OptimizerBase.php — 10Web Booster – Website speed optimization, Cache & Page Speed optimizer trunk

tenweb-speed-optimizer / includes / OptimizerBase.php

tenweb-speed-optimizer / includes Last commit date

OptimizerBase.php

599 lines

1	<?php
2
3	namespace TenWebOptimizer;
4
5	/*
6	* Base class other (more-specific) classes inherit from.
7	*/
8	if (!defined('ABSPATH')) {
9	exit;
10	}
11
12	abstract class OptimizerBase
13	{
14	/**
15	* Holds content being processed (html, scripts, styles)
16	*
17	* @var string
18	*/
19	protected $content = '';
20
21	/**
22	* Controls debug logging.
23	*
24	* @var bool
25	*/
26	public $debug_log = false;
27
28	/** @var string */
29	public $cdn_url = '';
30
31	public function __construct($content)
32	{
33	$this->content = $content;
34	}
35
36	/**
37	* Reads the page and collects tags.
38	*
39	* @param array $options options
40	*
41	* @return bool
42	*/
43	abstract public function read($options);
44
45	/**
46	* Joins and optimizes collected things.
47	*
48	* @return bool
49	*/
50	abstract public function optimize();
51
52	/**
53	* Caches the things.
54	*
55	* @return void
56	*/
57	abstract public function cache();
58
59	/**
60	* Returns the content
61	*
62	* @return string
63	*/
64	abstract public function getcontent();
65
66	/**
67	* Tranfsorms a given URL to a full local filepath if possible.
68	* Returns local filepath or false.
69	*
70	* @param string $url URL to transform
71	*
72	* @return bool\|string
73	*/
74	public function getpath($url)
75	{
76	if (false !== strpos($url, '%')) {
77	$url = urldecode($url);
78	}
79	$site_host = wp_parse_url(TWO_WP_SITE_URL, PHP_URL_HOST);
80	$content_host = wp_parse_url(TWO_WP_ROOT_URL, PHP_URL_HOST);
81	// Normalizing attempts...
82	$double_slash_position = strpos($url, '//');
83
84	if (0 === $double_slash_position) {
85	if (is_ssl()) {
86	$url = 'https:' . $url;
87	} else {
88	$url = 'http:' . $url;
89	}
90	} elseif ((false === $double_slash_position) && (false === strpos($url, $site_host))) {
91	if (TWO_WP_SITE_URL === $site_host) {
92	$url = TWO_WP_SITE_URL . $url;
93	} else {
94	$url = TWO_WP_SITE_URL . OptimizerUtils::path_canonicalize($url);
95	}
96	}
97
98	if ($site_host !== $content_host) {
99	$url = str_replace(TWO_WP_CONTENT_URL, TWO_WP_SITE_URL . TWO_WP_CONTENT_NAME, $url);
100	}
101	// First check; hostname wp site should be hostname of url!
102	$url_host = @parse_url($url, PHP_URL_HOST); // @codingStandardsIgnoreLine
103
104	if ($url_host !== $site_host) {
105	$multidomains = [];
106	$multidomains_wpml = apply_filters('wpml_setting', [], 'language_domains');
107
108	if (!empty($multidomains_wpml)) {
109	$multidomains = array_map([$this, 'get_url_hostname'], $multidomains_wpml);
110	}
111
112	if (!empty($this->cdn_url)) {
113	$multidomains[] = wp_parse_url($this->cdn_url, PHP_URL_HOST);
114	}
115
116	if (!empty($multidomains)) {
117	if (in_array($url_host, $multidomains)) {
118	$url = str_replace($url_host, $site_host, $url);
119	} else {
120	return false;
121	}
122	} else {
123	return false;
124	}
125	}
126	// Try to remove "wp root url" from url while not minding http<>https.
127	$tmp_ao_root = preg_replace('/https?:/', '', TWO_WP_ROOT_URL);
128
129	if ($site_host !== $content_host) {
130	// As we replaced the content-domain with the site-domain, we should match against that.
131	$tmp_ao_root = preg_replace('/https?:/', '', TWO_WP_SITE_URL);
132	}
133	$tmp_url = preg_replace('/https?:/', '', $url);
134	$path = str_replace($tmp_ao_root, '', $tmp_url);
135
136	// If path starts with :// or //, this is not a URL in the WP context and
137	// we have to assume we can't aggregate.
138	if (preg_match('#^:?//#', $path)) {
139	// External script/css (adsense, etc).
140	return false;
141	}
142	// Prepend with WP_ROOT_DIR to have full path to file.
143	$path = str_replace('//', '/', WP_ROOT_DIR . $path);
144
145	// Final check: does file exist and is it readable?
146	if (file_exists($path) && is_file($path) && is_readable($path)) {
147	return $path;
148	} else {
149	return false;
150	}
151	}
152
153	/**
154	* Returns the hostname part of a given $url if we're able to parse it.
155	* If not, it returns the original url (prefixed with http:// scheme in case
156	* it was missing).
157	* Used as callback for WPML multidomains filter.
158	*
159	* @param string $url URL
160	*
161	* @return string
162	*/
163	protected function get_url_hostname($url)
164	{
165	// Checking that the url starts with something vaguely resembling a protocol.
166	if ((0 !== strpos($url, 'http')) && (0 !== strpos($url, '//'))) {
167	$url = 'http://' . $url;
168	}
169	// Grab the hostname.
170	$hostname = wp_parse_url($url, PHP_URL_HOST);
171
172	// Fallback when parse_url() fails.
173	if (empty($hostname)) {
174	$hostname = $url;
175	}
176
177	return $hostname;
178	}
179
180	/**
181	* Hides everything between noptimize-comment tags.
182	*
183	* @param string $markup markup to process
184	*
185	* @return string
186	*/
187	protected function hide_noptimize($markup)
188	{
189	return $this->replace_contents_with_marker_if_exists('NOPTIMIZE', '/<!--\s?noptimize\s?-->/', '#<!--\s?noptimize\s?-->.*?<!--\s?/\s?noptimize\s?-->#is', $markup);
190	}
191
192	/**
193	* Unhide noptimize-tags.
194	*
195	* @param string $markup markup to process
196	*
197	* @return string
198	*/
199	protected function restore_noptimize($markup)
200	{
201	return $this->restore_marked_content('NOPTIMIZE', $markup);
202	}
203
204	/**
205	* Hides "iehacks" content.
206	*
207	* @param string $markup markup to process
208	*
209	* @return string
210	*/
211	protected function hide_iehacks($markup)
212	{
213	return $this->replace_contents_with_marker_if_exists('IEHACK', // Marker name...
214	'<!--[if', // Invalid regex, will fallback to search using strpos()...
215	'#<!--\[if.*?\[endif\]-->#is', // Replacement regex...
216	$markup);
217	}
218
219	/**
220	* Restores "hidden" iehacks content.
221	*
222	* @param string $markup markup to process
223	*
224	* @return string
225	*/
226	protected function restore_iehacks($markup)
227	{
228	return $this->restore_marked_content('IEHACK', $markup);
229	}
230
231	/**
232	* "Hides" content within HTML comments using a regex-based replacement
233	* if HTML comment markers are found.
234	* `<!--example-->` becomes `%%COMMENTS%%ZXhhbXBsZQ==%%COMMENTS%%`
235	*
236	* @param string $markup markup to process
237	*
238	* @return string
239	*/
240	protected function hide_comments($markup)
241	{
242	return $this->replace_contents_with_marker_if_exists('COMMENTS', '<!--', '#<!--.*?-->#is', $markup);
243	}
244
245	/**
246	* Restores original HTML comment markers inside a string whose HTML
247	* comments have been "hidden" by using `hide_comments()`.
248	*
249	* @param string $markup markup to process
250	*
251	* @return string
252	*/
253	protected function restore_comments($markup)
254	{
255	return $this->restore_marked_content('COMMENTS', $markup);
256	}
257
258	/**
259	* Replaces the given URL with the CDN-version of it when CDN replacement
260	* is supposed to be done.
261	*
262	* @param string $url URL to process
263	*
264	* @return string
265	*/
266	public function url_replace_cdn($url)
267	{
268	// For 2.3 back-compat in which cdn-ing appeared to be automatically
269	// including WP subfolder/subdirectory into account as part of cdn-ing,
270	// even though it might've caused serious troubles in certain edge-cases.
271	$cdn_url = OptimizerUtils::tweak_cdn_url_if_needed($this->cdn_url);
272
273	// Allows API/filter to further tweak the cdn url...
274	if (!empty($cdn_url)) {
275	$this->debug_log('before=' . $url);
276	// Simple str_replace-based approach fails when $url is protocol-or-host-relative.
277	$is_protocol_relative = OptimizerUtils::is_protocol_relative($url);
278	$is_host_relative = (!$is_protocol_relative && ('/' === $url[0]));
279	$cdn_url = rtrim($cdn_url, '/');
280
281	if ($is_host_relative) {
282	// Prepending host-relative urls with the cdn url.
283	$url = $cdn_url . $url;
284	} else {
285	// Either a protocol-relative or "regular" url, replacing it either way.
286	if ($is_protocol_relative) {
287	// Massage $site_url so that simple str_replace() still "works" by
288	// searching for the protocol-relative version of TWO_WP_SITE_URL.
289	$site_url = str_replace(['http:', 'https:'], '', TWO_WP_SITE_URL);
290	} else {
291	$site_url = TWO_WP_SITE_URL;
292	}
293	$this->debug_log('`' . $site_url . '` -> `' . $cdn_url . '` in `' . $url . '`');
294	$url = str_replace($site_url, $cdn_url, $url);
295	}
296	$this->debug_log('after=' . $url);
297	}
298
299	return $url;
300	}
301
302	/**
303	* Returns true if given `$tag` is found in the list of `$list_for_delay`.
304	*
305	* @param string $tag tag to search for
306	* @param array $list_for_delay list of scripts considered to be delayed
307	*
308	* @return bool
309	*/
310	protected function isfordelay($tag, $list_for_delay)
311	{
312	if (false !== strpos($tag, OptimizerScripts::TWO_NO_DELAYED_JS_ATTRIBUTE)) {
313	return false;
314	}
315
316	foreach ($list_for_delay as $match) {
317	if (false !== strpos($tag, $match)) {
318	return true;
319	}
320	}
321
322	return false;
323	}
324
325	/**
326	* Returns true if given `$tag` is found in the list of `$removables`.
327	*
328	* @param string $tag tag to search for
329	* @param array $removables list of things considered completely removable
330	*
331	* @return bool
332	*/
333	protected function isremovable($tag, $removables)
334	{
335	foreach ($removables as $match) {
336	if (false !== strpos($tag, $match)) {
337	return true;
338	}
339	}
340
341	return false;
342	}
343
344	/**
345	* Callback used in `self::inject_minified()`.
346	*
347	* @param array $matches regex matches
348	*
349	* @return string
350	*/
351	public function inject_minified_callback($matches)
352	{
353	static $conf = null;
354	$filepath = null;
355	$filehash = null;
356	$parts = explode('\|', $matches[1]);
357
358	if (!empty($parts)) {
359	$filepath = isset($parts[0]) ? base64_decode($parts[0]) : null;
360	$filehash = isset($parts[1]) ? $parts[1] : null;
361	}
362
363	// Bail early if something's not right...
364	if (!$filepath \|\| !$filehash) {
365	return "\n";
366	}
367	$filecontent = file_get_contents($filepath); // phpcs:ignore
368	// Some things are differently handled for css/js...
369	$is_js_file = ('.js' === substr($filepath, -3, 3));
370	$is_css_file = false;
371
372	if (!$is_js_file) {
373	$is_css_file = ('.css' === substr($filepath, -4, 4));
374	}
375	// BOMs being nuked here unconditionally (regardless of where they are)!
376	$filecontent = preg_replace("#\x{EF}\x{BB}\x{BF}#", '', $filecontent);
377
378	// Remove comments and blank lines.
379	if ($is_js_file) {
380	$filecontent = preg_replace('#^\s\/\/.$#Um', '', $filecontent);
381	}
382	// Nuke un-important comments.
383	$filecontent = preg_replace('#^\s\/\[^!].\\/\s?#Um', '', $filecontent);
384	// Normalize newlines.
385	$filecontent = preg_replace('#(^[\r\n]\|[\r\n]+)[\s\t][\r\n]+#', "\n", $filecontent);
386
387	// JS specifics.
388	if ($is_js_file) {
389	// Append a semicolon at the end of js files if it's missing.
390	$last_char = substr($filecontent, -1, 1);
391
392	if (';' !== $last_char && '}' !== $last_char) {
393	$filecontent .= ';';
394	}
395	// Check if try/catch should be used.
396	$opt_js_try_catch = 'on';
397
398	if ('on' === $opt_js_try_catch) {
399	// It should, wrap in try/catch.
400	$filecontent = 'try{' . $filecontent . '}catch(e){}';
401	}
402	} elseif ($is_css_file) {
403	$filecontent = OptimizerStyles::fixurls($filepath, $filecontent);
404	} else {
405	$filecontent = '';
406	}
407
408	// phpcs:ignore Squiz.PHP.CommentedOutCode.Found
409	// Return modified (or empty) code/content.
410	return "\n" . $filecontent;
411	}
412
413	/**
414	* Inject already minified code in optimized JS/CSS.
415	*
416	* @param string $in markup
417	*
418	* @return string
419	*/
420	protected function inject_minified($in)
421	{
422	$out = $in;
423
424	if (false !== strpos($in, '%%INJECTLATER%%')) {
425	$out = preg_replace_callback('#\/\\!%%INJECTLATER' . TWO_HASH . '%%(.?)%%INJECTLATER%%\*\/#is', [
426	$this,
427	'inject_minified_callback',
428	], $in);
429	}
430
431	return $out;
432	}
433
434	/**
435	* Specialized method to create the INJECTLATER marker.
436	* These are somewhat "special", in the sense that they're additionally wrapped
437	* within an "exclamation mark style" comment, so that they're not stripped
438	* out by minifiers.
439	* They also currently contain the hash of the file's contents too (unlike other markers).
440	*
441	* @param string $filepath filepath
442	* @param string $hash hash
443	*
444	* @return string
445	*/
446	public static function build_injectlater_marker($filepath, $hash)
447	{
448	$contents = '/!' . self::build_marker('INJECTLATER', $filepath, $hash) . '/';
449
450	return $contents;
451	}
452
453	/**
454	* Creates and returns a `%%`-style named marker which holds
455	* the base64 encoded `$data`.
456	* If `$hash` is provided, it's appended to the base64 encoded string
457	* using `\|` as the separator (in order to support building the
458	* somewhat special/different INJECTLATER marker).
459	*
460	* @param string $name marker name
461	* @param string $data marker data which will be base64-encoded
462	* @param string\|null $hash optional
463	*
464	* @return string
465	*/
466	public static function build_marker($name, $data, $hash = null)
467	{
468	// Start the marker, add the data.
469	$marker = '%%' . $name . TWO_HASH . '%%' . base64_encode($data);
470
471	// Add the hash if provided.
472	if (null !== $hash) {
473	$marker .= '\|' . $hash;
474	}
475	// Close the marker.
476	$marker .= '%%' . $name . '%%';
477
478	return $marker;
479	}
480
481	public static function replace_contents_with_marker_if_exists($marker, $search, $re_replace_pattern, $content)
482	{
483	$found = false;
484	$is_regex = OptimizerUtils::str_is_valid_regex($search);
485
486	if ($is_regex) {
487	$found = preg_match($search, $content);
488	} else {
489	$found = (false !== strpos($content, $search));
490	}
491
492	if ($found) {
493	$content = preg_replace_callback($re_replace_pattern, function ($matches) use ($marker) {
494	return OptimizerBase::build_marker($marker, $matches[0]);
495	}, $content);
496	}
497
498	return $content;
499	}
500
501	public static function restore_marked_content($marker, $content)
502	{
503	if (false !== strpos($content, $marker)) {
504	$content = preg_replace_callback('#%%' . $marker . TWO_HASH . '%%(.*?)%%' . $marker . '%%#is', function ($matches) {
505	return base64_decode($matches[1]);
506	}, $content);
507	}
508
509	return $content;
510	}
511
512	/**
513	* Logs given `$data` for debugging purposes (when debug logging is on).
514	*
515	* @param mixed $data data to log
516	*
517	* @return void
518	*/
519	protected function debug_log($data)
520	{
521	if (!isset($this->debug_log) \|\| !$this->debug_log) {
522	return;
523	}
524
525	if (!is_string($data) && !is_resource($data)) {
526	$data = var_export($data, true); // phpcs:ignore
527	}
528	error_log($data); // phpcs:ignore
529	}
530
531	/**
532	* Checks if a single local css/js file can be minified and returns source if so.
533	*
534	* @param string $filepath filepath
535	*
536	* @return bool\|string to be minified code or false
537	*/
538	protected function prepare_minify_single($filepath)
539	{
540	// Decide what we're dealing with, return false if we don't know.
541	if ($this->str_ends_in($filepath, '.js')) {
542	$type = 'js';
543	} elseif ($this->str_ends_in($filepath, '.css')) {
544	$type = 'css';
545	} else {
546	return false;
547	}
548	// Bail if it looks like its already minifed (by having -min or .min
549	// in filename) or if it looks like WP jquery.js (which is minified).
550	$minified_variants = [
551	'-min.' . $type,
552	'.min.' . $type,
553	'js/jquery/jquery.js',
554	];
555
556	foreach ($minified_variants as $ending) {
557	if ($this->str_ends_in($filepath, $ending)) {
558	return false;
559	}
560	}
561	// Get file contents, bail if empty.
562	$contents = file_get_contents($filepath); // phpcs:ignore
563
564	return $contents;
565	}
566
567	protected function build_minify_single_url(OptimizerCache $cache)
568	{
569	$url = TWO_CACHE_URL . $cache->getname(true);
570	$url = $this->url_replace_cdn($url);
571
572	return $url;
573	}
574
575	/**
576	* Returns true if given $str ends with given $test.
577	*
578	* @param string $str string to check
579	* @param string $test ending to match
580	*
581	* @return bool
582	*/
583	protected function str_ends_in($str, $test)
584	{
585	// @codingStandardsIgnoreStart
586	// substr_compare() is bugged on 5.5.11: https://3v4l.org/qGYBH
587	// return ( 0 === substr_compare( $str, $test, -strlen( $test ) ) );
588	// @codingStandardsIgnoreEnd
589	$length = strlen($test);
590
591	return substr($str, -$length, $length) === $test;
592	}
593
594	protected function remove_from_html($tag)
595	{
596	$this->content = str_replace($tag, '', $this->content);
597	}
598	}
599