During `wp import`, cache remote files locally for subsequent imports

import-cache.php

<?php

/**
 * Run with `wp --require=import-cache.php import ...
 */
WP_CLI::add_hook( 'after_wp_load', function(){

	// Only intercept HTTP requests when the importer is running
	if ( ! defined( 'WP_IMPORTING') || ! WP_IMPORTING ) {
		return;
	}

	$get_cache = function() {
		// @todo this should maybe be configurable in some way
		$dir = dirname( __FILE__ ) . '/import-cache/';
		if ( ! is_dir( $dir ) ) {
			@mkdir( $dir );
		}
		// @todo File cache defaults to 6 months, 300 MB. Maybe this should be configurable?
		return new WP_CLI\FileCache( $dir, 15552000, 314572800 );
	};

	add_filter( 'pre_http_request', function( $retval, $r, $url ) use ( $get_cache ) {

		$key = md5( $url );
		$cache = $get_cache();
		if ( $cache->has( $key ) ){
			WP_CLI::log( " - Importing {$url} from cache..." );
			$body = $cache->read( $key );
			return array(
				// @todo probably want to return some semi-accurate header
				'headers' => array(),
				'body' => $body,
				'response' => array(
					'code' => 200,
					'message' => 'OK',
				),
				'cookies' => array(),
				'filename' => null,
			);
		}

		return $retval;

	}, 10, 3 );

	// @todo validate against expected content types
	add_filter( 'http_response', function( $response, $args, $url ) use( $get_cache ) {
		if ( 200 !== wp_remote_retrieve_response_code( $response ) ) {
			return $response;
		}
		WP_CLI::log( " - Saving {$url} to cache..." );
		$get_cache()->write( md5( $url ), wp_remote_retrieve_body( $response ) );
		return $response;
	}, 10, 3 );

});