server:
  port: 18082
spring:
  ai:
    dashscope:
      api-key: sk-8718a83408d7443b9544cdfbXXXXX
    deepseek: ## 这一行是你选择的LLM模型，如果是openai，这里就填openai， base-url就是填对应厂商的地址
      api-key: "sk-199324596dbb4308afcb77d4XXXXX"
      base-url: "https://api.deepseek.com"
      chat:
        options:
          model: deepseek-chat
      embedding:
        enabled: false
    ollama:
      chat:
        options:
          model: gemma3:4b
          #model: qwen2.5vl:3b   moondream:latest qwen2.5vl:3b-q4_K_M   moondream:v2
          #model: qwen2:0.5b
          #temperature: 0.1
          # 指定默认使用的模型，也可以在代码中动态覆盖
          #model: gemma3:4b
          ####最小的轻量多模态模型（约 1.5GB）
          #model: minicpm-v:latest
      base-url: http://192.168.91.164:11434
  main: #允许 Bean 覆盖
    allow-bean-definition-overriding: true

public class MultiPlatformAndModelOptions {
    private String platform;    // 平台标识：dashscope/deepseek/ollama
    private String model;       // 模型名称：qwen-plus/deepseek-chat/llama3
    private Double temperature; // 温度参数，控制生成随机性
    public String getPlatform() {
        return platform;
    }
    public void setPlatform(String platform) {
        this.platform = platform;
    }
    public String getModel() {
        return model;
    }
    public void setModel(String model) {
        this.model = model;
    }
    public Double getTemperature() {
        return temperature;
    }
    public void setTemperature(Double temperature) {
        this.temperature = temperature;
    }
}

import com.alibaba.cloud.ai.dashscope.chat.DashScopeChatModel;
import com.sb.multiplatform18083.tool.MultiPlatformAndModelOptions;
import jakarta.servlet.http.HttpServletResponse;
import org.springframework.ai.chat.client.ChatClient;
import org.springframework.ai.chat.model.ChatModel;
import org.springframework.ai.chat.prompt.ChatOptions;
import org.springframework.ai.deepseek.DeepSeekChatModel;
import org.springframework.ai.ollama.OllamaChatModel;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.RequestParam;
import org.springframework.web.bind.annotation.RestController;
import reactor.core.publisher.Flux;
import java.util.HashMap;
import java.util.Map;
@RestController
public class MultiPlatformAndModelController {
    private final Map<String, ChatModel> platforms = new HashMap<>();
    // 构造器注入所有 ChatModel
    public MultiPlatformAndModelController(
            DashScopeChatModel dashScopeChatModel,
            DeepSeekChatModel deepSeekChatModel,
            OllamaChatModel ollamaChatModel) {
        platforms.put("dashscope", dashScopeChatModel);
        platforms.put("ollama", ollamaChatModel);
        platforms.put("deepseek", deepSeekChatModel);
    }
    @GetMapping(value = "/chat", produces = "text/event-stream;charset=UTF-8")
    public Flux<String> chat(
            @RequestParam(value = "message") String message,
            @RequestParam(value = "platform") String platform,
            @RequestParam(value = "model") String model,
            @RequestParam(value = "temperature", defaultValue = "0.7") Float temperature
            , HttpServletResponse response // 1. 注入 HttpServletResponse
    ) {
        // 2. 【核心修复】强制设置响应内容类型和编码
        response.setContentType("text/event-stream;charset=UTF-8");
        // 防止浏览器或中间件缓存流数据
        response.setHeader("Cache-Control", "no-cache");
        // 手动封装参数（比直接用对象更稳定，不会出现参数解析失败）
        MultiPlatformAndModelOptions options = new MultiPlatformAndModelOptions();
        options.setPlatform(platform);
        options.setModel(model);
        options.setTemperature(Double.valueOf(temperature));
        // 1. 根据平台获取对应的 ChatModel
        ChatModel chatModel = platforms.get(options.getPlatform());
        if (chatModel == null) {
            return Flux.just("错误：不支持的平台：" + options.getPlatform());
        }
        // 2. 构建 ChatClient，动态配置模型参数
        ChatClient chatClient = ChatClient.builder(chatModel)
                .defaultOptions(ChatOptions.builder()
                        .temperature(options.getTemperature())
                        .model(options.getModel())
                        .build())
                .build();
        // 3. 发起流式对话请求
        return chatClient.prompt()
                .user(message)
                .stream()
                .content();
    }
}