xref: /plugin/deeplautotranslate/action.php (revision ff327fe62164f36215cc111c2e178e34f292f457)
13832d0abSNetali<?php
23832d0abSNetali/**
33832d0abSNetali * Deepl Autotranslate Plugin
43832d0abSNetali *
53832d0abSNetali * @author     Jennifer Graul <me@netali.de>
63832d0abSNetali */
73832d0abSNetali
83832d0abSNetaliif(!defined('DOKU_INC')) die();
93832d0abSNetali
1081931e50SNetaliuse \dokuwiki\HTTP\DokuHTTPClient;
113c636ad3SNetaliuse \dokuwiki\plugin\deeplautotranslate\MenuItem;
1281931e50SNetali
133832d0abSNetaliclass action_plugin_deeplautotranslate extends DokuWiki_Action_Plugin {
143832d0abSNetali
153832d0abSNetali    // manual mapping of ISO-languages to DeepL-languages to fix inconsistent naming
163832d0abSNetali    private $langs = [
173832d0abSNetali        'bg' => 'BG',
183832d0abSNetali        'cs' => 'CS',
193832d0abSNetali        'da' => 'DA',
203832d0abSNetali        'de' => 'DE',
213832d0abSNetali        'de-informal' => 'DE',
223832d0abSNetali        'el' => 'EL',
233832d0abSNetali        'en' => 'EN-GB',
243832d0abSNetali        'es' => 'ES',
253832d0abSNetali        'et' => 'ET',
263832d0abSNetali        'fi' => 'FI',
273832d0abSNetali        'fr' => 'FR',
283832d0abSNetali        'hu' => 'HU',
293832d0abSNetali        'hu-formal' => 'HU',
303832d0abSNetali        'it' => 'IT',
313832d0abSNetali        'ja' => 'JA',
323832d0abSNetali        'lt' => 'LT',
333832d0abSNetali        'lv' => 'LV',
343832d0abSNetali        'nl' => 'NL',
353832d0abSNetali        'pl' => 'PL',
363832d0abSNetali        'pt' => 'PT-PT',
373832d0abSNetali        'ro' => 'RO',
383832d0abSNetali        'ru' => 'RU',
393832d0abSNetali        'sk' => 'SK',
403832d0abSNetali        'sl' => 'SL',
413832d0abSNetali        'sv' => 'SV',
423832d0abSNetali        'zh' => 'ZH'
433832d0abSNetali    ];
443832d0abSNetali
453832d0abSNetali    /**
463832d0abSNetali     * Register its handlers with the DokuWiki's event controller
473832d0abSNetali     */
483832d0abSNetali    public function register(Doku_Event_Handler $controller) {
49153e4498SNetali        $controller->register_hook('ACTION_ACT_PREPROCESS','BEFORE', $this, 'preprocess');
503832d0abSNetali        $controller->register_hook('COMMON_PAGETPL_LOAD','AFTER', $this, 'autotrans_editor');
513c636ad3SNetali        $controller->register_hook('MENU_ITEMS_ASSEMBLY', 'AFTER', $this, 'add_menu_button');
523c636ad3SNetali    }
533c636ad3SNetali
540180404cSNetali    public function add_menu_button(Doku_Event $event): void {
55153e4498SNetali        global $ID;
56bbb1fba9SNetali        global $ACT;
57bbb1fba9SNetali
58bbb1fba9SNetali        if ($ACT != 'show') return;
59153e4498SNetali
603c636ad3SNetali        if ($event->data['view'] != 'page') return;
613c636ad3SNetali
623c636ad3SNetali        if (!$this->getConf('show_button')) return;
63153e4498SNetali
64153e4498SNetali        $split_id = explode(':', $ID);
65153e4498SNetali        $lang_ns = array_shift($split_id);
66153e4498SNetali        // check if we are in a language namespace
67153e4498SNetali        if (array_key_exists($lang_ns, $this->langs)) {
68*ff327fe6SNetali            if($this->getConf('default_lang_in_ns') and $lang_ns === $this->get_default_lang()) {
6999da9a08SNetali                // if the default lang is in a namespace and we are in that namespace --> check for push translation
7099da9a08SNetali                if (!$this->check_do_push_translate()) return;
7199da9a08SNetali            } else {
72153e4498SNetali                // in language namespace --> check if we should translate
733c636ad3SNetali                if (!$this->check_do_translation(true)) return;
7499da9a08SNetali            }
75153e4498SNetali        } else {
7699da9a08SNetali            // do not show the button if we are not in a language namespace and the default language is in a namespace
7799da9a08SNetali            if($this->getConf('default_lang_in_ns')) return;
7899da9a08SNetali            // not in language namespace and default language is npt in a namespace --> check if we should show the push translate button
79153e4498SNetali            if (!$this->check_do_push_translate()) return;
80153e4498SNetali        }
813c636ad3SNetali
823c636ad3SNetali        array_splice($event->data['items'], -1, 0, [new MenuItem()]);
833832d0abSNetali    }
843832d0abSNetali
85153e4498SNetali    public function preprocess(Doku_Event  $event, $param): void {
863832d0abSNetali        global $ID;
873c636ad3SNetali
883c636ad3SNetali        // check if action is show or translate
893c636ad3SNetali        if ($event->data != 'show' and $event->data != 'translate') return;
903c636ad3SNetali
91153e4498SNetali        $split_id = explode(':', $ID);
92153e4498SNetali        $lang_ns = array_shift($split_id);
93153e4498SNetali        // check if we are in a language namespace
94153e4498SNetali        if (array_key_exists($lang_ns, $this->langs)) {
95*ff327fe6SNetali            if($this->getConf('default_lang_in_ns') and $lang_ns === $this->get_default_lang()) {
9699da9a08SNetali                // if the default lang is in a namespace and we are in that namespace --> push translate
9799da9a08SNetali                $this->push_translate($event);
9899da9a08SNetali            } else {
9999da9a08SNetali                // in language namespace --> autotrans direct
100153e4498SNetali                $this->autotrans_direct($event);
10199da9a08SNetali            }
102153e4498SNetali        } else {
103153e4498SNetali            // not in language namespace --> push translate
104153e4498SNetali            $this->push_translate($event);
105153e4498SNetali        }
106153e4498SNetali    }
107153e4498SNetali
108153e4498SNetali    private function autotrans_direct(Doku_Event $event): void {
109153e4498SNetali        global $ID;
110153e4498SNetali
1113c636ad3SNetali        // abort if action is translate and the translate button is disabled
1123c636ad3SNetali        if ($event->data == 'translate' and !$this->getConf('show_button')) return;
1133c636ad3SNetali
1143c636ad3SNetali        // do nothing on show action when mode is not direct
1153c636ad3SNetali        if ($event->data == 'show' and $this->get_mode() != 'direct') return;
1163c636ad3SNetali
1173c636ad3SNetali        // allow translation of existing pages is we are in the translate action
1183c636ad3SNetali        $allow_existing = ($event->data == 'translate');
1193c636ad3SNetali
1203c636ad3SNetali        // reset action to show
1213c636ad3SNetali        $event->data = 'show';
1223c636ad3SNetali
123153e4498SNetali        if (!$this->check_do_translation($allow_existing)) {
124153e4498SNetali            return;
125153e4498SNetali        }
1263832d0abSNetali
1270180404cSNetali        $org_page_info = $this->get_org_page_info();
1280180404cSNetali        $translated_text = $this->deepl_translate($org_page_info["text"], $this->get_target_lang(), $org_page_info["ns"]);
1293832d0abSNetali
130153e4498SNetali        if ($translated_text === '') {
131153e4498SNetali            return;
132153e4498SNetali        }
1333832d0abSNetali
1343832d0abSNetali        saveWikiText($ID, $translated_text, 'Automatic translation');
1353832d0abSNetali
136153e4498SNetali        msg($this->getLang('msg_translation_success'), 1);
137153e4498SNetali
1383c636ad3SNetali        // reload the page after translation
1393c636ad3SNetali        send_redirect(wl($ID));
1403832d0abSNetali    }
1413832d0abSNetali
142153e4498SNetali    public function autotrans_editor(Doku_Event $event, $param): void {
1433832d0abSNetali        if ($this->get_mode() != 'editor') return;
1443832d0abSNetali
1453832d0abSNetali        if (!$this->check_do_translation()) return;
1463832d0abSNetali
1470180404cSNetali        $org_page_info = $this->get_org_page_info();
1483832d0abSNetali
1490180404cSNetali        $event->data['tpl'] = $this->deepl_translate($org_page_info["text"], $this->get_target_lang(), $org_page_info["ns"]);
1503832d0abSNetali    }
1513832d0abSNetali
152153e4498SNetali    private function push_translate(Doku_Event $event): void {
153153e4498SNetali        global $ID;
154153e4498SNetali
155153e4498SNetali        // check if action is translate
156153e4498SNetali        if ($event->data != 'translate') return;
157153e4498SNetali
158153e4498SNetali        // check if button is enabled
159153e4498SNetali        if (!$this->getConf('show_button')) {
160153e4498SNetali            send_redirect(wl($ID));
161153e4498SNetali            return;
162153e4498SNetali        }
163153e4498SNetali
164153e4498SNetali        if (!$this->check_do_push_translate()) {
165153e4498SNetali            send_redirect(wl($ID));
166153e4498SNetali            return;
167153e4498SNetali        }
168153e4498SNetali
169153e4498SNetali        // push translate
170153e4498SNetali        $push_langs = $this->get_push_langs();
171153e4498SNetali        $org_page_text = rawWiki($ID);
172153e4498SNetali        foreach ($push_langs as $lang) {
173153e4498SNetali            // skip invalid languages
174153e4498SNetali            if (!array_key_exists($lang, $this->langs)) {
175153e4498SNetali                msg($this->getLang('msg_translation_fail_invalid_lang') . $lang, -1);
176153e4498SNetali                continue;
177153e4498SNetali            }
178153e4498SNetali
17999da9a08SNetali            if ($this->getConf('default_lang_in_ns')) {
18099da9a08SNetali                // if default lang is in ns: replace language namespace in ID
18199da9a08SNetali                $split_id = explode(':', $ID);
18299da9a08SNetali                array_shift($split_id);
18399da9a08SNetali                $lang_id = implode(':', $split_id);
18499da9a08SNetali                $lang_id = $lang . ':' . $lang_id;
18599da9a08SNetali            } else {
18699da9a08SNetali                // if default lang is not in ns: add language namespace to ID
187153e4498SNetali                $lang_id = $lang . ':' . $ID;
18899da9a08SNetali            }
189153e4498SNetali
190153e4498SNetali            // check permissions
191153e4498SNetali            $perm = auth_quickaclcheck($ID);
192153e4498SNetali            $exists = page_exists($lang_id);
193153e4498SNetali            if (($exists and $perm < AUTH_EDIT) or (!$exists and $perm < AUTH_CREATE)) {
194153e4498SNetali                msg($this->getLang('msg_translation_fail_no_permissions') . $lang_id, -1);
195153e4498SNetali                continue;
196153e4498SNetali            }
197153e4498SNetali
1980180404cSNetali            $translated_text = $this->deepl_translate($org_page_text, $lang, getNS($ID));
199153e4498SNetali            saveWikiText($lang_id, $translated_text, 'Automatic push translation');
200153e4498SNetali        }
201153e4498SNetali
202153e4498SNetali        msg($this->getLang('msg_translation_success'), 1);
203153e4498SNetali
204153e4498SNetali        // reload the page after translation to clear the action
205153e4498SNetali        send_redirect(wl($ID));
206153e4498SNetali    }
207153e4498SNetali
2083832d0abSNetali    private function get_mode(): string {
2093832d0abSNetali        global $ID;
2103832d0abSNetali        if ($this->getConf('editor_regex')) {
2113832d0abSNetali            if (preg_match('/' . $this->getConf('editor_regex') . '/', $ID) === 1) return 'editor';
2123832d0abSNetali        }
2133832d0abSNetali        if ($this->getConf('direct_regex')) {
2143832d0abSNetali            if (preg_match('/' . $this->getConf('direct_regex') . '/', $ID) === 1) return 'direct';
2153832d0abSNetali        }
2163832d0abSNetali        return $this->getConf('mode');
2173832d0abSNetali    }
2183832d0abSNetali
2193832d0abSNetali    private function get_target_lang(): string {
2203832d0abSNetali        global $ID;
2213832d0abSNetali        $split_id = explode(':', $ID);
2223832d0abSNetali        return array_shift($split_id);
2233832d0abSNetali    }
2243832d0abSNetali
225*ff327fe6SNetali    private function get_default_lang(): string {
226*ff327fe6SNetali        global $conf;
227*ff327fe6SNetali
228*ff327fe6SNetali        if (empty($conf['lang_before_translation'])) {
229*ff327fe6SNetali            $default_lang = $conf['lang'];
230*ff327fe6SNetali        } else {
231*ff327fe6SNetali            $default_lang = $conf['lang_before_translation'];
232*ff327fe6SNetali        }
233*ff327fe6SNetali
234*ff327fe6SNetali        return $default_lang;
235*ff327fe6SNetali    }
236*ff327fe6SNetali
2370180404cSNetali    private function get_org_page_info(): array {
2383832d0abSNetali        global $ID;
2393832d0abSNetali
2403832d0abSNetali        $split_id = explode(':', $ID);
2413832d0abSNetali        array_shift($split_id);
2423832d0abSNetali        $org_id = implode(':', $split_id);
2433832d0abSNetali
24499da9a08SNetali        // if default lang is in ns: add default ns in front of org id
24599da9a08SNetali        if ($this->getConf('default_lang_in_ns')) {
246*ff327fe6SNetali            $org_id = $this->get_default_lang() . ':' . $org_id;
24799da9a08SNetali        }
24899da9a08SNetali
2490180404cSNetali        return array("ns" => getNS($org_id), "text" => rawWiki($org_id));
2503832d0abSNetali    }
2513832d0abSNetali
2523c636ad3SNetali    private function check_do_translation($allow_existing = false): bool {
2533832d0abSNetali        global $INFO;
2543832d0abSNetali        global $ID;
2553832d0abSNetali
2563c636ad3SNetali        // only translate if the current page does not exist
2573c636ad3SNetali        if ($INFO['exists'] and !$allow_existing) return false;
2583c636ad3SNetali
2593c636ad3SNetali        // permission check
2603c636ad3SNetali        $perm = auth_quickaclcheck($ID);
2613c636ad3SNetali        if (($INFO['exists'] and $perm < AUTH_EDIT) or (!$INFO['exists'] and $perm < AUTH_CREATE)) return false;
2623c636ad3SNetali
2633832d0abSNetali        // skip blacklisted namespaces and pages
2643832d0abSNetali        if ($this->getConf('blacklist_regex')) {
2653832d0abSNetali            if (preg_match('/' . $this->getConf('blacklist_regex') . '/', $ID) === 1) return false;
2663832d0abSNetali        }
2673832d0abSNetali
2683832d0abSNetali        $split_id = explode(':', $ID);
2693832d0abSNetali        $lang_ns = array_shift($split_id);
2703832d0abSNetali        // only translate if the current page is in a language namespace
2713832d0abSNetali        if (!array_key_exists($lang_ns, $this->langs)) return false;
2723832d0abSNetali
2733832d0abSNetali        $org_id = implode(':', $split_id);
27499da9a08SNetali
27599da9a08SNetali        // if default lang is in ns: add default ns in front of org id
27699da9a08SNetali        if ($this->getConf('default_lang_in_ns')) {
277*ff327fe6SNetali            $org_id = $this->get_default_lang() . ':' . $org_id;
27899da9a08SNetali        }
27999da9a08SNetali
2803832d0abSNetali        // check if the original page exists
2813832d0abSNetali        if (!page_exists($org_id)) return false;
2823832d0abSNetali
2833832d0abSNetali        return true;
2843832d0abSNetali    }
2853832d0abSNetali
286153e4498SNetali    private function check_do_push_translate(): bool {
287153e4498SNetali        global $ID;
28899da9a08SNetali        global $INFO;
28999da9a08SNetali
29099da9a08SNetali        if (!$INFO['exists']) return false;
29199da9a08SNetali
29299da9a08SNetali        // if default language is in namespace: only allow push translation from that namespace
29399da9a08SNetali        if($this->getConf('default_lang_in_ns')) {
29499da9a08SNetali            $split_id = explode(':', $ID);
29599da9a08SNetali            $lang_ns = array_shift($split_id);
29699da9a08SNetali
297*ff327fe6SNetali            if ($lang_ns !== $this->get_default_lang()) return false;
29899da9a08SNetali        }
299153e4498SNetali
300153e4498SNetali        $push_langs = $this->get_push_langs();
301153e4498SNetali        // push_langs empty --> push_translate disabled --> abort
302153e4498SNetali        if (empty($push_langs)) return false;
303153e4498SNetali
304153e4498SNetali        // skip blacklisted namespaces and pages
305153e4498SNetali        if ($this->getConf('blacklist_regex')) {
306153e4498SNetali            // blacklist regex match --> abort
307153e4498SNetali            if (preg_match('/' . $this->getConf('blacklist_regex') . '/', $ID) === 1) return false;
308153e4498SNetali        }
309153e4498SNetali
310153e4498SNetali        return true;
311153e4498SNetali    }
312153e4498SNetali
3130180404cSNetali    private function deepl_translate($text, $target_lang, $org_ns): string {
314153e4498SNetali        if (!trim($this->getConf('api_key'))) return '';
3153832d0abSNetali
3160180404cSNetali        $text = $this->patch_links($text, $target_lang, $org_ns);
3170180404cSNetali
3183832d0abSNetali        $text = $this->insert_ignore_tags($text);
3193832d0abSNetali
3203832d0abSNetali        $data = [
3213832d0abSNetali            'auth_key' => $this->getConf('api_key'),
3220180404cSNetali            'target_lang' => $this->langs[$target_lang],
3233832d0abSNetali            'tag_handling' => 'xml',
3240180404cSNetali            'ignore_tags' => 'ignore',
3253832d0abSNetali            'text' => $text
3263832d0abSNetali        ];
3273832d0abSNetali
3283832d0abSNetali        if ($this->getConf('api') == 'free') {
32981931e50SNetali            $url = 'https://api-free.deepl.com/v2/translate';
3303832d0abSNetali        } else {
33181931e50SNetali            $url = 'https://api.deepl.com/v2/translate';
3323832d0abSNetali        }
3333832d0abSNetali
33481931e50SNetali        $http = new DokuHTTPClient();
33581931e50SNetali        $raw_response = $http->post($url, $data);
3363832d0abSNetali
3375f8ab21dSNetali        if ($http->status >= 400) {
3385f8ab21dSNetali            // add error messages
3395f8ab21dSNetali            switch ($http->status) {
3405f8ab21dSNetali                case 403:
3415f8ab21dSNetali                    msg($this->getLang('msg_translation_fail_bad_key'), -1);
3425f8ab21dSNetali                    break;
3435f8ab21dSNetali                case 456:
3445f8ab21dSNetali                    msg($this->getLang('msg_translation_fail_quota_exceeded'), -1);
3455f8ab21dSNetali                    break;
3465f8ab21dSNetali                default:
3475f8ab21dSNetali                    msg($this->getLang('msg_translation_fail'), -1);
3485f8ab21dSNetali                    break;
3495f8ab21dSNetali            }
3505f8ab21dSNetali
3513832d0abSNetali            // if any error occurred return an empty string
3525f8ab21dSNetali            return '';
3535f8ab21dSNetali        }
3543832d0abSNetali
3553832d0abSNetali        $json_response = json_decode($raw_response, true);
3563832d0abSNetali        $translated_text = $json_response['translations'][0]['text'];
3573832d0abSNetali
3583832d0abSNetali        $translated_text = $this->remove_ignore_tags($translated_text);
3593832d0abSNetali
3603832d0abSNetali        return $translated_text;
3613832d0abSNetali    }
3623832d0abSNetali
363153e4498SNetali    private function get_push_langs(): array {
364153e4498SNetali        $push_langs = trim($this->getConf('push_langs'));
365153e4498SNetali
366153e4498SNetali        if ($push_langs === '') return array();
367153e4498SNetali
368153e4498SNetali        return explode(' ', $push_langs);
369153e4498SNetali    }
370153e4498SNetali
3710180404cSNetali    private function patch_links($text, $target_lang, $ns): string {
3720180404cSNetali        /*
3730180404cSNetali         * 1. Find links in [[ aa:bb ]] or [[ aa:bb | cc ]]
3740180404cSNetali         * 2. Extract aa:bb
3750180404cSNetali         * 3. Check if lang:aa:bb exists
3760180404cSNetali         * 3.1. --> Yes --> replace
3770180404cSNetali         * 3.2. --> No --> leave it as it is
3780180404cSNetali         */
3793832d0abSNetali
3800180404cSNetali
3810180404cSNetali        /*
3820180404cSNetali         * LINKS
3830180404cSNetali         */
3840180404cSNetali
3850180404cSNetali        preg_match_all('/\[\[([\s\S]*?)(\|([\s\S]*?))?]]/', $text, $matches, PREG_SET_ORDER);
3860180404cSNetali
3870180404cSNetali        foreach ($matches as $match) {
3880180404cSNetali
3890180404cSNetali            if (strpos($match[1], '://') !== false) {
3900180404cSNetali                // external link --> skip
3910180404cSNetali                continue;
3920180404cSNetali            }
3930180404cSNetali
3940180404cSNetali            $resolved_id = $match[1];
3950180404cSNetali
3960180404cSNetali            resolve_pageid($ns, $resolved_id, $exists);
3970180404cSNetali
3980180404cSNetali            if (!$exists) {
3990180404cSNetali                // redlink --> skip
4000180404cSNetali                continue;
4010180404cSNetali            }
4020180404cSNetali
4030180404cSNetali            $lang_id = $target_lang . ':' . $resolved_id;
4040180404cSNetali
4050180404cSNetali            if (!page_exists($lang_id)) {
4060180404cSNetali                // Page in target lang does not exist --> skip
4070180404cSNetali                continue;
4080180404cSNetali            }
4090180404cSNetali
4100180404cSNetali            $new_link = '[[' . $lang_id . $match[2] . ']]';
4110180404cSNetali
4120180404cSNetali            $text = str_replace($match[0], $new_link, $text);
4130180404cSNetali
4140180404cSNetali        }
4150180404cSNetali
4160180404cSNetali        /*
4170180404cSNetali         * MEDIA
4180180404cSNetali         */
4190180404cSNetali
4200180404cSNetali        preg_match_all('/\{\{([\s\S]*?)(\?[\s\S]*?)?(\|([\s\S]*?))?}}/', $text, $matches, PREG_SET_ORDER);
4210180404cSNetali
4220180404cSNetali        foreach ($matches as $match) {
4230180404cSNetali
4240180404cSNetali            if (strpos($match[1], '://') !== false) {
4250180404cSNetali                // external image --> skip
4260180404cSNetali                continue;
4270180404cSNetali            }
4280180404cSNetali
4290180404cSNetali            $resolved_id = $match[1];
4300180404cSNetali
4310180404cSNetali            resolve_mediaid($ns, $resolved_id, $exists);
4320180404cSNetali
4330180404cSNetali            if (!$exists) {
4340180404cSNetali                // redlink --> skip
4350180404cSNetali                continue;
4360180404cSNetali            }
4370180404cSNetali
4380180404cSNetali            $lang_id = $target_lang . ':' . $resolved_id;
4390180404cSNetali
4400180404cSNetali            $lang_id_fn = mediaFN($lang_id);
4410180404cSNetali
4420180404cSNetali            if (!file_exists($lang_id_fn)) {
4430180404cSNetali                // media in target lang does not exist --> skip
4440180404cSNetali                continue;
4450180404cSNetali            }
4460180404cSNetali
4470180404cSNetali            $new_link = '{{' . $lang_id . $match[2] . $match[3] . '}}';
4480180404cSNetali
4490180404cSNetali            $text = str_replace($match[0], $new_link, $text);
4500180404cSNetali
4510180404cSNetali        }
4520180404cSNetali
4530180404cSNetali        return $text;
4540180404cSNetali    }
4550180404cSNetali
4560180404cSNetali    private function insert_ignore_tags($text): string {
4570180404cSNetali        // ignore every other xml-like tags (the tags themselves, not their content), otherwise deepl would break the formatting
4580180404cSNetali        $text = preg_replace('/<[\s\S]+?>/', '<ignore>${0}</ignore>', $text);
4590180404cSNetali
4601cd781c4SNetali        // prevent deepl from breaking headings
4611cd781c4SNetali        $text = preg_replace('/={1,6}/', '<ignore>${0}</ignore>', $text);
4621cd781c4SNetali
4630180404cSNetali        // fix for the template plugin
4640180404cSNetali        $text = preg_replace('/\{\{template>[\s\S]*?}}/', '<ignore>${0}</ignore>', $text);
4650180404cSNetali
4660180404cSNetali        // ignore link/media ids but translate the text (if existing)
4670180404cSNetali        $text = preg_replace('/\[\[([\s\S]*?)((\|)([\s\S]*?))?]]/', '<ignore>[[${1}${3}</ignore>${4}<ignore>]]</ignore>', $text);
4680180404cSNetali        $text = preg_replace('/\{\{([\s\S]*?)(\?[\s\S]*?)?((\|)([\s\S]*?))?}}/', '<ignore>{{${1}${2}${4}</ignore>${5}<ignore>}}</ignore>', $text);
4690180404cSNetali
4700180404cSNetali        // prevent deepl from doing strange things with dokuwiki syntax
4710180404cSNetali        $text = str_replace("''", "<ignore>''</ignore>", $text);
4720180404cSNetali        $text = str_replace("\\\\", "<ignore>\\\\</ignore>", $text);
4730180404cSNetali
4740180404cSNetali        // ignore code tags
4750180404cSNetali        $text = preg_replace('/(<php[\s\S]*?>[\s\S]*?<\/php>)/', '<ignore>${1}</ignore>', $text);
476057940f7SNetali        $text = preg_replace('/(<file[\s\S]*?>[\s\S]*?<\/file>)/', '<ignore>${1}</ignore>', $text);
477057940f7SNetali        $text = preg_replace('/(<code[\s\S]*?>[\s\S]*?<\/code>)/', '<ignore>${1}</ignore>', $text);
478057940f7SNetali
4790180404cSNetali        // ignore the expressions from the ignore list
4803832d0abSNetali        $ignored_expressions = explode(':', $this->getConf('ignored_expressions'));
4813832d0abSNetali
4823832d0abSNetali        foreach ($ignored_expressions as $expression) {
4833832d0abSNetali            $text = str_replace($expression, '<ignore>' . $expression . '</ignore>', $text);
4843832d0abSNetali        }
4853832d0abSNetali
4863832d0abSNetali        return $text;
4873832d0abSNetali    }
4883832d0abSNetali
4893832d0abSNetali    private function remove_ignore_tags($text): string {
4903832d0abSNetali        $ignored_expressions = explode(':', $this->getConf('ignored_expressions'));
4913832d0abSNetali
4923832d0abSNetali        foreach ($ignored_expressions as $expression) {
4933832d0abSNetali            $text = str_replace('<ignore>' . $expression . '</ignore>', $expression, $text);
4943832d0abSNetali        }
4953832d0abSNetali
4960180404cSNetali        $text = preg_replace('/<ignore>\[\[([\s\S]*?)(\|)?(<\/ignore>)([\s\S]*?)?<ignore>]]<\/ignore>/', '[[${1}${2}${4}]]', $text);
4970180404cSNetali        $text = preg_replace('/<ignore>\{\{([\s\S]*?)(\|)?(<\/ignore>)([\s\S]*?)?<ignore>}}<\/ignore>/', '{{${1}${2}${4}}}', $text);
4980180404cSNetali
4990180404cSNetali        $text = str_replace("<ignore>''</ignore>", "''", $text);
5000180404cSNetali        $text = str_replace("<ignore>\\\\</ignore>", "\\\\", $text);
5010180404cSNetali
5020180404cSNetali        $text = preg_replace('/<ignore>(<php[\s\S]*?>[\s\S]*?<\/php>)<\/ignore>/', '${1}', $text);
5030180404cSNetali        $text = preg_replace('/<ignore>(<file[\s\S]*?>[\s\S]*?<\/file>)<\/ignore>/', '${1}', $text);
5040180404cSNetali        $text = preg_replace('/<ignore>(<code[\s\S]*?>[\s\S]*?<\/code>)<\/ignore>/', '${1}', $text);
5050180404cSNetali
5060180404cSNetali        // fix for the template plugin
5070180404cSNetali        $text = preg_replace('/<ignore>(\{\{template>[\s\S]*?}})<\/ignore>/', '${1}', $text);
5080180404cSNetali
5091cd781c4SNetali        // prevent deepl from breaking headings
5101cd781c4SNetali        $text = preg_replace('/<ignore>(={1,6})<\/ignore>/','${1}', $text);
5111cd781c4SNetali
5121cd781c4SNetali        // ignore every other xml-like tags (the tags themselves, not their content), otherwise deepl would break the formatting
5131cd781c4SNetali        $text = preg_replace('/<ignore>(<[\s\S]+?>)<\/ignore>/', '${1}', $text);
5141cd781c4SNetali
5150180404cSNetali        // restore < and > for example from arrows (-->) in wikitext
5160180404cSNetali        $text = str_replace('&gt;', '>', $text);
5170180404cSNetali        $text = str_replace('&lt;', '<', $text);
5180180404cSNetali
5193832d0abSNetali        return $text;
5203832d0abSNetali    }
5213832d0abSNetali}
5223832d0abSNetali
523