(improvement)(headless) Add support for the Ollama provider in the frontend and optimize the code (#1270)

chat/server/src/main/java/com/tencent/supersonic/chat/server/executor/PlainTextExecutor.java

@@ -7,7 +7,6 @@ import com.tencent.supersonic.chat.server.persistence.repository.ChatQueryReposi
 import com.tencent.supersonic.chat.server.pojo.ChatExecuteContext;
 import com.tencent.supersonic.chat.server.service.AgentService;
 import com.tencent.supersonic.common.util.ContextUtils;
-import com.tencent.supersonic.common.util.S2ChatModelProvider;
 import com.tencent.supersonic.headless.api.pojo.response.ParseResp;
 import com.tencent.supersonic.headless.api.pojo.response.QueryResult;
 import com.tencent.supersonic.headless.api.pojo.response.QueryState;
@@ -16,6 +15,7 @@ import dev.langchain4j.model.chat.ChatLanguageModel;
 import dev.langchain4j.model.input.Prompt;
 import dev.langchain4j.model.input.PromptTemplate;
 import dev.langchain4j.model.output.Response;
+import dev.langchain4j.model.provider.ChatLanguageModelProvider;
 
 import java.util.Collections;
 import java.util.List;
@@ -46,7 +46,7 @@ public class PlainTextExecutor implements ChatExecutor {
         AgentService agentService = ContextUtils.getBean(AgentService.class);
         Agent chatAgent = agentService.getAgent(chatExecuteContext.getAgentId());
 
-        ChatLanguageModel chatLanguageModel = S2ChatModelProvider.provide(chatAgent.getLlmConfig());
+        ChatLanguageModel chatLanguageModel = ChatLanguageModelProvider.provide(chatAgent.getLlmConfig());
         Response<AiMessage> response = chatLanguageModel.generate(prompt.toUserMessage());
 
         QueryResult result = new QueryResult();

chat/server/src/main/java/com/tencent/supersonic/chat/server/memory/MemoryReviewTask.java

@@ -4,10 +4,10 @@ import com.tencent.supersonic.chat.api.pojo.enums.MemoryReviewResult;
 import com.tencent.supersonic.chat.server.agent.Agent;
 import com.tencent.supersonic.chat.server.service.AgentService;
 import com.tencent.supersonic.chat.server.service.MemoryService;
-import com.tencent.supersonic.common.util.S2ChatModelProvider;
 import dev.langchain4j.model.chat.ChatLanguageModel;
 import dev.langchain4j.model.input.Prompt;
 import dev.langchain4j.model.input.PromptTemplate;
+import dev.langchain4j.model.provider.ChatLanguageModelProvider;
 import lombok.extern.slf4j.Slf4j;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -54,7 +54,7 @@ public class MemoryReviewTask {
                     Prompt prompt = PromptTemplate.from(promptStr).apply(Collections.EMPTY_MAP);
 
                     keyPipelineLog.info("MemoryReviewTask reqPrompt:{}", promptStr);
-                    ChatLanguageModel chatLanguageModel = S2ChatModelProvider.provide(chatAgent.getLlmConfig());
+                    ChatLanguageModel chatLanguageModel = ChatLanguageModelProvider.provide(chatAgent.getLlmConfig());
                     String response = chatLanguageModel.generate(prompt.toUserMessage()).content().text();
                     keyPipelineLog.info("MemoryReviewTask modelResp:{}", response);
 

chat/server/src/main/java/com/tencent/supersonic/chat/server/parser/NL2SQLParser.java

@@ -8,7 +8,6 @@ import com.tencent.supersonic.chat.server.pojo.ChatParseContext;
 import com.tencent.supersonic.chat.server.util.QueryReqConverter;
 import com.tencent.supersonic.common.config.LLMConfig;
 import com.tencent.supersonic.common.util.ContextUtils;
-import com.tencent.supersonic.common.util.S2ChatModelProvider;
 import com.tencent.supersonic.headless.api.pojo.SchemaElement;
 import com.tencent.supersonic.headless.api.pojo.SchemaElementMatch;
 import com.tencent.supersonic.headless.api.pojo.SchemaElementType;
@@ -24,6 +23,7 @@ import dev.langchain4j.model.chat.ChatLanguageModel;
 import dev.langchain4j.model.input.Prompt;
 import dev.langchain4j.model.input.PromptTemplate;
 import dev.langchain4j.model.output.Response;
+import dev.langchain4j.model.provider.ChatLanguageModelProvider;
 import lombok.Builder;
 import lombok.Data;
 import lombok.extern.slf4j.Slf4j;
@@ -180,7 +180,7 @@ public class NL2SQLParser implements ChatParser {
         Prompt prompt = PromptTemplate.from(promptStr).apply(Collections.EMPTY_MAP);
         keyPipelineLog.info("NL2SQLParser reqPrompt:{}", promptStr);
 
-        ChatLanguageModel chatLanguageModel = S2ChatModelProvider.provide(context.getLlmConfig());
+        ChatLanguageModel chatLanguageModel = ChatLanguageModelProvider.provide(context.getLlmConfig());
         Response<AiMessage> response = chatLanguageModel.generate(prompt.toUserMessage());
 
         String result = response.content().text();

chat/server/src/main/java/com/tencent/supersonic/chat/server/util/LLMConnHelper.java

@@ -2,8 +2,8 @@ package com.tencent.supersonic.chat.server.util;
 
 import com.tencent.supersonic.common.config.LLMConfig;
 import com.tencent.supersonic.common.pojo.exception.InvalidArgumentException;
-import com.tencent.supersonic.common.util.S2ChatModelProvider;
 import dev.langchain4j.model.chat.ChatLanguageModel;
+import dev.langchain4j.model.provider.ChatLanguageModelProvider;
 import lombok.extern.slf4j.Slf4j;
 import org.apache.commons.lang3.StringUtils;
 
@@ -14,7 +14,7 @@ public class LLMConnHelper {
             if (llmConfig == null || StringUtils.isBlank(llmConfig.getBaseUrl())) {
                 return false;
             }
-            ChatLanguageModel chatLanguageModel = S2ChatModelProvider.provide(llmConfig);
+            ChatLanguageModel chatLanguageModel = ChatLanguageModelProvider.provide(llmConfig);
             String response = chatLanguageModel.generate("Hi there");
             return StringUtils.isNotEmpty(response) ? true : false;
         } catch (Exception e) {

common/src/main/java/com/tencent/supersonic/common/pojo/enums/S2ModelProvider.java

@@ -1,9 +0,0 @@
-package com.tencent.supersonic.common.pojo.enums;
-
-public enum S2ModelProvider {
-
-    OPEN_AI,
-    HUGGING_FACE,
-    LOCAL_AI,
-    IN_PROCESS
-}

common/src/main/java/com/tencent/supersonic/common/util/S2ChatModelProvider.java

@@ -1,41 +0,0 @@
-package com.tencent.supersonic.common.util;
-
-import com.tencent.supersonic.common.config.LLMConfig;
-import com.tencent.supersonic.common.pojo.enums.S2ModelProvider;
-import dev.langchain4j.model.chat.ChatLanguageModel;
-import dev.langchain4j.model.localai.LocalAiChatModel;
-import dev.langchain4j.model.openai.OpenAiChatModel;
-import org.apache.commons.lang3.StringUtils;
-
-import java.time.Duration;
-
-public class S2ChatModelProvider {
-
-    public static ChatLanguageModel provide(LLMConfig llmConfig) {
-        ChatLanguageModel chatLanguageModel = ContextUtils.getBean(ChatLanguageModel.class);
-        if (llmConfig == null || StringUtils.isBlank(llmConfig.getProvider())
-                || StringUtils.isBlank(llmConfig.getBaseUrl())) {
-            return chatLanguageModel;
-        }
-        if (S2ModelProvider.OPEN_AI.name().equalsIgnoreCase(llmConfig.getProvider())) {
-            return OpenAiChatModel
-                    .builder()
-                    .baseUrl(llmConfig.getBaseUrl())
-                    .modelName(llmConfig.getModelName())
-                    .apiKey(llmConfig.keyDecrypt())
-                    .temperature(llmConfig.getTemperature())
-                    .timeout(Duration.ofSeconds(llmConfig.getTimeOut()))
-                    .build();
-        } else if (S2ModelProvider.LOCAL_AI.name().equalsIgnoreCase(llmConfig.getProvider())) {
-            return LocalAiChatModel
-                    .builder()
-                    .baseUrl(llmConfig.getBaseUrl())
-                    .modelName(llmConfig.getModelName())
-                    .temperature(llmConfig.getTemperature())
-                    .timeout(Duration.ofSeconds(llmConfig.getTimeOut()))
-                    .build();
-        }
-        throw new RuntimeException("unsupported provider: " + llmConfig.getProvider());
-    }
-
-}

common/src/main/java/dev/langchain4j/model/provider/ChatLanguageModelFactory.java

@@ -0,0 +1,8 @@
+package dev.langchain4j.model.provider;
+
+import com.tencent.supersonic.common.config.LLMConfig;
+import dev.langchain4j.model.chat.ChatLanguageModel;
+
+public interface ChatLanguageModelFactory {
+    ChatLanguageModel create(LLMConfig llmConfig);
+}

common/src/main/java/dev/langchain4j/model/provider/ChatLanguageModelProvider.java

@@ -0,0 +1,33 @@
+package dev.langchain4j.model.provider;
+
+import com.tencent.supersonic.common.config.LLMConfig;
+import com.tencent.supersonic.common.util.ContextUtils;
+import dev.langchain4j.model.chat.ChatLanguageModel;
+import org.apache.commons.lang3.StringUtils;
+
+import java.util.HashMap;
+import java.util.Map;
+
+public class ChatLanguageModelProvider {
+    private static final Map<String, ChatLanguageModelFactory> factories = new HashMap<>();
+
+    static {
+        factories.put(ModelProvider.OPEN_AI.name(), new OpenAiChatModelFactory());
+        factories.put(ModelProvider.LOCAL_AI.name(), new LocalAiChatModelFactory());
+        factories.put(ModelProvider.OLLAMA.name(), new OllamaChatModelFactory());
+    }
+
+    public static ChatLanguageModel provide(LLMConfig llmConfig) {
+        if (llmConfig == null || StringUtils.isBlank(llmConfig.getProvider())
+                || StringUtils.isBlank(llmConfig.getBaseUrl())) {
+            return ContextUtils.getBean(ChatLanguageModel.class);
+        }
+
+        ChatLanguageModelFactory factory = factories.get(llmConfig.getProvider().toUpperCase());
+        if (factory != null) {
+            return factory.create(llmConfig);
+        }
+
+        throw new RuntimeException("Unsupported provider: " + llmConfig.getProvider());
+    }
+}

common/src/main/java/dev/langchain4j/model/provider/LocalAiChatModelFactory.java

@@ -0,0 +1,20 @@
+package dev.langchain4j.model.provider;
+
+import com.tencent.supersonic.common.config.LLMConfig;
+import dev.langchain4j.model.chat.ChatLanguageModel;
+import dev.langchain4j.model.localai.LocalAiChatModel;
+
+import java.time.Duration;
+
+public class LocalAiChatModelFactory implements ChatLanguageModelFactory {
+    @Override
+    public ChatLanguageModel create(LLMConfig llmConfig) {
+        return LocalAiChatModel
+                .builder()
+                .baseUrl(llmConfig.getBaseUrl())
+                .modelName(llmConfig.getModelName())
+                .temperature(llmConfig.getTemperature())
+                .timeout(Duration.ofSeconds(llmConfig.getTimeOut()))
+                .build();
+    }
+}

common/src/main/java/dev/langchain4j/model/provider/ModelProvider.java

@@ -0,0 +1,9 @@
+package dev.langchain4j.model.provider;
+
+public enum ModelProvider {
+    OPEN_AI,
+    HUGGING_FACE,
+    LOCAL_AI,
+    IN_PROCESS,
+    OLLAMA
+}

common/src/main/java/dev/langchain4j/model/provider/OllamaChatModelFactory.java

@@ -0,0 +1,20 @@
+package dev.langchain4j.model.provider;
+
+import com.tencent.supersonic.common.config.LLMConfig;
+import dev.langchain4j.model.chat.ChatLanguageModel;
+import dev.langchain4j.model.ollama.OllamaChatModel;
+
+import java.time.Duration;
+
+public class OllamaChatModelFactory implements ChatLanguageModelFactory {
+    @Override
+    public ChatLanguageModel create(LLMConfig llmConfig) {
+        return OllamaChatModel
+                .builder()
+                .baseUrl(llmConfig.getBaseUrl())
+                .modelName(llmConfig.getModelName())
+                .temperature(llmConfig.getTemperature())
+                .timeout(Duration.ofSeconds(llmConfig.getTimeOut()))
+                .build();
+    }
+}

common/src/main/java/dev/langchain4j/model/provider/OpenAiChatModelFactory.java

@@ -0,0 +1,21 @@
+package dev.langchain4j.model.provider;
+
+import com.tencent.supersonic.common.config.LLMConfig;
+import dev.langchain4j.model.chat.ChatLanguageModel;
+import dev.langchain4j.model.openai.OpenAiChatModel;
+
+import java.time.Duration;
+
+public class OpenAiChatModelFactory implements ChatLanguageModelFactory {
+    @Override
+    public ChatLanguageModel create(LLMConfig llmConfig) {
+        return OpenAiChatModel
+                .builder()
+                .baseUrl(llmConfig.getBaseUrl())
+                .modelName(llmConfig.getModelName())
+                .apiKey(llmConfig.keyDecrypt())
+                .temperature(llmConfig.getTemperature())
+                .timeout(Duration.ofSeconds(llmConfig.getTimeOut()))
+                .build();
+    }
+}

headless/chat/src/main/java/com/tencent/supersonic/headless/chat/parser/llm/SqlGenStrategy.java

@@ -3,8 +3,8 @@ package com.tencent.supersonic.headless.chat.parser.llm;
 import com.tencent.supersonic.common.config.LLMConfig;
 import com.tencent.supersonic.headless.chat.query.llm.s2sql.LLMReq;
 import com.tencent.supersonic.headless.chat.query.llm.s2sql.LLMResp;
-import com.tencent.supersonic.common.util.S2ChatModelProvider;
 import dev.langchain4j.model.chat.ChatLanguageModel;
+import dev.langchain4j.model.provider.ChatLanguageModelProvider;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 import org.springframework.beans.factory.InitializingBean;
@@ -24,7 +24,7 @@ public abstract class SqlGenStrategy implements InitializingBean {
     protected PromptHelper promptHelper;
 
     protected ChatLanguageModel getChatLanguageModel(LLMConfig llmConfig) {
-        return S2ChatModelProvider.provide(llmConfig);
+        return ChatLanguageModelProvider.provide(llmConfig);
     }
 
     abstract LLMResp generate(LLMReq llmReq);

launchers/standalone/src/main/resources/langchain4j-config.yaml

@@ -13,4 +13,25 @@ langchain4j:
     embedding-model:
       model-name: bge-small-zh
     embedding-store:
-      persist-path: /tmp
+      persist-path: /tmp
+#  ollama:
+#    chat-model:
+#      base-url: http://localhost:11434
+#      api-key: demo
+#      model-name: qwen:0.5b
+#      temperature: 0.0
+#      timeout: PT60S
+
+#  chroma:
+#    embedding-store:
+#      baseUrl: http://0.0.0.0:8000
+#      timeout: 120s
+
+#  milvus:
+#    embedding-store:
+#      host: localhost
+#      port: 2379
+#      uri: http://0.0.0.0:19530
+#      token: demo
+#      dimension: 512
+#      timeout: 120s

launchers/standalone/src/test/resources/langchain4j-config.yaml

@@ -13,4 +13,25 @@ langchain4j:
     embedding-model:
       model-name: bge-small-zh
     embedding-store:
-      persist-path: /tmp
+      persist-path: /tmp
+#  ollama:
+#    chat-model:
+#      base-url: http://localhost:11434
+#      api-key: demo
+#      model-name: qwen:0.5b
+#      temperature: 0.0
+#      timeout: PT60S
+
+#  chroma:
+#    embedding-store:
+#      baseUrl: http://0.0.0.0:8000
+#      timeout: 120s
+
+#  milvus:
+#    embedding-store:
+#      host: localhost
+#      port: 2379
+#      uri: http://0.0.0.0:19530
+#      token: demo
+#      dimension: 512
+#      timeout: 120s