Function extract

src/extractors/root-extractor.js:205–270

(extractor = GenericExtractor, opts)

const RootExtractor = {
  extract(extractor = GenericExtractor, opts) {
    const { contentOnly, extractedTitle } = opts;
    // This is the generic extractor. Run its extract method
    if (extractor.domain === '*') return extractor.extract(opts);

    opts = {
      ...opts,
      extractor,
    };

    if (contentOnly) {
      const content = extractResult({
        ...opts,
        type: 'content',
        extractHtml: true,
        title: extractedTitle,
      });
      return {
        content,
      };
    }
    const title = extractResult({ ...opts, type: 'title' });
    const date_published = extractResult({ ...opts, type: 'date_published' });
    const author = extractResult({ ...opts, type: 'author' });
    const next_page_url = extractResult({ ...opts, type: 'next_page_url' });
    const content = extractResult({
      ...opts,
      type: 'content',
      extractHtml: true,
      title,
    });
    const lead_image_url = extractResult({
      ...opts,
      type: 'lead_image_url',
      content,
    });
    const excerpt = extractResult({ ...opts, type: 'excerpt', content });
    const dek = extractResult({ ...opts, type: 'dek', content, excerpt });
    const word_count = extractResult({ ...opts, type: 'word_count', content });
    const direction = extractResult({ ...opts, type: 'direction', title });
    const { url, domain } = extractResult({
      ...opts,
      type: 'url_and_domain',
    }) || { url: null, domain: null };

    let extendedResults = {};
    if (extractor.extend) {
      extendedResults = selectExtendedTypes(extractor.extend, opts);
    }

    return {
      title,
      content,
      author,
      date_published,
      lead_image_url,
      dek,
      next_page_url,