13832d0abSNetali<?php 23832d0abSNetali/** 33832d0abSNetali * Deepl Autotranslate Plugin 43832d0abSNetali * 53832d0abSNetali * @author Jennifer Graul <me@netali.de> 63832d0abSNetali */ 73832d0abSNetali 83832d0abSNetaliif(!defined('DOKU_INC')) die(); 93832d0abSNetali 10*81931e50SNetaliuse \dokuwiki\HTTP\DokuHTTPClient; 11*81931e50SNetali 123832d0abSNetaliclass action_plugin_deeplautotranslate extends DokuWiki_Action_Plugin { 133832d0abSNetali 143832d0abSNetali // manual mapping of ISO-languages to DeepL-languages to fix inconsistent naming 153832d0abSNetali private $langs = [ 163832d0abSNetali 'bg' => 'BG', 173832d0abSNetali 'cs' => 'CS', 183832d0abSNetali 'da' => 'DA', 193832d0abSNetali 'de' => 'DE', 203832d0abSNetali 'de-informal' => 'DE', 213832d0abSNetali 'el' => 'EL', 223832d0abSNetali 'en' => 'EN-GB', 233832d0abSNetali 'es' => 'ES', 243832d0abSNetali 'et' => 'ET', 253832d0abSNetali 'fi' => 'FI', 263832d0abSNetali 'fr' => 'FR', 273832d0abSNetali 'hu' => 'HU', 283832d0abSNetali 'hu-formal' => 'HU', 293832d0abSNetali 'it' => 'IT', 303832d0abSNetali 'ja' => 'JA', 313832d0abSNetali 'lt' => 'LT', 323832d0abSNetali 'lv' => 'LV', 333832d0abSNetali 'nl' => 'NL', 343832d0abSNetali 'pl' => 'PL', 353832d0abSNetali 'pt' => 'PT-PT', 363832d0abSNetali 'ro' => 'RO', 373832d0abSNetali 'ru' => 'RU', 383832d0abSNetali 'sk' => 'SK', 393832d0abSNetali 'sl' => 'SL', 403832d0abSNetali 'sv' => 'SV', 413832d0abSNetali 'zh' => 'ZH' 423832d0abSNetali ]; 433832d0abSNetali 443832d0abSNetali /** 453832d0abSNetali * Register its handlers with the DokuWiki's event controller 463832d0abSNetali */ 473832d0abSNetali public function register(Doku_Event_Handler $controller) { 483832d0abSNetali $controller->register_hook('ACTION_ACT_PREPROCESS','BEFORE', $this, 'autotrans_direct'); 493832d0abSNetali $controller->register_hook('COMMON_PAGETPL_LOAD','AFTER', $this, 'autotrans_editor'); 503832d0abSNetali } 513832d0abSNetali 523832d0abSNetali public function autotrans_direct(Doku_Event $event, $param) { 533832d0abSNetali if ($this->get_mode() != 'direct') return; 543832d0abSNetali if ($event->data != 'show') return; 553832d0abSNetali 563832d0abSNetali if (!$this->check_do_translation()) return; 573832d0abSNetali 583832d0abSNetali global $ID; 593832d0abSNetali global $INFO; 603832d0abSNetali 613832d0abSNetali $org_page_text = $this->get_org_page_text(); 623832d0abSNetali $translated_text = $this->deepl_translate($org_page_text, $this->langs[$this->get_target_lang()]); 633832d0abSNetali 643832d0abSNetali if ($translated_text === '') { 653832d0abSNetali return; 663832d0abSNetali } 673832d0abSNetali 683832d0abSNetali saveWikiText($ID, $translated_text, 'Automatic translation'); 693832d0abSNetali 703832d0abSNetali $INFO = pageinfo(); 713832d0abSNetali } 723832d0abSNetali 733832d0abSNetali public function autotrans_editor(Doku_Event $event, $param) { 743832d0abSNetali if ($this->get_mode() != 'editor') return; 753832d0abSNetali 763832d0abSNetali if (!$this->check_do_translation()) return; 773832d0abSNetali 783832d0abSNetali $org_page_text = $this->get_org_page_text(); 793832d0abSNetali 803832d0abSNetali $event->data['tpl'] = $this->deepl_translate($org_page_text, $this->langs[$this->get_target_lang()]); 813832d0abSNetali } 823832d0abSNetali 833832d0abSNetali private function get_mode(): string { 843832d0abSNetali global $ID; 853832d0abSNetali if ($this->getConf('editor_regex')) { 863832d0abSNetali if (preg_match('/' . $this->getConf('editor_regex') . '/', $ID) === 1) return 'editor'; 873832d0abSNetali } 883832d0abSNetali if ($this->getConf('direct_regex')) { 893832d0abSNetali if (preg_match('/' . $this->getConf('direct_regex') . '/', $ID) === 1) return 'direct'; 903832d0abSNetali } 913832d0abSNetali return $this->getConf('mode'); 923832d0abSNetali } 933832d0abSNetali 943832d0abSNetali private function get_target_lang(): string { 953832d0abSNetali global $ID; 963832d0abSNetali $split_id = explode(':', $ID); 973832d0abSNetali return array_shift($split_id); 983832d0abSNetali } 993832d0abSNetali 1003832d0abSNetali private function get_org_page_text(): string { 1013832d0abSNetali global $ID; 1023832d0abSNetali 1033832d0abSNetali $split_id = explode(':', $ID); 1043832d0abSNetali array_shift($split_id); 1053832d0abSNetali $org_id = implode(':', $split_id); 1063832d0abSNetali 1073832d0abSNetali return rawWiki($org_id); 1083832d0abSNetali } 1093832d0abSNetali 1103832d0abSNetali private function check_do_translation(): bool { 1113832d0abSNetali global $INFO; 1123832d0abSNetali // only translate if the current page does not exist 1133832d0abSNetali if ($INFO['exists']) return false; 1143832d0abSNetali 1153832d0abSNetali global $ID; 1163832d0abSNetali 1173832d0abSNetali // skip blacklisted namespaces and pages 1183832d0abSNetali if ($this->getConf('blacklist_regex')) { 1193832d0abSNetali if (preg_match('/' . $this->getConf('blacklist_regex') . '/', $ID) === 1) return false; 1203832d0abSNetali } 1213832d0abSNetali 1223832d0abSNetali $split_id = explode(':', $ID); 1233832d0abSNetali $lang_ns = array_shift($split_id); 1243832d0abSNetali // only translate if the current page is in a language namespace 1253832d0abSNetali if (!array_key_exists($lang_ns, $this->langs)) return false; 1263832d0abSNetali 1273832d0abSNetali $org_id = implode(':', $split_id); 1283832d0abSNetali // check if the original page exists 1293832d0abSNetali if (!page_exists($org_id)) return false; 1303832d0abSNetali 1313832d0abSNetali return true; 1323832d0abSNetali } 1333832d0abSNetali 1343832d0abSNetali private function deepl_translate($text, $target_lang): string { 1353832d0abSNetali if (!$this->getConf('api_key')) return ''; 1363832d0abSNetali 1373832d0abSNetali $text = $this->insert_ignore_tags($text); 1383832d0abSNetali 1393832d0abSNetali $data = [ 1403832d0abSNetali 'auth_key' => $this->getConf('api_key'), 1413832d0abSNetali 'target_lang' => $target_lang, 1423832d0abSNetali 'tag_handling' => 'xml', 1433832d0abSNetali 'ignore_tags' => 'ignore', 1443832d0abSNetali 'text' => $text 1453832d0abSNetali ]; 1463832d0abSNetali 1473832d0abSNetali if ($this->getConf('api') == 'free') { 148*81931e50SNetali $url = 'https://api-free.deepl.com/v2/translate'; 1493832d0abSNetali } else { 150*81931e50SNetali $url = 'https://api.deepl.com/v2/translate'; 1513832d0abSNetali } 1523832d0abSNetali 153*81931e50SNetali $http = new DokuHTTPClient(); 154*81931e50SNetali $raw_response = $http->post($url, $data); 1553832d0abSNetali 1563832d0abSNetali // if any error occurred return an empty string 157*81931e50SNetali if ($http->status >= 400) return ''; 1583832d0abSNetali 1593832d0abSNetali $json_response = json_decode($raw_response, true); 1603832d0abSNetali $translated_text = $json_response['translations'][0]['text']; 1613832d0abSNetali 1623832d0abSNetali $translated_text = $this->remove_ignore_tags($translated_text); 1633832d0abSNetali 1643832d0abSNetali return $translated_text; 1653832d0abSNetali } 1663832d0abSNetali 1673832d0abSNetali private function insert_ignore_tags($text): string { 1683832d0abSNetali $text = str_replace('[[', '<ignore>[[', $text); 1693832d0abSNetali $text = str_replace('{{', '<ignore>{{', $text); 1703832d0abSNetali $text = str_replace(']]', ']]</ignore>', $text); 1713832d0abSNetali $text = str_replace('}}', '}}</ignore>', $text); 1723832d0abSNetali 1733832d0abSNetali $ignored_expressions = explode(':', $this->getConf('ignored_expressions')); 1743832d0abSNetali 1753832d0abSNetali foreach ($ignored_expressions as $expression) { 1763832d0abSNetali $text = str_replace($expression, '<ignore>' . $expression . '</ignore>', $text); 1773832d0abSNetali } 1783832d0abSNetali 1793832d0abSNetali return $text; 1803832d0abSNetali } 1813832d0abSNetali 1823832d0abSNetali private function remove_ignore_tags($text): string { 1833832d0abSNetali $text = str_replace('<ignore>[[', '[[', $text); 1843832d0abSNetali $text = str_replace('<ignore>{{', '{{', $text); 1853832d0abSNetali $text = str_replace(']]</ignore>', ']]', $text); 1863832d0abSNetali $text = str_replace('}}</ignore>', '}}', $text); 1873832d0abSNetali 1883832d0abSNetali $ignored_expressions = explode(':', $this->getConf('ignored_expressions')); 1893832d0abSNetali 1903832d0abSNetali foreach ($ignored_expressions as $expression) { 1913832d0abSNetali $text = str_replace('<ignore>' . $expression . '</ignore>', $expression, $text); 1923832d0abSNetali } 1933832d0abSNetali 1943832d0abSNetali return $text; 1953832d0abSNetali } 1963832d0abSNetali} 1973832d0abSNetali 198