Use SearchWP to make WooCommerce Downloadable Product Document content searchable

functions.php

<?php

// Tell SearchWP to parse WooCommerce Downloadable Product downloads for document content.
// The content will be extracted from downloadable documents where possible
// and stored as extra metadata with a key of `swp_wc_doc_content`
add_filter( 'searchwp_extra_metadata', function( $extra_metadata, $the_post ) {
	if ( 'product' !== get_post_type( $the_post ) || ! class_exists( 'WC_Product' ) ) {
		return $extra_metadata;
	}

	$product = new WC_Product( $the_post );
	$downloads = $product->get_downloads();

	if ( empty( $downloads ) ) {
		return $extra_metadata;
	}

	// WooCommerce only stores a hashed ID, the filename, and the URL to the file
	// but we need to retrieve the Media library ID for each downloadble file.
	$upload_dir = wp_upload_dir();
	foreach ( $downloads as $key => $download ) {
		$relative_file_location = str_replace( trailingslashit( $upload_dir['baseurl'] ), '', $download['data']['file'] );

		// We can use the relative file location to retrieve the post ID we need to parse the PDFs.
		$file_id = get_posts( array(
			'post_type'   => 'attachment',
			'post_status' => 'inherit',
			'fields'      => 'ids',
			'meta_query'  => array(
				array(
					'key'   => '_wp_attached_file',
					'value' => $relative_file_location,
				),
			),
		) );

		if ( empty( $file_id ) ) {
			continue;
		}

		// Use SearchWP's parser to extract document content.
		$parser = new SearchWPDocumentParser( $file_id[0] );

		$product_content = ! empty( $parser->stored_content ) ? $parser->stored_content : $parser->extract_document_content();

		// Store the extracted content as extra metadata.
		$extra_metadata['swp_wc_doc_content'][] = $product_content;
	}

	return $extra_metadata;
}, 10, 2 );

// Add a Custom Field key for our WooCommerce Downloadable Product extracted Document Content.
add_filter( 'searchwp_custom_field_keys', function( $keys ) {
	$keys[] = 'swp_wc_doc_content';

	return $keys;
} );