/** * Options for HTML simplification */ export interface SimplifyHtmlOptions { /** Keep align attribute on paragraphs (default: true) */ keepAlignment?: boolean; /** Remove empty paragraphs (default: true) */ removeEmptyParagraphs?: boolean; /** Throw error if an alinea image hash is missing from mapping (default: false) */ strictAlineas?: boolean; } /** * Simplifies Word-generated HTML to clean, semantic HTML * * This function takes complex HTML generated by Word (typically from Assemblée nationale documents) * and converts it to simple, clean HTML with only essential semantic elements: * -
for paragraphs
* - for bold text
* - for italic text
* - for superscript
* - for subscript
* -
for line breaks
* - to
for headings
* - for links (href preserved)
*
* @param html - The input HTML string (Word-generated)
* @param options - Options for simplification
* @returns Simplified HTML string
*/
export declare function simplifyWordHtml(html: string, options?: SimplifyHtmlOptions): string;
/**
* Simplifies Word-generated HTML and returns a complete HTML document
*/
export declare function simplifyWordHtmlToDocument(html: string, options?: SimplifyHtmlOptions): string;