/**
 * Extracts text from a PDF buffer and returns it as markdown.
 * Preserves paragraph breaks. Returns empty string on failure.
 */
export declare function extractPdfToMarkdown(data: Buffer, signal?: AbortSignal): Promise<string>;
/**
 * Converts a server-rendered HTML fragment to markdown.
 *
 * Used for Atlas documentation content resolved via the metadata API (whose
 * `content` field is server-rendered HTML, unlike the client-rendered SPA shell
 * served at the raw `.htm` URL). Returns an empty string for empty/whitespace
 * input.
 */
export declare function extractHtmlToMarkdown(html: string): string;
/**
 * Count of visible (non-tag, non-whitespace-collapsed) text characters in an
 * HTML string. Used to detect client-rendered SPA shells, whose visible text is
 * negligible (~21 chars, e.g. just "Salesforce Developers") even though the raw
 * markup is large.
 */
export declare function visibleTextLength(html: string): number;
/**
 * Extracts text from a DOCX buffer and returns it as markdown.
 * Uses mammoth to convert to HTML, then converts HTML to markdown.
 * Returns empty string on failure.
 */
export declare function extractDocxToMarkdown(data: Buffer): Promise<string>;