gptranslate
Last commit date
assets
1 year ago
flags
1 year ago
language
1 year ago
gptranslate.php
1 year ago
multilang-routing.php
1 year ago
readme.txt
1 year ago
serverside-translations.php
1 year ago
settings.php
1 year ago
simplehtmldom.php
1 year ago
uninstall.php
1 year ago
serverside-translations.php
620 lines
| 1 | <?php |
| 2 | if (!defined('ABSPATH')) exit; |
| 3 | |
| 4 | // Hook to apply server-side translations before output is sent to the browser |
| 5 | add_action ( 'template_redirect', function () { |
| 6 | if (is_admin ()) { |
| 7 | return; |
| 8 | } |
| 9 | |
| 10 | $raw_request_uri = isset( $_SERVER['REQUEST_URI'] ) ? sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'])) : ''; |
| 11 | $uri = esc_url_raw( $raw_request_uri ); |
| 12 | |
| 13 | $settings = get_option ( 'gptranslate_options', [ ] ); |
| 14 | |
| 15 | if (empty ( $settings ['serverside_translations'] ) || $settings ['serverside_translations'] != 1) |
| 16 | return; |
| 17 | |
| 18 | // Skip static assets and admin/API routes |
| 19 | if (preg_match ( '#\.(ico|png|jpe?g|gif|svg|css|js|woff2?|ttf|eot|mp4|webm)$#i', $uri ) || strpos ( $uri, '/wp-' ) === 0 || strpos ( $uri, '/wp-json/' ) === 0) { |
| 20 | return; |
| 21 | } |
| 22 | |
| 23 | function normalizeText($text) { |
| 24 | $text = html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 25 | $text = str_replace(["\xC2\xA0", "\n", "\r", "\t"], ' ', $text); // NBSP e spazi strani |
| 26 | $text = preg_replace('/\s+/u', ' ', $text); // spazi multipli → singolo spazio |
| 27 | $text = trim($text); |
| 28 | return $text; |
| 29 | } |
| 30 | |
| 31 | ob_start ( function ($html) use ($uri, $settings) { |
| 32 | global $wpdb; |
| 33 | |
| 34 | $originalLang = $settings ['language'] ?? ''; |
| 35 | |
| 36 | if (empty ( $originalLang ) || $originalLang === $translatedLang) |
| 37 | return $html; |
| 38 | |
| 39 | if (! defined ( 'GPTRANSLATE_CURRENT_LANG' )) { |
| 40 | return $html; |
| 41 | } |
| 42 | |
| 43 | $translatedLang = GPTRANSLATE_CURRENT_LANG; |
| 44 | |
| 45 | // Exceptions and management for subfolder installations |
| 46 | if($settings['subfolder_installation']) { |
| 47 | // Explode URI into parts |
| 48 | $uri_parts = explode('/', ltrim($uri, '/')); |
| 49 | |
| 50 | // Remove the first part (subfolder) |
| 51 | array_shift($uri_parts); |
| 52 | |
| 53 | // Rebuild the URI |
| 54 | $uri = '/' . implode('/', $uri_parts); |
| 55 | } |
| 56 | |
| 57 | $pageLink = rtrim ( get_site_url (), '/' ) . '/' . ltrim ( GPTRANSLATE_CURRENT_LANG, '/' ) . $uri; |
| 58 | |
| 59 | if ( $settings ['serverside_translations_urldecode'] ) { |
| 60 | $pageLink = urldecode ( $pageLink ); |
| 61 | } |
| 62 | |
| 63 | if ($settings ['serverside_translations_ignore_querystring'] == 1) { |
| 64 | // Remove query string |
| 65 | $pageLink = strtok ( $pageLink, '?' ); |
| 66 | } elseif (! empty ( $_SERVER ['QUERY_STRING'] )) { |
| 67 | // Sanitize and rebuild query string |
| 68 | $raw_qs = isset($_SERVER['QUERY_STRING']) ? wp_unslash($_SERVER['QUERY_STRING']) : ''; // phpcs:ignore WordPress.Security.ValidatedSanitizedInput.InputNotSanitized |
| 69 | |
| 70 | parse_str ( $raw_qs, $qs_args ); |
| 71 | |
| 72 | $clean_args = [ ]; |
| 73 | foreach ( $qs_args as $key => $value ) { |
| 74 | $clean_key = sanitize_key ( $key ); |
| 75 | $clean_value = is_array ( $value ) ? array_map ( 'sanitize_text_field', $value ) : sanitize_text_field ( $value ); |
| 76 | $clean_args [$clean_key] = $clean_value; |
| 77 | } |
| 78 | |
| 79 | $clean_qs = http_build_query ( $clean_args, '', '&', PHP_QUERY_RFC3986 ); |
| 80 | |
| 81 | if ($clean_qs !== '') { |
| 82 | $pageLink = strtok ( $pageLink, '?' ) . '?' . $clean_qs; |
| 83 | } |
| 84 | } |
| 85 | |
| 86 | if ( $settings ['serverside_translations_urlencode_space'] ) { |
| 87 | $pageLink = str_ireplace ( ' ', '%20', $pageLink ); |
| 88 | } |
| 89 | |
| 90 | $table = $wpdb->prefix . 'gptranslate'; |
| 91 | |
| 92 | if( $settings ['rewrite_language_alias'] ) { |
| 93 | $row = $wpdb->get_row ( $wpdb->prepare ( "SELECT pagelink, translations, alt_translations FROM {$wpdb->prefix}gptranslate" . // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching |
| 94 | "\n WHERE ( pagelink = %s OR pagelink = %s OR translated_alias = %s OR translated_alias = %s) AND languageoriginal = %s AND languagetranslated = %s AND published = 1", |
| 95 | rtrim($pageLink, '/'), rtrim($pageLink, '/') . '/', rtrim($pageLink, '/'), rtrim($pageLink, '/') . '/', $originalLang, $translatedLang ), ARRAY_A ); |
| 96 | } else { |
| 97 | $row = $wpdb->get_row ( $wpdb->prepare ( "SELECT pagelink, translations, alt_translations FROM {$wpdb->prefix}gptranslate" . // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared, WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching |
| 98 | "\n WHERE ( pagelink = %s OR pagelink = %s ) AND languageoriginal = %s AND languagetranslated = %s AND published = 1", |
| 99 | rtrim($pageLink, '/'), rtrim($pageLink, '/') . '/', $originalLang, $translatedLang ), ARRAY_A ); |
| 100 | } |
| 101 | |
| 102 | if (! $row) { |
| 103 | return $html; |
| 104 | } |
| 105 | |
| 106 | // Process replacement method |
| 107 | if ($settings ['serverside_translations_method'] == 'regex') { |
| 108 | $translations = json_decode ( $row ['translations'], true ) ?? [ ]; |
| 109 | $altTranslations = json_decode ( $row ['alt_translations'], true ) ?? [ ]; |
| 110 | uksort ( $translations, fn ($a, $b) => strlen ( $b ) - strlen ( $a ) ); |
| 111 | uksort ( $altTranslations, fn ($a, $b) => strlen ( $b ) - strlen ( $a ) ); |
| 112 | |
| 113 | $caseInsensitive = ! empty ( $settings ['serverside_translations_caseinsensitive'] ); |
| 114 | $matchQuotes = ! empty ( $settings ['serverside_translations_matchquotes'] ); |
| 115 | $excludedPatterns = [ ]; |
| 116 | |
| 117 | $excludedCss = preg_replace ( '/,+/', ',', str_ireplace ( [ |
| 118 | "\r", |
| 119 | "\n", |
| 120 | '"' |
| 121 | ], [ |
| 122 | ',', |
| 123 | ',', |
| 124 | '' |
| 125 | ], $settings ['css_selector_serverside_leafnodes_excluded'] ?? '') ); |
| 126 | $excludedCss = array_filter ( array_map ( 'trim', explode ( ',', $excludedCss ) ) ); |
| 127 | |
| 128 | foreach ( $excludedCss as $selector ) { |
| 129 | if (preg_match ( '/^([a-z0-9]+)\.(.+)$/i', $selector, $m )) { |
| 130 | $excludedPatterns [] = '/<' . preg_quote ( $m [1], '/' ) . '(?=[^>]*\sclass\s*=\s*["\'][^"\']*\b' . preg_quote ( $m [2], '/' ) . '\b)[^>]*>/i'; |
| 131 | } elseif (preg_match ( '/^\.(.+)$/', $selector, $m )) { |
| 132 | $excludedPatterns [] = '/<([a-z0-9]+)(?=[^>]*\sclass\s*=\s*["\'][^"\']*\b' . preg_quote ( $m [1], '/' ) . '\b)[^>]*>/i'; |
| 133 | } |
| 134 | } |
| 135 | |
| 136 | $segments = preg_split ( '/(<[^>]+>)/i', $html, - 1, PREG_SPLIT_DELIM_CAPTURE ); |
| 137 | |
| 138 | $skipStack = [ ]; |
| 139 | foreach ( $segments as $index => $segment ) { |
| 140 | if (preg_match ( '/^<\s*(script|style)(\s|>)/i', $segment, $matches )) { |
| 141 | $skipStack [] = strtolower ( $matches [1] ); |
| 142 | } elseif (preg_match ( '/<\/\s*(script|style)[^>]*>/i', $segment, $matches )) { // <--- FIX QUI |
| 143 | $tag = strtolower ( $matches [1] ); |
| 144 | if (! empty ( $skipStack ) && end ( $skipStack ) === $tag) { |
| 145 | array_pop ( $skipStack ); |
| 146 | } |
| 147 | } elseif (preg_match ( '/^<\s*([a-zA-Z0-9]+)/', $segment, $tagMatch )) { |
| 148 | $tagName = strtolower ( $tagMatch [1] ); |
| 149 | foreach ( $excludedPatterns as $pattern ) { |
| 150 | if (preg_match ( $pattern, $segment )) { |
| 151 | $skipStack [] = $tagName; |
| 152 | break; |
| 153 | } |
| 154 | } |
| 155 | } elseif (preg_match ( '/^<\/\s*([a-zA-Z0-9]+)/', $segment, $tagMatch )) { |
| 156 | $tagName = strtolower ( $tagMatch [1] ); |
| 157 | if (! empty ( $skipStack ) && end ( $skipStack ) === $tagName) { |
| 158 | array_pop ( $skipStack ); |
| 159 | } |
| 160 | } elseif (! preg_match ( '/^<[^>]+>$/', $segment )) { |
| 161 | if (empty ( $skipStack )) { |
| 162 | if ($matchQuotes) { |
| 163 | $segment = str_ireplace ( '"', "'", $segment ); |
| 164 | } |
| 165 | foreach ( $translations as $originalText => $translatedText ) { |
| 166 | $originalText = normalizeText($originalText); |
| 167 | $segment = normalizeText($segment); |
| 168 | |
| 169 | $prevSegment = $segment; |
| 170 | if ($caseInsensitive) { |
| 171 | $segment = str_ireplace ( trim ( $originalText ), $translatedText, $segment ); |
| 172 | } else { |
| 173 | $segment = str_replace ( trim ( $originalText ), $translatedText, $segment ); |
| 174 | } |
| 175 | if ($segment !== $prevSegment) { |
| 176 | break; |
| 177 | } |
| 178 | } |
| 179 | } |
| 180 | } |
| 181 | |
| 182 | $segments [$index] = $segment; |
| 183 | } |
| 184 | |
| 185 | $html = implode ( '', $segments ); |
| 186 | |
| 187 | // Replace alt and title in images |
| 188 | if (! empty ( $settings ['translate_altimages'] )) { |
| 189 | $html = preg_replace_callback ( '/<img[^>]*\b(alt|title)\s*=\s*([\"\'])(.*?)\2[^>]*>/i', function ($matches) use ($altTranslations, $caseInsensitive, $matchQuotes) { |
| 190 | $attr = $matches [1]; |
| 191 | $quote = $matches [2]; |
| 192 | $value = $matches [3]; |
| 193 | if ($matchQuotes) |
| 194 | $value = str_ireplace ( '"', "'", $value ); |
| 195 | foreach ( $altTranslations as $original => $translated ) { |
| 196 | $original = normalizeText($original); |
| 197 | $value = normalizeText($value); |
| 198 | |
| 199 | $value = $caseInsensitive ? str_ireplace ( trim ( $original ), $translated, $value ) : str_replace ( trim ( $original ), $translated, $value ); |
| 200 | } |
| 201 | return preg_replace ( '/\b' . $attr . '\s*=\s*["\'].*?["\']/', "$attr=$quote$value$quote", $matches [0] ); |
| 202 | }, $html ); |
| 203 | } |
| 204 | |
| 205 | // Add skip marker |
| 206 | $html = preg_replace ( '/<body/i', '<body data-gptranslateskip="1" data-gptranslateoriginalalias="' . $row['pagelink'] . '"', $html, 1 ); |
| 207 | } elseif ($settings ['serverside_translations_method'] == 'domdocument') { |
| 208 | // Solution 2: classic DOMDocument approach, effective but could cause closing tags and encoding issues |
| 209 | try { |
| 210 | $translationsArray = json_decode ( $row['translations'], true ) ?? [ ]; |
| 211 | $altTranslationsArray = json_decode ( $row['alt_translations'] ?? '', true ) ?: [ ]; |
| 212 | |
| 213 | // Sort the translation keys in descending order by length. |
| 214 | uksort ( $translationsArray, function ( $a, $b ) { |
| 215 | return strlen ( $b ) - strlen ( $a ); |
| 216 | } ); |
| 217 | |
| 218 | // Sort the alt translation keys in descending order by length. |
| 219 | uksort ( $altTranslationsArray, function ( $a, $b ) { |
| 220 | return strlen ( $b ) - strlen ( $a ); |
| 221 | } ); |
| 222 | |
| 223 | // Initialize DOMDocument without converting the body content via mb_convert_encoding. |
| 224 | $doc = new DOMDocument (); |
| 225 | libxml_use_internal_errors ( true ); |
| 226 | // Use an XML encoding hack to inform DOMDocument about UTF-8 while preserving original characters. |
| 227 | $doc->loadHTML ( '<?xml encoding="UTF-8">' . $html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD ); |
| 228 | libxml_clear_errors (); |
| 229 | |
| 230 | // Use XPath to locate text nodes, skipping those in <script> or <style> tags. |
| 231 | $xpath = new DOMXPath ( $doc ); |
| 232 | $cssSelectorLeafnodesExcluded = str_ireplace ( '"', '', trim ( trim ( preg_replace ( '/,+/', ',', str_ireplace ( [ "\r", "\n" ], ",", $settings['css_selector_serverside_leafnodes_excluded'] ?? '' ) ) ), ',' ) ); |
| 233 | $excludedNodes = [ ]; |
| 234 | |
| 235 | if (! empty ( $cssSelectorLeafnodesExcluded )) { |
| 236 | $selectors = explode ( ',', $cssSelectorLeafnodesExcluded ); |
| 237 | $xpathQueries = [ ]; |
| 238 | |
| 239 | foreach ( $selectors as $selector ) { |
| 240 | $selector = trim ( $selector ); |
| 241 | if (empty ( $selector )) |
| 242 | continue; |
| 243 | |
| 244 | // Convert CSS selector to XPath |
| 245 | $selector = preg_replace ( '/\s+/', ' ', $selector ); // Normalize spaces |
| 246 | $selectorParts = explode ( ' ', $selector ); |
| 247 | |
| 248 | $xpathQuery = ''; |
| 249 | foreach ( $selectorParts as $part ) { |
| 250 | if (preg_match ( '/^([a-zA-Z0-9_-]+)?(\.[a-zA-Z0-9_-]+)?(#[a-zA-Z0-9_-]+)?$/', $part, $matches )) { |
| 251 | $tag = ! empty ( $matches[1] ) ? $matches[1] : '*'; // If no tag, use '*' |
| 252 | $class = ! empty ( $matches[2] ) ? substr ( $matches[2], 1 ) : ''; // Remove leading '.' |
| 253 | $id = ! empty ( $matches[3] ) ? substr ( $matches[3], 1 ) : ''; // Remove leading '#' |
| 254 | |
| 255 | $conditions = [ ]; |
| 256 | if ($class) { |
| 257 | $conditions[] = "contains(concat(' ', normalize-space(@class), ' '), ' $class ')"; |
| 258 | } |
| 259 | if ($id) { |
| 260 | $conditions[] = "@id='$id'"; |
| 261 | } |
| 262 | |
| 263 | $xpathQuery .= "//{$tag}" . (! empty ( $conditions ) ? "[" . implode ( " and ", $conditions ) . "]" : ""); |
| 264 | } |
| 265 | } |
| 266 | |
| 267 | if (! empty ( $xpathQuery )) { |
| 268 | $xpathQueries[] = $xpathQuery; |
| 269 | } |
| 270 | } |
| 271 | |
| 272 | // Execute all XPath queries and collect excluded nodes |
| 273 | foreach ( $xpathQueries as $query ) { |
| 274 | foreach ( $xpath->query ( $query ) as $excludedNode ) { |
| 275 | $excludedNodes[] = $excludedNode; |
| 276 | } |
| 277 | } |
| 278 | } |
| 279 | |
| 280 | $textNodes = $xpath->query ( '//text()[not(ancestor::script) and not(ancestor::style) and normalize-space()]' ); |
| 281 | $caseInsensitive = ! empty ( $settings['serverside_translations_caseinsensitive'] ); |
| 282 | $matchQuotes = ! empty ( $settings['serverside_translations_matchquotes'] ); |
| 283 | |
| 284 | // Process each text node with translation replacements. |
| 285 | foreach ( $textNodes as $textNode ) { |
| 286 | if ($cssSelectorLeafnodesExcluded) { |
| 287 | $skipNode = false; |
| 288 | foreach ( $excludedNodes as $excludedNode ) { |
| 289 | if ($excludedNode->isSameNode ( $textNode->parentNode )) { |
| 290 | $skipNode = true; |
| 291 | break; |
| 292 | } |
| 293 | } |
| 294 | if ($skipNode) |
| 295 | continue; |
| 296 | } |
| 297 | |
| 298 | $textContent = $textNode->nodeValue; |
| 299 | if ($matchQuotes) { |
| 300 | $textContent = str_ireplace ( '"', "'", $textContent ); |
| 301 | } |
| 302 | foreach ( $translationsArray as $originalText => $translatedText ) { |
| 303 | $originalText = normalizeText($originalText); |
| 304 | $textContent = normalizeText($textContent); |
| 305 | |
| 306 | $prevTextContent = $textContent; // Save the current state |
| 307 | if ($caseInsensitive) { |
| 308 | $textContent = str_ireplace ( trim ( $originalText ), $translatedText, $textContent ); |
| 309 | } else { |
| 310 | $textContent = str_replace ( trim ( $originalText ), $translatedText, $textContent ); |
| 311 | } |
| 312 | // Break out of the loop if a replacement occurred. |
| 313 | if ($textContent !== $prevTextContent) { |
| 314 | break; |
| 315 | } |
| 316 | } |
| 317 | $textNode->nodeValue = $textContent; |
| 318 | } |
| 319 | |
| 320 | // Check if also images 'alt' are enabled to be translated |
| 321 | if (! empty ( $settings['translate_altimages'] )) { |
| 322 | $imgNodesAlt = $xpath->query ( '//img[@alt]' ); |
| 323 | foreach ( $imgNodesAlt as $imgNode ) { |
| 324 | $altText = $imgNode->getAttribute ( 'alt' ); |
| 325 | if ($matchQuotes) { |
| 326 | $altText = str_ireplace ( '"', "'", $altText ); |
| 327 | } |
| 328 | foreach ( $altTranslationsArray as $originalText => $translatedText ) { |
| 329 | $originalText = normalizeText($originalText); |
| 330 | $altText = normalizeText($altText); |
| 331 | |
| 332 | $prevAltText = $altText; |
| 333 | if ($caseInsensitive) { |
| 334 | $altText = str_ireplace ( trim ( $originalText ), $translatedText, $altText ); |
| 335 | } else { |
| 336 | $altText = str_replace ( trim ( $originalText ), $translatedText, $altText ); |
| 337 | } |
| 338 | if ($altText !== $prevAltText) { |
| 339 | break; |
| 340 | } |
| 341 | } |
| 342 | $imgNode->setAttribute ( 'alt', $altText ); |
| 343 | } |
| 344 | |
| 345 | $imgNodesTitle = $xpath->query ( '//img[@title]' ); |
| 346 | foreach ( $imgNodesTitle as $imgNode ) { |
| 347 | $titleText = $imgNode->getAttribute ( 'title' ); |
| 348 | if ($matchQuotes) { |
| 349 | $titleText = str_ireplace ( '"', "'", $titleText ); |
| 350 | } |
| 351 | foreach ( $altTranslationsArray as $originalText => $translatedText ) { |
| 352 | $originalText = normalizeText($originalText); |
| 353 | $titleText = normalizeText($titleText); |
| 354 | |
| 355 | $prevTitleText = $titleText; |
| 356 | if ($caseInsensitive) { |
| 357 | $titleText = str_ireplace ( trim ( $originalText ), $translatedText, $titleText ); |
| 358 | } else { |
| 359 | $titleText = str_replace ( trim ( $originalText ), $translatedText, $titleText ); |
| 360 | } |
| 361 | if ($titleText !== $prevTitleText) { |
| 362 | break; |
| 363 | } |
| 364 | } |
| 365 | $imgNode->setAttribute ( 'title', $titleText ); |
| 366 | } |
| 367 | } |
| 368 | |
| 369 | // Optionally, add an attribute to the <body> tag to signal that translations were applied. |
| 370 | $bodyTag = $doc->getElementsByTagName ( 'body' )->item ( 0 ); |
| 371 | if ($bodyTag) { |
| 372 | $bodyTag->setAttribute ( 'data-gptranslateskip', '1' ); |
| 373 | $bodyTag->setAttribute ( 'data-gptranslateoriginalalias', htmlspecialchars($row['pagelink'], ENT_QUOTES, 'UTF-8') ); |
| 374 | } |
| 375 | |
| 376 | // Save the modified HTML. |
| 377 | $html = $doc->saveHTML (); |
| 378 | |
| 379 | // Decode HTML entities back into UTF-8 characters. |
| 380 | $html = html_entity_decode ( $html, ENT_QUOTES | ENT_HTML5, 'UTF-8' ); |
| 381 | |
| 382 | } catch ( Exception $e ) { |
| 383 | // Handle exception if needed |
| 384 | } |
| 385 | } elseif ($settings['serverside_translations_method'] == 'simplehtmldom') { |
| 386 | require_once plugin_dir_path(__FILE__) . 'simplehtmldom.php'; |
| 387 | |
| 388 | // Helper: checks if $child is inside (or is) $parent. |
| 389 | function nodeIsInsideExcluded($child, $excludedNodes) { |
| 390 | while ($child !== null) { |
| 391 | foreach ($excludedNodes as $ex) { |
| 392 | if ($child === $ex) { |
| 393 | return true; |
| 394 | } |
| 395 | } |
| 396 | $child = $child->parent; |
| 397 | } |
| 398 | return false; |
| 399 | } |
| 400 | |
| 401 | // Recursive function to process all text nodes. |
| 402 | function processTextNodes($node, $excludedNodes, $translationsArray, $caseInsensitive, $matchQuotes) { |
| 403 | if ($node->tag === 'text') { |
| 404 | if (! nodeIsInsideExcluded($node, $excludedNodes) && trim($node->innertext) && $node->innertext != "\t") { |
| 405 | $text = $node->innertext; |
| 406 | if ($matchQuotes) { |
| 407 | $text = str_ireplace('"', "'", $text); |
| 408 | } |
| 409 | |
| 410 | // Apply translations without overriding longer translations |
| 411 | $processedParts = []; |
| 412 | foreach ($translationsArray as $originalText => $translatedText) { |
| 413 | $originalText = normalizeText($originalText); |
| 414 | $text = normalizeText($text); |
| 415 | |
| 416 | if ($caseInsensitive) { |
| 417 | $text = preg_replace_callback( |
| 418 | '/(?<!\w)' . preg_quote(trim($originalText), '/') . '(?!\w)/ui', |
| 419 | function ($matches) use ($translatedText, &$processedParts) { |
| 420 | if (in_array($matches[0], $processedParts, true)) { |
| 421 | return $matches[0]; |
| 422 | } |
| 423 | $processedParts[] = $translatedText; |
| 424 | return $translatedText; |
| 425 | }, |
| 426 | $text |
| 427 | ); |
| 428 | } else { |
| 429 | $text = preg_replace_callback( |
| 430 | '/(?<!\w)' . preg_quote(trim($originalText), '/') . '(?!\w)/', |
| 431 | function ($matches) use ($translatedText, &$processedParts) { |
| 432 | if (in_array($matches[0], $processedParts, true)) { |
| 433 | return $matches[0]; |
| 434 | } |
| 435 | $processedParts[] = $translatedText; |
| 436 | return $translatedText; |
| 437 | }, |
| 438 | $text |
| 439 | ); |
| 440 | } |
| 441 | } |
| 442 | $node->innertext = $text; |
| 443 | } |
| 444 | } else { |
| 445 | $tagLower = strtolower($node->tag ?? ''); |
| 446 | if (in_array($tagLower, ['script', 'style'])) { |
| 447 | return; |
| 448 | } |
| 449 | if (isset($node->nodes) && is_array($node->nodes)) { |
| 450 | foreach ($node->nodes as $child) { |
| 451 | processTextNodes($child, $excludedNodes, $translationsArray, $caseInsensitive, $matchQuotes); |
| 452 | } |
| 453 | } |
| 454 | } |
| 455 | } |
| 456 | |
| 457 | // Main processing code. |
| 458 | try { |
| 459 | $translationsArray = json_decode($row['translations'], true) ?? []; |
| 460 | $altTranslationsArray = json_decode($row['alt_translations'] ?? '', true) ?: []; |
| 461 | |
| 462 | // Sort translations by descending key length |
| 463 | uksort($translationsArray, function ($a, $b) { |
| 464 | return strlen($b) - strlen($a); |
| 465 | }); |
| 466 | |
| 467 | // Sort alt translations by descending key length |
| 468 | uksort($altTranslationsArray, function ($a, $b) { |
| 469 | return strlen($b) - strlen($a); |
| 470 | }); |
| 471 | |
| 472 | $htmlObj = gptranslate_simplehtmldom_str_get_html($html); |
| 473 | |
| 474 | $cssSelectorLeafnodesExcluded = str_ireplace( |
| 475 | '"', |
| 476 | '', |
| 477 | trim(trim(preg_replace('/,+/', ',', str_ireplace(["\r", "\n"], ",", $settings['css_selector_serverside_leafnodes_excluded'] ?? ''))), ',') |
| 478 | ); |
| 479 | |
| 480 | $excludedNodes = []; |
| 481 | if (! empty($cssSelectorLeafnodesExcluded)) { |
| 482 | $selectors = explode(',', $cssSelectorLeafnodesExcluded); |
| 483 | foreach ($selectors as $selector) { |
| 484 | $selector = trim($selector); |
| 485 | if (! empty($selector)) { |
| 486 | $foundNodes = $htmlObj->find($selector); |
| 487 | foreach ($foundNodes as $node) { |
| 488 | $excludedNodes[] = $node; |
| 489 | } |
| 490 | } |
| 491 | } |
| 492 | } |
| 493 | |
| 494 | $caseInsensitive = ! empty($settings['serverside_translations_caseinsensitive']); |
| 495 | $matchQuotes = ! empty($settings['serverside_translations_matchquotes']); |
| 496 | |
| 497 | processTextNodes($htmlObj, $excludedNodes, $translationsArray, $caseInsensitive, $matchQuotes); |
| 498 | |
| 499 | // Check if also images 'alt' are enabled to be translated |
| 500 | if (! empty($settings['translate_altimages'])) { |
| 501 | foreach ($htmlObj->find('img[alt]') as $imgNode) { |
| 502 | $altText = $imgNode->alt; |
| 503 | if ($matchQuotes) { |
| 504 | $altText = str_ireplace('"', "'", $altText); |
| 505 | } |
| 506 | foreach ($altTranslationsArray as $originalText => $translatedText) { |
| 507 | $originalText = normalizeText($originalText); |
| 508 | $altText = normalizeText($altText); |
| 509 | |
| 510 | $prevAltText = $altText; |
| 511 | if ($caseInsensitive) { |
| 512 | $altText = str_ireplace(trim($originalText), $translatedText, $altText); |
| 513 | } else { |
| 514 | $altText = str_replace(trim($originalText), $translatedText, $altText); |
| 515 | } |
| 516 | if ($altText !== $prevAltText) { |
| 517 | break; |
| 518 | } |
| 519 | } |
| 520 | $imgNode->alt = $altText; |
| 521 | } |
| 522 | |
| 523 | foreach ($htmlObj->find('img[title]') as $imgNode) { |
| 524 | $titleText = $imgNode->title; |
| 525 | if ($matchQuotes) { |
| 526 | $titleText = str_ireplace('"', "'", $titleText); |
| 527 | } |
| 528 | foreach ($altTranslationsArray as $originalText => $translatedText) { |
| 529 | $originalText = normalizeText($originalText); |
| 530 | $titleText = normalizeText($titleText); |
| 531 | |
| 532 | $prevTitleText = $titleText; |
| 533 | if ($caseInsensitive) { |
| 534 | $titleText = str_ireplace(trim($originalText), $translatedText, $titleText); |
| 535 | } else { |
| 536 | $titleText = str_replace(trim($originalText), $translatedText, $titleText); |
| 537 | } |
| 538 | if ($titleText !== $prevTitleText) { |
| 539 | break; |
| 540 | } |
| 541 | } |
| 542 | $imgNode->title = $titleText; |
| 543 | } |
| 544 | } |
| 545 | |
| 546 | if ($bodyElement = $htmlObj->find('body', 0)) { |
| 547 | $bodyElement->setAttribute('data-gptranslateskip', '1'); |
| 548 | $bodyElement->setAttribute ( 'data-gptranslateoriginalalias', htmlspecialchars($row['pagelink'], ENT_QUOTES, 'UTF-8') ); |
| 549 | } |
| 550 | |
| 551 | $modifiedHtml = $htmlObj->save(); |
| 552 | $modifiedHtml = html_entity_decode($modifiedHtml, ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 553 | |
| 554 | $html = $modifiedHtml; |
| 555 | } catch (Exception $e) { |
| 556 | // Handle exceptions as needed |
| 557 | } |
| 558 | } elseif ($settings['serverside_translations_method'] == 'strireplace') { |
| 559 | // Solution 3: simplest approach, unconditional str_ireplace that could cause unintentional replacements |
| 560 | try { |
| 561 | $translationsArray = json_decode($row['translations'], true) ?? []; |
| 562 | $altTranslationsArray = json_decode($row['alt_translations'] ?? '', true) ?: []; |
| 563 | |
| 564 | // Do body page translations replacements |
| 565 | uksort($translationsArray, function ($a, $b) { |
| 566 | return strlen($b) - strlen($a); |
| 567 | }); |
| 568 | |
| 569 | $caseInsensitive = !empty($settings['serverside_translations_caseinsensitive']); |
| 570 | $matchQuotes = !empty($settings['serverside_translations_matchquotes']); |
| 571 | |
| 572 | foreach ($translationsArray as $originalText => $translatedText) { |
| 573 | $originalText = normalizeText($originalText); |
| 574 | $html = normalizeText($html); |
| 575 | |
| 576 | if ($caseInsensitive) { |
| 577 | $html = str_ireplace(trim($originalText), $translatedText, $html); |
| 578 | |
| 579 | // Check also if both single quotes or double quotes should be checked to replace |
| 580 | if ($matchQuotes && strpos($originalText, "'") !== false) { |
| 581 | $originalTextAlt = str_ireplace("'", '"', $originalText); |
| 582 | $html = str_ireplace(trim($originalTextAlt), $translatedText, $html); |
| 583 | } |
| 584 | } else { |
| 585 | $html = str_replace(trim($originalText), $translatedText, $html); |
| 586 | |
| 587 | if ($matchQuotes && strpos($originalText, "'") !== false) { |
| 588 | $originalTextAlt = str_replace("'", '"', $originalText); |
| 589 | $html = str_replace(trim($originalTextAlt), $translatedText, $html); |
| 590 | } |
| 591 | } |
| 592 | } |
| 593 | |
| 594 | // Check if also images 'alt' are enabled to be translated |
| 595 | if (!empty($settings['translate_altimages'])) { |
| 596 | foreach ($altTranslationsArray as $originalAlt => $translatedAlt) { |
| 597 | if ($matchQuotes) { |
| 598 | $originalAlt = str_ireplace('"', "'", $originalAlt); |
| 599 | } |
| 600 | if ($caseInsensitive) { |
| 601 | $html = str_ireplace('alt="' . trim($originalAlt) . '"', 'alt="' . $translatedAlt . '"', $html); |
| 602 | $html = str_ireplace("alt='" . trim($originalAlt) . "'", "alt='" . $translatedAlt . "'", $html); |
| 603 | } else { |
| 604 | $html = str_replace('alt="' . trim($originalAlt) . '"', 'alt="' . $translatedAlt . '"', $html); |
| 605 | $html = str_replace("alt='" . trim($originalAlt) . "'", "alt='" . $translatedAlt . "'", $html); |
| 606 | } |
| 607 | } |
| 608 | } |
| 609 | |
| 610 | $html = str_ireplace('<body', '<body data-gptranslateskip="1" data-gptranslateoriginalalias="' . $row['pagelink'] . '"', $html); |
| 611 | |
| 612 | } catch (Exception $e) { |
| 613 | // Handle exceptions as needed |
| 614 | } |
| 615 | } |
| 616 | |
| 617 | return $html; |
| 618 | } ); |
| 619 | } ); |
| 620 |