectorstore.VectorStore;
 import org.springframework.context.annotation.Bean;
 import org.springframework.context.annotation.Configuration;
+import org.springframework.http.client.reactive.ReactorClientHttpConnector;
 import org.springframework.jdbc.core.JdbcTemplate;
+import org.springframework.web.reactive.function.client.WebClient;
+import reactor.netty.http.HttpProtocol;
+import reactor.netty.http.client.HttpClient;
 
 import java.util.ArrayList;
 import java.util.List;
@@ -25,7 +30,13 @@ import java.util.List;
 @Configuration
 public class CommonConfiguration {
 
-
+    @Bean
+    public WebClient.Builder openAiWebClientBuilder() {
+        return WebClient.builder()
+                .clientConnector(new ReactorClientHttpConnector(
+                        HttpClient.create().protocol(HttpProtocol.HTTP11)
+                ));
+    }
     @Bean
     public ChatMemory chatMemory(JdbcTemplate jdbcTemplate, PostgresChatMemoryDialect postgresChatMemoryDialect) {
         ChatMemoryRepository chatMemoryRepository = JdbcChatMemoryRepository.builder()

ask-data-ai/ask-data-ai-biz/src/main/java/com/ask/controller/ChatController.java

@@ -109,7 +109,7 @@ public class ChatController {
         Message userMessage = new UserMessage(message);
         Prompt prompt = new Prompt(List.of(systemMessage, userMessage));
 
-        return FluxUtils.wrapDeepSeekStream(openAiChatClient.prompt(prompt)
+        return FluxUtils.wrapDeepSeekStream(deepseekChatClient.prompt(prompt)
                 .advisors(messageChatMemoryAdvisor)
                 .advisors(a -> a.param(ChatMemory.CONVERSATION_ID, conversationId))
                 .stream()
@@ -145,7 +145,7 @@ public class ChatController {
         String userPrompt = ragPromptService.createRagPrompt(message, context, historyMemory);
 
         StringBuilder contentBuilder = new StringBuilder();
-        return FluxUtils.wrapDeepSeekStream(openAiChatClient.prompt()
+        return FluxUtils.wrapDeepSeekStream(deepseekChatClient.prompt()
                         .user(userPrompt)
                         .system("你是一个智能助手，基于以下上下文和历史对话回答问题,请用简洁的语言回答问题，并确保答案准确,要求" +
                                 "1.以 Markdown 格式输出")

ask-data-ai/ask-data-ai-biz/src/main/java/com/ask/service/impl/AskVectorStoreServiceImpl.java

@@ -42,7 +42,7 @@ import java.util.stream.Collectors;
 public class AskVectorStoreServiceImpl extends ServiceImpl<AskVectorStoreMapper, AskVectorStore> implements AskVectorStoreService {
 
     @Autowired
-    private EmbeddingModel embeddingModel;
+    private EmbeddingModel ollamaEmbeddingModel;
 
     @Autowired
     private JdbcTemplate jdbcTemplate;
@@ -74,7 +74,7 @@ public class AskVectorStoreServiceImpl extends ServiceImpl<AskVectorStoreMapper,
                     String result = (title == null || title.trim().isEmpty()) ?
                             (content == null ? "" : content) :
                             title.trim() + "\n" + (content == null ? "" : content);
-                    return embeddingModel.embed(result);
+                    return ollamaEmbeddingModel.embed(result);
                 })
                 .toList();
 

ask-data-ai/ask-data-ai-boot/src/main/resources/application.yml

@@ -18,6 +18,8 @@ spring:
     password: e5d039e4ba5246068
     driver-class-name: org.postgresql.Driver
   ai:
+    model:
+      embedding: ollama
     vectorstore:
       pgvector:
         index-type: HNSW
@@ -38,11 +40,7 @@ spring:
       chat:
         options:
           model: deepseek-r1-0528
-      embedding:
-        base-url: https://dashscope.aliyuncs.com/compatible-mode
-        api-key: sk-ae96ff281ff644c992843c64a711a950
-        options:
-          model: text-embedding-v4
+
     deepseek:
       base-url: https://dashscope.aliyuncs.com/compatible-mode/v1
       api-key: sk-ae96ff281ff644c992843c64a711a950
@@ -50,6 +48,11 @@ spring:
         enabled: true
         options:
           model: deepseek-r1-0528
+    ollama:
+      base-url: http://127.0.0.1:11434
+      embedding:
+        options:
+          model: nomic-embed-text
 
 mybatis-plus:
   mapper-locations: classpath*:/mapper/*Mapper.xml # mapper文件位置