/**
 * Text extraction for web captures (RCL-01) — the HTML/MHTML arm of
 * `extractText`, kept here so `cap-documents.ts` stays a dispatch table.
 *
 * @module v1/cli/knowledge/capture-text
 */
export declare const HTML_EXTENSIONS: Set<string>;
export declare const MHTML_EXTENSIONS: Set<string>;
export declare function isCaptureExtension(ext: string): boolean;
/**
 * Clean text for an archived page.
 *
 * PRECEDENCE — when the file sits in a capture envelope, the sibling
 * `readable.md` wins. It was produced by Readability against the LIVE DOM,
 * with lazy images resolved, the consent modal dismissed and the site's own JS
 * finished; nothing derived from the archived markup can beat that. Our own
 * extractor is the fallback for archives nobody cleaned.
 *
 * Returns '' rather than throwing — an unreadable capture is a document with
 * no text, not a failed ingest.
 */
export declare function extractCaptureText(absolutePath: string, ext: string): string;
//# sourceMappingURL=capture-text.d.ts.map