content = $content; } /** * Reads the page and collects tags. * * @param array $options Options. * * @return bool */ abstract public function read( $options ); /** * Joins and optimizes collected things. * * @return bool */ abstract public function minify(); /** * Caches the things. * * @return void */ abstract public function cache(); /** * Returns the content * * @return string */ abstract public function getcontent(); /** * Tranfsorms a given URL to a full local filepath if possible. * Returns local filepath or false. * * @param string $url URL to transform. * * @return bool|string */ public function getpath( $url ) { $url = apply_filters( 'autoptimize_filter_cssjs_alter_url', $url ); if ( false !== strpos( $url, '%' ) ) { $url = urldecode( $url ); } $site_host = parse_url( AUTOPTIMIZE_WP_SITE_URL, PHP_URL_HOST ); $content_host = parse_url( AUTOPTIMIZE_WP_ROOT_URL, PHP_URL_HOST ); // Normalizing attempts... $double_slash_position = strpos( $url, '//' ); if ( 0 === $double_slash_position ) { if ( is_ssl() ) { $url = 'https:' . $url; } else { $url = 'http:' . $url; } } elseif ( ( false === $double_slash_position ) && ( false === strpos( $url, $site_host ) ) ) { if ( AUTOPTIMIZE_WP_SITE_URL === $site_host ) { $url = AUTOPTIMIZE_WP_SITE_URL . $url; } else { $url = AUTOPTIMIZE_WP_SITE_URL . autoptimizeUtils::path_canonicalize( $url ); } } if ( $site_host !== $content_host ) { $url = str_replace( AUTOPTIMIZE_WP_CONTENT_URL, AUTOPTIMIZE_WP_SITE_URL . AUTOPTIMIZE_WP_CONTENT_NAME, $url ); } // First check; hostname wp site should be hostname of url! $url_host = @parse_url( $url, PHP_URL_HOST ); // @codingStandardsIgnoreLine if ( $url_host !== $site_host ) { /** * First try to get all domains from WPML (if available) * then explicitely declare $this->cdn_url as OK as well * then apply own filter autoptimize_filter_cssjs_multidomain takes an array of hostnames * each item in that array will be considered part of the same WP multisite installation */ $multidomains = array(); $multidomains_wpml = apply_filters( 'wpml_setting', array(), 'language_domains' ); if ( ! empty( $multidomains_wpml ) ) { $multidomains = array_map( array( $this, 'get_url_hostname' ), $multidomains_wpml ); } if ( ! empty( $this->cdn_url ) ) { $multidomains[] = parse_url( $this->cdn_url, PHP_URL_HOST ); } $multidomains = apply_filters( 'autoptimize_filter_cssjs_multidomain', $multidomains ); if ( ! empty( $multidomains ) ) { if ( in_array( $url_host, $multidomains ) ) { $url = str_replace( $url_host, $site_host, $url ); } else { return false; } } else { return false; } } // Try to remove "wp root url" from url while not minding http<>https. $tmp_ao_root = preg_replace( '/https?:/', '', AUTOPTIMIZE_WP_ROOT_URL ); if ( $site_host !== $content_host ) { // As we replaced the content-domain with the site-domain, we should match against that. $tmp_ao_root = preg_replace( '/https?:/', '', AUTOPTIMIZE_WP_SITE_URL ); } $tmp_url = preg_replace( '/https?:/', '', $url ); $path = str_replace( $tmp_ao_root, '', $tmp_url ); // If path starts with :// or //, this is not a URL in the WP context and // we have to assume we can't aggregate. if ( preg_match( '#^:?//#', $path ) ) { // External script/css (adsense, etc). return false; } // Prepend with WP_ROOT_DIR to have full path to file. $path = str_replace( '//', '/', WP_ROOT_DIR . $path ); // Final check: does file exist and is it readable? if ( file_exists( $path ) && is_file( $path ) && is_readable( $path ) ) { return $path; } else { return false; } } /** * Returns the hostname part of a given $url if we're able to parse it. * If not, it returns the original url (prefixed with http:// scheme in case * it was missing). * Used as callback for WPML multidomains filter. * * @param string $url URL. * * @return string */ protected function get_url_hostname( $url ) { // Checking that the url starts with something vaguely resembling a protocol. if ( ( 0 !== strpos( $url, 'http' ) ) && ( 0 !== strpos( $url, '//' ) ) ) { $url = 'http://' . $url; } // Grab the hostname. $hostname = parse_url( $url, PHP_URL_HOST ); // Fallback when parse_url() fails. if ( empty( $hostname ) ) { $hostname = $url; } return $hostname; } /** * Hides everything between noptimize-comment tags. * * @param string $markup Markup to process. * * @return string */ protected function hide_noptimize( $markup ) { return $this->replace_contents_with_marker_if_exists( 'NOPTIMIZE', '//', '#.*?#is', $markup ); } /** * Unhide noptimize-tags. * * @param string $markup Markup to process. * * @return string */ protected function restore_noptimize( $markup ) { return $this->restore_marked_content( 'NOPTIMIZE', $markup ); } /** * Hides "iehacks" content. * * @param string $markup Markup to process. * * @return string */ protected function hide_iehacks( $markup ) { return $this->replace_contents_with_marker_if_exists( 'IEHACK', // Marker name... '#is', // Replacement regex... $markup ); } /** * Restores "hidden" iehacks content. * * @param string $markup Markup to process. * * @return string */ protected function restore_iehacks( $markup ) { return $this->restore_marked_content( 'IEHACK', $markup ); } /** * "Hides" content within HTML comments using a regex-based replacement * if HTML comment markers are found. * `` becomes `%%COMMENTS%%ZXhhbXBsZQ==%%COMMENTS%%` * * @param string $markup Markup to process. * * @return string */ protected function hide_comments( $markup ) { return $this->replace_contents_with_marker_if_exists( 'COMMENTS', '#is', $markup ); } /** * Restores original HTML comment markers inside a string whose HTML * comments have been "hidden" by using `hide_comments()`. * * @param string $markup Markup to process. * * @return string */ protected function restore_comments( $markup ) { return $this->restore_marked_content( 'COMMENTS', $markup ); } /** * Replaces the given URL with the CDN-version of it when CDN replacement * is supposed to be done. * * @param string $url URL to process. * * @return string */ public function url_replace_cdn( $url ) { // For 2.3 back-compat in which cdn-ing appeared to be automatically // including WP subfolder/subdirectory into account as part of cdn-ing, // even though it might've caused serious troubles in certain edge-cases. $cdn_url = autoptimizeUtils::tweak_cdn_url_if_needed( $this->cdn_url ); // Allows API/filter to further tweak the cdn url... $cdn_url = apply_filters( 'autoptimize_filter_base_cdnurl', $cdn_url ); if ( ! empty( $cdn_url ) ) { $this->debug_log( 'before=' . $url ); // Simple str_replace-based approach fails when $url is protocol-or-host-relative. $is_protocol_relative = autoptimizeUtils::is_protocol_relative( $url ); $is_host_relative = ( ! $is_protocol_relative && ( '/' === $url{0} ) ); $cdn_url = rtrim( $cdn_url, '/' ); if ( $is_host_relative ) { // Prepending host-relative urls with the cdn url. $url = $cdn_url . $url; } else { // Either a protocol-relative or "regular" url, replacing it either way. if ( $is_protocol_relative ) { // Massage $site_url so that simple str_replace() still "works" by // searching for the protocol-relative version of AUTOPTIMIZE_WP_SITE_URL. $site_url = str_replace( array( 'http:', 'https:' ), '', AUTOPTIMIZE_WP_SITE_URL ); } else { $site_url = AUTOPTIMIZE_WP_SITE_URL; } $this->debug_log( '`' . $site_url . '` -> `' . $cdn_url . '` in `' . $url . '`' ); $url = str_replace( $site_url, $cdn_url, $url ); } $this->debug_log( 'after=' . $url ); } // Allow API filter to take further care of CDN replacement. $url = apply_filters( 'autoptimize_filter_base_replace_cdn', $url ); return $url; } /** * Injects/replaces the given payload markup into `$this->content` * at the specified location. * If the specified tag cannot be found, the payload is appended into * $this->content along with a warning wrapped inside tags. * * @param string $payload Markup to inject. * @param array $where Array specifying the tag name and method of injection. * Index 0 is the tag name (i.e., ``). * Index 1 specifies ˛'before', 'after' or 'replace'. Defaults to 'before'. * * @return void */ protected function inject_in_html( $payload, $where ) { $warned = false; $position = autoptimizeUtils::strpos( $this->content, $where[0] ); if ( false !== $position ) { // Found the tag, setup content/injection as specified. if ( 'after' === $where[1] ) { $content = $where[0] . $payload; } elseif ( 'replace' === $where[1] ) { $content = $payload; } else { $content = $payload . $where[0]; } // Place where specified. $this->content = autoptimizeUtils::substr_replace( $this->content, $content, $position, // Using plain strlen() should be safe here for now, since // we're not searching for multibyte chars here still... strlen( $where[0] ) ); } else { // Couldn't find what was specified, just append and add a warning. $this->content .= $payload; if ( ! $warned ) { $tag_display = str_replace( array( '<', '>' ), '', $where[0] ); $this->content .= ''; $warned = true; } } } /** * Returns true if given `$tag` is found in the list of `$removables`. * * @param string $tag Tag to search for. * @param array $removables List of things considered completely removable. * * @return bool */ protected function isremovable( $tag, $removables ) { foreach ( $removables as $match ) { if ( false !== strpos( $tag, $match ) ) { return true; } } return false; } /** * Callback used in `self::inject_minified()`. * * @param array $matches Regex matches. * * @return string */ public function inject_minified_callback( $matches ) { static $conf = null; if ( null === $conf ) { $conf = autoptimizeConfig::instance(); } /** * $matches[1] holds the whole match caught by regex in self::inject_minified(), * so we take that and split the string on `|`. * First element is the filepath, second is the md5 hash of contents * the filepath had when it was being processed. * If we don't have those, we'll bail out early. */ $filepath = null; $filehash = null; // Grab the parts we need. $parts = explode( '|', $matches[1] ); if ( ! empty( $parts ) ) { $filepath = isset( $parts[0] ) ? base64_decode( $parts[0] ) : null; $filehash = isset( $parts[1] ) ? $parts[1] : null; } // Bail early if something's not right... if ( ! $filepath || ! $filehash ) { return "\n"; } $filecontent = file_get_contents( $filepath ); // Some things are differently handled for css/js... $is_js_file = ( '.js' === substr( $filepath, -3, 3 ) ); $is_css_file = false; if ( ! $is_js_file ) { $is_css_file = ( '.css' === substr( $filepath, -4, 4 ) ); } // BOMs being nuked here unconditionally (regardless of where they are)! $filecontent = preg_replace( "#\x{EF}\x{BB}\x{BF}#", '', $filecontent ); // Remove comments and blank lines. if ( $is_js_file ) { $filecontent = preg_replace( '#^\s*\/\/.*$#Um', '', $filecontent ); } // Nuke un-important comments. $filecontent = preg_replace( '#^\s*\/\*[^!].*\*\/\s?#Um', '', $filecontent ); // Normalize newlines. $filecontent = preg_replace( '#(^[\r\n]*|[\r\n]+)[\s\t]*[\r\n]+#', "\n", $filecontent ); // JS specifics. if ( $is_js_file ) { // Append a semicolon at the end of js files if it's missing. $last_char = substr( $filecontent, -1, 1 ); if ( ';' !== $last_char && '}' !== $last_char ) { $filecontent .= ';'; } // Check if try/catch should be used. $opt_js_try_catch = $conf->get( 'autoptimize_js_trycatch' ); if ( 'on' === $opt_js_try_catch ) { // It should, wrap in try/catch. $filecontent = 'try{' . $filecontent . '}catch(e){}'; } } elseif ( $is_css_file ) { $filecontent = autoptimizeStyles::fixurls( $filepath, $filecontent ); } else { $filecontent = ''; } // Return modified (or empty!) code/content. return "\n" . $filecontent; } /** * Inject already minified code in optimized JS/CSS. * * @param string $in Markup. * * @return string */ protected function inject_minified( $in ) { $out = $in; if ( false !== strpos( $in, '%%INJECTLATER%%' ) ) { $out = preg_replace_callback( '#\/\*\!%%INJECTLATER' . AUTOPTIMIZE_HASH . '%%(.*?)%%INJECTLATER%%\*\/#is', array( $this, 'inject_minified_callback' ), $in ); } return $out; } /** * Specialized method to create the INJECTLATER marker. * These are somewhat "special", in the sense that they're additionally wrapped * within an "exclamation mark style" comment, so that they're not stripped * out by minifiers. * They also currently contain the hash of the file's contents too (unlike other markers). * * @param string $filepath Filepath. * @param string $hash Hash. * * @return string */ public static function build_injectlater_marker( $filepath, $hash ) { $contents = '/*!' . self::build_marker( 'INJECTLATER', $filepath, $hash ) . '*/'; return $contents; } /** * Creates and returns a `%%`-style named marker which holds * the base64 encoded `$data`. * If `$hash` is provided, it's appended to the base64 encoded string * using `|` as the separator (in order to support building the * somewhat special/different INJECTLATER marker). * * @param string $name Marker name. * @param string $data Marker data which will be base64-encoded. * @param string|null $hash Optional. * * @return string */ public static function build_marker( $name, $data, $hash = null ) { // Start the marker, add the data. $marker = '%%' . $name . AUTOPTIMIZE_HASH . '%%' . base64_encode( $data ); // Add the hash if provided. if ( null !== $hash ) { $marker .= '|' . $hash; } // Close the marker. $marker .= '%%' . $name . '%%'; return $marker; } /** * Searches for `$search` in `$content` (using either `preg_match()` * or `strpos()`, depending on whether `$search` is a valid regex pattern or not). * If something is found, it replaces `$content` using `$re_replace_pattern`, * effectively creating our named markers (`%%{$marker}%%`. * These are then at some point replaced back to their actual/original/modified * contents using `autoptimizeBase::restore_marked_content()`. * * @param string $marker Marker name (without percent characters). * @param string $search A string or full blown regex pattern to search for in $content. Uses `strpos()` or `preg_match()`. * @param string $re_replace_pattern Regex pattern to use when replacing contents. * @param string $content Content to work on. * * @return string */ public static function replace_contents_with_marker_if_exists( $marker, $search, $re_replace_pattern, $content ) { $found = false; $is_regex = autoptimizeUtils::str_is_valid_regex( $search ); if ( $is_regex ) { $found = preg_match( $search, $content ); } else { $found = ( false !== strpos( $content, $search ) ); } if ( $found ) { $content = preg_replace_callback( $re_replace_pattern, function( $matches ) use ( $marker ) { return autoptimizeBase::build_marker( $marker, $matches[0] ); }, $content ); } return $content; } /** * Complements `autoptimizeBase::replace_contents_with_marker_if_exists()`. * * @param string $marker Marker. * @param string $content Markup. * * @return string */ public static function restore_marked_content( $marker, $content ) { if ( false !== strpos( $content, $marker ) ) { $content = preg_replace_callback( '#%%' . $marker . AUTOPTIMIZE_HASH . '%%(.*?)%%' . $marker . '%%#is', function ( $matches ) { return base64_decode( $matches[1] ); }, $content ); } return $content; } /** * Logs given `$data` for debugging purposes (when debug logging is on). * * @param mixed $data Data to log. * * @return void */ protected function debug_log( $data ) { if ( ! isset( $this->debug_log ) || ! $this->debug_log ) { return; } if ( ! is_string( $data ) && ! is_resource( $data ) ) { $data = var_export( $data, true ); } error_log( $data ); } /** * Checks if a single local css/js file can be minified and returns source if so. * * @param string $filepath Filepath. * * @return bool|string to be minified code or false. */ protected function prepare_minify_single( $filepath ) { // Decide what we're dealing with, return false if we don't know. if ( $this->str_ends_in( $filepath, '.js' ) ) { $type = 'js'; } elseif ( $this->str_ends_in( $filepath, '.css' ) ) { $type = 'css'; } else { return false; } // Bail if it looks like its already minifed (by having -min or .min // in filename) or if it looks like WP jquery.js (which is minified). $minified_variants = array( '-min.' . $type, '.min.' . $type, 'js/jquery/jquery.js', ); foreach ( $minified_variants as $ending ) { if ( $this->str_ends_in( $filepath, $ending ) ) { return false; } } // Get file contents, bail if empty. $contents = file_get_contents( $filepath ); return $contents; } /** * Given an autoptimizeCache instance returns the (maybe cdn-ed) url of * the cached file. * * @param autoptimizeCache $cache autoptimizeCache instance. * * @return string */ protected function build_minify_single_url( autoptimizeCache $cache ) { $url = AUTOPTIMIZE_CACHE_URL . $cache->getname(); // CDN-replace the resulting URL if needed... $url = $this->url_replace_cdn( $url ); return $url; } /** * Returns true if given $str ends with given $test. * * @param string $str String to check. * @param string $test Ending to match. * * @return bool */ protected function str_ends_in( $str, $test ) { // @codingStandardsIgnoreStart // substr_compare() is bugged on 5.5.11: https://3v4l.org/qGYBH // return ( 0 === substr_compare( $str, $test, -strlen( $test ) ) ); // @codingStandardsIgnoreEnd $length = strlen( $test ); return ( substr( $str, -$length, $length ) === $test ); } }