| 1 |
<?php |
| 2 |
// If this file is called directly, abort. |
| 3 |
if (!defined('ABSPATH')) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
|
| 8 |
# uses the simple html dom library |
| 9 |
use simplehtmldom\HtmlWeb; |
| 10 |
use simplehtmldom\HtmlDocument; |
| 11 |
|
| 12 |
/** |
| 13 |
* This Class Handles |
| 14 |
*/ |
| 15 |
|
| 16 |
Class Metasync_otto_html{ |
| 17 |
|
| 18 |
# html dom |
| 19 |
private $dom; |
| 20 |
|
| 21 |
# the file path to save to |
| 22 |
private $html_file; |
| 23 |
|
| 24 |
# uuid of site |
| 25 |
private $site_uuid; |
| 26 |
|
| 27 |
# the otto endpoint url |
| 28 |
private $otto_end_point; |
| 29 |
|
| 30 |
# PERFORMANCE OPTIMIZATION: Defer DOM reload until final save |
| 31 |
# Reduces 6-10 serialize/deserialize cycles to just 1 |
| 32 |
private $deferred_reload = false; |
| 33 |
|
| 34 |
# PERFORMANCE OPTIMIZATION: Cache commonly accessed DOM elements |
| 35 |
# Eliminates 8-10 full DOM traversals per page |
| 36 |
private $cached_elements = []; |
| 37 |
|
| 38 |
/** |
| 39 |
* Escape bare < characters in text content that SimpleHtmlDom misinterprets as HTML tags. |
| 40 |
* Example: "<4 microns" gets parsed as a tag, corrupting the DOM and breaking page layout. |
| 41 |
* Only targets < followed by digits (never valid HTML tag starts). |
| 42 |
* Protects <script> and <style> blocks where < appears in code. |
| 43 |
*/ |
| 44 |
private function sanitize_text_less_than($html) { |
| 45 |
$protected = []; |
| 46 |
$html = preg_replace_callback( |
| 47 |
'/<(script|style)([\s>])(.*?)<\/\1>/si', |
| 48 |
function($match) use (&$protected) { |
| 49 |
$placeholder = '<!--METASYNC_PROTECTED_' . count($protected) . '-->'; |
| 50 |
$protected[$placeholder] = $match[0]; |
| 51 |
return $placeholder; |
| 52 |
}, |
| 53 |
$html |
| 54 |
); |
| 55 |
|
| 56 |
$html = preg_replace('/<(?=\d)/', '<', $html); |
| 57 |
|
| 58 |
if (!empty($protected)) { |
| 59 |
$html = str_replace(array_keys($protected), array_values($protected), $html); |
| 60 |
} |
| 61 |
|
| 62 |
return $html; |
| 63 |
} |
| 64 |
|
| 65 |
/** |
| 66 |
* WP-355: Fix malformed self-closing non-void HTML tags before DOM parsing. |
| 67 |
* |
| 68 |
* Some themes (e.g. Bootstrap Component Blox) output malformed tags like: |
| 69 |
* <ul/ class="sub-menu dropdown-menu" /> |
| 70 |
* SimpleHtmlDom interprets <ul/ as a void element, strips all attributes, |
| 71 |
* and re-serializes as bare <ul> — breaking Bootstrap dropdown menus. |
| 72 |
* |
| 73 |
* This normalizes such tags into proper opening tags: |
| 74 |
* <ul class="sub-menu dropdown-menu"> |
| 75 |
* |
| 76 |
* @param string $html Raw HTML before DOM parsing. |
| 77 |
* @return string HTML with malformed self-closing non-void tags fixed. |
| 78 |
*/ |
| 79 |
private function fix_malformed_self_closing_tags($html) { |
| 80 |
# Non-void elements that must never be self-closing |
| 81 |
$tags = 'ul|ol|li|div|span|a|p|h[1-6]|section|header|footer|nav|main|article|aside' |
| 82 |
. '|table|thead|tbody|tfoot|tr|td|th|caption|colgroup' |
| 83 |
. '|form|fieldset|legend|label|select|option|optgroup|textarea|button' |
| 84 |
. '|figure|figcaption|details|summary|blockquote|pre|code|dl|dt|dd' |
| 85 |
. '|video|audio|canvas|iframe|object|embed|map|picture|source'; |
| 86 |
|
| 87 |
# Pattern 1: <tag/ attr...> or <tag/ attr... /> — slash after tag name |
| 88 |
# e.g. <ul/ class="sub-menu dropdown-menu" /> |
| 89 |
$html = preg_replace( |
| 90 |
'#<(' . $tags . ')/(\s[^>]*?)\s*/>#si', |
| 91 |
'<$1$2>', |
| 92 |
$html |
| 93 |
); |
| 94 |
|
| 95 |
# Pattern 2: <tag/ attr...> without trailing slash |
| 96 |
# e.g. <ul/ class="sub-menu"> |
| 97 |
$html = preg_replace( |
| 98 |
'#<(' . $tags . ')/(\s)#si', |
| 99 |
'<$1$2', |
| 100 |
$html |
| 101 |
); |
| 102 |
|
| 103 |
# Pattern 3: <tag attr... /> — proper self-closing syntax on non-void element |
| 104 |
# e.g. <ul class="sub-menu" /> |
| 105 |
$html = preg_replace( |
| 106 |
'#<(' . $tags . ')(\s[^>]*?)\s*/>#si', |
| 107 |
'<$1$2>', |
| 108 |
$html |
| 109 |
); |
| 110 |
|
| 111 |
return $html; |
| 112 |
} |
| 113 |
|
| 114 |
/** |
| 115 |
* Restore case-sensitive SVG and HTML5 attributes that SimpleHtmlDom lowercases. |
| 116 |
* The DOM parser's $lowercase=true flag (required for find() queries) lowercases |
| 117 |
* all attribute names, breaking case-sensitive attributes like viewBox. |
| 118 |
* Applied on final HTML output after all DOM processing is complete. |
| 119 |
*/ |
| 120 |
private function restore_case_sensitive_attributes($html) { |
| 121 |
static $map = [ |
| 122 |
' viewbox=' => ' viewBox=', |
| 123 |
' preserveaspectratio=' => ' preserveAspectRatio=', |
| 124 |
' controlslist=' => ' controlsList=', |
| 125 |
]; |
| 126 |
return str_ireplace(array_keys($map), array_values($map), $html); |
| 127 |
} |
| 128 |
|
| 129 |
/** |
| 130 |
* WP-355 / WP-315: Capture ALL <style> blocks before DOM processing. |
| 131 |
* |
| 132 |
* SimpleHtmlDom can lose or corrupt <style> blocks during parse/serialize, |
| 133 |
* especially those without an id attribute (common in custom themes). |
| 134 |
* This captures every <style> block so we can re-inject any that are lost. |
| 135 |
* |
| 136 |
* @param string $html Raw HTML before DOM parsing. |
| 137 |
* @return array Keyed by id (or content hash for anonymous styles) → full <style> tag. |
| 138 |
*/ |
| 139 |
private function capture_style_blocks($html) { |
| 140 |
$blocks = []; |
| 141 |
if (preg_match_all('/<style[^>]*>.*?<\/style>/si', $html, $matches)) { |
| 142 |
foreach ($matches[0] as $style_tag) { |
| 143 |
if (preg_match('/id=["\']([^"\']+)["\']/', $style_tag, $id_match)) { |
| 144 |
$key = $id_match[1]; |
| 145 |
} else { |
| 146 |
$key = 'anon_' . md5($style_tag); |
| 147 |
} |
| 148 |
$blocks[$key] = $style_tag; |
| 149 |
} |
| 150 |
} |
| 151 |
return $blocks; |
| 152 |
} |
| 153 |
|
| 154 |
/** |
| 155 |
* WP-355 / WP-315: Re-inject any <style> blocks lost during DOM processing. |
| 156 |
* |
| 157 |
* Compares the captured blocks against the final HTML and re-injects any |
| 158 |
* that disappeared. Named blocks (with id) are checked by id; anonymous |
| 159 |
* blocks are checked by exact content match. |
| 160 |
* |
| 161 |
* @param string $html Processed HTML after DOM modifications. |
| 162 |
* @param array $original_blocks Blocks returned by capture_style_blocks(). |
| 163 |
* @return string HTML with lost style blocks restored. |
| 164 |
*/ |
| 165 |
private function restore_lost_style_blocks($html, $original_blocks) { |
| 166 |
if (empty($original_blocks)) { |
| 167 |
return $html; |
| 168 |
} |
| 169 |
|
| 170 |
foreach ($original_blocks as $key => $style_tag) { |
| 171 |
if (strpos($key, 'anon_') === 0) { |
| 172 |
# Anonymous block — check if exact content is still present |
| 173 |
if (strpos($html, $style_tag) === false) { |
| 174 |
$html = str_replace('</head>', $style_tag . "\n" . '</head>', $html); |
| 175 |
} |
| 176 |
} else { |
| 177 |
# Named block — check if its id is still in the document |
| 178 |
if (strpos($html, 'id="' . $key . '"') === false |
| 179 |
&& strpos($html, "id='" . $key . "'") === false) { |
| 180 |
$html = str_replace('</head>', $style_tag . "\n" . '</head>', $html); |
| 181 |
} |
| 182 |
} |
| 183 |
} |
| 184 |
|
| 185 |
return $html; |
| 186 |
} |
| 187 |
|
| 188 |
# |
| 189 |
function __construct($otto_uuid){ |
| 190 |
|
| 191 |
# set the site uuid using the provided string |
| 192 |
$this->site_uuid = $otto_uuid; |
| 193 |
|
| 194 |
# Use endpoint manager if available, otherwise fallback to production |
| 195 |
if (class_exists('Metasync_Endpoint_Manager')) { |
| 196 |
$this->otto_end_point = Metasync_Endpoint_Manager::get_endpoint('OTTO_URL_DETAILS'); |
| 197 |
} else { |
| 198 |
$this->otto_end_point = 'https://sa.searchatlas.com/api/v2/otto-url-details'; |
| 199 |
} |
| 200 |
|
| 201 |
# laod the simple html dom parser with UTF-8 charset to handle special characters |
| 202 |
$this->dom = new HtmlDocument(null, true, true, 'UTF-8', false); |
| 203 |
} |
| 204 |
|
| 205 |
/** |
| 206 |
* Check Route Method |
| 207 |
* @param route : The route to check |
| 208 |
* @param path : The path of the html file to save |
| 209 |
*/ |
| 210 |
function process_route($route, $file_path){ |
| 211 |
|
| 212 |
# Construct the full endpoint URL with query parameters |
| 213 |
$url_with_params = add_query_arg( |
| 214 |
[ |
| 215 |
'url' => $route, |
| 216 |
'uuid' => $this->site_uuid, |
| 217 |
], |
| 218 |
$this->otto_end_point |
| 219 |
); |
| 220 |
|
| 221 |
# PERFORMANCE FIX: Add timeout to prevent blocking |
| 222 |
$args = array( |
| 223 |
'timeout' => 5, // 5 second max timeout (allow time for redirects) |
| 224 |
'redirection' => 5, // CRITICAL FIX: Allow redirects (API returns 301) |
| 225 |
'user-agent' => 'MetaSync-OTTO-SSR/2.0', |
| 226 |
'sslverify' => true |
| 227 |
); |
| 228 |
|
| 229 |
# Perform the GET request with timeout |
| 230 |
$response = wp_remote_get($url_with_params, $args); |
| 231 |
|
| 232 |
# Check for errors |
| 233 |
if (is_wp_error($response)) { |
| 234 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': API call failed - ' . $response->get_error_message()); |
| 235 |
return false; |
| 236 |
} |
| 237 |
|
| 238 |
# get the response body |
| 239 |
$body = wp_remote_retrieve_body($response); |
| 240 |
|
| 241 |
# Get the response code |
| 242 |
$response_code = wp_remote_retrieve_response_code($response); |
| 243 |
|
| 244 |
# if no change data skip |
| 245 |
if (empty($body) || $response_code !== 200){ |
| 246 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': API returned empty or non-200. Code: ' . $response_code); |
| 247 |
return false; |
| 248 |
} |
| 249 |
|
| 250 |
# set the html file path |
| 251 |
$this->html_file = $file_path; |
| 252 |
|
| 253 |
# load change data |
| 254 |
$change_data = json_decode($body, true); |
| 255 |
|
| 256 |
# Process with the fetched data |
| 257 |
return $this->process_route_with_data($route, $change_data, $file_path); |
| 258 |
} |
| 259 |
|
| 260 |
/** |
| 261 |
* Process route with pre-fetched suggestions data |
| 262 |
* OPTION 1: Used when data comes from transient cache |
| 263 |
* @param route : The route to check |
| 264 |
* @param change_data : Pre-fetched OTTO suggestions data |
| 265 |
* @param path : The path of the html file to save |
| 266 |
*/ |
| 267 |
function process_route_with_data($route, $change_data, $file_path){ |
| 268 |
|
| 269 |
if (empty($change_data) || !is_array($change_data)) { |
| 270 |
return false; |
| 271 |
} |
| 272 |
|
| 273 |
# set the html file path |
| 274 |
$this->html_file = $file_path; |
| 275 |
|
| 276 |
# Analyze what Otto is providing and store for conditional SEO blocking |
| 277 |
$has_otto_title = false; |
| 278 |
$otto_description_tags = []; // Track specific description tags Otto provides |
| 279 |
|
| 280 |
if (!empty($change_data['header_replacements']) && is_array($change_data['header_replacements'])) { |
| 281 |
foreach ($change_data['header_replacements'] as $item) { |
| 282 |
if (!empty($item['type'])) { |
| 283 |
# Check if Otto has title |
| 284 |
if ($item['type'] == 'title' && !empty($item['recommended_value'])) { |
| 285 |
$has_otto_title = true; |
| 286 |
} |
| 287 |
# Check if Otto has description - track specific tag types |
| 288 |
if ($item['type'] == 'meta') { |
| 289 |
# Check for meta[name=description] |
| 290 |
if (!empty($item['name']) && $item['name'] == 'description' && !empty($item['recommended_value'])) { |
| 291 |
$otto_description_tags[] = 'meta[name=description]'; |
| 292 |
} |
| 293 |
# Check for meta[property=og:description] |
| 294 |
if (!empty($item['property']) && $item['property'] == 'og:description' && !empty($item['recommended_value'])) { |
| 295 |
$otto_description_tags[] = 'meta[property=og:description]'; |
| 296 |
} |
| 297 |
# Check for meta[name=twitter:description] |
| 298 |
if (!empty($item['name']) && $item['name'] == 'twitter:description' && !empty($item['recommended_value'])) { |
| 299 |
$otto_description_tags[] = 'meta[name=twitter:description]'; |
| 300 |
} |
| 301 |
} |
| 302 |
} |
| 303 |
} |
| 304 |
} |
| 305 |
|
| 306 |
# Check header_html_insertion for description - must have non-empty content value |
| 307 |
if (!empty($change_data['header_html_insertion'])) { |
| 308 |
if (preg_match('/<meta[^>]*name=["\']description["\'][^>]*content=["\']([^"\']+)["\'][^>]*>/i', $change_data['header_html_insertion'])) { |
| 309 |
$otto_description_tags[] = 'meta[name=description]'; |
| 310 |
} |
| 311 |
} |
| 312 |
|
| 313 |
# Remove duplicates |
| 314 |
$otto_description_tags = array_unique($otto_description_tags); |
| 315 |
|
| 316 |
# Store blocking flags to pass to handle_route_html |
| 317 |
# This will be added to the internal fetch URL as parameters |
| 318 |
$change_data['_otto_blocking'] = array( |
| 319 |
'block_title' => $has_otto_title, |
| 320 |
'block_description_tags' => $otto_description_tags // Pass array of specific tags to remove |
| 321 |
); |
| 322 |
|
| 323 |
# Process the route with the suggestions data |
| 324 |
return $this->handle_route_html($route, $change_data); |
| 325 |
|
| 326 |
} |
| 327 |
|
| 328 |
# function to get tag attributes |
| 329 |
function get_tag_attributes($tag){ |
| 330 |
|
| 331 |
# Extract existing attributes of the <body> tag |
| 332 |
$attributes = []; |
| 333 |
|
| 334 |
|
| 335 |
# set the tag attributes |
| 336 |
$tag_attributes = []; |
| 337 |
|
| 338 |
|
| 339 |
# check that the tag attributes |
| 340 |
if(!is_object($tag) || !method_exists($tag, 'getAllAttributes')){ |
| 341 |
return ''; |
| 342 |
} |
| 343 |
|
| 344 |
# get the tag attributes |
| 345 |
$tag_attributes = $tag->getAllAttributes(); |
| 346 |
|
| 347 |
# loop all attributes |
| 348 |
foreach ($tag_attributes as $key => $value) { |
| 349 |
|
| 350 |
if ($value == 1) { |
| 351 |
|
| 352 |
# Handle boolean attributes |
| 353 |
$attributes[] = htmlspecialchars($key, ENT_QUOTES); |
| 354 |
} else { |
| 355 |
|
| 356 |
# Handle attributes with values |
| 357 |
$attributes[] = $key . '="' . htmlspecialchars($value, ENT_QUOTES) . '"'; |
| 358 |
} |
| 359 |
} |
| 360 |
|
| 361 |
# Convert attributes array to a string |
| 362 |
$attributes_string = !empty($attributes) ? ' ' . implode(' ', $attributes) : ''; |
| 363 |
|
| 364 |
# return the attributes string |
| 365 |
return $attributes_string; |
| 366 |
} |
| 367 |
|
| 368 |
function handle_route_html($route, $replacement_data){ |
| 369 |
|
| 370 |
# Detect if current page uses Brizy and disable SG Cache if so |
| 371 |
# Using global function defined in otto_pixel.php |
| 372 |
if (function_exists('metasync_otto_disable_sg_cache_for_brizy')) { |
| 373 |
metasync_otto_disable_sg_cache_for_brizy(); |
| 374 |
} |
| 375 |
|
| 376 |
# lablel the Otto Route |
| 377 |
# label otto requests to avoid loops |
| 378 |
// $request_body = add_query_arg( |
| 379 |
// [ |
| 380 |
// 'is_otto_page_fetch' => 1 |
| 381 |
// ], |
| 382 |
// $route |
| 383 |
// ); |
| 384 |
# Add blocking flags as URL parameters (no database writes!) |
| 385 |
$url_params = ['is_otto_page_fetch' => 1]; |
| 386 |
|
| 387 |
# Add blocking flags if available |
| 388 |
if (!empty($replacement_data['_otto_blocking'])) { |
| 389 |
$url_params['otto_block_title'] = $replacement_data['_otto_blocking']['block_title'] ? '1' : '0'; |
| 390 |
# For HTTP fetch path, check if any description tags need blocking |
| 391 |
$block_description_tags = $replacement_data['_otto_blocking']['block_description_tags'] ?? []; |
| 392 |
$url_params['otto_block_desc'] = !empty($block_description_tags) ? '1' : '0'; |
| 393 |
} |
| 394 |
|
| 395 |
$request_body = add_query_arg($url_params, $route); |
| 396 |
|
| 397 |
# TUNNEL/PROXY SUPPORT: If site is behind a tunnel (ngrok, zrok, etc.) |
| 398 |
# and loopback requests fail, try using localhost instead |
| 399 |
$request_body = apply_filters('metasync_otto_internal_fetch_url', $request_body, $route); |
| 400 |
# set cookie header var |
| 401 |
$cookie_header = ''; |
| 402 |
|
| 403 |
# loop cookies to set header |
| 404 |
foreach ($_COOKIE as $name => $value) { |
| 405 |
|
| 406 |
# handle array values by converting to string |
| 407 |
$cookie_value = is_array($value) ? serialize($value) : $value; |
| 408 |
|
| 409 |
# add cookie to header |
| 410 |
# $cookie_header .= $name . '=' . $value . '; '; |
| 411 |
$cookie_header .= $name . '=' . $cookie_value . '; '; |
| 412 |
} |
| 413 |
|
| 414 |
# trim the string |
| 415 |
$cookie_header = rtrim($cookie_header, '; '); |
| 416 |
|
| 417 |
# Allow timeout customization for slow tunnel environments |
| 418 |
$fetch_timeout = apply_filters('metasync_otto_internal_fetch_timeout', 5); |
| 419 |
|
| 420 |
$args = array( |
| 421 |
'sslverify' => false, // Disabled for localhost/tunnel environments |
| 422 |
'timeout' => $fetch_timeout, // Configurable timeout for tunnels |
| 423 |
'redirection' => 5, |
| 424 |
'httpversion' => '1.1', |
| 425 |
'headers' => array( |
| 426 |
'Cookie' => $cookie_header, |
| 427 |
'Cache-Control' => 'no-cache, no-store, must-revalidate', |
| 428 |
'Pragma' => 'no-cache', |
| 429 |
'X-OTTO-Internal-Fetch' => '1', |
| 430 |
'User-Agent' => 'MetaSync-OTTO-SSR/3.0', |
| 431 |
'X-Forwarded-Host' => $_SERVER['HTTP_HOST'] ?? '', // Preserve original host for tunnels |
| 432 |
) |
| 433 |
); |
| 434 |
|
| 435 |
# get the associateed route html |
| 436 |
$route_html = wp_remote_get($request_body, $args); |
| 437 |
|
| 438 |
# Check for timeout or connection errors |
| 439 |
if (is_wp_error($route_html)) { |
| 440 |
$error_msg = $route_html->get_error_message(); |
| 441 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ' DEBUG: FAILED - wp_remote_get error: ' . $error_msg . ' for route: ' . $route); |
| 442 |
return false; |
| 443 |
} |
| 444 |
|
| 445 |
# get body |
| 446 |
$html_body = wp_remote_retrieve_body($route_html); |
| 447 |
|
| 448 |
# Get the response code |
| 449 |
$response_code = wp_remote_retrieve_response_code($route_html); |
| 450 |
|
| 451 |
|
| 452 |
# check not empty |
| 453 |
if(empty($html_body) || $response_code !== 200){ |
| 454 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ' DEBUG: FAILED - Empty body or non-200 status for route: ' . $route); |
| 455 |
return false; |
| 456 |
} |
| 457 |
|
| 458 |
# Remove XML declaration |
| 459 |
$html_body = preg_replace('/<\?xml[^?]*\?>\s*/i', '', $html_body); |
| 460 |
|
| 461 |
# WP-355 / WP-315: Save ALL original <style> blocks before DOM processing. |
| 462 |
# SimpleHtmlDom can lose or corrupt style blocks during parse/serialize — |
| 463 |
# not only Divi's id-tagged ones but also anonymous <style> blocks used by |
| 464 |
# custom themes for nav collapse CSS, responsive breakpoints, etc. |
| 465 |
$original_style_blocks = $this->capture_style_blocks($html_body); |
| 466 |
|
| 467 |
# Escape bare < in text content (e.g. "<4 microns") before DOM parsing |
| 468 |
$html_body = $this->sanitize_text_less_than($html_body); |
| 469 |
|
| 470 |
# WP-355: Fix malformed self-closing non-void tags (e.g. <ul/ class="...">) |
| 471 |
# before DOM parsing — SimpleHtmlDom strips attributes from these. |
| 472 |
$html_body = $this->fix_malformed_self_closing_tags($html_body); |
| 473 |
|
| 474 |
# now that the html is not empty |
| 475 |
# load it into the simple html dom |
| 476 |
$this->dom->load($html_body, true, false); |
| 477 |
|
| 478 |
# Force UTF-8 charset to preserve emojis and special characters |
| 479 |
# This overrides any charset detection from HTML meta tags |
| 480 |
$this->dom->_charset = 'UTF-8'; |
| 481 |
$this->dom->_target_charset = 'UTF-8'; |
| 482 |
|
| 483 |
# PERFORMANCE OPTIMIZATION: Pre-cache commonly accessed DOM elements |
| 484 |
# This eliminates 8-10 full DOM traversals per page |
| 485 |
$this->cache_elements(); |
| 486 |
|
| 487 |
# PERFORMANCE OPTIMIZATION: Enable deferred reload to skip intermediate reloads |
| 488 |
# This reduces 6-10 DOM serialize/deserialize cycles to just 1 final reload |
| 489 |
$this->deferred_reload = true; |
| 490 |
|
| 491 |
# now lets do the magic |
| 492 |
|
| 493 |
# COMPATIBILITY FIX: Transform 'value' to 'recommended_value' if needed |
| 494 |
# Some API versions return 'value' instead of 'recommended_value' |
| 495 |
if (!empty($replacement_data['header_replacements'])) { |
| 496 |
foreach ($replacement_data['header_replacements'] as &$item) { |
| 497 |
if (isset($item['value']) && !isset($item['recommended_value'])) { |
| 498 |
$item['recommended_value'] = $item['value']; |
| 499 |
} |
| 500 |
} |
| 501 |
unset($item); // Break reference |
| 502 |
} |
| 503 |
|
| 504 |
# start the header html insertion |
| 505 |
$this->insert_header_html($replacement_data); |
| 506 |
|
| 507 |
# now we do the header replacements |
| 508 |
$this->do_header_replacements($replacement_data); |
| 509 |
|
| 510 |
# now do the body replacements |
| 511 |
$this->do_body_replacements($replacement_data); |
| 512 |
|
| 513 |
# now do the footer insertions |
| 514 |
$this->do_footer_html_insertion($replacement_data); |
| 515 |
|
| 516 |
# final cleanup: ensure metasync_optimized attribute is removed from AMP pages |
| 517 |
$this->cleanup_amp_metasync_attribute(); |
| 518 |
|
| 519 |
# CRITICAL FIX: SimpleHtmlDom save() doesn't persist outertext/innertext changes |
| 520 |
# Use the same manual string replacement approach as process_html_directly |
| 521 |
$this->deferred_reload = false; |
| 522 |
|
| 523 |
# Get the HTML as string |
| 524 |
$result_html = $this->dom->save(); |
| 525 |
|
| 526 |
# Apply manual replacements (same logic as process_html_directly) |
| 527 |
|
| 528 |
# Apply header replacements manually |
| 529 |
if (!empty($replacement_data['header_replacements'])) { |
| 530 |
foreach ($replacement_data['header_replacements'] as $item) { |
| 531 |
$type = $item['type'] ?? ''; |
| 532 |
$value = $item['recommended_value'] ?? $item['value'] ?? ''; |
| 533 |
|
| 534 |
if (empty($value)) continue; |
| 535 |
|
| 536 |
if ($type === 'title') { |
| 537 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 538 |
$title_tag = '<title>' . $new_value . '</title>'; |
| 539 |
$result_html = preg_replace_callback('/<title[^>]*>.*?<\/title>/is', function ($m) use ($title_tag) { |
| 540 |
return $title_tag; |
| 541 |
}, $result_html, 1); |
| 542 |
} elseif ($type === 'meta') { |
| 543 |
$name = $item['name'] ?? ''; |
| 544 |
$property = $item['property'] ?? ''; |
| 545 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 546 |
|
| 547 |
if (!empty($name)) { |
| 548 |
$pattern = '/<meta\s+(?:name\s*=\s*["\']' . preg_quote($name, '/') . '["\']\s+content\s*=\s*["\'][^"\']*["\']|content\s*=\s*["\'][^"\']*["\']\s+name\s*=\s*["\']' . preg_quote($name, '/') . '["\'])\s*\/?>/i'; |
| 549 |
$replacement = '<meta name="' . htmlspecialchars($name, ENT_QUOTES, 'UTF-8') . '" content="' . $new_value . '">'; |
| 550 |
$count = 0; |
| 551 |
$result_html = preg_replace_callback($pattern, function ($m) use ($replacement) { |
| 552 |
return $replacement; |
| 553 |
}, $result_html, -1, $count); |
| 554 |
|
| 555 |
if ($count === 0) { |
| 556 |
$result_html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($replacement) { |
| 557 |
return $m[1] . "\n" . $replacement; |
| 558 |
}, $result_html, 1); |
| 559 |
} |
| 560 |
} elseif (!empty($property)) { |
| 561 |
$pattern = '/<meta\s+property\s*=\s*["\']' . preg_quote($property, '/') . '["\']\s+content\s*=\s*["\'][^"\']*["\']\s*\/?>/i'; |
| 562 |
$replacement = '<meta property="' . htmlspecialchars($property, ENT_QUOTES, 'UTF-8') . '" content="' . $new_value . '">'; |
| 563 |
$count = 0; |
| 564 |
$result_html = preg_replace_callback($pattern, function ($m) use ($replacement) { |
| 565 |
return $replacement; |
| 566 |
}, $result_html, -1, $count); |
| 567 |
|
| 568 |
if ($count === 0) { |
| 569 |
$result_html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($replacement) { |
| 570 |
return $m[1] . "\n" . $replacement; |
| 571 |
}, $result_html, 1); |
| 572 |
} |
| 573 |
} |
| 574 |
} elseif ($type === 'h1' || $type === 'heading') { |
| 575 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 576 |
$result_html = preg_replace_callback( |
| 577 |
'/<h1([^>]*)>.*?<\/h1>/is', |
| 578 |
function ($m) use ($new_value) { |
| 579 |
return '<h1' . $m[1] . '>' . $new_value . '</h1>'; |
| 580 |
}, |
| 581 |
$result_html, |
| 582 |
1 |
| 583 |
); |
| 584 |
} |
| 585 |
} |
| 586 |
} |
| 587 |
|
| 588 |
# String-based heading fallback for body_substitutions |
| 589 |
# DOM changes via SimpleHtmlDom don't persist on Divi/page-builder sites |
| 590 |
if (!empty($replacement_data['body_substitutions']['headings']) && is_array($replacement_data['body_substitutions']['headings'])) { |
| 591 |
foreach ($replacement_data['body_substitutions']['headings'] as $heading) { |
| 592 |
if (empty($heading['type']) || empty($heading['current_value']) || empty($heading['recommended_value'])) { |
| 593 |
continue; |
| 594 |
} |
| 595 |
|
| 596 |
$heading_type = preg_quote($heading['type'], '/'); |
| 597 |
$current_value = trim(preg_replace('/\s+/', ' ', html_entity_decode($heading['current_value'], ENT_QUOTES, 'UTF-8'))); |
| 598 |
$recommended_value = htmlspecialchars($heading['recommended_value'], ENT_QUOTES, 'UTF-8'); |
| 599 |
|
| 600 |
$result_html = preg_replace_callback( |
| 601 |
'/(<' . $heading_type . '(?:\s[^>]*)?>)(.*?)(<\/' . $heading_type . '>)/is', |
| 602 |
function ($m) use ($current_value, $recommended_value) { |
| 603 |
$inner_text = trim(preg_replace('/\s+/', ' ', html_entity_decode(strip_tags($m[2]), ENT_QUOTES, 'UTF-8'))); |
| 604 |
if ($inner_text === $current_value) { |
| 605 |
return $m[1] . $recommended_value . $m[3]; |
| 606 |
} |
| 607 |
return $m[0]; |
| 608 |
}, |
| 609 |
$result_html, |
| 610 |
-1 |
| 611 |
); |
| 612 |
} |
| 613 |
} |
| 614 |
|
| 615 |
# CRITICAL FIX: Apply image alt text manually via string replacement |
| 616 |
# DOM changes via SimpleHtmlDom don't persist on Oxygen/page-builder sites using HTTP render path |
| 617 |
$result_html = $this->apply_image_alt_text_via_string($result_html, $replacement_data); |
| 618 |
|
| 619 |
# Apply insertions — only if DOM insertion didn't already apply it |
| 620 |
if (!empty($replacement_data['header_html_insertion'])) { |
| 621 |
$header_html_check = trim($replacement_data['header_html_insertion']); |
| 622 |
if (strpos($result_html, $header_html_check) === false) { |
| 623 |
$header_html_insertion = preg_replace( |
| 624 |
'/<script(\s[^>]*)type\s*=\s*(["\'])application\/ld\+json\2/i', |
| 625 |
'<script$1type=$2application/ld+json$2 data-otto="true"', |
| 626 |
$replacement_data['header_html_insertion'] |
| 627 |
); |
| 628 |
$safe_header = str_replace(array('\\', '$'), array('\\\\', '\\$'), $header_html_insertion); |
| 629 |
$result_html = preg_replace('/(<\/head>)/i', $safe_header . "\n" . '$1', $result_html, 1); |
| 630 |
} |
| 631 |
} |
| 632 |
if (!empty($replacement_data['body_top_html_insertion'])) { |
| 633 |
$body_top_check = trim($replacement_data['body_top_html_insertion']); |
| 634 |
if (strpos($result_html, $body_top_check) === false) { |
| 635 |
$safe_body_top = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['body_top_html_insertion']); |
| 636 |
$result_html = preg_replace('/(<body[^>]*>)/i', '$1' . "\n" . $safe_body_top, $result_html, 1); |
| 637 |
} |
| 638 |
} |
| 639 |
if (!empty($replacement_data['body_bottom_html_insertion'])) { |
| 640 |
$body_bottom_check = trim($replacement_data['body_bottom_html_insertion']); |
| 641 |
if (strpos($result_html, $body_bottom_check) === false) { |
| 642 |
$safe_body_bottom = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['body_bottom_html_insertion']); |
| 643 |
$result_html = preg_replace('/(<\/body>)/i', $safe_body_bottom . "\n" . '$1', $result_html, 1); |
| 644 |
} |
| 645 |
} |
| 646 |
if (!empty($replacement_data['footer_html_insertion'])) { |
| 647 |
$safe_footer = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['footer_html_insertion']); |
| 648 |
$result_html = preg_replace('/(<\/html>)/i', $safe_footer . "\n" . '$1', $result_html, 1); |
| 649 |
} |
| 650 |
|
| 651 |
# DEDUPLICATION: Remove duplicate <title>, OG, Twitter tags, canonical, and JSON-LD schema |
| 652 |
$result_html = $this->deduplicate_title_tags($result_html); |
| 653 |
$result_html = $this->deduplicate_og_twitter_tags($result_html); |
| 654 |
$result_html = $this->deduplicate_schema_tags($result_html); |
| 655 |
$result_html = $this->deduplicate_canonical_tags($result_html); |
| 656 |
|
| 657 |
# Ensure metasync_optimized attribute on <head> (post-serialization so dom->clear() can't wipe it) |
| 658 |
if (!$this->is_amp_page() && strpos($result_html, 'metasync_optimized') === false) { |
| 659 |
$result_html = preg_replace('/<head(\s|>)/i', '<head metasync_optimized$1', $result_html, 1); |
| 660 |
} |
| 661 |
|
| 662 |
$result_html = $this->restore_case_sensitive_attributes($result_html); |
| 663 |
|
| 664 |
# WP-355 / WP-315: Re-inject any <style> blocks lost during processing. |
| 665 |
$result_html = $this->restore_lost_style_blocks($result_html, $original_style_blocks); |
| 666 |
|
| 667 |
# WP-315: Fix Divi 5 shortcode framework class renumbering (HTTP path) |
| 668 |
$result_html = $this->fix_divi_class_renumbering($result_html); |
| 669 |
|
| 670 |
# WP-315: Clean OTTO internal fetch params from HTML output. |
| 671 |
# The HTTP render uses wp_remote_get with ?is_otto_page_fetch=1&otto_block_title=1&otto_block_desc=1 |
| 672 |
# These leak into form action URLs, canonical links, etc. in the rendered HTML. |
| 673 |
$result_html = $this->clean_otto_fetch_params($result_html); |
| 674 |
|
| 675 |
return $result_html; |
| 676 |
} |
| 677 |
|
| 678 |
# do the footer html insertion |
| 679 |
function do_footer_html_insertion($replacement_data){ |
| 680 |
|
| 681 |
# check that we have footer html |
| 682 |
if(empty($replacement_data['footer_html_insertion'])){ |
| 683 |
return; |
| 684 |
} |
| 685 |
|
| 686 |
# OPTIMIZED: Use cached element instead of DOM traversal |
| 687 |
$footer = $this->get_cached_element('footer'); |
| 688 |
|
| 689 |
# check that footer is object |
| 690 |
if(!is_object($footer) || !isset($footer->innertext, $footer->outertext)){ |
| 691 |
return; |
| 692 |
} |
| 693 |
|
| 694 |
# get the tag attributes |
| 695 |
$attributes_string = $this->get_tag_attributes($footer); |
| 696 |
|
| 697 |
# now do the actual html replacements |
| 698 |
$footer->outertext = '<footer' . $attributes_string . '>' . $footer->innertext . $replacement_data['footer_html_insertion'].'</footer>'; |
| 699 |
|
| 700 |
# save the document |
| 701 |
$this->save_reload(); |
| 702 |
} |
| 703 |
|
| 704 |
# do body replacements |
| 705 |
function do_body_replacements($replacement_data){ |
| 706 |
|
| 707 |
# start body top html replacements |
| 708 |
$this->do_body_top_html($replacement_data); |
| 709 |
|
| 710 |
# start the body bottom html replacements |
| 711 |
$this->do_body_bottom_html($replacement_data); |
| 712 |
|
| 713 |
# do the body substitutions |
| 714 |
$this->do_body_substitutions($replacement_data); |
| 715 |
|
| 716 |
# Check if the feature is enabled in general settings (Post/Page Editor Settings) |
| 717 |
$general_settings = get_option('metasync_options')['general'] ?? []; |
| 718 |
if (!empty($general_settings['open_external_links']) && $general_settings['open_external_links'] == '1') { |
| 719 |
$this->add_target_blank_to_external_links(); |
| 720 |
} |
| 721 |
|
| 722 |
# Check if the feature is enabled in seo_controls settings (Indexation Control) |
| 723 |
$seo_controls = get_option('metasync_options')['seo_controls'] ?? []; |
| 724 |
if (!empty($seo_controls['add_nofollow_to_external_links']) && $seo_controls['add_nofollow_to_external_links'] === 'true') { |
| 725 |
$this->add_nofollow_to_external_links(); |
| 726 |
} |
| 727 |
} |
| 728 |
|
| 729 |
# body substitutions data |
| 730 |
function do_body_substitutions($replacement_data){ |
| 731 |
|
| 732 |
# check that we have an array of substitutions |
| 733 |
if(empty($replacement_data['body_substitutions']) || !is_array($replacement_data['body_substitutions'])){ |
| 734 |
return; |
| 735 |
} |
| 736 |
|
| 737 |
# now work on different substitution keys |
| 738 |
foreach ($replacement_data['body_substitutions'] as $key => $value) { |
| 739 |
|
| 740 |
# check key categories |
| 741 |
if($key == 'images'){ |
| 742 |
# do image replacements |
| 743 |
$this->handle_images($value); |
| 744 |
} |
| 745 |
elseif($key == 'headings'){ |
| 746 |
# do heading repalcements |
| 747 |
$this->do_heading_body_substitutions($value); |
| 748 |
} |
| 749 |
elseif($key == 'links'){ |
| 750 |
# do link replacements |
| 751 |
$this->do_link_body_substitutions($value); |
| 752 |
} |
| 753 |
|
| 754 |
} |
| 755 |
|
| 756 |
# save the document |
| 757 |
$this->save_reload(); |
| 758 |
|
| 759 |
} |
| 760 |
|
| 761 |
/** |
| 762 |
* START BODY SUBSTITUTION FUNCTIONS |
| 763 |
* @see do_body_substitutions(); |
| 764 |
*/ |
| 765 |
|
| 766 |
# image substitions |
| 767 |
function handle_images($image_data){ |
| 768 |
|
| 769 |
if (empty($image_data) || !is_array($image_data)) { |
| 770 |
return; |
| 771 |
} |
| 772 |
|
| 773 |
# OPTIMIZED: Use cached images instead of DOM traversal |
| 774 |
$images = $this->get_cached_element('imgs', []); |
| 775 |
|
| 776 |
if (empty($images)) { |
| 777 |
return; |
| 778 |
} |
| 779 |
|
| 780 |
# WP-425: index suggestions by URL path as well, so images authored with |
| 781 |
# relative srcs still match OTTO's absolute-URL keys (and vice versa). |
| 782 |
$alt_lookup = $this->build_image_alt_lookup($image_data); |
| 783 |
|
| 784 |
# PERFORMANCE OPTIMIZATION: O(n²) reduced to O(n) |
| 785 |
# Single pass with hash map lookup instead of nested loop |
| 786 |
foreach($images AS $key => $image){ |
| 787 |
# Get image src |
| 788 |
$image_src = $image->src; |
| 789 |
|
| 790 |
if (empty($image_src)) { |
| 791 |
continue; |
| 792 |
} |
| 793 |
|
| 794 |
# Hash map lookup O(1) instead of loop O(n) |
| 795 |
$alt_text = $this->lookup_image_alt($alt_lookup, $image_src); |
| 796 |
if ($alt_text !== null) { |
| 797 |
# Set alt text - Note: This may not persist in all cases |
| 798 |
# Manual string replacement in process_html_directly() ensures it's applied |
| 799 |
$new_alt = htmlspecialchars($alt_text, ENT_QUOTES, 'UTF-8'); |
| 800 |
|
| 801 |
# Get current img tag HTML and update alt attribute |
| 802 |
$current_html = $image->outertext; |
| 803 |
|
| 804 |
# Remove existing alt attribute (if any) |
| 805 |
$updated_html = preg_replace('/\s+alt=(["\'])[^"\']*\1/', '', $current_html); |
| 806 |
|
| 807 |
# Insert new alt attribute after the opening <img |
| 808 |
$updated_html = preg_replace('/^<img\s/', '<img alt="' . str_replace('$', '\\$', $new_alt) . '" ', $updated_html); |
| 809 |
|
| 810 |
# Update the element |
| 811 |
$image->outertext = $updated_html; |
| 812 |
|
| 813 |
$multi_view_attr = $image->getAttribute('data-et-multi-view'); |
| 814 |
if (!empty($multi_view_attr)) { |
| 815 |
$this->update_divi_multi_view_alt($image, $alt_text); |
| 816 |
} |
| 817 |
} |
| 818 |
} |
| 819 |
} |
| 820 |
|
| 821 |
# WP-425: memoized site host — home_url() runs its filter chain on every |
| 822 |
# call and these helpers run per suggestion and per img. |
| 823 |
private ?string $site_host_cache = null; |
| 824 |
|
| 825 |
/** |
| 826 |
* Get this site's host (lowercase) for same-host URL comparisons. |
| 827 |
* |
| 828 |
* @return string |
| 829 |
*/ |
| 830 |
private function get_site_host() { |
| 831 |
if ($this->site_host_cache === null) { |
| 832 |
$host = wp_parse_url(home_url(), PHP_URL_HOST); |
| 833 |
$this->site_host_cache = is_string($host) ? strtolower($host) : ''; |
| 834 |
} |
| 835 |
return $this->site_host_cache; |
| 836 |
} |
| 837 |
|
| 838 |
/** |
| 839 |
* WP-425: Candidate lookup keys for an image URL — the URL itself plus, |
| 840 |
* for relative URLs or absolute/protocol-relative URLs on this site's |
| 841 |
* host, the bare URL path. URLs on a different host only ever match |
| 842 |
* exactly, so a same path on a foreign host cannot produce a false match. |
| 843 |
* |
| 844 |
* @param string $url Image URL (absolute, protocol-relative, or relative) |
| 845 |
* @return array Candidate keys, original URL first |
| 846 |
*/ |
| 847 |
private function get_image_url_variants($url) { |
| 848 |
$variants = [$url]; |
| 849 |
|
| 850 |
if (!is_string($url) || $url === '') { |
| 851 |
return $variants; |
| 852 |
} |
| 853 |
|
| 854 |
$host = wp_parse_url($url, PHP_URL_HOST); |
| 855 |
if (!empty($host) && strtolower($host) !== $this->get_site_host()) { |
| 856 |
# Foreign host: exact match only |
| 857 |
return $variants; |
| 858 |
} |
| 859 |
|
| 860 |
$path = wp_parse_url($url, PHP_URL_PATH); |
| 861 |
if (is_string($path) && $path !== '' && $path !== $url) { |
| 862 |
$variants[] = $path; |
| 863 |
} |
| 864 |
|
| 865 |
return $variants; |
| 866 |
} |
| 867 |
|
| 868 |
/** |
| 869 |
* WP-425: Build the image alt lookup keyed by every URL variant of each |
| 870 |
* OTTO suggestion, so relative img srcs match absolute suggestion URLs. |
| 871 |
* |
| 872 |
* @param array $image_data OTTO body_substitutions.images map (url => alt) |
| 873 |
* @return array Expanded lookup (url-or-path => alt) |
| 874 |
*/ |
| 875 |
private function build_image_alt_lookup($image_data) { |
| 876 |
$lookup = []; |
| 877 |
|
| 878 |
foreach ($image_data as $image_url => $alt_text) { |
| 879 |
foreach ($this->get_image_url_variants($image_url) as $variant) { |
| 880 |
# First suggestion wins when two keys collapse to the same |
| 881 |
# path (e.g. a relative and an absolute key for one image); |
| 882 |
# exact-src matches still take priority in lookup_image_alt(). |
| 883 |
if (!isset($lookup[$variant])) { |
| 884 |
$lookup[$variant] = $alt_text; |
| 885 |
} |
| 886 |
} |
| 887 |
} |
| 888 |
|
| 889 |
return $lookup; |
| 890 |
} |
| 891 |
|
| 892 |
/** |
| 893 |
* WP-425: Resolve the OTTO alt text for an img src, trying the exact src |
| 894 |
* first and then its same-host path variant. |
| 895 |
* |
| 896 |
* @param array $lookup Expanded lookup from build_image_alt_lookup() |
| 897 |
* @param string $image_src The img element's src attribute |
| 898 |
* @return string|null Alt text, or null when no suggestion matches |
| 899 |
*/ |
| 900 |
private function lookup_image_alt($lookup, $image_src) { |
| 901 |
foreach ($this->get_image_url_variants($image_src) as $variant) { |
| 902 |
if (isset($lookup[$variant])) { |
| 903 |
return $lookup[$variant]; |
| 904 |
} |
| 905 |
} |
| 906 |
|
| 907 |
return null; |
| 908 |
} |
| 909 |
|
| 910 |
/** |
| 911 |
* Update Divi's multi-view data attribute with alt text |
| 912 |
* Divi stores image attributes in a JSON structure within data-et-multi-view |
| 913 |
* |
| 914 |
* @param object $image The image DOM element |
| 915 |
* @param string $alt_text The alt text to set |
| 916 |
*/ |
| 917 |
private function update_divi_multi_view_alt($image, $alt_text) { |
| 918 |
$multi_view_attr = $image->getAttribute('data-et-multi-view'); |
| 919 |
|
| 920 |
if (!empty($multi_view_attr)) { |
| 921 |
try { |
| 922 |
# Decode the JSON |
| 923 |
$multi_view_data = json_decode($multi_view_attr, true); |
| 924 |
|
| 925 |
if ($multi_view_data && isset($multi_view_data['schema']['attrs'])) { |
| 926 |
# Update alt in desktop view |
| 927 |
if (isset($multi_view_data['schema']['attrs']['desktop'])) { |
| 928 |
$multi_view_data['schema']['attrs']['desktop']['alt'] = $alt_text; |
| 929 |
} |
| 930 |
|
| 931 |
# Update alt in other views if they exist (phone, tablet, etc.) |
| 932 |
foreach ($multi_view_data['schema']['attrs'] as $view => $attrs) { |
| 933 |
if (isset($attrs['alt'])) { |
| 934 |
$multi_view_data['schema']['attrs'][$view]['alt'] = $alt_text; |
| 935 |
} |
| 936 |
} |
| 937 |
|
| 938 |
# Encode back to JSON and update the attribute |
| 939 |
$updated_json = json_encode($multi_view_data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE); |
| 940 |
$image->setAttribute('data-et-multi-view', $updated_json); |
| 941 |
|
| 942 |
# error_log('MetaSync OTTO DEBUG: Updated Divi multi-view alt text'); |
| 943 |
} |
| 944 |
} catch (Exception $e) { |
| 945 |
# If JSON decode fails, just log and continue |
| 946 |
# error_log('MetaSync OTTO DEBUG: Failed to update Divi multi-view - ' . $e->getMessage()); |
| 947 |
} |
| 948 |
} |
| 949 |
} |
| 950 |
|
| 951 |
/** |
| 952 |
* Apply image alt text via string replacement on raw HTML. |
| 953 |
* Used as a fallback when DOM changes don't persist (Oxygen, Divi, page-builder sites). |
| 954 |
* |
| 955 |
* Known limitation (pre-existing): the `<img[^>]*src=` pattern can match a |
| 956 |
* `data-src`/`data-lazy-src` attribute that ends with the target URL before |
| 957 |
* the real `src`, mis-attributing the alt on lazy-loaded markup. Tracked as |
| 958 |
* a follow-up to WP-425. |
| 959 |
* |
| 960 |
* @param string $html The HTML to process |
| 961 |
* @param array $replacement_data The replacement data containing body_substitutions.images |
| 962 |
* @return string Modified HTML with alt text applied |
| 963 |
*/ |
| 964 |
private function apply_image_alt_text_via_string($html, $replacement_data) { |
| 965 |
if (empty($replacement_data['body_substitutions']['images']) || !is_array($replacement_data['body_substitutions']['images'])) { |
| 966 |
return $html; |
| 967 |
} |
| 968 |
|
| 969 |
foreach ($replacement_data['body_substitutions']['images'] as $image_url => $alt_text) { |
| 970 |
# WP-425: prefilter and match on the same-host path variant too, so |
| 971 |
# relative img srcs match OTTO's absolute suggestion URLs. |
| 972 |
$src_pattern = $this->get_image_src_pattern($image_url); |
| 973 |
|
| 974 |
if (empty($alt_text) || $src_pattern === null || strpos($html, $src_pattern['needle']) === false) { |
| 975 |
continue; |
| 976 |
} |
| 977 |
|
| 978 |
$escaped_alt = htmlspecialchars($alt_text, ENT_QUOTES, 'UTF-8'); |
| 979 |
$img_pattern = '/<img[^>]*src=["\']' . $src_pattern['pattern'] . '["\'][^>]*>/i'; |
| 980 |
|
| 981 |
if (preg_match_all($img_pattern, $html, $img_matches)) { |
| 982 |
foreach ($img_matches[0] as $original_img) { |
| 983 |
if (strpos($original_img, $escaped_alt) !== false) { |
| 984 |
continue; |
| 985 |
} |
| 986 |
|
| 987 |
# Remove ALL existing alt attributes |
| 988 |
$new_img = preg_replace('/\s+alt\s*=\s*(["\'])[^"\']*\1/i', '', $original_img); |
| 989 |
$new_img = preg_replace('/<img\s+alt\s*=\s*(["\'])[^"\']*\1\s*/i', '<img ', $new_img); |
| 990 |
|
| 991 |
# Add single alt attribute after <img |
| 992 |
$new_img = preg_replace('/^<img\s*/i', '<img alt="' . str_replace('$', '\\$', $escaped_alt) . '" ', $new_img); |
| 993 |
|
| 994 |
# Update data-et-multi-view JSON if present (Divi theme) |
| 995 |
if (strpos($new_img, 'data-et-multi-view') !== false) { |
| 996 |
$new_img = preg_replace_callback( |
| 997 |
'/data-et-multi-view="([^"]+)"/i', |
| 998 |
function($mv_matches) use ($alt_text) { |
| 999 |
$json_str = html_entity_decode($mv_matches[1], ENT_QUOTES, 'UTF-8'); |
| 1000 |
$json_data = json_decode($json_str, true); |
| 1001 |
|
| 1002 |
if ($json_data && isset($json_data['schema']['attrs'])) { |
| 1003 |
foreach ($json_data['schema']['attrs'] as &$attrs) { |
| 1004 |
if (array_key_exists('alt', $attrs)) { |
| 1005 |
$attrs['alt'] = $alt_text; |
| 1006 |
} |
| 1007 |
} |
| 1008 |
unset($attrs); |
| 1009 |
$new_json = json_encode($json_data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE); |
| 1010 |
return 'data-et-multi-view="' . str_replace('"', '"', $new_json) . '"'; |
| 1011 |
} |
| 1012 |
return $mv_matches[0]; |
| 1013 |
}, |
| 1014 |
$new_img |
| 1015 |
); |
| 1016 |
} |
| 1017 |
|
| 1018 |
$html = str_replace($original_img, $new_img, $html); |
| 1019 |
} |
| 1020 |
} |
| 1021 |
} |
| 1022 |
|
| 1023 |
return $html; |
| 1024 |
} |
| 1025 |
|
| 1026 |
/** |
| 1027 |
* WP-425: Build the src match for an OTTO image suggestion URL. |
| 1028 |
* |
| 1029 |
* For relative URLs or absolute/protocol-relative URLs on this site's |
| 1030 |
* host, the regex matches the URL path with an optional scheme+host |
| 1031 |
* prefix — so a relative img src matches an absolute suggestion URL and |
| 1032 |
* vice versa. URLs on a foreign host keep the original exact match, so a |
| 1033 |
* same path on a different host cannot produce a false positive. |
| 1034 |
* |
| 1035 |
* @param string $image_url OTTO suggestion URL |
| 1036 |
* @return array|null ['needle' => strpos prefilter string, |
| 1037 |
* 'pattern' => regex fragment ('/' delimiter)] or null |
| 1038 |
*/ |
| 1039 |
private function get_image_src_pattern($image_url) { |
| 1040 |
if (!is_string($image_url) || $image_url === '') { |
| 1041 |
return null; |
| 1042 |
} |
| 1043 |
|
| 1044 |
$host = wp_parse_url($image_url, PHP_URL_HOST); |
| 1045 |
$site_host = $this->get_site_host(); |
| 1046 |
|
| 1047 |
# Foreign host: exact match only |
| 1048 |
if (!empty($host) && strtolower($host) !== $site_host) { |
| 1049 |
return ['needle' => $image_url, 'pattern' => preg_quote($image_url, '/')]; |
| 1050 |
} |
| 1051 |
|
| 1052 |
$path = wp_parse_url($image_url, PHP_URL_PATH); |
| 1053 |
if (!is_string($path) || $path === '') { |
| 1054 |
return ['needle' => $image_url, 'pattern' => preg_quote($image_url, '/')]; |
| 1055 |
} |
| 1056 |
|
| 1057 |
$prefix = $site_host !== '' ? '(?:(?:https?:)?\/\/' . preg_quote($site_host, '/') . ')?' : ''; |
| 1058 |
|
| 1059 |
# Allow an optional query/fragment after the path (e.g. WordPress |
| 1060 |
# `?ver=` cache-busters) so src="/up/a.jpg?ver=2" still matches. |
| 1061 |
return ['needle' => $path, 'pattern' => $prefix . preg_quote($path, '/') . '(?:[?#][^"\']*)?']; |
| 1062 |
} |
| 1063 |
|
| 1064 |
# heading substitutions |
| 1065 |
function do_heading_body_substitutions($heading_data){ |
| 1066 |
|
| 1067 |
# loop all data |
| 1068 |
foreach($heading_data AS $okey => $heading){ |
| 1069 |
|
| 1070 |
# find all occurences of the heading type |
| 1071 |
$occurences = $this->dom->find($heading['type']); |
| 1072 |
|
| 1073 |
# loop all occurences |
| 1074 |
foreach ($occurences as $ikey => $heading_old) { |
| 1075 |
|
| 1076 |
# get the header text |
| 1077 |
$text = $heading_old->text(); |
| 1078 |
|
| 1079 |
# check matching text |
| 1080 |
if(trim($heading['current_value'] ?? '') == trim($text ?? '')){ |
| 1081 |
|
| 1082 |
# replace entire tag output — preserves attributes, removes all children (elements + text nodes) |
| 1083 |
$outer = $heading_old->outertext; |
| 1084 |
$open_end = strpos($outer, '>'); |
| 1085 |
if ($open_end !== false) { |
| 1086 |
$open_tag = substr($outer, 0, $open_end + 1); |
| 1087 |
$close_tag = '</' . $heading['type'] . '>'; |
| 1088 |
$heading_old->outertext = $open_tag . $heading['recommended_value'] . $close_tag; |
| 1089 |
} |
| 1090 |
|
| 1091 |
} |
| 1092 |
} |
| 1093 |
|
| 1094 |
} |
| 1095 |
|
| 1096 |
# save and reload the dom |
| 1097 |
$this->save_reload(); |
| 1098 |
} |
| 1099 |
|
| 1100 |
# link replacements |
| 1101 |
function do_link_body_substitutions($swap_data){ |
| 1102 |
|
| 1103 |
# find all links in the body |
| 1104 |
$links = $this->dom->find('a'); |
| 1105 |
|
| 1106 |
# loop all links check if we have them in swap data |
| 1107 |
foreach ($links as $key => $link) { |
| 1108 |
|
| 1109 |
# check if link matches |
| 1110 |
if(!empty($swap_data[$link->href])){ |
| 1111 |
|
| 1112 |
# replace the link |
| 1113 |
$link->href = $swap_data[$link->href]; |
| 1114 |
} |
| 1115 |
|
| 1116 |
} |
| 1117 |
} |
| 1118 |
|
| 1119 |
/** |
| 1120 |
* Add rel="nofollow" attribute to external links |
| 1121 |
* Works for both buffer and HTTP rendering methods |
| 1122 |
* Only processes links that don't already have rel="nofollow" |
| 1123 |
*/ |
| 1124 |
function add_nofollow_to_external_links(){ |
| 1125 |
|
| 1126 |
# find all links in the body |
| 1127 |
$links = $this->dom->find('a'); |
| 1128 |
|
| 1129 |
# get home URL for comparison |
| 1130 |
$home_url = rtrim(home_url(), '/'); |
| 1131 |
$home_url_lower = strtolower($home_url); |
| 1132 |
|
| 1133 |
# loop through all links |
| 1134 |
foreach ($links as $key => $link) { |
| 1135 |
|
| 1136 |
# get the href attribute |
| 1137 |
$href = $link->href ?? ''; |
| 1138 |
|
| 1139 |
# skip if href is empty |
| 1140 |
if (empty($href)) { |
| 1141 |
continue; |
| 1142 |
} |
| 1143 |
|
| 1144 |
# check if link is external |
| 1145 |
if ($this->is_external_link($href, $home_url, $home_url_lower)) { |
| 1146 |
# get existing rel attribute |
| 1147 |
$existing_rel = $link->rel ?? ''; |
| 1148 |
|
| 1149 |
# check if nofollow already exists |
| 1150 |
if (empty($existing_rel)) { |
| 1151 |
# no rel attribute, add nofollow |
| 1152 |
$link->rel = 'nofollow'; |
| 1153 |
} elseif (strpos($existing_rel, 'nofollow') === false) { |
| 1154 |
# rel exists but nofollow is not present, add it |
| 1155 |
$link->rel = trim($existing_rel . ' nofollow'); |
| 1156 |
} |
| 1157 |
# if nofollow already exists, do nothing |
| 1158 |
} |
| 1159 |
} |
| 1160 |
|
| 1161 |
# Note: No need to call save_reload() here - it's called at the end of process_html_directly() |
| 1162 |
# This avoids redundant DOM save/reload operations and improves performance |
| 1163 |
} |
| 1164 |
|
| 1165 |
/** |
| 1166 |
* Check if a URL is external |
| 1167 |
* |
| 1168 |
* @param string $url The URL to check |
| 1169 |
* @param string $home_url The home URL without trailing slash |
| 1170 |
* @param string $home_url_lower Lowercase version of home URL |
| 1171 |
* @return bool True if external, false if internal |
| 1172 |
*/ |
| 1173 |
private function is_external_link($url, $home_url, $home_url_lower) { |
| 1174 |
# Empty or anchor-only links are internal |
| 1175 |
if (empty($url) || $url === '#' || strpos($url, '#') === 0) { |
| 1176 |
return false; |
| 1177 |
} |
| 1178 |
|
| 1179 |
# Relative URLs (starting with /) are internal |
| 1180 |
if (strpos($url, '/') === 0 && strpos($url, '//') !== 0) { |
| 1181 |
return false; |
| 1182 |
} |
| 1183 |
|
| 1184 |
# Check if URL starts with home URL (case-insensitive) |
| 1185 |
$url_lower = strtolower($url); |
| 1186 |
if (strpos($url_lower, $home_url_lower) === 0) { |
| 1187 |
return false; |
| 1188 |
} |
| 1189 |
|
| 1190 |
# If it's a protocol-relative URL (//example.com), check if it matches our domain |
| 1191 |
if (strpos($url, '//') === 0) { |
| 1192 |
$parsed_home = parse_url($home_url); |
| 1193 |
$parsed_link = parse_url($url); |
| 1194 |
|
| 1195 |
if (isset($parsed_home['host']) && isset($parsed_link['host'])) { |
| 1196 |
if (strtolower($parsed_home['host']) === strtolower($parsed_link['host'])) { |
| 1197 |
return false; |
| 1198 |
} |
| 1199 |
} |
| 1200 |
} |
| 1201 |
|
| 1202 |
# All other URLs are considered external |
| 1203 |
return true; |
| 1204 |
} |
| 1205 |
|
| 1206 |
/** |
| 1207 |
* Add target="_blank" attribute to external links |
| 1208 |
* Works for both buffer and HTTP rendering methods |
| 1209 |
* Only processes links that don't already have a target attribute |
| 1210 |
*/ |
| 1211 |
function add_target_blank_to_external_links(){ |
| 1212 |
|
| 1213 |
# find all links in the body |
| 1214 |
$links = $this->dom->find('a'); |
| 1215 |
|
| 1216 |
# get home URL for comparison |
| 1217 |
$home_url = rtrim(home_url(), '/'); |
| 1218 |
$home_url_lower = strtolower($home_url); |
| 1219 |
|
| 1220 |
# loop through all links |
| 1221 |
foreach ($links as $key => $link) { |
| 1222 |
|
| 1223 |
# get the href attribute |
| 1224 |
$href = $link->href ?? ''; |
| 1225 |
|
| 1226 |
# skip if href is empty or already has target attribute |
| 1227 |
if (empty($href) || !empty($link->target)) { |
| 1228 |
continue; |
| 1229 |
} |
| 1230 |
|
| 1231 |
# check if link is external |
| 1232 |
if ($this->is_external_link($href, $home_url, $home_url_lower)) { |
| 1233 |
# add target="_blank" attribute |
| 1234 |
$link->target = '_blank'; |
| 1235 |
|
| 1236 |
# add rel="noopener noreferrer" for security |
| 1237 |
$existing_rel = $link->rel ?? ''; |
| 1238 |
if (empty($existing_rel)) { |
| 1239 |
$link->rel = 'noopener noreferrer'; |
| 1240 |
} elseif (strpos($existing_rel, 'noopener') === false) { |
| 1241 |
$link->rel = trim($existing_rel . ' noopener noreferrer'); |
| 1242 |
} |
| 1243 |
} |
| 1244 |
} |
| 1245 |
|
| 1246 |
# Note: No need to call save_reload() here - it's called at the end of process_html_directly() |
| 1247 |
# This avoids redundant DOM save/reload operations and improves performance |
| 1248 |
} |
| 1249 |
|
| 1250 |
# body bottom html replacement code |
| 1251 |
function do_body_bottom_html($insert_data){ |
| 1252 |
|
| 1253 |
# check that data is availbale |
| 1254 |
if(empty($insert_data['body_bottom_html_insertion'])){ |
| 1255 |
return; |
| 1256 |
} |
| 1257 |
|
| 1258 |
# OPTIMIZED: Use cached body element |
| 1259 |
$body = $this->get_cached_element('body'); |
| 1260 |
|
| 1261 |
# set the link property if not empty |
| 1262 |
if(empty($body->outertext)){ |
| 1263 |
return; |
| 1264 |
} |
| 1265 |
|
| 1266 |
|
| 1267 |
# get the tag attributes |
| 1268 |
$attributes_string = $this->get_tag_attributes($body); |
| 1269 |
|
| 1270 |
# now do the actual html replacements |
| 1271 |
$body->outertext = '<body' . $attributes_string . '>' . $body->innertext . $insert_data['body_bottom_html_insertion'].'</body>'; |
| 1272 |
|
| 1273 |
# save the document |
| 1274 |
$this->save_reload(); |
| 1275 |
} |
| 1276 |
|
| 1277 |
# body top html replacement |
| 1278 |
function do_body_top_html($insert_data){ |
| 1279 |
|
| 1280 |
# check that data is availbale |
| 1281 |
if(empty($insert_data['body_top_html_insertion'])){ |
| 1282 |
return; |
| 1283 |
} |
| 1284 |
|
| 1285 |
# OPTIMIZED: Use cached body element |
| 1286 |
$body = $this->get_cached_element('body'); |
| 1287 |
|
| 1288 |
# get the tag attributes |
| 1289 |
$attributes_string = $this->get_tag_attributes($body); |
| 1290 |
|
| 1291 |
# now do the actual html replacements |
| 1292 |
$body->outertext = '<body' . $attributes_string . '>'.$insert_data['body_top_html_insertion'].$body->innertext . '</body>'; |
| 1293 |
} |
| 1294 |
|
| 1295 |
# this function does the header replacements |
| 1296 |
function do_header_replacements($replacement_data){ |
| 1297 |
|
| 1298 |
# check that we have header replacements |
| 1299 |
if(empty($replacement_data['header_replacements']) || !is_array($replacement_data['header_replacements'])){ |
| 1300 |
return; |
| 1301 |
} |
| 1302 |
|
| 1303 |
# Check for custom SEO values from MetaSync SEO Sidebar |
| 1304 |
# Custom values take absolute priority over OTTO suggestions |
| 1305 |
$custom_seo_title = ''; |
| 1306 |
$custom_seo_description = ''; |
| 1307 |
|
| 1308 |
if (function_exists('is_singular') && is_singular()) { |
| 1309 |
$post_id = get_the_ID(); |
| 1310 |
if ($post_id) { |
| 1311 |
$custom_seo_title = get_post_meta($post_id, '_metasync_seo_title', true); |
| 1312 |
$custom_seo_description = get_post_meta($post_id, '_metasync_seo_desc', true); |
| 1313 |
} |
| 1314 |
} |
| 1315 |
|
| 1316 |
# now lets do the replacement work |
| 1317 |
foreach($replacement_data['header_replacements'] AS $key => $data){ |
| 1318 |
|
| 1319 |
# skip cases where type is not specified |
| 1320 |
if(empty($data['type'])){ |
| 1321 |
continue; |
| 1322 |
} |
| 1323 |
|
| 1324 |
# handle title - skip if custom SEO title exists or no value |
| 1325 |
if($data['type'] == 'title'){ |
| 1326 |
if (!empty($custom_seo_title)) { |
| 1327 |
# Skip title replacement - custom SEO title takes priority |
| 1328 |
continue; |
| 1329 |
} |
| 1330 |
|
| 1331 |
# Skip if OTTO has no title value |
| 1332 |
if (empty(trim($data['recommended_value'] ?? $data['value'] ?? ''))) { |
| 1333 |
continue; |
| 1334 |
} |
| 1335 |
|
| 1336 |
# handle the title logic |
| 1337 |
$this->replace_title($data); |
| 1338 |
|
| 1339 |
# |
| 1340 |
continue; |
| 1341 |
} |
| 1342 |
|
| 1343 |
# handle canonical links |
| 1344 |
if($data['type'] == 'link' && $data['rel'] === 'canonical'){ |
| 1345 |
|
| 1346 |
# Protect manually-set canonical from OTTO override |
| 1347 |
if (function_exists('is_singular') && is_singular()) { |
| 1348 |
$post_id = get_the_ID(); |
| 1349 |
if ($post_id) { |
| 1350 |
$custom_canonical = get_post_meta($post_id, '_metasync_canonical_url', true); |
| 1351 |
if (empty($custom_canonical)) { |
| 1352 |
$custom_canonical = get_post_meta($post_id, 'meta_canonical', true); |
| 1353 |
// Handle legacy array values |
| 1354 |
if (is_array($custom_canonical)) { |
| 1355 |
$custom_canonical = reset($custom_canonical) ?: ''; |
| 1356 |
} |
| 1357 |
} |
| 1358 |
if (!empty($custom_canonical)) { |
| 1359 |
# Manual canonical takes priority — skip OTTO override |
| 1360 |
continue; |
| 1361 |
} |
| 1362 |
} |
| 1363 |
} |
| 1364 |
|
| 1365 |
# find the cannonical dom element |
| 1366 |
$link = $this->dom->find('link[rel="canonical"]', 0); |
| 1367 |
|
| 1368 |
# set the link property if not empty |
| 1369 |
if(!empty($link->href)){ |
| 1370 |
$link->href = $data['recommended_value'] ?? $link->href; |
| 1371 |
} |
| 1372 |
|
| 1373 |
# |
| 1374 |
continue; |
| 1375 |
} |
| 1376 |
|
| 1377 |
# work on other elemenets not titlte |
| 1378 |
$this->handle_meta_element($data); |
| 1379 |
} |
| 1380 |
} |
| 1381 |
|
| 1382 |
# function to handle meta elements other than title |
| 1383 |
function handle_meta_element($data){ |
| 1384 |
|
| 1385 |
# Skip if OTTO has no value to set - prevents overwriting existing tags (e.g. Yoast) |
| 1386 |
# with empty content when OTTO has no recommendation for this meta field |
| 1387 |
$recommended_value = $data['recommended_value'] ?? $data['value'] ?? ''; |
| 1388 |
if (empty(trim($recommended_value))) { |
| 1389 |
return; |
| 1390 |
} |
| 1391 |
|
| 1392 |
# Check if this is a description meta tag and if custom description exists |
| 1393 |
# Custom values take absolute priority over OTTO suggestions |
| 1394 |
$name = $data['name'] ?? false; |
| 1395 |
$property = $data['property'] ?? false; |
| 1396 |
|
| 1397 |
# Protect manually-set robots meta from OTTO override |
| 1398 |
# If the user set noindex via Common Robots Meta or meta_robots, honour it |
| 1399 |
if (!empty($name) && $name === 'robots') { |
| 1400 |
if (function_exists('is_singular') && is_singular()) { |
| 1401 |
$post_id = get_the_ID(); |
| 1402 |
if ($post_id) { |
| 1403 |
$manual_robots = get_post_meta($post_id, 'meta_robots', true); |
| 1404 |
if (!empty($manual_robots) && stripos($manual_robots, 'noindex') !== false) { |
| 1405 |
# Manual noindex takes priority — skip OTTO override |
| 1406 |
return; |
| 1407 |
} |
| 1408 |
$common_robots = get_post_meta($post_id, 'metasync_common_robots', true); |
| 1409 |
if (is_array($common_robots) && !empty($common_robots['noindex'])) { |
| 1410 |
# Common Robots Meta has noindex checked — skip OTTO override |
| 1411 |
return; |
| 1412 |
} |
| 1413 |
} |
| 1414 |
} |
| 1415 |
} |
| 1416 |
|
| 1417 |
# Check for custom SEO description |
| 1418 |
if (!empty($name) && $name === 'description') { |
| 1419 |
if (function_exists('is_singular') && is_singular()) { |
| 1420 |
$post_id = get_the_ID(); |
| 1421 |
if ($post_id) { |
| 1422 |
$custom_seo_description = get_post_meta($post_id, '_metasync_seo_desc', true); |
| 1423 |
if (!empty($custom_seo_description)) { |
| 1424 |
# Custom description exists, skip OTTO's suggestion |
| 1425 |
return; |
| 1426 |
} |
| 1427 |
} |
| 1428 |
} |
| 1429 |
} |
| 1430 |
|
| 1431 |
# extract property value |
| 1432 |
$property = $data['property'] ?? false; |
| 1433 |
|
| 1434 |
# extract name value |
| 1435 |
$name = $data['name'] ?? false; |
| 1436 |
|
| 1437 |
# set the selector |
| 1438 |
$meta_selector = ''; |
| 1439 |
|
| 1440 |
# extend selector |
| 1441 |
if(!empty($name)){ |
| 1442 |
$meta_selector .= 'meta[name="' . trim($name) . '"]'; |
| 1443 |
} |
| 1444 |
|
| 1445 |
# extent if property is defined |
| 1446 |
if(!empty($property)){ |
| 1447 |
if (!empty($meta_selector)) { |
| 1448 |
$meta_selector .= ','; |
| 1449 |
} |
| 1450 |
$meta_selector .= 'meta[property="' . trim($property) . '"]'; |
| 1451 |
} |
| 1452 |
|
| 1453 |
# find the meta gat in the dom |
| 1454 |
$meta_tag = $this->dom->find($meta_selector, 0); |
| 1455 |
|
| 1456 |
# if tag not exists add it |
| 1457 |
if(empty($meta_tag)){ |
| 1458 |
if($data['type'] == 'meta'){ |
| 1459 |
|
| 1460 |
# get the attribute |
| 1461 |
$attribute = $property ? 'property' : 'name'; |
| 1462 |
|
| 1463 |
# call the create metatag function |
| 1464 |
$result = $this->create_metatag($attribute, $data); |
| 1465 |
} |
| 1466 |
|
| 1467 |
# return after creation |
| 1468 |
return; |
| 1469 |
} |
| 1470 |
|
| 1471 |
# CRITICAL FIX: Clear cache and get fresh reference |
| 1472 |
# Preserve 'imgs' key for later use by handle_images() |
| 1473 |
$preserved_imgs = $this->cached_elements['imgs'] ?? null; |
| 1474 |
$this->cached_elements = []; |
| 1475 |
if ($preserved_imgs !== null) { |
| 1476 |
$this->cached_elements['imgs'] = $preserved_imgs; |
| 1477 |
} |
| 1478 |
|
| 1479 |
# Get fresh meta tag reference using same selector |
| 1480 |
$meta_tag_fresh = $this->dom->find($meta_selector, 0); |
| 1481 |
|
| 1482 |
if ($meta_tag_fresh) { |
| 1483 |
# Use outertext for replacement |
| 1484 |
$new_value = htmlspecialchars($data['recommended_value'] ?? '', ENT_QUOTES, 'UTF-8'); |
| 1485 |
|
| 1486 |
# Determine attribute name |
| 1487 |
$attr_name = !empty($data['name']) ? 'name' : 'property'; |
| 1488 |
$attr_value = !empty($data['name']) ? $data['name'] : ($data['property'] ?? ''); |
| 1489 |
|
| 1490 |
# Build new meta tag |
| 1491 |
$meta_tag_fresh->outertext = '<meta ' . $attr_name . '="' . htmlspecialchars($attr_value, ENT_QUOTES, 'UTF-8') . '" content="' . $new_value . '">'; |
| 1492 |
} |
| 1493 |
|
| 1494 |
|
| 1495 |
} |
| 1496 |
|
| 1497 |
# function to handle the page title |
| 1498 |
function replace_title($title_data){ |
| 1499 |
|
| 1500 |
# find the title |
| 1501 |
$title = $this->dom->find('title', 0) ?? false; |
| 1502 |
|
| 1503 |
# if none |
| 1504 |
if($title === false){ |
| 1505 |
return $this->create_title($title_data); |
| 1506 |
} |
| 1507 |
|
| 1508 |
# CRITICAL FIX: Clear element cache and get fresh reference |
| 1509 |
# Preserve 'imgs' key for later use by handle_images() |
| 1510 |
$preserved_imgs = $this->cached_elements['imgs'] ?? null; |
| 1511 |
$this->cached_elements = []; |
| 1512 |
if ($preserved_imgs !== null) { |
| 1513 |
$this->cached_elements['imgs'] = $preserved_imgs; |
| 1514 |
} |
| 1515 |
|
| 1516 |
# Get fresh title element reference |
| 1517 |
$title_fresh = $this->dom->find('title', 0); |
| 1518 |
|
| 1519 |
if ($title_fresh) { |
| 1520 |
# Use outertext for replacement |
| 1521 |
$new_value = htmlspecialchars($title_data['recommended_value'], ENT_QUOTES, 'UTF-8'); |
| 1522 |
$title_fresh->outertext = '<title>' . $new_value . '</title>'; |
| 1523 |
} |
| 1524 |
|
| 1525 |
# Don't save_reload here - will happen at the end |
| 1526 |
# $this->save_reload(); |
| 1527 |
} |
| 1528 |
|
| 1529 |
# Function to create a title when it's missing |
| 1530 |
function create_title($title_data) { |
| 1531 |
|
| 1532 |
# Find the <head> tag |
| 1533 |
$head = $this->dom->find('head', 0); |
| 1534 |
|
| 1535 |
# Construct the <title> tag HTML |
| 1536 |
$title_html = '<title>' . htmlspecialchars($title_data['recommended_value'], ENT_QUOTES) . '</title>'; |
| 1537 |
|
| 1538 |
if (empty($head)) { |
| 1539 |
return false; |
| 1540 |
} |
| 1541 |
|
| 1542 |
# Extract existing attributes of the <head> tag |
| 1543 |
$attributes = []; |
| 1544 |
|
| 1545 |
# get the tag attributes |
| 1546 |
$tag_attributes = $head->getAllAttributes(); |
| 1547 |
|
| 1548 |
# loop all attributes |
| 1549 |
foreach ($tag_attributes as $key => $value) { |
| 1550 |
|
| 1551 |
if ($value == 1) { |
| 1552 |
|
| 1553 |
# Handle boolean attributes |
| 1554 |
$attributes[] = htmlspecialchars($key, ENT_QUOTES); |
| 1555 |
} else { |
| 1556 |
|
| 1557 |
# Handle attributes with values |
| 1558 |
$attributes[] = $key . '="' . htmlspecialchars($value, ENT_QUOTES) . '"'; |
| 1559 |
} |
| 1560 |
} |
| 1561 |
|
| 1562 |
# Convert attributes array to a string |
| 1563 |
$attributes_string = !empty($attributes) ? ' ' . implode(' ', $attributes) : ''; |
| 1564 |
|
| 1565 |
# Rebuild the <head> tag, inserting the <title> at the beginning |
| 1566 |
$head->outertext = '<head' . $attributes_string . '>' . $title_html . $head->innertext . '</head>'; |
| 1567 |
|
| 1568 |
# save and reload DOM |
| 1569 |
$this->save_reload(); |
| 1570 |
} |
| 1571 |
|
| 1572 |
# function to create meta tag if none existss |
| 1573 |
function create_metatag($attribute, $data){ |
| 1574 |
|
| 1575 |
# Find the <head> tag |
| 1576 |
$head = $this->dom->find('head', 0); |
| 1577 |
|
| 1578 |
# Construct the meta tag HTML (no spaces around = for standard HTML and regex compatibility) |
| 1579 |
$meta_tag = '<meta '.$attribute.'="'.htmlspecialchars($data[$attribute], ENT_QUOTES, 'UTF-8').'" content="'.htmlspecialchars($data['recommended_value'], ENT_QUOTES, 'UTF-8').'">'; |
| 1580 |
|
| 1581 |
if (empty($head)) { |
| 1582 |
return false; |
| 1583 |
} |
| 1584 |
|
| 1585 |
# Extract existing attributes of the <head> tag |
| 1586 |
$attributes = []; |
| 1587 |
|
| 1588 |
# get the tag attributes |
| 1589 |
$tag_attributes = $head->getAllAttributes(); |
| 1590 |
|
| 1591 |
# loop all attributes |
| 1592 |
foreach ($tag_attributes as $key => $value) { |
| 1593 |
|
| 1594 |
if ($value == 1) { |
| 1595 |
|
| 1596 |
# Handle boolean attributes |
| 1597 |
$attributes[] = htmlspecialchars($key, ENT_QUOTES); |
| 1598 |
} else { |
| 1599 |
|
| 1600 |
# Handle attributes with values |
| 1601 |
$attributes[] = $key . '="' . htmlspecialchars($value, ENT_QUOTES) . '"'; |
| 1602 |
} |
| 1603 |
} |
| 1604 |
|
| 1605 |
# Convert attributes array to a string |
| 1606 |
$attributes_string = !empty($attributes) ? ' ' . implode(' ', $attributes) : ''; |
| 1607 |
|
| 1608 |
# Rebuild the <head> tag, inserting the <title> at the beginning |
| 1609 |
$head->outertext = '<head' . $attributes_string . '>' . $meta_tag . $head->innertext . '</head>'; |
| 1610 |
|
| 1611 |
# save and reload DOM |
| 1612 |
$this->save_reload(); |
| 1613 |
} |
| 1614 |
|
| 1615 |
/** |
| 1616 |
* Remove duplicate <title> tags from HTML, keeping the first (OTTO's) value. |
| 1617 |
* |
| 1618 |
* OTTO's title replacement is always applied first (limit=1 or DOM manipulation), |
| 1619 |
* so the first <title> tag holds the authoritative value. If SEO plugin conflicts |
| 1620 |
* produce additional <title> tags, this strips all and re-inserts one. |
| 1621 |
* |
| 1622 |
* @param string $html Full HTML document. |
| 1623 |
* @return string HTML with at most one <title> tag. |
| 1624 |
*/ |
| 1625 |
private function deduplicate_title_tags($html) { |
| 1626 |
$title_count = preg_match_all('/<title[^>]*>.*?<\/title>/is', $html, $title_matches); |
| 1627 |
if ($title_count <= 1) { |
| 1628 |
return $html; |
| 1629 |
} |
| 1630 |
|
| 1631 |
# Capture the first title's inner text (OTTO's replacement) |
| 1632 |
preg_match('/<title[^>]*>(.*?)<\/title>/is', $html, $first_title); |
| 1633 |
$authoritative_title = isset($first_title[1]) ? $first_title[1] : ''; |
| 1634 |
|
| 1635 |
# Strip all <title> tags |
| 1636 |
$html = preg_replace('/<title[^>]*>.*?<\/title>/is', '', $html); |
| 1637 |
|
| 1638 |
# Re-insert a single <title> after <head> using preg_replace_callback to prevent |
| 1639 |
# backreference injection when title contains $ followed by digits (e.g. "$50 off") |
| 1640 |
$title_tag = '<title>' . $authoritative_title . '</title>'; |
| 1641 |
$html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($title_tag) { |
| 1642 |
return $m[1] . $title_tag; |
| 1643 |
}, $html, 1); |
| 1644 |
|
| 1645 |
return $html; |
| 1646 |
} |
| 1647 |
|
| 1648 |
/** |
| 1649 |
* Remove duplicate OG and Twitter meta tags from HTML after OTTO processing. |
| 1650 |
* |
| 1651 |
* OTTO injects its tags with `data-otto-pixel` or `data-otto` attributes. |
| 1652 |
* Legacy MetaSync output and third-party SEO plugins may also emit the same |
| 1653 |
* OG/Twitter properties. This method runs at the buffer level — after all |
| 1654 |
* sources have written their tags — and keeps only the OTTO version when |
| 1655 |
* duplicates exist. |
| 1656 |
* |
| 1657 |
* Strategy per property (e.g. og:description): |
| 1658 |
* - If OTTO tag exists (has data-otto marker) → remove all non-OTTO duplicates |
| 1659 |
* - If no OTTO tag exists → keep the first occurrence, remove the rest |
| 1660 |
* |
| 1661 |
* @param string $html Full HTML document. |
| 1662 |
* @return string HTML with at most one tag per OG/Twitter property. |
| 1663 |
*/ |
| 1664 |
private function deduplicate_og_twitter_tags($html) { |
| 1665 |
# OG properties to deduplicate |
| 1666 |
$og_properties = [ |
| 1667 |
'og:title', 'og:description', 'og:url', 'og:type', |
| 1668 |
'og:locale', 'og:site_name', 'og:image', |
| 1669 |
]; |
| 1670 |
|
| 1671 |
foreach ($og_properties as $prop) { |
| 1672 |
$html = $this->deduplicate_meta_by_attr($html, 'property', $prop); |
| 1673 |
} |
| 1674 |
|
| 1675 |
# Twitter names to deduplicate |
| 1676 |
$twitter_names = [ |
| 1677 |
'twitter:title', 'twitter:description', 'twitter:card', |
| 1678 |
'twitter:image', 'twitter:site', |
| 1679 |
]; |
| 1680 |
|
| 1681 |
foreach ($twitter_names as $name) { |
| 1682 |
$html = $this->deduplicate_meta_by_attr($html, 'name', $name); |
| 1683 |
} |
| 1684 |
|
| 1685 |
return $html; |
| 1686 |
} |
| 1687 |
|
| 1688 |
/** |
| 1689 |
* Deduplicate meta tags by a specific attribute (property= or name=). |
| 1690 |
* |
| 1691 |
* When duplicates exist and one carries a data-otto marker, keep only |
| 1692 |
* the OTTO version. Otherwise keep the first occurrence. |
| 1693 |
* |
| 1694 |
* @param string $html Full HTML. |
| 1695 |
* @param string $attr_name Attribute name: 'property' or 'name'. |
| 1696 |
* @param string $attr_val Attribute value: e.g. 'og:title' or 'twitter:description'. |
| 1697 |
* @return string |
| 1698 |
*/ |
| 1699 |
private function deduplicate_meta_by_attr($html, $attr_name, $attr_val) { |
| 1700 |
$escaped = preg_quote($attr_val, '/'); |
| 1701 |
# Match all <meta> tags with this attribute value (both attr orderings) |
| 1702 |
$pattern = '/<meta\s[^>]*' . preg_quote($attr_name, '/') . '\s*=\s*["\']' . $escaped . '["\'][^>]*\/?>/i'; |
| 1703 |
|
| 1704 |
if (preg_match_all($pattern, $html, $matches) <= 1) { |
| 1705 |
return $html; # 0 or 1 — nothing to deduplicate |
| 1706 |
} |
| 1707 |
|
| 1708 |
$all_tags = $matches[0]; |
| 1709 |
|
| 1710 |
# Find the OTTO tag (has data-otto-pixel or data-otto attribute) |
| 1711 |
$otto_tag = null; |
| 1712 |
foreach ($all_tags as $tag) { |
| 1713 |
if (stripos($tag, 'data-otto') !== false) { |
| 1714 |
$otto_tag = $tag; |
| 1715 |
break; |
| 1716 |
} |
| 1717 |
} |
| 1718 |
|
| 1719 |
# Determine the keeper: OTTO tag if present, otherwise the first tag |
| 1720 |
$keeper = $otto_tag ?: $all_tags[0]; |
| 1721 |
|
| 1722 |
# Remove all occurrences, then re-insert the keeper at the first position |
| 1723 |
$first_replaced = false; |
| 1724 |
$html = preg_replace_callback($pattern, function ($m) use ($keeper, &$first_replaced) { |
| 1725 |
if (!$first_replaced) { |
| 1726 |
$first_replaced = true; |
| 1727 |
return $keeper; |
| 1728 |
} |
| 1729 |
return ''; # Remove subsequent duplicates |
| 1730 |
}, $html); |
| 1731 |
|
| 1732 |
return $html; |
| 1733 |
} |
| 1734 |
|
| 1735 |
/** |
| 1736 |
* Remove duplicate <link rel="canonical"> tags from HTML. |
| 1737 |
* |
| 1738 |
* When OTTO injects a canonical via header_html_insertion and MetaSync's |
| 1739 |
* SEO output (or WordPress core) has already emitted one, keep only the |
| 1740 |
* OTTO version (identified by data-otto marker). If no OTTO tag exists, |
| 1741 |
* keep the first occurrence. |
| 1742 |
* |
| 1743 |
* @param string $html Full HTML document. |
| 1744 |
* @return string HTML with at most one canonical tag. |
| 1745 |
*/ |
| 1746 |
private function deduplicate_canonical_tags($html) { |
| 1747 |
$pattern = '/<link\s[^>]*rel=["\']canonical["\'][^>]*\/?>/i'; |
| 1748 |
|
| 1749 |
if (preg_match_all($pattern, $html, $matches) <= 1) { |
| 1750 |
return $html; |
| 1751 |
} |
| 1752 |
|
| 1753 |
$all_tags = $matches[0]; |
| 1754 |
|
| 1755 |
$otto_tag = null; |
| 1756 |
foreach ($all_tags as $tag) { |
| 1757 |
if (stripos($tag, 'data-otto') !== false) { |
| 1758 |
$otto_tag = $tag; |
| 1759 |
break; |
| 1760 |
} |
| 1761 |
} |
| 1762 |
|
| 1763 |
$keeper = $otto_tag ?: $all_tags[0]; |
| 1764 |
|
| 1765 |
$first_replaced = false; |
| 1766 |
$html = preg_replace_callback($pattern, function ($m) use ($keeper, &$first_replaced) { |
| 1767 |
if (!$first_replaced) { |
| 1768 |
$first_replaced = true; |
| 1769 |
return $keeper; |
| 1770 |
} |
| 1771 |
return ''; |
| 1772 |
}, $html); |
| 1773 |
|
| 1774 |
return $html; |
| 1775 |
} |
| 1776 |
|
| 1777 |
/** |
| 1778 |
* Deduplicate JSON-LD schema blocks. |
| 1779 |
* |
| 1780 |
* When OTTO and a third-party SEO plugin both inject <script type="application/ld+json"> |
| 1781 |
* blocks, keep OTTO's version for any @type that appears in both. |
| 1782 |
* Third-party blocks whose @type is not covered by OTTO are preserved. |
| 1783 |
* |
| 1784 |
* @param string $html Full HTML. |
| 1785 |
* @return string |
| 1786 |
*/ |
| 1787 |
private function deduplicate_schema_tags($html) { |
| 1788 |
// Find all JSON-LD script blocks |
| 1789 |
$pattern = '/<script(\s[^>]*)type\s*=\s*(["\'])application\/ld\+json\2[^>]*>\s*([\s\S]*?)<\/script>/i'; |
| 1790 |
if (preg_match_all($pattern, $html, $matches, PREG_SET_ORDER) <= 1) { |
| 1791 |
return $html; |
| 1792 |
} |
| 1793 |
|
| 1794 |
$otto_by_type = []; // @type => decoded JSON object |
| 1795 |
$third_by_type = []; // @type => decoded JSON object |
| 1796 |
$otto_graph = []; // entries from OTTO @graph blocks |
| 1797 |
$third_graph = []; // entries from third-party @graph blocks |
| 1798 |
|
| 1799 |
foreach ($matches as $m) { |
| 1800 |
$attrs = $m[1]; |
| 1801 |
$json_str = $m[3]; |
| 1802 |
$decoded = json_decode($json_str, true); |
| 1803 |
if (json_last_error() !== JSON_ERROR_NONE || !is_array($decoded)) { |
| 1804 |
continue; // skip unparseable blocks — leave them in place |
| 1805 |
} |
| 1806 |
$is_otto = stripos($attrs, 'data-otto') !== false; |
| 1807 |
|
| 1808 |
if (isset($decoded['@graph']) && is_array($decoded['@graph'])) { |
| 1809 |
foreach ($decoded['@graph'] as $entry) { |
| 1810 |
if (!isset($entry['@type'])) continue; |
| 1811 |
// JSON-LD allows @type to be a string OR an array of strings. |
| 1812 |
// Rank Math routinely emits multi-typed entries (e.g. ["Person", "Organization"]). |
| 1813 |
// Using an array as an offset throws a fatal on PHP 8+, so normalize to scalar. |
| 1814 |
$type = is_array($entry['@type']) |
| 1815 |
? (string) reset($entry['@type']) |
| 1816 |
: (string) $entry['@type']; |
| 1817 |
if ($type === '') continue; |
| 1818 |
if ($is_otto) { |
| 1819 |
$otto_graph[$type] = $entry; |
| 1820 |
} else { |
| 1821 |
$third_graph[$type] = $entry; |
| 1822 |
} |
| 1823 |
} |
| 1824 |
} elseif (isset($decoded['@type'])) { |
| 1825 |
$type = is_array($decoded['@type']) |
| 1826 |
? (string) reset($decoded['@type']) |
| 1827 |
: (string) $decoded['@type']; |
| 1828 |
if ($type === '') continue; |
| 1829 |
if ($is_otto) { |
| 1830 |
$otto_by_type[$type] = $decoded; |
| 1831 |
} else { |
| 1832 |
$third_by_type[$type] = $decoded; |
| 1833 |
} |
| 1834 |
} |
| 1835 |
} |
| 1836 |
|
| 1837 |
// If OTTO provided no schema at all, nothing to deduplicate |
| 1838 |
if (empty($otto_by_type) && empty($otto_graph)) { |
| 1839 |
return $html; |
| 1840 |
} |
| 1841 |
|
| 1842 |
// Remove all JSON-LD blocks from HTML |
| 1843 |
$html = preg_replace($pattern, '', $html); |
| 1844 |
|
| 1845 |
// Re-insert flat (non-@graph) blocks: OTTO wins for matching @type |
| 1846 |
$kept = array_merge($otto_by_type, array_diff_key($third_by_type, $otto_by_type)); |
| 1847 |
$rebuilt = ''; |
| 1848 |
foreach ($kept as $decoded) { |
| 1849 |
$rebuilt .= '<script type="application/ld+json" data-otto="true">' . |
| 1850 |
wp_json_encode($decoded) . "</script>\n"; |
| 1851 |
} |
| 1852 |
|
| 1853 |
// Re-insert merged @graph block (if any entries exist) |
| 1854 |
$merged_graph = array_merge($third_graph, $otto_graph); // OTTO wins on duplicate @type |
| 1855 |
if (!empty($merged_graph)) { |
| 1856 |
$graph_obj = ['@context' => 'https://schema.org', '@graph' => array_values($merged_graph)]; |
| 1857 |
$rebuilt .= '<script type="application/ld+json" data-otto="true">' . |
| 1858 |
wp_json_encode($graph_obj) . "</script>\n"; |
| 1859 |
} |
| 1860 |
|
| 1861 |
// Re-inject before </head> |
| 1862 |
if (!empty($rebuilt)) { |
| 1863 |
$html = preg_replace_callback('/(<\/head>)/i', function ($m) use ($rebuilt) { |
| 1864 |
return $rebuilt . $m[1]; |
| 1865 |
}, $html, 1); |
| 1866 |
} |
| 1867 |
|
| 1868 |
return $html; |
| 1869 |
} |
| 1870 |
|
| 1871 |
# function to detect if current page is an AMP page |
| 1872 |
function is_amp_page(){ |
| 1873 |
|
| 1874 |
# Check if URL path contains /amp/ |
| 1875 |
$current_url = $_SERVER['REQUEST_URI'] ?? ''; |
| 1876 |
if (strpos($current_url, '/amp/') !== false) { |
| 1877 |
return true; |
| 1878 |
} |
| 1879 |
|
| 1880 |
# Check if URL ends with /amp |
| 1881 |
if (preg_match('/\/amp\/?$/', $current_url)) { |
| 1882 |
return true; |
| 1883 |
} |
| 1884 |
|
| 1885 |
# Check if amp=1 query parameter is present |
| 1886 |
if (isset($_GET['amp']) && $_GET['amp'] == '1') { |
| 1887 |
return true; |
| 1888 |
} |
| 1889 |
|
| 1890 |
# Check for other common AMP query parameters |
| 1891 |
if (isset($_GET['amp']) && !empty($_GET['amp'])) { |
| 1892 |
return true; |
| 1893 |
} |
| 1894 |
|
| 1895 |
return false; |
| 1896 |
} |
| 1897 |
|
| 1898 |
# this function insterts header html to the dom |
| 1899 |
function insert_header_html($data){ |
| 1900 |
|
| 1901 |
# check that we have the header html |
| 1902 |
if(empty($data['header_html_insertion'])){ |
| 1903 |
# |
| 1904 |
return; |
| 1905 |
} |
| 1906 |
|
| 1907 |
# append the/ html at the start of the header |
| 1908 |
$head = $this->dom->find('head', 0); |
| 1909 |
|
| 1910 |
if ($head) { |
| 1911 |
|
| 1912 |
# Check if this is an AMP page - if so, don't add metasync_optimized attribute |
| 1913 |
$is_amp_page = $this->is_amp_page(); |
| 1914 |
|
| 1915 |
# Append the new HTML at the start of the <head> tag |
| 1916 |
# For AMP pages: use clean <head> tag without metasync_optimized attribute |
| 1917 |
# For non-AMP pages: add metasync_optimized attribute to <head> tag |
| 1918 |
if ($is_amp_page) { |
| 1919 |
$head->outertext = '<head>' .$data['header_html_insertion']. $head->innertext . '</head>'; |
| 1920 |
} else { |
| 1921 |
$head->outertext = '<head metasync_optimized>' .$data['header_html_insertion']. $head->innertext . '</head>'; |
| 1922 |
} |
| 1923 |
|
| 1924 |
} |
| 1925 |
|
| 1926 |
# save and reload DOM |
| 1927 |
$this->save_reload(); |
| 1928 |
} |
| 1929 |
|
| 1930 |
# function to forcefully remove metasync_optimized attribute from head on AMP pages |
| 1931 |
function cleanup_amp_metasync_attribute(){ |
| 1932 |
|
| 1933 |
# Only proceed if this is an AMP page |
| 1934 |
if (!$this->is_amp_page()) { |
| 1935 |
return; |
| 1936 |
} |
| 1937 |
|
| 1938 |
# Find the head tag |
| 1939 |
$head = $this->dom->find('head', 0); |
| 1940 |
|
| 1941 |
if (!$head) { |
| 1942 |
return; |
| 1943 |
} |
| 1944 |
|
| 1945 |
# Check if head has metasync_optimized attribute |
| 1946 |
$head_html = $head->outertext; |
| 1947 |
|
| 1948 |
# If metasync_optimized attribute is found, remove it |
| 1949 |
if (strpos($head_html, 'metasync_optimized') !== false) { |
| 1950 |
|
| 1951 |
# Remove the metasync_optimized attribute from the head tag |
| 1952 |
# This handles various formats: <head metasync_optimized>, <head metasync_optimized=""> etc. |
| 1953 |
$cleaned_head_html = preg_replace('/\s*metasync_optimized(?:="[^"]*")?/', '', $head_html); |
| 1954 |
|
| 1955 |
# Update the head element |
| 1956 |
$head->outertext = $cleaned_head_html; |
| 1957 |
|
| 1958 |
} |
| 1959 |
} |
| 1960 |
|
| 1961 |
/** |
| 1962 |
* PERFORMANCE OPTIMIZATION: Pre-cache commonly accessed DOM elements |
| 1963 |
* Reduces 8-10 full DOM traversals to 1 initial traversal |
| 1964 |
* Call this once after loading HTML, before processing |
| 1965 |
*/ |
| 1966 |
private function cache_elements() { |
| 1967 |
if (!$this->dom) { |
| 1968 |
return; |
| 1969 |
} |
| 1970 |
|
| 1971 |
# Cache all commonly accessed elements in one pass |
| 1972 |
$this->cached_elements = [ |
| 1973 |
'html' => $this->dom->find('html', 0), |
| 1974 |
'head' => $this->dom->find('head', 0), |
| 1975 |
'body' => $this->dom->find('body', 0), |
| 1976 |
'footer' => $this->dom->find('footer', 0), |
| 1977 |
'title' => $this->dom->find('title', 0), |
| 1978 |
'imgs' => $this->dom->find('img'), |
| 1979 |
'links' => $this->dom->find('a'), |
| 1980 |
'canonical' => $this->dom->find('link[rel="canonical"]', 0), |
| 1981 |
]; |
| 1982 |
} |
| 1983 |
|
| 1984 |
/** |
| 1985 |
* Get cached element by key, with fallback to DOM find |
| 1986 |
* @param string $key Element key from cache |
| 1987 |
* @param mixed $fallback Fallback value if not cached |
| 1988 |
* @return mixed Cached element or fallback |
| 1989 |
*/ |
| 1990 |
private function get_cached_element($key, $fallback = null) { |
| 1991 |
return $this->cached_elements[$key] ?? $fallback; |
| 1992 |
} |
| 1993 |
|
| 1994 |
# this function saves are reloads the dom for modifications to avoid conflict |
| 1995 |
function save_reload(){ |
| 1996 |
|
| 1997 |
# PERFORMANCE OPTIMIZATION: Skip reload if deferred |
| 1998 |
# This reduces multiple serialize/deserialize cycles to just one final reload |
| 1999 |
if ($this->deferred_reload) { |
| 2000 |
return; |
| 2001 |
} |
| 2002 |
|
| 2003 |
# Cleanup metasync_optimized attribute on AMP pages before saving |
| 2004 |
$this->cleanup_amp_metasync_attribute(); |
| 2005 |
|
| 2006 |
# DISABLED: Cache file creation temporarily disabled |
| 2007 |
|
| 2008 |
# if(file_put_contents($this->html_file, $this->dom)){ |
| 2009 |
|
| 2010 |
# load the modified file to the DOM |
| 2011 |
# $this->dom = new HtmlDocument($this->html_file ); |
| 2012 |
# } |
| 2013 |
|
| 2014 |
# this code is to be replaced in future |
| 2015 |
# reson for adding is to prevent caching logged in user pages |
| 2016 |
# why not just skip saving? it broke the DOM Library |
| 2017 |
# check user is logged in clear the file |
| 2018 |
|
| 2019 |
# if(is_user_logged_in()) { |
| 2020 |
# unlink($this->html_file); |
| 2021 |
# } |
| 2022 |
|
| 2023 |
# MEMORY-BASED RELOAD: Instead of file operations, reload DOM from current HTML string |
| 2024 |
# This prevents DOM breaking while avoiding cache file creation |
| 2025 |
if($this->dom){ |
| 2026 |
# Get current DOM as HTML string |
| 2027 |
$current_html = $this->dom->save(); |
| 2028 |
|
| 2029 |
# Reload DOM from the HTML string to refresh internal state |
| 2030 |
# This replaces the file save/reload cycle that SimpleHtmlDOM expects |
| 2031 |
$current_html = $this->sanitize_text_less_than($current_html); |
| 2032 |
$this->dom->load($current_html, true, false); |
| 2033 |
|
| 2034 |
# Force UTF-8 charset after reload to preserve emojis |
| 2035 |
$this->dom->_charset = 'UTF-8'; |
| 2036 |
$this->dom->_target_charset = 'UTF-8'; |
| 2037 |
} |
| 2038 |
|
| 2039 |
} |
| 2040 |
|
| 2041 |
/** |
| 2042 |
* Process HTML directly without HTTP request (for buffer approach) |
| 2043 |
* This is the FAST path - eliminates the internal wp_remote_get call |
| 2044 |
* |
| 2045 |
* CRITICAL: This method is called from the output buffer callback. |
| 2046 |
* If it fails, we must return false so the original HTML can be used. |
| 2047 |
* |
| 2048 |
* @param string $html The raw HTML captured from output buffer |
| 2049 |
* @param array $replacement_data OTTO suggestions/replacement data |
| 2050 |
* @return HtmlDocument|false Modified DOM or false on failure |
| 2051 |
* @since 2.6.0 |
| 2052 |
*/ |
| 2053 |
function process_html_directly($html, $replacement_data) { |
| 2054 |
try { |
| 2055 |
# Validate inputs |
| 2056 |
if (empty($html) || empty($replacement_data) || !is_array($replacement_data)) { |
| 2057 |
return false; |
| 2058 |
} |
| 2059 |
|
| 2060 |
# Validate HTML is actual HTML content |
| 2061 |
if (stripos($html, '<html') === false && stripos($html, '<!DOCTYPE') === false) { |
| 2062 |
return false; |
| 2063 |
} |
| 2064 |
|
| 2065 |
# Remove XML declaration if present |
| 2066 |
$html = preg_replace('/<\?xml[^?]*\?>\s*/i', '', $html); |
| 2067 |
|
| 2068 |
# WP-355 / WP-315: Save ALL original <style> blocks before DOM processing. |
| 2069 |
$original_style_blocks = $this->capture_style_blocks($html); |
| 2070 |
|
| 2071 |
# Escape bare < in text content before DOM parsing |
| 2072 |
$html = $this->sanitize_text_less_than($html); |
| 2073 |
|
| 2074 |
# WP-355: Fix malformed self-closing non-void tags (e.g. <ul/ class="...">) |
| 2075 |
$html = $this->fix_malformed_self_closing_tags($html); |
| 2076 |
|
| 2077 |
# Load HTML into DOM |
| 2078 |
$this->dom->load($html, true, false); |
| 2079 |
|
| 2080 |
# Force UTF-8 charset to preserve emojis and special characters |
| 2081 |
$this->dom->_charset = 'UTF-8'; |
| 2082 |
$this->dom->_target_charset = 'UTF-8'; |
| 2083 |
|
| 2084 |
# PERFORMANCE OPTIMIZATION: Pre-cache commonly accessed DOM elements |
| 2085 |
# This eliminates 8-10 full DOM traversals per page |
| 2086 |
$this->cache_elements(); |
| 2087 |
|
| 2088 |
# PERFORMANCE OPTIMIZATION: Enable deferred reload to skip intermediate reloads |
| 2089 |
# This reduces 6-10 DOM serialize/deserialize cycles to just 1 final reload |
| 2090 |
$this->deferred_reload = true; |
| 2091 |
|
| 2092 |
# Apply blocking flags if available (for SEO plugin coordination) |
| 2093 |
if (!empty($replacement_data['_otto_blocking'])) { |
| 2094 |
$block_title = $replacement_data['_otto_blocking']['block_title'] ?? false; |
| 2095 |
$block_description_tags = $replacement_data['_otto_blocking']['block_description_tags'] ?? []; |
| 2096 |
|
| 2097 |
# Remove SEO plugin meta tags if OTTO is providing them |
| 2098 |
if ($block_title || !empty($block_description_tags)) { |
| 2099 |
$this->remove_conflicting_seo_tags($block_title, $block_description_tags); |
| 2100 |
} |
| 2101 |
} |
| 2102 |
|
| 2103 |
# Apply all OTTO modifications (same as handle_route_html but without HTTP fetch) |
| 2104 |
|
| 2105 |
# COMPATIBILITY FIX: Transform 'value' to 'recommended_value' if needed |
| 2106 |
# Some API versions return 'value' instead of 'recommended_value' |
| 2107 |
if (!empty($replacement_data['header_replacements'])) { |
| 2108 |
foreach ($replacement_data['header_replacements'] as &$item) { |
| 2109 |
if (isset($item['value']) && !isset($item['recommended_value'])) { |
| 2110 |
$item['recommended_value'] = $item['value']; |
| 2111 |
} |
| 2112 |
} |
| 2113 |
unset($item); // Break reference |
| 2114 |
} |
| 2115 |
|
| 2116 |
# 1. Header HTML insertion |
| 2117 |
$this->insert_header_html($replacement_data); |
| 2118 |
|
| 2119 |
# 2. Header replacements (title, meta, canonical) |
| 2120 |
$this->do_header_replacements($replacement_data); |
| 2121 |
|
| 2122 |
# 3. Body replacements (top, bottom, substitutions) |
| 2123 |
$this->do_body_replacements($replacement_data); |
| 2124 |
|
| 2125 |
# 4. Footer HTML insertion |
| 2126 |
$this->do_footer_html_insertion($replacement_data); |
| 2127 |
|
| 2128 |
# 5. Final cleanup for AMP pages |
| 2129 |
$this->cleanup_amp_metasync_attribute(); |
| 2130 |
|
| 2131 |
# CRITICAL FIX: Clear cached elements and force DOM to refresh |
| 2132 |
$this->deferred_reload = false; |
| 2133 |
$this->cached_elements = []; // Clear our custom cache |
| 2134 |
|
| 2135 |
# Clear SimpleHtmlDom's internal cache |
| 2136 |
if (method_exists($this->dom, 'clear')) { |
| 2137 |
$this->dom->clear(); |
| 2138 |
} |
| 2139 |
|
| 2140 |
|
| 2141 |
# Try getting HTML via root element instead of save() |
| 2142 |
$root = $this->dom->root; |
| 2143 |
if ($root && isset($root->outertext)) { |
| 2144 |
$result_html = $root->outertext; |
| 2145 |
} else { |
| 2146 |
# Fallback to save() method |
| 2147 |
$result_html = $this->dom->save(); |
| 2148 |
} |
| 2149 |
|
| 2150 |
# DEBUG: Check if DOM changes persisted |
| 2151 |
if (preg_match('/<title[^>]*>(.*?)<\/title>/is', $result_html, $matches)) { |
| 2152 |
} |
| 2153 |
|
| 2154 |
# Apply header replacements manually via string replacement |
| 2155 |
if (!empty($replacement_data['header_replacements'])) { |
| 2156 |
|
| 2157 |
foreach ($replacement_data['header_replacements'] as $idx => $item) { |
| 2158 |
$type = $item['type'] ?? ''; |
| 2159 |
$value = $item['recommended_value'] ?? $item['value'] ?? ''; |
| 2160 |
|
| 2161 |
|
| 2162 |
if (empty($value)) { |
| 2163 |
continue; |
| 2164 |
} |
| 2165 |
|
| 2166 |
if ($type === 'title') { |
| 2167 |
# Replace title tag if present; otherwise insert (e.g. when Yoast was blocked and no <title> was output) |
| 2168 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 2169 |
$title_tag = '<title>' . $new_value . '</title>'; |
| 2170 |
$replaced = preg_replace_callback('/<title[^>]*>.*?<\/title>/is', function ($m) use ($title_tag) { |
| 2171 |
return $title_tag; |
| 2172 |
}, $result_html, 1); |
| 2173 |
if ($replaced === $result_html && strpos($result_html, '<title') === false) { |
| 2174 |
# No <title> in document (common when Yoast is blocked) — insert after <head> |
| 2175 |
$result_html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($title_tag) { |
| 2176 |
return $m[1] . "\n" . $title_tag; |
| 2177 |
}, $result_html, 1); |
| 2178 |
} else { |
| 2179 |
$result_html = $replaced; |
| 2180 |
} |
| 2181 |
} elseif ($type === 'meta') { |
| 2182 |
$name = $item['name'] ?? ''; |
| 2183 |
$property = $item['property'] ?? ''; |
| 2184 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 2185 |
|
| 2186 |
if (!empty($name)) { |
| 2187 |
# MEMORY OPTIMIZED: Count meta tags without storing all matches |
| 2188 |
# preg_match_all requires $matches, so we pass it but unset immediately |
| 2189 |
# Use \s*=\s* to match attributes with or without spaces around = |
| 2190 |
$before_count = preg_match_all('/<meta[^>]+name\s*=\s*["\']' . preg_quote($name, '/') . '["\'][^>]*>/i', $result_html, $before_matches); |
| 2191 |
|
| 2192 |
# Free memory immediately after use |
| 2193 |
unset($before_matches); |
| 2194 |
|
| 2195 |
# IMPROVED FIX: Remove ALL existing meta tags with this name |
| 2196 |
# This pattern matches ANY meta tag with name="X" regardless of: |
| 2197 |
# - Attribute order (name before/after content) |
| 2198 |
# - Additional attributes (id, class, data-*, etc.) |
| 2199 |
# - Quote style (single or double quotes) |
| 2200 |
# - Whitespace variations |
| 2201 |
# The pattern uses [^>]* to match ANY characters until the closing > |
| 2202 |
$removed_count = preg_replace_callback( |
| 2203 |
'/<meta\s+[^>]*name\s*=\s*["\']' . preg_quote($name, '/') . '["\'][^>]*>/i', |
| 2204 |
function($match) { |
| 2205 |
return ''; // Remove the tag |
| 2206 |
}, |
| 2207 |
$result_html, |
| 2208 |
-1, // Remove all occurrences |
| 2209 |
$total_removed |
| 2210 |
); |
| 2211 |
|
| 2212 |
# Update the HTML with removed tags |
| 2213 |
if ($total_removed > 0) { |
| 2214 |
$result_html = $removed_count; |
| 2215 |
} |
| 2216 |
|
| 2217 |
|
| 2218 |
# MEMORY OPTIMIZED: Verify removal - count again but free memory immediately |
| 2219 |
$after_count = preg_match_all('/<meta[^>]+name\s*=\s*["\']' . preg_quote($name, '/') . '["\'][^>]*>/i', $result_html, $after_matches); |
| 2220 |
unset($after_matches); |
| 2221 |
|
| 2222 |
# Now insert ONE new meta tag at the TOP of <head> |
| 2223 |
$replacement = '<meta name="' . htmlspecialchars($name, ENT_QUOTES, 'UTF-8') . '" content="' . $new_value . '" data-otto="true">'; |
| 2224 |
$result_html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($replacement) { |
| 2225 |
return $m[1] . "\n" . $replacement; |
| 2226 |
}, $result_html, 1); |
| 2227 |
} elseif (!empty($property)) { |
| 2228 |
# Replace meta property tag |
| 2229 |
$pattern = '/<meta\s+property\s*=\s*["\']' . preg_quote($property, '/') . '["\']\s+content\s*=\s*["\'][^"\']*["\']\s*\/?>/i'; |
| 2230 |
$replacement = '<meta property="' . htmlspecialchars($property, ENT_QUOTES, 'UTF-8') . '" content="' . $new_value . '">'; |
| 2231 |
|
| 2232 |
if (preg_match($pattern, $result_html)) { |
| 2233 |
$result_html = preg_replace_callback($pattern, function ($m) use ($replacement) { |
| 2234 |
return $replacement; |
| 2235 |
}, $result_html, 1); |
| 2236 |
} else { |
| 2237 |
# Meta tag doesn't exist, insert it in head |
| 2238 |
$result_html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($replacement) { |
| 2239 |
return $m[1] . "\n" . $replacement; |
| 2240 |
}, $result_html, 1); |
| 2241 |
} |
| 2242 |
} |
| 2243 |
} elseif ($type === 'h1' || $type === 'heading') { |
| 2244 |
# Replace first H1 tag, preserving attributes |
| 2245 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 2246 |
$result_html = preg_replace_callback( |
| 2247 |
'/<h1([^>]*)>.*?<\/h1>/is', |
| 2248 |
function ($m) use ($new_value) { |
| 2249 |
return '<h1' . $m[1] . '>' . $new_value . '</h1>'; |
| 2250 |
}, |
| 2251 |
$result_html, |
| 2252 |
1 |
| 2253 |
); |
| 2254 |
} |
| 2255 |
} |
| 2256 |
} |
| 2257 |
|
| 2258 |
# Apply header HTML insertion (for schema, etc.) — only if DOM insertion didn't already apply it |
| 2259 |
if (!empty($replacement_data['header_html_insertion'])) { |
| 2260 |
$header_html_check = trim($replacement_data['header_html_insertion']); |
| 2261 |
if (strpos($result_html, $header_html_check) === false) { |
| 2262 |
$header_html_insertion = preg_replace( |
| 2263 |
'/<script(\s[^>]*)type\s*=\s*(["\'])application\/ld\+json\2/i', |
| 2264 |
'<script$1type=$2application/ld+json$2 data-otto="true"', |
| 2265 |
$replacement_data['header_html_insertion'] |
| 2266 |
); |
| 2267 |
$header_html = str_replace(array('\\', '$'), array('\\\\', '\\$'), $header_html_insertion); |
| 2268 |
# Insert before </head> |
| 2269 |
$result_html = preg_replace('/(<\/head>)/i', $header_html . "\n" . '$1', $result_html, 1); |
| 2270 |
} |
| 2271 |
} |
| 2272 |
|
| 2273 |
# When Otto block has a title: ensure only ONE <title> (remove non-Otto, then keep first only) |
| 2274 |
if (!empty($replacement_data['header_html_insertion']) && stripos($replacement_data['header_html_insertion'], '<title') !== false) { |
| 2275 |
$result_html = preg_replace( |
| 2276 |
'/<title(?![^>]*data-otto-pixel\s*=\s*["\']dynamic-seo["\'])[^>]*>.*?<\/title>\s*/is', |
| 2277 |
'', |
| 2278 |
$result_html |
| 2279 |
); |
| 2280 |
$first = true; |
| 2281 |
$result_html = preg_replace_callback( |
| 2282 |
'/<title[^>]*>.*?<\/title>\s*/is', |
| 2283 |
function ($m) use (&$first) { |
| 2284 |
if ($first) { |
| 2285 |
$first = false; |
| 2286 |
return $m[0]; |
| 2287 |
} |
| 2288 |
return ''; |
| 2289 |
}, |
| 2290 |
$result_html |
| 2291 |
); |
| 2292 |
} |
| 2293 |
|
| 2294 |
# Apply body top HTML insertion — only if DOM insertion didn't already apply it |
| 2295 |
if (!empty($replacement_data['body_top_html_insertion'])) { |
| 2296 |
$body_top_check = trim($replacement_data['body_top_html_insertion']); |
| 2297 |
if (strpos($result_html, $body_top_check) === false) { |
| 2298 |
$body_top_html = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['body_top_html_insertion']); |
| 2299 |
# Insert after <body> |
| 2300 |
$result_html = preg_replace('/(<body[^>]*>)/i', '$1' . "\n" . $body_top_html, $result_html, 1); |
| 2301 |
} |
| 2302 |
} |
| 2303 |
|
| 2304 |
# Apply body bottom HTML insertion — only if DOM insertion didn't already apply it |
| 2305 |
if (!empty($replacement_data['body_bottom_html_insertion'])) { |
| 2306 |
$body_bottom_check = trim($replacement_data['body_bottom_html_insertion']); |
| 2307 |
if (strpos($result_html, $body_bottom_check) === false) { |
| 2308 |
$body_bottom_html = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['body_bottom_html_insertion']); |
| 2309 |
# Insert before </body> |
| 2310 |
$result_html = preg_replace('/(<\/body>)/i', $body_bottom_html . "\n" . '$1', $result_html, 1); |
| 2311 |
} |
| 2312 |
} |
| 2313 |
|
| 2314 |
# Apply footer HTML insertion |
| 2315 |
if (!empty($replacement_data['footer_html_insertion'])) { |
| 2316 |
$footer_html = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['footer_html_insertion']); |
| 2317 |
# Insert before </html> |
| 2318 |
$result_html = preg_replace('/(<\/html>)/i', $footer_html . "\n" . '$1', $result_html, 1); |
| 2319 |
} |
| 2320 |
|
| 2321 |
# CRITICAL FIX: Apply image alt text manually via string replacement |
| 2322 |
# DOM changes don't persist, must use string replacement |
| 2323 |
$result_html = $this->apply_image_alt_text_via_string($result_html, $replacement_data); |
| 2324 |
|
| 2325 |
# String-based heading fallback for body_substitutions |
| 2326 |
# DOM changes via SimpleHtmlDom don't persist on Divi/page-builder sites |
| 2327 |
if (!empty($replacement_data['body_substitutions']['headings']) && is_array($replacement_data['body_substitutions']['headings'])) { |
| 2328 |
foreach ($replacement_data['body_substitutions']['headings'] as $heading) { |
| 2329 |
if (empty($heading['type']) || empty($heading['current_value']) || empty($heading['recommended_value'])) { |
| 2330 |
continue; |
| 2331 |
} |
| 2332 |
|
| 2333 |
$heading_type = preg_quote($heading['type'], '/'); |
| 2334 |
$current_value = trim(preg_replace('/\s+/', ' ', html_entity_decode($heading['current_value'], ENT_QUOTES, 'UTF-8'))); |
| 2335 |
$recommended_value = htmlspecialchars($heading['recommended_value'], ENT_QUOTES, 'UTF-8'); |
| 2336 |
|
| 2337 |
$result_html = preg_replace_callback( |
| 2338 |
'/(<' . $heading_type . '(?:\s[^>]*)?>)(.*?)(<\/' . $heading_type . '>)/is', |
| 2339 |
function ($m) use ($current_value, $recommended_value) { |
| 2340 |
$inner_text = trim(preg_replace('/\s+/', ' ', html_entity_decode(strip_tags($m[2]), ENT_QUOTES, 'UTF-8'))); |
| 2341 |
if ($inner_text === $current_value) { |
| 2342 |
return $m[1] . $recommended_value . $m[3]; |
| 2343 |
} |
| 2344 |
return $m[0]; |
| 2345 |
}, |
| 2346 |
$result_html, |
| 2347 |
-1 |
| 2348 |
); |
| 2349 |
} |
| 2350 |
} |
| 2351 |
|
| 2352 |
# DEBUG: Check what we're returning |
| 2353 |
if (preg_match('/<title[^>]*>(.*?)<\/title>/is', $result_html, $matches)) { |
| 2354 |
} |
| 2355 |
|
| 2356 |
# FINAL VERIFICATION: Count meta descriptions in returned HTML |
| 2357 |
$final_meta_count = preg_match_all('/<meta[^>]+name=["\']description["\'][^>]*>/i', $result_html, $final_meta_matches); |
| 2358 |
if ($final_meta_count > 0) { |
| 2359 |
foreach ($final_meta_matches[0] as $idx => $meta) { |
| 2360 |
} |
| 2361 |
} |
| 2362 |
|
| 2363 |
if ($final_meta_count > 1) { |
| 2364 |
} |
| 2365 |
|
| 2366 |
# AGGRESSIVE DUPLICATE REMOVAL: Remove any meta description without data-otto marker |
| 2367 |
# Only runs when OTTO has actually inserted its own description (data-otto marker present) |
| 2368 |
# This prevents stripping Yoast/plugin descriptions when OTTO has no description to replace |
| 2369 |
|
| 2370 |
$removal_count = 0; |
| 2371 |
$otto_has_description = (bool) preg_match('/<meta[^>]*name\s*=\s*["\']description["\'][^>]*data-otto[^>]*>/i', $result_html); |
| 2372 |
if ($otto_has_description) { |
| 2373 |
$result_html = preg_replace_callback( |
| 2374 |
'/<meta\s+([^>]*name\s*=\s*["\']description["\'][^>]*)>/i', |
| 2375 |
function($match) use (&$removal_count) { |
| 2376 |
# Keep only if it has data-otto="true" |
| 2377 |
if (stripos($match[1], 'data-otto') !== false) { |
| 2378 |
return $match[0]; // Keep OTTO's meta tag |
| 2379 |
} |
| 2380 |
# Remove any other meta description |
| 2381 |
$removal_count++; |
| 2382 |
return ''; |
| 2383 |
}, |
| 2384 |
$result_html |
| 2385 |
); |
| 2386 |
} |
| 2387 |
|
| 2388 |
# DEDUPLICATION: Remove duplicate <title>, OG, Twitter tags, canonical, and JSON-LD schema |
| 2389 |
$result_html = $this->deduplicate_title_tags($result_html); |
| 2390 |
$result_html = $this->deduplicate_og_twitter_tags($result_html); |
| 2391 |
$result_html = $this->deduplicate_schema_tags($result_html); |
| 2392 |
$result_html = $this->deduplicate_canonical_tags($result_html); |
| 2393 |
|
| 2394 |
# MEMORY OPTIMIZED: Free all large objects and arrays before returning |
| 2395 |
# This ensures memory is released immediately, especially important for high-traffic sites |
| 2396 |
unset($final_meta_matches, $matches); |
| 2397 |
|
| 2398 |
# Clear SimpleHtmlDom internal cache to free memory |
| 2399 |
# Note: We don't unset $this->dom as the object may be reused |
| 2400 |
if ($this->dom && method_exists($this->dom, 'clear')) { |
| 2401 |
$this->dom->clear(); |
| 2402 |
} |
| 2403 |
|
| 2404 |
# Clear element cache array |
| 2405 |
$this->cached_elements = []; |
| 2406 |
|
| 2407 |
# Ensure metasync_optimized attribute on <head> (post-serialization so dom->clear() can't wipe it) |
| 2408 |
if (!$this->is_amp_page() && strpos($result_html, 'metasync_optimized') === false) { |
| 2409 |
$result_html = preg_replace('/<head(\s|>)/i', '<head metasync_optimized$1', $result_html, 1); |
| 2410 |
} |
| 2411 |
|
| 2412 |
$result_html = $this->restore_case_sensitive_attributes($result_html); |
| 2413 |
|
| 2414 |
# WP-355 / WP-315: Re-inject any <style> blocks lost during processing. |
| 2415 |
$result_html = $this->restore_lost_style_blocks($result_html, $original_style_blocks); |
| 2416 |
|
| 2417 |
# WP-315: Fix Divi 5 shortcode framework class renumbering |
| 2418 |
$result_html = $this->fix_divi_class_renumbering($result_html); |
| 2419 |
|
| 2420 |
# WP-315: Clean OTTO internal fetch params from HTML output |
| 2421 |
$result_html = $this->clean_otto_fetch_params($result_html); |
| 2422 |
|
| 2423 |
return $result_html; |
| 2424 |
|
| 2425 |
} catch (Exception $e) { |
| 2426 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': Exception in process_html_directly - ' . $e->getMessage()); |
| 2427 |
return false; |
| 2428 |
} catch (Error $e) { |
| 2429 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': Error in process_html_directly - ' . $e->getMessage()); |
| 2430 |
return false; |
| 2431 |
} |
| 2432 |
} |
| 2433 |
|
| 2434 |
/** |
| 2435 |
* Remove conflicting SEO plugin meta tags from DOM |
| 2436 |
* Called when OTTO is providing its own title/description |
| 2437 |
* |
| 2438 |
* @param bool $remove_title Remove title-related tags |
| 2439 |
* @param array|bool $remove_description_tags Array of specific description tag selectors to remove, or false/empty array for none |
| 2440 |
* @since 2.6.0 |
| 2441 |
*/ |
| 2442 |
private function remove_conflicting_seo_tags($remove_title = false, $remove_description_tags = []) { |
| 2443 |
if (!$this->dom) { |
| 2444 |
return; |
| 2445 |
} |
| 2446 |
|
| 2447 |
# Find head element |
| 2448 |
$head = $this->dom->find('head', 0); |
| 2449 |
if (!$head) { |
| 2450 |
return; |
| 2451 |
} |
| 2452 |
|
| 2453 |
# Check for custom SEO values from MetaSync SEO Sidebar |
| 2454 |
# Custom values take absolute priority over OTTO suggestions |
| 2455 |
$has_custom_title = false; |
| 2456 |
$has_custom_description = false; |
| 2457 |
|
| 2458 |
if (function_exists('is_singular') && is_singular()) { |
| 2459 |
$post_id = get_the_ID(); |
| 2460 |
if ($post_id) { |
| 2461 |
$custom_seo_title = get_post_meta($post_id, '_metasync_seo_title', true); |
| 2462 |
$custom_seo_description = get_post_meta($post_id, '_metasync_seo_desc', true); |
| 2463 |
$has_custom_title = !empty($custom_seo_title); |
| 2464 |
$has_custom_description = !empty($custom_seo_description); |
| 2465 |
} |
| 2466 |
} |
| 2467 |
|
| 2468 |
# Handle description tags - only remove specific tags that Otto is providing |
| 2469 |
if (!empty($remove_description_tags) && is_array($remove_description_tags) && !$has_custom_description) { |
| 2470 |
# Remove only the specific description tags that Otto is providing |
| 2471 |
foreach ($remove_description_tags as $selector) { |
| 2472 |
$tags = $this->dom->find($selector); |
| 2473 |
foreach ($tags as $tag) { |
| 2474 |
# Remove the tag |
| 2475 |
$tag->outertext = ''; |
| 2476 |
} |
| 2477 |
} |
| 2478 |
} |
| 2479 |
|
| 2480 |
if ($remove_title && !$has_custom_title) { |
| 2481 |
# Remove Open Graph and Twitter title tags (keep main <title>) |
| 2482 |
$title_selectors = [ |
| 2483 |
'meta[property=og:title]', |
| 2484 |
'meta[name=twitter:title]', |
| 2485 |
]; |
| 2486 |
|
| 2487 |
foreach ($title_selectors as $selector) { |
| 2488 |
$tags = $this->dom->find($selector); |
| 2489 |
foreach ($tags as $tag) { |
| 2490 |
$tag->outertext = ''; # Remove the tag |
| 2491 |
} |
| 2492 |
} |
| 2493 |
} |
| 2494 |
|
| 2495 |
# Save changes |
| 2496 |
$this->save_reload(); |
| 2497 |
} |
| 2498 |
|
| 2499 |
/** |
| 2500 |
* Clean OTTO internal fetch query parameters from rendered HTML (WP-315). |
| 2501 |
* |
| 2502 |
* @param string $html The rendered HTML |
| 2503 |
* @return string HTML with OTTO fetch params removed |
| 2504 |
*/ |
| 2505 |
private function clean_otto_fetch_params($html) { |
| 2506 |
$patterns = [ |
| 2507 |
'/[?&]is_otto_page_fetch=1/', |
| 2508 |
'/[?&]otto_block_title=1/', |
| 2509 |
'/[?&]otto_block_desc=1/', |
| 2510 |
'/[?&]amp;is_otto_page_fetch=1/', |
| 2511 |
'/[?&]amp;otto_block_title=1/', |
| 2512 |
'/[?&]amp;otto_block_desc=1/', |
| 2513 |
]; |
| 2514 |
$html = preg_replace($patterns, '', $html); |
| 2515 |
|
| 2516 |
# WP-398: Protect <script>/<style> bodies before the URL cleanup below. |
| 2517 |
# The href/src/action matcher can otherwise match JS that looks like an |
| 2518 |
# attribute (e.g. location.href="/x??y", el.src="a??b") and collapse the |
| 2519 |
# ?? inside it. Swap each block for a placeholder comment, run the cleanup, |
| 2520 |
# then restore byte-for-byte. |
| 2521 |
$protected = []; |
| 2522 |
$html = preg_replace_callback( |
| 2523 |
'/<(script|style)(\b[^>]*)>([\s\S]*?)<\/\1>/i', |
| 2524 |
function ($m) use (&$protected) { |
| 2525 |
$key = '<!--METASYNC_FETCHCLEAN_' . count($protected) . '-->'; |
| 2526 |
$protected[$key] = $m[0]; |
| 2527 |
return $key; |
| 2528 |
}, |
| 2529 |
$html |
| 2530 |
); |
| 2531 |
|
| 2532 |
# WP-398: Normalize leftover query-string separators (?? / ?& / ?#) created |
| 2533 |
# by the param removal above — but ONLY inside URL attributes (href/src/action). |
| 2534 |
# The previous implementation ran str_replace('??','?') over the ENTIRE document, |
| 2535 |
# which corrupted inline JavaScript nullish operators (?? / ??=) and any literal |
| 2536 |
# "??" in page text — breaking Slider Revolution and other modern inline scripts. |
| 2537 |
$html = preg_replace_callback( |
| 2538 |
'/\b(href|src|action)(\s*=\s*)(["\'])(.*?)\3/is', |
| 2539 |
function ($m) { |
| 2540 |
$url = str_replace(['??', '?&', '?#'], ['?', '?', '#'], $m[4]); |
| 2541 |
return $m[1] . $m[2] . $m[3] . $url . $m[3]; |
| 2542 |
}, |
| 2543 |
$html |
| 2544 |
); |
| 2545 |
|
| 2546 |
# Restore protected <script>/<style> blocks byte-for-byte. |
| 2547 |
if (!empty($protected)) { |
| 2548 |
$html = str_replace(array_keys($protected), array_values($protected), $html); |
| 2549 |
} |
| 2550 |
|
| 2551 |
return $html; |
| 2552 |
} |
| 2553 |
|
| 2554 |
/** |
| 2555 |
* Fix Divi 5 shortcode framework class renumbering (WP-315) |
| 2556 |
* |
| 2557 |
* When Divi 5's shortcode framework loads during the internal HTTP fetch, |
| 2558 |
* it uses shared counters across all template parts (header + page + footer), |
| 2559 |
* causing page content element classes to be offset (et_pb_section_0 becomes |
| 2560 |
* et_pb_section_4, etc.). The cached CSS references 0-based numbering. |
| 2561 |
* |
| 2562 |
* Detects the offset per element type and remaps back to 0-based. |
| 2563 |
* |
| 2564 |
* @param string $html The processed HTML string |
| 2565 |
* @return string HTML with corrected Divi class numbering |
| 2566 |
*/ |
| 2567 |
private function fix_divi_class_renumbering($html) { |
| 2568 |
# Only run on Divi sites |
| 2569 |
if (!defined('ET_CORE_VERSION')) { |
| 2570 |
return $html; |
| 2571 |
} |
| 2572 |
|
| 2573 |
# Only apply to Divi pages with template builder |
| 2574 |
if (strpos($html, 'et-l et-l--post') === false) { |
| 2575 |
return $html; |
| 2576 |
} |
| 2577 |
|
| 2578 |
# Locate the page content area (between et-l--post and et-l--footer) |
| 2579 |
$page_start = strpos($html, 'et-l et-l--post'); |
| 2580 |
if ($page_start === false) { |
| 2581 |
return $html; |
| 2582 |
} |
| 2583 |
|
| 2584 |
$footer_start = strpos($html, 'et-l et-l--footer', $page_start); |
| 2585 |
if ($footer_start === false) { |
| 2586 |
$footer_start = strlen($html); |
| 2587 |
} |
| 2588 |
|
| 2589 |
$page_content = substr($html, $page_start, $footer_start - $page_start); |
| 2590 |
|
| 2591 |
|
| 2592 |
# Divi element types that get renumbered by the shortcode framework. |
| 2593 |
# IMPORTANT: 'blog' and 'portfolio' MUST be included — OTTO's HTTP render |
| 2594 |
# shifts their numbering (et_pb_blog_0 → et_pb_blog_1) because the internal |
| 2595 |
# wp_remote_get includes the Theme Builder header template in Divi's counter. |
| 2596 |
# If the blog module class doesn't match between page 1 (OTTO-processed) and |
| 2597 |
# page 2 (AJAX, no OTTO), Divi's pagination JS can't find the container → empty results (WP-315). |
| 2598 |
$types = [ |
| 2599 |
'section', 'row', 'column', 'text', 'blurb', 'toggle', |
| 2600 |
'button', 'image', 'group_carousel', |
| 2601 |
'blog', 'portfolio', 'filterable_portfolio', 'shop', |
| 2602 |
'heading', 'divider', 'code', 'icon', 'contact_form_7', |
| 2603 |
]; |
| 2604 |
|
| 2605 |
$offsets = []; |
| 2606 |
|
| 2607 |
foreach ($types as $type) { |
| 2608 |
# Find the FIRST numbered instance in page content (not _tb_ suffixed) |
| 2609 |
if (preg_match('/et_pb_' . preg_quote($type, '/') . '_(\d+)(?!_tb_)/', $page_content, $m)) { |
| 2610 |
$first_num = (int) $m[1]; |
| 2611 |
if ($first_num > 0) { |
| 2612 |
$offsets[$type] = $first_num; |
| 2613 |
} |
| 2614 |
} |
| 2615 |
} |
| 2616 |
|
| 2617 |
if (empty($offsets)) { |
| 2618 |
return $html; |
| 2619 |
} |
| 2620 |
|
| 2621 |
# Remap each element type back to 0-based numbering |
| 2622 |
foreach ($offsets as $etype => $offset) { |
| 2623 |
$escaped = preg_quote($etype, '/'); |
| 2624 |
$html = preg_replace_callback( |
| 2625 |
'/et_pb_' . $escaped . '_(\d+)(?!_tb_)/', |
| 2626 |
function ($m) use ($etype, $offset) { |
| 2627 |
$num = (int) $m[1]; |
| 2628 |
if ($num >= $offset) { |
| 2629 |
return 'et_pb_' . $etype . '_' . ($num - $offset); |
| 2630 |
} |
| 2631 |
return $m[0]; |
| 2632 |
}, |
| 2633 |
$html |
| 2634 |
); |
| 2635 |
} |
| 2636 |
|
| 2637 |
# Remove duplicate JS variable declarations from the shortcode framework |
| 2638 |
foreach (['et_pb_custom', 'et_frontend_scripts', 'et_builder_utils_params'] as $var_name) { |
| 2639 |
$needle = 'var ' . $var_name . ' = '; |
| 2640 |
$first = strpos($html, $needle); |
| 2641 |
if ($first !== false) { |
| 2642 |
$second = strpos($html, $needle, $first + strlen($needle)); |
| 2643 |
if ($second !== false) { |
| 2644 |
$end = strpos($html, ";\n", $second); |
| 2645 |
if ($end !== false) { |
| 2646 |
$html = substr($html, 0, $second) . substr($html, $end + 2); |
| 2647 |
} |
| 2648 |
} |
| 2649 |
} |
| 2650 |
} |
| 2651 |
|
| 2652 |
# Deduplicate animation data (shortcode framework creates duplicate entries) |
| 2653 |
if (preg_match('/var diviElementAnimationData = (\[.*?\]);/s', $html, $am)) { |
| 2654 |
$data = json_decode($am[1], true); |
| 2655 |
if (is_array($data)) { |
| 2656 |
$seen = []; |
| 2657 |
$unique = []; |
| 2658 |
foreach (array_reverse($data) as $e) { |
| 2659 |
$k = $e['class'] ?? ''; |
| 2660 |
if ($k !== '' && !isset($seen[$k])) { |
| 2661 |
$seen[$k] = true; |
| 2662 |
array_unshift($unique, $e); |
| 2663 |
} |
| 2664 |
} |
| 2665 |
if (count($unique) < count($data)) { |
| 2666 |
$html = str_replace( |
| 2667 |
$am[0], |
| 2668 |
'var diviElementAnimationData = ' . json_encode($unique, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE) . ';', |
| 2669 |
$html |
| 2670 |
); |
| 2671 |
} |
| 2672 |
} |
| 2673 |
} |
| 2674 |
|
| 2675 |
# Deduplicate multiview data |
| 2676 |
if (preg_match('/var diviElementMultiViewData = (\[.*?\]);/s', $html, $mv)) { |
| 2677 |
$data = json_decode($mv[1], true); |
| 2678 |
if (is_array($data)) { |
| 2679 |
$seen = []; |
| 2680 |
$unique = []; |
| 2681 |
foreach (array_reverse($data) as $e) { |
| 2682 |
$k = ($e['selector'] ?? '') . '|' . ($e['action'] ?? ''); |
| 2683 |
if ($k !== '|' && !isset($seen[$k])) { |
| 2684 |
$seen[$k] = true; |
| 2685 |
array_unshift($unique, $e); |
| 2686 |
} |
| 2687 |
} |
| 2688 |
if (count($unique) < count($data)) { |
| 2689 |
$html = str_replace( |
| 2690 |
$mv[0], |
| 2691 |
'var diviElementMultiViewData = ' . json_encode($unique, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE) . ';', |
| 2692 |
$html |
| 2693 |
); |
| 2694 |
} |
| 2695 |
} |
| 2696 |
} |
| 2697 |
|
| 2698 |
return $html; |
| 2699 |
} |
| 2700 |
|
| 2701 |
|
| 2702 |
} |
