Function select

src/extractors/root-extractor.js:68–175

(opts)

}

export function select(opts) {
  const { $, type, extractionOpts, extractHtml = false } = opts;
  // Skip if there's not extraction for this type
  if (!extractionOpts) return null;

  // If a string is hardcoded for a type (e.g., Wikipedia
  // contributors), return the string
  if (typeof extractionOpts === 'string') return extractionOpts;

  const { selectors, defaultCleaner = true, allowMultiple } = extractionOpts;

  const overrideAllowMultiple = type === 'lead_image_url' || allowMultiple;

  const matchingSelector = findMatchingSelector(
    $,
    selectors,
    extractHtml,
    overrideAllowMultiple
  );

  if (!matchingSelector) return null;

  function transformAndClean($node) {
    makeLinksAbsolute($node, $, opts.url || '');
    cleanBySelectors($node, $, extractionOpts);
    transformElements($node, $, extractionOpts);
    return $node;
  }

  function selectHtml() {
    // If the selector type requests html as its return type
    // transform and clean the element with provided selectors
    let $content;

    // If matching selector is an array, we're considering this a
    // multi-match selection, which allows the parser to choose several
    // selectors to include in the result. Note that all selectors in the
    // array must match in order for this selector to trigger
    if (Array.isArray(matchingSelector)) {
      $content = $(matchingSelector.join(','));
      const $wrapper = $('<div></div>');
      $content.each((_, element) => {
        $wrapper.append(element);
      });

      $content = $wrapper;
    } else {
      $content = $(matchingSelector);
    }

    // Wrap in div so transformation can take place on root element
    $content.wrap($('<div></div>'));
    $content = $content.parent();
    $content = transformAndClean($content);
    if (Cleaners[type]) {
      Cleaners[type]($content, { ...opts, defaultCleaner });
    }