Changeset 3455577

Timestamp:

02/06/2026 05:30:28 PM (2 weeks ago)

Author:

gkanters

Message:

New release 2.2.5

Location:

ai-translate/trunk/includes

Files:

• class-ai-dom.php (modified) (1 diff)
• class-ai-ob.php (modified) (4 diffs)

ai-translate/trunk/includes/class-ai-dom.php

@@ -348 +348 @@
      * @return string
      */
-    private static function extractAndReplace($html, $tagName, &$placeholders, &$counter)
+    public static function extractAndReplace($html, $tagName, &$placeholders, &$counter)
     {
         $result = '';

ai-translate/trunk/includes/class-ai-ob.php

@@ -380 +380 @@
                 preg_match('/<body\b[^>]*>([\s\S]*?)<\/body>/i', (string) $html, $mOrig)) {
                 $newInner = (string) $mNew[1];
-                $html2 = (string) preg_replace('/(<body\b[^>]*>)[\s\S]*?(<\/body>)/i', '$1' . $newInner . '$2', (string) $html, 1);
+                // Use preg_replace_callback instead of preg_replace to prevent $n back-reference
+                // interpretation in the replacement string. Content like bash code blocks may contain
+                // $1, $2 etc. which preg_replace would interpret as capture group references,
+                // injecting </body> mid-content and truncating the page.
+                $html2 = (string) preg_replace_callback(
+                    '/(<body\b[^>]*>)[\s\S]*?(<\/body>)/i',
+                    function ($m) use ($newInner) {
+                        return $m[1] . $newInner . $m[2];
+                    },
+                    (string) $html,
+                    1
+                );
             }
             
@@ -407 +418 @@
         }
 
+        // Protect script/style tags from DOMDocument corruption in SEO/URL pass.
+        // DOMDocument::saveHTML() can encode & as &amp; inside <script>/<style>,
+        // breaking JavaScript operators like && and CSS syntax.
+        $seoPlaceholders = [];
+        $seoPlaceholderCounter = 0;
+        $html2Protected = $html2;
+        $html2Protected = AI_DOM::extractAndReplace($html2Protected, 'script', $seoPlaceholders, $seoPlaceholderCounter);
+        $html2Protected = AI_DOM::extractAndReplace($html2Protected, 'style', $seoPlaceholders, $seoPlaceholderCounter);
+
         // Combined SEO + URL pass: single DOM parse instead of two separate ones
         $doc = new \DOMDocument();
         $internalErrors = libxml_use_internal_errors(true);
-        $htmlToLoad = AI_DOM::ensureUtf8($html2);
+        $htmlToLoad = AI_DOM::ensureUtf8($html2Protected);
         $doc->loadHTML('<?xml encoding="utf-8" ?>' . $htmlToLoad, LIBXML_HTML_NODEFDTD);
         libxml_clear_errors();
@@ -424 +444 @@
             if (stripos($html3, '<!DOCTYPE') === false) {
                 $html3 = $docMatch[1] . "\n" . $html3;
+            }
+        }
+
+        // Restore script/style tags from placeholders
+        if (!empty($seoPlaceholders)) {
+            foreach ($seoPlaceholders as $placeholderId => $original_tag) {
+                $escapedId = preg_quote($placeholderId, '/');
+                $pattern = '/<div\s+[^>]*data-ai-placeholder=["\']' . $escapedId . '["\'][^>]*><\/div>/is';
+                if (preg_match($pattern, $html3)) {
+                    $html3 = preg_replace($pattern, $original_tag, $html3, 1);
+                }
             }
         }
@@ -686 +717 @@
         if (function_exists('is_front_page') && is_front_page()) {
             $front_page_id = (int) get_option('page_on_front');
+            $paged = get_query_var('paged', 0);
             if ($front_page_id > 0) {
                 // Static front page: use post ID
                 return 'post:' . $front_page_id;
             } else {
-                // Posts listing homepage: always use normalized path
-                // Normalize to '/' to prevent /en/, /en, //, etc. from creating different caches
-                return 'path:' . md5('/');
+                // Posts listing homepage: include page number to differentiate paginated pages.
+                // is_front_page() returns true for ALL paginated blog pages when show_on_front='posts',
+                // so without paged in the key, /page/2 would serve the cached /page/1 content.
+                $path_key = ($paged > 1) ? '/page/' . $paged : '/';
+                return 'path:' . md5($path_key);
             }
         }