| 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 |
* 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 |
* Quote unquoted attribute values before DOM parsing. |
| 131 |
* |
| 132 |
* Per the HTML spec an unquoted attribute value ends at whitespace, so |
| 133 |
* browsers read this correctly: |
| 134 |
* |
| 135 |
* data-autoplay-viewport=80% |
| 136 |
* data-mousewheel="false" |
| 137 |
* data-type="carousel" |
| 138 |
* |
| 139 |
* SimpleHtmlDom does not stop at the whitespace. It runs to the next quote |
| 140 |
* character, so the value becomes |
| 141 |
* `80%\n data-mousewheel="false"\n data-type="carousel"...` and EVERY |
| 142 |
* following attribute on the tag is swallowed into it and lost. |
| 143 |
* |
| 144 |
* Supreme Modules' image carousel emits exactly that markup. The attributes |
| 145 |
* lost after it include data-type, data-slider-orientation and |
| 146 |
* data-slideshow-to-show, so Swiper initialises on defaults, stacks all |
| 147 |
* slides and grows the page by one slide-set per autoplay tick until it is |
| 148 |
* tens of thousands of pixels tall. |
| 149 |
* |
| 150 |
* Normalising to `data-autoplay-viewport="80%"` before parsing makes the |
| 151 |
* markup unambiguous, so no attribute is lost. Already-quoted values are |
| 152 |
* left untouched, and <script>/<style> bodies are shielded so their |
| 153 |
* contents are never mistaken for tags. |
| 154 |
* |
| 155 |
* @param string $html Raw HTML before DOM parsing. |
| 156 |
* @return string HTML with every unquoted attribute value quoted. |
| 157 |
*/ |
| 158 |
private function quote_unquoted_attribute_values($html) { |
| 159 |
# Shield <script>/<style> bodies — their contents can contain '<' and '=' |
| 160 |
# and must never be treated as markup. |
| 161 |
$protected = array(); |
| 162 |
$html = preg_replace_callback( |
| 163 |
'/<(script|style)\b([^>]*)>(.*?)<\/\1>/is', |
| 164 |
function ($m) use (&$protected) { |
| 165 |
$token = '<!--METASYNC_ATTRQ_' . count($protected) . '-->'; |
| 166 |
$protected[$token] = $m[0]; |
| 167 |
return $token; |
| 168 |
}, |
| 169 |
$html |
| 170 |
); |
| 171 |
|
| 172 |
# Walk real tags only. The alternation lets a quoted value contain '>' |
| 173 |
# so the tag boundary is found correctly. Possessive quantifiers (*+ / ++) |
| 174 |
# are deliberate: the three branches are mutually exclusive on their first |
| 175 |
# character, so backtracking can never produce a different match, and |
| 176 |
# forbidding it removes the catastrophic-backtracking risk this shape |
| 177 |
# otherwise carries on markup with many quotes and no '>'. |
| 178 |
$result = preg_replace_callback( |
| 179 |
'/<([a-zA-Z][a-zA-Z0-9:-]*+)((?:[^>"\']++|"[^"]*+"|\'[^\']*+\')*+)>/s', |
| 180 |
function ($m) { |
| 181 |
if (strpos($m[2], '=') === false) { |
| 182 |
return $m[0]; |
| 183 |
} |
| 184 |
|
| 185 |
$attrs = $this->quote_unquoted_values_in_attr_region($m[2]); |
| 186 |
|
| 187 |
# null means nothing needed changing — return the ORIGINAL tag so |
| 188 |
# a document of well-formed HTML comes back byte-identical rather |
| 189 |
# than merely equivalent. |
| 190 |
return $attrs === null ? $m[0] : '<' . $m[1] . $attrs . '>'; |
| 191 |
}, |
| 192 |
$html |
| 193 |
); |
| 194 |
|
| 195 |
if ($result === null) { |
| 196 |
$result = $html; |
| 197 |
} |
| 198 |
|
| 199 |
if (!empty($protected)) { |
| 200 |
$result = str_replace(array_keys($protected), array_values($protected), $result); |
| 201 |
} |
| 202 |
|
| 203 |
return $result; |
| 204 |
} |
| 205 |
|
| 206 |
/** |
| 207 |
* Quote the unquoted values inside one tag's attribute region. |
| 208 |
* |
| 209 |
* Must be a tokeniser, not a regex. A flat pattern over the region cannot |
| 210 |
* tell an attribute from ordinary text inside an already-quoted value, and |
| 211 |
* would wreck common markup — `content="width=device-width, |
| 212 |
* initial-scale=1"` looks like it contains an `initial-scale=1` attribute. |
| 213 |
* Quoted values are therefore copied through verbatim and never inspected. |
| 214 |
* |
| 215 |
* @param string $attrs Attribute region: everything between the tag name and '>'. |
| 216 |
* @return string|null Rewritten region, or null when nothing needed changing. |
| 217 |
*/ |
| 218 |
private function quote_unquoted_values_in_attr_region($attrs) { |
| 219 |
$len = strlen($attrs); |
| 220 |
$out = ''; |
| 221 |
$i = 0; |
| 222 |
$changed = false; |
| 223 |
|
| 224 |
while ($i < $len) { |
| 225 |
$c = $attrs[$i]; |
| 226 |
|
| 227 |
# Whitespace between attributes — preserve exactly. |
| 228 |
if ($c === ' ' || $c === "\t" || $c === "\n" || $c === "\r" || $c === "\f" || $c === "\v") { |
| 229 |
$out .= $c; |
| 230 |
$i++; |
| 231 |
continue; |
| 232 |
} |
| 233 |
|
| 234 |
# Attribute name. |
| 235 |
if (preg_match('/\G[a-zA-Z_:][a-zA-Z0-9_:.\-]*+/', $attrs, $nm, 0, $i)) { |
| 236 |
$out .= $nm[0]; |
| 237 |
$i += strlen($nm[0]); |
| 238 |
|
| 239 |
# Look past optional whitespace for '='. No '=' means a valueless |
| 240 |
# boolean attribute; leave it alone. |
| 241 |
$j = $i; |
| 242 |
while ($j < $len && strpos(" \t\n\r\f\v", $attrs[$j]) !== false) { |
| 243 |
$j++; |
| 244 |
} |
| 245 |
if ($j >= $len || $attrs[$j] !== '=') { |
| 246 |
continue; |
| 247 |
} |
| 248 |
|
| 249 |
$out .= substr($attrs, $i, $j - $i) . '='; |
| 250 |
$i = $j + 1; |
| 251 |
|
| 252 |
# Whitespace after '='. |
| 253 |
while ($i < $len && strpos(" \t\n\r\f\v", $attrs[$i]) !== false) { |
| 254 |
$out .= $attrs[$i]; |
| 255 |
$i++; |
| 256 |
} |
| 257 |
|
| 258 |
if ($i >= $len) { |
| 259 |
continue; |
| 260 |
} |
| 261 |
|
| 262 |
# Quoted value — copy verbatim through the closing quote so its |
| 263 |
# contents are never treated as attributes. |
| 264 |
if ($attrs[$i] === '"' || $attrs[$i] === "'") { |
| 265 |
$quote = $attrs[$i]; |
| 266 |
$end = strpos($attrs, $quote, $i + 1); |
| 267 |
if ($end === false) { |
| 268 |
$out .= substr($attrs, $i); |
| 269 |
$i = $len; |
| 270 |
} else { |
| 271 |
$out .= substr($attrs, $i, $end - $i + 1); |
| 272 |
$i = $end + 1; |
| 273 |
} |
| 274 |
continue; |
| 275 |
} |
| 276 |
|
| 277 |
# Unquoted value — runs to the next whitespace. ('>' cannot occur: |
| 278 |
# it would have ended the tag before this region.) |
| 279 |
$start = $i; |
| 280 |
while ($i < $len && strpos(" \t\n\r\f\v", $attrs[$i]) === false) { |
| 281 |
$i++; |
| 282 |
} |
| 283 |
$value = substr($attrs, $start, $i - $start); |
| 284 |
|
| 285 |
if ($value === '') { |
| 286 |
continue; |
| 287 |
} |
| 288 |
|
| 289 |
# Only rewrite when it actually prevents a loss. SimpleHtmlDom |
| 290 |
# mis-parses an unquoted value by running from it to the NEXT |
| 291 |
# QUOTE CHARACTER, swallowing everything between. With no quote |
| 292 |
# left in this region there is nothing to swallow and the value |
| 293 |
# already parses correctly — so leave that markup byte-identical |
| 294 |
# rather than churn it. (OceanWP's `style=color:#;` as the final |
| 295 |
# attribute is the common benign case.) |
| 296 |
$rest = substr($attrs, $i); |
| 297 |
if (strpos($rest, '"') === false && strpos($rest, "'") === false) { |
| 298 |
$out .= $value; |
| 299 |
continue; |
| 300 |
} |
| 301 |
|
| 302 |
$out .= '"' . str_replace('"', '"', $value) . '"'; |
| 303 |
$changed = true; |
| 304 |
continue; |
| 305 |
} |
| 306 |
|
| 307 |
# Anything else (stray punctuation, a '/' before '>', malformed input) |
| 308 |
# — copy through untouched. |
| 309 |
$out .= $c; |
| 310 |
$i++; |
| 311 |
} |
| 312 |
|
| 313 |
return $changed ? $out : null; |
| 314 |
} |
| 315 |
|
| 316 |
/** |
| 317 |
* / Capture ALL <style> blocks before DOM processing. |
| 318 |
* |
| 319 |
* SimpleHtmlDom can lose or corrupt <style> blocks during parse/serialize, |
| 320 |
* especially those without an id attribute (common in custom themes). |
| 321 |
* This captures every <style> block so we can re-inject any that are lost. |
| 322 |
* |
| 323 |
* @param string $html Raw HTML before DOM parsing. |
| 324 |
* @return array Keyed by id (or content hash for anonymous styles) → full <style> tag. |
| 325 |
*/ |
| 326 |
private function capture_style_blocks($html) { |
| 327 |
$blocks = []; |
| 328 |
if (preg_match_all('/<style[^>]*>.*?<\/style>/si', $html, $matches)) { |
| 329 |
foreach ($matches[0] as $style_tag) { |
| 330 |
if (preg_match('/id=["\']([^"\']+)["\']/', $style_tag, $id_match)) { |
| 331 |
$key = $id_match[1]; |
| 332 |
} else { |
| 333 |
$key = 'anon_' . md5($style_tag); |
| 334 |
} |
| 335 |
$blocks[$key] = $style_tag; |
| 336 |
} |
| 337 |
} |
| 338 |
return $blocks; |
| 339 |
} |
| 340 |
|
| 341 |
/** |
| 342 |
* / Re-inject any <style> blocks lost during DOM processing. |
| 343 |
* |
| 344 |
* Compares the captured blocks against the final HTML and re-injects any |
| 345 |
* that disappeared. Named blocks (with id) are checked by id; anonymous |
| 346 |
* blocks are checked by exact content match. |
| 347 |
* |
| 348 |
* @param string $html Processed HTML after DOM modifications. |
| 349 |
* @param array $original_blocks Blocks returned by capture_style_blocks(). |
| 350 |
* @return string HTML with lost style blocks restored. |
| 351 |
*/ |
| 352 |
private function restore_lost_style_blocks($html, $original_blocks) { |
| 353 |
if (empty($original_blocks)) { |
| 354 |
return $html; |
| 355 |
} |
| 356 |
|
| 357 |
foreach ($original_blocks as $key => $style_tag) { |
| 358 |
if (strpos($key, 'anon_') === 0) { |
| 359 |
# Anonymous block — check if exact content is still present |
| 360 |
if (strpos($html, $style_tag) === false) { |
| 361 |
$html = str_replace('</head>', $style_tag . "\n" . '</head>', $html); |
| 362 |
} |
| 363 |
} else { |
| 364 |
# Named block — check if its id is still in the document |
| 365 |
if (strpos($html, 'id="' . $key . '"') === false |
| 366 |
&& strpos($html, "id='" . $key . "'") === false) { |
| 367 |
$html = str_replace('</head>', $style_tag . "\n" . '</head>', $html); |
| 368 |
} |
| 369 |
} |
| 370 |
} |
| 371 |
|
| 372 |
return $html; |
| 373 |
} |
| 374 |
|
| 375 |
/** |
| 376 |
* Build the new inner HTML for a heading whose text OTTO is replacing, |
| 377 |
* preserving the author's inline formatting. |
| 378 |
* |
| 379 |
* Headings are matched by their stripped text, so the recommended value is |
| 380 |
* plain text with no way to know how it maps onto the original markup. |
| 381 |
* Writing it back directly would discard whatever was inside — a coloured |
| 382 |
* <span>, <em>, <br> line breaks — and visibly restyle the page. |
| 383 |
* |
| 384 |
* Two cases are safe: |
| 385 |
* - the heading is already plain text -> replace it |
| 386 |
* - the heading is ONE nested inline wrapper -> keep the wrappers, |
| 387 |
* chain around a single run of text swap only the text |
| 388 |
* |
| 389 |
* Anything else is ambiguous: sibling elements, <br>-separated lines, or |
| 390 |
* text mixed with elements have no single place the new sentence belongs. |
| 391 |
* Those return null so the caller leaves the heading alone — an unapplied |
| 392 |
* suggestion is better than silently restyling the customer's content. |
| 393 |
* |
| 394 |
* @param string $inner_html Current inner HTML of the heading. |
| 395 |
* @param string $replacement_text New text, already escaped by the caller. |
| 396 |
* @return string|null New inner HTML, or null to skip this heading. |
| 397 |
*/ |
| 398 |
public static function build_heading_inner_html($inner_html, $replacement_text) { |
| 399 |
$inner = trim($inner_html); |
| 400 |
|
| 401 |
# Plain text — nothing to preserve. |
| 402 |
if (strpos($inner, '<') === false) { |
| 403 |
return $replacement_text; |
| 404 |
} |
| 405 |
|
| 406 |
# Inline elements whose wrapper carries the styling worth keeping. |
| 407 |
# Deliberately excludes <br>: it encodes line structure, not styling, |
| 408 |
# and its presence means the heading has more than one text run. |
| 409 |
$inline = 'span|em|strong|b|i|u|a|small|mark|abbr|cite|q|s|sub|sup|font'; |
| 410 |
|
| 411 |
$open = ''; |
| 412 |
$close = ''; |
| 413 |
|
| 414 |
# Peel wrappers outermost-in. The greedy (.*) means a heading with |
| 415 |
# sibling elements (e.g. <em>A</em><br><em>B</em>) leaves stray tags in |
| 416 |
# $inner, which the ambiguity check below then rejects. |
| 417 |
while (preg_match('#^<(' . $inline . ')(\s[^>]*)?>(.*)</\1\s*>$#is', $inner, $m)) { |
| 418 |
# Group 2 (the attribute list) is optional but always captured as '' |
| 419 |
# when absent, because group 3 follows it. |
| 420 |
$open .= '<' . $m[1] . $m[2] . '>'; |
| 421 |
$close = '</' . $m[1] . '>' . $close; |
| 422 |
$inner = trim($m[3]); |
| 423 |
|
| 424 |
if (strpos($inner, '<') === false) { |
| 425 |
return $open . $replacement_text . $close; |
| 426 |
} |
| 427 |
} |
| 428 |
|
| 429 |
# Still markup left after peeling — ambiguous, so don't touch it. |
| 430 |
return null; |
| 431 |
} |
| 432 |
|
| 433 |
/** |
| 434 |
* / Protect entity-encoded `srcdoc` attribute values before DOM parsing. |
| 435 |
* |
| 436 |
* Lazy-loaded YouTube/video facades embed a full HTML document inside an |
| 437 |
* iframe's `srcdoc` attribute, entity-encoded: |
| 438 |
* <iframe srcdoc="<style>*{overflow:hidden}...</style><a>..."></iframe> |
| 439 |
* SimpleHtmlDom decodes that value and re-emits the inner <style>/<a>/<img> |
| 440 |
* as REAL nodes hoisted into the document <head>. The facade's global CSS |
| 441 |
* (`*{overflow:hidden}`, `img,span{position:absolute;width:100%;...}`) then |
| 442 |
* nukes the entire page layout — the page renders blank below the header even |
| 443 |
* though all body content is still present (so the element-loss guard can't |
| 444 |
* see it). We swap each srcdoc value for an opaque token before parsing and |
| 445 |
* restore it verbatim afterward, so SimpleHtmlDom never touches it. |
| 446 |
* |
| 447 |
* @param string $html Raw HTML before DOM parsing. |
| 448 |
* @param array $store (out) token => original encoded value. |
| 449 |
* @return string HTML with srcdoc values tokenized. |
| 450 |
*/ |
| 451 |
private function protect_srcdoc_attributes($html, &$store) { |
| 452 |
$store = array(); |
| 453 |
$index = 0; |
| 454 |
return preg_replace_callback( |
| 455 |
'/\bsrcdoc\s*=\s*(["\'])(.*?)\1/is', |
| 456 |
function ($m) use (&$store, &$index) { |
| 457 |
$token = '__METASYNC_SRCDOC_' . ($index++) . '__'; |
| 458 |
$store[$token] = $m[2]; |
| 459 |
return 'srcdoc="' . $token . '"'; |
| 460 |
}, |
| 461 |
$html |
| 462 |
); |
| 463 |
} |
| 464 |
|
| 465 |
/** |
| 466 |
* Restore srcdoc values tokenized by protect_srcdoc_attributes(). |
| 467 |
* Tokens are unique and never altered by SimpleHtmlDom, so a plain |
| 468 |
* str_replace on the final output is safe. |
| 469 |
* |
| 470 |
* @param string $html Processed HTML. |
| 471 |
* @param array $store Token map from protect_srcdoc_attributes(). |
| 472 |
* @return string |
| 473 |
*/ |
| 474 |
private function restore_srcdoc_attributes($html, $store) { |
| 475 |
if (empty($store)) { |
| 476 |
return $html; |
| 477 |
} |
| 478 |
return str_replace(array_keys($store), array_values($store), $html); |
| 479 |
} |
| 480 |
|
| 481 |
/** |
| 482 |
* Safety net — strip any lazy-video FACADE <style> that leaked into the |
| 483 |
* document as a real stylesheet. |
| 484 |
* |
| 485 |
* Lazy YouTube/video facades carry this exact CSS reset inside their iframe |
| 486 |
* srcdoc: *{...overflow:hidden} html,body{height:100%} img,span{position:absolute;width:100%;...} |
| 487 |
* If anything (SimpleHtmlDom attribute-decode, capture/restore_lost_style_blocks) |
| 488 |
* hoists it out of the srcdoc into a real <style>, it absolutely-positions every |
| 489 |
* image and span on the page and clips all overflow — rendering the page blank |
| 490 |
* below the header. No legitimate global stylesheet ever does this to img,span, |
| 491 |
* so removing such a block is safe. |
| 492 |
* |
| 493 |
* Must run AFTER srcdoc values are tokenized (protect_srcdoc_attributes) so the |
| 494 |
* encoded copy still living inside the iframe attribute is never matched. |
| 495 |
* |
| 496 |
* @param string $html |
| 497 |
* @return string |
| 498 |
*/ |
| 499 |
/** |
| 500 |
* Crash-free guard for full-document string transforms. |
| 501 |
* |
| 502 |
* PCRE operations (preg_replace / preg_replace_callback) return NULL when they |
| 503 |
* hit a limit — most notably "JIT stack limit exhausted" on Divi/page-builder |
| 504 |
* pages whose large inline <style> blocks defeat tempered-greedy patterns. |
| 505 |
* Historically that NULL propagated through the rest of the pipeline, so |
| 506 |
* process_html_directly() returned empty HTML, the render strategy's 50%-size |
| 507 |
* sanity check rejected it, and the ORIGINAL un-optimised page was served — |
| 508 |
* OTTO silently did nothing. |
| 509 |
* |
| 510 |
* This guard keeps the last-good HTML whenever a transform yields NULL (or an |
| 511 |
* unexpectedly empty string), so a single failing regex can no longer discard |
| 512 |
* every OTTO optimisation. It is defence-in-depth: the specific pattern that |
| 513 |
* triggered (strip_hoisted_facade_styles) has been rewritten to be |
| 514 |
* non-backtracking, but this guard still protects the rest of the pipeline. |
| 515 |
* |
| 516 |
* @param string|null $new Result of the transform. |
| 517 |
* @param string $prev HTML before the transform (fallback value). |
| 518 |
* @param string $where Transform name, for the log line. |
| 519 |
* @return string |
| 520 |
*/ |
| 521 |
private function otto_guard_html($new, $prev, $where = '') { |
| 522 |
if ($new === null || (is_string($new) && $new === '' && $prev !== '')) { |
| 523 |
# A full-document transform ($where) returned NULL (e.g. a PCRE |
| 524 |
# limit such as "JIT stack limit exhausted"). Keep the last-good |
| 525 |
# HTML so one failed step can't discard every OTTO optimisation. |
| 526 |
# Intentionally silent — not logged, to keep customer logs clean. |
| 527 |
return $prev; |
| 528 |
} |
| 529 |
return $new; |
| 530 |
} |
| 531 |
|
| 532 |
private function strip_hoisted_facade_styles($html) { |
| 533 |
# Non-backtracking implementation. |
| 534 |
# |
| 535 |
# The previous single-regex approach used two tempered-greedy segments |
| 536 |
# ((?:(?!</style>).)*?) to keep the match inside one <style> block. That |
| 537 |
# runs a negative lookahead for every character, and on Divi/page-builder |
| 538 |
# pages whose inline <style> blocks are tens of KB it overflows the PCRE |
| 539 |
# JIT stack ("JIT stack limit exhausted"). preg_replace() then returns |
| 540 |
# NULL, which discarded the entire OTTO-modified document. |
| 541 |
# |
| 542 |
# Instead, isolate each <style>…</style> block with a single lazy .*? |
| 543 |
# (cheap, no per-char lookahead) and test only that bounded block for the |
| 544 |
# facade signature. The signature check runs on one small block at a time, |
| 545 |
# so neither pattern can strain the JIT stack. |
| 546 |
if (!is_string($html) || stripos($html, '<style') === false) { |
| 547 |
return $html; |
| 548 |
} |
| 549 |
|
| 550 |
$out = preg_replace_callback( |
| 551 |
'#<style\b[^>]*>.*?</style>#is', |
| 552 |
static function ($m) { |
| 553 |
# Facade reset: "img,span{ … position:absolute … }" inside this block. |
| 554 |
if (preg_match('#\bimg\s*,\s*span\s*\{[^}]*position\s*:\s*absolute[^}]*\}#is', $m[0])) { |
| 555 |
return ''; |
| 556 |
} |
| 557 |
return $m[0]; |
| 558 |
}, |
| 559 |
$html |
| 560 |
); |
| 561 |
|
| 562 |
# Belt-and-suspenders: if PCRE still fails for any reason, keep the input. |
| 563 |
return $out === null ? $html : $out; |
| 564 |
} |
| 565 |
|
| 566 |
/** |
| 567 |
* Apply OTTO's canonical via string replacement (reliable fallback). |
| 568 |
* |
| 569 |
* Why the DOM path fails: do_header_replacements() sets the canonical with a |
| 570 |
* DOM attribute edit ($link->href = …). That edit works in isolation, but |
| 571 |
* insert_header_html() runs *earlier* in the same pipeline and reassigns |
| 572 |
* $head->outertext to a raw string (to inject OTTO's schema). In SimpleHtmlDom, |
| 573 |
* assigning ->outertext freezes that node — <head> now serializes as that |
| 574 |
* literal string and no longer re-renders from its child node tree. The |
| 575 |
* canonical <link> is captured inside that frozen string with its OLD href, so |
| 576 |
* the later $link->href edit is never reflected in the output and the SEO |
| 577 |
* plugin's (Yoast/Rank Math/AIOSEO) canonical always wins. This is the same |
| 578 |
* head-freeze that forced title/meta onto string-replacement fallbacks; |
| 579 |
* canonical was the one case that never got one. |
| 580 |
* |
| 581 |
* This pass runs on the final serialized HTML (after the freeze), so it is |
| 582 |
* immune to the ordering problem. It guarantees exactly one canonical — |
| 583 |
* OTTO's: it removes every existing <link rel="canonical"> and inserts OTTO's |
| 584 |
* recommended value (marked data-otto="true") right after <head>. Manual |
| 585 |
* canonicals (_metasync_canonical_url / meta_canonical) still take priority, |
| 586 |
* matching the DOM path's protection. |
| 587 |
* |
| 588 |
* @param string $html Serialized HTML. |
| 589 |
* @param array $replacement_data OTTO suggestions. |
| 590 |
* @return string |
| 591 |
*/ |
| 592 |
private function apply_canonical_via_string($html, $replacement_data) { |
| 593 |
if (!is_string($html) || empty($replacement_data['header_replacements']) || !is_array($replacement_data['header_replacements'])) { |
| 594 |
return $html; |
| 595 |
} |
| 596 |
|
| 597 |
# Find OTTO's canonical recommendation. |
| 598 |
$canonical = ''; |
| 599 |
foreach ($replacement_data['header_replacements'] as $item) { |
| 600 |
if (($item['type'] ?? '') === 'link' && ($item['rel'] ?? '') === 'canonical') { |
| 601 |
$canonical = $item['recommended_value'] ?? $item['value'] ?? ''; |
| 602 |
break; |
| 603 |
} |
| 604 |
} |
| 605 |
if (empty($canonical) || !is_string($canonical)) { |
| 606 |
return $html; |
| 607 |
} |
| 608 |
|
| 609 |
# Validate OTTO's own suggestion too: if the platform payload |
| 610 |
# carries a corrupted/non-URL canonical, leave the document untouched. |
| 611 |
$canonical = Metasync_Canonical_Sanitizer::sanitize($canonical); |
| 612 |
if ($canonical === '') { |
| 613 |
return $html; |
| 614 |
} |
| 615 |
|
| 616 |
# Respect a manually-set canonical (same protection as the DOM path). |
| 617 |
# Validated: a legacy row corrupted to "Array" must not count |
| 618 |
# as a manual canonical — that would suppress OTTO's correct value. |
| 619 |
if (function_exists('is_singular') && is_singular()) { |
| 620 |
$post_id = function_exists('get_the_ID') ? get_the_ID() : 0; |
| 621 |
if ($post_id) { |
| 622 |
$custom = Metasync_Canonical_Sanitizer::sanitize(get_post_meta($post_id, '_metasync_canonical_url', true)); |
| 623 |
if ($custom === '') { |
| 624 |
$custom = Metasync_Canonical_Sanitizer::sanitize(get_post_meta($post_id, 'meta_canonical', true)); |
| 625 |
} |
| 626 |
if ($custom !== '') { |
| 627 |
return $html; # manual canonical wins — leave the document untouched |
| 628 |
} |
| 629 |
} |
| 630 |
} |
| 631 |
|
| 632 |
# Only proceed if there is a <head> to place the tag in (never end up with zero canonical). |
| 633 |
if (!preg_match('#<head\b[^>]*>#i', $html)) { |
| 634 |
return $html; |
| 635 |
} |
| 636 |
|
| 637 |
$tag = '<link rel="canonical" href="' . htmlspecialchars($canonical, ENT_QUOTES, 'UTF-8') . '" data-otto="true" />'; |
| 638 |
|
| 639 |
# Remove every existing canonical link (bounded per-tag pattern; null-safe). |
| 640 |
$stripped = preg_replace('#<link\b[^>]*\brel=(["\'])canonical\1[^>]*>\s*#i', '', $html); |
| 641 |
if (is_string($stripped)) { |
| 642 |
$html = $stripped; |
| 643 |
} |
| 644 |
|
| 645 |
# Insert OTTO's canonical right after <head> (callback avoids $/\ interpolation from the URL). |
| 646 |
$inserted = preg_replace_callback('#(<head\b[^>]*>)#i', function ($m) use ($tag) { |
| 647 |
return $m[1] . "\n" . $tag; |
| 648 |
}, $html, 1); |
| 649 |
if (is_string($inserted)) { |
| 650 |
$html = $inserted; |
| 651 |
} |
| 652 |
|
| 653 |
return $html; |
| 654 |
} |
| 655 |
|
| 656 |
/** |
| 657 |
* Keep the charset declaration within the first 1024 bytes of <head>. |
| 658 |
* |
| 659 |
* OTTO injects meta tags + JSON-LD schema at the top of <head>, which can push |
| 660 |
* the theme's <meta charset="utf-8"> past the 1024-byte limit that browsers |
| 661 |
* enforce for in-document charset detection (HTML spec). When that happens — |
| 662 |
* and a cached/CDN response is served without an HTTP charset header — the |
| 663 |
* document falls back to the locale encoding (Windows-1252). External |
| 664 |
* stylesheets that declare no @charset of their own (e.g. a theme rule like |
| 665 |
* `content:"\2713"` written as a raw UTF-8 ✓) then inherit that wrong encoding |
| 666 |
* and render as mojibake ("âœ"" instead of "✓"). |
| 667 |
* |
| 668 |
* We guarantee a <meta charset="UTF-8"> as the first child of <head> whenever |
| 669 |
* one isn't already present within the first 1024 bytes. Harmless when a valid |
| 670 |
* early charset already exists (we skip), and a duplicate later declaration is |
| 671 |
* ignored by the browser (first one wins). |
| 672 |
* |
| 673 |
* @param string $html |
| 674 |
* @return string |
| 675 |
*/ |
| 676 |
private function ensure_early_charset_meta($html) { |
| 677 |
if (!preg_match('/<head\b[^>]*>/i', $html, $m, PREG_OFFSET_CAPTURE)) { |
| 678 |
return $html; |
| 679 |
} |
| 680 |
$inner_start = $m[0][1] + strlen($m[0][0]); |
| 681 |
|
| 682 |
# Already declared early enough? Leave it alone (also covers AMP, which |
| 683 |
# requires charset as the first child). |
| 684 |
$window = substr($html, $inner_start, 1024); |
| 685 |
if (preg_match('/<meta[^>]*charset/i', $window)) { |
| 686 |
return $html; |
| 687 |
} |
| 688 |
|
| 689 |
return substr($html, 0, $inner_start) |
| 690 |
. '<meta charset="UTF-8">' |
| 691 |
. substr($html, $inner_start); |
| 692 |
} |
| 693 |
|
| 694 |
# |
| 695 |
function __construct($otto_uuid){ |
| 696 |
|
| 697 |
# set the site uuid using the provided string |
| 698 |
$this->site_uuid = $otto_uuid; |
| 699 |
|
| 700 |
# Use endpoint manager if available, otherwise fallback to production |
| 701 |
if (class_exists('Metasync_Endpoint_Manager')) { |
| 702 |
$this->otto_end_point = Metasync_Endpoint_Manager::get_endpoint('OTTO_URL_DETAILS'); |
| 703 |
} else { |
| 704 |
$this->otto_end_point = 'https://sa.searchatlas.com/api/v2/otto-url-details'; |
| 705 |
} |
| 706 |
|
| 707 |
# laod the simple html dom parser with UTF-8 charset to handle special characters |
| 708 |
$this->dom = new HtmlDocument(null, true, true, 'UTF-8', false); |
| 709 |
} |
| 710 |
|
| 711 |
/** |
| 712 |
* Check Route Method |
| 713 |
* @param route : The route to check |
| 714 |
* @param path : The path of the html file to save |
| 715 |
*/ |
| 716 |
function process_route($route, $file_path){ |
| 717 |
|
| 718 |
# Construct the full endpoint URL with query parameters |
| 719 |
$url_with_params = add_query_arg( |
| 720 |
[ |
| 721 |
'url' => $route, |
| 722 |
'uuid' => $this->site_uuid, |
| 723 |
], |
| 724 |
$this->otto_end_point |
| 725 |
); |
| 726 |
|
| 727 |
# PERFORMANCE FIX: Add timeout to prevent blocking |
| 728 |
$args = array( |
| 729 |
'timeout' => 5, // 5 second max timeout (allow time for redirects) |
| 730 |
'redirection' => 5, // CRITICAL FIX: Allow redirects (API returns 301) |
| 731 |
'user-agent' => 'MetaSync-OTTO-SSR/2.0', |
| 732 |
'sslverify' => true |
| 733 |
); |
| 734 |
|
| 735 |
# Perform the GET request with timeout |
| 736 |
$response = wp_remote_get($url_with_params, $args); |
| 737 |
|
| 738 |
# Check for errors |
| 739 |
if (is_wp_error($response)) { |
| 740 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': API call failed - ' . $response->get_error_message()); |
| 741 |
return false; |
| 742 |
} |
| 743 |
|
| 744 |
# get the response body |
| 745 |
$body = wp_remote_retrieve_body($response); |
| 746 |
|
| 747 |
# Get the response code |
| 748 |
$response_code = wp_remote_retrieve_response_code($response); |
| 749 |
|
| 750 |
# if no change data skip |
| 751 |
if (empty($body) || $response_code !== 200){ |
| 752 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': API returned empty or non-200. Code: ' . $response_code); |
| 753 |
return false; |
| 754 |
} |
| 755 |
|
| 756 |
# set the html file path |
| 757 |
$this->html_file = $file_path; |
| 758 |
|
| 759 |
# load change data |
| 760 |
$change_data = json_decode($body, true); |
| 761 |
|
| 762 |
# Process with the fetched data |
| 763 |
return $this->process_route_with_data($route, $change_data, $file_path); |
| 764 |
} |
| 765 |
|
| 766 |
/** |
| 767 |
* Process route with pre-fetched suggestions data |
| 768 |
* OPTION 1: Used when data comes from transient cache |
| 769 |
* @param route : The route to check |
| 770 |
* @param change_data : Pre-fetched OTTO suggestions data |
| 771 |
* @param path : The path of the html file to save |
| 772 |
*/ |
| 773 |
function process_route_with_data($route, $change_data, $file_path){ |
| 774 |
|
| 775 |
if (empty($change_data) || !is_array($change_data)) { |
| 776 |
return false; |
| 777 |
} |
| 778 |
|
| 779 |
# set the html file path |
| 780 |
$this->html_file = $file_path; |
| 781 |
|
| 782 |
# Analyze what Otto is providing and store for conditional SEO blocking |
| 783 |
$has_otto_title = false; |
| 784 |
$otto_description_tags = []; // Track specific description tags Otto provides |
| 785 |
|
| 786 |
if (!empty($change_data['header_replacements']) && is_array($change_data['header_replacements'])) { |
| 787 |
foreach ($change_data['header_replacements'] as $item) { |
| 788 |
if (!empty($item['type'])) { |
| 789 |
# Check if Otto has title |
| 790 |
if ($item['type'] == 'title' && !empty($item['recommended_value'])) { |
| 791 |
$has_otto_title = true; |
| 792 |
} |
| 793 |
# Check if Otto has description - track specific tag types |
| 794 |
if ($item['type'] == 'meta') { |
| 795 |
# Check for meta[name=description] |
| 796 |
if (!empty($item['name']) && $item['name'] == 'description' && !empty($item['recommended_value'])) { |
| 797 |
$otto_description_tags[] = 'meta[name=description]'; |
| 798 |
} |
| 799 |
# Check for meta[property=og:description] |
| 800 |
if (!empty($item['property']) && $item['property'] == 'og:description' && !empty($item['recommended_value'])) { |
| 801 |
$otto_description_tags[] = 'meta[property=og:description]'; |
| 802 |
} |
| 803 |
# Check for meta[name=twitter:description] |
| 804 |
if (!empty($item['name']) && $item['name'] == 'twitter:description' && !empty($item['recommended_value'])) { |
| 805 |
$otto_description_tags[] = 'meta[name=twitter:description]'; |
| 806 |
} |
| 807 |
} |
| 808 |
} |
| 809 |
} |
| 810 |
} |
| 811 |
|
| 812 |
# Check header_html_insertion for description - must have non-empty content value |
| 813 |
if (!empty($change_data['header_html_insertion'])) { |
| 814 |
if (preg_match('/<meta[^>]*name=["\']description["\'][^>]*content=["\']([^"\']+)["\'][^>]*>/i', $change_data['header_html_insertion'])) { |
| 815 |
$otto_description_tags[] = 'meta[name=description]'; |
| 816 |
} |
| 817 |
} |
| 818 |
|
| 819 |
# Remove duplicates |
| 820 |
$otto_description_tags = array_unique($otto_description_tags); |
| 821 |
|
| 822 |
# Store blocking flags to pass to handle_route_html |
| 823 |
# This will be added to the internal fetch URL as parameters |
| 824 |
$change_data['_otto_blocking'] = array( |
| 825 |
'block_title' => $has_otto_title, |
| 826 |
'block_description_tags' => $otto_description_tags // Pass array of specific tags to remove |
| 827 |
); |
| 828 |
|
| 829 |
# Process the route with the suggestions data |
| 830 |
return $this->handle_route_html($route, $change_data); |
| 831 |
|
| 832 |
} |
| 833 |
|
| 834 |
# function to get tag attributes |
| 835 |
function get_tag_attributes($tag){ |
| 836 |
|
| 837 |
# Extract existing attributes of the <body> tag |
| 838 |
$attributes = []; |
| 839 |
|
| 840 |
|
| 841 |
# set the tag attributes |
| 842 |
$tag_attributes = []; |
| 843 |
|
| 844 |
|
| 845 |
# check that the tag attributes |
| 846 |
if(!is_object($tag) || !method_exists($tag, 'getAllAttributes')){ |
| 847 |
return ''; |
| 848 |
} |
| 849 |
|
| 850 |
# get the tag attributes |
| 851 |
$tag_attributes = $tag->getAllAttributes(); |
| 852 |
|
| 853 |
# loop all attributes |
| 854 |
foreach ($tag_attributes as $key => $value) { |
| 855 |
|
| 856 |
if ($value == 1) { |
| 857 |
|
| 858 |
# Handle boolean attributes |
| 859 |
$attributes[] = htmlspecialchars($key, ENT_QUOTES); |
| 860 |
} else { |
| 861 |
|
| 862 |
# Handle attributes with values |
| 863 |
$attributes[] = $key . '="' . htmlspecialchars($value, ENT_QUOTES) . '"'; |
| 864 |
} |
| 865 |
} |
| 866 |
|
| 867 |
# Convert attributes array to a string |
| 868 |
$attributes_string = !empty($attributes) ? ' ' . implode(' ', $attributes) : ''; |
| 869 |
|
| 870 |
# return the attributes string |
| 871 |
return $attributes_string; |
| 872 |
} |
| 873 |
|
| 874 |
function handle_route_html($route, $replacement_data){ |
| 875 |
|
| 876 |
# Detect if current page uses Brizy and disable SG Cache if so |
| 877 |
# Using global function defined in otto_pixel.php |
| 878 |
if (function_exists('metasync_otto_disable_sg_cache_for_brizy')) { |
| 879 |
metasync_otto_disable_sg_cache_for_brizy(); |
| 880 |
} |
| 881 |
|
| 882 |
# lablel the Otto Route |
| 883 |
# label otto requests to avoid loops |
| 884 |
// $request_body = add_query_arg( |
| 885 |
// [ |
| 886 |
// 'is_otto_page_fetch' => 1 |
| 887 |
// ], |
| 888 |
// $route |
| 889 |
// ); |
| 890 |
# Add blocking flags as URL parameters (no database writes!) |
| 891 |
$url_params = ['is_otto_page_fetch' => 1]; |
| 892 |
|
| 893 |
# Add blocking flags if available |
| 894 |
if (!empty($replacement_data['_otto_blocking'])) { |
| 895 |
$url_params['otto_block_title'] = $replacement_data['_otto_blocking']['block_title'] ? '1' : '0'; |
| 896 |
# For HTTP fetch path, check if any description tags need blocking |
| 897 |
$block_description_tags = $replacement_data['_otto_blocking']['block_description_tags'] ?? []; |
| 898 |
$url_params['otto_block_desc'] = !empty($block_description_tags) ? '1' : '0'; |
| 899 |
} |
| 900 |
|
| 901 |
$request_body = add_query_arg($url_params, $route); |
| 902 |
|
| 903 |
# TUNNEL/PROXY SUPPORT: If site is behind a tunnel (ngrok, zrok, etc.) |
| 904 |
# and loopback requests fail, try using localhost instead |
| 905 |
$request_body = apply_filters('metasync_otto_internal_fetch_url', $request_body, $route); |
| 906 |
# set cookie header var |
| 907 |
$cookie_header = ''; |
| 908 |
|
| 909 |
# loop cookies to set header |
| 910 |
foreach ($_COOKIE as $name => $value) { |
| 911 |
|
| 912 |
# handle array values by converting to string |
| 913 |
$cookie_value = is_array($value) ? serialize($value) : $value; |
| 914 |
|
| 915 |
# add cookie to header |
| 916 |
# $cookie_header .= $name . '=' . $value . '; '; |
| 917 |
$cookie_header .= $name . '=' . $cookie_value . '; '; |
| 918 |
} |
| 919 |
|
| 920 |
# trim the string |
| 921 |
$cookie_header = rtrim($cookie_header, '; '); |
| 922 |
|
| 923 |
# Allow timeout customization for slow tunnel environments |
| 924 |
$fetch_timeout = apply_filters('metasync_otto_internal_fetch_timeout', 5); |
| 925 |
|
| 926 |
$args = array( |
| 927 |
'sslverify' => false, // Disabled for localhost/tunnel environments |
| 928 |
'timeout' => $fetch_timeout, // Configurable timeout for tunnels |
| 929 |
'redirection' => 5, |
| 930 |
'httpversion' => '1.1', |
| 931 |
'headers' => array( |
| 932 |
'Cookie' => $cookie_header, |
| 933 |
'Cache-Control' => 'no-cache, no-store, must-revalidate', |
| 934 |
'Pragma' => 'no-cache', |
| 935 |
'X-OTTO-Internal-Fetch' => '1', |
| 936 |
'User-Agent' => 'MetaSync-OTTO-SSR/3.0', |
| 937 |
'X-Forwarded-Host' => $_SERVER['HTTP_HOST'] ?? '', // Preserve original host for tunnels |
| 938 |
) |
| 939 |
); |
| 940 |
|
| 941 |
# get the associateed route html |
| 942 |
$route_html = wp_remote_get($request_body, $args); |
| 943 |
|
| 944 |
# Check for timeout or connection errors |
| 945 |
if (is_wp_error($route_html)) { |
| 946 |
$error_msg = $route_html->get_error_message(); |
| 947 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ' DEBUG: FAILED - wp_remote_get error: ' . $error_msg . ' for route: ' . $route); |
| 948 |
return false; |
| 949 |
} |
| 950 |
|
| 951 |
# get body |
| 952 |
$html_body = wp_remote_retrieve_body($route_html); |
| 953 |
|
| 954 |
# Get the response code |
| 955 |
$response_code = wp_remote_retrieve_response_code($route_html); |
| 956 |
|
| 957 |
|
| 958 |
# check not empty |
| 959 |
if(empty($html_body) || $response_code !== 200){ |
| 960 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ' DEBUG: FAILED - Empty body or non-200 status for route: ' . $route); |
| 961 |
return false; |
| 962 |
} |
| 963 |
|
| 964 |
# Skip DOM processing on oversized documents to avoid fatal OOM. |
| 965 |
if (class_exists('Metasync_Otto_Render_Strategy') |
| 966 |
&& !Metasync_Otto_Render_Strategy::is_document_processable(strlen($html_body)) |
| 967 |
) { |
| 968 |
Metasync_Otto_Render_Strategy::log_oversized_skip('handle_route_html', strlen($html_body)); |
| 969 |
return false; |
| 970 |
} |
| 971 |
|
| 972 |
# Remove XML declaration |
| 973 |
$html_body = preg_replace('/<\?xml[^?]*\?>\s*/i', '', $html_body); |
| 974 |
|
| 975 |
# / Save ALL original <style> blocks before DOM processing. |
| 976 |
# SimpleHtmlDom can lose or corrupt style blocks during parse/serialize — |
| 977 |
# not only Divi's id-tagged ones but also anonymous <style> blocks used by |
| 978 |
# custom themes for nav collapse CSS, responsive breakpoints, etc. |
| 979 |
$original_style_blocks = $this->capture_style_blocks($html_body); |
| 980 |
|
| 981 |
# Shield entity-encoded srcdoc values (lazy YouTube/video facades) |
| 982 |
# so SimpleHtmlDom can't decode them and hoist their global <style> into <head>. |
| 983 |
$srcdoc_store = array(); |
| 984 |
$html_body = $this->protect_srcdoc_attributes($html_body, $srcdoc_store); |
| 985 |
|
| 986 |
# Escape bare < in text content (e.g. "<4 microns") before DOM parsing |
| 987 |
$html_body = $this->sanitize_text_less_than($html_body); |
| 988 |
|
| 989 |
# Fix malformed self-closing non-void tags (e.g. <ul/ class="...">) |
| 990 |
# before DOM parsing — SimpleHtmlDom strips attributes from these. |
| 991 |
$html_body = $this->fix_malformed_self_closing_tags($html_body); |
| 992 |
|
| 993 |
# Quote unquoted attribute values (e.g. data-autoplay-viewport=80%) — |
| 994 |
# SimpleHtmlDom swallows every following attribute into such a value. |
| 995 |
$html_body = $this->quote_unquoted_attribute_values($html_body); |
| 996 |
|
| 997 |
# now that the html is not empty |
| 998 |
# load it into the simple html dom |
| 999 |
$this->dom->load($html_body, true, false); |
| 1000 |
|
| 1001 |
# Force UTF-8 charset to preserve emojis and special characters |
| 1002 |
# This overrides any charset detection from HTML meta tags |
| 1003 |
$this->dom->_charset = 'UTF-8'; |
| 1004 |
$this->dom->_target_charset = 'UTF-8'; |
| 1005 |
|
| 1006 |
# PERFORMANCE OPTIMIZATION: Pre-cache commonly accessed DOM elements |
| 1007 |
# This eliminates 8-10 full DOM traversals per page |
| 1008 |
$this->cache_elements(); |
| 1009 |
|
| 1010 |
# PERFORMANCE OPTIMIZATION: Enable deferred reload to skip intermediate reloads |
| 1011 |
# This reduces 6-10 DOM serialize/deserialize cycles to just 1 final reload |
| 1012 |
$this->deferred_reload = true; |
| 1013 |
|
| 1014 |
# now lets do the magic |
| 1015 |
|
| 1016 |
# COMPATIBILITY FIX: Transform 'value' to 'recommended_value' if needed |
| 1017 |
# Some API versions return 'value' instead of 'recommended_value' |
| 1018 |
if (!empty($replacement_data['header_replacements'])) { |
| 1019 |
foreach ($replacement_data['header_replacements'] as &$item) { |
| 1020 |
if (isset($item['value']) && !isset($item['recommended_value'])) { |
| 1021 |
$item['recommended_value'] = $item['value']; |
| 1022 |
} |
| 1023 |
} |
| 1024 |
unset($item); // Break reference |
| 1025 |
} |
| 1026 |
|
| 1027 |
# start the header html insertion |
| 1028 |
$this->insert_header_html($replacement_data); |
| 1029 |
|
| 1030 |
# now we do the header replacements |
| 1031 |
$this->do_header_replacements($replacement_data); |
| 1032 |
|
| 1033 |
# now do the body replacements |
| 1034 |
$this->do_body_replacements($replacement_data); |
| 1035 |
|
| 1036 |
# now do the footer insertions |
| 1037 |
$this->do_footer_html_insertion($replacement_data); |
| 1038 |
|
| 1039 |
# final cleanup: ensure metasync_optimized attribute is removed from AMP pages |
| 1040 |
$this->cleanup_amp_metasync_attribute(); |
| 1041 |
|
| 1042 |
# CRITICAL FIX: SimpleHtmlDom save() doesn't persist outertext/innertext changes |
| 1043 |
# Use the same manual string replacement approach as process_html_directly |
| 1044 |
$this->deferred_reload = false; |
| 1045 |
|
| 1046 |
# Get the HTML as string |
| 1047 |
$result_html = $this->dom->save(); |
| 1048 |
|
| 1049 |
# Apply manual replacements (same logic as process_html_directly) |
| 1050 |
|
| 1051 |
# Apply header replacements manually |
| 1052 |
if (!empty($replacement_data['header_replacements'])) { |
| 1053 |
foreach ($replacement_data['header_replacements'] as $item) { |
| 1054 |
$type = $item['type'] ?? ''; |
| 1055 |
$value = $item['recommended_value'] ?? $item['value'] ?? ''; |
| 1056 |
|
| 1057 |
if (empty($value)) continue; |
| 1058 |
|
| 1059 |
if ($type === 'title') { |
| 1060 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 1061 |
$title_tag = '<title>' . $new_value . '</title>'; |
| 1062 |
$result_html = preg_replace_callback('/<title[^>]*>.*?<\/title>/is', function ($m) use ($title_tag) { |
| 1063 |
return $title_tag; |
| 1064 |
}, $result_html, 1); |
| 1065 |
} elseif ($type === 'meta') { |
| 1066 |
$name = $item['name'] ?? ''; |
| 1067 |
$property = $item['property'] ?? ''; |
| 1068 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 1069 |
|
| 1070 |
if (!empty($name)) { |
| 1071 |
$pattern = '/<meta\s+(?:name\s*=\s*["\']' . preg_quote($name, '/') . '["\']\s+content\s*=\s*["\'][^"\']*["\']|content\s*=\s*["\'][^"\']*["\']\s+name\s*=\s*["\']' . preg_quote($name, '/') . '["\'])\s*\/?>/i'; |
| 1072 |
$replacement = '<meta name="' . htmlspecialchars($name, ENT_QUOTES, 'UTF-8') . '" content="' . $new_value . '">'; |
| 1073 |
$count = 0; |
| 1074 |
$result_html = preg_replace_callback($pattern, function ($m) use ($replacement) { |
| 1075 |
return $replacement; |
| 1076 |
}, $result_html, -1, $count); |
| 1077 |
|
| 1078 |
if ($count === 0) { |
| 1079 |
$result_html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($replacement) { |
| 1080 |
return $m[1] . "\n" . $replacement; |
| 1081 |
}, $result_html, 1); |
| 1082 |
} |
| 1083 |
} elseif (!empty($property)) { |
| 1084 |
$pattern = '/<meta\s+property\s*=\s*["\']' . preg_quote($property, '/') . '["\']\s+content\s*=\s*["\'][^"\']*["\']\s*\/?>/i'; |
| 1085 |
$replacement = '<meta property="' . htmlspecialchars($property, ENT_QUOTES, 'UTF-8') . '" content="' . $new_value . '">'; |
| 1086 |
$count = 0; |
| 1087 |
$result_html = preg_replace_callback($pattern, function ($m) use ($replacement) { |
| 1088 |
return $replacement; |
| 1089 |
}, $result_html, -1, $count); |
| 1090 |
|
| 1091 |
if ($count === 0) { |
| 1092 |
$result_html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($replacement) { |
| 1093 |
return $m[1] . "\n" . $replacement; |
| 1094 |
}, $result_html, 1); |
| 1095 |
} |
| 1096 |
} |
| 1097 |
} elseif ($type === 'h1' || $type === 'heading') { |
| 1098 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 1099 |
$result_html = preg_replace_callback( |
| 1100 |
'/<h1([^>]*)>.*?<\/h1>/is', |
| 1101 |
function ($m) use ($new_value) { |
| 1102 |
return '<h1' . $m[1] . '>' . $new_value . '</h1>'; |
| 1103 |
}, |
| 1104 |
$result_html, |
| 1105 |
1 |
| 1106 |
); |
| 1107 |
} |
| 1108 |
} |
| 1109 |
} |
| 1110 |
|
| 1111 |
# String-based heading fallback for body_substitutions |
| 1112 |
# DOM changes via SimpleHtmlDom don't persist on Divi/page-builder sites |
| 1113 |
if (!empty($replacement_data['body_substitutions']['headings']) && is_array($replacement_data['body_substitutions']['headings'])) { |
| 1114 |
foreach ($replacement_data['body_substitutions']['headings'] as $heading) { |
| 1115 |
if (empty($heading['type']) || empty($heading['current_value']) || empty($heading['recommended_value'])) { |
| 1116 |
continue; |
| 1117 |
} |
| 1118 |
|
| 1119 |
$heading_type = preg_quote($heading['type'], '/'); |
| 1120 |
$current_value = trim(preg_replace('/\s+/', ' ', html_entity_decode($heading['current_value'], ENT_QUOTES, 'UTF-8'))); |
| 1121 |
$recommended_value = htmlspecialchars($heading['recommended_value'], ENT_QUOTES, 'UTF-8'); |
| 1122 |
|
| 1123 |
$result_html = preg_replace_callback( |
| 1124 |
'/(<' . $heading_type . '(?:\s[^>]*)?>)(.*?)(<\/' . $heading_type . '>)/is', |
| 1125 |
function ($m) use ($current_value, $recommended_value) { |
| 1126 |
$inner_text = trim(preg_replace('/\s+/', ' ', html_entity_decode(strip_tags($m[2]), ENT_QUOTES, 'UTF-8'))); |
| 1127 |
if ($inner_text !== $current_value) { |
| 1128 |
return $m[0]; |
| 1129 |
} |
| 1130 |
# Keep the author's inline formatting; skip when it can't |
| 1131 |
# be preserved rather than flattening the heading. |
| 1132 |
$new_inner = self::build_heading_inner_html($m[2], $recommended_value); |
| 1133 |
if ($new_inner === null) { |
| 1134 |
return $m[0]; |
| 1135 |
} |
| 1136 |
return $m[1] . $new_inner . $m[3]; |
| 1137 |
}, |
| 1138 |
$result_html, |
| 1139 |
-1 |
| 1140 |
); |
| 1141 |
} |
| 1142 |
} |
| 1143 |
|
| 1144 |
# CRITICAL FIX: Apply image alt text manually via string replacement |
| 1145 |
# DOM changes via SimpleHtmlDom don't persist on Oxygen/page-builder sites using HTTP render path |
| 1146 |
$result_html = $this->otto_guard_html($this->apply_image_alt_text_via_string($result_html, $replacement_data), $result_html, 'apply_image_alt_text_via_string'); |
| 1147 |
|
| 1148 |
# Apply insertions — only if DOM insertion didn't already apply it |
| 1149 |
if (!empty($replacement_data['header_html_insertion'])) { |
| 1150 |
$header_html_check = trim($replacement_data['header_html_insertion']); |
| 1151 |
if (strpos($result_html, $header_html_check) === false) { |
| 1152 |
$header_html_insertion = preg_replace( |
| 1153 |
'/<script(\s[^>]*)type\s*=\s*(["\'])application\/ld\+json\2/i', |
| 1154 |
'<script$1type=$2application/ld+json$2 data-otto="true"', |
| 1155 |
$replacement_data['header_html_insertion'] |
| 1156 |
); |
| 1157 |
$safe_header = str_replace(array('\\', '$'), array('\\\\', '\\$'), $header_html_insertion); |
| 1158 |
$result_html = preg_replace('/(<\/head>)/i', $safe_header . "\n" . '$1', $result_html, 1); |
| 1159 |
} |
| 1160 |
} |
| 1161 |
if (!empty($replacement_data['body_top_html_insertion'])) { |
| 1162 |
$body_top_check = trim($replacement_data['body_top_html_insertion']); |
| 1163 |
if (strpos($result_html, $body_top_check) === false) { |
| 1164 |
$safe_body_top = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['body_top_html_insertion']); |
| 1165 |
$result_html = preg_replace('/(<body[^>]*>)/i', '$1' . "\n" . $safe_body_top, $result_html, 1); |
| 1166 |
} |
| 1167 |
} |
| 1168 |
if (!empty($replacement_data['body_bottom_html_insertion'])) { |
| 1169 |
$body_bottom_check = trim($replacement_data['body_bottom_html_insertion']); |
| 1170 |
if (strpos($result_html, $body_bottom_check) === false) { |
| 1171 |
$safe_body_bottom = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['body_bottom_html_insertion']); |
| 1172 |
$result_html = preg_replace('/(<\/body>)/i', $safe_body_bottom . "\n" . '$1', $result_html, 1); |
| 1173 |
} |
| 1174 |
} |
| 1175 |
if (!empty($replacement_data['footer_html_insertion'])) { |
| 1176 |
$safe_footer = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['footer_html_insertion']); |
| 1177 |
$result_html = preg_replace('/(<\/html>)/i', $safe_footer . "\n" . '$1', $result_html, 1); |
| 1178 |
} |
| 1179 |
|
| 1180 |
# Apply OTTO's canonical via string replacement (DOM edit is clobbered by the insert_header_html head-freeze; see apply_canonical_via_string). |
| 1181 |
$result_html = $this->apply_canonical_via_string($result_html, $replacement_data); |
| 1182 |
|
| 1183 |
# DEDUPLICATION: Remove duplicate <title>, meta description, OG, Twitter tags, canonical, and JSON-LD schema |
| 1184 |
$result_html = $this->otto_guard_html($this->deduplicate_title_tags($result_html), $result_html, 'deduplicate_title_tags'); |
| 1185 |
$result_html = $this->otto_guard_html($this->deduplicate_description_tags($result_html), $result_html, 'deduplicate_description_tags'); |
| 1186 |
$result_html = $this->otto_guard_html($this->deduplicate_og_twitter_tags($result_html), $result_html, 'deduplicate_og_twitter_tags'); |
| 1187 |
$result_html = $this->otto_guard_html($this->apply_metabox_og_precedence($result_html), $result_html, 'apply_metabox_og_precedence'); |
| 1188 |
$result_html = $this->otto_guard_html($this->apply_custom_seo_precedence($result_html), $result_html, 'apply_custom_seo_precedence'); |
| 1189 |
$result_html = $this->otto_guard_html($this->deduplicate_schema_tags($result_html), $result_html, 'deduplicate_schema_tags'); |
| 1190 |
$result_html = $this->otto_guard_html($this->deduplicate_canonical_tags($result_html), $result_html, 'deduplicate_canonical_tags'); |
| 1191 |
|
| 1192 |
# Ensure metasync_optimized attribute on <head> (post-serialization so dom->clear() can't wipe it) |
| 1193 |
if (!$this->is_amp_page() && strpos($result_html, 'metasync_optimized') === false) { |
| 1194 |
$result_html = preg_replace('/<head(\s|>)/i', '<head metasync_optimized$1', $result_html, 1); |
| 1195 |
} |
| 1196 |
|
| 1197 |
$result_html = $this->otto_guard_html($this->restore_case_sensitive_attributes($result_html), $result_html, 'restore_case_sensitive_attributes'); |
| 1198 |
|
| 1199 |
# / Re-inject any <style> blocks lost during processing. |
| 1200 |
$result_html = $this->otto_guard_html($this->restore_lost_style_blocks($result_html, $original_style_blocks), $result_html, 'restore_lost_style_blocks'); |
| 1201 |
|
| 1202 |
# Fix Divi 5 shortcode framework class renumbering (HTTP path) |
| 1203 |
$result_html = $this->otto_guard_html($this->fix_divi_class_renumbering($result_html), $result_html, 'fix_divi_class_renumbering'); |
| 1204 |
|
| 1205 |
# Clean OTTO internal fetch params from HTML output. |
| 1206 |
# The HTTP render uses wp_remote_get with ?is_otto_page_fetch=1&otto_block_title=1&otto_block_desc=1 |
| 1207 |
# These leak into form action URLs, canonical links, etc. in the rendered HTML. |
| 1208 |
$result_html = $this->otto_guard_html($this->clean_otto_fetch_params($result_html), $result_html, 'clean_otto_fetch_params'); |
| 1209 |
|
| 1210 |
# Remove any lazy-video facade <style> hoisted into the page |
| 1211 |
# (runs while srcdoc is still tokenized, so the iframe's own copy is safe). |
| 1212 |
$result_html = $this->otto_guard_html($this->strip_hoisted_facade_styles($result_html), $result_html, 'strip_hoisted_facade_styles'); |
| 1213 |
|
| 1214 |
# Restore the original encoded srcdoc values (undo tokenization). |
| 1215 |
$result_html = $this->otto_guard_html($this->restore_srcdoc_attributes($result_html, $srcdoc_store), $result_html, 'restore_srcdoc_attributes'); |
| 1216 |
|
| 1217 |
# Keep <meta charset> within the first 1024 bytes so external CSS |
| 1218 |
# (e.g. checkmark content:"✓") doesn't mojibake on cached responses. |
| 1219 |
$result_html = $this->otto_guard_html($this->ensure_early_charset_meta($result_html), $result_html, 'ensure_early_charset_meta'); |
| 1220 |
|
| 1221 |
# undo double-encoded numeric/hex character references produced by the |
| 1222 |
# bundled simplehtmldom serializer from attributes like data-x-icon-s="". |
| 1223 |
$result_html = $this->otto_guard_html($this->repair_double_encoded_entities($result_html), $result_html, 'repair_double_encoded_entities'); |
| 1224 |
|
| 1225 |
# Remove any lazy-video facade <style> hoisted into the page |
| 1226 |
# (runs while srcdoc is still tokenized, so the iframe's own copy is safe). |
| 1227 |
$result_html = $this->strip_hoisted_facade_styles($result_html); |
| 1228 |
|
| 1229 |
# Restore the original encoded srcdoc values (undo tokenization). |
| 1230 |
$result_html = $this->restore_srcdoc_attributes($result_html, $srcdoc_store); |
| 1231 |
|
| 1232 |
# Keep <meta charset> within the first 1024 bytes so external CSS |
| 1233 |
# (e.g. checkmark content:"✓") doesn't mojibake on cached responses. |
| 1234 |
$result_html = $this->ensure_early_charset_meta($result_html); |
| 1235 |
|
| 1236 |
# undo double-encoded numeric/hex character references produced by the |
| 1237 |
# bundled simplehtmldom serializer from attributes like data-x-icon-s="". |
| 1238 |
$result_html = $this->repair_double_encoded_entities($result_html); |
| 1239 |
|
| 1240 |
return $result_html; |
| 1241 |
} |
| 1242 |
|
| 1243 |
/** |
| 1244 |
* Repair double-encoded character references in serialized HTML. |
| 1245 |
* |
| 1246 |
* The bundled simplehtmldom serializer (HtmlNode::makeup(), reached via |
| 1247 |
* $this->dom->save() and $root->outertext) runs every attribute value through |
| 1248 |
* htmlentities(), which escapes the leading '&' of numeric/hex character |
| 1249 |
* references. Theme icon attributes such as data-x-icon-s="" |
| 1250 |
* (Themeco X/Cornerstone) therefore become "&#xf3c5", so the browser |
| 1251 |
* prints the literal text instead of drawing the icon glyph. |
| 1252 |
* |
| 1253 |
* Restores only NUMERIC and HEX character references (&#NNN; / &#xHHHH;). |
| 1254 |
* Bare ampersands in query strings (?a=1&b=2 -> ?a=1&b=2) and named |
| 1255 |
* entities are intentionally left escaped. |
| 1256 |
* |
| 1257 |
* @param string $html Serialized HTML from save()/outertext. |
| 1258 |
* @return string |
| 1259 |
*/ |
| 1260 |
private function repair_double_encoded_entities($html) { |
| 1261 |
# strpos guard: skip the regex entirely when there is nothing to repair |
| 1262 |
if (!is_string($html) || strpos($html, '&#') === false) { |
| 1263 |
return $html; |
| 1264 |
} |
| 1265 |
return preg_replace('/&(#x[0-9a-fA-F]+;?|#[0-9]+;?)/', '&$1', $html); |
| 1266 |
} |
| 1267 |
|
| 1268 |
# do the footer html insertion |
| 1269 |
function do_footer_html_insertion($replacement_data){ |
| 1270 |
|
| 1271 |
# check that we have footer html |
| 1272 |
if(empty($replacement_data['footer_html_insertion'])){ |
| 1273 |
return; |
| 1274 |
} |
| 1275 |
|
| 1276 |
# OPTIMIZED: Use cached element instead of DOM traversal |
| 1277 |
$footer = $this->get_cached_element('footer'); |
| 1278 |
|
| 1279 |
# check that footer is object |
| 1280 |
if(!is_object($footer) || !isset($footer->innertext, $footer->outertext)){ |
| 1281 |
return; |
| 1282 |
} |
| 1283 |
|
| 1284 |
# get the tag attributes |
| 1285 |
$attributes_string = $this->get_tag_attributes($footer); |
| 1286 |
|
| 1287 |
# now do the actual html replacements |
| 1288 |
$footer->outertext = '<footer' . $attributes_string . '>' . $footer->innertext . $replacement_data['footer_html_insertion'].'</footer>'; |
| 1289 |
|
| 1290 |
# save the document |
| 1291 |
$this->save_reload(); |
| 1292 |
} |
| 1293 |
|
| 1294 |
# do body replacements |
| 1295 |
function do_body_replacements($replacement_data){ |
| 1296 |
|
| 1297 |
# start body top html replacements |
| 1298 |
$this->do_body_top_html($replacement_data); |
| 1299 |
|
| 1300 |
# start the body bottom html replacements |
| 1301 |
$this->do_body_bottom_html($replacement_data); |
| 1302 |
|
| 1303 |
# do the body substitutions |
| 1304 |
$this->do_body_substitutions($replacement_data); |
| 1305 |
|
| 1306 |
# Check if the feature is enabled in general settings (Post/Page Editor Settings) |
| 1307 |
$general_settings = get_option('metasync_options')['general'] ?? []; |
| 1308 |
if (!empty($general_settings['open_external_links']) && $general_settings['open_external_links'] == '1') { |
| 1309 |
$this->add_target_blank_to_external_links(); |
| 1310 |
} |
| 1311 |
|
| 1312 |
# Check if the feature is enabled in seo_controls settings (Indexation Control) |
| 1313 |
$seo_controls = get_option('metasync_options')['seo_controls'] ?? []; |
| 1314 |
if (!empty($seo_controls['add_nofollow_to_external_links']) && $seo_controls['add_nofollow_to_external_links'] === 'true') { |
| 1315 |
$this->add_nofollow_to_external_links(); |
| 1316 |
} |
| 1317 |
} |
| 1318 |
|
| 1319 |
# body substitutions data |
| 1320 |
function do_body_substitutions($replacement_data){ |
| 1321 |
|
| 1322 |
# check that we have an array of substitutions |
| 1323 |
if(empty($replacement_data['body_substitutions']) || !is_array($replacement_data['body_substitutions'])){ |
| 1324 |
return; |
| 1325 |
} |
| 1326 |
|
| 1327 |
# now work on different substitution keys |
| 1328 |
foreach ($replacement_data['body_substitutions'] as $key => $value) { |
| 1329 |
|
| 1330 |
# check key categories |
| 1331 |
if($key == 'images'){ |
| 1332 |
# do image replacements |
| 1333 |
$this->handle_images($value); |
| 1334 |
} |
| 1335 |
elseif($key == 'headings'){ |
| 1336 |
# do heading repalcements |
| 1337 |
$this->do_heading_body_substitutions($value); |
| 1338 |
} |
| 1339 |
elseif($key == 'links'){ |
| 1340 |
# do link replacements |
| 1341 |
$this->do_link_body_substitutions($value); |
| 1342 |
} |
| 1343 |
|
| 1344 |
} |
| 1345 |
|
| 1346 |
# save the document |
| 1347 |
$this->save_reload(); |
| 1348 |
|
| 1349 |
} |
| 1350 |
|
| 1351 |
/** |
| 1352 |
* START BODY SUBSTITUTION FUNCTIONS |
| 1353 |
* @see do_body_substitutions(); |
| 1354 |
*/ |
| 1355 |
|
| 1356 |
# image substitions |
| 1357 |
function handle_images($image_data){ |
| 1358 |
|
| 1359 |
if (empty($image_data) || !is_array($image_data)) { |
| 1360 |
return; |
| 1361 |
} |
| 1362 |
|
| 1363 |
# OPTIMIZED: Use cached images instead of DOM traversal |
| 1364 |
$images = $this->get_cached_element('imgs', []); |
| 1365 |
|
| 1366 |
if (empty($images)) { |
| 1367 |
return; |
| 1368 |
} |
| 1369 |
|
| 1370 |
# index suggestions by URL path as well, so images authored with |
| 1371 |
# relative srcs still match OTTO's absolute-URL keys (and vice versa). |
| 1372 |
$alt_lookup = $this->build_image_alt_lookup($image_data); |
| 1373 |
|
| 1374 |
# PERFORMANCE OPTIMIZATION: O(n²) reduced to O(n) |
| 1375 |
# Single pass with hash map lookup instead of nested loop |
| 1376 |
foreach($images AS $key => $image){ |
| 1377 |
# Get image src |
| 1378 |
$image_src = $image->src; |
| 1379 |
|
| 1380 |
if (empty($image_src)) { |
| 1381 |
continue; |
| 1382 |
} |
| 1383 |
|
| 1384 |
# Hash map lookup O(1) instead of loop O(n) |
| 1385 |
$alt_text = $this->lookup_image_alt($alt_lookup, $image_src); |
| 1386 |
if ($alt_text !== null) { |
| 1387 |
# Set alt text - Note: This may not persist in all cases |
| 1388 |
# Manual string replacement in process_html_directly() ensures it's applied |
| 1389 |
$new_alt = htmlspecialchars($alt_text, ENT_QUOTES, 'UTF-8'); |
| 1390 |
|
| 1391 |
# Get current img tag HTML and update alt attribute |
| 1392 |
$current_html = $image->outertext; |
| 1393 |
|
| 1394 |
# Remove existing alt attribute (if any) |
| 1395 |
$updated_html = preg_replace('/\s+alt=(["\'])[^"\']*\1/', '', $current_html); |
| 1396 |
|
| 1397 |
# Insert new alt attribute after the opening <img |
| 1398 |
$updated_html = preg_replace('/^<img\s/', '<img alt="' . str_replace('$', '\\$', $new_alt) . '" ', $updated_html); |
| 1399 |
|
| 1400 |
# Update the element |
| 1401 |
$image->outertext = $updated_html; |
| 1402 |
|
| 1403 |
$multi_view_attr = $image->getAttribute('data-et-multi-view'); |
| 1404 |
if (!empty($multi_view_attr)) { |
| 1405 |
$this->update_divi_multi_view_alt($image, $alt_text); |
| 1406 |
} |
| 1407 |
} |
| 1408 |
} |
| 1409 |
} |
| 1410 |
|
| 1411 |
# memoized site host — home_url() runs its filter chain on every |
| 1412 |
# call and these helpers run per suggestion and per img. |
| 1413 |
private ?string $site_host_cache = null; |
| 1414 |
|
| 1415 |
/** |
| 1416 |
* Get this site's host (lowercase) for same-host URL comparisons. |
| 1417 |
* |
| 1418 |
* @return string |
| 1419 |
*/ |
| 1420 |
private function get_site_host() { |
| 1421 |
if ($this->site_host_cache === null) { |
| 1422 |
$host = wp_parse_url(home_url(), PHP_URL_HOST); |
| 1423 |
$this->site_host_cache = is_string($host) ? strtolower($host) : ''; |
| 1424 |
} |
| 1425 |
return $this->site_host_cache; |
| 1426 |
} |
| 1427 |
|
| 1428 |
/** |
| 1429 |
* Candidate lookup keys for an image URL — the URL itself plus, |
| 1430 |
* for relative URLs or absolute/protocol-relative URLs on this site's |
| 1431 |
* host, the bare URL path. URLs on a different host only ever match |
| 1432 |
* exactly, so a same path on a foreign host cannot produce a false match. |
| 1433 |
* |
| 1434 |
* @param string $url Image URL (absolute, protocol-relative, or relative) |
| 1435 |
* @return array Candidate keys, original URL first |
| 1436 |
*/ |
| 1437 |
private function get_image_url_variants($url) { |
| 1438 |
$variants = [$url]; |
| 1439 |
|
| 1440 |
if (!is_string($url) || $url === '') { |
| 1441 |
return $variants; |
| 1442 |
} |
| 1443 |
|
| 1444 |
$host = wp_parse_url($url, PHP_URL_HOST); |
| 1445 |
if (!empty($host) && strtolower($host) !== $this->get_site_host()) { |
| 1446 |
# Foreign host: exact match only |
| 1447 |
return $variants; |
| 1448 |
} |
| 1449 |
|
| 1450 |
$path = wp_parse_url($url, PHP_URL_PATH); |
| 1451 |
if (is_string($path) && $path !== '' && $path !== $url) { |
| 1452 |
$variants[] = $path; |
| 1453 |
} |
| 1454 |
|
| 1455 |
return $variants; |
| 1456 |
} |
| 1457 |
|
| 1458 |
/** |
| 1459 |
* Build the image alt lookup keyed by every URL variant of each |
| 1460 |
* OTTO suggestion, so relative img srcs match absolute suggestion URLs. |
| 1461 |
* |
| 1462 |
* @param array $image_data OTTO body_substitutions.images map (url => alt) |
| 1463 |
* @return array Expanded lookup (url-or-path => alt) |
| 1464 |
*/ |
| 1465 |
private function build_image_alt_lookup($image_data) { |
| 1466 |
$lookup = []; |
| 1467 |
|
| 1468 |
foreach ($image_data as $image_url => $alt_text) { |
| 1469 |
foreach ($this->get_image_url_variants($image_url) as $variant) { |
| 1470 |
# First suggestion wins when two keys collapse to the same |
| 1471 |
# path (e.g. a relative and an absolute key for one image); |
| 1472 |
# exact-src matches still take priority in lookup_image_alt(). |
| 1473 |
if (!isset($lookup[$variant])) { |
| 1474 |
$lookup[$variant] = $alt_text; |
| 1475 |
} |
| 1476 |
} |
| 1477 |
} |
| 1478 |
|
| 1479 |
return $lookup; |
| 1480 |
} |
| 1481 |
|
| 1482 |
/** |
| 1483 |
* Resolve the OTTO alt text for an img src, trying the exact src |
| 1484 |
* first and then its same-host path variant. |
| 1485 |
* |
| 1486 |
* @param array $lookup Expanded lookup from build_image_alt_lookup() |
| 1487 |
* @param string $image_src The img element's src attribute |
| 1488 |
* @return string|null Alt text, or null when no suggestion matches |
| 1489 |
*/ |
| 1490 |
private function lookup_image_alt($lookup, $image_src) { |
| 1491 |
foreach ($this->get_image_url_variants($image_src) as $variant) { |
| 1492 |
if (isset($lookup[$variant])) { |
| 1493 |
return $lookup[$variant]; |
| 1494 |
} |
| 1495 |
} |
| 1496 |
|
| 1497 |
return null; |
| 1498 |
} |
| 1499 |
|
| 1500 |
/** |
| 1501 |
* Update Divi's multi-view data attribute with alt text |
| 1502 |
* Divi stores image attributes in a JSON structure within data-et-multi-view |
| 1503 |
* |
| 1504 |
* @param object $image The image DOM element |
| 1505 |
* @param string $alt_text The alt text to set |
| 1506 |
*/ |
| 1507 |
private function update_divi_multi_view_alt($image, $alt_text) { |
| 1508 |
$multi_view_attr = $image->getAttribute('data-et-multi-view'); |
| 1509 |
|
| 1510 |
if (!empty($multi_view_attr)) { |
| 1511 |
try { |
| 1512 |
# Decode the JSON |
| 1513 |
$multi_view_data = json_decode($multi_view_attr, true); |
| 1514 |
|
| 1515 |
if ($multi_view_data && isset($multi_view_data['schema']['attrs'])) { |
| 1516 |
# Update alt in desktop view |
| 1517 |
if (isset($multi_view_data['schema']['attrs']['desktop'])) { |
| 1518 |
$multi_view_data['schema']['attrs']['desktop']['alt'] = $alt_text; |
| 1519 |
} |
| 1520 |
|
| 1521 |
# Update alt in other views if they exist (phone, tablet, etc.) |
| 1522 |
foreach ($multi_view_data['schema']['attrs'] as $view => $attrs) { |
| 1523 |
if (isset($attrs['alt'])) { |
| 1524 |
$multi_view_data['schema']['attrs'][$view]['alt'] = $alt_text; |
| 1525 |
} |
| 1526 |
} |
| 1527 |
|
| 1528 |
# Encode back to JSON and update the attribute |
| 1529 |
$updated_json = json_encode($multi_view_data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE); |
| 1530 |
$image->setAttribute('data-et-multi-view', $updated_json); |
| 1531 |
|
| 1532 |
# error_log('MetaSync OTTO DEBUG: Updated Divi multi-view alt text'); |
| 1533 |
} |
| 1534 |
} catch (Exception $e) { |
| 1535 |
# If JSON decode fails, just log and continue |
| 1536 |
# error_log('MetaSync OTTO DEBUG: Failed to update Divi multi-view - ' . $e->getMessage()); |
| 1537 |
} |
| 1538 |
} |
| 1539 |
} |
| 1540 |
|
| 1541 |
/** |
| 1542 |
* Apply image alt text via string replacement on raw HTML. |
| 1543 |
* Used as a fallback when DOM changes don't persist (Oxygen, Divi, page-builder sites). |
| 1544 |
* |
| 1545 |
* Known limitation (pre-existing): the `<img[^>]*src=` pattern can match a |
| 1546 |
* `data-src`/`data-lazy-src` attribute that ends with the target URL before |
| 1547 |
* the real `src`, mis-attributing the alt on lazy-loaded markup. Tracked as |
| 1548 |
* a follow-up to. |
| 1549 |
* |
| 1550 |
* @param string $html The HTML to process |
| 1551 |
* @param array $replacement_data The replacement data containing body_substitutions.images |
| 1552 |
* @return string Modified HTML with alt text applied |
| 1553 |
*/ |
| 1554 |
private function apply_image_alt_text_via_string($html, $replacement_data) { |
| 1555 |
if (empty($replacement_data['body_substitutions']['images']) || !is_array($replacement_data['body_substitutions']['images'])) { |
| 1556 |
return $html; |
| 1557 |
} |
| 1558 |
|
| 1559 |
foreach ($replacement_data['body_substitutions']['images'] as $image_url => $alt_text) { |
| 1560 |
# prefilter and match on the same-host path variant too, so |
| 1561 |
# relative img srcs match OTTO's absolute suggestion URLs. |
| 1562 |
$src_pattern = $this->get_image_src_pattern($image_url); |
| 1563 |
|
| 1564 |
if (empty($alt_text) || $src_pattern === null || strpos($html, $src_pattern['needle']) === false) { |
| 1565 |
continue; |
| 1566 |
} |
| 1567 |
|
| 1568 |
$escaped_alt = htmlspecialchars($alt_text, ENT_QUOTES, 'UTF-8'); |
| 1569 |
$img_pattern = '/<img[^>]*src=["\']' . $src_pattern['pattern'] . '["\'][^>]*>/i'; |
| 1570 |
|
| 1571 |
if (preg_match_all($img_pattern, $html, $img_matches)) { |
| 1572 |
foreach ($img_matches[0] as $original_img) { |
| 1573 |
if (strpos($original_img, $escaped_alt) !== false) { |
| 1574 |
continue; |
| 1575 |
} |
| 1576 |
|
| 1577 |
# Remove ALL existing alt attributes |
| 1578 |
$new_img = preg_replace('/\s+alt\s*=\s*(["\'])[^"\']*\1/i', '', $original_img); |
| 1579 |
$new_img = preg_replace('/<img\s+alt\s*=\s*(["\'])[^"\']*\1\s*/i', '<img ', $new_img); |
| 1580 |
|
| 1581 |
# Add single alt attribute after <img |
| 1582 |
$new_img = preg_replace('/^<img\s*/i', '<img alt="' . str_replace('$', '\\$', $escaped_alt) . '" ', $new_img); |
| 1583 |
|
| 1584 |
# Update data-et-multi-view JSON if present (Divi theme) |
| 1585 |
if (strpos($new_img, 'data-et-multi-view') !== false) { |
| 1586 |
$new_img = preg_replace_callback( |
| 1587 |
'/data-et-multi-view="([^"]+)"/i', |
| 1588 |
function($mv_matches) use ($alt_text) { |
| 1589 |
$json_str = html_entity_decode($mv_matches[1], ENT_QUOTES, 'UTF-8'); |
| 1590 |
$json_data = json_decode($json_str, true); |
| 1591 |
|
| 1592 |
if ($json_data && isset($json_data['schema']['attrs'])) { |
| 1593 |
foreach ($json_data['schema']['attrs'] as &$attrs) { |
| 1594 |
if (array_key_exists('alt', $attrs)) { |
| 1595 |
$attrs['alt'] = $alt_text; |
| 1596 |
} |
| 1597 |
} |
| 1598 |
unset($attrs); |
| 1599 |
$new_json = json_encode($json_data, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE); |
| 1600 |
return 'data-et-multi-view="' . str_replace('"', '"', $new_json) . '"'; |
| 1601 |
} |
| 1602 |
return $mv_matches[0]; |
| 1603 |
}, |
| 1604 |
$new_img |
| 1605 |
); |
| 1606 |
} |
| 1607 |
|
| 1608 |
$html = str_replace($original_img, $new_img, $html); |
| 1609 |
} |
| 1610 |
} |
| 1611 |
} |
| 1612 |
|
| 1613 |
return $html; |
| 1614 |
} |
| 1615 |
|
| 1616 |
/** |
| 1617 |
* Build the src match for an OTTO image suggestion URL. |
| 1618 |
* |
| 1619 |
* For relative URLs or absolute/protocol-relative URLs on this site's |
| 1620 |
* host, the regex matches the URL path with an optional scheme+host |
| 1621 |
* prefix — so a relative img src matches an absolute suggestion URL and |
| 1622 |
* vice versa. URLs on a foreign host keep the original exact match, so a |
| 1623 |
* same path on a different host cannot produce a false positive. |
| 1624 |
* |
| 1625 |
* @param string|int $image_url OTTO suggestion URL (may arrive as an int |
| 1626 |
* when it originates from an array key that PHP coerced) |
| 1627 |
* @return array|null ['needle' => strpos prefilter string, |
| 1628 |
* 'pattern' => regex fragment ('/' delimiter)] or null |
| 1629 |
*/ |
| 1630 |
private function get_image_src_pattern($image_url) { |
| 1631 |
if (!is_string($image_url) || $image_url === '') { |
| 1632 |
return null; |
| 1633 |
} |
| 1634 |
|
| 1635 |
$host = wp_parse_url($image_url, PHP_URL_HOST); |
| 1636 |
$site_host = $this->get_site_host(); |
| 1637 |
|
| 1638 |
# Foreign host: exact match only |
| 1639 |
if (!empty($host) && strtolower($host) !== $site_host) { |
| 1640 |
return ['needle' => $image_url, 'pattern' => preg_quote($image_url, '/')]; |
| 1641 |
} |
| 1642 |
|
| 1643 |
$path = wp_parse_url($image_url, PHP_URL_PATH); |
| 1644 |
if (!is_string($path) || $path === '') { |
| 1645 |
return ['needle' => $image_url, 'pattern' => preg_quote($image_url, '/')]; |
| 1646 |
} |
| 1647 |
|
| 1648 |
$prefix = $site_host !== '' ? '(?:(?:https?:)?\/\/' . preg_quote($site_host, '/') . ')?' : ''; |
| 1649 |
|
| 1650 |
# Allow an optional query/fragment after the path (e.g. WordPress |
| 1651 |
# `?ver=` cache-busters) so src="/up/a.jpg?ver=2" still matches. |
| 1652 |
return ['needle' => $path, 'pattern' => $prefix . preg_quote($path, '/') . '(?:[?#][^"\']*)?']; |
| 1653 |
} |
| 1654 |
|
| 1655 |
# heading substitutions |
| 1656 |
function do_heading_body_substitutions($heading_data){ |
| 1657 |
|
| 1658 |
# loop all data |
| 1659 |
foreach($heading_data AS $okey => $heading){ |
| 1660 |
|
| 1661 |
# find all occurences of the heading type |
| 1662 |
$occurences = $this->dom->find($heading['type']); |
| 1663 |
|
| 1664 |
# loop all occurences |
| 1665 |
foreach ($occurences as $ikey => $heading_old) { |
| 1666 |
|
| 1667 |
# get the header text |
| 1668 |
$text = $heading_old->text(); |
| 1669 |
|
| 1670 |
# check matching text |
| 1671 |
if(trim($heading['current_value'] ?? '') == trim($text ?? '')){ |
| 1672 |
|
| 1673 |
# Rebuild the inner HTML, keeping the author's inline |
| 1674 |
# formatting. Returns null when the markup can't be preserved |
| 1675 |
# unambiguously — leave that heading alone rather than strip |
| 1676 |
# its <span style>/<em>/<br> and restyle the page. |
| 1677 |
$new_inner = self::build_heading_inner_html( |
| 1678 |
$heading_old->innertext, |
| 1679 |
$heading['recommended_value'] |
| 1680 |
); |
| 1681 |
|
| 1682 |
if ($new_inner === null) { |
| 1683 |
continue; |
| 1684 |
} |
| 1685 |
|
| 1686 |
# replace the tag output, preserving the opening tag's attributes |
| 1687 |
$outer = $heading_old->outertext; |
| 1688 |
$open_end = strpos($outer, '>'); |
| 1689 |
if ($open_end !== false) { |
| 1690 |
$open_tag = substr($outer, 0, $open_end + 1); |
| 1691 |
$close_tag = '</' . $heading['type'] . '>'; |
| 1692 |
$heading_old->outertext = $open_tag . $new_inner . $close_tag; |
| 1693 |
} |
| 1694 |
|
| 1695 |
} |
| 1696 |
} |
| 1697 |
|
| 1698 |
} |
| 1699 |
|
| 1700 |
# save and reload the dom |
| 1701 |
$this->save_reload(); |
| 1702 |
} |
| 1703 |
|
| 1704 |
# link replacements |
| 1705 |
function do_link_body_substitutions($swap_data){ |
| 1706 |
|
| 1707 |
# find all links in the body |
| 1708 |
$links = $this->dom->find('a'); |
| 1709 |
|
| 1710 |
# loop all links check if we have them in swap data |
| 1711 |
foreach ($links as $key => $link) { |
| 1712 |
|
| 1713 |
# check if link matches |
| 1714 |
if(!empty($swap_data[$link->href])){ |
| 1715 |
|
| 1716 |
# replace the link |
| 1717 |
$link->href = $swap_data[$link->href]; |
| 1718 |
} |
| 1719 |
|
| 1720 |
} |
| 1721 |
} |
| 1722 |
|
| 1723 |
/** |
| 1724 |
* Add rel="nofollow" attribute to external links |
| 1725 |
* Works for both buffer and HTTP rendering methods |
| 1726 |
* Only processes links that don't already have rel="nofollow" |
| 1727 |
*/ |
| 1728 |
function add_nofollow_to_external_links(){ |
| 1729 |
|
| 1730 |
# find all links in the body |
| 1731 |
$links = $this->dom->find('a'); |
| 1732 |
|
| 1733 |
# get home URL for comparison |
| 1734 |
$home_url = rtrim(home_url(), '/'); |
| 1735 |
$home_url_lower = strtolower($home_url); |
| 1736 |
|
| 1737 |
# loop through all links |
| 1738 |
foreach ($links as $key => $link) { |
| 1739 |
|
| 1740 |
# get the href attribute |
| 1741 |
$href = $link->href ?? ''; |
| 1742 |
|
| 1743 |
# skip if href is empty |
| 1744 |
if (empty($href)) { |
| 1745 |
continue; |
| 1746 |
} |
| 1747 |
|
| 1748 |
# check if link is external |
| 1749 |
if ($this->is_external_link($href, $home_url, $home_url_lower)) { |
| 1750 |
# get existing rel attribute |
| 1751 |
$existing_rel = $link->rel ?? ''; |
| 1752 |
|
| 1753 |
# check if nofollow already exists |
| 1754 |
if (empty($existing_rel)) { |
| 1755 |
# no rel attribute, add nofollow |
| 1756 |
$link->rel = 'nofollow'; |
| 1757 |
} elseif (strpos($existing_rel, 'nofollow') === false) { |
| 1758 |
# rel exists but nofollow is not present, add it |
| 1759 |
$link->rel = trim($existing_rel . ' nofollow'); |
| 1760 |
} |
| 1761 |
# if nofollow already exists, do nothing |
| 1762 |
} |
| 1763 |
} |
| 1764 |
|
| 1765 |
# Note: No need to call save_reload() here - it's called at the end of process_html_directly() |
| 1766 |
# This avoids redundant DOM save/reload operations and improves performance |
| 1767 |
} |
| 1768 |
|
| 1769 |
/** |
| 1770 |
* Check if a URL is external |
| 1771 |
* |
| 1772 |
* @param string $url The URL to check |
| 1773 |
* @param string $home_url The home URL without trailing slash |
| 1774 |
* @param string $home_url_lower Lowercase version of home URL |
| 1775 |
* @return bool True if external, false if internal |
| 1776 |
*/ |
| 1777 |
private function is_external_link($url, $home_url, $home_url_lower) { |
| 1778 |
# Empty or anchor-only links are internal |
| 1779 |
if (empty($url) || $url === '#' || strpos($url, '#') === 0) { |
| 1780 |
return false; |
| 1781 |
} |
| 1782 |
|
| 1783 |
# Relative URLs (starting with /) are internal |
| 1784 |
if (strpos($url, '/') === 0 && strpos($url, '//') !== 0) { |
| 1785 |
return false; |
| 1786 |
} |
| 1787 |
|
| 1788 |
# Check if URL starts with home URL (case-insensitive) |
| 1789 |
$url_lower = strtolower($url); |
| 1790 |
if (strpos($url_lower, $home_url_lower) === 0) { |
| 1791 |
return false; |
| 1792 |
} |
| 1793 |
|
| 1794 |
# If it's a protocol-relative URL (//example.com), check if it matches our domain |
| 1795 |
if (strpos($url, '//') === 0) { |
| 1796 |
$parsed_home = parse_url($home_url); |
| 1797 |
$parsed_link = parse_url($url); |
| 1798 |
|
| 1799 |
if (isset($parsed_home['host']) && isset($parsed_link['host'])) { |
| 1800 |
if (strtolower($parsed_home['host']) === strtolower($parsed_link['host'])) { |
| 1801 |
return false; |
| 1802 |
} |
| 1803 |
} |
| 1804 |
} |
| 1805 |
|
| 1806 |
# All other URLs are considered external |
| 1807 |
return true; |
| 1808 |
} |
| 1809 |
|
| 1810 |
/** |
| 1811 |
* Add target="_blank" attribute to external links |
| 1812 |
* Works for both buffer and HTTP rendering methods |
| 1813 |
* Only processes links that don't already have a target attribute |
| 1814 |
*/ |
| 1815 |
function add_target_blank_to_external_links(){ |
| 1816 |
|
| 1817 |
# find all links in the body |
| 1818 |
$links = $this->dom->find('a'); |
| 1819 |
|
| 1820 |
# get home URL for comparison |
| 1821 |
$home_url = rtrim(home_url(), '/'); |
| 1822 |
$home_url_lower = strtolower($home_url); |
| 1823 |
|
| 1824 |
# loop through all links |
| 1825 |
foreach ($links as $key => $link) { |
| 1826 |
|
| 1827 |
# get the href attribute |
| 1828 |
$href = $link->href ?? ''; |
| 1829 |
|
| 1830 |
# skip if href is empty or already has target attribute |
| 1831 |
if (empty($href) || !empty($link->target)) { |
| 1832 |
continue; |
| 1833 |
} |
| 1834 |
|
| 1835 |
# check if link is external |
| 1836 |
if ($this->is_external_link($href, $home_url, $home_url_lower)) { |
| 1837 |
# add target="_blank" attribute |
| 1838 |
$link->target = '_blank'; |
| 1839 |
|
| 1840 |
# add rel="noopener noreferrer" for security |
| 1841 |
$existing_rel = $link->rel ?? ''; |
| 1842 |
if (empty($existing_rel)) { |
| 1843 |
$link->rel = 'noopener noreferrer'; |
| 1844 |
} elseif (strpos($existing_rel, 'noopener') === false) { |
| 1845 |
$link->rel = trim($existing_rel . ' noopener noreferrer'); |
| 1846 |
} |
| 1847 |
} |
| 1848 |
} |
| 1849 |
|
| 1850 |
# Note: No need to call save_reload() here - it's called at the end of process_html_directly() |
| 1851 |
# This avoids redundant DOM save/reload operations and improves performance |
| 1852 |
} |
| 1853 |
|
| 1854 |
# body bottom html replacement code |
| 1855 |
function do_body_bottom_html($insert_data){ |
| 1856 |
|
| 1857 |
# check that data is availbale |
| 1858 |
if(empty($insert_data['body_bottom_html_insertion'])){ |
| 1859 |
return; |
| 1860 |
} |
| 1861 |
|
| 1862 |
# OPTIMIZED: Use cached body element |
| 1863 |
$body = $this->get_cached_element('body'); |
| 1864 |
|
| 1865 |
# set the link property if not empty |
| 1866 |
if(empty($body->outertext)){ |
| 1867 |
return; |
| 1868 |
} |
| 1869 |
|
| 1870 |
|
| 1871 |
# get the tag attributes |
| 1872 |
$attributes_string = $this->get_tag_attributes($body); |
| 1873 |
|
| 1874 |
# now do the actual html replacements |
| 1875 |
$body->outertext = '<body' . $attributes_string . '>' . $body->innertext . $insert_data['body_bottom_html_insertion'].'</body>'; |
| 1876 |
|
| 1877 |
# save the document |
| 1878 |
$this->save_reload(); |
| 1879 |
} |
| 1880 |
|
| 1881 |
# body top html replacement |
| 1882 |
function do_body_top_html($insert_data){ |
| 1883 |
|
| 1884 |
# check that data is availbale |
| 1885 |
if(empty($insert_data['body_top_html_insertion'])){ |
| 1886 |
return; |
| 1887 |
} |
| 1888 |
|
| 1889 |
# OPTIMIZED: Use cached body element |
| 1890 |
$body = $this->get_cached_element('body'); |
| 1891 |
|
| 1892 |
# get the tag attributes |
| 1893 |
$attributes_string = $this->get_tag_attributes($body); |
| 1894 |
|
| 1895 |
# now do the actual html replacements |
| 1896 |
$body->outertext = '<body' . $attributes_string . '>'.$insert_data['body_top_html_insertion'].$body->innertext . '</body>'; |
| 1897 |
} |
| 1898 |
|
| 1899 |
/** |
| 1900 |
* Resolve the current singular post id reliably inside the OTTO output |
| 1901 |
* buffer / direct-render context. |
| 1902 |
* |
| 1903 |
* get_the_ID() depends on the global $post/loop state, which is not reliable |
| 1904 |
* once WordPress has finished rendering and the buffer/shutdown pass runs. |
| 1905 |
* get_queried_object_id() reads the parsed main query and stays correct there |
| 1906 |
* (same approach as apply_metabox_og_precedence()). Restricted to |
| 1907 |
* singular views: on archives/home the queried object id can be a term or |
| 1908 |
* user id, which must not be read as post meta. |
| 1909 |
* |
| 1910 |
* @return int Post id, or 0 when not applicable. |
| 1911 |
*/ |
| 1912 |
private function get_buffer_post_id() { |
| 1913 |
if (!function_exists('is_singular') || !is_singular()) { |
| 1914 |
return 0; |
| 1915 |
} |
| 1916 |
$post_id = function_exists('get_queried_object_id') ? get_queried_object_id() : 0; |
| 1917 |
return $post_id ? (int) $post_id : 0; |
| 1918 |
} |
| 1919 |
|
| 1920 |
/** |
| 1921 |
* Replace the first <meta $attr="$val"> content with $content_escaped (marked |
| 1922 |
* data-metasync-seo="custom"); insert one after <head> if none is present. |
| 1923 |
* Helper for apply_custom_seo_precedence(). |
| 1924 |
* |
| 1925 |
* @param string $html |
| 1926 |
* @param string $attr "name" or "property" |
| 1927 |
* @param string $val e.g. "description", "og:description" |
| 1928 |
* @param string $content_escaped Already HTML-escaped content value |
| 1929 |
* @return string |
| 1930 |
*/ |
| 1931 |
private function force_custom_meta($html, $attr, $val, $content_escaped) { |
| 1932 |
$tag = '<meta ' . $attr . '="' . $val . '" content="' . $content_escaped . '" data-metasync-seo="custom" />'; |
| 1933 |
$pattern = '/<meta\s[^>]*' . preg_quote($attr, '/') . '\s*=\s*["\']' . preg_quote($val, '/') . '["\'][^>]*\/?>/i'; |
| 1934 |
$count = 0; |
| 1935 |
$new = preg_replace_callback($pattern, function ($m) use ($tag) { |
| 1936 |
return $tag; |
| 1937 |
}, $html, 1, $count); |
| 1938 |
if ($count > 0) { |
| 1939 |
return $new; |
| 1940 |
} |
| 1941 |
# None present — insert after <head>. |
| 1942 |
return preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($tag) { |
| 1943 |
return $m[1] . "\n" . $tag; |
| 1944 |
}, $html, 1); |
| 1945 |
} |
| 1946 |
|
| 1947 |
/** |
| 1948 |
* Enforce the SEO-sidebar custom title/description as the FINAL word over OTTO. |
| 1949 |
* |
| 1950 |
* The DOM-pass guards (do_header_replacements/handle_meta_element) already skip |
| 1951 |
* OTTO for these fields, but later steps re-introduce OTTO's value: |
| 1952 |
* - the manual string-replacement title pass re-applies OTTO's title; |
| 1953 |
* - OTTO's description arrives via header_html_insertion (data-otto) and the |
| 1954 |
* dedup keeps it whenever no data-metasync-seo tag is present — which is |
| 1955 |
* exactly the case when the post is synced to a third-party SEO plugin and |
| 1956 |
* the sidebar deferred emitting its own tag (output_seo_meta_description()). |
| 1957 |
* |
| 1958 |
* Running LAST — after all dedup/precedence steps — this makes the user's |
| 1959 |
* explicit sidebar value authoritative regardless of how the tag was produced, |
| 1960 |
* mirroring apply_metabox_og_precedence() for the OG meta box. Singular views |
| 1961 |
* only; a value must be non-empty (explicitly set) to take precedence. |
| 1962 |
* |
| 1963 |
* @param string $html Full HTML document (post-dedup). |
| 1964 |
* @return string |
| 1965 |
*/ |
| 1966 |
private function apply_custom_seo_precedence($html) { |
| 1967 |
if ($html === '') { |
| 1968 |
return $html; |
| 1969 |
} |
| 1970 |
$post_id = $this->get_buffer_post_id(); |
| 1971 |
if (!$post_id) { |
| 1972 |
return $html; |
| 1973 |
} |
| 1974 |
|
| 1975 |
# Title — force the custom sidebar title into the (single, post-dedup) <title>. |
| 1976 |
$custom_title = get_post_meta($post_id, '_metasync_seo_title', true); |
| 1977 |
if (!empty($custom_title) && is_string($custom_title)) { |
| 1978 |
$title_tag = '<title>' . htmlspecialchars($custom_title, ENT_QUOTES, 'UTF-8') . '</title>'; |
| 1979 |
$count = 0; |
| 1980 |
$replaced = preg_replace_callback('/<title[^>]*>.*?<\/title>/is', function ($m) use ($title_tag) { |
| 1981 |
return $title_tag; |
| 1982 |
}, $html, 1, $count); |
| 1983 |
if ($count > 0) { |
| 1984 |
$html = $replaced; |
| 1985 |
} elseif (stripos($html, '<title') === false) { |
| 1986 |
$html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($title_tag) { |
| 1987 |
return $m[1] . "\n" . $title_tag; |
| 1988 |
}, $html, 1); |
| 1989 |
} |
| 1990 |
} |
| 1991 |
|
| 1992 |
# Description — force the custom sidebar description into the SEO |
| 1993 |
# <meta name="description"> only. og:/twitter: description are social tags |
| 1994 |
# owned by the OG meta box and apply_metabox_og_precedence() (which runs |
| 1995 |
# earlier); this pass must NOT touch them, or it would override a |
| 1996 |
# separately user-set og:description with the plain SEO description. |
| 1997 |
$custom_desc = get_post_meta($post_id, '_metasync_seo_desc', true); |
| 1998 |
if (!empty($custom_desc) && is_string($custom_desc)) { |
| 1999 |
$esc = htmlspecialchars($custom_desc, ENT_QUOTES, 'UTF-8'); |
| 2000 |
$html = $this->force_custom_meta($html, 'name', 'description', $esc); |
| 2001 |
} |
| 2002 |
|
| 2003 |
return $html; |
| 2004 |
} |
| 2005 |
|
| 2006 |
# this function does the header replacements |
| 2007 |
function do_header_replacements($replacement_data){ |
| 2008 |
|
| 2009 |
# check that we have header replacements |
| 2010 |
if(empty($replacement_data['header_replacements']) || !is_array($replacement_data['header_replacements'])){ |
| 2011 |
return; |
| 2012 |
} |
| 2013 |
|
| 2014 |
# Check for custom SEO values from MetaSync SEO Sidebar |
| 2015 |
# Custom values take absolute priority over OTTO suggestions |
| 2016 |
$custom_seo_title = ''; |
| 2017 |
$custom_seo_description = ''; |
| 2018 |
|
| 2019 |
if (function_exists('is_singular') && is_singular()) { |
| 2020 |
$post_id = get_the_ID(); |
| 2021 |
if ($post_id) { |
| 2022 |
$custom_seo_title = get_post_meta($post_id, '_metasync_seo_title', true); |
| 2023 |
$custom_seo_description = get_post_meta($post_id, '_metasync_seo_desc', true); |
| 2024 |
} |
| 2025 |
} |
| 2026 |
|
| 2027 |
# now lets do the replacement work |
| 2028 |
foreach($replacement_data['header_replacements'] AS $key => $data){ |
| 2029 |
|
| 2030 |
# skip cases where type is not specified |
| 2031 |
if(empty($data['type'])){ |
| 2032 |
continue; |
| 2033 |
} |
| 2034 |
|
| 2035 |
# handle title - skip if custom SEO title exists or no value |
| 2036 |
if($data['type'] == 'title'){ |
| 2037 |
if (!empty($custom_seo_title)) { |
| 2038 |
# Skip title replacement - custom SEO title takes priority |
| 2039 |
continue; |
| 2040 |
} |
| 2041 |
|
| 2042 |
# Skip if OTTO has no title value |
| 2043 |
if (empty(trim($data['recommended_value'] ?? $data['value'] ?? ''))) { |
| 2044 |
continue; |
| 2045 |
} |
| 2046 |
|
| 2047 |
# handle the title logic |
| 2048 |
$this->replace_title($data); |
| 2049 |
|
| 2050 |
# |
| 2051 |
continue; |
| 2052 |
} |
| 2053 |
|
| 2054 |
# handle canonical links |
| 2055 |
if($data['type'] == 'link' && $data['rel'] === 'canonical'){ |
| 2056 |
|
| 2057 |
# Protect manually-set canonical from OTTO override. |
| 2058 |
# Validated: a legacy row corrupted to "Array" must not |
| 2059 |
# count as a manual canonical and block OTTO's correct value. |
| 2060 |
if (function_exists('is_singular') && is_singular()) { |
| 2061 |
$post_id = get_the_ID(); |
| 2062 |
if ($post_id) { |
| 2063 |
$custom_canonical = Metasync_Canonical_Sanitizer::sanitize(get_post_meta($post_id, '_metasync_canonical_url', true)); |
| 2064 |
if ($custom_canonical === '') { |
| 2065 |
$custom_canonical = Metasync_Canonical_Sanitizer::sanitize(get_post_meta($post_id, 'meta_canonical', true)); |
| 2066 |
} |
| 2067 |
if ($custom_canonical !== '') { |
| 2068 |
# Manual canonical takes priority — skip OTTO override |
| 2069 |
continue; |
| 2070 |
} |
| 2071 |
} |
| 2072 |
} |
| 2073 |
|
| 2074 |
# find the cannonical dom element |
| 2075 |
$link = $this->dom->find('link[rel="canonical"]', 0); |
| 2076 |
|
| 2077 |
# set the link property if not empty — only to a validated URL |
| 2078 |
# never inject a corrupted/non-URL suggestion. |
| 2079 |
if(!empty($link->href)){ |
| 2080 |
$otto_canonical = Metasync_Canonical_Sanitizer::sanitize($data['recommended_value'] ?? ''); |
| 2081 |
if ($otto_canonical !== '') { |
| 2082 |
$link->href = $otto_canonical; |
| 2083 |
} |
| 2084 |
} |
| 2085 |
|
| 2086 |
# |
| 2087 |
continue; |
| 2088 |
} |
| 2089 |
|
| 2090 |
# work on other elemenets not titlte |
| 2091 |
$this->handle_meta_element($data); |
| 2092 |
} |
| 2093 |
} |
| 2094 |
|
| 2095 |
# function to handle meta elements other than title |
| 2096 |
function handle_meta_element($data){ |
| 2097 |
|
| 2098 |
# Skip if OTTO has no value to set - prevents overwriting existing tags (e.g. Yoast) |
| 2099 |
# with empty content when OTTO has no recommendation for this meta field |
| 2100 |
$recommended_value = $data['recommended_value'] ?? $data['value'] ?? ''; |
| 2101 |
if (is_object($recommended_value) || is_array($recommended_value)) { |
| 2102 |
return; |
| 2103 |
} |
| 2104 |
if (!is_string($recommended_value)) { |
| 2105 |
$recommended_value = (string) $recommended_value; |
| 2106 |
} |
| 2107 |
if (empty(trim($recommended_value))) { |
| 2108 |
return; |
| 2109 |
} |
| 2110 |
|
| 2111 |
# Check if this is a description meta tag and if custom description exists |
| 2112 |
# Custom values take absolute priority over OTTO suggestions |
| 2113 |
$name = $data['name'] ?? false; |
| 2114 |
$property = $data['property'] ?? false; |
| 2115 |
|
| 2116 |
# Protect manually-set robots meta from OTTO override |
| 2117 |
# If the user set noindex via Common Robots Meta or meta_robots, honour it |
| 2118 |
if (!empty($name) && $name === 'robots') { |
| 2119 |
if (function_exists('is_singular') && is_singular()) { |
| 2120 |
$post_id = get_the_ID(); |
| 2121 |
if ($post_id) { |
| 2122 |
$manual_robots = get_post_meta($post_id, 'meta_robots', true); |
| 2123 |
if (!empty($manual_robots) && stripos($manual_robots, 'noindex') !== false) { |
| 2124 |
# Manual noindex takes priority — skip OTTO override |
| 2125 |
return; |
| 2126 |
} |
| 2127 |
$common_robots = get_post_meta($post_id, 'metasync_common_robots', true); |
| 2128 |
if (is_array($common_robots) && !empty($common_robots['noindex'])) { |
| 2129 |
# Common Robots Meta has noindex checked — skip OTTO override |
| 2130 |
return; |
| 2131 |
} |
| 2132 |
} |
| 2133 |
} |
| 2134 |
} |
| 2135 |
|
| 2136 |
# Check for custom SEO description |
| 2137 |
if (!empty($name) && $name === 'description') { |
| 2138 |
if (function_exists('is_singular') && is_singular()) { |
| 2139 |
$post_id = get_the_ID(); |
| 2140 |
if ($post_id) { |
| 2141 |
$custom_seo_description = get_post_meta($post_id, '_metasync_seo_desc', true); |
| 2142 |
if (!empty($custom_seo_description)) { |
| 2143 |
# Custom description exists, skip OTTO's suggestion |
| 2144 |
return; |
| 2145 |
} |
| 2146 |
} |
| 2147 |
} |
| 2148 |
} |
| 2149 |
|
| 2150 |
# extract property value |
| 2151 |
$property = $data['property'] ?? false; |
| 2152 |
|
| 2153 |
# extract name value |
| 2154 |
$name = $data['name'] ?? false; |
| 2155 |
|
| 2156 |
# set the selector |
| 2157 |
$meta_selector = ''; |
| 2158 |
|
| 2159 |
# extend selector |
| 2160 |
if(!empty($name)){ |
| 2161 |
$meta_selector .= 'meta[name="' . trim($name) . '"]'; |
| 2162 |
} |
| 2163 |
|
| 2164 |
# extent if property is defined |
| 2165 |
if(!empty($property)){ |
| 2166 |
if (!empty($meta_selector)) { |
| 2167 |
$meta_selector .= ','; |
| 2168 |
} |
| 2169 |
$meta_selector .= 'meta[property="' . trim($property) . '"]'; |
| 2170 |
} |
| 2171 |
|
| 2172 |
# find the meta gat in the dom |
| 2173 |
$meta_tag = $this->dom->find($meta_selector, 0); |
| 2174 |
|
| 2175 |
# if tag not exists add it |
| 2176 |
if(empty($meta_tag)){ |
| 2177 |
if($data['type'] == 'meta'){ |
| 2178 |
|
| 2179 |
# get the attribute |
| 2180 |
$attribute = $property ? 'property' : 'name'; |
| 2181 |
|
| 2182 |
# call the create metatag function |
| 2183 |
$result = $this->create_metatag($attribute, $data); |
| 2184 |
} |
| 2185 |
|
| 2186 |
# return after creation |
| 2187 |
return; |
| 2188 |
} |
| 2189 |
|
| 2190 |
# CRITICAL FIX: Clear cache and get fresh reference |
| 2191 |
# Preserve 'imgs' key for later use by handle_images() |
| 2192 |
$preserved_imgs = $this->cached_elements['imgs'] ?? null; |
| 2193 |
$this->cached_elements = []; |
| 2194 |
if ($preserved_imgs !== null) { |
| 2195 |
$this->cached_elements['imgs'] = $preserved_imgs; |
| 2196 |
} |
| 2197 |
|
| 2198 |
# Get fresh meta tag reference using same selector |
| 2199 |
$meta_tag_fresh = $this->dom->find($meta_selector, 0); |
| 2200 |
|
| 2201 |
if ($meta_tag_fresh) { |
| 2202 |
# Use outertext for replacement |
| 2203 |
$new_value = htmlspecialchars($recommended_value, ENT_QUOTES, 'UTF-8'); |
| 2204 |
|
| 2205 |
# Determine attribute name |
| 2206 |
$attr_name = !empty($data['name']) ? 'name' : 'property'; |
| 2207 |
$attr_value = !empty($data['name']) ? $data['name'] : ($data['property'] ?? ''); |
| 2208 |
|
| 2209 |
# Build new meta tag |
| 2210 |
$meta_tag_fresh->outertext = '<meta ' . $attr_name . '="' . htmlspecialchars($attr_value, ENT_QUOTES, 'UTF-8') . '" content="' . $new_value . '">'; |
| 2211 |
} |
| 2212 |
|
| 2213 |
|
| 2214 |
} |
| 2215 |
|
| 2216 |
# function to handle the page title |
| 2217 |
function replace_title($title_data){ |
| 2218 |
|
| 2219 |
# find the title |
| 2220 |
$title = $this->dom->find('title', 0) ?? false; |
| 2221 |
|
| 2222 |
# if none |
| 2223 |
if($title === false){ |
| 2224 |
return $this->create_title($title_data); |
| 2225 |
} |
| 2226 |
|
| 2227 |
# CRITICAL FIX: Clear element cache and get fresh reference |
| 2228 |
# Preserve 'imgs' key for later use by handle_images() |
| 2229 |
$preserved_imgs = $this->cached_elements['imgs'] ?? null; |
| 2230 |
$this->cached_elements = []; |
| 2231 |
if ($preserved_imgs !== null) { |
| 2232 |
$this->cached_elements['imgs'] = $preserved_imgs; |
| 2233 |
} |
| 2234 |
|
| 2235 |
# Get fresh title element reference |
| 2236 |
$title_fresh = $this->dom->find('title', 0); |
| 2237 |
|
| 2238 |
if ($title_fresh) { |
| 2239 |
# Use outertext for replacement |
| 2240 |
$new_value = htmlspecialchars($title_data['recommended_value'], ENT_QUOTES, 'UTF-8'); |
| 2241 |
$title_fresh->outertext = '<title>' . $new_value . '</title>'; |
| 2242 |
} |
| 2243 |
|
| 2244 |
# Don't save_reload here - will happen at the end |
| 2245 |
# $this->save_reload(); |
| 2246 |
} |
| 2247 |
|
| 2248 |
# Function to create a title when it's missing |
| 2249 |
function create_title($title_data) { |
| 2250 |
|
| 2251 |
# Find the <head> tag |
| 2252 |
$head = $this->dom->find('head', 0); |
| 2253 |
|
| 2254 |
# Construct the <title> tag HTML |
| 2255 |
$title_html = '<title>' . htmlspecialchars($title_data['recommended_value'], ENT_QUOTES) . '</title>'; |
| 2256 |
|
| 2257 |
if (empty($head)) { |
| 2258 |
return false; |
| 2259 |
} |
| 2260 |
|
| 2261 |
# Extract existing attributes of the <head> tag |
| 2262 |
$attributes = []; |
| 2263 |
|
| 2264 |
# get the tag attributes |
| 2265 |
$tag_attributes = $head->getAllAttributes(); |
| 2266 |
|
| 2267 |
# loop all attributes |
| 2268 |
foreach ($tag_attributes as $key => $value) { |
| 2269 |
|
| 2270 |
if ($value == 1) { |
| 2271 |
|
| 2272 |
# Handle boolean attributes |
| 2273 |
$attributes[] = htmlspecialchars($key, ENT_QUOTES); |
| 2274 |
} else { |
| 2275 |
|
| 2276 |
# Handle attributes with values |
| 2277 |
$attributes[] = $key . '="' . htmlspecialchars($value, ENT_QUOTES) . '"'; |
| 2278 |
} |
| 2279 |
} |
| 2280 |
|
| 2281 |
# Convert attributes array to a string |
| 2282 |
$attributes_string = !empty($attributes) ? ' ' . implode(' ', $attributes) : ''; |
| 2283 |
|
| 2284 |
# Rebuild the <head> tag, inserting the <title> at the beginning |
| 2285 |
$head->outertext = '<head' . $attributes_string . '>' . $title_html . $head->innertext . '</head>'; |
| 2286 |
|
| 2287 |
# save and reload DOM |
| 2288 |
$this->save_reload(); |
| 2289 |
} |
| 2290 |
|
| 2291 |
# function to create meta tag if none existss |
| 2292 |
function create_metatag($attribute, $data){ |
| 2293 |
|
| 2294 |
# Find the <head> tag |
| 2295 |
$head = $this->dom->find('head', 0); |
| 2296 |
|
| 2297 |
# Construct the meta tag HTML (no spaces around = for standard HTML and regex compatibility) |
| 2298 |
$meta_tag = '<meta '.$attribute.'="'.htmlspecialchars($data[$attribute], ENT_QUOTES, 'UTF-8').'" content="'.htmlspecialchars($data['recommended_value'], ENT_QUOTES, 'UTF-8').'">'; |
| 2299 |
|
| 2300 |
if (empty($head)) { |
| 2301 |
return false; |
| 2302 |
} |
| 2303 |
|
| 2304 |
# Extract existing attributes of the <head> tag |
| 2305 |
$attributes = []; |
| 2306 |
|
| 2307 |
# get the tag attributes |
| 2308 |
$tag_attributes = $head->getAllAttributes(); |
| 2309 |
|
| 2310 |
# loop all attributes |
| 2311 |
foreach ($tag_attributes as $key => $value) { |
| 2312 |
|
| 2313 |
if ($value == 1) { |
| 2314 |
|
| 2315 |
# Handle boolean attributes |
| 2316 |
$attributes[] = htmlspecialchars($key, ENT_QUOTES); |
| 2317 |
} else { |
| 2318 |
|
| 2319 |
# Handle attributes with values |
| 2320 |
$attributes[] = $key . '="' . htmlspecialchars($value, ENT_QUOTES) . '"'; |
| 2321 |
} |
| 2322 |
} |
| 2323 |
|
| 2324 |
# Convert attributes array to a string |
| 2325 |
$attributes_string = !empty($attributes) ? ' ' . implode(' ', $attributes) : ''; |
| 2326 |
|
| 2327 |
# Rebuild the <head> tag, inserting the <title> at the beginning |
| 2328 |
$head->outertext = '<head' . $attributes_string . '>' . $meta_tag . $head->innertext . '</head>'; |
| 2329 |
|
| 2330 |
# save and reload DOM |
| 2331 |
$this->save_reload(); |
| 2332 |
} |
| 2333 |
|
| 2334 |
/** |
| 2335 |
* Remove duplicate <title> tags from HTML, keeping the first (OTTO's) value. |
| 2336 |
* |
| 2337 |
* OTTO's title replacement is always applied first (limit=1 or DOM manipulation), |
| 2338 |
* so the first <title> tag holds the authoritative value. If SEO plugin conflicts |
| 2339 |
* produce additional <title> tags, this strips all and re-inserts one. |
| 2340 |
* |
| 2341 |
* @param string $html Full HTML document. |
| 2342 |
* @return string HTML with at most one <title> tag. |
| 2343 |
*/ |
| 2344 |
private function deduplicate_title_tags($html) { |
| 2345 |
$title_count = preg_match_all('/<title[^>]*>.*?<\/title>/is', $html, $title_matches); |
| 2346 |
if ($title_count <= 1) { |
| 2347 |
return $html; |
| 2348 |
} |
| 2349 |
|
| 2350 |
# Capture the first title's inner text (OTTO's replacement) |
| 2351 |
preg_match('/<title[^>]*>(.*?)<\/title>/is', $html, $first_title); |
| 2352 |
$authoritative_title = isset($first_title[1]) ? $first_title[1] : ''; |
| 2353 |
|
| 2354 |
# Strip all <title> tags |
| 2355 |
$html = preg_replace('/<title[^>]*>.*?<\/title>/is', '', $html); |
| 2356 |
|
| 2357 |
# Re-insert a single <title> after <head> using preg_replace_callback to prevent |
| 2358 |
# backreference injection when title contains $ followed by digits (e.g. "$50 off") |
| 2359 |
$title_tag = '<title>' . $authoritative_title . '</title>'; |
| 2360 |
$html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($title_tag) { |
| 2361 |
return $m[1] . $title_tag; |
| 2362 |
}, $html, 1); |
| 2363 |
|
| 2364 |
return $html; |
| 2365 |
} |
| 2366 |
|
| 2367 |
/** |
| 2368 |
* Remove duplicate OG and Twitter meta tags from HTML after OTTO processing. |
| 2369 |
* |
| 2370 |
* OTTO injects its tags with `data-otto-pixel` or `data-otto` attributes. |
| 2371 |
* Legacy MetaSync output and third-party SEO plugins may also emit the same |
| 2372 |
* OG/Twitter properties. This method runs at the buffer level — after all |
| 2373 |
* sources have written their tags — and keeps only the OTTO version when |
| 2374 |
* duplicates exist. |
| 2375 |
* |
| 2376 |
* Strategy per property (e.g. og:description): |
| 2377 |
* - If OTTO tag exists (has data-otto marker) → remove all non-OTTO duplicates |
| 2378 |
* - If no OTTO tag exists → keep the first occurrence, remove the rest |
| 2379 |
* |
| 2380 |
* @param string $html Full HTML document. |
| 2381 |
* @return string HTML with at most one tag per OG/Twitter property. |
| 2382 |
*/ |
| 2383 |
private function deduplicate_og_twitter_tags($html) { |
| 2384 |
# OG properties to deduplicate |
| 2385 |
$og_properties = [ |
| 2386 |
'og:title', 'og:description', 'og:url', 'og:type', |
| 2387 |
'og:locale', 'og:site_name', 'og:image', |
| 2388 |
]; |
| 2389 |
|
| 2390 |
foreach ($og_properties as $prop) { |
| 2391 |
$html = $this->deduplicate_meta_by_attr($html, 'property', $prop); |
| 2392 |
} |
| 2393 |
|
| 2394 |
# Twitter names to deduplicate |
| 2395 |
$twitter_names = [ |
| 2396 |
'twitter:title', 'twitter:description', 'twitter:card', |
| 2397 |
'twitter:image', 'twitter:site', |
| 2398 |
]; |
| 2399 |
|
| 2400 |
foreach ($twitter_names as $name) { |
| 2401 |
$html = $this->deduplicate_meta_by_attr($html, 'name', $name); |
| 2402 |
} |
| 2403 |
|
| 2404 |
return $html; |
| 2405 |
} |
| 2406 |
|
| 2407 |
/** |
| 2408 |
* Apply the per-post "Social Media & Open Graph" meta box precedence over OTTO. |
| 2409 |
* |
| 2410 |
* Precedence per tag: an explicit meta box value set by the user wins over OTTO; |
| 2411 |
* a field left blank falls back to OTTO's value; if OTTO also has none, a sensible |
| 2412 |
* default fills the gap. Mirrors how the custom SEO sidebar values take priority |
| 2413 |
* over OTTO (see do_header_replacements()). |
| 2414 |
* |
| 2415 |
* Runs *after* deduplicate_og_twitter_tags(), and only within <head>. For a field the |
| 2416 |
* user explicitly set it removes any existing tag (OTTO's) and re-adds the user's |
| 2417 |
* value; for the rest it only adds a tag when absent — so OTTO stays authoritative |
| 2418 |
* where the user left a field blank and no duplicate is produced. OTTO never emits an |
| 2419 |
* og:image, so this also surfaces a user-set social image on an OTTO-managed page. |
| 2420 |
* |
| 2421 |
* Only applies to singular posts/pages: the meta box values are per-post, and on |
| 2422 |
* archives/home get_queried_object_id() can return a term or user id. |
| 2423 |
* |
| 2424 |
* @param string $html Full HTML document (post-dedup). |
| 2425 |
* @return string |
| 2426 |
*/ |
| 2427 |
private function apply_metabox_og_precedence($html) { |
| 2428 |
if (!function_exists('get_queried_object_id') || !function_exists('is_singular')) { |
| 2429 |
return $html; |
| 2430 |
} |
| 2431 |
# Per-post OG only applies to singular posts/pages. On archives/home the queried |
| 2432 |
# object id can be a term or user id, which must not be read as post meta. |
| 2433 |
if (!is_singular()) { |
| 2434 |
return $html; |
| 2435 |
} |
| 2436 |
$post_id = get_queried_object_id(); |
| 2437 |
if (!$post_id) { |
| 2438 |
return $html; |
| 2439 |
} |
| 2440 |
|
| 2441 |
# Respect the meta box opt-out (only an explicit '0' disables it) |
| 2442 |
if (get_post_meta($post_id, '_metasync_og_enabled', true) === '0') { |
| 2443 |
return $html; |
| 2444 |
} |
| 2445 |
|
| 2446 |
# Explicit (user-typed) meta box values. A non-empty value means the user set |
| 2447 |
# this field themselves and it must take priority over OTTO. |
| 2448 |
$og_title_set = get_post_meta($post_id, '_metasync_og_title', true); |
| 2449 |
$og_desc_set = get_post_meta($post_id, '_metasync_og_description', true); |
| 2450 |
$og_image_set = get_post_meta($post_id, '_metasync_og_image', true); |
| 2451 |
$og_type_set = get_post_meta($post_id, '_metasync_og_type', true); |
| 2452 |
$tw_card_set = get_post_meta($post_id, '_metasync_twitter_card', true); |
| 2453 |
$tw_site_set = get_post_meta($post_id, '_metasync_twitter_site', true); |
| 2454 |
$tw_title_set = get_post_meta($post_id, '_metasync_twitter_title', true); |
| 2455 |
$tw_desc_set = get_post_meta($post_id, '_metasync_twitter_description', true); |
| 2456 |
$tw_image_set = get_post_meta($post_id, '_metasync_twitter_image', true); |
| 2457 |
$tw_image_alt_set = get_post_meta($post_id, '_metasync_twitter_image_alt', true); |
| 2458 |
|
| 2459 |
# Default OG values the meta box pre-fills (post title / generated excerpt / |
| 2460 |
# featured image). The meta box PERSISTS these defaults on save, so a non-empty |
| 2461 |
# _metasync_og_* value alone does not prove the user customized it — a field is a |
| 2462 |
# genuine override only when its stored value differs from this default. Reuse the |
| 2463 |
# emitter's own resolver so the defaults match exactly (falling back to empty |
| 2464 |
# defaults, i.e. treat nothing as customized, if the instance is unavailable). |
| 2465 |
$defaults = ['title' => '', 'description' => '', 'image' => '']; |
| 2466 |
if (class_exists('Metasync_OpenGraph') && Metasync_OpenGraph::get_instance()) { |
| 2467 |
$defaults = Metasync_OpenGraph::get_instance()->get_default_og_values($post_id); |
| 2468 |
} |
| 2469 |
|
| 2470 |
$title_is_custom = ($og_title_set !== '' && $og_title_set !== $defaults['title']); |
| 2471 |
$desc_is_custom = ($og_desc_set !== '' && $og_desc_set !== $defaults['description']); |
| 2472 |
$image_is_custom = ($og_image_set !== '' && $og_image_set !== $defaults['image']); |
| 2473 |
|
| 2474 |
# Resolved values (explicit value first, then the meta box default). |
| 2475 |
$og_title = $og_title_set !== '' ? $og_title_set : $defaults['title']; |
| 2476 |
$og_desc = $og_desc_set !== '' ? $og_desc_set : $defaults['description']; |
| 2477 |
$og_image = $og_image_set !== '' ? $og_image_set : $defaults['image']; |
| 2478 |
$og_type = $og_type_set ?: 'article'; |
| 2479 |
$site_name = get_bloginfo('name'); |
| 2480 |
|
| 2481 |
# Twitter inherits the OG value when its own field is blank; a user-customized |
| 2482 |
# og:image therefore also overrides OTTO's twitter:image (the meta box treats them |
| 2483 |
# as one). A twitter field is custom if its own value differs from the default OR |
| 2484 |
# the OG field it inherits was customized. |
| 2485 |
$tw_card = $tw_card_set ?: 'summary_large_image'; |
| 2486 |
$tw_title = $tw_title_set !== '' ? $tw_title_set : $og_title; |
| 2487 |
$tw_desc = $tw_desc_set !== '' ? $tw_desc_set : $og_desc; |
| 2488 |
$tw_image = $tw_image_set !== '' ? $tw_image_set : $og_image; |
| 2489 |
|
| 2490 |
$tw_title_is_custom = ($tw_title_set !== '' && $tw_title_set !== $defaults['title']) || $title_is_custom; |
| 2491 |
$tw_desc_is_custom = ($tw_desc_set !== '' && $tw_desc_set !== $defaults['description']) || $desc_is_custom; |
| 2492 |
$tw_image_is_custom = ($tw_image_set !== '' && $tw_image_set !== $defaults['image']) || $image_is_custom; |
| 2493 |
|
| 2494 |
# twitter:site: per-post value, else the site-wide handle from Social Meta settings. |
| 2495 |
# twitter:site / twitter:image:alt are not auto-defaulted by the meta box, so a |
| 2496 |
# non-empty per-post value is itself the override signal. |
| 2497 |
$tw_site = $tw_site_set; |
| 2498 |
if ($tw_site === '' && class_exists('Metasync')) { |
| 2499 |
$social = Metasync::get_option('social_meta'); |
| 2500 |
if (is_array($social) && !empty($social['twitter_username'])) { |
| 2501 |
$tw_site = '@' . ltrim($social['twitter_username'], '@'); |
| 2502 |
} |
| 2503 |
} |
| 2504 |
|
| 2505 |
# og:type / twitter:card are dropdowns that always carry a value, so they override |
| 2506 |
# OTTO only when the user picked a non-default option. og:url is auto-populated with |
| 2507 |
# the permalink on save, so it is fill-only (never an override) to avoid clobbering |
| 2508 |
# OTTO's canonical URL with a stale/auto value. og:site_name is a site default. |
| 2509 |
$og_type_override = ($og_type_set !== '' && $og_type_set !== 'article'); |
| 2510 |
$tw_card_override = ($tw_card_set !== '' && $tw_card_set !== 'summary_large_image'); |
| 2511 |
|
| 2512 |
# [attr, property/name, value, is_url, is_override] |
| 2513 |
$tags = [ |
| 2514 |
['property', 'og:title', $og_title, false, $title_is_custom], |
| 2515 |
['property', 'og:description', $og_desc, false, $desc_is_custom], |
| 2516 |
['property', 'og:image', $og_image, true, $image_is_custom], |
| 2517 |
['property', 'og:url', get_permalink($post_id), true, false], |
| 2518 |
['property', 'og:type', $og_type, false, $og_type_override], |
| 2519 |
['property', 'og:site_name', $site_name, false, false], |
| 2520 |
['name', 'twitter:card', $tw_card, false, $tw_card_override], |
| 2521 |
['name', 'twitter:site', $tw_site, false, $tw_site_set !== ''], |
| 2522 |
['name', 'twitter:title', $tw_title, false, $tw_title_is_custom], |
| 2523 |
['name', 'twitter:description', $tw_desc, false, $tw_desc_is_custom], |
| 2524 |
['name', 'twitter:image', $tw_image, true, $tw_image_is_custom], |
| 2525 |
['name', 'twitter:image:alt', $tw_image_alt_set, false, $tw_image_alt_set !== ''], |
| 2526 |
]; |
| 2527 |
|
| 2528 |
# Operate only within <head> so meta tags in the body are never touched. |
| 2529 |
if (!preg_match('/(<head\b[^>]*>)(.*?)(<\/head>)/is', $html, $hm)) { |
| 2530 |
return $html; |
| 2531 |
} |
| 2532 |
$head = $hm[2]; |
| 2533 |
|
| 2534 |
# 1) Remove existing head tags for user-overridden properties so the user's value wins. |
| 2535 |
foreach ($tags as $t) { |
| 2536 |
list($attr, $key, $val, $is_url, $is_override) = $t; |
| 2537 |
if ($is_override && $val !== '' && $val !== null) { |
| 2538 |
$stripped = preg_replace( |
| 2539 |
'/<meta\s[^>]*' . $attr . '\s*=\s*(["\'])' . preg_quote($key, '/') . '\1[^>]*>\s*/i', |
| 2540 |
'', |
| 2541 |
$head |
| 2542 |
); |
| 2543 |
if ($stripped !== null) { |
| 2544 |
$head = $stripped; |
| 2545 |
} |
| 2546 |
} |
| 2547 |
} |
| 2548 |
|
| 2549 |
# 2) Build additions: overrides (removed above) + gap-fills for anything still absent. |
| 2550 |
# |
| 2551 |
# When a third-party SEO plugin is active the conflict handler may have suppressed |
| 2552 |
# that plugin's og:title / twitter:title so MetaSync can supply the replacement |
| 2553 |
# (Metasync_SEO_Sidebar::output_seo_meta_description). Gap-filling those two tags |
| 2554 |
# here would inject the meta-box post-title default over that replacement, so skip |
| 2555 |
# them in that case. User-customized overrides are unaffected — they were already |
| 2556 |
# applied above. |
| 2557 |
$seo_plugin_active = class_exists('Metasync_SEO_Conflict_Handler') |
| 2558 |
&& Metasync_SEO_Conflict_Handler::get_instance()->has_active_seo_plugin(); |
| 2559 |
|
| 2560 |
$additions = ''; |
| 2561 |
foreach ($tags as $t) { |
| 2562 |
list($attr, $key, $val, $is_url, $is_override) = $t; |
| 2563 |
if ($val === '' || $val === null) { |
| 2564 |
continue; |
| 2565 |
} |
| 2566 |
if (!$is_override && $seo_plugin_active && ($key === 'og:title' || $key === 'twitter:title')) { |
| 2567 |
continue; # never gap-fill the post title over a third-party plugin's own title |
| 2568 |
} |
| 2569 |
if (!$is_override && preg_match('/<meta\s[^>]*' . $attr . '\s*=\s*(["\'])' . preg_quote($key, '/') . '\1/i', $head)) { |
| 2570 |
continue; # OTTO (or another source) already provides it and the user did not override |
| 2571 |
} |
| 2572 |
$content = $is_url ? esc_url($val) : esc_attr($val); |
| 2573 |
if ($content === '') { |
| 2574 |
continue; |
| 2575 |
} |
| 2576 |
$marker = $is_override ? 'override' : 'fill'; |
| 2577 |
$additions .= '<meta ' . $attr . '="' . $key . '" content="' . $content . '" data-metasync-og="' . $marker . '">' . "\n"; |
| 2578 |
} |
| 2579 |
|
| 2580 |
# Reassemble the head block (str_replace avoids backreference issues from $ in values). |
| 2581 |
if ($head === $hm[2] && $additions === '') { |
| 2582 |
return $html; # nothing removed, nothing added |
| 2583 |
} |
| 2584 |
$new_head_block = $hm[1] . $head . $additions . $hm[3]; |
| 2585 |
return str_replace($hm[0], $new_head_block, $html); |
| 2586 |
} |
| 2587 |
|
| 2588 |
/** |
| 2589 |
* Deduplicate meta tags by a specific attribute (property= or name=). |
| 2590 |
* |
| 2591 |
* When duplicates exist and one carries a data-otto marker, keep only |
| 2592 |
* the OTTO version. Otherwise keep the first occurrence. |
| 2593 |
* |
| 2594 |
* @param string $html Full HTML. |
| 2595 |
* @param string $attr_name Attribute name: 'property' or 'name'. |
| 2596 |
* @param string $attr_val Attribute value: e.g. 'og:title' or 'twitter:description'. |
| 2597 |
* @return string |
| 2598 |
*/ |
| 2599 |
private function deduplicate_meta_by_attr($html, $attr_name, $attr_val) { |
| 2600 |
$escaped = preg_quote($attr_val, '/'); |
| 2601 |
# Match all <meta> tags with this attribute value (both attr orderings) |
| 2602 |
$pattern = '/<meta\s[^>]*' . preg_quote($attr_name, '/') . '\s*=\s*["\']' . $escaped . '["\'][^>]*\/?>/i'; |
| 2603 |
|
| 2604 |
if (preg_match_all($pattern, $html, $matches) <= 1) { |
| 2605 |
return $html; # 0 or 1 — nothing to deduplicate |
| 2606 |
} |
| 2607 |
|
| 2608 |
$all_tags = $matches[0]; |
| 2609 |
|
| 2610 |
# Choose the keeper by precedence: custom sidebar → OTTO → first. |
| 2611 |
# A user-set SEO sidebar value (data-metasync-seo="custom") must win over |
| 2612 |
# OTTO, matching deduplicate_description_tags() and the sidebar's documented |
| 2613 |
# "custom always wins over OTTO" intent. Without this, the generic keeper |
| 2614 |
# (OTTO-or-first) drops the sidebar's og:/twitter: description in favor of |
| 2615 |
# OTTO's injected tag. |
| 2616 |
$custom_tag = null; |
| 2617 |
$otto_tag = null; |
| 2618 |
foreach ($all_tags as $tag) { |
| 2619 |
if ($custom_tag === null && stripos($tag, 'data-metasync-seo') !== false) { |
| 2620 |
$custom_tag = $tag; |
| 2621 |
} |
| 2622 |
if ($otto_tag === null && stripos($tag, 'data-otto') !== false) { |
| 2623 |
$otto_tag = $tag; |
| 2624 |
} |
| 2625 |
} |
| 2626 |
|
| 2627 |
# Determine the keeper: custom sidebar tag, else OTTO tag, else the first tag |
| 2628 |
$keeper = $custom_tag ?: ($otto_tag ?: $all_tags[0]); |
| 2629 |
|
| 2630 |
# Remove all occurrences, then re-insert the keeper at the first position |
| 2631 |
$first_replaced = false; |
| 2632 |
$html = preg_replace_callback($pattern, function ($m) use ($keeper, &$first_replaced) { |
| 2633 |
if (!$first_replaced) { |
| 2634 |
$first_replaced = true; |
| 2635 |
return $keeper; |
| 2636 |
} |
| 2637 |
return ''; # Remove subsequent duplicates |
| 2638 |
}, $html); |
| 2639 |
|
| 2640 |
return $html; |
| 2641 |
} |
| 2642 |
|
| 2643 |
/** |
| 2644 |
* Deduplicate <meta name="description"> tags after OTTO processing. |
| 2645 |
* |
| 2646 |
* Unlike title, og, twitter, and canonical tags, the plain name="description" |
| 2647 |
* tag had NO dedup pass, so a page could end up with 2-3 copies when several |
| 2648 |
* subsystems each emit one: |
| 2649 |
* - OTTO backend payload (header_html_insertion) → data-otto-pixel="dynamic-seo" |
| 2650 |
* (spliced in additively before </head>, never replacing an existing tag) |
| 2651 |
* - MetaSync's persisted-meta wp_head hook → data-metasync-otto="true" |
| 2652 |
* - MetaSync SEO sidebar custom value → data-metasync-seo="custom" |
| 2653 |
* |
| 2654 |
* This runs at the buffer level — after every source has written its tag — |
| 2655 |
* and keeps exactly ONE, by precedence (matching the SEO sidebar's documented |
| 2656 |
* "custom always wins over OTTO" intent): |
| 2657 |
* 1. custom sidebar (data-metasync-seo) |
| 2658 |
* 2. OTTO (data-otto-pixel OR data-metasync-otto OR data-otto) |
| 2659 |
* 3. first occurrence |
| 2660 |
* |
| 2661 |
* Note: the generic deduplicate_meta_by_attr() keeper-detection only matches |
| 2662 |
* substring "data-otto", which misses "data-metasync-otto"; this method checks |
| 2663 |
* both OTTO markers explicitly, so either OTTO source is recognized. |
| 2664 |
* |
| 2665 |
* @param string $html Full HTML document. |
| 2666 |
* @return string HTML with at most one <meta name="description">. |
| 2667 |
*/ |
| 2668 |
private function deduplicate_description_tags($html) { |
| 2669 |
if (!is_string($html) || $html === '') { |
| 2670 |
return $html; |
| 2671 |
} |
| 2672 |
|
| 2673 |
# Match <meta ... name="description" ...> in either attribute order. |
| 2674 |
# [^>]* is bounded to a single tag; name="twitter:description" is NOT matched |
| 2675 |
# because the opening quote must be immediately followed by "description". |
| 2676 |
$pattern = '/<meta\s[^>]*name\s*=\s*["\']description["\'][^>]*\/?>/i'; |
| 2677 |
|
| 2678 |
if (preg_match_all($pattern, $html, $matches) <= 1) { |
| 2679 |
return $html; # 0 or 1 — nothing to deduplicate |
| 2680 |
} |
| 2681 |
|
| 2682 |
$all_tags = $matches[0]; |
| 2683 |
|
| 2684 |
# Choose the keeper by precedence: custom sidebar → OTTO → first. |
| 2685 |
$custom_tag = null; |
| 2686 |
$otto_tag = null; |
| 2687 |
foreach ($all_tags as $tag) { |
| 2688 |
if ($custom_tag === null && stripos($tag, 'data-metasync-seo') !== false) { |
| 2689 |
$custom_tag = $tag; |
| 2690 |
} |
| 2691 |
if ($otto_tag === null && ( |
| 2692 |
stripos($tag, 'data-otto-pixel') !== false || |
| 2693 |
stripos($tag, 'data-metasync-otto') !== false || |
| 2694 |
stripos($tag, 'data-otto') !== false |
| 2695 |
)) { |
| 2696 |
$otto_tag = $tag; |
| 2697 |
} |
| 2698 |
} |
| 2699 |
$keeper = $custom_tag ?: ($otto_tag ?: $all_tags[0]); |
| 2700 |
|
| 2701 |
# Remove all occurrences, re-inserting the keeper at the first position. |
| 2702 |
# Callback form avoids backreference injection when the description content |
| 2703 |
# contains $ followed by digits (e.g. "$50 off"). |
| 2704 |
$first_replaced = false; |
| 2705 |
$html = preg_replace_callback($pattern, function ($m) use ($keeper, &$first_replaced) { |
| 2706 |
if (!$first_replaced) { |
| 2707 |
$first_replaced = true; |
| 2708 |
return $keeper; |
| 2709 |
} |
| 2710 |
return ''; # Remove subsequent duplicates |
| 2711 |
}, $html); |
| 2712 |
|
| 2713 |
return $html; |
| 2714 |
} |
| 2715 |
|
| 2716 |
/** |
| 2717 |
* Remove duplicate <link rel="canonical"> tags from HTML. |
| 2718 |
* |
| 2719 |
* When OTTO injects a canonical via header_html_insertion and MetaSync's |
| 2720 |
* SEO output (or WordPress core) has already emitted one, keep only the |
| 2721 |
* OTTO version (identified by data-otto marker). If no OTTO tag exists, |
| 2722 |
* keep the first occurrence. |
| 2723 |
* |
| 2724 |
* @param string $html Full HTML document. |
| 2725 |
* @return string HTML with at most one canonical tag. |
| 2726 |
*/ |
| 2727 |
private function deduplicate_canonical_tags($html) { |
| 2728 |
$pattern = '/<link\s[^>]*rel=["\']canonical["\'][^>]*\/?>/i'; |
| 2729 |
|
| 2730 |
if (preg_match_all($pattern, $html, $matches) <= 1) { |
| 2731 |
return $html; |
| 2732 |
} |
| 2733 |
|
| 2734 |
$all_tags = $matches[0]; |
| 2735 |
|
| 2736 |
$otto_tag = null; |
| 2737 |
foreach ($all_tags as $tag) { |
| 2738 |
if (stripos($tag, 'data-otto') !== false) { |
| 2739 |
$otto_tag = $tag; |
| 2740 |
break; |
| 2741 |
} |
| 2742 |
} |
| 2743 |
|
| 2744 |
$keeper = $otto_tag ?: $all_tags[0]; |
| 2745 |
|
| 2746 |
$first_replaced = false; |
| 2747 |
$html = preg_replace_callback($pattern, function ($m) use ($keeper, &$first_replaced) { |
| 2748 |
if (!$first_replaced) { |
| 2749 |
$first_replaced = true; |
| 2750 |
return $keeper; |
| 2751 |
} |
| 2752 |
return ''; |
| 2753 |
}, $html); |
| 2754 |
|
| 2755 |
return $html; |
| 2756 |
} |
| 2757 |
|
| 2758 |
/** |
| 2759 |
* Deduplicate JSON-LD schema blocks. |
| 2760 |
* |
| 2761 |
* When OTTO and a third-party SEO plugin both inject <script type="application/ld+json"> |
| 2762 |
* blocks, keep OTTO's version for any @type that appears in both. |
| 2763 |
* Third-party blocks whose @type is not covered by OTTO are preserved. |
| 2764 |
* |
| 2765 |
* @param string $html Full HTML. |
| 2766 |
* @return string |
| 2767 |
*/ |
| 2768 |
private function deduplicate_schema_tags($html) { |
| 2769 |
// Find all JSON-LD script blocks |
| 2770 |
$pattern = '/<script(\s[^>]*)type\s*=\s*(["\'])application\/ld\+json\2[^>]*>\s*([\s\S]*?)<\/script>/i'; |
| 2771 |
if (preg_match_all($pattern, $html, $matches, PREG_SET_ORDER) <= 1) { |
| 2772 |
return $html; |
| 2773 |
} |
| 2774 |
|
| 2775 |
$otto_by_type = []; // @type => decoded JSON object |
| 2776 |
$third_by_type = []; // @type => decoded JSON object |
| 2777 |
$otto_graph = []; // entries from OTTO @graph blocks |
| 2778 |
$third_graph = []; // entries from third-party @graph blocks |
| 2779 |
|
| 2780 |
foreach ($matches as $m) { |
| 2781 |
$attrs = $m[1]; |
| 2782 |
$json_str = $m[3]; |
| 2783 |
$decoded = json_decode($json_str, true); |
| 2784 |
if (json_last_error() !== JSON_ERROR_NONE || !is_array($decoded)) { |
| 2785 |
continue; // skip unparseable blocks — leave them in place |
| 2786 |
} |
| 2787 |
$is_otto = stripos($attrs, 'data-otto') !== false; |
| 2788 |
|
| 2789 |
if (isset($decoded['@graph']) && is_array($decoded['@graph'])) { |
| 2790 |
foreach ($decoded['@graph'] as $entry) { |
| 2791 |
if (!isset($entry['@type'])) continue; |
| 2792 |
// JSON-LD allows @type to be a string OR an array of strings. |
| 2793 |
// Rank Math routinely emits multi-typed entries (e.g. ["Person", "Organization"]). |
| 2794 |
// Using an array as an offset throws a fatal on PHP 8+, so normalize to scalar. |
| 2795 |
$type = is_array($entry['@type']) |
| 2796 |
? (string) reset($entry['@type']) |
| 2797 |
: (string) $entry['@type']; |
| 2798 |
if ($type === '') continue; |
| 2799 |
if ($is_otto) { |
| 2800 |
$otto_graph[$type] = $entry; |
| 2801 |
} else { |
| 2802 |
$third_graph[$type] = $entry; |
| 2803 |
} |
| 2804 |
} |
| 2805 |
} elseif (isset($decoded['@type'])) { |
| 2806 |
$type = is_array($decoded['@type']) |
| 2807 |
? (string) reset($decoded['@type']) |
| 2808 |
: (string) $decoded['@type']; |
| 2809 |
if ($type === '') continue; |
| 2810 |
if ($is_otto) { |
| 2811 |
$otto_by_type[$type] = $decoded; |
| 2812 |
} else { |
| 2813 |
$third_by_type[$type] = $decoded; |
| 2814 |
} |
| 2815 |
} |
| 2816 |
} |
| 2817 |
|
| 2818 |
// If OTTO provided no schema at all, nothing to deduplicate |
| 2819 |
if (empty($otto_by_type) && empty($otto_graph)) { |
| 2820 |
return $html; |
| 2821 |
} |
| 2822 |
|
| 2823 |
// Remove all JSON-LD blocks from HTML |
| 2824 |
$html = preg_replace($pattern, '', $html); |
| 2825 |
|
| 2826 |
// Re-insert flat (non-@graph) blocks: OTTO wins for matching @type |
| 2827 |
$kept = array_merge($otto_by_type, array_diff_key($third_by_type, $otto_by_type)); |
| 2828 |
$rebuilt = ''; |
| 2829 |
foreach ($kept as $decoded) { |
| 2830 |
$rebuilt .= '<script type="application/ld+json" data-otto="true">' . |
| 2831 |
wp_json_encode($decoded) . "</script>\n"; |
| 2832 |
} |
| 2833 |
|
| 2834 |
// Re-insert merged @graph block (if any entries exist) |
| 2835 |
$merged_graph = array_merge($third_graph, $otto_graph); // OTTO wins on duplicate @type |
| 2836 |
if (!empty($merged_graph)) { |
| 2837 |
$graph_obj = ['@context' => 'https://schema.org', '@graph' => array_values($merged_graph)]; |
| 2838 |
$rebuilt .= '<script type="application/ld+json" data-otto="true">' . |
| 2839 |
wp_json_encode($graph_obj) . "</script>\n"; |
| 2840 |
} |
| 2841 |
|
| 2842 |
// Re-inject before </head> |
| 2843 |
if (!empty($rebuilt)) { |
| 2844 |
$html = preg_replace_callback('/(<\/head>)/i', function ($m) use ($rebuilt) { |
| 2845 |
return $rebuilt . $m[1]; |
| 2846 |
}, $html, 1); |
| 2847 |
} |
| 2848 |
|
| 2849 |
return $html; |
| 2850 |
} |
| 2851 |
|
| 2852 |
# function to detect if current page is an AMP page |
| 2853 |
function is_amp_page(){ |
| 2854 |
|
| 2855 |
# Check if URL path contains /amp/ |
| 2856 |
$current_url = $_SERVER['REQUEST_URI'] ?? ''; |
| 2857 |
if (strpos($current_url, '/amp/') !== false) { |
| 2858 |
return true; |
| 2859 |
} |
| 2860 |
|
| 2861 |
# Check if URL ends with /amp |
| 2862 |
if (preg_match('/\/amp\/?$/', $current_url)) { |
| 2863 |
return true; |
| 2864 |
} |
| 2865 |
|
| 2866 |
# Check if amp=1 query parameter is present |
| 2867 |
if (isset($_GET['amp']) && $_GET['amp'] == '1') { |
| 2868 |
return true; |
| 2869 |
} |
| 2870 |
|
| 2871 |
# Check for other common AMP query parameters |
| 2872 |
if (isset($_GET['amp']) && !empty($_GET['amp'])) { |
| 2873 |
return true; |
| 2874 |
} |
| 2875 |
|
| 2876 |
return false; |
| 2877 |
} |
| 2878 |
|
| 2879 |
# this function insterts header html to the dom |
| 2880 |
function insert_header_html($data){ |
| 2881 |
|
| 2882 |
# check that we have the header html |
| 2883 |
if(empty($data['header_html_insertion'])){ |
| 2884 |
# |
| 2885 |
return; |
| 2886 |
} |
| 2887 |
|
| 2888 |
# append the/ html at the start of the header |
| 2889 |
$head = $this->dom->find('head', 0); |
| 2890 |
|
| 2891 |
if ($head) { |
| 2892 |
|
| 2893 |
# Check if this is an AMP page - if so, don't add metasync_optimized attribute |
| 2894 |
$is_amp_page = $this->is_amp_page(); |
| 2895 |
|
| 2896 |
# Append the new HTML at the start of the <head> tag |
| 2897 |
# For AMP pages: use clean <head> tag without metasync_optimized attribute |
| 2898 |
# For non-AMP pages: add metasync_optimized attribute to <head> tag |
| 2899 |
if ($is_amp_page) { |
| 2900 |
$head->outertext = '<head>' .$data['header_html_insertion']. $head->innertext . '</head>'; |
| 2901 |
} else { |
| 2902 |
$head->outertext = '<head metasync_optimized>' .$data['header_html_insertion']. $head->innertext . '</head>'; |
| 2903 |
} |
| 2904 |
|
| 2905 |
} |
| 2906 |
|
| 2907 |
# save and reload DOM |
| 2908 |
$this->save_reload(); |
| 2909 |
} |
| 2910 |
|
| 2911 |
# function to forcefully remove metasync_optimized attribute from head on AMP pages |
| 2912 |
function cleanup_amp_metasync_attribute(){ |
| 2913 |
|
| 2914 |
# Only proceed if this is an AMP page |
| 2915 |
if (!$this->is_amp_page()) { |
| 2916 |
return; |
| 2917 |
} |
| 2918 |
|
| 2919 |
# Find the head tag |
| 2920 |
$head = $this->dom->find('head', 0); |
| 2921 |
|
| 2922 |
if (!$head) { |
| 2923 |
return; |
| 2924 |
} |
| 2925 |
|
| 2926 |
# Check if head has metasync_optimized attribute |
| 2927 |
$head_html = $head->outertext; |
| 2928 |
|
| 2929 |
# If metasync_optimized attribute is found, remove it |
| 2930 |
if (strpos($head_html, 'metasync_optimized') !== false) { |
| 2931 |
|
| 2932 |
# Remove the metasync_optimized attribute from the head tag |
| 2933 |
# This handles various formats: <head metasync_optimized>, <head metasync_optimized=""> etc. |
| 2934 |
$cleaned_head_html = preg_replace('/\s*metasync_optimized(?:="[^"]*")?/', '', $head_html); |
| 2935 |
|
| 2936 |
# Update the head element |
| 2937 |
$head->outertext = $cleaned_head_html; |
| 2938 |
|
| 2939 |
} |
| 2940 |
} |
| 2941 |
|
| 2942 |
/** |
| 2943 |
* PERFORMANCE OPTIMIZATION: Pre-cache commonly accessed DOM elements |
| 2944 |
* Reduces 8-10 full DOM traversals to 1 initial traversal |
| 2945 |
* Call this once after loading HTML, before processing |
| 2946 |
*/ |
| 2947 |
private function cache_elements() { |
| 2948 |
if (!$this->dom) { |
| 2949 |
return; |
| 2950 |
} |
| 2951 |
|
| 2952 |
# Cache all commonly accessed elements in one pass |
| 2953 |
$this->cached_elements = [ |
| 2954 |
'html' => $this->dom->find('html', 0), |
| 2955 |
'head' => $this->dom->find('head', 0), |
| 2956 |
'body' => $this->dom->find('body', 0), |
| 2957 |
'footer' => $this->dom->find('footer', 0), |
| 2958 |
'title' => $this->dom->find('title', 0), |
| 2959 |
'imgs' => $this->dom->find('img'), |
| 2960 |
'links' => $this->dom->find('a'), |
| 2961 |
'canonical' => $this->dom->find('link[rel="canonical"]', 0), |
| 2962 |
]; |
| 2963 |
} |
| 2964 |
|
| 2965 |
/** |
| 2966 |
* Get cached element by key, with fallback to DOM find |
| 2967 |
* @param string $key Element key from cache |
| 2968 |
* @param mixed $fallback Fallback value if not cached |
| 2969 |
* @return mixed Cached element or fallback |
| 2970 |
*/ |
| 2971 |
private function get_cached_element($key, $fallback = null) { |
| 2972 |
return $this->cached_elements[$key] ?? $fallback; |
| 2973 |
} |
| 2974 |
|
| 2975 |
# this function saves are reloads the dom for modifications to avoid conflict |
| 2976 |
function save_reload(){ |
| 2977 |
|
| 2978 |
# PERFORMANCE OPTIMIZATION: Skip reload if deferred |
| 2979 |
# This reduces multiple serialize/deserialize cycles to just one final reload |
| 2980 |
if ($this->deferred_reload) { |
| 2981 |
return; |
| 2982 |
} |
| 2983 |
|
| 2984 |
# Cleanup metasync_optimized attribute on AMP pages before saving |
| 2985 |
$this->cleanup_amp_metasync_attribute(); |
| 2986 |
|
| 2987 |
# DISABLED: Cache file creation temporarily disabled |
| 2988 |
|
| 2989 |
# if(file_put_contents($this->html_file, $this->dom)){ |
| 2990 |
|
| 2991 |
# load the modified file to the DOM |
| 2992 |
# $this->dom = new HtmlDocument($this->html_file ); |
| 2993 |
# } |
| 2994 |
|
| 2995 |
# this code is to be replaced in future |
| 2996 |
# reson for adding is to prevent caching logged in user pages |
| 2997 |
# why not just skip saving? it broke the DOM Library |
| 2998 |
# check user is logged in clear the file |
| 2999 |
|
| 3000 |
# if(is_user_logged_in()) { |
| 3001 |
# unlink($this->html_file); |
| 3002 |
# } |
| 3003 |
|
| 3004 |
# MEMORY-BASED RELOAD: Instead of file operations, reload DOM from current HTML string |
| 3005 |
# This prevents DOM breaking while avoiding cache file creation |
| 3006 |
if($this->dom){ |
| 3007 |
# Get current DOM as HTML string |
| 3008 |
$current_html = $this->dom->save(); |
| 3009 |
|
| 3010 |
# Reload DOM from the HTML string to refresh internal state |
| 3011 |
# This replaces the file save/reload cycle that SimpleHtmlDOM expects |
| 3012 |
$current_html = $this->sanitize_text_less_than($current_html); |
| 3013 |
$this->dom->load($current_html, true, false); |
| 3014 |
|
| 3015 |
# Force UTF-8 charset after reload to preserve emojis |
| 3016 |
$this->dom->_charset = 'UTF-8'; |
| 3017 |
$this->dom->_target_charset = 'UTF-8'; |
| 3018 |
} |
| 3019 |
|
| 3020 |
} |
| 3021 |
|
| 3022 |
/** |
| 3023 |
* Explicitly free the SimpleHtmlDom node tree and reclaim memory. |
| 3024 |
* |
| 3025 |
* SimpleHtmlDom nodes hold circular parent/child references, so simply |
| 3026 |
* dropping the document does not free the tree until request shutdown. |
| 3027 |
* HtmlNode::clear() breaks those cycles; we then drop the document, force a |
| 3028 |
* GC pass, and re-instantiate a fresh empty parser so the instance stays |
| 3029 |
* reusable for any subsequent route on the same request. |
| 3030 |
*/ |
| 3031 |
function free_dom(){ |
| 3032 |
if (isset($this->dom)) { |
| 3033 |
# Break the node tree's parent/child cycles so GC can reclaim it now. |
| 3034 |
if (isset($this->dom->root) && is_object($this->dom->root) |
| 3035 |
&& method_exists($this->dom->root, 'clear')) { |
| 3036 |
$this->dom->root->clear(); |
| 3037 |
} |
| 3038 |
unset($this->dom); |
| 3039 |
} |
| 3040 |
|
| 3041 |
# Fresh, empty parser — mirrors the constructor so the object is reusable. |
| 3042 |
$this->dom = new HtmlDocument(null, true, true, 'UTF-8', false); |
| 3043 |
|
| 3044 |
if (function_exists('gc_collect_cycles')) { |
| 3045 |
gc_collect_cycles(); |
| 3046 |
} |
| 3047 |
} |
| 3048 |
|
| 3049 |
/** |
| 3050 |
* Process HTML directly without HTTP request (for buffer approach) |
| 3051 |
* This is the FAST path - eliminates the internal wp_remote_get call |
| 3052 |
* |
| 3053 |
* CRITICAL: This method is called from the output buffer callback. |
| 3054 |
* If it fails, we must return false so the original HTML can be used. |
| 3055 |
* |
| 3056 |
* @param string $html The raw HTML captured from output buffer |
| 3057 |
* @param array $replacement_data OTTO suggestions/replacement data |
| 3058 |
* @return HtmlDocument|false Modified DOM or false on failure |
| 3059 |
* @since 2.6.0 |
| 3060 |
*/ |
| 3061 |
function process_html_directly($html, $replacement_data) { |
| 3062 |
try { |
| 3063 |
# Validate inputs |
| 3064 |
if (empty($html) || empty($replacement_data) || !is_array($replacement_data)) { |
| 3065 |
return false; |
| 3066 |
} |
| 3067 |
|
| 3068 |
# Validate HTML is actual HTML content |
| 3069 |
if (stripos($html, '<html') === false && stripos($html, '<!DOCTYPE') === false) { |
| 3070 |
return false; |
| 3071 |
} |
| 3072 |
|
| 3073 |
# Skip DOM processing on oversized documents to avoid fatal OOM. |
| 3074 |
# Returning false makes the buffer callback serve the original HTML |
| 3075 |
# unmodified — the page still renders, just without OTTO changes. |
| 3076 |
if (class_exists('Metasync_Otto_Render_Strategy') |
| 3077 |
&& !Metasync_Otto_Render_Strategy::is_document_processable(strlen($html)) |
| 3078 |
) { |
| 3079 |
Metasync_Otto_Render_Strategy::log_oversized_skip('process_html_directly', strlen($html)); |
| 3080 |
return false; |
| 3081 |
} |
| 3082 |
|
| 3083 |
# Remove XML declaration if present |
| 3084 |
$html = preg_replace('/<\?xml[^?]*\?>\s*/i', '', $html); |
| 3085 |
|
| 3086 |
# / Save ALL original <style> blocks before DOM processing. |
| 3087 |
$original_style_blocks = $this->capture_style_blocks($html); |
| 3088 |
|
| 3089 |
# Shield entity-encoded srcdoc values (lazy YouTube/video facades) |
| 3090 |
# so SimpleHtmlDom can't decode them and hoist their global <style> into <head>. |
| 3091 |
$srcdoc_store = array(); |
| 3092 |
$html = $this->protect_srcdoc_attributes($html, $srcdoc_store); |
| 3093 |
|
| 3094 |
# Escape bare < in text content before DOM parsing |
| 3095 |
$html = $this->sanitize_text_less_than($html); |
| 3096 |
|
| 3097 |
# Fix malformed self-closing non-void tags (e.g. <ul/ class="...">) |
| 3098 |
$html = $this->fix_malformed_self_closing_tags($html); |
| 3099 |
|
| 3100 |
# Quote unquoted attribute values (e.g. data-autoplay-viewport=80%) — |
| 3101 |
# SimpleHtmlDom swallows every following attribute into such a value. |
| 3102 |
$html = $this->quote_unquoted_attribute_values($html); |
| 3103 |
|
| 3104 |
# Load HTML into DOM |
| 3105 |
$this->dom->load($html, true, false); |
| 3106 |
|
| 3107 |
# Force UTF-8 charset to preserve emojis and special characters |
| 3108 |
$this->dom->_charset = 'UTF-8'; |
| 3109 |
$this->dom->_target_charset = 'UTF-8'; |
| 3110 |
|
| 3111 |
# PERFORMANCE OPTIMIZATION: Pre-cache commonly accessed DOM elements |
| 3112 |
# This eliminates 8-10 full DOM traversals per page |
| 3113 |
$this->cache_elements(); |
| 3114 |
|
| 3115 |
# PERFORMANCE OPTIMIZATION: Enable deferred reload to skip intermediate reloads |
| 3116 |
# This reduces 6-10 DOM serialize/deserialize cycles to just 1 final reload |
| 3117 |
$this->deferred_reload = true; |
| 3118 |
|
| 3119 |
# Apply blocking flags if available (for SEO plugin coordination) |
| 3120 |
if (!empty($replacement_data['_otto_blocking'])) { |
| 3121 |
$block_title = $replacement_data['_otto_blocking']['block_title'] ?? false; |
| 3122 |
$block_description_tags = $replacement_data['_otto_blocking']['block_description_tags'] ?? []; |
| 3123 |
|
| 3124 |
# Remove SEO plugin meta tags if OTTO is providing them |
| 3125 |
if ($block_title || !empty($block_description_tags)) { |
| 3126 |
$this->remove_conflicting_seo_tags($block_title, $block_description_tags); |
| 3127 |
} |
| 3128 |
} |
| 3129 |
|
| 3130 |
# Apply all OTTO modifications (same as handle_route_html but without HTTP fetch) |
| 3131 |
|
| 3132 |
# COMPATIBILITY FIX: Transform 'value' to 'recommended_value' if needed |
| 3133 |
# Some API versions return 'value' instead of 'recommended_value' |
| 3134 |
if (!empty($replacement_data['header_replacements'])) { |
| 3135 |
foreach ($replacement_data['header_replacements'] as &$item) { |
| 3136 |
if (isset($item['value']) && !isset($item['recommended_value'])) { |
| 3137 |
$item['recommended_value'] = $item['value']; |
| 3138 |
} |
| 3139 |
} |
| 3140 |
unset($item); // Break reference |
| 3141 |
} |
| 3142 |
|
| 3143 |
# 1. Header HTML insertion |
| 3144 |
$this->insert_header_html($replacement_data); |
| 3145 |
|
| 3146 |
# 2. Header replacements (title, meta, canonical) |
| 3147 |
$this->do_header_replacements($replacement_data); |
| 3148 |
|
| 3149 |
# 3. Body replacements (top, bottom, substitutions) |
| 3150 |
$this->do_body_replacements($replacement_data); |
| 3151 |
|
| 3152 |
# 4. Footer HTML insertion |
| 3153 |
$this->do_footer_html_insertion($replacement_data); |
| 3154 |
|
| 3155 |
# 5. Final cleanup for AMP pages |
| 3156 |
$this->cleanup_amp_metasync_attribute(); |
| 3157 |
|
| 3158 |
# CRITICAL FIX: Clear cached elements and force DOM to refresh |
| 3159 |
$this->deferred_reload = false; |
| 3160 |
$this->cached_elements = []; // Clear our custom cache |
| 3161 |
|
| 3162 |
# Clear SimpleHtmlDom's internal cache |
| 3163 |
if (method_exists($this->dom, 'clear')) { |
| 3164 |
$this->dom->clear(); |
| 3165 |
} |
| 3166 |
|
| 3167 |
|
| 3168 |
# Try getting HTML via root element instead of save() |
| 3169 |
$root = $this->dom->root; |
| 3170 |
if ($root && isset($root->outertext)) { |
| 3171 |
$result_html = $root->outertext; |
| 3172 |
} else { |
| 3173 |
# Fallback to save() method |
| 3174 |
$result_html = $this->dom->save(); |
| 3175 |
} |
| 3176 |
|
| 3177 |
# DEBUG: Check if DOM changes persisted |
| 3178 |
if (preg_match('/<title[^>]*>(.*?)<\/title>/is', $result_html, $matches)) { |
| 3179 |
} |
| 3180 |
|
| 3181 |
# Apply header replacements manually via string replacement |
| 3182 |
if (!empty($replacement_data['header_replacements'])) { |
| 3183 |
|
| 3184 |
foreach ($replacement_data['header_replacements'] as $idx => $item) { |
| 3185 |
$type = $item['type'] ?? ''; |
| 3186 |
$value = $item['recommended_value'] ?? $item['value'] ?? ''; |
| 3187 |
|
| 3188 |
|
| 3189 |
if (empty($value)) { |
| 3190 |
continue; |
| 3191 |
} |
| 3192 |
|
| 3193 |
if ($type === 'title') { |
| 3194 |
# Replace title tag if present; otherwise insert (e.g. when Yoast was blocked and no <title> was output) |
| 3195 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 3196 |
$title_tag = '<title>' . $new_value . '</title>'; |
| 3197 |
$replaced = preg_replace_callback('/<title[^>]*>.*?<\/title>/is', function ($m) use ($title_tag) { |
| 3198 |
return $title_tag; |
| 3199 |
}, $result_html, 1); |
| 3200 |
if ($replaced === $result_html && strpos($result_html, '<title') === false) { |
| 3201 |
# No <title> in document (common when Yoast is blocked) — insert after <head> |
| 3202 |
$result_html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($title_tag) { |
| 3203 |
return $m[1] . "\n" . $title_tag; |
| 3204 |
}, $result_html, 1); |
| 3205 |
} else { |
| 3206 |
$result_html = $replaced; |
| 3207 |
} |
| 3208 |
} elseif ($type === 'meta') { |
| 3209 |
$name = $item['name'] ?? ''; |
| 3210 |
$property = $item['property'] ?? ''; |
| 3211 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 3212 |
|
| 3213 |
if (!empty($name)) { |
| 3214 |
# MEMORY OPTIMIZED: Count meta tags without storing all matches |
| 3215 |
# preg_match_all requires $matches, so we pass it but unset immediately |
| 3216 |
# Use \s*=\s* to match attributes with or without spaces around = |
| 3217 |
$before_count = preg_match_all('/<meta[^>]+name\s*=\s*["\']' . preg_quote($name, '/') . '["\'][^>]*>/i', $result_html, $before_matches); |
| 3218 |
|
| 3219 |
# Free memory immediately after use |
| 3220 |
unset($before_matches); |
| 3221 |
|
| 3222 |
# IMPROVED FIX: Remove ALL existing meta tags with this name |
| 3223 |
# This pattern matches ANY meta tag with name="X" regardless of: |
| 3224 |
# - Attribute order (name before/after content) |
| 3225 |
# - Additional attributes (id, class, data-*, etc.) |
| 3226 |
# - Quote style (single or double quotes) |
| 3227 |
# - Whitespace variations |
| 3228 |
# The pattern uses [^>]* to match ANY characters until the closing > |
| 3229 |
$removed_count = preg_replace_callback( |
| 3230 |
'/<meta\s+[^>]*name\s*=\s*["\']' . preg_quote($name, '/') . '["\'][^>]*>/i', |
| 3231 |
function($match) { |
| 3232 |
return ''; // Remove the tag |
| 3233 |
}, |
| 3234 |
$result_html, |
| 3235 |
-1, // Remove all occurrences |
| 3236 |
$total_removed |
| 3237 |
); |
| 3238 |
|
| 3239 |
# Update the HTML with removed tags |
| 3240 |
if ($total_removed > 0) { |
| 3241 |
$result_html = $removed_count; |
| 3242 |
} |
| 3243 |
|
| 3244 |
|
| 3245 |
# MEMORY OPTIMIZED: Verify removal - count again but free memory immediately |
| 3246 |
$after_count = preg_match_all('/<meta[^>]+name\s*=\s*["\']' . preg_quote($name, '/') . '["\'][^>]*>/i', $result_html, $after_matches); |
| 3247 |
unset($after_matches); |
| 3248 |
|
| 3249 |
# Now insert ONE new meta tag at the TOP of <head> |
| 3250 |
$replacement = '<meta name="' . htmlspecialchars($name, ENT_QUOTES, 'UTF-8') . '" content="' . $new_value . '" data-otto="true">'; |
| 3251 |
$result_html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($replacement) { |
| 3252 |
return $m[1] . "\n" . $replacement; |
| 3253 |
}, $result_html, 1); |
| 3254 |
} elseif (!empty($property)) { |
| 3255 |
# Replace meta property tag |
| 3256 |
$pattern = '/<meta\s+property\s*=\s*["\']' . preg_quote($property, '/') . '["\']\s+content\s*=\s*["\'][^"\']*["\']\s*\/?>/i'; |
| 3257 |
$replacement = '<meta property="' . htmlspecialchars($property, ENT_QUOTES, 'UTF-8') . '" content="' . $new_value . '">'; |
| 3258 |
|
| 3259 |
if (preg_match($pattern, $result_html)) { |
| 3260 |
$result_html = preg_replace_callback($pattern, function ($m) use ($replacement) { |
| 3261 |
return $replacement; |
| 3262 |
}, $result_html, 1); |
| 3263 |
} else { |
| 3264 |
# Meta tag doesn't exist, insert it in head |
| 3265 |
$result_html = preg_replace_callback('/(<head[^>]*>)/i', function ($m) use ($replacement) { |
| 3266 |
return $m[1] . "\n" . $replacement; |
| 3267 |
}, $result_html, 1); |
| 3268 |
} |
| 3269 |
} |
| 3270 |
} elseif ($type === 'h1' || $type === 'heading') { |
| 3271 |
# Replace first H1 tag, preserving attributes |
| 3272 |
$new_value = htmlspecialchars($value, ENT_QUOTES, 'UTF-8'); |
| 3273 |
$result_html = preg_replace_callback( |
| 3274 |
'/<h1([^>]*)>.*?<\/h1>/is', |
| 3275 |
function ($m) use ($new_value) { |
| 3276 |
return '<h1' . $m[1] . '>' . $new_value . '</h1>'; |
| 3277 |
}, |
| 3278 |
$result_html, |
| 3279 |
1 |
| 3280 |
); |
| 3281 |
} |
| 3282 |
} |
| 3283 |
} |
| 3284 |
|
| 3285 |
# Apply header HTML insertion (for schema, etc.) — only if DOM insertion didn't already apply it |
| 3286 |
if (!empty($replacement_data['header_html_insertion'])) { |
| 3287 |
$header_html_check = trim($replacement_data['header_html_insertion']); |
| 3288 |
if (strpos($result_html, $header_html_check) === false) { |
| 3289 |
$header_html_insertion = preg_replace( |
| 3290 |
'/<script(\s[^>]*)type\s*=\s*(["\'])application\/ld\+json\2/i', |
| 3291 |
'<script$1type=$2application/ld+json$2 data-otto="true"', |
| 3292 |
$replacement_data['header_html_insertion'] |
| 3293 |
); |
| 3294 |
$header_html = str_replace(array('\\', '$'), array('\\\\', '\\$'), $header_html_insertion); |
| 3295 |
# Insert before </head> |
| 3296 |
$result_html = preg_replace('/(<\/head>)/i', $header_html . "\n" . '$1', $result_html, 1); |
| 3297 |
} |
| 3298 |
} |
| 3299 |
|
| 3300 |
# When Otto block has a title: ensure only ONE <title> (remove non-Otto, then keep first only) |
| 3301 |
if (!empty($replacement_data['header_html_insertion']) && stripos($replacement_data['header_html_insertion'], '<title') !== false) { |
| 3302 |
$result_html = preg_replace( |
| 3303 |
'/<title(?![^>]*data-otto-pixel\s*=\s*["\']dynamic-seo["\'])[^>]*>.*?<\/title>\s*/is', |
| 3304 |
'', |
| 3305 |
$result_html |
| 3306 |
); |
| 3307 |
$first = true; |
| 3308 |
$result_html = preg_replace_callback( |
| 3309 |
'/<title[^>]*>.*?<\/title>\s*/is', |
| 3310 |
function ($m) use (&$first) { |
| 3311 |
if ($first) { |
| 3312 |
$first = false; |
| 3313 |
return $m[0]; |
| 3314 |
} |
| 3315 |
return ''; |
| 3316 |
}, |
| 3317 |
$result_html |
| 3318 |
); |
| 3319 |
} |
| 3320 |
|
| 3321 |
# Apply body top HTML insertion — only if DOM insertion didn't already apply it |
| 3322 |
if (!empty($replacement_data['body_top_html_insertion'])) { |
| 3323 |
$body_top_check = trim($replacement_data['body_top_html_insertion']); |
| 3324 |
if (strpos($result_html, $body_top_check) === false) { |
| 3325 |
$body_top_html = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['body_top_html_insertion']); |
| 3326 |
# Insert after <body> |
| 3327 |
$result_html = preg_replace('/(<body[^>]*>)/i', '$1' . "\n" . $body_top_html, $result_html, 1); |
| 3328 |
} |
| 3329 |
} |
| 3330 |
|
| 3331 |
# Apply body bottom HTML insertion — only if DOM insertion didn't already apply it |
| 3332 |
if (!empty($replacement_data['body_bottom_html_insertion'])) { |
| 3333 |
$body_bottom_check = trim($replacement_data['body_bottom_html_insertion']); |
| 3334 |
if (strpos($result_html, $body_bottom_check) === false) { |
| 3335 |
$body_bottom_html = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['body_bottom_html_insertion']); |
| 3336 |
# Insert before </body> |
| 3337 |
$result_html = preg_replace('/(<\/body>)/i', $body_bottom_html . "\n" . '$1', $result_html, 1); |
| 3338 |
} |
| 3339 |
} |
| 3340 |
|
| 3341 |
# Apply footer HTML insertion |
| 3342 |
if (!empty($replacement_data['footer_html_insertion'])) { |
| 3343 |
$footer_html = str_replace(array('\\', '$'), array('\\\\', '\\$'), $replacement_data['footer_html_insertion']); |
| 3344 |
# Insert before </html> |
| 3345 |
$result_html = preg_replace('/(<\/html>)/i', $footer_html . "\n" . '$1', $result_html, 1); |
| 3346 |
} |
| 3347 |
|
| 3348 |
# CRITICAL FIX: Apply image alt text manually via string replacement |
| 3349 |
# DOM changes don't persist, must use string replacement |
| 3350 |
$result_html = $this->otto_guard_html($this->apply_image_alt_text_via_string($result_html, $replacement_data), $result_html, 'apply_image_alt_text_via_string'); |
| 3351 |
|
| 3352 |
# String-based heading fallback for body_substitutions |
| 3353 |
# DOM changes via SimpleHtmlDom don't persist on Divi/page-builder sites |
| 3354 |
if (!empty($replacement_data['body_substitutions']['headings']) && is_array($replacement_data['body_substitutions']['headings'])) { |
| 3355 |
foreach ($replacement_data['body_substitutions']['headings'] as $heading) { |
| 3356 |
if (empty($heading['type']) || empty($heading['current_value']) || empty($heading['recommended_value'])) { |
| 3357 |
continue; |
| 3358 |
} |
| 3359 |
|
| 3360 |
$heading_type = preg_quote($heading['type'], '/'); |
| 3361 |
$current_value = trim(preg_replace('/\s+/', ' ', html_entity_decode($heading['current_value'], ENT_QUOTES, 'UTF-8'))); |
| 3362 |
$recommended_value = htmlspecialchars($heading['recommended_value'], ENT_QUOTES, 'UTF-8'); |
| 3363 |
|
| 3364 |
$result_html = preg_replace_callback( |
| 3365 |
'/(<' . $heading_type . '(?:\s[^>]*)?>)(.*?)(<\/' . $heading_type . '>)/is', |
| 3366 |
function ($m) use ($current_value, $recommended_value) { |
| 3367 |
$inner_text = trim(preg_replace('/\s+/', ' ', html_entity_decode(strip_tags($m[2]), ENT_QUOTES, 'UTF-8'))); |
| 3368 |
if ($inner_text !== $current_value) { |
| 3369 |
return $m[0]; |
| 3370 |
} |
| 3371 |
# Keep the author's inline formatting; skip when it can't |
| 3372 |
# be preserved rather than flattening the heading. |
| 3373 |
$new_inner = self::build_heading_inner_html($m[2], $recommended_value); |
| 3374 |
if ($new_inner === null) { |
| 3375 |
return $m[0]; |
| 3376 |
} |
| 3377 |
return $m[1] . $new_inner . $m[3]; |
| 3378 |
}, |
| 3379 |
$result_html, |
| 3380 |
-1 |
| 3381 |
); |
| 3382 |
} |
| 3383 |
} |
| 3384 |
|
| 3385 |
# DEBUG: Check what we're returning |
| 3386 |
if (preg_match('/<title[^>]*>(.*?)<\/title>/is', $result_html, $matches)) { |
| 3387 |
} |
| 3388 |
|
| 3389 |
# FINAL VERIFICATION: Count meta descriptions in returned HTML |
| 3390 |
$final_meta_count = preg_match_all('/<meta[^>]+name=["\']description["\'][^>]*>/i', $result_html, $final_meta_matches); |
| 3391 |
if ($final_meta_count > 0) { |
| 3392 |
foreach ($final_meta_matches[0] as $idx => $meta) { |
| 3393 |
} |
| 3394 |
} |
| 3395 |
|
| 3396 |
if ($final_meta_count > 1) { |
| 3397 |
} |
| 3398 |
|
| 3399 |
# AGGRESSIVE DUPLICATE REMOVAL: Remove any meta description without data-otto marker |
| 3400 |
# Only runs when OTTO has actually inserted its own description (data-otto marker present) |
| 3401 |
# This prevents stripping Yoast/plugin descriptions when OTTO has no description to replace |
| 3402 |
|
| 3403 |
$removal_count = 0; |
| 3404 |
$otto_has_description = (bool) preg_match('/<meta[^>]*name\s*=\s*["\']description["\'][^>]*data-otto[^>]*>/i', $result_html); |
| 3405 |
if ($otto_has_description) { |
| 3406 |
$result_html = preg_replace_callback( |
| 3407 |
'/<meta\s+([^>]*name\s*=\s*["\']description["\'][^>]*)>/i', |
| 3408 |
function($match) use (&$removal_count) { |
| 3409 |
# Keep only if it has data-otto="true" |
| 3410 |
if (stripos($match[1], 'data-otto') !== false) { |
| 3411 |
return $match[0]; // Keep OTTO's meta tag |
| 3412 |
} |
| 3413 |
# Remove any other meta description |
| 3414 |
$removal_count++; |
| 3415 |
return ''; |
| 3416 |
}, |
| 3417 |
$result_html |
| 3418 |
); |
| 3419 |
} |
| 3420 |
|
| 3421 |
# Apply OTTO's canonical via string replacement (DOM edit is clobbered by the insert_header_html head-freeze; see apply_canonical_via_string). |
| 3422 |
$result_html = $this->apply_canonical_via_string($result_html, $replacement_data); |
| 3423 |
|
| 3424 |
# DEDUPLICATION: Remove duplicate <title>, meta description, OG, Twitter tags, canonical, and JSON-LD schema |
| 3425 |
$result_html = $this->otto_guard_html($this->deduplicate_title_tags($result_html), $result_html, 'deduplicate_title_tags'); |
| 3426 |
$result_html = $this->otto_guard_html($this->deduplicate_description_tags($result_html), $result_html, 'deduplicate_description_tags'); |
| 3427 |
$result_html = $this->otto_guard_html($this->deduplicate_og_twitter_tags($result_html), $result_html, 'deduplicate_og_twitter_tags'); |
| 3428 |
$result_html = $this->otto_guard_html($this->apply_metabox_og_precedence($result_html), $result_html, 'apply_metabox_og_precedence'); |
| 3429 |
$result_html = $this->otto_guard_html($this->apply_custom_seo_precedence($result_html), $result_html, 'apply_custom_seo_precedence'); |
| 3430 |
$result_html = $this->otto_guard_html($this->deduplicate_schema_tags($result_html), $result_html, 'deduplicate_schema_tags'); |
| 3431 |
$result_html = $this->otto_guard_html($this->deduplicate_canonical_tags($result_html), $result_html, 'deduplicate_canonical_tags'); |
| 3432 |
|
| 3433 |
# MEMORY OPTIMIZED: Free all large objects and arrays before returning |
| 3434 |
# This ensures memory is released immediately, especially important for high-traffic sites |
| 3435 |
unset($final_meta_matches, $matches); |
| 3436 |
|
| 3437 |
# Explicitly free the SimpleHtmlDom node tree now that the |
| 3438 |
# result has been serialized to a string. The bundled HtmlDocument has |
| 3439 |
# no clear() method, so the previous method_exists() guard was a no-op |
| 3440 |
# and the (circular-referenced) node tree lingered until object |
| 3441 |
# destruction. free_dom() breaks those references and reclaims memory. |
| 3442 |
$this->free_dom(); |
| 3443 |
|
| 3444 |
# Clear element cache array |
| 3445 |
$this->cached_elements = []; |
| 3446 |
|
| 3447 |
# Ensure metasync_optimized attribute on <head> (post-serialization so dom->clear() can't wipe it) |
| 3448 |
if (!$this->is_amp_page() && strpos($result_html, 'metasync_optimized') === false) { |
| 3449 |
$result_html = preg_replace('/<head(\s|>)/i', '<head metasync_optimized$1', $result_html, 1); |
| 3450 |
} |
| 3451 |
|
| 3452 |
$result_html = $this->otto_guard_html($this->restore_case_sensitive_attributes($result_html), $result_html, 'restore_case_sensitive_attributes'); |
| 3453 |
|
| 3454 |
# / Re-inject any <style> blocks lost during processing. |
| 3455 |
$result_html = $this->otto_guard_html($this->restore_lost_style_blocks($result_html, $original_style_blocks), $result_html, 'restore_lost_style_blocks'); |
| 3456 |
|
| 3457 |
# Do NOT renumber Divi classes on the buffer/in-place path. |
| 3458 |
# fix_divi_class_renumbering() exists to undo the index offset that |
| 3459 |
# Divi's shortcode framework introduces ONLY during the internal |
| 3460 |
# wp_remote_get fetch (handle_route_html), where the Theme Builder |
| 3461 |
# header is counted twice. process_html_directly() processes the |
| 3462 |
# page Divi already rendered normally — its module numbering and |
| 3463 |
# Divi's own CSS are already in sync. Renumbering here desynced every |
| 3464 |
# section/row/column/text/button index (e.g. et_pb_section_10 → _0), |
| 3465 |
# colliding page modules with header/global modules of the same |
| 3466 |
# number and collapsing layouts (Divi hero/slider) on Divi sites. |
| 3467 |
# The renumber stays only on the HTTP-fetch path (handle_route_html). |
| 3468 |
|
| 3469 |
# Clean OTTO internal fetch params from HTML output |
| 3470 |
$result_html = $this->otto_guard_html($this->clean_otto_fetch_params($result_html), $result_html, 'clean_otto_fetch_params'); |
| 3471 |
|
| 3472 |
# Remove any lazy-video facade <style> hoisted into the page |
| 3473 |
# (runs while srcdoc is still tokenized, so the iframe's own copy is safe). |
| 3474 |
$result_html = $this->otto_guard_html($this->strip_hoisted_facade_styles($result_html), $result_html, 'strip_hoisted_facade_styles'); |
| 3475 |
|
| 3476 |
# Restore the original encoded srcdoc values (undo tokenization). |
| 3477 |
$result_html = $this->otto_guard_html($this->restore_srcdoc_attributes($result_html, $srcdoc_store), $result_html, 'restore_srcdoc_attributes'); |
| 3478 |
|
| 3479 |
# Keep <meta charset> within the first 1024 bytes so external CSS |
| 3480 |
# (e.g. checkmark content:"✓") doesn't mojibake on cached responses. |
| 3481 |
$result_html = $this->otto_guard_html($this->ensure_early_charset_meta($result_html), $result_html, 'ensure_early_charset_meta'); |
| 3482 |
|
| 3483 |
# undo double-encoded numeric/hex character references produced by the |
| 3484 |
# bundled simplehtmldom serializer from attributes like data-x-icon-s="". |
| 3485 |
$result_html = $this->otto_guard_html($this->repair_double_encoded_entities($result_html), $result_html, 'repair_double_encoded_entities'); |
| 3486 |
|
| 3487 |
# Remove any lazy-video facade <style> hoisted into the page |
| 3488 |
# (runs while srcdoc is still tokenized, so the iframe's own copy is safe). |
| 3489 |
$result_html = $this->strip_hoisted_facade_styles($result_html); |
| 3490 |
|
| 3491 |
# Restore the original encoded srcdoc values (undo tokenization). |
| 3492 |
$result_html = $this->restore_srcdoc_attributes($result_html, $srcdoc_store); |
| 3493 |
|
| 3494 |
# Keep <meta charset> within the first 1024 bytes so external CSS |
| 3495 |
# (e.g. checkmark content:"✓") doesn't mojibake on cached responses. |
| 3496 |
$result_html = $this->ensure_early_charset_meta($result_html); |
| 3497 |
|
| 3498 |
# undo double-encoded numeric/hex character references produced by the |
| 3499 |
# bundled simplehtmldom serializer from attributes like data-x-icon-s="". |
| 3500 |
$result_html = $this->repair_double_encoded_entities($result_html); |
| 3501 |
|
| 3502 |
return $result_html; |
| 3503 |
|
| 3504 |
} catch (Exception $e) { |
| 3505 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': Exception in process_html_directly - ' . $e->getMessage()); |
| 3506 |
return false; |
| 3507 |
} catch (Error $e) { |
| 3508 |
error_log('MetaSync ' . Metasync::get_whitelabel_otto_name() . ': Error in process_html_directly - ' . $e->getMessage()); |
| 3509 |
return false; |
| 3510 |
} |
| 3511 |
} |
| 3512 |
|
| 3513 |
/** |
| 3514 |
* Remove conflicting SEO plugin meta tags from DOM |
| 3515 |
* Called when OTTO is providing its own title/description |
| 3516 |
* |
| 3517 |
* @param bool $remove_title Remove title-related tags |
| 3518 |
* @param array|bool $remove_description_tags Array of specific description tag selectors to remove, or false/empty array for none |
| 3519 |
* @since 2.6.0 |
| 3520 |
*/ |
| 3521 |
private function remove_conflicting_seo_tags($remove_title = false, $remove_description_tags = []) { |
| 3522 |
if (!$this->dom) { |
| 3523 |
return; |
| 3524 |
} |
| 3525 |
|
| 3526 |
# Find head element |
| 3527 |
$head = $this->dom->find('head', 0); |
| 3528 |
if (!$head) { |
| 3529 |
return; |
| 3530 |
} |
| 3531 |
|
| 3532 |
# Check for custom SEO values from MetaSync SEO Sidebar |
| 3533 |
# Custom values take absolute priority over OTTO suggestions |
| 3534 |
$has_custom_title = false; |
| 3535 |
$has_custom_description = false; |
| 3536 |
|
| 3537 |
if (function_exists('is_singular') && is_singular()) { |
| 3538 |
$post_id = get_the_ID(); |
| 3539 |
if ($post_id) { |
| 3540 |
$custom_seo_title = get_post_meta($post_id, '_metasync_seo_title', true); |
| 3541 |
$custom_seo_description = get_post_meta($post_id, '_metasync_seo_desc', true); |
| 3542 |
$has_custom_title = !empty($custom_seo_title); |
| 3543 |
$has_custom_description = !empty($custom_seo_description); |
| 3544 |
} |
| 3545 |
} |
| 3546 |
|
| 3547 |
# Handle description tags - only remove specific tags that Otto is providing |
| 3548 |
if (!empty($remove_description_tags) && is_array($remove_description_tags) && !$has_custom_description) { |
| 3549 |
# Remove only the specific description tags that Otto is providing |
| 3550 |
foreach ($remove_description_tags as $selector) { |
| 3551 |
$tags = $this->dom->find($selector); |
| 3552 |
foreach ($tags as $tag) { |
| 3553 |
# Remove the tag |
| 3554 |
$tag->outertext = ''; |
| 3555 |
} |
| 3556 |
} |
| 3557 |
} |
| 3558 |
|
| 3559 |
if ($remove_title && !$has_custom_title) { |
| 3560 |
# Remove Open Graph and Twitter title tags (keep main <title>) |
| 3561 |
$title_selectors = [ |
| 3562 |
'meta[property=og:title]', |
| 3563 |
'meta[name=twitter:title]', |
| 3564 |
]; |
| 3565 |
|
| 3566 |
foreach ($title_selectors as $selector) { |
| 3567 |
$tags = $this->dom->find($selector); |
| 3568 |
foreach ($tags as $tag) { |
| 3569 |
$tag->outertext = ''; # Remove the tag |
| 3570 |
} |
| 3571 |
} |
| 3572 |
} |
| 3573 |
|
| 3574 |
# Save changes |
| 3575 |
$this->save_reload(); |
| 3576 |
} |
| 3577 |
|
| 3578 |
/** |
| 3579 |
* Clean OTTO internal fetch query parameters from rendered HTML. |
| 3580 |
* |
| 3581 |
* @param string $html The rendered HTML |
| 3582 |
* @return string HTML with OTTO fetch params removed |
| 3583 |
*/ |
| 3584 |
private function clean_otto_fetch_params($html) { |
| 3585 |
$patterns = [ |
| 3586 |
'/[?&]is_otto_page_fetch=1/', |
| 3587 |
'/[?&]otto_block_title=1/', |
| 3588 |
'/[?&]otto_block_desc=1/', |
| 3589 |
'/[?&]amp;is_otto_page_fetch=1/', |
| 3590 |
'/[?&]amp;otto_block_title=1/', |
| 3591 |
'/[?&]amp;otto_block_desc=1/', |
| 3592 |
]; |
| 3593 |
$html = preg_replace($patterns, '', $html); |
| 3594 |
|
| 3595 |
# Protect <script>/<style> bodies before the URL cleanup below. |
| 3596 |
# The href/src/action matcher can otherwise match JS that looks like an |
| 3597 |
# attribute (e.g. location.href="/x??y", el.src="a??b") and collapse the |
| 3598 |
# ?? inside it. Swap each block for a placeholder comment, run the cleanup, |
| 3599 |
# then restore byte-for-byte. |
| 3600 |
$protected = []; |
| 3601 |
$html = preg_replace_callback( |
| 3602 |
'/<(script|style)(\b[^>]*)>([\s\S]*?)<\/\1>/i', |
| 3603 |
function ($m) use (&$protected) { |
| 3604 |
$key = '<!--METASYNC_FETCHCLEAN_' . count($protected) . '-->'; |
| 3605 |
$protected[$key] = $m[0]; |
| 3606 |
return $key; |
| 3607 |
}, |
| 3608 |
$html |
| 3609 |
); |
| 3610 |
|
| 3611 |
# Normalize leftover query-string separators (?& / ?#) created by the param |
| 3612 |
# removal above — but ONLY inside URL attributes (href/src/action). |
| 3613 |
# |
| 3614 |
# IMPORTANT: never collapse "??". It is valid, load-bearing syntax for |
| 3615 |
# file-concatenation endpoints — Jetpack Boost (/_jb_static/??<hash>) and |
| 3616 |
# WordPress core (load-styles.php??... / load-scripts.php??...). Collapsing |
| 3617 |
# "??" -> "?" points those at a different URL that returns HTTP 400 / empty, |
| 3618 |
# so every affected CSS/JS bundle fails to load and the page renders broken |
| 3619 |
# (Kadence + Jetpack Boost "defer non-essential CSS" was the reported case). |
| 3620 |
# Removing an OTTO fetch param never produces a "??" anyway, so there is |
| 3621 |
# nothing to normalize there. |
| 3622 |
# (A previous implementation ran str_replace('??','?') over the ENTIRE |
| 3623 |
# document, which additionally corrupted inline JavaScript nullish operators.) |
| 3624 |
$html = preg_replace_callback( |
| 3625 |
'/\b(href|src|action)(\s*=\s*)(["\'])(.*?)\3/is', |
| 3626 |
function ($m) { |
| 3627 |
$url = str_replace(['?&', '?#'], ['?', '#'], $m[4]); |
| 3628 |
return $m[1] . $m[2] . $m[3] . $url . $m[3]; |
| 3629 |
}, |
| 3630 |
$html |
| 3631 |
); |
| 3632 |
|
| 3633 |
# Restore protected <script>/<style> blocks byte-for-byte. |
| 3634 |
if (!empty($protected)) { |
| 3635 |
$html = str_replace(array_keys($protected), array_values($protected), $html); |
| 3636 |
} |
| 3637 |
|
| 3638 |
return $html; |
| 3639 |
} |
| 3640 |
|
| 3641 |
/** |
| 3642 |
* Fix Divi 5 shortcode framework class renumbering |
| 3643 |
* |
| 3644 |
* When Divi 5's shortcode framework loads during the internal HTTP fetch, |
| 3645 |
* it uses shared counters across all template parts (header + page + footer), |
| 3646 |
* causing page content element classes to be offset (et_pb_section_0 becomes |
| 3647 |
* et_pb_section_4, etc.). The cached CSS references 0-based numbering. |
| 3648 |
* |
| 3649 |
* Detects the offset per element type and remaps back to 0-based. |
| 3650 |
* |
| 3651 |
* @param string $html The processed HTML string |
| 3652 |
* @return string HTML with corrected Divi class numbering |
| 3653 |
*/ |
| 3654 |
private function fix_divi_class_renumbering($html) { |
| 3655 |
# Only run on Divi sites |
| 3656 |
if (!defined('ET_CORE_VERSION')) { |
| 3657 |
return $html; |
| 3658 |
} |
| 3659 |
|
| 3660 |
# Only apply to Divi pages with template builder |
| 3661 |
if (strpos($html, 'et-l et-l--post') === false) { |
| 3662 |
return $html; |
| 3663 |
} |
| 3664 |
|
| 3665 |
# Locate the page content area (between et-l--post and et-l--footer) |
| 3666 |
$page_start = strpos($html, 'et-l et-l--post'); |
| 3667 |
if ($page_start === false) { |
| 3668 |
return $html; |
| 3669 |
} |
| 3670 |
|
| 3671 |
$footer_start = strpos($html, 'et-l et-l--footer', $page_start); |
| 3672 |
if ($footer_start === false) { |
| 3673 |
$footer_start = strlen($html); |
| 3674 |
} |
| 3675 |
|
| 3676 |
$page_content = substr($html, $page_start, $footer_start - $page_start); |
| 3677 |
|
| 3678 |
|
| 3679 |
# Divi element types that get renumbered by the shortcode framework. |
| 3680 |
# IMPORTANT: 'blog' and 'portfolio' MUST be included — OTTO's HTTP render |
| 3681 |
# shifts their numbering (et_pb_blog_0 → et_pb_blog_1) because the internal |
| 3682 |
# wp_remote_get includes the Theme Builder header template in Divi's counter. |
| 3683 |
# If the blog module class doesn't match between page 1 (OTTO-processed) and |
| 3684 |
# page 2 (AJAX, no OTTO), Divi's pagination JS can't find the container → empty results. |
| 3685 |
$types = [ |
| 3686 |
'section', 'row', 'column', 'text', 'blurb', 'toggle', |
| 3687 |
'button', 'image', 'group_carousel', |
| 3688 |
'blog', 'portfolio', 'filterable_portfolio', 'shop', |
| 3689 |
'heading', 'divider', 'code', 'icon', 'contact_form_7', |
| 3690 |
]; |
| 3691 |
|
| 3692 |
$offsets = []; |
| 3693 |
|
| 3694 |
# Lookahead excludes two non-instance class shapes so we never rewrite them: |
| 3695 |
# _tb_ → Theme Builder index suffix (et_pb_column_0_tb_header) |
| 3696 |
# _\d → Divi column WIDTH fractions (et_pb_column_4_4, _1_2, _1_3, …). |
| 3697 |
# The width fraction is NOT an instance counter; rewriting it (e.g. |
| 3698 |
# et_pb_column_4_4 → et_pb_column_0_4) strips the column's width rule from |
| 3699 |
# Divi's static stylesheet (.et_pb_column_4_4{width:100%}) and collapses the |
| 3700 |
# layout. Bare instance indices (et_pb_column_16, followed by space/quote) |
| 3701 |
# still match and renumber as intended. |
| 3702 |
$instance_lookahead = '(?!_(?:tb_|\d))'; |
| 3703 |
|
| 3704 |
foreach ($types as $type) { |
| 3705 |
# Find the FIRST numbered instance in page content (not _tb_ or width-fraction) |
| 3706 |
if (preg_match('/et_pb_' . preg_quote($type, '/') . '_(\d+)' . $instance_lookahead . '/', $page_content, $m)) { |
| 3707 |
$first_num = (int) $m[1]; |
| 3708 |
if ($first_num > 0) { |
| 3709 |
$offsets[$type] = $first_num; |
| 3710 |
} |
| 3711 |
} |
| 3712 |
} |
| 3713 |
|
| 3714 |
if (empty($offsets)) { |
| 3715 |
return $html; |
| 3716 |
} |
| 3717 |
|
| 3718 |
# Remap each element type back to 0-based numbering |
| 3719 |
foreach ($offsets as $etype => $offset) { |
| 3720 |
$escaped = preg_quote($etype, '/'); |
| 3721 |
$html = preg_replace_callback( |
| 3722 |
'/et_pb_' . $escaped . '_(\d+)' . $instance_lookahead . '/', |
| 3723 |
function ($m) use ($etype, $offset) { |
| 3724 |
$num = (int) $m[1]; |
| 3725 |
if ($num >= $offset) { |
| 3726 |
return 'et_pb_' . $etype . '_' . ($num - $offset); |
| 3727 |
} |
| 3728 |
return $m[0]; |
| 3729 |
}, |
| 3730 |
$html |
| 3731 |
); |
| 3732 |
} |
| 3733 |
|
| 3734 |
# Remove duplicate JS variable declarations from the shortcode framework |
| 3735 |
foreach (['et_pb_custom', 'et_frontend_scripts', 'et_builder_utils_params'] as $var_name) { |
| 3736 |
$needle = 'var ' . $var_name . ' = '; |
| 3737 |
$first = strpos($html, $needle); |
| 3738 |
if ($first !== false) { |
| 3739 |
$second = strpos($html, $needle, $first + strlen($needle)); |
| 3740 |
if ($second !== false) { |
| 3741 |
$end = strpos($html, ";\n", $second); |
| 3742 |
if ($end !== false) { |
| 3743 |
$html = substr($html, 0, $second) . substr($html, $end + 2); |
| 3744 |
} |
| 3745 |
} |
| 3746 |
} |
| 3747 |
} |
| 3748 |
|
| 3749 |
# Deduplicate animation data (shortcode framework creates duplicate entries) |
| 3750 |
if (preg_match('/var diviElementAnimationData = (\[.*?\]);/s', $html, $am)) { |
| 3751 |
$data = json_decode($am[1], true); |
| 3752 |
if (is_array($data)) { |
| 3753 |
$seen = []; |
| 3754 |
$unique = []; |
| 3755 |
foreach (array_reverse($data) as $e) { |
| 3756 |
$k = $e['class'] ?? ''; |
| 3757 |
if ($k !== '' && !isset($seen[$k])) { |
| 3758 |
$seen[$k] = true; |
| 3759 |
array_unshift($unique, $e); |
| 3760 |
} |
| 3761 |
} |
| 3762 |
if (count($unique) < count($data)) { |
| 3763 |
$html = str_replace( |
| 3764 |
$am[0], |
| 3765 |
'var diviElementAnimationData = ' . json_encode($unique, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE) . ';', |
| 3766 |
$html |
| 3767 |
); |
| 3768 |
} |
| 3769 |
} |
| 3770 |
} |
| 3771 |
|
| 3772 |
# Deduplicate multiview data |
| 3773 |
if (preg_match('/var diviElementMultiViewData = (\[.*?\]);/s', $html, $mv)) { |
| 3774 |
$data = json_decode($mv[1], true); |
| 3775 |
if (is_array($data)) { |
| 3776 |
$seen = []; |
| 3777 |
$unique = []; |
| 3778 |
foreach (array_reverse($data) as $e) { |
| 3779 |
$k = ($e['selector'] ?? '') . '|' . ($e['action'] ?? ''); |
| 3780 |
if ($k !== '|' && !isset($seen[$k])) { |
| 3781 |
$seen[$k] = true; |
| 3782 |
array_unshift($unique, $e); |
| 3783 |
} |
| 3784 |
} |
| 3785 |
if (count($unique) < count($data)) { |
| 3786 |
$html = str_replace( |
| 3787 |
$mv[0], |
| 3788 |
'var diviElementMultiViewData = ' . json_encode($unique, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE) . ';', |
| 3789 |
$html |
| 3790 |
); |
| 3791 |
} |
| 3792 |
} |
| 3793 |
} |
| 3794 |
|
| 3795 |
return $html; |
| 3796 |
} |
| 3797 |
|
| 3798 |
|
| 3799 |
} |
