/**
 * Content Extraction Utilities
 *
 * Shared utilities for extracting and processing content from web pages.
 * Used by ContentIntelligence and site handlers.
 */
import type { ContentMapping } from '../types/api-patterns.js';
import { detectPageLanguage as detectLanguage, type LanguageDetectionResult, type FieldCategory } from './language-aware-extraction.js';
export { detectLanguage as detectPageLanguage };
export type { LanguageDetectionResult, FieldCategory };
/**
 * Get a value from an object using dot notation path
 * Supports array access like "items[0].title"
 */
export declare function getValueAtPath(obj: unknown, path: string): unknown;
/**
 * Check if an object has a field at the given path (supports dot notation and array access)
 */
export declare function hasFieldAtPath(obj: unknown, path: string): boolean;
/**
 * Get a string value from an object at the given path
 */
export declare function getStringAtPath(obj: unknown, path: string): string | null;
/**
 * Check if a string contains HTML content
 */
export declare function isHtmlContent(str: string): boolean;
/**
 * Simple HTML to plain text converter
 */
export declare function htmlToPlainText(html: string): string;
/**
 * Extract text content from structured data
 */
export declare function extractTextFromStructured(data: unknown): string;
/**
 * Extract text recursively from an object, filtering out code/URLs
 */
export declare function extractTextFromObject(obj: unknown, depth?: number): string;
/**
 * Extract content from API response using contentMapping
 * Handles HTML content by converting to plain text and markdown
 */
export declare function extractContentFromMapping(data: unknown, mapping: ContentMapping, turndownFn?: (html: string) => string): {
    title: string;
    text: string;
    markdown: string;
};
/**
 * Confidence thresholds for pattern application
 */
export declare const PATTERN_CONFIDENCE: {
    readonly MIN: 0.3;
    readonly MEDIUM: 0.5;
    readonly HIGH: 0.8;
};
/**
 * Get confidence level string from numeric score
 */
export declare function getConfidenceLevel(confidence: number): 'high' | 'medium' | 'low';
/**
 * Extract text content from structured data with language awareness
 *
 * Tries field names in the detected language first, then falls back to English.
 * This enables extraction from multilingual content like government portals.
 *
 * @param data - The data object to extract from
 * @param language - The detected page language (ISO 639-1 code)
 * @returns Extracted text content
 */
export declare function extractTextFromStructuredLanguageAware(data: unknown, language: string): string;
/**
 * Extract content from API response using contentMapping with language awareness
 *
 * First tries the mapping as-is, then tries language-specific field variants
 *
 * @param data - The data object to extract from
 * @param mapping - The content mapping
 * @param language - The detected page language
 * @param turndownFn - Optional HTML to markdown converter
 * @returns Extracted title, text, and markdown
 */
export declare function extractContentFromMappingLanguageAware(data: unknown, mapping: ContentMapping, language: string, turndownFn?: (html: string) => string): {
    title: string;
    text: string;
    markdown: string;
};
//# sourceMappingURL=content-extraction-utils.d.ts.map