// author: simon-pierre boucher <contact@spboucher.ai>
export interface PageMetadata {
  title?: string;
  description?: string;
  author?: string;
  canonical?: string;
  lang?: string;
  image?: string;
  siteName?: string;
  type?: string;
  publishedTime?: string;
  modifiedTime?: string;
  keywords?: string[];
}

export interface StructuredData {
  jsonld: unknown[];
  openGraph: Record<string, string>;
  twitter: Record<string, string>;
}

export interface ExtractOptions {
  readonly url: string;
  readonly onlyMainContent?: boolean;
  readonly includeTags?: readonly string[];
  readonly excludeTags?: readonly string[];
}

export interface ExtractResult {
  readonly markdown: string;
  readonly html: string;
  readonly links: PageLink[];
  readonly metadata: PageMetadata;
  readonly structured: StructuredData;
  readonly textLength: number;
}

export interface PageLink {
  readonly url: string;
  readonly text: string;
  readonly rel: string | null;
  readonly isInternal: boolean;
}

export const PIPELINE_VERSION = "0.1.0";