/**
 * Agentic QE v3 - LRU Cache for LLM Responses
 * ADR-011: LLM Provider System for Quality Engineering
 *
 * Provides efficient caching of LLM responses to reduce costs and latency
 * for repeated queries. Uses LRU (Least Recently Used) eviction policy.
 */
import { CacheEntry, LLMCacheConfig, LLMCacheStats, LLMResponse, EmbeddingResponse, CompletionResponse } from './interfaces';
/**
 * Default cache configuration
 */
export declare const DEFAULT_CACHE_CONFIG: LLMCacheConfig;
/**
 * Type for cacheable response types
 */
export type CacheableResponse = LLMResponse | EmbeddingResponse | CompletionResponse;
/**
 * LRU Cache implementation for LLM responses
 */
export declare class LLMCache<T = CacheableResponse> {
    private cache;
    private accessOrder;
    private readonly config;
    private hits;
    private misses;
    private evictions;
    constructor(config?: Partial<LLMCacheConfig>);
    /**
     * Generate a cache key from request parameters
     */
    static generateKey(type: 'generation' | 'embedding' | 'completion', input: string, options?: {
        model?: string;
        temperature?: number;
        maxTokens?: number;
        systemPrompt?: string;
    }): string;
    /**
     * Get a cached value
     */
    get(key: string): T | undefined;
    /**
     * Store a value in the cache
     */
    set(key: string, value: T, ttlMs?: number): void;
    /**
     * Check if a key exists and is not expired
     */
    has(key: string): boolean;
    /**
     * Delete a cached value
     */
    delete(key: string): boolean;
    /**
     * Clear all cached values
     */
    clear(): void;
    /**
     * Get cache statistics
     */
    getStats(): LLMCacheStats;
    /**
     * Get all cached keys
     */
    keys(): string[];
    /**
     * Get entries for export/persistence
     */
    entries(): Array<[string, CacheEntry<T>]>;
    /**
     * Import entries from persistence
     */
    import(entries: Array<[string, CacheEntry<T>]>): void;
    /**
     * Prune expired entries
     */
    pruneExpired(): number;
    /**
     * Update the LRU access order
     */
    private updateAccessOrder;
    /**
     * Evict the least recently used entry
     */
    private evictLRU;
    /**
     * Estimate the size of a cached value in bytes
     */
    private estimateSize;
}
/**
 * Specialized cache for LLM responses with request-specific key generation
 */
export declare class LLMResponseCache {
    private generationCache;
    private embeddingCache;
    private completionCache;
    private readonly config;
    constructor(config?: Partial<LLMCacheConfig>);
    /**
     * Get a cached generation response
     */
    getGeneration(input: string, options?: {
        model?: string;
        temperature?: number;
        maxTokens?: number;
        systemPrompt?: string;
    }): LLMResponse | undefined;
    /**
     * Cache a generation response
     */
    setGeneration(input: string, response: LLMResponse, options?: {
        model?: string;
        temperature?: number;
        maxTokens?: number;
        systemPrompt?: string;
    }, ttlMs?: number): void;
    /**
     * Get a cached embedding response
     */
    getEmbedding(text: string, options?: {
        model?: string;
    }): EmbeddingResponse | undefined;
    /**
     * Cache an embedding response
     */
    setEmbedding(text: string, response: EmbeddingResponse, options?: {
        model?: string;
    }, ttlMs?: number): void;
    /**
     * Get a cached completion response
     */
    getCompletion(prompt: string, options?: {
        model?: string;
        temperature?: number;
        maxTokens?: number;
    }): CompletionResponse | undefined;
    /**
     * Cache a completion response
     */
    setCompletion(prompt: string, response: CompletionResponse, options?: {
        model?: string;
        temperature?: number;
        maxTokens?: number;
    }, ttlMs?: number): void;
    /**
     * Get combined cache statistics
     */
    getStats(): {
        generation: LLMCacheStats;
        embedding: LLMCacheStats;
        completion: LLMCacheStats;
        total: LLMCacheStats;
    };
    /**
     * Clear all caches
     */
    clear(): void;
    /**
     * Prune expired entries from all caches
     */
    pruneExpired(): number;
}
//# sourceMappingURL=cache.d.ts.map