tags // We split the HTML into top-level block elements and translate each $chunks = $this->split_into_chunks($html); foreach ($chunks as $chunk) { $chunk = trim($chunk); if (empty($chunk)) { continue; } // Check if the chunk has meaningful translatable text $text_only = wp_strip_all_tags($chunk); $text_only = html_entity_decode($text_only, ENT_QUOTES | ENT_HTML5, 'UTF-8'); $text_only = trim(preg_replace('/\s+/', ' ', $text_only)); if (mb_strlen($text_only) < 2 || !preg_match('/[\p{L}\p{N}]/u', $text_only)) { continue; } // Avoid duplicates if (!in_array($chunk, $strings, true)) { $strings[] = $chunk; } } return $strings; } /** * Split HTML content into semantic chunks for translation * Each chunk is a top-level block element (p, h1-h6, ul, ol, blockquote, table, div, etc.) * * @param string $html * @return array */ private function split_into_chunks($html) { $chunks = []; // Block-level elements that we want to split on $block_tags = 'p|h[1-6]|ul|ol|li|blockquote|table|thead|tbody|tr|th|td|div|section|article|header|footer|figure|figcaption|pre|address|dl|dt|dd|details|summary|hr|br'; // Try to split by top-level block elements // Match: optional whitespace, opening tag, content (greedy), closing tag $pattern = '#(<(?:' . $block_tags . ')(?:\s[^>]*)?>[\s\S]*?)#i'; // Split by block elements, capturing positions in a single pass if (preg_match_all($pattern, $html, $positioned_matches, PREG_OFFSET_CAPTURE)) { $last_end = 0; foreach ($positioned_matches[0] as $match) { $match_text = $match[0]; $match_pos = $match[1]; // Check for text between blocks $between = substr($html, $last_end, $match_pos - $last_end); $between = trim($between); if (!empty($between) && $this->has_translatable_text($between)) { $chunks[] = $between; } // Add the block element as a chunk $chunks[] = $match_text; $last_end = $match_pos + strlen($match_text); } // Handle trailing content $trailing = trim(substr($html, $last_end)); if (!empty($trailing) && $this->has_translatable_text($trailing)) { $chunks[] = $trailing; } } else { // No block elements found — split by double newlines (\n\n) if present, else treat entire content as one chunk $paragraphs = preg_split('/(\r\n\r\n|\n\n)/u', $html); if (count($paragraphs) > 1) { foreach ($paragraphs as $para) { $para = trim($para); if (!empty($para) && $this->has_translatable_text($para)) { $chunks[] = $para; } } } else { $chunks[] = trim($html); } } return $chunks; } /** * Check if text has translatable content */ private function has_translatable_text($text) { $stripped = wp_strip_all_tags($text); $stripped = html_entity_decode($stripped, ENT_QUOTES | ENT_HTML5, 'UTF-8'); $stripped = trim(preg_replace('/\s+/', ' ', $stripped)); return (mb_strlen($stripped) >= 2 && preg_match('/[\p{L}\p{N}]/u', $stripped)); } /** * Apply translations to HTML content * Replaces entire chunk with its translated version * * @param string $html Original HTML content * @param array $translations Map of original => translated text * @return string Translated HTML */ public function apply_translations($html, $translations) { if (empty($translations)) { return $html; } $result = $html; // Sort by length descending to replace longest strings first uksort($translations, function($a, $b) { return strlen($b) - strlen($a); }); foreach ($translations as $original => $translated) { if (empty($original) || empty($translated) || $original === $translated) { continue; } // Replace all occurrences — handles duplicate paragraphs // Safe because we're replacing whole semantic HTML chunks $result = str_replace($original, $translated, $result); } return $result; } }