apple-hig-mcp/dist/services/crawlee-hig.service.d.ts

High-performance MCP server providing instant access to Apple's Human Interface Guidelines via hybrid static/dynamic content delivery

/**
 * CrawleeHIGService
 *
 * Modern replacement for HIGScraper using Crawlee's PlaywrightCrawler
 * for JavaScript-capable content extraction from Apple's SPA-based HIG website.
 *
 * Achieves 95%+ real content extraction by properly handling Apple's
 * Single Page Application architecture.
 */
import type { HIGSection, ApplePlatform, HIGCategory, SearchResult, ScrapingConfig } from '../types.js';
import type { HIGCache } from '../cache.js';
export interface ContentExtractionResult {
    content: string;
    quality: number;
    extractionMethod: 'crawlee' | 'fallback';
    timestamp: Date;
}
export interface CrawleeConfig extends ScrapingConfig {
    maxConcurrency: number;
    browserOptions: {
        headless: boolean;
        viewport: {
            width: number;
            height: number;
        };
        args: string[];
    };
    waitOptions: {
        networkIdle: number;
        timeout: number;
    };
}
export declare class CrawleeHIGService {
    private cache;
    private discoveryService;
    private config;
    private requestCount;
    private lastRequestTime;
    constructor(cache: HIGCache);
    /**
     * Discover all HIG sections using dynamic discovery
     */
    discoverSections(): Promise<HIGSection[]>;
    /**
     * Fetch content for a specific HIG section using Crawlee
     */
    fetchSectionContent(section: HIGSection): Promise<HIGSection>;
    /**
     * Extract content using direct Playwright browser instead of Crawlee
     */
    private extractContentWithCrawlee;
    /**
     * Extract content from a loaded page
     */
    private extractPageContent;
    /**
     * Extract content using main content strategy
     */
    private extractWithMainContentStrategy;
    /**
     * Extract content using article strategy
     */
    private extractWithArticleStrategy;
    /**
     * Extract content using body strategy
     */
    private extractWithBodyStrategy;
    /**
     * Extract content using fallback strategy
     */
    private extractWithFallbackStrategy;
    /**
     * Convert DOM element to markdown (to be injected into page context)
     */
    private convertElementToMarkdown;
    /**
     * Clean and normalize markdown content
     */
    private cleanMarkdownContent;
    /**
     * Calculate content quality score (0-1)
     */
    private calculateContentQuality;
    /**
     * Respect rate limiting
     */
    private respectRateLimit;
    /**
     * Get minimal fallback content when all extraction methods fail
     */
    private getMinimalFallbackContent;
    /**
     * Search content using existing discovery and extraction capabilities
     */
    searchContent(query: string, platform?: ApplePlatform, category?: HIGCategory, limit?: number): Promise<SearchResult[]>;
    /**
     * Clean up service resources
     */
    teardown(): Promise<void>;
}
//# sourceMappingURL=crawlee-hig.service.d.ts.map