← All changes | webp/class-wpo-webp-alter-html.php +85 -144 3.2.5 → 4.7.0 View file →

		@@ -1,27 +1,35 @@
1	1	<?php
2	2
3		-if (!defined('WPO_VERSION')) die('No direct access allowed');
	3	+if (!defined('ABSPATH')) die('No direct access allowed');
4	4
5	5	if (!class_exists('WPO_WebP_Alter_HTML')) :
6	6
7	7	class WPO_WebP_Alter_HTML {
8	8
9		- private $tags = array('img', 'source', 'input', 'iframe', 'div', 'li', 'link', 'a', 'section', 'video');
	9	+ /**
	10	+ * HTML tags that may contain image-related attributes eligible for WebP replacement.
	11	+ */
	12	+ const SUPPORTED_TAGS = array('img', 'source', 'input', 'iframe', 'div', 'li', 'link', 'a', 'section', 'video');
10	13
11	14	/**
12		- * Constructor
	15	+ * Regex pattern to identify image-related attribute names such as src, srcset,
	16	+ * poster, and common data-* attributes used for lazy loading or responsive images.
13	17	*/
	18	+ const IMAGE_ATTRIBUTE_PATTERN = '#^(src|srcset|poster|(data-[^=]*(lazy|small|slide|img|large|src|thumb|source|set|bg-url)[^=]*))$#i';
	19	+
	20	+ /**
	21	+ * Constructor — private to enforce singleton pattern.
	22	+ */
14	23	private function __construct() {
15		- add_action('template_redirect', array($this, 'start'), 9999);
16	24	}
17	25
18	26	/**
19		- * Returns singleton instance
	27	+ * Returns singleton instance.
20	28	*
21		- * @return WPO_WebP_Alter_HTML
	29	+ * @return self
22	30	*/
23		- public static function get_instance() {
	31	+ public static function get_instance(): self {
24	32	static $_instance = null;
25	33	if (null === $_instance) {
26	34	$_instance = new self();
27	35	}
		@@ -26,49 +34,45 @@
26	34	$_instance = new self();
27	35	}
28	36	return $_instance;
29	37	}
30		-
31		- /**
32		- * Start to alter html in output buffer
33		- */
34		- public function start() {
35		- if (!is_admin() || (defined('DOING_AJAX') && DOING_AJAX)) {
36		- ob_start(array(__CLASS__, 'alter_html'));
37		- }
38		- }
39	38
40	39	/**
41		- * Alter html to change image related tags to specify webp images
	40	+ * Alter HTML to replace image-related attributes with WebP URLs where available.
42	41	*
43		- * @param string $html - HTML document as string
	42	+ * @param string $html HTML document as string.
44	43	* @return string
45	44	*/
46		- public function alter_html($html) {
47		-
48		- if (!$this->is_valid_html($html)) return $html;
	45	+ public function alter_html(string $html): string {
	46	+ if (apply_filters('wpo_disable_webp_alter_html', false)) {
	47	+ return $html;
	48	+ }
49	49
50		- $this->maybe_include_simple_html_dom();
	50	+ // MAX_FILE_SIZE is defined in simple_html_dom — ensure it has a safe default.
	51	+ WP_Optimize_Utils::define_maxfile_size_constant();
51	52
52		- $dom = str_get_html($html, false, false, 'UTF-8', false);
	53	+ $dom = WP_Optimize_Utils::get_simple_html_dom_object($html);
53	54
54		- // MAX_FILE_SIZE is defined in simple_html_dom.
55		- // For safety sake, we make sure it is defined before using
56		- defined('MAX_FILE_SIZE') || define('MAX_FILE_SIZE', 600000);
57		-
58	55	if (false === $dom) {
59		- if (strlen($html) > MAX_FILE_SIZE) {
60		- return '<!-- Alter HTML was skipped because the HTML is too big to process! ' .
61		- '(limit is set to ' . MAX_FILE_SIZE . ' bytes) -->' . "\n" . $html;
62		- }
63		- return '<!-- Alter HTML was skipped because the helper library refused to process the html -->' .
64		- "\n" . $html;
	56	+ return $this->get_skipped_html($html);
65	57	}
66	58
67		- // Replace attributes (src, srcset, data-src, etc)
68		- foreach ($this->tags as $tag) {
69		- $elems = $dom->find($tag);
70		- foreach ($elems as $elem) {
	59	+ $this->replace_image_attributes($dom);
	60	+
	61	+ return $dom->save();
	62	+ }
	63	+
	64	+ /**
	65	+ * Walks all supported tags in the DOM and replaces image attribute values
	66	+ * with their WebP counterparts when available.
	67	+ *
	68	+ * @param simplehtmldom\HtmlDocument $dom A simple_html_dom instance.
	69	+ *
	70	+ * @return void
	71	+ */
	72	+ private function replace_image_attributes($dom): void {
	73	+ foreach (self::SUPPORTED_TAGS as $tag) {
	74	+ foreach ($dom->find($tag) as $elem) {
71	75	$attributes = $elem->getAllAttributes();
72	76	foreach ($attributes as $attr_name => $attr_value) {
73	77	if ($this->is_image_attribute($attr_name)) {
74	78	$elem->setAttribute($attr_name, $this->handle_attribute($attr_value));
		@@ -75,79 +79,49 @@
75	79	}
76	80	}
77	81	}
78	82	}
79		-
80		- return $dom->save();
81	83	}
82	84
83	85	/**
84		- * Checks whether supplied string is a valid html document or not
	86	+ * Returns the original HTML annotated with a comment explaining why
	87	+ * the WebP alteration was skipped.
85	88	*
86		- * @param string $html - HTML document as string
87		- * @return bool
	89	+ * @param string $html The original HTML document.
	90	+ * @return string
88	91	*/
89		- private function is_valid_html($html) {
90		- if (is_feed()) return false;
91		-
92		- // To prevent issue with `simple_html_dom` class
93		- // Exit if it doesn't look like HTML
94		- // https://github.com/rosell-dk/webp-express/issues/228
95		- if (!preg_match("#^\\s*<#", $html)) return false;
96		-
97		- if ('' == $html) return false;
98		- return true;
99		- }
100		-
101		- /**
102		- * Include simple html dom script if not available
103		- */
104		- private function maybe_include_simple_html_dom() {
105		- if (!function_exists('str_get_html')) {
106		- require_once WPO_PLUGIN_MAIN_PATH . 'vendor/simplehtmldom/simplehtmldom/simple_html_dom.php';
	92	+ private function get_skipped_html($html): string {
	93	+ if (strlen($html) > MAX_FILE_SIZE) {
	94	+ return $html . "\n" . "<!-- Alter HTML was skipped because the HTML is too big to process! " .
	95	+ "(limit is set to " . MAX_FILE_SIZE . " bytes) -->";
107	96	}
	97	+ return $html . "\n" . "<!-- Alter HTML was skipped because the helper library refused to process the html -->";
108	98	}
109	99
110	100	/**
111		- * Append image urls with `.webp` extension
	101	+ * Returns the URL with a .webp extension appended if a WebP version exists on disk.
112	102	*
113		- * @param strinng $url - Image URL
	103	+ * @param string $url URL of the original image.
114	104	* @return string
115	105	*/
116		- private function replace_url($url) {
117		- return $url . '.webp';
118		- }
119		-
120		- /**
121		- * If webp version for supplied image url is available then replace extension
122		- *
123		- * @param string $url - URL of image
124		- * @return string
125		- */
126		- private function maybe_replace_url($url) {
	106	+ private function maybe_replace_url($url): string {
127	107	if ($this->is_webp_version_available($url)) {
128		- $url = $this->replace_url($url);
	108	+ $url .= '.webp';
129	109	}
130	110	return $url;
131	111	}
132	112
133	113	/**
134		- * Modifies src attribute value, if needed
	114	+ * Processes a srcset attribute value, replacing each image URL with its
	115	+ * WebP counterpart when available.
135	116	*
136		- * @param string $attr_value
137		- * @return string
138		- */
139		- private function handle_src($attr_value) {
140		- return $this->maybe_replace_url($attr_value);
141		- }
142		-
143		- /**
144		- * Modified `srcset` attribute value, if needed
	117	+ * A srcset value is a comma-separated list of entries such as
	118	+ * "image.jpg 520w" or "image.jpg 2x" or just "image.jpg".
145	119	*
146		- * @param string $attr_value
	120	+ * @param string $attr_value The raw srcset attribute value.
147	121	* @return string
148	122	*/
149		- private function handle_srcset($attr_value) {
	123	+ private function handle_srcset($attr_value): string {
150	124	// $attr_value is ie: <img data-x="1.jpg 1000w, 2.jpg">
151	125	$srcset_arr = explode(',', $attr_value);
152	126	foreach ($srcset_arr as $i => $srcset_entry) {
153	127	// $srcset_entry is ie "image.jpg 520w", but can also lack width, ie just "image.jpg"
		@@ -153,9 +127,10 @@
153	127	// $srcset_entry is ie "image.jpg 520w", but can also lack width, ie just "image.jpg"
154	128	// it can also be ie "image.jpg 2x"
155	129	$srcset_entry = trim($srcset_entry);
156	130	$entry_parts = preg_split('/\s+/', $srcset_entry, 2);
157		- if (count($entry_parts) == 2) {
	131	+
	132	+ if (is_array($entry_parts) && 2 === count($entry_parts)) {
158	133	list($src, $descriptors) = $entry_parts;
159	134	} else {
160	135	$src = $srcset_entry;
161	136	$descriptors = null;
		@@ -167,89 +142,55 @@
167	142	return implode(', ', $srcset_arr);
168	143	}
169	144
170	145	/**
171		- * Decides whether given value is a `srcset` or not
	146	+ * Determines whether an attribute value resembles a srcset declaration
	147	+ * (contains a width or pixel-density descriptor).
172	148	*
	149	+ * @param string $value The attribute value to inspect.
173	150	* @return bool
174	151	*/
175		- private function looks_like_srcset($value) {
176		- if (preg_match('#\s\d*(w|x)#', $value)) {
177		- return true;
178		- }
179		- return false;
	152	+ private function looks_like_srcset($value): bool {
	153	+ return (bool) preg_match('#\s\d*(w|x)#', $value);
180	154	}
181	155
182	156	/**
183		- * Handle attribute value based on attribute name, src or srcset
	157	+ * Dispatches attribute value handling based on whether it looks like a
	158	+ * srcset or a plain image source URL.
184	159	*
	160	+ * @param string $value The attribute value.
185	161	* @return string
186	162	*/
187		- private function handle_attribute($value) {
	163	+ private function handle_attribute($value): string {
188	164	if ($this->looks_like_srcset($value)) {
189	165	return $this->handle_srcset($value);
190	166	}
191		- return $this->handle_src($value);
	167	+ return $this->maybe_replace_url($value);
192	168	}
193	169
194	170	/**
195		- * Decide whether given attribute name is an image attribute or not
	171	+ * Determines whether an attribute name is image-related and should be
	172	+ * considered for WebP replacement.
196	173	*
	174	+ * @param string $attr_name The attribute name.
197	175	* @return bool
198	176	*/
199		- private function is_image_attribute($attr_name) {
200		- return preg_match('#^(src|srcset|poster|(data-[^=]*(lazy|small|slide|img|large|src|thumb|source|set|bg-url)[^=]*))$#i', $attr_name);
	177	+ private function is_image_attribute($attr_name): bool {
	178	+ return (bool) preg_match(self::IMAGE_ATTRIBUTE_PATTERN, $attr_name);
201	179	}
202	180
203	181	/**
204		- * Does webp image file exists
	182	+ * Checks whether a WebP version of the given image URL exists on disk.
205	183	*
206		- * @param string $url
207		- * @return boolean
	184	+ * @param string $url The original image URL.
	185	+ * @return bool
208	186	*/
209		- private function is_webp_version_available($url) {
210		- $filename = $this->get_file_path($url);
	187	+ private function is_webp_version_available($url): bool {
	188	+ $filename = WP_Optimize_Utils::get_file_path($url);
	189	+ if (empty($filename)) {
	190	+ return false;
	191	+ }
211	192	return file_exists($filename . '.webp');
212		- }
213		-
214		- /**
215		- * Get the file path
216		- *
217		- * @param string $url
218		- * @return string
219		- */
220		- private function get_file_path($url) {
221		- if (is_multisite()) {
222		- if (function_exists('get_main_site_id')) {
223		- $site_id = get_main_site_id();
224		- } else {
225		- $network = get_network();
226		- $site_id = $network->site_id;
227		- }
228		- switch_to_blog($site_id);
229		- }
230		- $upload_dir = wp_upload_dir();
231		- $uploads_url = trailingslashit($upload_dir['baseurl']);
232		- $uploads_dir = trailingslashit($upload_dir['basedir']);
233		- if (is_multisite()) {
234		- restore_current_blog();
235		- }
236		- $possible_urls = array(
237		- WP_CONTENT_URL => WP_CONTENT_DIR,
238		- WP_PLUGIN_URL => WP_PLUGIN_DIR,
239		- $uploads_url => $uploads_dir,
240		- get_template_directory_uri() => get_template_directory(),
241		- includes_url() => preg_replace('/wp-content$/', trailingslashit('wp-includes'), WP_CONTENT_DIR),
242		- );
243		- $file = '';
244		- foreach ($possible_urls as $possible_url => $path) {
245		- $pos = strpos($url, $possible_url);
246		- if (false !== $pos) {
247		- $file = substr_replace($url, $path, $pos, strlen($possible_url));
248		- break;
249		- }
250		- }
251		- return $file;
252	193	}
253	194	}
254	195
255	196	endif;