🔧 chore(ai): 增强 AI 提供商管理系统

- 添加智谱、通义和豆包等新的 AI 模型提供商支持
- 改进 AI 提供商工厂的缓存管理机制，增加 30 分钟缓存清理
- 优化错误处理和模型信息展示
- 统一本地化消息的使用
- 重构 OpenAI Provider，提取基类以复用代码
- 改进各提供商的代码结构和异常处理

src/ai/AIProviderFactory.ts

@@ -5,19 +5,34 @@ import { AIProvider, ConfigKeys } from "../config/types";
 import { ConfigurationManager } from "../config/ConfigurationManager";
 import { VSCodeProvider } from "./providers/VscodeProvider";
 import { LocalizationManager } from "../utils/LocalizationManager";
+import { ZhipuAIProvider } from "./providers/ZhipuAIProvider";
+import { DashScopeProvider } from "./providers/DashScopeProvider";
+import { DoubaoProvider } from "./providers/DoubaoProvider";
 
 export class AIProviderFactory {
   private static providers: Map<string, AIProviderInterface> = new Map();
+  private static readonly PROVIDER_CACHE_TTL = 1000 * 60 * 30; // 30分钟缓存
+  private static providerTimestamps: Map<string, number> = new Map();
+
+  private static cleanStaleProviders() {
+    const now = Date.now();
+    for (const [id, timestamp] of this.providerTimestamps.entries()) {
+      if (now - timestamp > this.PROVIDER_CACHE_TTL) {
+        this.providers.delete(id);
+        this.providerTimestamps.delete(id);
+      }
+    }
+  }
 
   public static getProvider(type?: string): AIProviderInterface {
-    // 如果未指定类型，使用默认提供商
+    this.cleanStaleProviders();
     const providerType =
       type ||
       ConfigurationManager.getInstance().getConfig<string>("PROVIDER") ||
       AIProvider.OPENAI;
 
     let provider = this.providers.get(providerType);
-    console.log("provider", providerType);
+
     if (!provider) {
       switch (providerType.toLowerCase()) {
         case AIProvider.OPENAI:
@@ -29,6 +44,15 @@ export class AIProviderFactory {
         case AIProvider.VSCODE:
           provider = new VSCodeProvider();
           break;
+        case AIProvider.ZHIPU:
+          provider = new ZhipuAIProvider();
+          break;
+        case AIProvider.DASHSCOPE:
+          provider = new DashScopeProvider();
+          break;
+        case AIProvider.DOUBAO:
+          provider = new DoubaoProvider();
+          break;
         default:
           throw new Error(
             LocalizationManager.getInstance().format(
@@ -38,14 +62,21 @@ export class AIProviderFactory {
           );
       }
       this.providers.set(providerType, provider);
+      this.providerTimestamps.set(providerType, Date.now());
     }
 
     return provider;
   }
 
   public static getAllProviders(): AIProviderInterface[] {
-    // 返回所有可用的 AI Provider 实例
-    return [new OpenAIProvider(), new OllamaProvider(), new VSCodeProvider()];
+    return [
+      new OpenAIProvider(),
+      new OllamaProvider(),
+      new VSCodeProvider(),
+      new ZhipuAIProvider(),
+      new DashScopeProvider(),
+      new DoubaoProvider(),
+    ];
   }
 
   public static reinitializeProvider(providerId: string): void {

src/ai/providers/BaseOpenAIProvider.ts

@@ -0,0 +1,100 @@
+import OpenAI from "openai";
+import { ChatCompletionMessageParam } from "openai/resources";
+import { AIProvider, AIRequestParams, AIResponse, AIModel } from "../types";
+import { generateWithRetry, getSystemPrompt } from "../utils/generateHelper";
+
+export interface OpenAIProviderConfig {
+  apiKey: string;
+  baseURL?: string;
+  apiVersion?: string;
+  providerId: string;
+  providerName: string;
+  defaultModel?: string;
+  models: AIModel[];
+}
+
+export abstract class BaseOpenAIProvider implements AIProvider {
+  protected openai: OpenAI;
+  protected config: OpenAIProviderConfig;
+  protected provider: { id: string; name: string };
+
+  constructor(config: OpenAIProviderConfig) {
+    this.config = config;
+    this.provider = {
+      id: config.providerId,
+      name: config.providerName,
+    };
+    this.openai = this.createClient();
+  }
+
+  protected createClient(): OpenAI {
+    const config: any = {
+      apiKey: this.config.apiKey,
+    };
+
+    if (this.config.baseURL) {
+      config.baseURL = this.config.baseURL;
+      if (this.config.apiKey) {
+        // config.defaultQuery = { "api-version": this.config.apiVersion };
+        config.defaultHeaders = { "api-key": this.config.apiKey };
+      }
+    }
+    console.log("config", config);
+
+    return new OpenAI(config);
+  }
+
+  async generateResponse(params: AIRequestParams): Promise<AIResponse> {
+    return generateWithRetry(
+      params,
+      async (truncatedDiff) => {
+        const messages: ChatCompletionMessageParam[] = [
+          {
+            role: "system",
+            content: getSystemPrompt(params),
+          },
+          {
+            role: "user",
+            content: truncatedDiff,
+          },
+        ];
+
+        const completion = await this.openai.chat.completions.create({
+          model:
+            (params.model && params.model.id) ||
+            this.config.defaultModel ||
+            "gpt-3.5-turbo",
+          messages,
+        });
+
+        return {
+          content: completion.choices[0]?.message?.content || "",
+          usage: {
+            promptTokens: completion.usage?.prompt_tokens,
+            completionTokens: completion.usage?.completion_tokens,
+            totalTokens: completion.usage?.total_tokens,
+          },
+        };
+      },
+      {
+        initialMaxLength: params.model?.maxTokens?.input || 16385,
+        provider: this.getId(),
+      }
+    );
+  }
+
+  async getModels(): Promise<AIModel[]> {
+    return Promise.resolve(this.config.models);
+  }
+
+  getName(): string {
+    return this.provider.name;
+  }
+
+  getId(): string {
+    return this.provider.id;
+  }
+
+  abstract isAvailable(): Promise<boolean>;
+  abstract refreshModels(): Promise<string[]>;
+}

src/ai/providers/DashScopeProvider.ts

@@ -0,0 +1,145 @@
+import { BaseOpenAIProvider } from "./BaseOpenAIProvider";
+import { ConfigurationManager } from "../../config/ConfigurationManager";
+import { AIModel } from "../types";
+
+const dashscopeModels: AIModel[] = [
+  {
+    id: "qwen-max",
+    name: "Qwen Max (稳定版) - 旗舰模型: 强大的理解和生成能力",
+    maxTokens: { input: 30720, output: 8192 },
+    provider: { id: "dashscope", name: "DashScope" },
+    capabilities: {
+      streaming: true,
+      functionCalling: true,
+    },
+    cost: {
+      input: 0.02,
+      output: 0.06,
+    },
+  },
+  {
+    id: "qwen-max-latest",
+    name: "Qwen Max (最新版) - 旗舰实验版: 最新的模型改进和优化",
+    maxTokens: { input: 30720, output: 8192 },
+    provider: { id: "dashscope", name: "DashScope" },
+    capabilities: {
+      streaming: true,
+      functionCalling: true,
+    },
+    cost: {
+      input: 0.02,
+      output: 0.06,
+    },
+  },
+  {
+    id: "qwen-plus",
+    name: "Qwen Plus (稳定版) - 增强版: 性能与成本的最佳平衡",
+    maxTokens: { input: 129024, output: 8192 },
+    provider: { id: "dashscope", name: "DashScope" },
+    capabilities: {
+      streaming: true,
+      functionCalling: true,
+    },
+    cost: {
+      input: 0.0008,
+      output: 0.002,
+    },
+  },
+  {
+    id: "qwen-plus-latest",
+    name: "Qwen Plus (最新版) - 增强实验版: 新特性和优化的测试版本",
+    maxTokens: { input: 129024, output: 8192 },
+    provider: { id: "dashscope", name: "DashScope" },
+    capabilities: {
+      streaming: true,
+      functionCalling: true,
+    },
+    cost: {
+      input: 0.0008,
+      output: 0.002,
+    },
+  },
+  {
+    id: "qwen-turbo",
+    name: "Qwen Turbo (稳定版) - 快速版: 高性价比的日常对话模型",
+    maxTokens: { input: 129024, output: 8192 },
+    provider: { id: "dashscope", name: "DashScope" },
+    default: true,
+    capabilities: {
+      streaming: true,
+      functionCalling: true,
+    },
+    cost: {
+      input: 0.0003,
+      output: 0.0006,
+    },
+  },
+  {
+    id: "qwen-turbo-latest",
+    name: "Qwen Turbo (最新版) - 快速实验版: 优化推理速度的最新版本",
+    maxTokens: { input: 1000000, output: 8192 },
+    provider: { id: "dashscope", name: "DashScope" },
+    capabilities: {
+      streaming: true,
+      functionCalling: true,
+    },
+    cost: {
+      input: 0.0003,
+      output: 0.0006,
+    },
+  },
+  {
+    id: "qwen-coder-turbo",
+    name: "Qwen Coder Turbo (稳定版) - 编程专用: 代码生成和分析的专业模型",
+    maxTokens: { input: 129024, output: 8192 },
+    provider: { id: "dashscope", name: "DashScope" },
+    capabilities: {
+      streaming: true,
+      functionCalling: true,
+    },
+    cost: {
+      input: 0.002,
+      output: 0.006,
+    },
+  },
+  {
+    id: "qwen-coder-turbo-latest",
+    name: "Qwen Coder Turbo (最新版) - 编程实验版: 最新的代码辅助功能",
+    maxTokens: { input: 129024, output: 8192 },
+    provider: { id: "dashscope", name: "DashScope" },
+    capabilities: {
+      streaming: true,
+      functionCalling: true,
+    },
+    cost: {
+      input: 0.002,
+      output: 0.006,
+    },
+  },
+];
+
+export class DashScopeProvider extends BaseOpenAIProvider {
+  constructor() {
+    const configManager = ConfigurationManager.getInstance();
+    super({
+      apiKey: configManager.getConfig<string>("DASHSCOPE_API_KEY", false),
+      baseURL: "https://api.dashscope.com/v1/services/chat/completions",
+      providerId: "dashscope",
+      providerName: "DashScope",
+      models: dashscopeModels,
+      defaultModel: "qwen-turbo",
+    });
+  }
+
+  async isAvailable(): Promise<boolean> {
+    try {
+      return !!this.config.apiKey;
+    } catch {
+      return false;
+    }
+  }
+
+  async refreshModels(): Promise<string[]> {
+    return Promise.resolve(dashscopeModels.map((m) => m.id));
+  }
+}