html-content-processor/dist/page-type-detector.d.ts

A professional library for processing, cleaning, filtering, and converting HTML content to Markdown. Features advanced customization options, presets, plugin support, fluent API, and TypeScript integration for reliable content extraction.

/**
 * Page Type Detector - Automatically detects page types and suggests optimal filtering parameters
 * Supports search engines, blogs, news sites, documentation, e-commerce, and more
 */
import { FilterOptions } from './types';
export interface PageTypeResult {
    /** Detected page type */
    type: PageType;
    /** Confidence score (0-1) */
    confidence: number;
    /** Recommended filter options */
    filterOptions: FilterOptions;
    /** Detection reasons */
    reasons: string[];
    /** Page characteristics */
    characteristics: PageCharacteristics;
}
export type PageType = 'search-engine' | 'blog' | 'news' | 'documentation' | 'e-commerce' | 'social-media' | 'forum' | 'landing-page' | 'article' | 'unknown';
export interface PageCharacteristics {
    /** Has search functionality */
    hasSearch: boolean;
    /** Has navigation menu */
    hasNavigation: boolean;
    /** Has article content */
    hasArticleContent: boolean;
    /** Has product listings */
    hasProductListings: boolean;
    /** Has social features */
    hasSocialFeatures: boolean;
    /** Has comments section */
    hasComments: boolean;
    /** Has code blocks */
    hasCodeBlocks: boolean;
    /** Link density ratio */
    linkDensity: number;
    /** Text to HTML ratio */
    textDensity: number;
    /** Number of forms */
    formCount: number;
    /** Number of images */
    imageCount: number;
}
export declare class PageTypeDetector {
    private detectionRules;
    /**
     * Detects the page type based on HTML content and optional URL
     */
    detectPageType(html: string, url?: string): Promise<PageTypeResult>;
    /**
     * Analyze page characteristics
     */
    private analyzePageCharacteristics;
    /**
     * Calculate scores for each page type
     */
    private calculateTypeScores;
    /**
     * Apply URL-based hints for page type detection
     */
    private applyUrlHints;
    /**
     * Get optimal filter options for detected page type
     */
    private getFilterOptionsForType;
    private hasSearchElements;
    private hasNavigationElements;
    private hasArticleElements;
    private hasProductElements;
    private hasSocialElements;
    private hasCommentElements;
    private hasCodeElements;
    private hasSearchEngineIndicators;
    private hasBlogIndicators;
    private hasNewsIndicators;
    private hasDocIndicators;
    private hasForumIndicators;
    private createUnknownResult;
}
export declare const pageTypeDetector: PageTypeDetector;