@scenemesh/entity-engine-aimodule 1.0.0

package/dist/chunk-TDRKKUNT.mjs

@@ -0,0 +1,357 @@
+// src/core/ai-embeddings.ts
+import { EventEmitter } from "events";
+import {
+  embed,
+  embedMany,
+  cosineSimilarity
+} from "ai";
+var SupportedEmbeddingModels = {
+  // Qwen
+  "qwen:text-embedding-v3": {
+    provider: "qwen",
+    model: "text-embedding-v3",
+    dimensions: 1024,
+    description: "Qwen latest embedding model with 1024 dimensions"
+  }
+};
+var AIEmbeddingsIntegration = class extends EventEmitter {
+  constructor() {
+    super();
+    this.initialized = false;
+    this.requestCounter = 0;
+    this.embeddingCache = /* @__PURE__ */ new Map();
+    this.usageStats = /* @__PURE__ */ new Map();
+  }
+  /**
+   * 初始化
+   */
+  async initialize() {
+    if (this.initialized) {
+      return;
+    }
+    try {
+      this.emit("ai_embeddings:initializing");
+      this.initialized = true;
+      this.emit("ai_embeddings:initialized");
+    } catch (error) {
+      this.emit("ai_embeddings:initialization_failed", { error });
+      throw error;
+    }
+  }
+  /**
+   * 生成请求ID
+   */
+  generateRequestId() {
+    return `ai-embeddings-${Date.now()}-${++this.requestCounter}`;
+  }
+  // Single value embedding methods
+  /**
+   * Embed single value - embed() API
+   */
+  async embed(options) {
+    return this.embedSingleValue(options);
+  }
+  /**
+   * Embed single value - complete embed() implementation
+   */
+  async embedSingleValue(options) {
+    if (!this.initialized) {
+      throw new Error("AIEmbeddingsIntegration not initialized. Call initialize() first.");
+    }
+    const requestId = this.generateRequestId();
+    const startTime = Date.now();
+    try {
+      this.emit("ai_embeddings:embed_started", {
+        requestId,
+        model: typeof options.model === "string" ? options.model : "unknown",
+        valueLength: options.value.length,
+        hasProviderOptions: !!options.providerOptions,
+        maxRetries: options.maxRetries || 2
+      });
+      const cacheKey = this.generateCacheKey(options.model, options.value, options.providerOptions);
+      if (this.embeddingCache.has(cacheKey)) {
+        const cached = this.embeddingCache.get(cacheKey);
+        this.emit("ai_embeddings:embed_cache_hit", { requestId, cacheKey });
+        return cached;
+      }
+      const result = await embed({
+        model: options.model,
+        value: options.value,
+        providerOptions: options.providerOptions,
+        maxRetries: options.maxRetries,
+        abortSignal: options.abortSignal,
+        headers: options.headers
+      });
+      this.embeddingCache.set(cacheKey, result);
+      this.recordUsage(options.model, 1, result.usage.tokens || 0);
+      this.emit("ai_embeddings:embed_completed", {
+        requestId,
+        latency: Date.now() - startTime,
+        embeddingDimensions: result.embedding.length,
+        usage: result.usage,
+        cached: false
+      });
+      return result;
+    } catch (error) {
+      this.emit("ai_embeddings:embed_failed", {
+        requestId,
+        error: error.message,
+        latency: Date.now() - startTime,
+        errorType: error.constructor.name
+      });
+      throw error;
+    }
+  }
+  // Batch embedding methods
+  /**
+   * Embed multiple values - embedMany() API
+   */
+  async embedMany(options) {
+    return this.embedManyValues(options);
+  }
+  /**
+   * Embed multiple values - complete embedMany() implementation  
+   */
+  async embedManyValues(options) {
+    if (!this.initialized) {
+      throw new Error("AIEmbeddingsIntegration not initialized. Call initialize() first.");
+    }
+    const requestId = this.generateRequestId();
+    const startTime = Date.now();
+    try {
+      this.emit("ai_embeddings:embed_many_started", {
+        requestId,
+        model: typeof options.model === "string" ? options.model : "unknown",
+        valuesCount: options.values.length,
+        maxParallelCalls: options.maxParallelCalls,
+        hasProviderOptions: !!options.providerOptions,
+        maxRetries: options.maxRetries || 2
+      });
+      const result = await embedMany({
+        model: options.model,
+        values: options.values,
+        maxParallelCalls: options.maxParallelCalls,
+        providerOptions: options.providerOptions,
+        maxRetries: options.maxRetries,
+        abortSignal: options.abortSignal,
+        headers: options.headers
+      });
+      this.recordUsage(options.model, options.values.length, result.usage.tokens || 0);
+      this.emit("ai_embeddings:embed_many_completed", {
+        requestId,
+        latency: Date.now() - startTime,
+        valuesCount: options.values.length,
+        embeddingsCount: result.embeddings.length,
+        embeddingDimensions: result.embeddings[0]?.length || 0,
+        usage: result.usage
+      });
+      return result;
+    } catch (error) {
+      this.emit("ai_embeddings:embed_many_failed", {
+        requestId,
+        error: error.message,
+        latency: Date.now() - startTime,
+        errorType: error.constructor.name,
+        valuesCount: options.values.length
+      });
+      throw error;
+    }
+  }
+  // Similarity calculation methods
+  /**
+   * Cosine similarity calculation - cosineSimilarity() implementation
+   */
+  calculateCosineSimilarity(vector1, vector2) {
+    if (vector1.length !== vector2.length) {
+      throw new Error("Vectors must have the same dimensions");
+    }
+    const similarity = cosineSimilarity(vector1, vector2);
+    this.emit("ai_embeddings:similarity_calculated", {
+      method: "cosine",
+      similarity,
+      vector1Dimensions: vector1.length,
+      vector2Dimensions: vector2.length
+    });
+    return similarity;
+  }
+  /**
+   * 批量相似度计算
+   */
+  calculateSimilarities(targetVector, vectors, method = "cosine") {
+    return vectors.map((vector) => ({
+      similarity: method === "cosine" ? this.calculateCosineSimilarity(targetVector, vector) : method === "euclidean" ? this.calculateEuclideanSimilarity(targetVector, vector) : this.calculateDotProduct(targetVector, vector),
+      vector1: targetVector,
+      vector2: vector,
+      method
+    }));
+  }
+  /**
+   * 欧几里得距离相似度
+   */
+  calculateEuclideanSimilarity(vector1, vector2) {
+    if (vector1.length !== vector2.length) {
+      throw new Error("Vectors must have the same dimensions");
+    }
+    const sumSquared = vector1.reduce((sum, val, i) => {
+      const diff = val - vector2[i];
+      return sum + diff * diff;
+    }, 0);
+    return 1 / (1 + Math.sqrt(sumSquared));
+  }
+  /**
+   * 点积相似度
+   */
+  calculateDotProduct(vector1, vector2) {
+    if (vector1.length !== vector2.length) {
+      throw new Error("Vectors must have the same dimensions");
+    }
+    return vector1.reduce((sum, val, i) => sum + val * vector2[i], 0);
+  }
+  // Similarity search and ranking
+  /**
+   * Find most similar embeddings
+   */
+  findMostSimilar(queryVector, candidateVectors, topK = 5, method = "cosine") {
+    const similarities = this.calculateSimilarities(queryVector, candidateVectors, method);
+    return similarities.map((result, index) => ({
+      index,
+      similarity: result.similarity,
+      vector: result.vector2
+    })).sort((a, b) => b.similarity - a.similarity).slice(0, topK);
+  }
+  /**
+   * 文本相似性搜索
+   */
+  async searchSimilarTexts(queryText, candidateTexts, model, topK = 5, options) {
+    const queryResult = await this.embedSingleValue({
+      model,
+      value: queryText,
+      ...options
+    });
+    const candidatesResult = await this.embedManyValues({
+      model,
+      values: candidateTexts,
+      ...options
+    });
+    const similarities = this.findMostSimilar(
+      queryResult.embedding,
+      candidatesResult.embeddings,
+      topK
+    );
+    return similarities.map((result) => ({
+      index: result.index,
+      text: candidateTexts[result.index],
+      similarity: result.similarity
+    }));
+  }
+  // Utility methods and cache management
+  /**
+   * Generate cache key
+   */
+  generateCacheKey(model, value, providerOptions) {
+    const modelKey = typeof model === "string" ? model : "unknown";
+    const optionsKey = providerOptions ? JSON.stringify(providerOptions) : "";
+    return `${modelKey}:${value}:${optionsKey}`;
+  }
+  /**
+   * Record usage statistics
+   */
+  recordUsage(model, embeddingCount, tokens) {
+    const modelKey = typeof model === "string" ? model : "unknown";
+    const existing = this.usageStats.get(modelKey);
+    if (existing) {
+      existing.totalEmbeddings += embeddingCount;
+      existing.totalTokens += tokens;
+      existing.lastUsed = /* @__PURE__ */ new Date();
+    } else {
+      this.usageStats.set(modelKey, {
+        totalEmbeddings: embeddingCount,
+        totalTokens: tokens,
+        lastUsed: /* @__PURE__ */ new Date()
+      });
+    }
+  }
+  /**
+   * 获取支持的模型信息
+   */
+  getSupportedModels() {
+    return Object.values(SupportedEmbeddingModels);
+  }
+  /**
+   * 根据提供商筛选模型
+   */
+  getModelsByProvider(provider) {
+    return Object.values(SupportedEmbeddingModels).filter((model) => model.provider === provider);
+  }
+  /**
+   * 获取使用统计
+   */
+  getUsageStats() {
+    const stats = {};
+    this.usageStats.forEach((stat, model) => {
+      stats[model] = { ...stat };
+    });
+    return stats;
+  }
+  /**
+   * 清理缓存
+   */
+  clearCache() {
+    this.embeddingCache.clear();
+    this.emit("ai_embeddings:cache_cleared");
+  }
+  /**
+   * 获取缓存统计
+   */
+  getCacheStats() {
+    return {
+      size: this.embeddingCache.size,
+      keys: Array.from(this.embeddingCache.keys())
+    };
+  }
+  /**
+   * 检查初始化状态
+   */
+  isInitialized() {
+    return this.initialized;
+  }
+  /**
+   * 获取统计信息
+   */
+  getStats() {
+    const totalUsage = Array.from(this.usageStats.values()).reduce(
+      (sum, stat) => ({
+        embeddings: sum.embeddings + stat.totalEmbeddings,
+        tokens: sum.tokens + stat.totalTokens
+      }),
+      { embeddings: 0, tokens: 0 }
+    );
+    return {
+      isInitialized: this.initialized,
+      requestCount: this.requestCounter,
+      cacheSize: this.embeddingCache.size,
+      totalUsage
+    };
+  }
+  /**
+   * 销毁实例
+   */
+  destroy() {
+    this.embeddingCache.clear();
+    this.usageStats.clear();
+    this.removeAllListeners();
+    this.initialized = false;
+    this.requestCounter = 0;
+  }
+};
+var ai_embeddings_default = AIEmbeddingsIntegration;
+
+export {
+  embed,
+  embedMany,
+  cosineSimilarity,
+  SupportedEmbeddingModels,
+  AIEmbeddingsIntegration,
+  ai_embeddings_default
+};
+//# sourceMappingURL=chunk-TDRKKUNT.mjs.map

package/dist/chunk-TDRKKUNT.mjs.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/core/ai-embeddings.ts"],"sourcesContent":["/**\n * AI Embeddings Complete Integration\n * \n * Complete embeddings functionality:\n * - embed: Single value embedding\n * - embedMany: Batch embedding  \n * - cosineSimilarity: Similarity calculation\n * - All configuration parameters supported\n * - Provider-specific configurations\n */\n\nimport { EventEmitter } from 'events';\nimport {\n  embed,\n  embedMany,\n  cosineSimilarity,\n  type EmbeddingModel,\n  type LanguageModelUsage,\n  type LanguageModelRequestMetadata,\n  type LanguageModelResponseMetadata\n} from 'ai';\n\n// Type Definitions\n\n/**\n * Complete embedding model configuration\n */\nexport interface EmbeddingModelConfig {\n  // 基础模型\n  model: EmbeddingModel;\n  \n  // 提供商特定选项 - 官方文档支持\n  providerOptions?: Record<string, any>; // 如 { openai: { dimensions: 512 } }\n  \n  // 控制参数\n  maxRetries?: number; // 最大重试次数，默认 2\n  abortSignal?: AbortSignal; // 取消信号\n  headers?: Record<string, string>; // 自定义 HTTP 头\n}\n\n/**\n * 批量嵌入配置 - embedMany 专用参数\n */\nexport interface EmbedManyConfig extends EmbeddingModelConfig {\n  maxParallelCalls?: number; // 并行请求数 - 官方文档功能\n}\n\n/**\n * 单值嵌入选项 - embed() 完整参数\n */\nexport interface EmbedOptions extends EmbeddingModelConfig {\n  value: string; // 要嵌入的文本\n}\n\n/**\n * 批量嵌入选项 - embedMany() 完整参数  \n */\nexport interface EmbedManyOptions extends EmbedManyConfig {\n  values: string[]; // 要嵌入的文本数组\n}\n\n/**\n * 嵌入结果类型 - 官方规范\n */\nexport interface EmbedResult {\n  embedding: number[]; // 嵌入向量\n  usage: LanguageModelUsage; // 令牌使用情况\n  request?: LanguageModelRequestMetadata; // 请求元数据\n  response?: LanguageModelResponseMetadata; // 响应元数据\n}\n\n/**\n * 批量嵌入结果类型 - 官方规范\n */\nexport interface EmbedManyResult {\n  embeddings: number[][]; // 嵌入向量数组，与输入顺序一致\n  usage: LanguageModelUsage; // 令牌使用情况\n  request?: LanguageModelRequestMetadata; // 请求元数据\n  response?: LanguageModelResponseMetadata; // 响应元数据\n}\n\n/**\n * 相似度计算结果\n */\nexport interface SimilarityResult {\n  similarity: number;\n  vector1: number[];\n  vector2: number[];\n  method: 'cosine' | 'euclidean' | 'dot';\n}\n\n/**\n * 嵌入提供商信息 - 基于官方文档表格\n */\nexport interface EmbeddingProviderInfo {\n  provider: string;\n  model: string;\n  dimensions: number;\n  description?: string;\n}\n\n// Predefined provider model configurations\n\n/**\n * Supported embedding models\n */\nexport const SupportedEmbeddingModels: Record<string, EmbeddingProviderInfo> = {\n  // Qwen\n  'qwen:text-embedding-v3': {\n    provider: 'qwen',\n    model: 'text-embedding-v3',\n    dimensions: 1024,\n    description: 'Qwen latest embedding model with 1024 dimensions'\n  }\n};\n\n// Core Embeddings Integration Class\n\n/**\n * Complete embeddings integration class\n */\nexport class AIEmbeddingsIntegration extends EventEmitter {\n  private initialized: boolean = false;\n  private requestCounter: number = 0;\n  private embeddingCache: Map<string, EmbedResult> = new Map();\n  private usageStats: Map<string, { totalEmbeddings: number; totalTokens: number; lastUsed: Date }> = new Map();\n\n  constructor() {\n    super();\n  }\n\n  /**\n   * 初始化\n   */\n  async initialize(): Promise<void> {\n    if (this.initialized) {\n      return;\n    }\n\n    try {\n      this.emit('ai_embeddings:initializing');\n      this.initialized = true;\n      this.emit('ai_embeddings:initialized');\n    } catch (error) {\n      this.emit('ai_embeddings:initialization_failed', { error });\n      throw error;\n    }\n  }\n\n  /**\n   * 生成请求ID\n   */\n  private generateRequestId(): string {\n    return `ai-embeddings-${Date.now()}-${++this.requestCounter}`;\n  }\n\n  // Single value embedding methods\n\n  /**\n   * Embed single value - embed() API\n   */\n  async embed(options: EmbedOptions): Promise<EmbedResult> {\n    return this.embedSingleValue(options);\n  }\n\n  /**\n   * Embed single value - complete embed() implementation\n   */\n  async embedSingleValue(options: EmbedOptions): Promise<EmbedResult> {\n    if (!this.initialized) {\n      throw new Error('AIEmbeddingsIntegration not initialized. Call initialize() first.');\n    }\n\n    const requestId = this.generateRequestId();\n    const startTime = Date.now();\n\n    try {\n      this.emit('ai_embeddings:embed_started', {\n        requestId,\n        model: typeof options.model === 'string' ? options.model : 'unknown',\n        valueLength: options.value.length,\n        hasProviderOptions: !!options.providerOptions,\n        maxRetries: options.maxRetries || 2\n      });\n\n      // 检查缓存\n      const cacheKey = this.generateCacheKey(options.model, options.value, options.providerOptions);\n      if (this.embeddingCache.has(cacheKey)) {\n        const cached = this.embeddingCache.get(cacheKey)!;\n        this.emit('ai_embeddings:embed_cache_hit', { requestId, cacheKey });\n        return cached;\n      }\n\n      // 调用官方 embed 函数\n      const result = await embed({\n        model: options.model,\n        value: options.value,\n        providerOptions: options.providerOptions,\n        maxRetries: options.maxRetries,\n        abortSignal: options.abortSignal,\n        headers: options.headers\n      });\n\n      // 缓存结果  \n      this.embeddingCache.set(cacheKey, result as any);\n\n      // 记录统计\n      this.recordUsage(options.model, 1, result.usage.tokens || 0);\n\n      this.emit('ai_embeddings:embed_completed', {\n        requestId,\n        latency: Date.now() - startTime,\n        embeddingDimensions: result.embedding.length,\n        usage: result.usage,\n        cached: false\n      });\n\n      return result as any;\n\n    } catch (error: any) {\n      this.emit('ai_embeddings:embed_failed', {\n        requestId,\n        error: error.message,\n        latency: Date.now() - startTime,\n        errorType: error.constructor.name\n      });\n\n      throw error;\n    }\n  }\n\n  // Batch embedding methods\n\n  /**\n   * Embed multiple values - embedMany() API\n   */\n  async embedMany(options: EmbedManyOptions): Promise<EmbedManyResult> {\n    return this.embedManyValues(options);\n  }\n\n  /**\n   * Embed multiple values - complete embedMany() implementation  \n   */\n  async embedManyValues(options: EmbedManyOptions): Promise<EmbedManyResult> {\n    if (!this.initialized) {\n      throw new Error('AIEmbeddingsIntegration not initialized. Call initialize() first.');\n    }\n\n    const requestId = this.generateRequestId();\n    const startTime = Date.now();\n\n    try {\n      this.emit('ai_embeddings:embed_many_started', {\n        requestId,\n        model: typeof options.model === 'string' ? options.model : 'unknown',\n        valuesCount: options.values.length,\n        maxParallelCalls: options.maxParallelCalls,\n        hasProviderOptions: !!options.providerOptions,\n        maxRetries: options.maxRetries || 2\n      });\n\n      // 调用官方 embedMany 函数\n      const result = await embedMany({\n        model: options.model,\n        values: options.values,\n        maxParallelCalls: options.maxParallelCalls,\n        providerOptions: options.providerOptions,\n        maxRetries: options.maxRetries,\n        abortSignal: options.abortSignal,\n        headers: options.headers\n      });\n\n      // 记录统计\n      this.recordUsage(options.model, options.values.length, result.usage.tokens || 0);\n\n      this.emit('ai_embeddings:embed_many_completed', {\n        requestId,\n        latency: Date.now() - startTime,\n        valuesCount: options.values.length,\n        embeddingsCount: result.embeddings.length,\n        embeddingDimensions: result.embeddings[0]?.length || 0,\n        usage: result.usage\n      });\n\n      return result as any;\n\n    } catch (error: any) {\n      this.emit('ai_embeddings:embed_many_failed', {\n        requestId,\n        error: error.message,\n        latency: Date.now() - startTime,\n        errorType: error.constructor.name,\n        valuesCount: options.values.length\n      });\n\n      throw error;\n    }\n  }\n\n  // Similarity calculation methods\n\n  /**\n   * Cosine similarity calculation - cosineSimilarity() implementation\n   */\n  calculateCosineSimilarity(vector1: number[], vector2: number[]): number {\n    if (vector1.length !== vector2.length) {\n      throw new Error('Vectors must have the same dimensions');\n    }\n\n    const similarity = cosineSimilarity(vector1, vector2);\n    \n    this.emit('ai_embeddings:similarity_calculated', {\n      method: 'cosine',\n      similarity,\n      vector1Dimensions: vector1.length,\n      vector2Dimensions: vector2.length\n    });\n\n    return similarity;\n  }\n\n  /**\n   * 批量相似度计算\n   */\n  calculateSimilarities(\n    targetVector: number[],\n    vectors: number[][],\n    method: 'cosine' | 'euclidean' | 'dot' = 'cosine'\n  ): SimilarityResult[] {\n    return vectors.map(vector => ({\n      similarity: method === 'cosine' \n        ? this.calculateCosineSimilarity(targetVector, vector)\n        : method === 'euclidean'\n        ? this.calculateEuclideanSimilarity(targetVector, vector)  \n        : this.calculateDotProduct(targetVector, vector),\n      vector1: targetVector,\n      vector2: vector,\n      method\n    }));\n  }\n\n  /**\n   * 欧几里得距离相似度\n   */\n  private calculateEuclideanSimilarity(vector1: number[], vector2: number[]): number {\n    if (vector1.length !== vector2.length) {\n      throw new Error('Vectors must have the same dimensions');\n    }\n\n    const sumSquared = vector1.reduce((sum, val, i) => {\n      const diff = val - vector2[i];\n      return sum + diff * diff;\n    }, 0);\n\n    // 转换为相似度（距离越小，相似度越高）\n    return 1 / (1 + Math.sqrt(sumSquared));\n  }\n\n  /**\n   * 点积相似度\n   */\n  private calculateDotProduct(vector1: number[], vector2: number[]): number {\n    if (vector1.length !== vector2.length) {\n      throw new Error('Vectors must have the same dimensions');\n    }\n\n    return vector1.reduce((sum, val, i) => sum + val * vector2[i], 0);\n  }\n\n  // Similarity search and ranking\n\n  /**\n   * Find most similar embeddings\n   */\n  findMostSimilar(\n    queryVector: number[],\n    candidateVectors: number[][],\n    topK: number = 5,\n    method: 'cosine' | 'euclidean' | 'dot' = 'cosine'\n  ): Array<{ index: number; similarity: number; vector: number[] }> {\n    const similarities = this.calculateSimilarities(queryVector, candidateVectors, method);\n    \n    return similarities\n      .map((result, index) => ({\n        index,\n        similarity: result.similarity,\n        vector: result.vector2\n      }))\n      .sort((a, b) => b.similarity - a.similarity) // 按相似度降序排列\n      .slice(0, topK);\n  }\n\n  /**\n   * 文本相似性搜索\n   */\n  async searchSimilarTexts(\n    queryText: string,\n    candidateTexts: string[],\n    model: EmbeddingModel,\n    topK: number = 5,\n    options?: Partial<EmbeddingModelConfig>\n  ): Promise<Array<{ index: number; text: string; similarity: number }>> {\n    // 嵌入查询文本和候选文本\n    const queryResult = await this.embedSingleValue({\n      model,\n      value: queryText,\n      ...options\n    });\n\n    const candidatesResult = await this.embedManyValues({\n      model,\n      values: candidateTexts,\n      ...options\n    });\n\n    // 计算相似度\n    const similarities = this.findMostSimilar(\n      queryResult.embedding,\n      candidatesResult.embeddings,\n      topK\n    );\n\n    return similarities.map(result => ({\n      index: result.index,\n      text: candidateTexts[result.index],\n      similarity: result.similarity\n    }));\n  }\n\n  // Utility methods and cache management\n\n  /**\n   * Generate cache key\n   */\n  private generateCacheKey(model: EmbeddingModel, value: string, providerOptions?: any): string {\n    const modelKey = typeof model === 'string' ? model : 'unknown';\n    const optionsKey = providerOptions ? JSON.stringify(providerOptions) : '';\n    return `${modelKey}:${value}:${optionsKey}`;\n  }\n\n  /**\n   * Record usage statistics\n   */\n  private recordUsage(model: EmbeddingModel, embeddingCount: number, tokens: number): void {\n    const modelKey = typeof model === 'string' ? model : 'unknown';\n    \n    const existing = this.usageStats.get(modelKey);\n    if (existing) {\n      existing.totalEmbeddings += embeddingCount;\n      existing.totalTokens += tokens;\n      existing.lastUsed = new Date();\n    } else {\n      this.usageStats.set(modelKey, {\n        totalEmbeddings: embeddingCount,\n        totalTokens: tokens,\n        lastUsed: new Date()\n      });\n    }\n  }\n\n  /**\n   * 获取支持的模型信息\n   */\n  getSupportedModels(): EmbeddingProviderInfo[] {\n    return Object.values(SupportedEmbeddingModels);\n  }\n\n  /**\n   * 根据提供商筛选模型\n   */\n  getModelsByProvider(provider: string): EmbeddingProviderInfo[] {\n    return Object.values(SupportedEmbeddingModels)\n      .filter(model => model.provider === provider);\n  }\n\n  /**\n   * 获取使用统计\n   */\n  getUsageStats(): Record<string, { totalEmbeddings: number; totalTokens: number; lastUsed: Date }> {\n    const stats: Record<string, { totalEmbeddings: number; totalTokens: number; lastUsed: Date }> = {};\n    this.usageStats.forEach((stat, model) => {\n      stats[model] = { ...stat };\n    });\n    return stats;\n  }\n\n  /**\n   * 清理缓存\n   */\n  clearCache(): void {\n    this.embeddingCache.clear();\n    this.emit('ai_embeddings:cache_cleared');\n  }\n\n  /**\n   * 获取缓存统计\n   */\n  getCacheStats(): { size: number; keys: string[] } {\n    return {\n      size: this.embeddingCache.size,\n      keys: Array.from(this.embeddingCache.keys())\n    };\n  }\n\n  /**\n   * 检查初始化状态\n   */\n  isInitialized(): boolean {\n    return this.initialized;\n  }\n\n  /**\n   * 获取统计信息\n   */\n  getStats(): {\n    isInitialized: boolean;\n    requestCount: number;\n    cacheSize: number;\n    totalUsage: { embeddings: number; tokens: number };\n  } {\n    const totalUsage = Array.from(this.usageStats.values()).reduce(\n      (sum, stat) => ({\n        embeddings: sum.embeddings + stat.totalEmbeddings,\n        tokens: sum.tokens + stat.totalTokens\n      }),\n      { embeddings: 0, tokens: 0 }\n    );\n\n    return {\n      isInitialized: this.initialized,\n      requestCount: this.requestCounter,\n      cacheSize: this.embeddingCache.size,\n      totalUsage\n    };\n  }\n\n  /**\n   * 销毁实例\n   */\n  destroy(): void {\n    this.embeddingCache.clear();\n    this.usageStats.clear();\n    this.removeAllListeners();\n    this.initialized = false;\n    this.requestCounter = 0;\n  }\n}\n\n// Exports\n\nexport {\n  // Core functions\n  embed,\n  embedMany,\n  cosineSimilarity,\n  // Core types\n  type EmbeddingModel,\n  \n  type LanguageModelUsage,\n  type LanguageModelRequestMetadata,\n  type LanguageModelResponseMetadata\n};\n\n// Default export\nexport default AIEmbeddingsIntegration;"],"mappings":";AAWA,SAAS,oBAAoB;AAC7B;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,OAKK;AAsFA,IAAM,2BAAkE;AAAA;AAAA,EAE7E,0BAA0B;AAAA,IACxB,UAAU;AAAA,IACV,OAAO;AAAA,IACP,YAAY;AAAA,IACZ,aAAa;AAAA,EACf;AACF;AAOO,IAAM,0BAAN,cAAsC,aAAa;AAAA,EAMxD,cAAc;AACZ,UAAM;AANR,SAAQ,cAAuB;AAC/B,SAAQ,iBAAyB;AACjC,SAAQ,iBAA2C,oBAAI,IAAI;AAC3D,SAAQ,aAA4F,oBAAI,IAAI;AAAA,EAI5G;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,aAA4B;AAChC,QAAI,KAAK,aAAa;AACpB;AAAA,IACF;AAEA,QAAI;AACF,WAAK,KAAK,4BAA4B;AACtC,WAAK,cAAc;AACnB,WAAK,KAAK,2BAA2B;AAAA,IACvC,SAAS,OAAO;AACd,WAAK,KAAK,uCAAuC,EAAE,MAAM,CAAC;AAC1D,YAAM;AAAA,IACR;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKQ,oBAA4B;AAClC,WAAO,iBAAiB,KAAK,IAAI,CAAC,IAAI,EAAE,KAAK,cAAc;AAAA,EAC7D;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,MAAM,MAAM,SAA6C;AACvD,WAAO,KAAK,iBAAiB,OAAO;AAAA,EACtC;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,iBAAiB,SAA6C;AAClE,QAAI,CAAC,KAAK,aAAa;AACrB,YAAM,IAAI,MAAM,mEAAmE;AAAA,IACrF;AAEA,UAAM,YAAY,KAAK,kBAAkB;AACzC,UAAM,YAAY,KAAK,IAAI;AAE3B,QAAI;AACF,WAAK,KAAK,+BAA+B;AAAA,QACvC;AAAA,QACA,OAAO,OAAO,QAAQ,UAAU,WAAW,QAAQ,QAAQ;AAAA,QAC3D,aAAa,QAAQ,MAAM;AAAA,QAC3B,oBAAoB,CAAC,CAAC,QAAQ;AAAA,QAC9B,YAAY,QAAQ,cAAc;AAAA,MACpC,CAAC;AAGD,YAAM,WAAW,KAAK,iBAAiB,QAAQ,OAAO,QAAQ,OAAO,QAAQ,eAAe;AAC5F,UAAI,KAAK,eAAe,IAAI,QAAQ,GAAG;AACrC,cAAM,SAAS,KAAK,eAAe,IAAI,QAAQ;AAC/C,aAAK,KAAK,iCAAiC,EAAE,WAAW,SAAS,CAAC;AAClE,eAAO;AAAA,MACT;AAGA,YAAM,SAAS,MAAM,MAAM;AAAA,QACzB,OAAO,QAAQ;AAAA,QACf,OAAO,QAAQ;AAAA,QACf,iBAAiB,QAAQ;AAAA,QACzB,YAAY,QAAQ;AAAA,QACpB,aAAa,QAAQ;AAAA,QACrB,SAAS,QAAQ;AAAA,MACnB,CAAC;AAGD,WAAK,eAAe,IAAI,UAAU,MAAa;AAG/C,WAAK,YAAY,QAAQ,OAAO,GAAG,OAAO,MAAM,UAAU,CAAC;AAE3D,WAAK,KAAK,iCAAiC;AAAA,QACzC;AAAA,QACA,SAAS,KAAK,IAAI,IAAI;AAAA,QACtB,qBAAqB,OAAO,UAAU;AAAA,QACtC,OAAO,OAAO;AAAA,QACd,QAAQ;AAAA,MACV,CAAC;AAED,aAAO;AAAA,IAET,SAAS,OAAY;AACnB,WAAK,KAAK,8BAA8B;AAAA,QACtC;AAAA,QACA,OAAO,MAAM;AAAA,QACb,SAAS,KAAK,IAAI,IAAI;AAAA,QACtB,WAAW,MAAM,YAAY;AAAA,MAC/B,CAAC;AAED,YAAM;AAAA,IACR;AAAA,EACF;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,MAAM,UAAU,SAAqD;AACnE,WAAO,KAAK,gBAAgB,OAAO;AAAA,EACrC;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,gBAAgB,SAAqD;AACzE,QAAI,CAAC,KAAK,aAAa;AACrB,YAAM,IAAI,MAAM,mEAAmE;AAAA,IACrF;AAEA,UAAM,YAAY,KAAK,kBAAkB;AACzC,UAAM,YAAY,KAAK,IAAI;AAE3B,QAAI;AACF,WAAK,KAAK,oCAAoC;AAAA,QAC5C;AAAA,QACA,OAAO,OAAO,QAAQ,UAAU,WAAW,QAAQ,QAAQ;AAAA,QAC3D,aAAa,QAAQ,OAAO;AAAA,QAC5B,kBAAkB,QAAQ;AAAA,QAC1B,oBAAoB,CAAC,CAAC,QAAQ;AAAA,QAC9B,YAAY,QAAQ,cAAc;AAAA,MACpC,CAAC;AAGD,YAAM,SAAS,MAAM,UAAU;AAAA,QAC7B,OAAO,QAAQ;AAAA,QACf,QAAQ,QAAQ;AAAA,QAChB,kBAAkB,QAAQ;AAAA,QAC1B,iBAAiB,QAAQ;AAAA,QACzB,YAAY,QAAQ;AAAA,QACpB,aAAa,QAAQ;AAAA,QACrB,SAAS,QAAQ;AAAA,MACnB,CAAC;AAGD,WAAK,YAAY,QAAQ,OAAO,QAAQ,OAAO,QAAQ,OAAO,MAAM,UAAU,CAAC;AAE/E,WAAK,KAAK,sCAAsC;AAAA,QAC9C;AAAA,QACA,SAAS,KAAK,IAAI,IAAI;AAAA,QACtB,aAAa,QAAQ,OAAO;AAAA,QAC5B,iBAAiB,OAAO,WAAW;AAAA,QACnC,qBAAqB,OAAO,WAAW,CAAC,GAAG,UAAU;AAAA,QACrD,OAAO,OAAO;AAAA,MAChB,CAAC;AAED,aAAO;AAAA,IAET,SAAS,OAAY;AACnB,WAAK,KAAK,mCAAmC;AAAA,QAC3C;AAAA,QACA,OAAO,MAAM;AAAA,QACb,SAAS,KAAK,IAAI,IAAI;AAAA,QACtB,WAAW,MAAM,YAAY;AAAA,QAC7B,aAAa,QAAQ,OAAO;AAAA,MAC9B,CAAC;AAED,YAAM;AAAA,IACR;AAAA,EACF;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,0BAA0B,SAAmB,SAA2B;AACtE,QAAI,QAAQ,WAAW,QAAQ,QAAQ;AACrC,YAAM,IAAI,MAAM,uCAAuC;AAAA,IACzD;AAEA,UAAM,aAAa,iBAAiB,SAAS,OAAO;AAEpD,SAAK,KAAK,uCAAuC;AAAA,MAC/C,QAAQ;AAAA,MACR;AAAA,MACA,mBAAmB,QAAQ;AAAA,MAC3B,mBAAmB,QAAQ;AAAA,IAC7B,CAAC;AAED,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA,EAKA,sBACE,cACA,SACA,SAAyC,UACrB;AACpB,WAAO,QAAQ,IAAI,aAAW;AAAA,MAC5B,YAAY,WAAW,WACnB,KAAK,0BAA0B,cAAc,MAAM,IACnD,WAAW,cACX,KAAK,6BAA6B,cAAc,MAAM,IACtD,KAAK,oBAAoB,cAAc,MAAM;AAAA,MACjD,SAAS;AAAA,MACT,SAAS;AAAA,MACT;AAAA,IACF,EAAE;AAAA,EACJ;AAAA;AAAA;AAAA;AAAA,EAKQ,6BAA6B,SAAmB,SAA2B;AACjF,QAAI,QAAQ,WAAW,QAAQ,QAAQ;AACrC,YAAM,IAAI,MAAM,uCAAuC;AAAA,IACzD;AAEA,UAAM,aAAa,QAAQ,OAAO,CAAC,KAAK,KAAK,MAAM;AACjD,YAAM,OAAO,MAAM,QAAQ,CAAC;AAC5B,aAAO,MAAM,OAAO;AAAA,IACtB,GAAG,CAAC;AAGJ,WAAO,KAAK,IAAI,KAAK,KAAK,UAAU;AAAA,EACtC;AAAA;AAAA;AAAA;AAAA,EAKQ,oBAAoB,SAAmB,SAA2B;AACxE,QAAI,QAAQ,WAAW,QAAQ,QAAQ;AACrC,YAAM,IAAI,MAAM,uCAAuC;AAAA,IACzD;AAEA,WAAO,QAAQ,OAAO,CAAC,KAAK,KAAK,MAAM,MAAM,MAAM,QAAQ,CAAC,GAAG,CAAC;AAAA,EAClE;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA,gBACE,aACA,kBACA,OAAe,GACf,SAAyC,UACuB;AAChE,UAAM,eAAe,KAAK,sBAAsB,aAAa,kBAAkB,MAAM;AAErF,WAAO,aACJ,IAAI,CAAC,QAAQ,WAAW;AAAA,MACvB;AAAA,MACA,YAAY,OAAO;AAAA,MACnB,QAAQ,OAAO;AAAA,IACjB,EAAE,EACD,KAAK,CAAC,GAAG,MAAM,EAAE,aAAa,EAAE,UAAU,EAC1C,MAAM,GAAG,IAAI;AAAA,EAClB;AAAA;AAAA;AAAA;AAAA,EAKA,MAAM,mBACJ,WACA,gBACA,OACA,OAAe,GACf,SACqE;AAErE,UAAM,cAAc,MAAM,KAAK,iBAAiB;AAAA,MAC9C;AAAA,MACA,OAAO;AAAA,MACP,GAAG;AAAA,IACL,CAAC;AAED,UAAM,mBAAmB,MAAM,KAAK,gBAAgB;AAAA,MAClD;AAAA,MACA,QAAQ;AAAA,MACR,GAAG;AAAA,IACL,CAAC;AAGD,UAAM,eAAe,KAAK;AAAA,MACxB,YAAY;AAAA,MACZ,iBAAiB;AAAA,MACjB;AAAA,IACF;AAEA,WAAO,aAAa,IAAI,aAAW;AAAA,MACjC,OAAO,OAAO;AAAA,MACd,MAAM,eAAe,OAAO,KAAK;AAAA,MACjC,YAAY,OAAO;AAAA,IACrB,EAAE;AAAA,EACJ;AAAA;AAAA;AAAA;AAAA;AAAA,EAOQ,iBAAiB,OAAuB,OAAe,iBAA+B;AAC5F,UAAM,WAAW,OAAO,UAAU,WAAW,QAAQ;AACrD,UAAM,aAAa,kBAAkB,KAAK,UAAU,eAAe,IAAI;AACvE,WAAO,GAAG,QAAQ,IAAI,KAAK,IAAI,UAAU;AAAA,EAC3C;AAAA;AAAA;AAAA;AAAA,EAKQ,YAAY,OAAuB,gBAAwB,QAAsB;AACvF,UAAM,WAAW,OAAO,UAAU,WAAW,QAAQ;AAErD,UAAM,WAAW,KAAK,WAAW,IAAI,QAAQ;AAC7C,QAAI,UAAU;AACZ,eAAS,mBAAmB;AAC5B,eAAS,eAAe;AACxB,eAAS,WAAW,oBAAI,KAAK;AAAA,IAC/B,OAAO;AACL,WAAK,WAAW,IAAI,UAAU;AAAA,QAC5B,iBAAiB;AAAA,QACjB,aAAa;AAAA,QACb,UAAU,oBAAI,KAAK;AAAA,MACrB,CAAC;AAAA,IACH;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,qBAA8C;AAC5C,WAAO,OAAO,OAAO,wBAAwB;AAAA,EAC/C;AAAA;AAAA;AAAA;AAAA,EAKA,oBAAoB,UAA2C;AAC7D,WAAO,OAAO,OAAO,wBAAwB,EAC1C,OAAO,WAAS,MAAM,aAAa,QAAQ;AAAA,EAChD;AAAA;AAAA;AAAA;AAAA,EAKA,gBAAkG;AAChG,UAAM,QAA0F,CAAC;AACjG,SAAK,WAAW,QAAQ,CAAC,MAAM,UAAU;AACvC,YAAM,KAAK,IAAI,EAAE,GAAG,KAAK;AAAA,IAC3B,CAAC;AACD,WAAO;AAAA,EACT;AAAA;AAAA;AAAA;AAAA,EAKA,aAAmB;AACjB,SAAK,eAAe,MAAM;AAC1B,SAAK,KAAK,6BAA6B;AAAA,EACzC;AAAA;AAAA;AAAA;AAAA,EAKA,gBAAkD;AAChD,WAAO;AAAA,MACL,MAAM,KAAK,eAAe;AAAA,MAC1B,MAAM,MAAM,KAAK,KAAK,eAAe,KAAK,CAAC;AAAA,IAC7C;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,gBAAyB;AACvB,WAAO,KAAK;AAAA,EACd;AAAA;AAAA;AAAA;AAAA,EAKA,WAKE;AACA,UAAM,aAAa,MAAM,KAAK,KAAK,WAAW,OAAO,CAAC,EAAE;AAAA,MACtD,CAAC,KAAK,UAAU;AAAA,QACd,YAAY,IAAI,aAAa,KAAK;AAAA,QAClC,QAAQ,IAAI,SAAS,KAAK;AAAA,MAC5B;AAAA,MACA,EAAE,YAAY,GAAG,QAAQ,EAAE;AAAA,IAC7B;AAEA,WAAO;AAAA,MACL,eAAe,KAAK;AAAA,MACpB,cAAc,KAAK;AAAA,MACnB,WAAW,KAAK,eAAe;AAAA,MAC/B;AAAA,IACF;AAAA,EACF;AAAA;AAAA;AAAA;AAAA,EAKA,UAAgB;AACd,SAAK,eAAe,MAAM;AAC1B,SAAK,WAAW,MAAM;AACtB,SAAK,mBAAmB;AACxB,SAAK,cAAc;AACnB,SAAK,iBAAiB;AAAA,EACxB;AACF;AAkBA,IAAO,wBAAQ;","names":[]}