magnitude-extract/dist/cli.d.cts

TypeScript DOM cleaning and structuring library

#!/usr/bin/env node
/**
 * CLI tool for extracting and converting web content to markdown
 * Usage: unstructured-ts <url> [options]
 */
interface CLIOptions {
    url: string;
    output?: string;
    includeMetadata?: boolean;
    includePageNumbers?: boolean;
    includeImages?: boolean;
    includeForms?: boolean;
    includeLinks?: boolean;
    skipNavigation?: boolean;
    minTextLength?: number;
    noTables?: boolean;
    verbose?: boolean;
    help?: boolean;
}
declare function fetchHTML(url: string): Promise<string>;
declare function parseArguments(args: string[]): CLIOptions;
declare function main(): Promise<void>;

export { fetchHTML, main, parseArguments };