/**
 * LLM-based content processor for web pages.
 *
 * Uses Cloudflare Workers AI to extract clean, semantic content from messy web pages.
 * Produces search-optimized content that matches how users ask questions.
 */
export interface LLMProcessorOptions {
    /** Cloudflare account ID */
    accountId: string;
    /** Cloudflare API token */
    apiToken: string;
    /** Enable verbose logging */
    verbose?: boolean;
    /** Request timeout in ms (default: 30000) */
    timeout?: number;
    /** Cloudflare Workers AI model to use (default: @cf/meta/llama-4-scout-17b-16e-instruct) */
    model?: string;
}
export interface PageContent {
    /** Page URL */
    url: string;
    /** Page title */
    title: string;
    /** Raw extracted content (markdown) */
    rawContent: string;
    /** Page category hint */
    category?: string;
}
export interface ProcessedPage {
    /** Page URL */
    url: string;
    /** Clean page title */
    title: string;
    /** Page type (floor-plan, community, general, etc.) */
    pageType: string;
    /** Brief summary of what this page is about (1-2 sentences) */
    summary: string;
    /** Clean content sections ready for embedding */
    sections: ProcessedSection[];
    /** Structured data extracted from the page */
    structuredData?: Record<string, string | number | string[]>;
}
export interface ProcessedSection {
    /** Section heading */
    heading: string;
    /** Clean content for this section */
    content: string;
}
/**
 * Detect page type from URL and content.
 */
export declare function detectPageType(url: string, content: string): string;
/**
 * Process a single page with LLM.
 */
export declare function processPageWithLLM(page: PageContent, options: LLMProcessorOptions): Promise<ProcessedPage>;
/**
 * Process multiple pages with LLM (with rate limiting).
 */
export declare function processPages(pages: PageContent[], options: LLMProcessorOptions & {
    concurrency?: number;
}): Promise<ProcessedPage[]>;
//# sourceMappingURL=llm-processor.d.ts.map