/**
 * EMBEDDINGS SUPPORT
 * Vector generation, similarity search, batching, caching
 */
export interface EmbeddingRequest {
    model: string;
    input: string | string[];
    encoding_format?: 'float' | 'base64';
    dimensions?: number;
}
export interface EmbeddingResponse {
    model: string;
    embeddings: Embedding[];
    usage: {
        prompt_tokens: number;
        total_tokens: number;
    };
}
export interface Embedding {
    index: number;
    embedding: number[];
    object: 'embedding';
}
export interface EmbeddingModel {
    provider: string;
    model: string;
    dimensions: number;
    max_input_tokens: number;
    cost_per_1k_tokens: number;
}
export declare class EmbeddingManager {
    private models;
    private cache;
    private batch_queue;
    private batch_timeout?;
    constructor();
    /**
     * Generate embeddings
     */
    generate(text: string | string[], model?: string): Promise<number[][]>;
    /**
     * Generate single embedding
     */
    private generateSingle;
    /**
     * Batch generate with automatic batching
     */
    generateBatch(text: string, model?: string): Promise<number[]>;
    /**
     * Calculate similarity
     */
    similarity(emb1: number[], emb2: number[]): number;
    /**
     * Find most similar
     */
    findMostSimilar(query: number[], candidates: number[][], top_k?: number): Array<{
        index: number;
        similarity: number;
    }>;
    /**
     * Get model info
     */
    getModel(model: string): EmbeddingModel | null;
    /**
     * List available models
     */
    listModels(): EmbeddingModel[];
    /**
     * Clear cache
     */
    clearCache(): void;
    /**
     * Get cache stats
     */
    getCacheStats(): {
        size: number;
        hit_rate: number;
    };
    private processBatch;
    private normalize;
    private getCacheKey;
    private initializeModels;
}
export interface VectorDocument {
    id: string;
    text: string;
    embedding: number[];
    metadata?: Record<string, unknown>;
}
export declare class VectorStore {
    private documents;
    private embedding_manager;
    constructor(embedding_manager?: EmbeddingManager);
    /**
     * Add document
     */
    add(id: string, text: string, metadata?: Record<string, unknown>, model?: string): Promise<void>;
    /**
     * Add many documents
     */
    addMany(documents: Array<{
        id: string;
        text: string;
        metadata?: Record<string, unknown>;
    }>, model?: string): Promise<void>;
    /**
     * Search by query
     */
    search(query: string, top_k?: number, model?: string, filter?: (doc: VectorDocument) => boolean): Promise<Array<{
        document: VectorDocument;
        similarity: number;
    }>>;
    /**
     * Search by vector
     */
    searchByVector(query_embedding: number[], top_k?: number, filter?: (doc: VectorDocument) => boolean): Array<{
        document: VectorDocument;
        similarity: number;
    }>;
    /**
     * Get document
     */
    get(id: string): VectorDocument | null;
    /**
     * Delete document
     */
    delete(id: string): boolean;
    /**
     * Get document count
     */
    count(): number;
    /**
     * Clear all documents
     */
    clear(): void;
}
export declare class EmbeddingBatchProcessor {
    private embedding_manager;
    constructor(embedding_manager?: EmbeddingManager);
    /**
     * Process texts in batches
     */
    processBatch(texts: string[], model?: string, batch_size?: number): Promise<number[][]>;
    /**
     * Process with progress callback
     */
    processWithProgress(texts: string[], on_progress: (completed: number, total: number) => void, model?: string, batch_size?: number): Promise<number[][]>;
}
//# sourceMappingURL=embedding-manager.d.ts.map