支持上传多份文件

llm-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/controller/CmcChatController.java

@@ -74,7 +74,8 @@ public class CmcChatController extends BaseController
     @PostMapping
     public AjaxResult add(@RequestBody CmcChat cmcChat)
     {
-        cmcChat.setChatId(new SnowFlake().generateId());
+        if (cmcChat.getChatId() == null)
+            cmcChat.setChatId(new SnowFlake().generateId());
         return toAjax(cmcChatService.insertCmcChat(cmcChat));
     }
 

llm-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/controller/CmcDocumentController.java

@@ -1,8 +1,13 @@
 package com.ruoyi.web.llm.controller;
 
+import java.io.File;
+import java.io.IOException;
+import java.util.ArrayList;
 import java.util.List;
 import javax.servlet.http.HttpServletResponse;
 
+import com.alibaba.fastjson2.JSONObject;
+import com.ruoyi.common.config.RuoYiConfig;
 import com.ruoyi.common.utils.SnowFlake;
 import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.web.bind.annotation.GetMapping;
@@ -21,6 +26,7 @@ import com.ruoyi.llm.domain.CmcDocument;
 import com.ruoyi.llm.service.ICmcDocumentService;
 import com.ruoyi.common.utils.poi.ExcelUtil;
 import com.ruoyi.common.core.page.TableDataInfo;
+import org.springframework.web.multipart.MultipartFile;
 
 /**
  * cmc聊天附件Controller
@@ -67,6 +73,35 @@ public class CmcDocumentController extends BaseController
         return success(cmcDocumentService.selectCmcDocumentByDocumentId(documentId));
     }
 
+    /**
+     * 上传外部文件
+     * @return
+     */
+    @GetMapping("/upload")
+    public JSONObject upload(MultipartFile[] fileList) throws IOException {
+        File profilePath = new File( RuoYiConfig.getProfile() + "/upload/rag/document" );
+        if (!profilePath.exists())
+            profilePath.mkdirs();
+        String chatId = new SnowFlake().generateId();
+        JSONObject jsonObject = new JSONObject();
+        jsonObject.put("chatId", chatId);
+        List<String> filenames = new ArrayList<>();
+        for (MultipartFile file : fileList) {
+            File transferFile = new File(profilePath + File.separator + file.getOriginalFilename());
+            if (!transferFile.exists()) {
+                file.transferTo(transferFile);
+            }
+            CmcDocument cmcDocument = new CmcDocument();
+            cmcDocument.setDocumentId(new SnowFlake().generateId());
+            cmcDocument.setChatId(chatId);
+            cmcDocument.setPath(file.getOriginalFilename());
+            cmcDocumentService.insertCmcDocument(cmcDocument);
+            filenames.add(file.getOriginalFilename());
+        }
+        jsonObject.put("filenames", filenames);
+        return jsonObject;
+    }
+
     /**
      * 新增cmc聊天附件
      */
@@ -74,7 +109,6 @@ public class CmcDocumentController extends BaseController
     @PostMapping
     public AjaxResult add(@RequestBody CmcDocument cmcDocument)
     {
-        cmcDocument.setDocumentId(new SnowFlake().generateId());
         return toAjax(cmcDocumentService.insertCmcDocument(cmcDocument));
     }
 

llm-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/controller/SessionController.java

@@ -1,9 +1,6 @@
 package com.ruoyi.web.llm.controller;
 
-import com.ruoyi.common.config.RuoYiConfig;
 import com.ruoyi.common.core.controller.BaseController;
-import com.ruoyi.llm.domain.CmcDocument;
-import com.ruoyi.llm.service.ICmcDocumentService;
 import com.ruoyi.web.llm.service.ILangChainMilvusService;
 import dev.langchain4j.model.embedding.EmbeddingModel;
 import dev.langchain4j.model.embedding.onnx.bgesmallzhv15.BgeSmallZhV15EmbeddingModel;
@@ -12,10 +9,8 @@ import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.web.bind.annotation.GetMapping;
 import org.springframework.web.bind.annotation.RequestMapping;
 import org.springframework.web.bind.annotation.RestController;
-import org.springframework.web.multipart.MultipartFile;
 import reactor.core.publisher.Flux;
 
-import java.io.File;
 import java.io.IOException;
 
 /**
@@ -31,9 +26,6 @@ public class SessionController extends BaseController
     @Autowired
     private ILangChainMilvusService langChainMilvusService;
 
-    @Autowired
-    private ICmcDocumentService cmcDocumentService;
-
     private static final EmbeddingModel embeddingModel = new BgeSmallZhV15EmbeddingModel();
 
     /**
@@ -44,33 +36,13 @@ public class SessionController extends BaseController
         return langChainMilvusService.generateAnswer(topicId, question, "http://192.168.28.188:8000/v1/chat/completions");
     }
 
-    /**
-     * 上传外部文件
-     * @return
-     */
-    @GetMapping("/upload")
-    public String upload(MultipartFile file) throws IOException {
-        File profilePath = new File( RuoYiConfig.getProfile() + "/upload/rag/document" );
-        if (!profilePath.exists())
-            profilePath.mkdirs();
-        File transferFile = new File( profilePath + File.separator + file.getOriginalFilename());
-        if (!transferFile.exists()) {
-            file.transferTo(transferFile);
-        }
-        return file.getOriginalFilename();
-    }
-
     /**
      * 调用LLM+RAG（外部文件）生成回答
      */
     @GetMapping("/answerWithDocument")
-    public Flux<AssistantMessage> answerWithDocument(String filename, String chatId, String topicId, String question) throws IOException
+    public Flux<AssistantMessage> answerWithDocument(String chatId, String question) throws IOException
     {
-        CmcDocument cmcDocument = new CmcDocument();
-        cmcDocument.setChatId(chatId);
-        cmcDocument.setPath(filename);
-        cmcDocumentService.insertCmcDocument(cmcDocument);
-        return langChainMilvusService.generateAnswerWithDocument(embeddingModel, filename, topicId, question, "http://192.168.28.188:8000/v1/chat/completions");
+        return langChainMilvusService.generateAnswerWithDocument(embeddingModel, chatId, question, "http://192.168.28.188:8000/v1/chat/completions");
     }
 
 }

llm-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/service/ILangChainMilvusService.java

@@ -47,6 +47,6 @@ public interface ILangChainMilvusService {
      * 调用LLM+RAG（外部文件）生成回答
      * @return
      */
-    public Flux<AssistantMessage> generateAnswerWithDocument(EmbeddingModel embeddingModel, String filename, String topicId, String question, String llmServiceUrl) throws IOException;
+    public Flux<AssistantMessage> generateAnswerWithDocument(EmbeddingModel embeddingModel, String chatId, String question, String llmServiceUrl) throws IOException;
 
 }

llm-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/service/impl/LangChainMilvusServiceImpl.java

@@ -3,7 +3,9 @@ package com.ruoyi.web.llm.service.impl;
 import com.alibaba.fastjson2.JSONObject;
 import com.ruoyi.common.config.RuoYiConfig;
 import com.ruoyi.llm.domain.CmcChat;
+import com.ruoyi.llm.domain.CmcDocument;
 import com.ruoyi.llm.service.ICmcChatService;
+import com.ruoyi.llm.service.ICmcDocumentService;
 import com.ruoyi.web.llm.service.ILangChainMilvusService;
 import dev.langchain4j.data.document.Document;
 import dev.langchain4j.data.document.parser.apache.pdfbox.ApachePdfBoxDocumentParser;
@@ -48,6 +50,9 @@ public class LangChainMilvusServiceImpl implements ILangChainMilvusService
     @Autowired
     private ICmcChatService cmcChatService;
 
+    @Autowired
+    private ICmcDocumentService cmcDocumentService;
+
     /**
      * 导入知识库文件
      */
@@ -146,12 +151,14 @@ public class LangChainMilvusServiceImpl implements ILangChainMilvusService
                 .apiKey("1")
                 .build();
 
-        CmcChat cmcChat = new CmcChat();
-        cmcChat.setTopicId(topicId);
-        List<CmcChat> cmcChatList = cmcChatService.selectCmcChatList(cmcChat);
-        for (CmcChat chat : cmcChatList) {
-            chatSession.addMessage(ChatMessage.ofUser(chat.getInput()));
-            chatSession.addMessage(ChatMessage.ofAssistant(chat.getOutput()));
+        if (topicId != null) {
+            CmcChat cmcChat = new CmcChat();
+            cmcChat.setTopicId(topicId);
+            List<CmcChat> cmcChatList = cmcChatService.selectCmcChatList(cmcChat);
+            for (CmcChat chat : cmcChatList) {
+                chatSession.addMessage(ChatMessage.ofUser(chat.getInput()));
+                chatSession.addMessage(ChatMessage.ofAssistant(chat.getOutput()));
+            }
         }
         chatSession.addMessage(ChatMessage.ofUser(prompt));
 
@@ -184,26 +191,30 @@ public class LangChainMilvusServiceImpl implements ILangChainMilvusService
      * 调用LLM生成回答
      */
     @Override
-    public Flux<AssistantMessage> generateAnswerWithDocument(EmbeddingModel embeddingModel, String filename, String topicId, String question, String llmServiceUrl) throws IOException {
-
-        File profilePath = new File( RuoYiConfig.getProfile() + "/upload/rag/document/" + filename);
-        List<TextSegment> segments = splitDocument(filename, profilePath);
-        List<Embedding> embeddings = embeddingModel.embedAll(segments).content();
-        InMemoryEmbeddingStore<TextSegment> embeddingStore = new InMemoryEmbeddingStore<>();
-        embeddingStore.addAll(embeddings, segments);
-        Embedding queryEmbedding = embeddingModel.embed(question).content();
-        EmbeddingSearchRequest embeddingSearchRequest = EmbeddingSearchRequest.builder()
-                .queryEmbedding(queryEmbedding)
-                .maxResults(1)
-                .build();
-        String contexts = embeddingStore.search(embeddingSearchRequest).matches().get(0).embedded().text();
-        String sb = "问题: " + question + "\n\n" +
-                "根据以下上下文回答问题：\n\n" +
-                "文件" + ": " +
-                filename + "\n\n" +
-                "上下文" + ": " +
-                contexts + "\n\n";
-        return generateAnswer(topicId, sb, llmServiceUrl);
+    public Flux<AssistantMessage> generateAnswerWithDocument(EmbeddingModel embeddingModel, String chatId, String question, String llmServiceUrl) throws IOException {
+        String topicId = cmcChatService.selectCmcChatByChatId(chatId).getTopicId();
+        CmcDocument cmcDocument = new CmcDocument();
+        cmcDocument.setChatId(chatId);
+        List<CmcDocument> documentList = cmcDocumentService.selectCmcDocumentList(cmcDocument);
+        StringBuilder sb = new StringBuilder("问题: " + question + "\n\n").append("根据以下上下文回答问题：\n\n");
+        for (CmcDocument document : documentList) {
+            File profilePath = new File(RuoYiConfig.getProfile() + "/upload/rag/document/" + document.getPath());
+            List<TextSegment> segments = splitDocument(document.getPath(), profilePath);
+            List<Embedding> embeddings = embeddingModel.embedAll(segments).content();
+            InMemoryEmbeddingStore<TextSegment> embeddingStore = new InMemoryEmbeddingStore<>();
+            embeddingStore.addAll(embeddings, segments);
+            Embedding queryEmbedding = embeddingModel.embed(question).content();
+            EmbeddingSearchRequest embeddingSearchRequest = EmbeddingSearchRequest.builder()
+                    .queryEmbedding(queryEmbedding)
+                    .maxResults(1)
+                    .build();
+            String contexts = embeddingStore.search(embeddingSearchRequest).matches().get(0).embedded().text();
+            sb.append("文件").append(": ")
+                    .append(document.getPath()).append("\n\n")
+                    .append("上下文").append(": ")
+                    .append(contexts).append("\n\n");
+        }
+        return generateAnswer(topicId, sb.toString(), llmServiceUrl);
     }
 
     /**
@@ -270,8 +281,4 @@ public class LangChainMilvusServiceImpl implements ILangChainMilvusService
         DocumentByParagraphSplitter splitter = new DocumentByParagraphSplitter(1000,200);
         return splitter.split(document);
     }
-    interface Assistant {
-
-        String chat(String message);
-    }
 }

llm-back/vllm_server.py

@@ -0,0 +1,117 @@
+from vllm import LLM, SamplingParams
+from fastapi import FastAPI, Request
+from fastapi.responses import StreamingResponse
+import uvicorn
+import time
+import json
+
+def create_vllm_server(
+    model: str,
+    served_model_name: str,
+    host: str,
+    port: int,
+    tensor_parallel_size: int,
+    top_p: float,
+    temperature: float,
+    max_tokens: int,
+    gpu_memory_utilization: float,
+    dtype: str,
+) -> FastAPI:
+    # 只初始化 LLM
+    llm = LLM(
+        model=model,
+        tensor_parallel_size=tensor_parallel_size,
+        gpu_memory_utilization=gpu_memory_utilization,
+        dtype=dtype,
+    )
+
+    sampling_params = SamplingParams(
+        temperature=temperature,
+        top_p=top_p,
+        max_tokens=max_tokens,
+    )
+
+    app = FastAPI()
+
+    @app.post("/v1/chat/completions")
+    async def chat_completions(request: Request):
+        try:
+            data = await request.json()
+            messages = data["messages"]
+            tools = data.get("tools")  # 支持 tools 参数
+            created_time = time.time()
+            request_id = f"chatcmpl-{int(time.time())}"
+
+            # 调用 llm.chat()，传入 tools
+            outputs = llm.chat(
+                messages=messages,
+                sampling_params=sampling_params,
+                tools=tools,
+            )
+            if data.get("stream"):
+                def generate():
+                    full_text = ""
+                    for output in outputs:
+                        new_text = output.outputs[0].text[len(full_text):]
+                        full_text = output.outputs[0].text
+                        response_data = {
+                            "id": request_id,
+                            "model": served_model_name,
+                            "created": created_time,
+                            "choices": [{
+                                "index": 0,
+                                "delta": {"content": new_text},
+                                "finish_reason": output.outputs[0].finish_reason,
+                            }],
+                        }
+                        yield f"data: {json.dumps(response_data)}\n\n"
+                    yield "data: [DONE]\n\n"
+
+                return StreamingResponse(generate(), media_type="text/event-stream")
+            else:
+                return {
+                    "id": request_id,
+                    "model": served_model_name,
+                    "created": created_time,
+                    "choices": [{
+                        "index": 0,
+                        "message": {
+                            "role": "assistant",
+                            "content": outputs[0].outputs[0].text,
+                        },
+                        "finish_reason": outputs[0].outputs[0].finish_reason,
+                    }],
+                }
+
+        except Exception as e:
+            return {"error": str(e)}, 400
+
+    return app
+
+if __name__ == "__main__":
+    # 配置参数
+    CONFIG = {
+        "model": "/mnt/d/Qwen/Qwen2.5-1.5B-Instruct",
+        "served_model_name": "Qwen2.5-1.5B-Instruct",
+        # "model": "/mnt/d/Deepseek/DeepSeek-R1-Distill-Qwen-1.5B",
+        # "served_model_name": "DeepSeek-R1-Distill-Qwen-1.5B",
+        "host": "172.25.231.226",
+        "port": 8000,
+        "tensor_parallel_size": 1,
+        "top_p": 0.9,
+        "temperature": 0.7,
+        "max_tokens": 8192,
+        "gpu_memory_utilization": 0.9,
+        "dtype": "float16"
+    }
+    
+    # 创建应用
+    app = create_vllm_server(**CONFIG)
+    
+    # 启动服务器
+    uvicorn.run(
+        app,
+        host=CONFIG["host"],
+        port=CONFIG["port"],
+        workers= 1,
+    )