gptranslate
Last commit date
assets
13 hours ago
flags
13 hours ago
includes
13 hours ago
language
13 hours ago
ajax-handler.php
13 hours ago
gptranslate.php
13 hours ago
multilang-routing.php
13 hours ago
readme.txt
13 hours ago
serverside-translations.php
13 hours ago
settings.php
13 hours ago
simplehtmldom.php
13 hours ago
uninstall.php
13 hours ago
serverside-translations.php
2279 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 | // Some themes/plugins (e.g. certain LMS course player templates) trigger the |
| 7 | // template_redirect action more than once within the same request. This closure |
| 8 | // declares several named global functions (normalizeText, processTextNodes, etc.); |
| 9 | // running it again would attempt to redeclare them and cause a PHP fatal error. |
| 10 | static $gptranslate_serverside_already_ran = false; |
| 11 | if ($gptranslate_serverside_already_ran) { |
| 12 | return; |
| 13 | } |
| 14 | $gptranslate_serverside_already_ran = true; |
| 15 | |
| 16 | if (is_admin ()) { |
| 17 | return; |
| 18 | } |
| 19 | |
| 20 | $raw_request_uri = isset( $_SERVER['REQUEST_URI'] ) ? sanitize_text_field(wp_unslash($_SERVER['REQUEST_URI'])) : ''; |
| 21 | $uri = esc_url_raw( $raw_request_uri ); |
| 22 | |
| 23 | $settings = get_option ( 'gptranslate_options', [ ] ); |
| 24 | |
| 25 | if (empty ( $settings ['serverside_translations'] ) || $settings ['serverside_translations'] != 1) |
| 26 | return; |
| 27 | |
| 28 | // Skip static assets and admin/API routes |
| 29 | if (preg_match ( '#\.(ico|png|jpe?g|gif|svg|css|js|woff2?|ttf|eot|mp4|webm)$#i', $uri ) || strpos ( $uri, '/wp-' ) === 0 || strpos ( $uri, '/wp-json/' ) === 0) { |
| 30 | return; |
| 31 | } |
| 32 | |
| 33 | function normalizeText($text) { |
| 34 | $text = html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 35 | $text = str_replace(["\xC2\xA0", "\n", "\r", "\t"], ' ', $text); // NBSP e spazi strani |
| 36 | $text = preg_replace('/\s+/u', ' ', $text); // spazi multipli → singolo spazio |
| 37 | $text = trim($text); |
| 38 | return $text; |
| 39 | } |
| 40 | |
| 41 | // Normalizza senza trim per preservare spazi iniziali/finali (utile per SimpleHTMLDOM). |
| 42 | function normalizeTextNoTrim($text) { |
| 43 | $text = html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 44 | $text = str_replace(["\xC2\xA0", "\n", "\r", "\t"], ' ', $text); // NBSP e spazi strani |
| 45 | $text = preg_replace('/\s+/u', ' ', $text); // spazi multipli -> singolo spazio |
| 46 | return $text; |
| 47 | } |
| 48 | |
| 49 | // Normalizza il testo per l'incremental translation detection con dictionary replacement |
| 50 | // Applica le stesse trasformazioni del client-side JS per far corrispondere i testi nel DB |
| 51 | function normalizeTextForIncremental($text, $settings, $originalLang, $translatedLang) { |
| 52 | // Normalizzazione base |
| 53 | $text = normalizeText($text); |
| 54 | |
| 55 | // Il dictionary viene decodificato e filtrato per coppia di lingue una sola volta: |
| 56 | // questa funzione gira per ogni nodo di testo della pagina. |
| 57 | static $compiledRulesCache = []; |
| 58 | $cacheKey = $originalLang . '|' . $translatedLang; |
| 59 | |
| 60 | if (!isset($compiledRulesCache[$cacheKey])) { |
| 61 | $compiledRules = []; |
| 62 | |
| 63 | // Carica il dictionary come array (è una stringa JSON) |
| 64 | $dictionary = isset($settings['words_leafnodes_excluded_bylanguage_repeatable']) |
| 65 | ? $settings['words_leafnodes_excluded_bylanguage_repeatable'] |
| 66 | : []; |
| 67 | |
| 68 | // Decodifica la stringa JSON se necessario |
| 69 | if (is_string($dictionary)) { |
| 70 | $dictionary = json_decode($dictionary, true); |
| 71 | } |
| 72 | |
| 73 | if (is_array($dictionary)) { |
| 74 | // Iterare su tutte le entries del dictionary |
| 75 | foreach ($dictionary as $entry) { |
| 76 | // Filtrare per lingua (fallback a "*" per all languages) |
| 77 | $langOriginal = isset($entry['langOriginal']) ? $entry['langOriginal'] : '*'; |
| 78 | $langTranslated = isset($entry['langTranslated']) ? $entry['langTranslated'] : '*'; |
| 79 | |
| 80 | // Verifica che le lingue corrispondano (considerando "*" come wildcard) |
| 81 | if (($langOriginal !== '*' && $langOriginal !== $originalLang) || ($langTranslated !== '*' && $langTranslated !== $translatedLang)) { |
| 82 | continue; |
| 83 | } |
| 84 | |
| 85 | // Parse words (CSV format: "word1,word2,word3") |
| 86 | $words = isset($entry['word']) ? array_map('trim', explode(',', $entry['word'])) : []; |
| 87 | |
| 88 | // Parse translations (CSV format: "trans1,trans2,trans3") |
| 89 | $translations = isset($entry['optionalTranslation']) ? array_map('trim', explode(',', $entry['optionalTranslation'])) : []; |
| 90 | |
| 91 | // Applica sostituzioni per indice (word[0] -> translation[0], ecc.) |
| 92 | foreach ($words as $i => $word) { |
| 93 | if (empty($word)) continue; |
| 94 | |
| 95 | $replacement = isset($translations[$i]) ? $translations[$i] : ''; |
| 96 | if (empty($replacement)) continue; |
| 97 | |
| 98 | // Word-boundary matching: sostituisce solo le parole intere, non substring |
| 99 | // Esempio: 'di' non sostituisce 'di' dentro 'modify' o 'indifferenziato' |
| 100 | $compiledRules[] = [ |
| 101 | 'pattern' => '/\b' . preg_quote($word, '/') . '\b/ui', |
| 102 | 'replacement' => $replacement, |
| 103 | ]; |
| 104 | } |
| 105 | } |
| 106 | } |
| 107 | |
| 108 | $compiledRulesCache[$cacheKey] = $compiledRules; |
| 109 | } |
| 110 | |
| 111 | foreach ($compiledRulesCache[$cacheKey] as $rule) { |
| 112 | $text = preg_replace($rule['pattern'], $rule['replacement'], $text); |
| 113 | } |
| 114 | |
| 115 | return $text; |
| 116 | } |
| 117 | |
| 118 | // Incremental exact-match applicato a un singolo nodo di testo, condiviso dai metodi |
| 119 | // regex / domdocument / simplehtmldom. |
| 120 | // Il testo viene sostituito solo su corrispondenza intera (o per frammenti delimitati |
| 121 | // dalle excluded words), mai a substring: contenuto nuovo resta intatto invece di |
| 122 | // essere corrotto da chiavi corte accumulate nel DB dalle incremental translations. |
| 123 | // $incr e' il contesto precalcolato una volta per pagina (vedi gpt_build_incremental_context). |
| 124 | function gpt_incremental_translate_text($text, $incr) { |
| 125 | // Extract leading/trailing whitespace from original text |
| 126 | preg_match('/^(\s*)(.*?)(\s*)$/s', $text, $wsMatches); |
| 127 | $leadingWhitespace = $wsMatches[1] ?? ''; |
| 128 | $trailingWhitespace = $wsMatches[3] ?? ''; |
| 129 | |
| 130 | // Apply dictionary replacement for incremental matching (matches client-side behavior) |
| 131 | $normalizedText = normalizeTextForIncremental($text, $incr['settings'], $incr['originalLang'], $incr['translatedLang']); |
| 132 | if ($incr['caseInsensitive']) { |
| 133 | $normalizedText = mb_strtolower($normalizedText, 'UTF-8'); |
| 134 | } |
| 135 | $normalizedText = trim($normalizedText); |
| 136 | |
| 137 | // Step 1: Exact match |
| 138 | if (isset($incr['lookup'][$normalizedText])) { |
| 139 | return $leadingWhitespace . $incr['lookup'][$normalizedText] . $trailingWhitespace; |
| 140 | } |
| 141 | |
| 142 | // Step 2: Excluded-words-aware fragment matching |
| 143 | // Mirrors client-side wrap_excluded_words behavior: client creates SEPARATE text nodes |
| 144 | // around excluded words, so DB stores fragments separately, NOT the full text. |
| 145 | // Example: "Now i add a new text and modify the page" with 'text' excluded |
| 146 | // Client creates 3 nodes: "Now i add a new ", "text" (preserved), " and modify the page" |
| 147 | // DB stores: "Now i add a new " -> "Ora aggiungo un nuovo ", " and modify the page" -> " e modificare la pagina" |
| 148 | // We split the text by excluded words, translate each fragment, preserve excluded words. |
| 149 | if ($incr['excludedPattern'] === '') { |
| 150 | return $text; |
| 151 | } |
| 152 | |
| 153 | // Split keeping excluded words as delimiters (PREG_SPLIT_DELIM_CAPTURE). |
| 154 | // Always case-insensitive here, regardless of $incr['caseInsensitive']: the client |
| 155 | // protects/wraps excluded words with a hardcoded case-insensitive match (ssProtectExcludedWords |
| 156 | // and the original wrap_excluded_words splitting both use the 'i' flag unconditionally), so a |
| 157 | // mixed-case source word (e.g. "Magnets" ... "magnets") is already split into separate DB |
| 158 | // fragments on the client side. Splitting here case-sensitively would leave the lowercase |
| 159 | // occurrences bundled into a larger segment that matches no stored key, and the paragraph |
| 160 | // would fall through untranslated even though every piece is already known. |
| 161 | $parts = preg_split( |
| 162 | '/((?<!\w)(?:' . $incr['excludedPattern'] . ')(?!\w))/ui', |
| 163 | $text, |
| 164 | -1, |
| 165 | PREG_SPLIT_DELIM_CAPTURE |
| 166 | ); |
| 167 | |
| 168 | if (!is_array($parts) || count($parts) < 2) { |
| 169 | return $text; |
| 170 | } |
| 171 | |
| 172 | $rebuiltText = ''; |
| 173 | $anyTranslated = false; |
| 174 | |
| 175 | foreach ($parts as $part) { |
| 176 | if ($part === '') { |
| 177 | continue; |
| 178 | } |
| 179 | |
| 180 | // Check if this part is an excluded word - preserve as-is |
| 181 | $excludedSetKey = strtolower(trim($part)); |
| 182 | if (isset($incr['excludedWordsSet'][$excludedSetKey])) { |
| 183 | $replacement = $incr['replacementMap'][$incr['excludedWordsSet'][$excludedSetKey]] ?? ''; |
| 184 | if ($replacement !== '') { |
| 185 | $rebuiltText .= $replacement; |
| 186 | $anyTranslated = true; |
| 187 | } else { |
| 188 | $rebuiltText .= $part; |
| 189 | } |
| 190 | continue; |
| 191 | } |
| 192 | |
| 193 | // Try to find an exact match for this fragment in DB |
| 194 | // Capture the fragment's own leading/trailing whitespace before trimming for the |
| 195 | // lookup key, same as the whole-text match above: the stored value never carries |
| 196 | // this whitespace (it was trimmed when saved), so pasting it in without restoring |
| 197 | // the fragment's original spacing glues it straight onto the adjacent word (e.g. |
| 198 | // "Magnetsysteme" + " und Baugruppen" -> "sistemi magnetici" + "e gruppi" becomes |
| 199 | // "sistemi magneticie gruppi" with no space at the join). |
| 200 | preg_match('/^(\s*)(.*?)(\s*)$/s', $part, $partWsMatches); |
| 201 | $partLeadingWhitespace = $partWsMatches[1] ?? ''; |
| 202 | $partTrailingWhitespace = $partWsMatches[3] ?? ''; |
| 203 | |
| 204 | $normalizedPart = normalizeTextForIncremental($part, $incr['settings'], $incr['originalLang'], $incr['translatedLang']); |
| 205 | if ($incr['caseInsensitive']) { |
| 206 | $normalizedPart = mb_strtolower($normalizedPart, 'UTF-8'); |
| 207 | } |
| 208 | $normalizedPart = trim($normalizedPart); |
| 209 | |
| 210 | if ($normalizedPart !== '' && isset($incr['lookup'][$normalizedPart])) { |
| 211 | $rebuiltText .= $partLeadingWhitespace . $incr['lookup'][$normalizedPart] . $partTrailingWhitespace; |
| 212 | $anyTranslated = true; |
| 213 | } else { |
| 214 | // Fragment not found in DB - keep original |
| 215 | $rebuiltText .= $part; |
| 216 | } |
| 217 | } |
| 218 | |
| 219 | return $anyTranslated ? $rebuiltText : $text; |
| 220 | } |
| 221 | |
| 222 | ob_start ( function ($html) use ($uri, $settings) { |
| 223 | global $wpdb; |
| 224 | |
| 225 | $originalLang = $settings ['language'] ?? ''; |
| 226 | |
| 227 | if (! defined ( 'GPTRANSLATE_CURRENT_LANG' )) { |
| 228 | return $html; |
| 229 | } |
| 230 | |
| 231 | $translatedLang = GPTRANSLATE_CURRENT_LANG; |
| 232 | |
| 233 | if (empty ( $originalLang ) || $originalLang === $translatedLang) |
| 234 | return $html; |
| 235 | |
| 236 | if ($settings['subfolder_installation'] && ! $settings ['rewrite_language_alias']) { |
| 237 | $uri_parts = explode('/', ltrim($uri, '/')); |
| 238 | |
| 239 | // Remove the first part (subfolder name) |
| 240 | $subfolder_prefix = '/' . array_shift($uri_parts); |
| 241 | |
| 242 | // Rebuild URI without subfolder |
| 243 | $uri = '/' . implode('/', $uri_parts); |
| 244 | } |
| 245 | |
| 246 | $pageLink = rtrim ( get_site_url (), '/' ) . '/' . ltrim ( GPTRANSLATE_CURRENT_LANG, '/' ) . $uri; |
| 247 | |
| 248 | if ( $settings ['serverside_translations_urldecode'] ) { |
| 249 | $pageLink = urldecode ( $pageLink ); |
| 250 | } |
| 251 | |
| 252 | if ($settings ['serverside_translations_ignore_querystring'] == 1) { |
| 253 | // Remove query string |
| 254 | $pageLink = strtok ( $pageLink, '?' ); |
| 255 | } elseif (! empty ( $_SERVER ['QUERY_STRING'] )) { |
| 256 | // Sanitize and rebuild query string |
| 257 | $raw_qs = isset($_SERVER['QUERY_STRING']) ? wp_unslash($_SERVER['QUERY_STRING']) : ''; // phpcs:ignore WordPress.Security.ValidatedSanitizedInput.InputNotSanitized |
| 258 | |
| 259 | parse_str ( $raw_qs, $qs_args ); |
| 260 | |
| 261 | $clean_args = [ ]; |
| 262 | foreach ( $qs_args as $key => $value ) { |
| 263 | $clean_key = sanitize_key ( $key ); |
| 264 | $clean_value = is_array ( $value ) ? array_map ( 'sanitize_text_field', $value ) : sanitize_text_field ( $value ); |
| 265 | $clean_args [$clean_key] = $clean_value; |
| 266 | } |
| 267 | |
| 268 | $clean_qs = http_build_query ( $clean_args, '', '&', PHP_QUERY_RFC3986 ); |
| 269 | |
| 270 | if ($clean_qs !== '') { |
| 271 | $pageLink = strtok ( $pageLink, '?' ) . '?' . $clean_qs; |
| 272 | } |
| 273 | } |
| 274 | |
| 275 | // Strip entire query string if any server-injected parameter is detected (e.g. q, fbclid) |
| 276 | // Acts as a conditional ignore_querystring=1 triggered by specific parameter names |
| 277 | if (!empty($settings['serverside_translations_strip_querystring_params']) && strpos($pageLink, '?') !== false) { |
| 278 | $stripParams = array_map('trim', explode(',', strtolower($settings['serverside_translations_strip_querystring_params']))); |
| 279 | $pageLinkQs = parse_url($pageLink, PHP_URL_QUERY); |
| 280 | if ($pageLinkQs) { |
| 281 | parse_str($pageLinkQs, $qsArgs); |
| 282 | foreach ($stripParams as $param) { |
| 283 | if (isset($qsArgs[$param])) { |
| 284 | $pageLink = strtok($pageLink, '?'); |
| 285 | break; |
| 286 | } |
| 287 | } |
| 288 | } |
| 289 | } |
| 290 | |
| 291 | if ( $settings ['serverside_translations_urlencode_space'] ) { |
| 292 | $pageLink = str_ireplace ( ' ', '%20', $pageLink ); |
| 293 | } |
| 294 | |
| 295 | $table = $wpdb->prefix . 'gptranslate'; |
| 296 | |
| 297 | if( $settings ['rewrite_language_url'] && $settings ['rewrite_language_alias'] ) { |
| 298 | $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 |
| 299 | "\n WHERE ( pagelink = %s OR pagelink = %s OR translated_alias = %s OR translated_alias = %s) AND languageoriginal = %s AND languagetranslated = %s AND published = 1", |
| 300 | rtrim($pageLink, '/'), rtrim($pageLink, '/') . '/', rtrim($pageLink, '/'), rtrim($pageLink, '/') . '/', $originalLang, $translatedLang ), ARRAY_A ); |
| 301 | } else { |
| 302 | $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 |
| 303 | "\n WHERE ( pagelink = %s OR pagelink = %s ) AND languageoriginal = %s AND languagetranslated = %s AND published = 1", |
| 304 | rtrim($pageLink, '/'), rtrim($pageLink, '/') . '/', $originalLang, $translatedLang ), ARRAY_A ); |
| 305 | } |
| 306 | |
| 307 | if (! $row) { |
| 308 | return $html; |
| 309 | } |
| 310 | |
| 311 | // Load translated aliases to replace on page links |
| 312 | $translatedAliasesMap = []; // Map absolute URLs |
| 313 | $translatedAliasesRelativeMap = []; // Map relative URL (pathname) |
| 314 | |
| 315 | // Store settings for keep context flatten tags |
| 316 | $flattenInnerFormattingTags = isset($settings['flatten_inner_formatting_tags']) ? $settings['flatten_inner_formatting_tags'] : 0; |
| 317 | $flattenInnerFormattingTagsToRemove = isset($settings['flatten_inner_formatting_tags_to_remove']) ? explode(',', $settings['flatten_inner_formatting_tags_to_remove']) : explode(',', 'strong,em,u,b,i'); |
| 318 | $wrapExcludedWords = isset($settings['wrap_excluded_words']) ? $settings['wrap_excluded_words'] : 0; |
| 319 | |
| 320 | if (!empty($settings['rewrite_page_links']) && !empty($settings['rewrite_language_url']) && !empty($settings['rewrite_language_alias'])) { |
| 321 | try { |
| 322 | $aliasesResult = $wpdb->get_results( $wpdb->prepare( |
| 323 | "SELECT pagelink, translated_alias" . |
| 324 | "\n FROM {$table}" . |
| 325 | "\n WHERE languagetranslated = %s" . |
| 326 | "\n AND published = 1", |
| 327 | $translatedLang |
| 328 | ), ARRAY_A ); |
| 329 | |
| 330 | if ($aliasesResult) { |
| 331 | $parsedRoot = trailingslashit(get_site_url()); |
| 332 | |
| 333 | foreach ($aliasesResult as $rowAliasResult) { |
| 334 | // NORMALIZE, WORDPRESS STYLE (KEEP TRAILING SLASH) |
| 335 | $pagelink = gpt_trailingslashit_url($rowAliasResult['pagelink']); |
| 336 | $translatedAlias = $rowAliasResult['translated_alias'] ?? ''; |
| 337 | |
| 338 | if (empty($translatedAlias)) continue; |
| 339 | |
| 340 | $translatedAlias = gpt_trailingslashit_url($translatedAlias); |
| 341 | |
| 342 | // First pass: use decoded pagelink and alias as key (DB stores decoded UTF-8) |
| 343 | $translatedAliasesMap[$pagelink] = $translatedAlias; |
| 344 | |
| 345 | // Second pass: use decoded relative path as key |
| 346 | $relativePath = str_replace($parsedRoot, '/', $pagelink); |
| 347 | $relativePath = parse_url($relativePath, PHP_URL_PATH); |
| 348 | if ($relativePath) { |
| 349 | $translatedAliasesRelativeMap[gpt_trailingslashit_url($relativePath)] = $translatedAlias; |
| 350 | } |
| 351 | } |
| 352 | } |
| 353 | } catch (Exception $e) { |
| 354 | // Silently fail |
| 355 | } |
| 356 | } |
| 357 | |
| 358 | // Check if incremental mode is enabled and preserve new texts option is active |
| 359 | $incrementalEnabledPreserveNewTexts = !empty($settings['incremental_enabled']) && !empty($settings['incremental_preserve_new_texts']); |
| 360 | |
| 361 | // Load excluded words list for incremental preserve mode |
| 362 | // When incremental_preserve_new_texts is ON, we use excluded words to do a "partial greedy match": |
| 363 | // - strip excluded words from the segment, then compare remainder with DB keys |
| 364 | // - this lets "Test new post" (Test=excluded) match DB key "new post" → "nuovo post" |
| 365 | // - without accidentally applying partial translations like "now"→"ora" to new untranslated text |
| 366 | $excludedWordsList = []; |
| 367 | $excludedWordsReplacementMap = []; |
| 368 | if ($incrementalEnabledPreserveNewTexts) { |
| 369 | // 1. Simple excluded words (words_leafnodes_excluded) |
| 370 | $rawExcluded = $settings['words_leafnodes_excluded'] ?? ''; |
| 371 | $rawExcluded = str_ireplace(["\r", "\n"], ",", $rawExcluded); |
| 372 | $rawExcluded = preg_replace('/,+/', ',', $rawExcluded); |
| 373 | $excludedWordsList = array_filter(array_map('trim', explode(',', $rawExcluded))); |
| 374 | foreach ($excludedWordsList as $word) { |
| 375 | $key = mb_strtolower(trim($word), 'UTF-8'); |
| 376 | if ($key !== '' && !isset($excludedWordsReplacementMap[$key])) { |
| 377 | $excludedWordsReplacementMap[$key] = ''; |
| 378 | } |
| 379 | } |
| 380 | |
| 381 | // 2. Dictionary (words_leafnodes_excluded_bylanguage_repeatable) |
| 382 | // Each entry has: word (CSV), langOriginal, langTranslated, optionalTranslation |
| 383 | // Only include entries that match the current language pair (* = any language) |
| 384 | $dictionaryRaw = $settings['words_leafnodes_excluded_bylanguage_repeatable'] ?? '[]'; |
| 385 | if (is_string($dictionaryRaw)) { |
| 386 | $dictionary = json_decode($dictionaryRaw, true) ?? []; |
| 387 | } else { |
| 388 | $dictionary = (array) $dictionaryRaw; |
| 389 | } |
| 390 | foreach ($dictionary as $entry) { |
| 391 | $entryLangOrig = $entry['langOriginal'] ?? '*'; |
| 392 | $entryLangTarget = $entry['langTranslated'] ?? '*'; |
| 393 | $matchesOrig = ($entryLangOrig === '*' || $entryLangOrig === $originalLang); |
| 394 | $matchesTarget = ($entryLangTarget === '*' || $entryLangTarget === $translatedLang); |
| 395 | if ($matchesOrig && $matchesTarget) { |
| 396 | $words = array_filter(array_map('trim', explode(',', $entry['word'] ?? ''))); |
| 397 | $translations = array_map('trim', explode(',', $entry['optionalTranslation'] ?? '')); |
| 398 | foreach ($words as $i => $word) { |
| 399 | if (!empty($word)) { |
| 400 | $excludedWordsList[] = $word; |
| 401 | $key = mb_strtolower(trim($word), 'UTF-8'); |
| 402 | $replacement = isset($translations[$i]) ? $translations[$i] : ''; |
| 403 | if ($key !== '' && !isset($excludedWordsReplacementMap[$key])) { |
| 404 | $excludedWordsReplacementMap[$key] = $replacement; |
| 405 | } elseif ($key !== '' && $excludedWordsReplacementMap[$key] === '' && $replacement !== '') { |
| 406 | $excludedWordsReplacementMap[$key] = $replacement; |
| 407 | } |
| 408 | } |
| 409 | } |
| 410 | } |
| 411 | } |
| 412 | |
| 413 | $excludedWordsList = array_unique($excludedWordsList); |
| 414 | } |
| 415 | |
| 416 | // Precalcola una sola volta per pagina tutto cio' che serve all'incremental exact-match. |
| 417 | // Prima la normalizzazione col dictionary veniva rifatta per ogni chiave del DB su ogni |
| 418 | // nodo di testo, rendendo il costo O(nodi x traduzioni x dictionary): su pagine lunghe |
| 419 | // con molte traduzioni accumulate questo poteva mandare la richiesta in timeout. |
| 420 | $incrementalContext = null; |
| 421 | if ($incrementalEnabledPreserveNewTexts) { |
| 422 | $incrCaseInsensitive = ! empty ( $settings ['serverside_translations_caseinsensitive'] ); |
| 423 | |
| 424 | $incrTranslations = json_decode ( $row ['translations'], true ) ?? [ ]; |
| 425 | // NO sort here: when two DB keys normalize to the same value, the original |
| 426 | // sequential loop picked whichever came FIRST in json_decode's (insertion) order, |
| 427 | // and "first wins" below only reproduces that if the array keeps its natural order. |
| 428 | $incrementalLookup = []; |
| 429 | foreach ( $incrTranslations as $originalText => $translatedText ) { |
| 430 | $normalizedOrig = normalizeTextForIncremental($originalText, $settings, $originalLang, $translatedLang); |
| 431 | if ($incrCaseInsensitive) { |
| 432 | $normalizedOrig = mb_strtolower($normalizedOrig, 'UTF-8'); |
| 433 | } |
| 434 | $normalizedOrig = trim($normalizedOrig); |
| 435 | if (!isset($incrementalLookup[$normalizedOrig])) { |
| 436 | $incrementalLookup[$normalizedOrig] = $translatedText; |
| 437 | } |
| 438 | } |
| 439 | |
| 440 | // La membership test originale usava strcasecmp (ASCII), mentre la chiave della |
| 441 | // replacement map e' mb_strtolower: la mappa qui conserva entrambe le forme. |
| 442 | // Ordina per lunghezza decrescente prima di costruire l'alternanza regex: altrimenti |
| 443 | // una voce piu' corta (es. "Haftkraft") puo' precedere nell'alternanza una voce piu' |
| 444 | // lunga e piu' specifica che la contiene come prefisso (es. "Haftkraft berechnen"), |
| 445 | // e la regex si ferma al primo match valido invece che al piu' lungo/corretto. |
| 446 | $excludedWordsListSorted = $excludedWordsList; |
| 447 | usort($excludedWordsListSorted, fn($a, $b) => mb_strlen($b, 'UTF-8') - mb_strlen($a, 'UTF-8')); |
| 448 | |
| 449 | $incrementalExcludedWordsSet = []; |
| 450 | $incrementalExcludedPattern = ''; |
| 451 | foreach ($excludedWordsListSorted as $excludedWord) { |
| 452 | $excludedWord = trim($excludedWord); |
| 453 | if ($excludedWord === '') { |
| 454 | continue; |
| 455 | } |
| 456 | $incrementalExcludedPattern .= ($incrementalExcludedPattern ? '|' : '') . preg_quote($excludedWord, '/'); |
| 457 | $setKey = strtolower($excludedWord); |
| 458 | if (!isset($incrementalExcludedWordsSet[$setKey])) { |
| 459 | $incrementalExcludedWordsSet[$setKey] = mb_strtolower($excludedWord, 'UTF-8'); |
| 460 | } |
| 461 | } |
| 462 | |
| 463 | $incrementalContext = [ |
| 464 | 'settings' => $settings, |
| 465 | 'originalLang' => $originalLang, |
| 466 | 'translatedLang' => $translatedLang, |
| 467 | 'caseInsensitive' => $incrCaseInsensitive, |
| 468 | 'lookup' => $incrementalLookup, |
| 469 | 'excludedPattern' => $incrementalExcludedPattern, |
| 470 | 'excludedWordsSet' => $incrementalExcludedWordsSet, |
| 471 | 'replacementMap' => $excludedWordsReplacementMap, |
| 472 | ]; |
| 473 | } |
| 474 | |
| 475 | // Process replacement method |
| 476 | if ($settings ['serverside_translations_method'] == 'regex') { |
| 477 | $translations = json_decode ( $row ['translations'], true ) ?? [ ]; |
| 478 | $altTranslations = json_decode ( $row ['alt_translations'], true ) ?? [ ]; |
| 479 | uksort ( $translations, fn ($a, $b) => strlen ( $b ) - strlen ( $a ) ); |
| 480 | uksort ( $altTranslations, fn ($a, $b) => strlen ( $b ) - strlen ( $a ) ); |
| 481 | |
| 482 | // Flatten formatting tags in the body to improve matching if enabled |
| 483 | if ($flattenInnerFormattingTags) { |
| 484 | $regexTagsToRemove = implode('|', array_map('preg_quote', $flattenInnerFormattingTagsToRemove)); |
| 485 | function gp_flatten_formatting_tags_regex($html, $regexTagsToRemove) { |
| 486 | $pattern = '~<(' . $regexTagsToRemove . ')(\s[^>]*)?>(.*?)</\1>~is'; |
| 487 | |
| 488 | // Ripeti finché non ci sono più match (gestisce annidamenti semplici) |
| 489 | $prev = null; |
| 490 | while ($prev !== $html) { |
| 491 | $prev = $html; |
| 492 | $html = preg_replace_callback($pattern, function ($m) { |
| 493 | // Preserva elementi che contengono form elements o media (select, input, textarea, button, img, ecc.) |
| 494 | if (preg_match('/<(select|input|textarea|button|option|optgroup|fieldset|datalist|output|label|form|img|video|audio|canvas|svg|iframe)[\s>\/]/i', $m[3])) { |
| 495 | return $m[0]; // restituisci il tag originale intatto |
| 496 | } |
| 497 | // Preserva elementi vuoti (es. <i> FontAwesome/icon fonts) |
| 498 | if (trim(str_replace([' ', "\xc2\xa0", "\xa0"], '', strip_tags($m[3]))) === '') { |
| 499 | return $m[0]; |
| 500 | } |
| 501 | // tieni SOLO il testo interno (se dentro c’è altro HTML lo butta via) |
| 502 | $inner = strip_tags($m[3]); |
| 503 | return $inner; |
| 504 | }, $html); |
| 505 | } |
| 506 | |
| 507 | return $html; |
| 508 | } |
| 509 | $html = gp_flatten_formatting_tags_regex($html, $regexTagsToRemove); |
| 510 | } |
| 511 | |
| 512 | $caseInsensitive = ! empty ( $settings ['serverside_translations_caseinsensitive'] ); |
| 513 | $matchQuotes = ! empty ( $settings ['serverside_translations_matchquotes'] ); |
| 514 | $excludedPatterns = [ ]; |
| 515 | |
| 516 | $excludedCss = preg_replace ( '/,+/', ',', str_ireplace ( [ |
| 517 | "\r", |
| 518 | "\n", |
| 519 | '"' |
| 520 | ], [ |
| 521 | ',', |
| 522 | ',', |
| 523 | '' |
| 524 | ], $settings ['css_selector_serverside_leafnodes_excluded'] ?? '') ); |
| 525 | $excludedCss = array_filter ( array_map ( 'trim', explode ( ',', $excludedCss ) ) ); |
| 526 | |
| 527 | foreach ( $excludedCss as $selector ) { |
| 528 | if (preg_match ( '/^([a-z0-9]+)\.(.+)$/i', $selector, $m )) { |
| 529 | $excludedPatterns [] = '/<' . preg_quote ( $m [1], '/' ) . '(?=[^>]*\sclass\s*=\s*["\'][^"\']*\b' . preg_quote ( $m [2], '/' ) . '\b)[^>]*>/i'; |
| 530 | } elseif (preg_match ( '/^\.(.+)$/', $selector, $m )) { |
| 531 | $excludedPatterns [] = '/<([a-z0-9]+)(?=[^>]*\sclass\s*=\s*["\'][^"\']*\b' . preg_quote ( $m [1], '/' ) . '\b)[^>]*>/i'; |
| 532 | } |
| 533 | } |
| 534 | |
| 535 | // PROTECT script/style content before splitting (if option enabled) |
| 536 | $protectedScriptsStyles = []; |
| 537 | |
| 538 | if (!empty($settings['protect_script_style'])) { |
| 539 | $protectCounter = 0; |
| 540 | |
| 541 | // Extract <script>...</script> tags |
| 542 | $html = preg_replace_callback( |
| 543 | '/<script\b[^>]*>.*?<\/script>/is', |
| 544 | function ($matches) use (&$protectedScriptsStyles, &$protectCounter) { |
| 545 | $placeholder = "___GPTRANSLATE_SCRIPT_" . ($protectCounter++) . "___"; |
| 546 | $protectedScriptsStyles[$placeholder] = $matches[0]; |
| 547 | return $placeholder; |
| 548 | }, |
| 549 | $html |
| 550 | ); |
| 551 | |
| 552 | // Extract <style>...</style> tags |
| 553 | $html = preg_replace_callback( |
| 554 | '/<style\b[^>]*>.*?<\/style>/is', |
| 555 | function ($matches) use (&$protectedScriptsStyles, &$protectCounter) { |
| 556 | $placeholder = "___GPTRANSLATE_STYLE_" . ($protectCounter++) . "___"; |
| 557 | $protectedScriptsStyles[$placeholder] = $matches[0]; |
| 558 | return $placeholder; |
| 559 | }, |
| 560 | $html |
| 561 | ); |
| 562 | } |
| 563 | |
| 564 | $segments = preg_split ( '/(<[^>]+>)/i', $html, - 1, PREG_SPLIT_DELIM_CAPTURE ); |
| 565 | |
| 566 | $skipStack = [ ]; |
| 567 | foreach ( $segments as $index => $segment ) { |
| 568 | if (preg_match ( '/^<\s*(script|style)(\s|>)/i', $segment, $matches )) { |
| 569 | $skipStack [] = strtolower ( $matches [1] ); |
| 570 | } elseif (preg_match ( '/<\/\s*(script|style)[^>]*>/i', $segment, $matches )) { // <--- FIX QUI |
| 571 | $tag = strtolower ( $matches [1] ); |
| 572 | if (! empty ( $skipStack ) && end ( $skipStack ) === $tag) { |
| 573 | array_pop ( $skipStack ); |
| 574 | } |
| 575 | } elseif (preg_match ( '/^<\s*([a-zA-Z0-9]+)/', $segment, $tagMatch )) { |
| 576 | $tagName = strtolower ( $tagMatch [1] ); |
| 577 | foreach ( $excludedPatterns as $pattern ) { |
| 578 | if (preg_match ( $pattern, $segment )) { |
| 579 | $skipStack [] = $tagName; |
| 580 | break; |
| 581 | } |
| 582 | } |
| 583 | } elseif (preg_match ( '/^<\/\s*([a-zA-Z0-9]+)/', $segment, $tagMatch )) { |
| 584 | $tagName = strtolower ( $tagMatch [1] ); |
| 585 | if (! empty ( $skipStack ) && end ( $skipStack ) === $tagName) { |
| 586 | array_pop ( $skipStack ); |
| 587 | } |
| 588 | } elseif (! preg_match ( '/^<[^>]+>$/', $segment )) { |
| 589 | if (empty ( $skipStack )) { |
| 590 | if ($matchQuotes) { |
| 591 | $segment = str_ireplace ( '"', "'", $segment ); |
| 592 | } |
| 593 | // Normalize once without trimming to preserve spaces between nodes |
| 594 | $segment = normalizeTextNoTrim($segment); |
| 595 | |
| 596 | // INCREMENTAL FIX: Use exact match only if incremental preserve new texts is enabled |
| 597 | if ($incrementalEnabledPreserveNewTexts) { |
| 598 | $segment = gpt_incremental_translate_text($segment, $incrementalContext); |
| 599 | } else { |
| 600 | // Normal pattern matching (standard behavior) |
| 601 | foreach ( $translations as $originalText => $translatedText ) { |
| 602 | $originalText = normalizeText($originalText); |
| 603 | |
| 604 | $prevSegment = $segment; |
| 605 | if ($wrapExcludedWords) { |
| 606 | // Use word boundaries to avoid sub-replacing inside already translated words |
| 607 | $pattern = '/' . (preg_match('/^\w/u', trim($originalText)) ? '(?<!\w)' : '') . preg_quote(trim($originalText), '/') . '(?!\w)/' . ($caseInsensitive ? 'ui' : 'u'); |
| 608 | $segment = preg_replace($pattern, $translatedText, $segment); |
| 609 | } else { |
| 610 | // Use word boundaries to avoid sub-replacing inside already translated words |
| 611 | $pattern = '/' . (preg_match('/^\w/u', trim($originalText)) ? '(?<!\w)' : '') . preg_quote(trim($originalText), '/') . '(?!\w)/' . ($caseInsensitive ? 'ui' : 'u'); |
| 612 | $segment = preg_replace($pattern, $translatedText, $segment); |
| 613 | if ($segment !== $prevSegment) { |
| 614 | break; |
| 615 | } |
| 616 | } |
| 617 | } |
| 618 | } |
| 619 | } |
| 620 | } |
| 621 | |
| 622 | $segments [$index] = $segment; |
| 623 | } |
| 624 | |
| 625 | $html = implode ( '', $segments ); |
| 626 | |
| 627 | // RESTORE protected script/style content (if protection was enabled) |
| 628 | if (!empty($settings['protect_script_style'])) { |
| 629 | foreach ($protectedScriptsStyles as $placeholder => $originalContent) { |
| 630 | $html = str_replace($placeholder, $originalContent, $html); |
| 631 | } |
| 632 | } |
| 633 | |
| 634 | // Replace alt and title in images |
| 635 | if (! empty ( $settings ['translate_altimages'] )) { |
| 636 | $html = preg_replace_callback ( '/<img[^>]*\b(alt|title)\s*=\s*([\"\'])(.*?)\2[^>]*>/i', function ($matches) use ($altTranslations, $caseInsensitive, $matchQuotes) { |
| 637 | $attr = $matches [1]; |
| 638 | $quote = $matches [2]; |
| 639 | $value = $matches [3]; |
| 640 | if ($matchQuotes) |
| 641 | $value = str_ireplace ( '"', "'", $value ); |
| 642 | foreach ( $altTranslations as $original => $translated ) { |
| 643 | $original = normalizeText($original); |
| 644 | $value = normalizeText($value); |
| 645 | |
| 646 | $value = $caseInsensitive ? str_ireplace ( trim ( $original ), $translated, $value ) : str_replace ( trim ( $original ), $translated, $value ); |
| 647 | } |
| 648 | return preg_replace ( '/\b' . $attr . '\s*=\s*["\'].*?["\']/', "$attr=$quote$value$quote", $matches [0] ); |
| 649 | }, $html ); |
| 650 | |
| 651 | // Check if also images 'src' are enabled to be translated |
| 652 | if (! empty ( $settings ['translate_srcimages'] )) { |
| 653 | $html = preg_replace_callback('/<img\b[^>]*\bsrc\s*=\s*(["\'])(.*?)\1[^>]*>/i', function ($matches) use ($altTranslations, $caseInsensitive, $matchQuotes) { |
| 654 | $originalTag = $matches[0]; |
| 655 | $quote = $matches[1]; |
| 656 | $srcValue = ltrim($matches[2], '/'); |
| 657 | |
| 658 | if ($matchQuotes) { |
| 659 | $srcValue = str_ireplace('"', "'", $srcValue); |
| 660 | } |
| 661 | |
| 662 | $translatedSrcValue = null; |
| 663 | foreach ($altTranslations as $originalText => $translatedText) { |
| 664 | $originalText = ltrim($originalText, '/'); |
| 665 | if ($caseInsensitive) { |
| 666 | if (strcasecmp(trim($originalText), $srcValue) === 0) { |
| 667 | $translatedSrcValue = $translatedText; |
| 668 | break; |
| 669 | } |
| 670 | } else { |
| 671 | if (trim($originalText) === $srcValue) { |
| 672 | $translatedSrcValue = $translatedText; |
| 673 | break; |
| 674 | } |
| 675 | } |
| 676 | } |
| 677 | |
| 678 | // If a translation was found, replace src and srcset |
| 679 | if ($translatedSrcValue !== null) { |
| 680 | // Replace the src attribute |
| 681 | $modifiedTag = preg_replace('/\bsrc\s*=\s*["\'].*?["\']/i', 'src=' . $quote . $translatedSrcValue . $quote, $originalTag); |
| 682 | |
| 683 | // Check if srcset exists and replace it with the translated src value |
| 684 | if (preg_match('/\bsrcset\s*=\s*(["\'])(.*?)\1/i', $modifiedTag, $srcsetMatch)) { |
| 685 | $srcsetQuote = $srcsetMatch[1]; |
| 686 | $modifiedTag = preg_replace('/\bsrcset\s*=\s*["\'].*?["\']/i', 'srcset=' . $srcsetQuote . $translatedSrcValue . $srcsetQuote, $modifiedTag); |
| 687 | } |
| 688 | |
| 689 | return $modifiedTag; |
| 690 | } |
| 691 | |
| 692 | return $originalTag; |
| 693 | }, $html); |
| 694 | } |
| 695 | |
| 696 | // Check if also iframes 'src' are enabled to be translated |
| 697 | if (! empty ( $settings ['translate_srciframes'] )) { |
| 698 | $html = preg_replace_callback('/<iframe\b[^>]*\bsrc\s*=\s*(["\'])(.*?)\1[^>]*>/i', function ($matches) use ($altTranslations, $caseInsensitive, $matchQuotes) { |
| 699 | $originalTag = $matches[0]; |
| 700 | $quote = $matches[1]; |
| 701 | $srcValue = ltrim($matches[2], '/'); |
| 702 | |
| 703 | if ($matchQuotes) { |
| 704 | $srcValue = str_ireplace('"', "'", $srcValue); |
| 705 | } |
| 706 | |
| 707 | $translatedSrcValue = null; |
| 708 | foreach ($altTranslations as $originalText => $translatedText) { |
| 709 | $originalText = ltrim($originalText, '/'); |
| 710 | if ($caseInsensitive) { |
| 711 | if (strcasecmp(trim($originalText), $srcValue) === 0) { |
| 712 | $translatedSrcValue = $translatedText; |
| 713 | break; |
| 714 | } |
| 715 | } else { |
| 716 | if (trim($originalText) === $srcValue) { |
| 717 | $translatedSrcValue = $translatedText; |
| 718 | break; |
| 719 | } |
| 720 | } |
| 721 | } |
| 722 | |
| 723 | if ($translatedSrcValue !== null) { |
| 724 | return preg_replace('/\bsrc\s*=\s*["\'].*?["\']/i', 'src=' . $quote . $translatedSrcValue . $quote, $originalTag); |
| 725 | } |
| 726 | |
| 727 | return $originalTag; |
| 728 | }, $html); |
| 729 | } |
| 730 | |
| 731 | // Check if also videos 'src' are enabled to be translated |
| 732 | if (! empty ( $settings ['translate_srcvideos'] )) { |
| 733 | $html = preg_replace_callback('/<(video|source)\b[^>]*\bsrc\s*=\s*(["\'])(.*?)\2[^>]*>/i', function ($matches) use ($altTranslations, $caseInsensitive, $matchQuotes) { |
| 734 | $originalTag = $matches[0]; |
| 735 | $quote = $matches[2]; |
| 736 | $srcValue = ltrim($matches[3], '/'); |
| 737 | |
| 738 | if ($matchQuotes) { |
| 739 | $srcValue = str_ireplace('"', "'", $srcValue); |
| 740 | } |
| 741 | |
| 742 | $translatedSrcValue = null; |
| 743 | foreach ($altTranslations as $originalText => $translatedText) { |
| 744 | $originalText = ltrim($originalText, '/'); |
| 745 | if ($caseInsensitive) { |
| 746 | if (strcasecmp(trim($originalText), $srcValue) === 0) { |
| 747 | $translatedSrcValue = $translatedText; |
| 748 | break; |
| 749 | } |
| 750 | } else { |
| 751 | if (trim($originalText) === $srcValue) { |
| 752 | $translatedSrcValue = $translatedText; |
| 753 | break; |
| 754 | } |
| 755 | } |
| 756 | } |
| 757 | |
| 758 | if ($translatedSrcValue !== null) { |
| 759 | return preg_replace('/\bsrc\s*=\s*["\'].*?["\']/i', 'src=' . $quote . $translatedSrcValue . $quote, $originalTag); |
| 760 | } |
| 761 | |
| 762 | return $originalTag; |
| 763 | }, $html); |
| 764 | } |
| 765 | } |
| 766 | |
| 767 | // Translate iframe locale parameter (e.g. Stripe payment forms) |
| 768 | if (! empty ( $settings ['translate_iframe_locale'] )) { |
| 769 | $currentLangCode = strtolower(explode('-', $currentLanguage)[0]); |
| 770 | $html = preg_replace('/(<iframe[^>]*\bsrc=["\'"])([^"\']*?locale=)[a-z]{2}([^"\']*["\'])/i', '$1${2}' . $currentLangCode . '$3', $html); |
| 771 | } |
| 772 | |
| 773 | // Translate <meta name="description" content="..."> |
| 774 | $html = preg_replace_callback( |
| 775 | '~<meta\s+(?:name|property)=["\'](?:description|og:description|twitter:description|dc\.description)["\']\s+content=["\'](.*?)["\'][^>]*>~i', |
| 776 | function ($matches) use ($altTranslations, $caseInsensitive, $matchQuotes) { |
| 777 | $originalTag = $matches[0]; |
| 778 | $contentValue = $matches[1]; |
| 779 | |
| 780 | // DECODIFICA HTML entities PRIMA del confronto |
| 781 | $contentValue = html_entity_decode($contentValue, ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 782 | |
| 783 | if ($matchQuotes) { |
| 784 | $contentValue = str_ireplace('"', "'", $contentValue); |
| 785 | } |
| 786 | |
| 787 | foreach ($altTranslations as $originalText => $translatedText) { |
| 788 | $prev = $contentValue; |
| 789 | if ($caseInsensitive) { |
| 790 | $contentValue = str_ireplace(trim($originalText), $translatedText, $contentValue); |
| 791 | } else { |
| 792 | $contentValue = str_replace(trim($originalText), $translatedText, $contentValue); |
| 793 | } |
| 794 | if ($contentValue !== $prev) break; |
| 795 | } |
| 796 | |
| 797 | // RI-ENCODIFICA per l'HTML output (sicurezza) |
| 798 | $contentValue = htmlspecialchars($contentValue, ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 799 | |
| 800 | return preg_replace( |
| 801 | '/content=["\'].*?["\']/i', |
| 802 | 'content="' . $contentValue . '"', |
| 803 | $originalTag |
| 804 | ); |
| 805 | }, |
| 806 | $html |
| 807 | ); |
| 808 | |
| 809 | // Translate <meta property="og:title"> with the same value as <title> tag |
| 810 | // Step 1: Extract the translated title content |
| 811 | if (preg_match('~<title[^>]*>(.*?)</title>~is', $html, $titleMatch)) { |
| 812 | $translatedTitle = trim(strip_tags($titleMatch[1])); |
| 813 | |
| 814 | // Step 2: Replace og:title content directly |
| 815 | $html = preg_replace( |
| 816 | '~(<meta\s+property=["\']og:title["\']\s+content=["\']).*?(["\'][^>]*>)~i', |
| 817 | '$1' . htmlspecialchars($translatedTitle, ENT_QUOTES, 'UTF-8') . '$2', |
| 818 | $html |
| 819 | ); |
| 820 | |
| 821 | // Step 3: Replace twitter:title content with the same translated title |
| 822 | $html = preg_replace( |
| 823 | '~(<meta\s+name=["\']twitter:title["\']\s+content=["\']).*?(["\'][^>]*>)~i', |
| 824 | '$1' . htmlspecialchars($translatedTitle, ENT_QUOTES, 'UTF-8') . '$2', |
| 825 | $html |
| 826 | ); |
| 827 | |
| 828 | // Step 4: Replace Dublin Core (SEOPress) dc.title content with the same translated title |
| 829 | $html = preg_replace( |
| 830 | '~(<meta\s+name=["\']dc\.title["\']\s+content=["\']).*?(["\'][^>]*>)~i', |
| 831 | '$1' . htmlspecialchars($translatedTitle, ENT_QUOTES, 'UTF-8') . '$2', |
| 832 | $html |
| 833 | ); |
| 834 | } |
| 835 | |
| 836 | // Add skip marker |
| 837 | $html = preg_replace ( '/<body/i', '<body data-gptranslateskip="1" data-gptranslateoriginalalias="' . $row['pagelink'] . '"', $html, 1 ); |
| 838 | |
| 839 | // Replace page links in <a href="..."> tags based on translated aliases |
| 840 | if (!empty($translatedAliasesMap) || !empty($translatedAliasesRelativeMap)) { |
| 841 | $exclusions = $settings['rewrite_page_links_exclusions'] ?? ''; |
| 842 | $page_exclusions = $settings['page_exclusions'] ?? ''; |
| 843 | $html = preg_replace_callback( |
| 844 | '/<a\s+([^>]*\s)?href\s*=\s*(["\'])(.*?)\2([^>]*)>/i', |
| 845 | function($matches) use ($translatedAliasesMap, $translatedAliasesRelativeMap, $settings, $exclusions, $page_exclusions) { |
| 846 | $fullTag = $matches[0]; |
| 847 | $quote = $matches[2]; |
| 848 | $href = $matches[3]; |
| 849 | |
| 850 | // Check if link should be excluded from rewriting |
| 851 | if (gptranslate_should_exclude_link($href, $exclusions, $page_exclusions)) { |
| 852 | return $fullTag; |
| 853 | } |
| 854 | |
| 855 | // Decode href attribute to match database entries |
| 856 | $decodedHref = rawurldecode(html_entity_decode($href, ENT_QUOTES, 'UTF-8')); |
| 857 | |
| 858 | // Extract hash/fragment if present |
| 859 | $hashFragment = ''; |
| 860 | if (strpos($decodedHref, '#') !== false) { |
| 861 | $parts = explode('#', $decodedHref, 2); |
| 862 | $decodedHref = $parts[0]; // URL without hash |
| 863 | $hashFragment = '#' . $parts[1]; // Save the hash |
| 864 | } |
| 865 | |
| 866 | $queryString = ''; |
| 867 | if ($settings ['ignore_querystring'] == 1 && strpos($decodedHref, '?') !== false) { |
| 868 | $parts = explode('?', $decodedHref, 2); |
| 869 | $decodedHref = $parts[0]; |
| 870 | $queryString = '?' . $parts[1]; |
| 871 | } |
| 872 | $originalPathBeforeNormalization = $decodedHref; |
| 873 | |
| 874 | $decodedHref = gpt_trailingslashit_url($decodedHref); |
| 875 | |
| 876 | $translatedAlias = null; |
| 877 | |
| 878 | // First pass, check in absolute URLs |
| 879 | if (isset($translatedAliasesMap[$decodedHref])) { |
| 880 | $translatedAlias = $translatedAliasesMap[$decodedHref]; |
| 881 | } |
| 882 | // Second pass, check in relative URLs (pathname) |
| 883 | elseif (isset($translatedAliasesRelativeMap[$decodedHref])) { |
| 884 | $translatedAlias = $translatedAliasesRelativeMap[$decodedHref]; |
| 885 | } |
| 886 | |
| 887 | // If found a translated alias, replace href and add data-originalhref if not present |
| 888 | if ($translatedAlias) { |
| 889 | $finalHref = $translatedAlias . $queryString . $hashFragment; |
| 890 | |
| 891 | $originalHrefFull = $originalPathBeforeNormalization . $queryString . $hashFragment; |
| 892 | |
| 893 | // Check if data-originalhref already exists in the tag |
| 894 | if (strpos($fullTag, 'data-originalhref') === false) { |
| 895 | $dataAttr = ' data-originalhref=' . $quote . htmlspecialchars($originalHrefFull, ENT_QUOTES, 'UTF-8') . $quote; |
| 896 | $newTag = str_ireplace('<a ', '<a' . $dataAttr . ' ', $fullTag); |
| 897 | $newTag = str_ireplace('href=' . $quote . $href . $quote, 'href=' . $quote . htmlspecialchars($finalHref, ENT_QUOTES, 'UTF-8') . $quote, $newTag); |
| 898 | return $newTag; |
| 899 | } else { |
| 900 | return str_ireplace('href=' . $quote . $href . $quote, 'href=' . $quote . htmlspecialchars($finalHref, ENT_QUOTES, 'UTF-8') . $quote, $fullTag); |
| 901 | } |
| 902 | } |
| 903 | |
| 904 | return $fullTag; |
| 905 | }, |
| 906 | $html |
| 907 | ); |
| 908 | |
| 909 | // Replace form action URLs: add language prefix and apply translated aliases |
| 910 | if (!empty($settings['rewrite_form_actions'])) { |
| 911 | $siteUrl = trailingslashit(get_site_url()); |
| 912 | $siteHost = parse_url($siteUrl, PHP_URL_HOST); |
| 913 | $knownLangs = (isset($settings['languages']) && is_array($settings['languages'])) ? array_map('strtolower', $settings['languages']) : [$originalLang, $translatedLang]; |
| 914 | $html = preg_replace_callback( |
| 915 | '/<form\s+([^>]*\s)?action\s*=\s*(["\'])(.*?)\2([^>]*)>/i', |
| 916 | function($matches) use ($translatedAliasesMap, $translatedAliasesRelativeMap, $settings, $translatedLang, $siteUrl, $siteHost, $knownLangs) { |
| 917 | $fullTag = $matches[0]; |
| 918 | $quote = $matches[2]; |
| 919 | $action = $matches[3]; |
| 920 | |
| 921 | $decodedAction = rawurldecode(html_entity_decode($action, ENT_QUOTES, 'UTF-8')); |
| 922 | |
| 923 | // Skip anchors, mailto, tel, javascript, .php endpoints |
| 924 | if (preg_match('/^(mailto:|tel:|javascript:|#)/i', $decodedAction)) return $fullTag; |
| 925 | if (preg_match('/\.php($|\?|\#)/i', $decodedAction)) return $fullTag; |
| 926 | |
| 927 | // Skip external URLs |
| 928 | if (preg_match('/^https?:\/\//i', $decodedAction)) { |
| 929 | $actionHost = parse_url($decodedAction, PHP_URL_HOST); |
| 930 | if ($actionHost !== $siteHost) return $fullTag; |
| 931 | } |
| 932 | |
| 933 | // Extract query string |
| 934 | $queryString = ''; |
| 935 | if ($settings['ignore_querystring'] == 1 && strpos($decodedAction, '?') !== false) { |
| 936 | $parts = explode('?', $decodedAction, 2); |
| 937 | $decodedAction = $parts[0]; |
| 938 | $queryString = '?' . $parts[1]; |
| 939 | } |
| 940 | $originalActionFull = $decodedAction . $queryString; |
| 941 | |
| 942 | // Extract relative path |
| 943 | $actionPath = $decodedAction; |
| 944 | $isAbsolute = false; |
| 945 | if (preg_match('/^https?:\/\//i', $actionPath)) { |
| 946 | $isAbsolute = true; |
| 947 | $actionPath = parse_url($actionPath, PHP_URL_PATH) ?: '/'; |
| 948 | } |
| 949 | |
| 950 | // Split path and determine language index |
| 951 | $pathParts = explode('/', $actionPath); |
| 952 | $langIndex = 1; |
| 953 | if (isset($pathParts[$langIndex]) && $pathParts[$langIndex] === 'index.php') { |
| 954 | $langIndex = 2; |
| 955 | } |
| 956 | if (!empty($settings['subfolder_installation'])) { |
| 957 | $langIndex = 2; |
| 958 | } |
| 959 | |
| 960 | // Check if already has a language prefix, replace or insert |
| 961 | if (isset($pathParts[$langIndex]) && in_array(strtolower($pathParts[$langIndex]), $knownLangs)) { |
| 962 | $pathParts[$langIndex] = $translatedLang; |
| 963 | } else { |
| 964 | array_splice($pathParts, $langIndex, 0, [$translatedLang]); |
| 965 | } |
| 966 | |
| 967 | // Rebuild path |
| 968 | $rebuiltPath = '/' . implode('/', array_filter($pathParts, function($p) { return $p !== ''; })); |
| 969 | $rebuiltPath = gpt_trailingslashit_url($rebuiltPath); |
| 970 | |
| 971 | // Rebuild full URL if original was absolute |
| 972 | $computedAction = $isAbsolute ? (rtrim($siteUrl, '/') . $rebuiltPath) : $rebuiltPath; |
| 973 | $computedAction = gpt_trailingslashit_url($computedAction); |
| 974 | |
| 975 | // Try alias match |
| 976 | $translatedAlias = null; |
| 977 | if (!empty($translatedAliasesMap) || !empty($translatedAliasesRelativeMap)) { |
| 978 | if (isset($translatedAliasesMap[$computedAction])) { |
| 979 | $translatedAlias = $translatedAliasesMap[$computedAction]; |
| 980 | } elseif (isset($translatedAliasesRelativeMap[$rebuiltPath])) { |
| 981 | $translatedAlias = $translatedAliasesRelativeMap[$rebuiltPath]; |
| 982 | } elseif (!$isAbsolute && !empty($settings['subfolder_installation'])) { |
| 983 | // Fallback: form action is relative and includes subfolder path |
| 984 | // Map keys strip the subfolder, so retry after stripping it |
| 985 | $siteBasePath = rtrim(parse_url(rtrim($siteUrl, '/'), PHP_URL_PATH) ?: '', '/'); |
| 986 | if ($siteBasePath !== '' && strpos($rebuiltPath, $siteBasePath . '/') === 0) { |
| 987 | $rebuiltPathNoSubfolder = substr($rebuiltPath, strlen($siteBasePath)); |
| 988 | if (isset($translatedAliasesRelativeMap[$rebuiltPathNoSubfolder])) { |
| 989 | $translatedAlias = $translatedAliasesRelativeMap[$rebuiltPathNoSubfolder]; |
| 990 | } |
| 991 | } |
| 992 | } |
| 993 | } |
| 994 | |
| 995 | $finalAction = $translatedAlias ? ($translatedAlias . $queryString) : ($computedAction . $queryString); |
| 996 | |
| 997 | if (strpos($fullTag, 'data-originalaction') === false) { |
| 998 | $dataAttr = ' data-originalaction=' . $quote . htmlspecialchars($originalActionFull, ENT_QUOTES, 'UTF-8') . $quote; |
| 999 | $newTag = str_ireplace('<form ', '<form' . $dataAttr . ' ', $fullTag); |
| 1000 | $newTag = str_ireplace('action=' . $quote . $action . $quote, 'action=' . $quote . htmlspecialchars($finalAction, ENT_QUOTES, 'UTF-8') . $quote, $newTag); |
| 1001 | return $newTag; |
| 1002 | } else { |
| 1003 | return str_ireplace('action=' . $quote . $action . $quote, 'action=' . $quote . htmlspecialchars($finalAction, ENT_QUOTES, 'UTF-8') . $quote, $fullTag); |
| 1004 | } |
| 1005 | }, |
| 1006 | $html |
| 1007 | ); |
| 1008 | } // end rewrite_form_actions |
| 1009 | } |
| 1010 | } elseif ($settings ['serverside_translations_method'] == 'domdocument') { |
| 1011 | // Solution 2: classic DOMDocument approach, effective but could cause closing tags and encoding issues |
| 1012 | try { |
| 1013 | $translationsArray = json_decode ( $row['translations'], true ) ?? [ ]; |
| 1014 | $altTranslationsArray = json_decode ( $row['alt_translations'] ?? '', true ) ?: [ ]; |
| 1015 | |
| 1016 | // Sort the translation keys in descending order by length. |
| 1017 | uksort ( $translationsArray, function ( $a, $b ) { |
| 1018 | return strlen ( $b ) - strlen ( $a ); |
| 1019 | } ); |
| 1020 | |
| 1021 | // Sort the alt translation keys in descending order by length. |
| 1022 | uksort ( $altTranslationsArray, function ( $a, $b ) { |
| 1023 | return strlen ( $b ) - strlen ( $a ); |
| 1024 | } ); |
| 1025 | |
| 1026 | // Flatten formatting tags in the body to improve matching if enabled |
| 1027 | if($flattenInnerFormattingTags) { |
| 1028 | function gp_flatten_inner_formatting_tags($html, $tagsToRemove) { |
| 1029 | $dom = new DOMDocument(); |
| 1030 | libxml_use_internal_errors(true); |
| 1031 | |
| 1032 | // Usa lo stesso metodo del codice principale con l'hack XML encoding |
| 1033 | $dom->loadHTML($html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD); |
| 1034 | |
| 1035 | libxml_clear_errors(); |
| 1036 | |
| 1037 | $xpath = new DOMXPath($dom); |
| 1038 | |
| 1039 | // Seleziona tutti i tag da rimuovere, ovunque |
| 1040 | $query = '//' . implode(' | //', array_map(function($t) { return strtolower($t); }, $tagsToRemove)); |
| 1041 | |
| 1042 | $nodes = $xpath->query($query); |
| 1043 | |
| 1044 | // Importante: sostituire dal "basso verso l'alto" per evitare problemi mentre modifichi il DOM |
| 1045 | // Convertiamo NodeList in array e invertiamo |
| 1046 | $toReplace = []; |
| 1047 | foreach ($nodes as $n) |
| 1048 | $toReplace[] = $n; |
| 1049 | $toReplace = array_reverse($toReplace); |
| 1050 | |
| 1051 | foreach ($toReplace as $node) { |
| 1052 | // Preserva elementi vuoti (es. <i> FontAwesome/icon fonts) |
| 1053 | if (trim(str_replace([' ', "\xc2\xa0", "\xa0"], '', $node->textContent)) === '') { |
| 1054 | continue; |
| 1055 | } |
| 1056 | // Preserva elementi che contengono form elements (select, input, textarea, button, ecc.) |
| 1057 | $interactiveTags = ['select', 'input', 'textarea', 'button', 'option', 'optgroup', 'fieldset', 'datalist', 'output', 'label', 'form', 'img', 'video', 'audio', 'canvas', 'svg', 'iframe']; |
| 1058 | $hasInteractive = false; |
| 1059 | foreach ($interactiveTags as $iTag) { |
| 1060 | if ($node->getElementsByTagName($iTag)->length > 0) { |
| 1061 | $hasInteractive = true; |
| 1062 | break; |
| 1063 | } |
| 1064 | } |
| 1065 | if ($hasInteractive) { |
| 1066 | continue; |
| 1067 | } |
| 1068 | |
| 1069 | $text = $node->textContent; // prende il testo "visibile" (quello che vuoi) |
| 1070 | $textNode = $dom->createTextNode($text); |
| 1071 | |
| 1072 | if ($node->parentNode) { |
| 1073 | $node->parentNode->replaceChild($textNode, $node); |
| 1074 | } |
| 1075 | } |
| 1076 | |
| 1077 | // Salva l'HTML modificato |
| 1078 | $out = $dom->saveHTML(); |
| 1079 | |
| 1080 | return $out; |
| 1081 | } |
| 1082 | $html = gp_flatten_inner_formatting_tags($html, $flattenInnerFormattingTagsToRemove); |
| 1083 | } |
| 1084 | |
| 1085 | // Initialize DOMDocument without converting the body content via mb_convert_encoding. |
| 1086 | $doc = new DOMDocument (); |
| 1087 | libxml_use_internal_errors ( true ); |
| 1088 | // Use an XML encoding hack to inform DOMDocument about UTF-8 while preserving original characters. |
| 1089 | |
| 1090 | // PROTECT script/style content before DOMDocument loading (if option enabled) |
| 1091 | $protectedScriptsStyles = []; |
| 1092 | if (!empty($settings['protect_script_style'])) { |
| 1093 | $protectCounter = 0; |
| 1094 | |
| 1095 | // Extract <script>...</script> tags and replace with marker meta tags |
| 1096 | $html = preg_replace_callback( |
| 1097 | '/<script\b[^>]*>.*?<\/script>/is', |
| 1098 | function ($matches) use (&$protectedScriptsStyles, &$protectCounter) { |
| 1099 | $placeholder = "___GPTRANSLATE_SCRIPT_" . ($protectCounter++) . "___"; |
| 1100 | $protectedScriptsStyles[$placeholder] = $matches[0]; |
| 1101 | // Use a data attribute to preserve position - DOMDocument won't move these |
| 1102 | return '<meta data-gptranslate-script="' . htmlspecialchars($placeholder, ENT_QUOTES) . '">'; |
| 1103 | }, |
| 1104 | $html |
| 1105 | ); |
| 1106 | |
| 1107 | // Extract <style>...</style> tags and replace with marker meta tags |
| 1108 | $html = preg_replace_callback( |
| 1109 | '/<style\b[^>]*>.*?<\/style>/is', |
| 1110 | function ($matches) use (&$protectedScriptsStyles, &$protectCounter) { |
| 1111 | $placeholder = "___GPTRANSLATE_STYLE_" . ($protectCounter++) . "___"; |
| 1112 | $protectedScriptsStyles[$placeholder] = $matches[0]; |
| 1113 | // Use a data attribute to preserve position |
| 1114 | return '<meta data-gptranslate-style="' . htmlspecialchars($placeholder, ENT_QUOTES) . '">'; |
| 1115 | }, |
| 1116 | $html |
| 1117 | ); |
| 1118 | } |
| 1119 | $doc->loadHTML ( '<?xml encoding="UTF-8">' . $html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD ); |
| 1120 | libxml_clear_errors (); |
| 1121 | |
| 1122 | // Use XPath to locate text nodes, skipping those in <script> or <style> tags. |
| 1123 | $xpath = new DOMXPath ( $doc ); |
| 1124 | $cssSelectorLeafnodesExcluded = str_ireplace ( '"', '', trim ( trim ( preg_replace ( '/,+/', ',', str_ireplace ( [ "\r", "\n" ], ",", $settings['css_selector_serverside_leafnodes_excluded'] ?? '' ) ) ), ',' ) ); |
| 1125 | $excludedNodes = [ ]; |
| 1126 | |
| 1127 | if (! empty ( $cssSelectorLeafnodesExcluded )) { |
| 1128 | $selectors = explode ( ',', $cssSelectorLeafnodesExcluded ); |
| 1129 | $xpathQueries = [ ]; |
| 1130 | |
| 1131 | foreach ( $selectors as $selector ) { |
| 1132 | $selector = trim ( $selector ); |
| 1133 | if (empty ( $selector )) |
| 1134 | continue; |
| 1135 | |
| 1136 | // Convert CSS selector to XPath |
| 1137 | $selector = preg_replace ( '/\s+/', ' ', $selector ); // Normalize spaces |
| 1138 | $selectorParts = explode ( ' ', $selector ); |
| 1139 | |
| 1140 | $xpathQuery = ''; |
| 1141 | foreach ( $selectorParts as $part ) { |
| 1142 | if (preg_match ( '/^([a-zA-Z0-9_-]+)?(\.[a-zA-Z0-9_-]+)?(#[a-zA-Z0-9_-]+)?$/', $part, $matches )) { |
| 1143 | $tag = ! empty ( $matches[1] ) ? $matches[1] : '*'; // If no tag, use '*' |
| 1144 | $class = ! empty ( $matches[2] ) ? substr ( $matches[2], 1 ) : ''; // Remove leading '.' |
| 1145 | $id = ! empty ( $matches[3] ) ? substr ( $matches[3], 1 ) : ''; // Remove leading '#' |
| 1146 | |
| 1147 | $conditions = [ ]; |
| 1148 | if ($class) { |
| 1149 | $conditions[] = "contains(concat(' ', normalize-space(@class), ' '), ' $class ')"; |
| 1150 | } |
| 1151 | if ($id) { |
| 1152 | $conditions[] = "@id='$id'"; |
| 1153 | } |
| 1154 | |
| 1155 | $xpathQuery .= "//{$tag}" . (! empty ( $conditions ) ? "[" . implode ( " and ", $conditions ) . "]" : ""); |
| 1156 | } |
| 1157 | } |
| 1158 | |
| 1159 | if (! empty ( $xpathQuery )) { |
| 1160 | $xpathQueries[] = $xpathQuery; |
| 1161 | } |
| 1162 | } |
| 1163 | |
| 1164 | // Execute all XPath queries and collect excluded nodes |
| 1165 | foreach ( $xpathQueries as $query ) { |
| 1166 | foreach ( $xpath->query ( $query ) as $excludedNode ) { |
| 1167 | $excludedNodes[] = $excludedNode; |
| 1168 | } |
| 1169 | } |
| 1170 | } |
| 1171 | |
| 1172 | $textNodes = $xpath->query ( '//text()[not(ancestor::script) and not(ancestor::style) and normalize-space()]' ); |
| 1173 | $caseInsensitive = ! empty ( $settings['serverside_translations_caseinsensitive'] ); |
| 1174 | $matchQuotes = ! empty ( $settings['serverside_translations_matchquotes'] ); |
| 1175 | |
| 1176 | // Process each text node with translation replacements. |
| 1177 | foreach ( $textNodes as $textNode ) { |
| 1178 | if ($cssSelectorLeafnodesExcluded) { |
| 1179 | $skipNode = false; |
| 1180 | foreach ( $excludedNodes as $excludedNode ) { |
| 1181 | if ($excludedNode->isSameNode ( $textNode->parentNode )) { |
| 1182 | $skipNode = true; |
| 1183 | break; |
| 1184 | } |
| 1185 | } |
| 1186 | if ($skipNode) |
| 1187 | continue; |
| 1188 | } |
| 1189 | |
| 1190 | $originalTextContent = $textNode->nodeValue; |
| 1191 | $textContent = $textNode->nodeValue; |
| 1192 | if ($matchQuotes) { |
| 1193 | $textContent = str_ireplace ( '"', "'", $textContent ); |
| 1194 | } |
| 1195 | // Normalize once without trimming to preserve spaces between nodes |
| 1196 | $textContent = normalizeTextNoTrim($textContent); |
| 1197 | |
| 1198 | // INCREMENTAL FIX: Use exact match only if incremental preserve new texts is enabled |
| 1199 | if ($incrementalEnabledPreserveNewTexts) { |
| 1200 | $textContent = gpt_incremental_translate_text($textContent, $incrementalContext); |
| 1201 | } else { |
| 1202 | // Normal pattern matching (standard behavior) |
| 1203 | foreach ( $translationsArray as $originalText => $translatedText ) { |
| 1204 | $originalText = normalizeText($originalText); |
| 1205 | |
| 1206 | $prevTextContent = $textContent; // Save the current state |
| 1207 | if ($wrapExcludedWords) { |
| 1208 | // Use word boundaries to avoid sub-replacing inside already translated words |
| 1209 | $pattern = '/' . (preg_match('/^\w/u', trim($originalText)) ? '(?<!\w)' : '') . preg_quote(trim($originalText), '/') . '(?!\w)/' . ($caseInsensitive ? 'ui' : 'u'); |
| 1210 | $textContent = preg_replace($pattern, $translatedText, $textContent); |
| 1211 | } else { |
| 1212 | // Use word boundaries to avoid sub-replacing inside already translated words |
| 1213 | $pattern = '/' . (preg_match('/^\w/u', trim($originalText)) ? '(?<!\w)' : '') . preg_quote(trim($originalText), '/') . '(?!\w)/' . ($caseInsensitive ? 'ui' : 'u'); |
| 1214 | $textContent = preg_replace($pattern, $translatedText, $textContent); |
| 1215 | if ($textContent !== $prevTextContent) { |
| 1216 | break; |
| 1217 | } |
| 1218 | } |
| 1219 | } |
| 1220 | } |
| 1221 | $textNode->nodeValue = $textContent; |
| 1222 | } |
| 1223 | |
| 1224 | // Replace page links in <a href="..."> tags based on translated aliases |
| 1225 | if (!empty($translatedAliasesMap) || !empty($translatedAliasesRelativeMap)) { |
| 1226 | $exclusions = $settings['rewrite_page_links_exclusions'] ?? ''; |
| 1227 | $page_exclusions = $settings['page_exclusions'] ?? ''; |
| 1228 | $linkNodes = $xpath->query('//a[@href]'); |
| 1229 | foreach ($linkNodes as $linkNode) { |
| 1230 | $href = $linkNode->getAttribute('href'); |
| 1231 | |
| 1232 | // Check if link should be excluded from rewriting |
| 1233 | if (gptranslate_should_exclude_link($href, $exclusions, $page_exclusions)) { |
| 1234 | continue; |
| 1235 | } |
| 1236 | |
| 1237 | // Decode href attribute (WordPress-style, KEEP trailing slash) |
| 1238 | $decodedHref = rawurldecode(html_entity_decode($href, ENT_QUOTES, 'UTF-8')); |
| 1239 | |
| 1240 | // Extract hash/fragment if present |
| 1241 | $hashFragment = ''; |
| 1242 | if (strpos($decodedHref, '#') !== false) { |
| 1243 | $parts = explode('#', $decodedHref, 2); |
| 1244 | $decodedHref = $parts[0]; // URL without hash |
| 1245 | $hashFragment = '#' . $parts[1]; // Save the hash |
| 1246 | } |
| 1247 | |
| 1248 | $queryString = ''; |
| 1249 | if ($settings['ignore_querystring'] == 1 && strpos($decodedHref, '?') !== false) { |
| 1250 | $parts = explode('?', $decodedHref, 2); |
| 1251 | $decodedHref = $parts[0]; |
| 1252 | $queryString = '?' . $parts[1]; |
| 1253 | } |
| 1254 | $originalPathBeforeNormalization = $decodedHref; |
| 1255 | |
| 1256 | $decodedHref = gpt_trailingslashit_url($decodedHref); |
| 1257 | |
| 1258 | $translatedAlias = null; |
| 1259 | |
| 1260 | // Cerca prima negli URL assoluti |
| 1261 | if (isset($translatedAliasesMap[$decodedHref])) { |
| 1262 | $translatedAlias = $translatedAliasesMap[$decodedHref]; |
| 1263 | } |
| 1264 | // Poi cerca negli URL relativi |
| 1265 | elseif (isset($translatedAliasesRelativeMap[$decodedHref])) { |
| 1266 | $translatedAlias = $translatedAliasesRelativeMap[$decodedHref]; |
| 1267 | } |
| 1268 | |
| 1269 | // Se trovato un alias tradotto, sostituisci |
| 1270 | if ($translatedAlias) { |
| 1271 | // Aggiungi data-originalhref se non presente |
| 1272 | if (!$linkNode->hasAttribute('data-originalhref')) { |
| 1273 | $linkNode->setAttribute('data-originalhref', $originalPathBeforeNormalization . $queryString . $hashFragment); |
| 1274 | } |
| 1275 | $linkNode->setAttribute('href', $translatedAlias . $queryString . $hashFragment); |
| 1276 | } |
| 1277 | } |
| 1278 | } |
| 1279 | |
| 1280 | // Replace form action URLs: add language prefix and apply translated aliases |
| 1281 | if (!empty($settings['rewrite_form_actions'])) { |
| 1282 | $siteUrl = trailingslashit(get_site_url()); |
| 1283 | $siteHost = parse_url($siteUrl, PHP_URL_HOST); |
| 1284 | $knownLangs = (isset($settings['languages']) && is_array($settings['languages'])) ? array_map('strtolower', $settings['languages']) : [$originalLang, $translatedLang]; |
| 1285 | $formNodes = $xpath->query('//form[@action]'); |
| 1286 | foreach ($formNodes as $formNode) { |
| 1287 | $action = $formNode->getAttribute('action'); |
| 1288 | $decodedAction = rawurldecode(html_entity_decode($action, ENT_QUOTES, 'UTF-8')); |
| 1289 | |
| 1290 | // Skip non-rewritable actions |
| 1291 | if (preg_match('/^(mailto:|tel:|javascript:|#)/i', $decodedAction)) continue; |
| 1292 | if (preg_match('/\.php($|\?|\#)/i', $decodedAction)) continue; |
| 1293 | if (preg_match('/^https?:\/\//i', $decodedAction) && parse_url($decodedAction, PHP_URL_HOST) !== $siteHost) continue; |
| 1294 | |
| 1295 | $queryString = ''; |
| 1296 | if ($settings['ignore_querystring'] == 1 && strpos($decodedAction, '?') !== false) { |
| 1297 | $parts = explode('?', $decodedAction, 2); |
| 1298 | $decodedAction = $parts[0]; |
| 1299 | $queryString = '?' . $parts[1]; |
| 1300 | } |
| 1301 | $originalActionFull = $decodedAction . $queryString; |
| 1302 | |
| 1303 | // Extract relative path |
| 1304 | $actionPath = $decodedAction; |
| 1305 | $isAbsolute = false; |
| 1306 | if (preg_match('/^https?:\/\//i', $actionPath)) { |
| 1307 | $isAbsolute = true; |
| 1308 | $actionPath = parse_url($actionPath, PHP_URL_PATH) ?: '/'; |
| 1309 | } |
| 1310 | |
| 1311 | // Add/replace language prefix |
| 1312 | $pathParts = explode('/', $actionPath); |
| 1313 | $langIndex = 1; |
| 1314 | if (isset($pathParts[$langIndex]) && $pathParts[$langIndex] === 'index.php') $langIndex = 2; |
| 1315 | if (!empty($settings['subfolder_installation'])) $langIndex = 2; |
| 1316 | |
| 1317 | if (isset($pathParts[$langIndex]) && in_array(strtolower($pathParts[$langIndex]), $knownLangs)) { |
| 1318 | $pathParts[$langIndex] = $translatedLang; |
| 1319 | } else { |
| 1320 | array_splice($pathParts, $langIndex, 0, [$translatedLang]); |
| 1321 | } |
| 1322 | |
| 1323 | $rebuiltPath = '/' . implode('/', array_filter($pathParts, function($p) { return $p !== ''; })); |
| 1324 | $rebuiltPath = gpt_trailingslashit_url($rebuiltPath); |
| 1325 | $computedAction = $isAbsolute ? (rtrim($siteUrl, '/') . $rebuiltPath) : $rebuiltPath; |
| 1326 | $computedAction = gpt_trailingslashit_url($computedAction); |
| 1327 | |
| 1328 | // Try alias match |
| 1329 | $translatedAlias = null; |
| 1330 | if (!empty($translatedAliasesMap) || !empty($translatedAliasesRelativeMap)) { |
| 1331 | if (isset($translatedAliasesMap[$computedAction])) { |
| 1332 | $translatedAlias = $translatedAliasesMap[$computedAction]; |
| 1333 | } elseif (isset($translatedAliasesRelativeMap[$rebuiltPath])) { |
| 1334 | $translatedAlias = $translatedAliasesRelativeMap[$rebuiltPath]; |
| 1335 | } elseif (!$isAbsolute && !empty($settings['subfolder_installation'])) { |
| 1336 | // Fallback: form action is relative and includes subfolder path |
| 1337 | // Map keys strip the subfolder, so retry after stripping it |
| 1338 | $siteBasePath = rtrim(parse_url(rtrim($siteUrl, '/'), PHP_URL_PATH) ?: '', '/'); |
| 1339 | if ($siteBasePath !== '' && strpos($rebuiltPath, $siteBasePath . '/') === 0) { |
| 1340 | $rebuiltPathNoSubfolder = substr($rebuiltPath, strlen($siteBasePath)); |
| 1341 | if (isset($translatedAliasesRelativeMap[$rebuiltPathNoSubfolder])) { |
| 1342 | $translatedAlias = $translatedAliasesRelativeMap[$rebuiltPathNoSubfolder]; |
| 1343 | } |
| 1344 | } |
| 1345 | } |
| 1346 | } |
| 1347 | |
| 1348 | $finalAction = $translatedAlias ? ($translatedAlias . $queryString) : ($computedAction . $queryString); |
| 1349 | |
| 1350 | if (!$formNode->hasAttribute('data-originalaction')) { |
| 1351 | $formNode->setAttribute('data-originalaction', $originalActionFull); |
| 1352 | } |
| 1353 | $formNode->setAttribute('action', $finalAction); |
| 1354 | } |
| 1355 | } |
| 1356 | |
| 1357 | // Check if also images 'alt' are enabled to be translated |
| 1358 | if (! empty ( $settings['translate_altimages'] )) { |
| 1359 | $imgNodesAlt = $xpath->query ( '//img[@alt]' ); |
| 1360 | foreach ( $imgNodesAlt as $imgNode ) { |
| 1361 | $altText = $imgNode->getAttribute ( 'alt' ); |
| 1362 | if ($matchQuotes) { |
| 1363 | $altText = str_ireplace ( '"', "'", $altText ); |
| 1364 | } |
| 1365 | foreach ( $altTranslationsArray as $originalText => $translatedText ) { |
| 1366 | $originalText = normalizeText($originalText); |
| 1367 | $altText = normalizeText($altText); |
| 1368 | |
| 1369 | $prevAltText = $altText; |
| 1370 | if ($caseInsensitive) { |
| 1371 | $altText = str_ireplace ( trim ( $originalText ), $translatedText, $altText ); |
| 1372 | } else { |
| 1373 | $altText = str_replace ( trim ( $originalText ), $translatedText, $altText ); |
| 1374 | } |
| 1375 | if ($altText !== $prevAltText) { |
| 1376 | break; |
| 1377 | } |
| 1378 | } |
| 1379 | $imgNode->setAttribute ( 'alt', $altText ); |
| 1380 | } |
| 1381 | |
| 1382 | $imgNodesTitle = $xpath->query ( '//img[@title]' ); |
| 1383 | foreach ( $imgNodesTitle as $imgNode ) { |
| 1384 | $titleText = $imgNode->getAttribute ( 'title' ); |
| 1385 | if ($matchQuotes) { |
| 1386 | $titleText = str_ireplace ( '"', "'", $titleText ); |
| 1387 | } |
| 1388 | foreach ( $altTranslationsArray as $originalText => $translatedText ) { |
| 1389 | $originalText = normalizeText($originalText); |
| 1390 | $titleText = normalizeText($titleText); |
| 1391 | |
| 1392 | $prevTitleText = $titleText; |
| 1393 | if ($caseInsensitive) { |
| 1394 | $titleText = str_ireplace ( trim ( $originalText ), $translatedText, $titleText ); |
| 1395 | } else { |
| 1396 | $titleText = str_replace ( trim ( $originalText ), $translatedText, $titleText ); |
| 1397 | } |
| 1398 | if ($titleText !== $prevTitleText) { |
| 1399 | break; |
| 1400 | } |
| 1401 | } |
| 1402 | $imgNode->setAttribute ( 'title', $titleText ); |
| 1403 | } |
| 1404 | |
| 1405 | // Check if also images 'src' are enabled to be translated |
| 1406 | if (! empty ( $settings ['translate_srcimages'] )) { |
| 1407 | $imgNodesSrc = $xpath->query ( '//img[@src]' ); |
| 1408 | foreach ( $imgNodesSrc as $imgNode ) { |
| 1409 | $srcValue = ltrim ( $imgNode->getAttribute ( 'src' ), '/' ); |
| 1410 | |
| 1411 | if ($matchQuotes) { |
| 1412 | $srcValue = str_ireplace ( '"', "'", $srcValue ); |
| 1413 | } |
| 1414 | |
| 1415 | $translatedSrcValue = null; |
| 1416 | foreach ( $altTranslationsArray as $originalText => $translatedText ) { |
| 1417 | $originalText = normalizeText($originalText); |
| 1418 | $srcValue = normalizeText($srcValue); |
| 1419 | |
| 1420 | $originalText = ltrim ( $originalText, '/' ); |
| 1421 | if ($caseInsensitive) { |
| 1422 | if (strcasecmp ( trim ( $originalText ), $srcValue ) === 0) { |
| 1423 | $translatedSrcValue = $translatedText; |
| 1424 | break; |
| 1425 | } |
| 1426 | } else { |
| 1427 | if (trim ( $originalText ) === $srcValue) { |
| 1428 | $translatedSrcValue = $translatedText; |
| 1429 | break; |
| 1430 | } |
| 1431 | } |
| 1432 | } |
| 1433 | |
| 1434 | // If a translation was found, replace src and srcset |
| 1435 | if ($translatedSrcValue !== null) { |
| 1436 | // Replace the src attribute |
| 1437 | $imgNode->setAttribute ( 'src', $translatedSrcValue ); |
| 1438 | |
| 1439 | // Check if srcset exists and replace it with the translated src value |
| 1440 | if ($imgNode->hasAttribute ( 'srcset' )) { |
| 1441 | $imgNode->setAttribute ( 'srcset', $translatedSrcValue ); |
| 1442 | } |
| 1443 | } |
| 1444 | } |
| 1445 | } |
| 1446 | |
| 1447 | // Check if also iframes 'src' are enabled to be translated |
| 1448 | if (! empty ( $settings ['translate_srciframes'] )) { |
| 1449 | $iframeNodesSrc = $xpath->query ( '//iframe[@src]' ); |
| 1450 | foreach ( $iframeNodesSrc as $iframeNode ) { |
| 1451 | $srcValue = ltrim ( $iframeNode->getAttribute ( 'src' ), '/' ); |
| 1452 | |
| 1453 | if ($matchQuotes) { |
| 1454 | $srcValue = str_ireplace ( '"', "'", $srcValue ); |
| 1455 | } |
| 1456 | |
| 1457 | $translatedSrcValue = null; |
| 1458 | foreach ( $altTranslationsArray as $originalText => $translatedText ) { |
| 1459 | $originalText = normalizeText($originalText); |
| 1460 | $srcValue = normalizeText($srcValue); |
| 1461 | |
| 1462 | $originalText = ltrim ( $originalText, '/' ); |
| 1463 | if ($caseInsensitive) { |
| 1464 | if (strcasecmp ( trim ( $originalText ), $srcValue ) === 0) { |
| 1465 | $translatedSrcValue = $translatedText; |
| 1466 | break; |
| 1467 | } |
| 1468 | } else { |
| 1469 | if (trim ( $originalText ) === $srcValue) { |
| 1470 | $translatedSrcValue = $translatedText; |
| 1471 | break; |
| 1472 | } |
| 1473 | } |
| 1474 | } |
| 1475 | |
| 1476 | if ($translatedSrcValue !== null) { |
| 1477 | $iframeNode->setAttribute ( 'src', $translatedSrcValue ); |
| 1478 | } |
| 1479 | } |
| 1480 | } |
| 1481 | |
| 1482 | // Check if also videos 'src' are enabled to be translated |
| 1483 | if (! empty ( $settings ['translate_srcvideos'] )) { |
| 1484 | $videoNodesSrc = $xpath->query ( '//video[@src] | //video//source[@src]' ); |
| 1485 | foreach ( $videoNodesSrc as $videoNode ) { |
| 1486 | $srcValue = ltrim ( $videoNode->getAttribute ( 'src' ), '/' ); |
| 1487 | |
| 1488 | if ($matchQuotes) { |
| 1489 | $srcValue = str_ireplace ( '"', "'", $srcValue ); |
| 1490 | } |
| 1491 | |
| 1492 | $translatedSrcValue = null; |
| 1493 | foreach ( $altTranslationsArray as $originalText => $translatedText ) { |
| 1494 | $originalText = normalizeText($originalText); |
| 1495 | $srcValue = normalizeText($srcValue); |
| 1496 | |
| 1497 | $originalText = ltrim ( $originalText, '/' ); |
| 1498 | if ($caseInsensitive) { |
| 1499 | if (strcasecmp ( trim ( $originalText ), $srcValue ) === 0) { |
| 1500 | $translatedSrcValue = $translatedText; |
| 1501 | break; |
| 1502 | } |
| 1503 | } else { |
| 1504 | if (trim ( $originalText ) === $srcValue) { |
| 1505 | $translatedSrcValue = $translatedText; |
| 1506 | break; |
| 1507 | } |
| 1508 | } |
| 1509 | } |
| 1510 | |
| 1511 | if ($translatedSrcValue !== null) { |
| 1512 | $videoNode->setAttribute ( 'src', $translatedSrcValue ); |
| 1513 | } |
| 1514 | } |
| 1515 | } |
| 1516 | } |
| 1517 | |
| 1518 | // Translate iframe locale parameter (e.g. Stripe payment forms) |
| 1519 | if (! empty ( $settings ['translate_iframe_locale'] )) { |
| 1520 | $currentLangCode = strtolower(explode('-', $currentLanguage)[0]); |
| 1521 | $iframeNodes = $xpath->query ( '//iframe[@src]' ); |
| 1522 | foreach ( $iframeNodes as $iframeNode ) { |
| 1523 | $src = $iframeNode->getAttribute ( 'src' ); |
| 1524 | if (strpos ( $src, 'locale=' ) !== false) { |
| 1525 | $translatedSrc = preg_replace ( '/locale=[a-z]{2}/i', 'locale=' . $currentLangCode, $src ); |
| 1526 | $iframeNode->setAttribute ( 'src', $translatedSrc ); |
| 1527 | } |
| 1528 | } |
| 1529 | } |
| 1530 | |
| 1531 | // Translate meta descriptions (standard, Open Graph, Twitter, Dublin Core) |
| 1532 | $metaDescriptions = $xpath->query( |
| 1533 | '//meta[ |
| 1534 | (@name="description") |
| 1535 | or (@name="twitter:description") |
| 1536 | or (@property="og:description") |
| 1537 | or (@name="dc.description") |
| 1538 | ]' |
| 1539 | ); |
| 1540 | foreach ($metaDescriptions as $meta) { |
| 1541 | $contentValue = $meta->getAttribute('content'); |
| 1542 | |
| 1543 | if ($matchQuotes) { |
| 1544 | $contentValue = str_ireplace('"', "'", $contentValue); |
| 1545 | } |
| 1546 | |
| 1547 | foreach ($altTranslationsArray as $originalText => $translatedText) { |
| 1548 | $prev = $contentValue; |
| 1549 | if ($caseInsensitive) { |
| 1550 | $contentValue = str_ireplace(trim($originalText), $translatedText, $contentValue); |
| 1551 | } else { |
| 1552 | $contentValue = str_replace(trim($originalText), $translatedText, $contentValue); |
| 1553 | } |
| 1554 | if ($contentValue !== $prev) break; |
| 1555 | } |
| 1556 | |
| 1557 | $meta->setAttribute('content', $contentValue); |
| 1558 | } |
| 1559 | |
| 1560 | $ogTitleNodes = $xpath->query('//meta[@property="og:title"]'); |
| 1561 | if ($ogTitleNodes->length > 0) { |
| 1562 | // Find the <title> tag to get the translated title |
| 1563 | $titleNodes = $xpath->query('//title'); |
| 1564 | |
| 1565 | if ($titleNodes->length > 0) { |
| 1566 | $titleNode = $titleNodes->item(0); |
| 1567 | $translatedTitle = $titleNode->textContent; |
| 1568 | |
| 1569 | // Set og:title with the same value as <title> |
| 1570 | foreach ($ogTitleNodes as $ogTitleNode) { |
| 1571 | $ogTitleNode->setAttribute('content', trim($translatedTitle)); |
| 1572 | } |
| 1573 | |
| 1574 | // Set twitter:title with the same value as <title> |
| 1575 | $twitterTitleNodes = $xpath->query('//meta[@name="twitter:title"]'); |
| 1576 | foreach ($twitterTitleNodes as $twitterTitleNode) { |
| 1577 | $twitterTitleNode->setAttribute('content', trim($translatedTitle)); |
| 1578 | } |
| 1579 | |
| 1580 | // Set Dublin Core (SEOPress) dc:title with the same value as <title> |
| 1581 | $dcTitleNodes = $xpath->query('//meta[@name="dc.title"]'); |
| 1582 | foreach ($dcTitleNodes as $dcTitleNode) { |
| 1583 | $dcTitleNode->setAttribute('content', trim($translatedTitle)); |
| 1584 | } |
| 1585 | } |
| 1586 | } |
| 1587 | |
| 1588 | // Optionally, add an attribute to the <body> tag to signal that translations were applied. |
| 1589 | $bodyTag = $doc->getElementsByTagName ( 'body' )->item ( 0 ); |
| 1590 | if ($bodyTag) { |
| 1591 | $bodyTag->setAttribute ( 'data-gptranslateskip', '1' ); |
| 1592 | $bodyTag->setAttribute ( 'data-gptranslateoriginalalias', htmlspecialchars($row['pagelink'], ENT_QUOTES, 'UTF-8') ); |
| 1593 | } |
| 1594 | |
| 1595 | // Save the modified HTML and strip the XML processing instruction |
| 1596 | // that was injected as a UTF-8 encoding hint for DOMDocument. |
| 1597 | $html = $doc->saveHTML (); |
| 1598 | $html = preg_replace ( '/<\?xml[^>]*>\s*/i', '', $html ); |
| 1599 | |
| 1600 | // RESTORE protected script/style content (if protection was enabled) |
| 1601 | if (!empty($settings['protect_script_style'])) { |
| 1602 | // Restore scripts from meta markers |
| 1603 | $html = preg_replace_callback( |
| 1604 | '/<meta\s+data-gptranslate-script="([^"]+)"[^>]*>/i', |
| 1605 | function ($matches) use ($protectedScriptsStyles) { |
| 1606 | $placeholder = $matches[1]; |
| 1607 | return isset($protectedScriptsStyles[$placeholder]) ? $protectedScriptsStyles[$placeholder] : $matches[0]; |
| 1608 | }, |
| 1609 | $html |
| 1610 | ); |
| 1611 | |
| 1612 | // Restore styles from meta markers |
| 1613 | $html = preg_replace_callback( |
| 1614 | '/<meta\s+data-gptranslate-style="([^"]+)"[^>]*>/i', |
| 1615 | function ($matches) use ($protectedScriptsStyles) { |
| 1616 | $placeholder = $matches[1]; |
| 1617 | return isset($protectedScriptsStyles[$placeholder]) ? $protectedScriptsStyles[$placeholder] : $matches[0]; |
| 1618 | }, |
| 1619 | $html |
| 1620 | ); |
| 1621 | } |
| 1622 | |
| 1623 | // Decode HTML entities back into UTF-8 characters. |
| 1624 | $html = html_entity_decode ( $html, ENT_QUOTES | ENT_HTML5, 'UTF-8' ); |
| 1625 | |
| 1626 | } catch ( Exception $e ) { |
| 1627 | // Handle exception if needed |
| 1628 | } |
| 1629 | } elseif ($settings['serverside_translations_method'] == 'simplehtmldom') { |
| 1630 | require_once plugin_dir_path(__FILE__) . 'simplehtmldom.php'; |
| 1631 | |
| 1632 | // Helper: checks if $child is inside (or is) $parent. |
| 1633 | function nodeIsInsideExcluded($child, $excludedNodes) { |
| 1634 | while ($child !== null) { |
| 1635 | foreach ($excludedNodes as $ex) { |
| 1636 | if ($child === $ex) { |
| 1637 | return true; |
| 1638 | } |
| 1639 | } |
| 1640 | $child = $child->parent; |
| 1641 | } |
| 1642 | return false; |
| 1643 | } |
| 1644 | |
| 1645 | // Helper: checks if $node is inside (or is) any of the specified tags. |
| 1646 | function nodeIsInsideTags($node, $tagNames) { |
| 1647 | $tagNames = array_map('strtolower', $tagNames); |
| 1648 | while ($node !== null) { |
| 1649 | $tag = strtolower($node->tag ?? ''); |
| 1650 | if (in_array($tag, $tagNames, true)) { |
| 1651 | return true; |
| 1652 | } |
| 1653 | $node = $node->parent; |
| 1654 | } |
| 1655 | return false; |
| 1656 | } |
| 1657 | |
| 1658 | // Recursive function to process all text nodes. |
| 1659 | function processTextNodes($node, $excludedNodes, $translationsArray, $caseInsensitive, $matchQuotes, $incrementalEnabledPreserveNewTexts, $incrementalContext) { |
| 1660 | if ($node->tag === 'text') { |
| 1661 | if (! nodeIsInsideExcluded($node, $excludedNodes) && ! nodeIsInsideTags($node, ['script', 'style', 'noscript']) && trim($node->innertext) && $node->innertext != "\t") { |
| 1662 | $text = $node->innertext; |
| 1663 | if ($matchQuotes) { |
| 1664 | $text = str_ireplace('"', "'", $text); |
| 1665 | } |
| 1666 | |
| 1667 | // Normalize once to avoid stripping spaces added by replacements |
| 1668 | $text = normalizeTextNoTrim($text); |
| 1669 | |
| 1670 | // INCREMENTAL FIX: Use exact match only if incremental preserve new texts is enabled |
| 1671 | if ($incrementalEnabledPreserveNewTexts) { |
| 1672 | $text = gpt_incremental_translate_text($text, $incrementalContext); |
| 1673 | } else { |
| 1674 | // Normal pattern matching (standard behavior) |
| 1675 | // Apply translations without overriding longer translations |
| 1676 | $processedParts = []; |
| 1677 | foreach ($translationsArray as $originalText => $translatedText) { |
| 1678 | $originalText = normalizeText($originalText); |
| 1679 | |
| 1680 | if ($caseInsensitive) { |
| 1681 | $text = preg_replace_callback( |
| 1682 | '/' . (preg_match('/^\w/u', trim($originalText)) ? '(?<!\w)' : '') . preg_quote(trim($originalText), '/') . '(?!\w)/ui', |
| 1683 | function ($matches) use ($translatedText, &$processedParts) { |
| 1684 | if (in_array($matches[0], $processedParts, true)) { |
| 1685 | return $matches[0]; |
| 1686 | } |
| 1687 | $processedParts[] = $translatedText; |
| 1688 | return $translatedText; |
| 1689 | }, |
| 1690 | $text |
| 1691 | ); |
| 1692 | } else { |
| 1693 | $text = preg_replace_callback( |
| 1694 | '/' . (preg_match('/^\w/u', trim($originalText)) ? '(?<!\w)' : '') . preg_quote(trim($originalText), '/') . '(?!\w)/', |
| 1695 | function ($matches) use ($translatedText, &$processedParts) { |
| 1696 | if (in_array($matches[0], $processedParts, true)) { |
| 1697 | return $matches[0]; |
| 1698 | } |
| 1699 | $processedParts[] = $translatedText; |
| 1700 | return $translatedText; |
| 1701 | }, |
| 1702 | $text |
| 1703 | ); |
| 1704 | } |
| 1705 | } |
| 1706 | } |
| 1707 | $node->innertext = $text; |
| 1708 | } |
| 1709 | } else { |
| 1710 | $tagLower = strtolower($node->tag ?? ''); |
| 1711 | if (in_array($tagLower, ['script', 'style'])) { |
| 1712 | return; |
| 1713 | } |
| 1714 | if (isset($node->nodes) && is_array($node->nodes)) { |
| 1715 | foreach ($node->nodes as $child) { |
| 1716 | processTextNodes($child, $excludedNodes, $translationsArray, $caseInsensitive, $matchQuotes, $incrementalEnabledPreserveNewTexts, $incrementalContext); |
| 1717 | } |
| 1718 | } |
| 1719 | } |
| 1720 | } |
| 1721 | |
| 1722 | // Main processing code. |
| 1723 | try { |
| 1724 | $translationsArray = json_decode($row['translations'], true) ?? []; |
| 1725 | $altTranslationsArray = json_decode($row['alt_translations'] ?? '', true) ?: []; |
| 1726 | |
| 1727 | // Sort translations by descending key length |
| 1728 | uksort($translationsArray, function ($a, $b) { |
| 1729 | return strlen($b) - strlen($a); |
| 1730 | }); |
| 1731 | |
| 1732 | // Sort alt translations by descending key length |
| 1733 | uksort($altTranslationsArray, function ($a, $b) { |
| 1734 | return strlen($b) - strlen($a); |
| 1735 | }); |
| 1736 | |
| 1737 | // Flatten formatting tags regex |
| 1738 | if($flattenInnerFormattingTags) { |
| 1739 | // Flatten formatting tags in the body to improve matching if enabled |
| 1740 | function gp_flatten_inner_formatting_tags_simplehtmldom($htmlString, $tagsToRemove ) { |
| 1741 | // Complete body parsing |
| 1742 | $doc = gptranslate_simplehtmldom_str_get_html ( $htmlString ); |
| 1743 | if (! $doc) |
| 1744 | return $htmlString; |
| 1745 | |
| 1746 | $selector = implode ( ',', $tagsToRemove ); |
| 1747 | |
| 1748 | $found = $doc->find($selector); |
| 1749 | if (!empty($found)) { |
| 1750 | for ($i = count($found) - 1; $i >= 0; $i--) { |
| 1751 | $n = $found[$i]; |
| 1752 | // Preserva elementi vuoti (es. <i> FontAwesome/icon fonts) |
| 1753 | if (trim(str_replace([' ', "\xc2\xa0", "\xa0"], '', $n->plaintext)) === '') { |
| 1754 | continue; |
| 1755 | } |
| 1756 | // Preserva elementi che contengono form elements (select, input, textarea, button, ecc.) |
| 1757 | $interactiveTags = ['select', 'input', 'textarea', 'button', 'option', 'optgroup', 'fieldset', 'datalist', 'output', 'label', 'form', 'img', 'video', 'audio', 'canvas', 'svg', 'iframe']; |
| 1758 | $hasInteractive = false; |
| 1759 | foreach ($interactiveTags as $iTag) { |
| 1760 | if (!empty($n->find($iTag))) { |
| 1761 | $hasInteractive = true; |
| 1762 | break; |
| 1763 | } |
| 1764 | } |
| 1765 | if ($hasInteractive) { |
| 1766 | continue; |
| 1767 | } |
| 1768 | $n->outertext = $n->plaintext; |
| 1769 | } |
| 1770 | } |
| 1771 | |
| 1772 | $out = $doc->save(); |
| 1773 | |
| 1774 | if (method_exists($doc, 'clear')) $doc->clear(); |
| 1775 | unset($doc); |
| 1776 | |
| 1777 | return $out; |
| 1778 | } |
| 1779 | $html = gp_flatten_inner_formatting_tags_simplehtmldom($html, $flattenInnerFormattingTagsToRemove); |
| 1780 | } |
| 1781 | |
| 1782 | // PROTECT script/style content before SimpleHtmlDom parsing (if option enabled) |
| 1783 | $protectedScriptsStyles = []; |
| 1784 | if (!empty($settings['protect_script_style'])) { |
| 1785 | $protectCounter = 0; |
| 1786 | $html = preg_replace_callback( |
| 1787 | '/<script[^>]*>.*?<\/script>/is', |
| 1788 | function ($matches) use (&$protectedScriptsStyles, &$protectCounter) { |
| 1789 | $placeholder = "___GPTRANSLATE_SCRIPT_" . ($protectCounter++) . "___"; |
| 1790 | $protectedScriptsStyles[$placeholder] = $matches[0]; |
| 1791 | return $placeholder; |
| 1792 | }, |
| 1793 | $html |
| 1794 | ); |
| 1795 | $html = preg_replace_callback( |
| 1796 | '/<style[^>]*>.*?<\/style>/is', |
| 1797 | function ($matches) use (&$protectedScriptsStyles, &$protectCounter) { |
| 1798 | $placeholder = "___GPTRANSLATE_STYLE_" . ($protectCounter++) . "___"; |
| 1799 | $protectedScriptsStyles[$placeholder] = $matches[0]; |
| 1800 | return $placeholder; |
| 1801 | }, |
| 1802 | $html |
| 1803 | ); |
| 1804 | } |
| 1805 | $htmlObj = gptranslate_simplehtmldom_str_get_html($html); |
| 1806 | |
| 1807 | $cssSelectorLeafnodesExcluded = str_ireplace( |
| 1808 | '"', |
| 1809 | '', |
| 1810 | trim(trim(preg_replace('/,+/', ',', str_ireplace(["\r", "\n"], ",", $settings['css_selector_serverside_leafnodes_excluded'] ?? ''))), ',') |
| 1811 | ); |
| 1812 | |
| 1813 | $excludedNodes = []; |
| 1814 | if (! empty($cssSelectorLeafnodesExcluded)) { |
| 1815 | $selectors = explode(',', $cssSelectorLeafnodesExcluded); |
| 1816 | foreach ($selectors as $selector) { |
| 1817 | $selector = trim($selector); |
| 1818 | if (! empty($selector)) { |
| 1819 | $foundNodes = $htmlObj->find($selector); |
| 1820 | foreach ($foundNodes as $node) { |
| 1821 | $excludedNodes[] = $node; |
| 1822 | } |
| 1823 | } |
| 1824 | } |
| 1825 | } |
| 1826 | |
| 1827 | $caseInsensitive = ! empty($settings['serverside_translations_caseinsensitive']); |
| 1828 | $matchQuotes = ! empty($settings['serverside_translations_matchquotes']); |
| 1829 | |
| 1830 | processTextNodes($htmlObj, $excludedNodes, $translationsArray, $caseInsensitive, $matchQuotes, $incrementalEnabledPreserveNewTexts, $incrementalContext); |
| 1831 | |
| 1832 | // Replace page links in <a href="..."> tags based on translated aliases |
| 1833 | if (!empty($translatedAliasesMap) || !empty($translatedAliasesRelativeMap)) { |
| 1834 | $exclusions = $settings['rewrite_page_links_exclusions'] ?? ''; |
| 1835 | $page_exclusions = $settings['page_exclusions'] ?? ''; |
| 1836 | foreach ($htmlObj->find('a[href]') as $linkNode) { |
| 1837 | $href = $linkNode->href; |
| 1838 | |
| 1839 | // Check if link should be excluded from rewriting |
| 1840 | if (gptranslate_should_exclude_link($href, $exclusions, $page_exclusions)) { |
| 1841 | continue; |
| 1842 | } |
| 1843 | |
| 1844 | // Decode href attribute (WordPress-style, KEEP trailing slash) |
| 1845 | $decodedHref = rawurldecode(html_entity_decode($href, ENT_QUOTES, 'UTF-8')); |
| 1846 | |
| 1847 | // Extract hash/fragment if present |
| 1848 | $hashFragment = ''; |
| 1849 | if (strpos($decodedHref, '#') !== false) { |
| 1850 | $parts = explode('#', $decodedHref, 2); |
| 1851 | $decodedHref = $parts[0]; // URL without hash |
| 1852 | $hashFragment = '#' . $parts[1]; // Save the hash |
| 1853 | } |
| 1854 | |
| 1855 | // Extract query string if ignore_querystring is enabled |
| 1856 | $queryString = ''; |
| 1857 | if ($settings['ignore_querystring'] == 1 && strpos($decodedHref, '?') !== false) { |
| 1858 | $parts = explode('?', $decodedHref, 2); |
| 1859 | $decodedHref = $parts[0]; |
| 1860 | $queryString = '?' . $parts[1]; |
| 1861 | } |
| 1862 | $originalPathBeforeNormalization = $decodedHref; |
| 1863 | |
| 1864 | $decodedHref = gpt_trailingslashit_url($decodedHref); |
| 1865 | |
| 1866 | $translatedAlias = null; |
| 1867 | |
| 1868 | // Cerca prima negli URL assoluti |
| 1869 | if (isset($translatedAliasesMap[$decodedHref])) { |
| 1870 | $translatedAlias = $translatedAliasesMap[$decodedHref]; |
| 1871 | } |
| 1872 | // Poi cerca negli URL relativi |
| 1873 | elseif (isset($translatedAliasesRelativeMap[$decodedHref])) { |
| 1874 | $translatedAlias = $translatedAliasesRelativeMap[$decodedHref]; |
| 1875 | } |
| 1876 | |
| 1877 | // Se trovato un alias tradotto, sostituisci |
| 1878 | if ($translatedAlias) { |
| 1879 | // Aggiungi data-originalhref se non presente |
| 1880 | if (!isset($linkNode->{'data-originalhref'})) { |
| 1881 | $linkNode->{'data-originalhref'} = $originalPathBeforeNormalization . $queryString . $hashFragment; |
| 1882 | } |
| 1883 | $linkNode->href = $translatedAlias . $queryString . $hashFragment; |
| 1884 | } |
| 1885 | } |
| 1886 | } |
| 1887 | |
| 1888 | // Replace form action URLs: add language prefix and apply translated aliases |
| 1889 | if (!empty($settings['rewrite_form_actions'])) { |
| 1890 | $siteUrl = trailingslashit(get_site_url()); |
| 1891 | $siteHost = parse_url($siteUrl, PHP_URL_HOST); |
| 1892 | $knownLangs = (isset($settings['languages']) && is_array($settings['languages'])) ? array_map('strtolower', $settings['languages']) : [$originalLang, $translatedLang]; |
| 1893 | foreach ($htmlObj->find('form[action]') as $formNode) { |
| 1894 | $action = $formNode->action; |
| 1895 | $decodedAction = rawurldecode(html_entity_decode($action, ENT_QUOTES, 'UTF-8')); |
| 1896 | |
| 1897 | // Skip non-rewritable actions |
| 1898 | if (preg_match('/^(mailto:|tel:|javascript:|#)/i', $decodedAction)) continue; |
| 1899 | if (preg_match('/\.php($|\?|\#)/i', $decodedAction)) continue; |
| 1900 | if (preg_match('/^https?:\/\//i', $decodedAction) && parse_url($decodedAction, PHP_URL_HOST) !== $siteHost) continue; |
| 1901 | |
| 1902 | $queryString = ''; |
| 1903 | if ($settings['ignore_querystring'] == 1 && strpos($decodedAction, '?') !== false) { |
| 1904 | $parts = explode('?', $decodedAction, 2); |
| 1905 | $decodedAction = $parts[0]; |
| 1906 | $queryString = '?' . $parts[1]; |
| 1907 | } |
| 1908 | $originalActionFull = $decodedAction . $queryString; |
| 1909 | |
| 1910 | // Extract relative path |
| 1911 | $actionPath = $decodedAction; |
| 1912 | $isAbsolute = false; |
| 1913 | if (preg_match('/^https?:\/\//i', $actionPath)) { |
| 1914 | $isAbsolute = true; |
| 1915 | $actionPath = parse_url($actionPath, PHP_URL_PATH) ?: '/'; |
| 1916 | } |
| 1917 | |
| 1918 | // Add/replace language prefix |
| 1919 | $pathParts = explode('/', $actionPath); |
| 1920 | $langIndex = 1; |
| 1921 | if (isset($pathParts[$langIndex]) && $pathParts[$langIndex] === 'index.php') $langIndex = 2; |
| 1922 | if (!empty($settings['subfolder_installation'])) $langIndex = 2; |
| 1923 | |
| 1924 | if (isset($pathParts[$langIndex]) && in_array(strtolower($pathParts[$langIndex]), $knownLangs)) { |
| 1925 | $pathParts[$langIndex] = $translatedLang; |
| 1926 | } else { |
| 1927 | array_splice($pathParts, $langIndex, 0, [$translatedLang]); |
| 1928 | } |
| 1929 | |
| 1930 | $rebuiltPath = '/' . implode('/', array_filter($pathParts, function($p) { return $p !== ''; })); |
| 1931 | $rebuiltPath = gpt_trailingslashit_url($rebuiltPath); |
| 1932 | $computedAction = $isAbsolute ? (rtrim($siteUrl, '/') . $rebuiltPath) : $rebuiltPath; |
| 1933 | $computedAction = gpt_trailingslashit_url($computedAction); |
| 1934 | |
| 1935 | // Try alias match |
| 1936 | $translatedAlias = null; |
| 1937 | if (!empty($translatedAliasesMap) || !empty($translatedAliasesRelativeMap)) { |
| 1938 | if (isset($translatedAliasesMap[$computedAction])) { |
| 1939 | $translatedAlias = $translatedAliasesMap[$computedAction]; |
| 1940 | } elseif (isset($translatedAliasesRelativeMap[$rebuiltPath])) { |
| 1941 | $translatedAlias = $translatedAliasesRelativeMap[$rebuiltPath]; |
| 1942 | } elseif (!$isAbsolute && !empty($settings['subfolder_installation'])) { |
| 1943 | // Fallback: form action is relative and includes subfolder path |
| 1944 | // Map keys strip the subfolder, so retry after stripping it |
| 1945 | $siteBasePath = rtrim(parse_url(rtrim($siteUrl, '/'), PHP_URL_PATH) ?: '', '/'); |
| 1946 | if ($siteBasePath !== '' && strpos($rebuiltPath, $siteBasePath . '/') === 0) { |
| 1947 | $rebuiltPathNoSubfolder = substr($rebuiltPath, strlen($siteBasePath)); |
| 1948 | if (isset($translatedAliasesRelativeMap[$rebuiltPathNoSubfolder])) { |
| 1949 | $translatedAlias = $translatedAliasesRelativeMap[$rebuiltPathNoSubfolder]; |
| 1950 | } |
| 1951 | } |
| 1952 | } |
| 1953 | } |
| 1954 | |
| 1955 | $finalAction = $translatedAlias ? ($translatedAlias . $queryString) : ($computedAction . $queryString); |
| 1956 | |
| 1957 | if (!isset($formNode->{'data-originalaction'})) { |
| 1958 | $formNode->{'data-originalaction'} = $originalActionFull; |
| 1959 | } |
| 1960 | $formNode->action = $finalAction; |
| 1961 | } |
| 1962 | } |
| 1963 | |
| 1964 | // Check if also images 'alt' are enabled to be translated |
| 1965 | if (! empty($settings['translate_altimages'])) { |
| 1966 | foreach ($htmlObj->find('img[alt]') as $imgNode) { |
| 1967 | $altText = $imgNode->alt; |
| 1968 | if ($matchQuotes) { |
| 1969 | $altText = str_ireplace('"', "'", $altText); |
| 1970 | } |
| 1971 | foreach ($altTranslationsArray as $originalText => $translatedText) { |
| 1972 | $originalText = normalizeText($originalText); |
| 1973 | $altText = normalizeText($altText); |
| 1974 | |
| 1975 | $prevAltText = $altText; |
| 1976 | if ($caseInsensitive) { |
| 1977 | $altText = str_ireplace(trim($originalText), $translatedText, $altText); |
| 1978 | } else { |
| 1979 | $altText = str_replace(trim($originalText), $translatedText, $altText); |
| 1980 | } |
| 1981 | if ($altText !== $prevAltText) { |
| 1982 | break; |
| 1983 | } |
| 1984 | } |
| 1985 | $imgNode->alt = $altText; |
| 1986 | } |
| 1987 | |
| 1988 | foreach ($htmlObj->find('img[title]') as $imgNode) { |
| 1989 | $titleText = $imgNode->title; |
| 1990 | if ($matchQuotes) { |
| 1991 | $titleText = str_ireplace('"', "'", $titleText); |
| 1992 | } |
| 1993 | foreach ($altTranslationsArray as $originalText => $translatedText) { |
| 1994 | $originalText = normalizeText($originalText); |
| 1995 | $titleText = normalizeText($titleText); |
| 1996 | |
| 1997 | $prevTitleText = $titleText; |
| 1998 | if ($caseInsensitive) { |
| 1999 | $titleText = str_ireplace(trim($originalText), $translatedText, $titleText); |
| 2000 | } else { |
| 2001 | $titleText = str_replace(trim($originalText), $translatedText, $titleText); |
| 2002 | } |
| 2003 | if ($titleText !== $prevTitleText) { |
| 2004 | break; |
| 2005 | } |
| 2006 | } |
| 2007 | $imgNode->title = $titleText; |
| 2008 | } |
| 2009 | |
| 2010 | // Check if also images 'src' are enabled to be translated |
| 2011 | if (! empty($settings['translate_srcimages'])) { |
| 2012 | foreach ($htmlObj->find('img[src]') as $imgNode) { |
| 2013 | $srcValue = ltrim($imgNode->src, '/'); |
| 2014 | |
| 2015 | if ($matchQuotes) { |
| 2016 | $srcValue = str_ireplace('"', "'", $srcValue); |
| 2017 | } |
| 2018 | |
| 2019 | $translatedSrcValue = null; |
| 2020 | foreach ($altTranslationsArray as $originalText => $translatedText) { |
| 2021 | $originalText = normalizeText($originalText); |
| 2022 | $srcValue = normalizeText($srcValue); |
| 2023 | |
| 2024 | $originalText = ltrim($originalText, '/'); |
| 2025 | if ($caseInsensitive) { |
| 2026 | if (strcasecmp(trim($originalText), $srcValue) === 0) { |
| 2027 | $translatedSrcValue = $translatedText; |
| 2028 | break; |
| 2029 | } |
| 2030 | } else { |
| 2031 | if (trim($originalText) === $srcValue) { |
| 2032 | $translatedSrcValue = $translatedText; |
| 2033 | break; |
| 2034 | } |
| 2035 | } |
| 2036 | } |
| 2037 | |
| 2038 | // If a translation was found, replace src and srcset |
| 2039 | if ($translatedSrcValue !== null) { |
| 2040 | // Replace the src attribute |
| 2041 | $imgNode->src = $translatedSrcValue; |
| 2042 | |
| 2043 | // Check if srcset exists and replace it with the translated src value |
| 2044 | if (isset($imgNode->srcset)) { |
| 2045 | $imgNode->srcset = $translatedSrcValue; |
| 2046 | } |
| 2047 | } |
| 2048 | } |
| 2049 | } |
| 2050 | |
| 2051 | // Check if also iframes 'src' are enabled to be translated |
| 2052 | if (! empty($settings['translate_srciframes'])) { |
| 2053 | foreach ($htmlObj->find('iframe[src]') as $iframeNode) { |
| 2054 | $srcValue = ltrim($iframeNode->src, '/'); |
| 2055 | |
| 2056 | if ($matchQuotes) { |
| 2057 | $srcValue = str_ireplace('"', "'", $srcValue); |
| 2058 | } |
| 2059 | |
| 2060 | $translatedSrcValue = null; |
| 2061 | foreach ($altTranslationsArray as $originalText => $translatedText) { |
| 2062 | $originalText = normalizeText($originalText); |
| 2063 | $srcValue = normalizeText($srcValue); |
| 2064 | |
| 2065 | $originalText = ltrim($originalText, '/'); |
| 2066 | if ($caseInsensitive) { |
| 2067 | if (strcasecmp(trim($originalText), $srcValue) === 0) { |
| 2068 | $translatedSrcValue = $translatedText; |
| 2069 | break; |
| 2070 | } |
| 2071 | } else { |
| 2072 | if (trim($originalText) === $srcValue) { |
| 2073 | $translatedSrcValue = $translatedText; |
| 2074 | break; |
| 2075 | } |
| 2076 | } |
| 2077 | } |
| 2078 | |
| 2079 | if ($translatedSrcValue !== null) { |
| 2080 | $iframeNode->src = $translatedSrcValue; |
| 2081 | } |
| 2082 | } |
| 2083 | } |
| 2084 | |
| 2085 | // Check if also videos 'src' are enabled to be translated |
| 2086 | if (! empty($settings['translate_srcvideos'])) { |
| 2087 | foreach ($htmlObj->find('video[src], video source[src]') as $videoNode) { |
| 2088 | $srcValue = ltrim($videoNode->src, '/'); |
| 2089 | |
| 2090 | if ($matchQuotes) { |
| 2091 | $srcValue = str_ireplace('"', "'", $srcValue); |
| 2092 | } |
| 2093 | |
| 2094 | $translatedSrcValue = null; |
| 2095 | foreach ($altTranslationsArray as $originalText => $translatedText) { |
| 2096 | $originalText = normalizeText($originalText); |
| 2097 | $srcValue = normalizeText($srcValue); |
| 2098 | |
| 2099 | $originalText = ltrim($originalText, '/'); |
| 2100 | if ($caseInsensitive) { |
| 2101 | if (strcasecmp(trim($originalText), $srcValue) === 0) { |
| 2102 | $translatedSrcValue = $translatedText; |
| 2103 | break; |
| 2104 | } |
| 2105 | } else { |
| 2106 | if (trim($originalText) === $srcValue) { |
| 2107 | $translatedSrcValue = $translatedText; |
| 2108 | break; |
| 2109 | } |
| 2110 | } |
| 2111 | } |
| 2112 | |
| 2113 | if ($translatedSrcValue !== null) { |
| 2114 | $videoNode->src = $translatedSrcValue; |
| 2115 | } |
| 2116 | } |
| 2117 | } |
| 2118 | } |
| 2119 | |
| 2120 | // Translate iframe locale parameter (e.g. Stripe payment forms) |
| 2121 | if (! empty($settings['translate_iframe_locale'])) { |
| 2122 | $currentLangCode = strtolower(explode('-', $currentLanguage)[0]); |
| 2123 | foreach ($htmlObj->find('iframe[src]') as $iframeNode) { |
| 2124 | $src = $iframeNode->src; |
| 2125 | if (strpos($src, 'locale=') !== false) { |
| 2126 | $iframeNode->src = preg_replace('/locale=[a-z]{2}/i', 'locale=' . $currentLangCode, $src); |
| 2127 | } |
| 2128 | } |
| 2129 | } |
| 2130 | |
| 2131 | // Translate <meta name="description"> |
| 2132 | foreach ($htmlObj->find('meta[name=description], meta[name=twitter:description], meta[property=og:description], meta[name=dc.description]') as $metaNode) { |
| 2133 | $contentValue = $metaNode->content; |
| 2134 | |
| 2135 | // DECODIFICA HTML entities PRIMA del confronto |
| 2136 | $contentValue = html_entity_decode($contentValue, ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 2137 | |
| 2138 | if ($matchQuotes) { |
| 2139 | $contentValue = str_ireplace('"', "'", $contentValue); |
| 2140 | } |
| 2141 | |
| 2142 | foreach ($altTranslationsArray as $originalText => $translatedText) { |
| 2143 | $prev = $contentValue; |
| 2144 | if ($caseInsensitive) { |
| 2145 | $contentValue = str_ireplace(trim($originalText), $translatedText, $contentValue); |
| 2146 | } else { |
| 2147 | $contentValue = str_replace(trim($originalText), $translatedText, $contentValue); |
| 2148 | } |
| 2149 | if ($contentValue !== $prev) break; |
| 2150 | } |
| 2151 | |
| 2152 | // RI-ENCODIFICA per l'HTML output (sicurezza) |
| 2153 | $contentValue = htmlspecialchars($contentValue, ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 2154 | |
| 2155 | $metaNode->content = $contentValue; |
| 2156 | } |
| 2157 | |
| 2158 | foreach ($htmlObj->find('meta[property=og:title]') as $metaOgTitleNode) { |
| 2159 | // Find the <title> tag to get the translated title |
| 2160 | $titleNode = $htmlObj->find('title', 0); |
| 2161 | |
| 2162 | if ($titleNode) { |
| 2163 | // Get the translated title text |
| 2164 | $translatedTitle = $titleNode->plaintext; |
| 2165 | |
| 2166 | // Set og:title with the same value as <title> |
| 2167 | $metaOgTitleNode->content = trim($translatedTitle); |
| 2168 | } |
| 2169 | } |
| 2170 | |
| 2171 | // Translate <meta name="twitter:title"> with the same value as <title> tag |
| 2172 | foreach ($htmlObj->find('meta[name=twitter:title]') as $metaTwitterTitleNode) { |
| 2173 | // Find the <title> tag to get the translated title |
| 2174 | $titleNode = $htmlObj->find('title', 0); |
| 2175 | |
| 2176 | if ($titleNode) { |
| 2177 | // Get the translated title text |
| 2178 | $translatedTitle = $titleNode->plaintext; |
| 2179 | |
| 2180 | // Set twitter:title with the same value as <title> |
| 2181 | $metaTwitterTitleNode->content = trim($translatedTitle); |
| 2182 | } |
| 2183 | } |
| 2184 | |
| 2185 | // Translate <meta name="dc:title"> (Dublin Core - SEOPress) with the same value as <title> tag |
| 2186 | foreach ($htmlObj->find('meta[name=dc.title]') as $metaDcTitleNode) { |
| 2187 | // Find the <title> tag to get the translated title |
| 2188 | $titleNode = $htmlObj->find('title', 0); |
| 2189 | |
| 2190 | if ($titleNode) { |
| 2191 | // Get the translated title text |
| 2192 | $translatedTitle = $titleNode->plaintext; |
| 2193 | |
| 2194 | // Set dc:title with the same value as <title> |
| 2195 | $metaDcTitleNode->content = trim($translatedTitle); |
| 2196 | } |
| 2197 | } |
| 2198 | |
| 2199 | if ($bodyElement = $htmlObj->find('body', 0)) { |
| 2200 | $bodyElement->setAttribute('data-gptranslateskip', '1'); |
| 2201 | $bodyElement->setAttribute ( 'data-gptranslateoriginalalias', htmlspecialchars($row['pagelink'], ENT_QUOTES, 'UTF-8') ); |
| 2202 | } |
| 2203 | |
| 2204 | $modifiedHtml = $htmlObj->save(); |
| 2205 | $modifiedHtml = html_entity_decode($modifiedHtml, ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 2206 | |
| 2207 | // RESTORE protected script/style content (if protection was enabled) |
| 2208 | if (!empty($settings['protect_script_style'])) { |
| 2209 | foreach ($protectedScriptsStyles as $placeholder => $originalContent) { |
| 2210 | $modifiedHtml = str_replace($placeholder, $originalContent, $modifiedHtml); |
| 2211 | } |
| 2212 | } |
| 2213 | $html = $modifiedHtml; |
| 2214 | } catch (Exception $e) { |
| 2215 | // Handle exceptions as needed |
| 2216 | } |
| 2217 | } elseif ($settings['serverside_translations_method'] == 'strireplace') { |
| 2218 | // Solution 3: simplest approach, unconditional str_ireplace that could cause unintentional replacements |
| 2219 | try { |
| 2220 | $translationsArray = json_decode($row['translations'], true) ?? []; |
| 2221 | $altTranslationsArray = json_decode($row['alt_translations'] ?? '', true) ?: []; |
| 2222 | |
| 2223 | // Do body page translations replacements |
| 2224 | uksort($translationsArray, function ($a, $b) { |
| 2225 | return strlen($b) - strlen($a); |
| 2226 | }); |
| 2227 | |
| 2228 | $caseInsensitive = !empty($settings['serverside_translations_caseinsensitive']); |
| 2229 | $matchQuotes = !empty($settings['serverside_translations_matchquotes']); |
| 2230 | |
| 2231 | foreach ($translationsArray as $originalText => $translatedText) { |
| 2232 | $originalText = normalizeText($originalText); |
| 2233 | $html = normalizeText($html); |
| 2234 | |
| 2235 | if ($caseInsensitive) { |
| 2236 | $html = str_ireplace(trim($originalText), $translatedText, $html); |
| 2237 | |
| 2238 | // Check also if both single quotes or double quotes should be checked to replace |
| 2239 | if ($matchQuotes && strpos($originalText, "'") !== false) { |
| 2240 | $originalTextAlt = str_ireplace("'", '"', $originalText); |
| 2241 | $html = str_ireplace(trim($originalTextAlt), $translatedText, $html); |
| 2242 | } |
| 2243 | } else { |
| 2244 | $html = str_replace(trim($originalText), $translatedText, $html); |
| 2245 | |
| 2246 | if ($matchQuotes && strpos($originalText, "'") !== false) { |
| 2247 | $originalTextAlt = str_replace("'", '"', $originalText); |
| 2248 | $html = str_replace(trim($originalTextAlt), $translatedText, $html); |
| 2249 | } |
| 2250 | } |
| 2251 | } |
| 2252 | |
| 2253 | // Check if also images 'alt' are enabled to be translated |
| 2254 | if (!empty($settings['translate_altimages'])) { |
| 2255 | foreach ($altTranslationsArray as $originalAlt => $translatedAlt) { |
| 2256 | if ($matchQuotes) { |
| 2257 | $originalAlt = str_ireplace('"', "'", $originalAlt); |
| 2258 | } |
| 2259 | if ($caseInsensitive) { |
| 2260 | $html = str_ireplace('alt="' . trim($originalAlt) . '"', 'alt="' . $translatedAlt . '"', $html); |
| 2261 | $html = str_ireplace("alt='" . trim($originalAlt) . "'", "alt='" . $translatedAlt . "'", $html); |
| 2262 | } else { |
| 2263 | $html = str_replace('alt="' . trim($originalAlt) . '"', 'alt="' . $translatedAlt . '"', $html); |
| 2264 | $html = str_replace("alt='" . trim($originalAlt) . "'", "alt='" . $translatedAlt . "'", $html); |
| 2265 | } |
| 2266 | } |
| 2267 | } |
| 2268 | |
| 2269 | $html = str_ireplace('<body', '<body data-gptranslateskip="1" data-gptranslateoriginalalias="' . $row['pagelink'] . '"', $html); |
| 2270 | |
| 2271 | } catch (Exception $e) { |
| 2272 | // Handle exceptions as needed |
| 2273 | } |
| 2274 | } |
| 2275 | |
| 2276 | return $html; |
| 2277 | } ); |
| 2278 | } ); |
| 2279 |