WPEP - Extract Image from different paths within WordPress from JSON data. Can be adapted.

WPEPAssetMatcher.php

<?php

namespace WPEPAddOnImportExport\Coordinator;

class AssetMatcher {

  public $website_url;

  public function __construct() {
    $this->website_url = get_site_url();
  }

  /**
   * @param $json
   * @return array
   */
  public function extract_from_json( $json ) {
    return $this->extract_from_array( json_decode( $json, true ) );
  }

  /**
   * @param $content_information
   * @return array
   */
  public function extract_from_array( $content_information ) {
    $response = $this->find_assets_from_entry( $content_information, $this->_get_extract_from_array_args() );

    return apply_filters( 'wpep_addon_ie_asset_matcher_result', array_unique( $response ), $content_information, $this->_get_extract_from_array_args() );
  }

  /**
   * We have the thumbnail id returned as the key, attachment_id => attachment_url
   * @return array
   */
  private function _get_extract_from_array_args() {
    $args = array(
      'courses' => array(
        'meta'  => array(
          'wpep_course_description',
          '_cic-thumb-image',
          '_cic-icon-image'
        ),
        'section_list' => array(
          'content',
          'lessons' => array( 'content' )
        )
      ),
      'taxonomies' => array(
        'course_category' => array(
          'meta' => array(
            'icon' => array(
              'url'
            )
          )
        )
      ),
      'videos'  => array(
        'post_information' => array( 'post_content' )
      ),
      'ebooks'  => array(
        'post_information' => array( 'post_content' ),
        'meta' => array( '_thumbnail_id' )
      )
    );

    $args = apply_filters( 'wpep_addon_ie_asset_matcher_args', $args );

    return $args;
  }

  /**
   * @param $entries
   * @param $path_map
   * @return array
   */
  public function find_assets_from_entries( $entries, $path_map ) {
    $response = array();

    foreach ($entries as $entry) {
      $response += $this->find_assets_from_entry( $entry, $path_map );
    }

    return $response;
  }

  /**
   * @param $entry
   * @param $path_map
   * @return array
   */
  public function find_assets_from_entry( $entry, $path_map ) {
    $response = array();

    foreach( $path_map as $path_index => $path ) {
      // We got to a section we care about, we're going to extract the information now.
      if( is_numeric( $path_index ) && isset( $entry[ $path ] ) ) {
        $response += $this->_find_assets_from_path_value( $entry[ $path ] );

        continue;
      }
      // We don't care about this case.
      if( !isset( $entry[ $path_index ] ) )
        continue;
      // We've reached an set of multiple entries, so we navigate accordingly.
      if( is_array( $entry[ $path_index ] ) && is_numeric( key( $entry[ $path_index ] ) ) ) {
        $response += $this->find_assets_from_entries( $entry[$path_index], $path );

        continue;
      }
      
      // Moving 1 level deeper into the array.
      if( is_array( $path ) ) {
        $response += $this->find_assets_from_entry( $entry[ $path_index ], $path );

        continue;
      }

    }

    return $response;
  }

  /**
   * @param $path_value
   * @return array
   */
  private function _find_assets_from_path_value( $path_value ) {
    $response = array();

    if( is_array( $path_value ) ) {
      foreach( $path_value as $value )
        $response += $this->_find_assets_from_string( $value );
    } else {
      $response += $this->_find_assets_from_string( $path_value );
    }

    return $response;
  }

  /**
   * @param $content
   * @return array
   */
  private function _find_assets_from_string( $content ) {
    $content = trim( $content );

    if( strpos( $content, $this->website_url ) === 0 ) {

      if( $this->_is_valid_resource( $content ) )
        return array( $this->_attachment_url_to_id( $content ) => $content );

      return array();
    }

    if( is_numeric( $content ) ) {
      $content = intval( $content );

      $attachment_url = wp_get_attachment_url( $content );

      if( $attachment_url != false )
        return array( $content => $attachment_url );

      return array();
    }

    $response = array();

    foreach( $this->_extract_from_dom( $content, '//img', 'src' ) as $target )
      if( strpos( $target, $this->website_url ) === 0 && $this->_is_valid_resource( $target ) )
        $response[ $this->_attachment_url_to_id( $target ) ] = $target;

    return $response;
  }

  /**
   * @param $content
   * @param string $element_path
   * @param bool $element_value
   * @return array
   */
  private function _extract_from_dom( $content, $element_path = '//img', $element_value = false ) {
    $doc = new \DOMDocument();
    $doc->loadHTML('<html><body>' . $content . '</body></html>');

    $xml = simplexml_import_dom($doc);

    $entries = $xml->xpath( $element_path );

    if( empty( $entries ) )
      return array();

    $response = array();

    foreach ( $entries as $entry ) {

      if( $element_value == false ) {
        $response[] = $entry;
        continue;
      }

      if( isset( $entry[ $element_value ] ) )
        $response[] = (string) $entry[ $element_value ]; // Stringify
    }

    return $response;
  }

  private function _is_valid_resource( $url ) {
    $url_headers = get_headers( $url, 1 );

    if( isset( $url_headers['Content-Type'] ) ){

      $type = strtolower($url_headers['Content-Type']);

      if( in_array( $type, array( 'image/png', 'image/jpg', 'image/jpeg', 'image/jpe', 'image/gif', 'image/tif', 'image/tiff', 'image/svg', 'image/ico', 'image/icon', 'image/x-icon' ) ) ){
        return true;
      }

    }

    return false;
  }

  private function _attachment_url_to_id( $image_url ) {
    global $wpdb;

    $attachment = $wpdb->get_col($wpdb->prepare("SELECT ID FROM $wpdb->posts WHERE guid='%s';", $image_url ));

    return isset( $attachment[0] ) ? $attachment[0] : md5( $image_url );
  }

}