xref: /plugin/deeplautotranslate/action.php (revision 8311ddaa338b01befbebadff30e27b61fbdab5b7)
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
16b33135efSNetali    private $langs = array(
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',
42*8311ddaaSnetali        'uk' => 'UK',
433832d0abSNetali        'zh' => 'ZH'
44b33135efSNetali    );
453832d0abSNetali
463832d0abSNetali    /**
473832d0abSNetali     * Register its handlers with the DokuWiki's event controller
483832d0abSNetali     */
493832d0abSNetali    public function register(Doku_Event_Handler $controller) {
50153e4498SNetali        $controller->register_hook('ACTION_ACT_PREPROCESS','BEFORE', $this, 'preprocess');
51b33135efSNetali        $controller->register_hook('COMMON_PAGETPL_LOAD','AFTER', $this, 'pagetpl_load');
52b33135efSNetali        $controller->register_hook('COMMON_WIKIPAGE_SAVE','AFTER', $this, 'update_glossary');
533c636ad3SNetali        $controller->register_hook('MENU_ITEMS_ASSEMBLY', 'AFTER', $this, 'add_menu_button');
543c636ad3SNetali    }
553c636ad3SNetali
560180404cSNetali    public function add_menu_button(Doku_Event $event): void {
57153e4498SNetali        global $ID;
58bbb1fba9SNetali        global $ACT;
59bbb1fba9SNetali
60bbb1fba9SNetali        if ($ACT != 'show') return;
61153e4498SNetali
623c636ad3SNetali        if ($event->data['view'] != 'page') return;
633c636ad3SNetali
643c636ad3SNetali        if (!$this->getConf('show_button')) return;
65153e4498SNetali
66b33135efSNetali        // no translations for the glossary namespace
67b33135efSNetali        if ($this->check_in_glossary_ns()) return;
68b33135efSNetali
69153e4498SNetali        $split_id = explode(':', $ID);
70153e4498SNetali        $lang_ns = array_shift($split_id);
71153e4498SNetali        // check if we are in a language namespace
72153e4498SNetali        if (array_key_exists($lang_ns, $this->langs)) {
73ff327fe6SNetali            if($this->getConf('default_lang_in_ns') and $lang_ns === $this->get_default_lang()) {
7499da9a08SNetali                // if the default lang is in a namespace and we are in that namespace --> check for push translation
7599da9a08SNetali                if (!$this->check_do_push_translate()) return;
7699da9a08SNetali            } else {
77153e4498SNetali                // in language namespace --> check if we should translate
783c636ad3SNetali                if (!$this->check_do_translation(true)) return;
7999da9a08SNetali            }
80153e4498SNetali        } else {
8199da9a08SNetali            // do not show the button if we are not in a language namespace and the default language is in a namespace
8299da9a08SNetali            if($this->getConf('default_lang_in_ns')) return;
83b33135efSNetali            // not in language namespace and default language is not in a namespace --> check if we should show the push translate button
84153e4498SNetali            if (!$this->check_do_push_translate()) return;
85153e4498SNetali        }
863c636ad3SNetali
873c636ad3SNetali        array_splice($event->data['items'], -1, 0, [new MenuItem()]);
883832d0abSNetali    }
893832d0abSNetali
90153e4498SNetali    public function preprocess(Doku_Event $event, $param): void {
913832d0abSNetali        global $ID;
923c636ad3SNetali
933c636ad3SNetali        // check if action is show or translate
943c636ad3SNetali        if ($event->data != 'show' and $event->data != 'translate') return;
953c636ad3SNetali
96b33135efSNetali        // redirect to glossary ns start if glossary ns is called
97b33135efSNetali        if ($this->check_in_glossary_ns() and $event->data == 'show' and $ID == $this->get_glossary_ns()) {
98b33135efSNetali            send_redirect(wl($this->get_glossary_ns() . ':start'));
99b33135efSNetali        }
100b33135efSNetali
101153e4498SNetali        $split_id = explode(':', $ID);
102153e4498SNetali        $lang_ns = array_shift($split_id);
103153e4498SNetali        // check if we are in a language namespace
104153e4498SNetali        if (array_key_exists($lang_ns, $this->langs)) {
105ff327fe6SNetali            if($this->getConf('default_lang_in_ns') and $lang_ns === $this->get_default_lang()) {
10699da9a08SNetali                // if the default lang is in a namespace and we are in that namespace --> push translate
10799da9a08SNetali                $this->push_translate($event);
10899da9a08SNetali            } else {
10999da9a08SNetali                // in language namespace --> autotrans direct
110153e4498SNetali                $this->autotrans_direct($event);
11199da9a08SNetali            }
112153e4498SNetali        } else {
113153e4498SNetali            // not in language namespace --> push translate
114153e4498SNetali            $this->push_translate($event);
115153e4498SNetali        }
116153e4498SNetali    }
117153e4498SNetali
118b33135efSNetali    public function pagetpl_load(Doku_Event $event, $param): void {
119b33135efSNetali        // handle glossary namespace init when we are in it
120b33135efSNetali        if ($this->check_in_glossary_ns()) {
121b33135efSNetali            $this->handle_glossary_init($event);
122b33135efSNetali            return;
123b33135efSNetali        }
124b33135efSNetali
125b33135efSNetali        $this->autotrans_editor($event);
126b33135efSNetali    }
127b33135efSNetali
128b33135efSNetali    public function update_glossary(Doku_Event $event, $param): void {
129b33135efSNetali        global $ID;
130b33135efSNetali        // this also checks if the glossary feature is enabled
131b33135efSNetali        if (!$this->check_in_glossary_ns()) return;
132b33135efSNetali
133b33135efSNetali        $glossary_ns = $this->get_glossary_ns();
134b33135efSNetali
135b33135efSNetali        // check if we are in a glossary definition
136b33135efSNetali        if(preg_match('/^' . $glossary_ns . ':(\w{2})_(\w{2})$/', $ID, $id_match)) {
137b33135efSNetali            $old_glossary_id = $this->get_glossary_id($id_match[1], $id_match[2]);
138b33135efSNetali            if ($event->data['changeType'] == DOKU_CHANGE_TYPE_DELETE) {
139b33135efSNetali                // page deleted --> delete glossary
140b33135efSNetali                if ($old_glossary_id) {
141b33135efSNetali                    $result = $this->delete_glossary($old_glossary_id);
142b33135efSNetali                    if ($result) {
143b33135efSNetali                        msg($this->getLang('msg_glossary_delete_success'), 1);
144b33135efSNetali                        $this->unset_glossary_id($id_match[1], $id_match[2]);
145b33135efSNetali                    }
146b33135efSNetali                }
147b33135efSNetali                return;
148b33135efSNetali            }
149b33135efSNetali
150b33135efSNetali            $entries = '';
151b33135efSNetali
152b33135efSNetali            // grep entries from definition table
153b33135efSNetali            preg_match_all('/[ \t]*\|(.*?)\|(.*?)\|/', $event->data['newContent'], $matches, PREG_SET_ORDER);
154b33135efSNetali            foreach ($matches as $match) {
155b33135efSNetali                $src = trim($match[1]);
156b33135efSNetali                $target = trim($match[2]);
157b33135efSNetali                if ($src == '' or $target == '') {
158b33135efSNetali                    msg($this->getLang('msg_glossary_empty_key'), -1);
159b33135efSNetali                    return;
160b33135efSNetali                }
161b33135efSNetali                $entries .=  $src . "\t" . $target . "\n";
162b33135efSNetali            }
163b33135efSNetali
164b33135efSNetali            if (empty($matches)) {
165b33135efSNetali                // no matches --> delete glossary
166b33135efSNetali                if ($old_glossary_id) {
167b33135efSNetali                    $result = $this->delete_glossary($old_glossary_id);
168b33135efSNetali                    if ($result) {
169b33135efSNetali                        msg($this->getLang('msg_glossary_delete_success'), 1);
170b33135efSNetali                        $this->unset_glossary_id($id_match[1], $id_match[2]);
171b33135efSNetali                    }
172b33135efSNetali                }
173b33135efSNetali                return;
174b33135efSNetali            }
175b33135efSNetali
176b33135efSNetali            $new_glossary_id = $this->create_glossary($id_match[1], $id_match[2], $entries);
177b33135efSNetali
178b33135efSNetali            if ($new_glossary_id) {
179b33135efSNetali                msg($this->getLang('msg_glossary_create_success'), 1);
180b33135efSNetali                $this->set_glossary_id($id_match[1], $id_match[2], $new_glossary_id);
181b33135efSNetali            } else {
182b33135efSNetali                return;
183b33135efSNetali            }
184b33135efSNetali
185b33135efSNetali            if ($old_glossary_id) $this->delete_glossary($old_glossary_id);
186b33135efSNetali        }
187b33135efSNetali    }
188b33135efSNetali
189153e4498SNetali    private function autotrans_direct(Doku_Event $event): void {
190153e4498SNetali        global $ID;
191153e4498SNetali
1923c636ad3SNetali        // abort if action is translate and the translate button is disabled
1933c636ad3SNetali        if ($event->data == 'translate' and !$this->getConf('show_button')) return;
1943c636ad3SNetali
1953c636ad3SNetali        // do nothing on show action when mode is not direct
1963c636ad3SNetali        if ($event->data == 'show' and $this->get_mode() != 'direct') return;
1973c636ad3SNetali
1983c636ad3SNetali        // allow translation of existing pages is we are in the translate action
1993c636ad3SNetali        $allow_existing = ($event->data == 'translate');
2003c636ad3SNetali
2013c636ad3SNetali        // reset action to show
2023c636ad3SNetali        $event->data = 'show';
2033c636ad3SNetali
204153e4498SNetali        if (!$this->check_do_translation($allow_existing)) {
205153e4498SNetali            return;
206153e4498SNetali        }
2073832d0abSNetali
2080180404cSNetali        $org_page_info = $this->get_org_page_info();
2090180404cSNetali        $translated_text = $this->deepl_translate($org_page_info["text"], $this->get_target_lang(), $org_page_info["ns"]);
2103832d0abSNetali
211153e4498SNetali        if ($translated_text === '') {
212153e4498SNetali            return;
213153e4498SNetali        }
2143832d0abSNetali
2153832d0abSNetali        saveWikiText($ID, $translated_text, 'Automatic translation');
2163832d0abSNetali
217153e4498SNetali        msg($this->getLang('msg_translation_success'), 1);
218153e4498SNetali
2193c636ad3SNetali        // reload the page after translation
2203c636ad3SNetali        send_redirect(wl($ID));
2213832d0abSNetali    }
2223832d0abSNetali
223b33135efSNetali    private function autotrans_editor(Doku_Event $event): void {
2243832d0abSNetali        if ($this->get_mode() != 'editor') return;
2253832d0abSNetali
2263832d0abSNetali        if (!$this->check_do_translation()) return;
2273832d0abSNetali
2280180404cSNetali        $org_page_info = $this->get_org_page_info();
2293832d0abSNetali
2300180404cSNetali        $event->data['tpl'] = $this->deepl_translate($org_page_info["text"], $this->get_target_lang(), $org_page_info["ns"]);
2313832d0abSNetali    }
2323832d0abSNetali
233153e4498SNetali    private function push_translate(Doku_Event $event): void {
234153e4498SNetali        global $ID;
235153e4498SNetali
236153e4498SNetali        // check if action is translate
237153e4498SNetali        if ($event->data != 'translate') return;
238153e4498SNetali
239153e4498SNetali        // check if button is enabled
240153e4498SNetali        if (!$this->getConf('show_button')) {
241153e4498SNetali            send_redirect(wl($ID));
242153e4498SNetali            return;
243153e4498SNetali        }
244153e4498SNetali
245153e4498SNetali        if (!$this->check_do_push_translate()) {
246153e4498SNetali            send_redirect(wl($ID));
247153e4498SNetali            return;
248153e4498SNetali        }
249153e4498SNetali
250153e4498SNetali        // push translate
251153e4498SNetali        $push_langs = $this->get_push_langs();
252153e4498SNetali        $org_page_text = rawWiki($ID);
253153e4498SNetali        foreach ($push_langs as $lang) {
254153e4498SNetali            // skip invalid languages
255153e4498SNetali            if (!array_key_exists($lang, $this->langs)) {
256153e4498SNetali                msg($this->getLang('msg_translation_fail_invalid_lang') . $lang, -1);
257153e4498SNetali                continue;
258153e4498SNetali            }
259153e4498SNetali
26099da9a08SNetali            if ($this->getConf('default_lang_in_ns')) {
26199da9a08SNetali                // if default lang is in ns: replace language namespace in ID
26299da9a08SNetali                $split_id = explode(':', $ID);
26399da9a08SNetali                array_shift($split_id);
26499da9a08SNetali                $lang_id = implode(':', $split_id);
26599da9a08SNetali                $lang_id = $lang . ':' . $lang_id;
26699da9a08SNetali            } else {
26799da9a08SNetali                // if default lang is not in ns: add language namespace to ID
268153e4498SNetali                $lang_id = $lang . ':' . $ID;
26999da9a08SNetali            }
270153e4498SNetali
271153e4498SNetali            // check permissions
272a3a51507SNetali            $perm = auth_quickaclcheck($lang_id);
273153e4498SNetali            $exists = page_exists($lang_id);
274153e4498SNetali            if (($exists and $perm < AUTH_EDIT) or (!$exists and $perm < AUTH_CREATE)) {
275153e4498SNetali                msg($this->getLang('msg_translation_fail_no_permissions') . $lang_id, -1);
276153e4498SNetali                continue;
277153e4498SNetali            }
278153e4498SNetali
2790180404cSNetali            $translated_text = $this->deepl_translate($org_page_text, $lang, getNS($ID));
280153e4498SNetali            saveWikiText($lang_id, $translated_text, 'Automatic push translation');
281153e4498SNetali        }
282153e4498SNetali
283153e4498SNetali        msg($this->getLang('msg_translation_success'), 1);
284153e4498SNetali
285153e4498SNetali        // reload the page after translation to clear the action
286153e4498SNetali        send_redirect(wl($ID));
287153e4498SNetali    }
288153e4498SNetali
289b33135efSNetali    private function handle_glossary_init(Doku_Event $event): void {
290b33135efSNetali        global $ID;
291b33135efSNetali
292b33135efSNetali        $glossary_ns = $this->get_glossary_ns();
293b33135efSNetali
294b33135efSNetali        // create glossary landing page
295b33135efSNetali        if ($ID == $glossary_ns . ':start') {
296b33135efSNetali            $landing_page_text = '====== ' . $this->getLang('glossary_landing_heading') . ' ======' . "\n";
297b33135efSNetali            $landing_page_text .= $this->getLang('glossary_landing_info_msg') . "\n";
298b33135efSNetali
299b33135efSNetali            $src_lang = substr($this->get_default_lang(), 0, 2);
300b33135efSNetali
301b33135efSNetali            $available_glossaries = $this->get_available_glossaries();
302b33135efSNetali            foreach ($available_glossaries as $glossary) {
303b33135efSNetali                if ($glossary['source_lang'] != $src_lang) continue;
304b33135efSNetali                // generate links to the available glossary pages
305b33135efSNetali                $landing_page_text .= '  * [[.:' . $glossary['source_lang'] . '_' . $glossary['target_lang'] . '|' . strtoupper($glossary['source_lang']) . ' -> ' . strtoupper($glossary['target_lang']) . ']]' . "\n";
306b33135efSNetali            }
307b33135efSNetali            $event->data['tpl'] = $landing_page_text;
308b33135efSNetali            return;
309b33135efSNetali        }
310b33135efSNetali
311b33135efSNetali        if (preg_match('/^' . $glossary_ns . ':(\w{2})_(\w{2})$/', $ID, $match)) {
312b33135efSNetali            // check if glossaries are supported for this language pair
313b33135efSNetali            if (!$this->check_glossary_supported($match[1], $match[2])) {
314b33135efSNetali                msg($this->getLang('msg_glossary_unsupported'), -1);
315b33135efSNetali                return;
316b33135efSNetali            }
317b33135efSNetali
318b33135efSNetali            $page_text = '====== ' . $this->getLang('glossary_definition_heading') . ': ' . strtoupper($match[1]) . ' -> ' . strtoupper($match[2]) . ' ======' . "\n";
319b33135efSNetali            $page_text .= $this->getLang('glossary_definition_help') . "\n\n";
320b33135efSNetali            $page_text .= '^ ' . strtoupper($match[1]) . ' ^ ' . strtoupper($match[2]) . ' ^' . "\n";
321b33135efSNetali
322b33135efSNetali            $event->data['tpl'] = $page_text;
323b33135efSNetali            return;
324b33135efSNetali        }
325b33135efSNetali    }
326b33135efSNetali
327b33135efSNetali    private function get_glossary_ns(): string {
328b33135efSNetali        return trim(strtolower($this->getConf('glossary_ns')));
329b33135efSNetali    }
330b33135efSNetali
3313832d0abSNetali    private function get_mode(): string {
3323832d0abSNetali        global $ID;
3333832d0abSNetali        if ($this->getConf('editor_regex')) {
3343832d0abSNetali            if (preg_match('/' . $this->getConf('editor_regex') . '/', $ID) === 1) return 'editor';
3353832d0abSNetali        }
3363832d0abSNetali        if ($this->getConf('direct_regex')) {
3373832d0abSNetali            if (preg_match('/' . $this->getConf('direct_regex') . '/', $ID) === 1) return 'direct';
3383832d0abSNetali        }
3393832d0abSNetali        return $this->getConf('mode');
3403832d0abSNetali    }
3413832d0abSNetali
3423832d0abSNetali    private function get_target_lang(): string {
3433832d0abSNetali        global $ID;
3443832d0abSNetali        $split_id = explode(':', $ID);
3453832d0abSNetali        return array_shift($split_id);
3463832d0abSNetali    }
3473832d0abSNetali
348ff327fe6SNetali    private function get_default_lang(): string {
349ff327fe6SNetali        global $conf;
350ff327fe6SNetali
351ff327fe6SNetali        if (empty($conf['lang_before_translation'])) {
352ff327fe6SNetali            $default_lang = $conf['lang'];
353ff327fe6SNetali        } else {
354ff327fe6SNetali            $default_lang = $conf['lang_before_translation'];
355ff327fe6SNetali        }
356ff327fe6SNetali
357ff327fe6SNetali        return $default_lang;
358ff327fe6SNetali    }
359ff327fe6SNetali
3600180404cSNetali    private function get_org_page_info(): array {
3613832d0abSNetali        global $ID;
3623832d0abSNetali
3633832d0abSNetali        $split_id = explode(':', $ID);
3643832d0abSNetali        array_shift($split_id);
3653832d0abSNetali        $org_id = implode(':', $split_id);
3663832d0abSNetali
36799da9a08SNetali        // if default lang is in ns: add default ns in front of org id
36899da9a08SNetali        if ($this->getConf('default_lang_in_ns')) {
369ff327fe6SNetali            $org_id = $this->get_default_lang() . ':' . $org_id;
37099da9a08SNetali        }
37199da9a08SNetali
3720180404cSNetali        return array("ns" => getNS($org_id), "text" => rawWiki($org_id));
3733832d0abSNetali    }
3743832d0abSNetali
375b33135efSNetali    private function get_available_glossaries(): array {
376b33135efSNetali        if (!trim($this->getConf('api_key'))) {
377b33135efSNetali            msg($this->getLang('msg_bad_key'), -1);
378b33135efSNetali            return array();
379b33135efSNetali        }
380b33135efSNetali
381b33135efSNetali        if ($this->getConf('api') == 'free') {
382b33135efSNetali            $url = 'https://api-free.deepl.com/v2/glossary-language-pairs';
383b33135efSNetali        } else {
384b33135efSNetali            $url = 'https://api.deepl.com/v2/glossary-language-pairs';
385b33135efSNetali        }
386b33135efSNetali
387b33135efSNetali        $http = new DokuHTTPClient();
388b33135efSNetali
389b33135efSNetali        $http->headers = array('Authorization' => 'DeepL-Auth-Key ' . $this->getConf('api_key'));
390b33135efSNetali
391b33135efSNetali        $raw_response = $http->get($url);
392b33135efSNetali
393b33135efSNetali        if ($http->status >= 400) {
394b33135efSNetali            // add error messages
395b33135efSNetali            switch ($http->status) {
396b33135efSNetali                case 403:
397b33135efSNetali                    msg($this->getLang('msg_bad_key'), -1);
398b33135efSNetali                    break;
399b33135efSNetali                default:
400b33135efSNetali                    msg($this->getLang('msg_glossary_fetch_fail'), -1);
401b33135efSNetali                    break;
402b33135efSNetali            }
403b33135efSNetali
404b33135efSNetali            // if any error occurred return an empty array
405b33135efSNetali            return array();
406b33135efSNetali        }
407b33135efSNetali
408b33135efSNetali        $json_response = json_decode($raw_response, true);
409b33135efSNetali
410b33135efSNetali        return $json_response['supported_languages'];
411b33135efSNetali    }
412b33135efSNetali
413b33135efSNetali    private function get_glossary_id($src, $target): string {
414b33135efSNetali        if (!file_exists(DOKU_CONF . 'deepl-glossaries.json')) return '';
415b33135efSNetali
416b33135efSNetali        $key = $src . "_" . $target;
417b33135efSNetali
418b33135efSNetali        $raw_json = file_get_contents(DOKU_CONF . 'deepl-glossaries.json');
419b33135efSNetali        $content = json_decode($raw_json, true);
420b33135efSNetali
421b33135efSNetali        if (array_key_exists($key, $content)) {
422b33135efSNetali            return $content[$key];
423b33135efSNetali        } else {
424b33135efSNetali            return '';
425b33135efSNetali        }
426b33135efSNetali    }
427b33135efSNetali
428b33135efSNetali    private function set_glossary_id($src, $target, $glossary_id): void {
429b33135efSNetali        if (file_exists(DOKU_CONF . 'deepl-glossaries.json')) {
430b33135efSNetali            $raw_json = file_get_contents(DOKU_CONF . 'deepl-glossaries.json');
431b33135efSNetali            $content = json_decode($raw_json, true);
432b33135efSNetali        } else {
433b33135efSNetali            $content = array();
434b33135efSNetali        }
435b33135efSNetali
436b33135efSNetali        $key = $src . "_" . $target;
437b33135efSNetali
438b33135efSNetali        $content[$key] = $glossary_id;
439b33135efSNetali
440b33135efSNetali        $raw_json = json_encode($content);
441b33135efSNetali        file_put_contents(DOKU_CONF . 'deepl-glossaries.json', $raw_json);
442b33135efSNetali    }
443b33135efSNetali
444b33135efSNetali    private function unset_glossary_id($src, $target): void {
445b33135efSNetali        if (file_exists(DOKU_CONF . 'deepl-glossaries.json')) {
446b33135efSNetali            $raw_json = file_get_contents(DOKU_CONF . 'deepl-glossaries.json');
447b33135efSNetali            $content = json_decode($raw_json, true);
448b33135efSNetali        } else {
449b33135efSNetali            return;
450b33135efSNetali        }
451b33135efSNetali
452b33135efSNetali        $key = $src . "_" . $target;
453b33135efSNetali
454b33135efSNetali        unset($content[$key]);
455b33135efSNetali
456b33135efSNetali        $raw_json = json_encode($content);
457b33135efSNetali        file_put_contents(DOKU_CONF . 'deepl-glossaries.json', $raw_json);
458b33135efSNetali    }
459b33135efSNetali
460b33135efSNetali    private function check_in_glossary_ns(): bool {
461b33135efSNetali        global $ID;
462b33135efSNetali
463b33135efSNetali        $glossary_ns = $this->get_glossary_ns();
464b33135efSNetali
465b33135efSNetali        // check if the glossary namespace is defined
466b33135efSNetali        if (!$glossary_ns) return false;
467b33135efSNetali
468b33135efSNetali        // check if we are in the glossary namespace
469b33135efSNetali        if (substr($ID, 0, strlen($glossary_ns)) == $glossary_ns) {
470b33135efSNetali            return true;
471b33135efSNetali        } else {
472b33135efSNetali            return false;
473b33135efSNetali        }
474b33135efSNetali    }
475b33135efSNetali
476b33135efSNetali    private function check_glossary_supported($src, $target): bool {
477b33135efSNetali        if(strlen($src) != 2 or strlen($target) != 2) return false;
478b33135efSNetali        $available_glossaries = $this->get_available_glossaries();
479b33135efSNetali        foreach ($available_glossaries as $glossary) {
480b33135efSNetali            if ($src == $glossary['source_lang'] and $target == $glossary['target_lang']) return true;
481b33135efSNetali        }
482b33135efSNetali        return false;
483b33135efSNetali    }
484b33135efSNetali
4853c636ad3SNetali    private function check_do_translation($allow_existing = false): bool {
4863832d0abSNetali        global $INFO;
4873832d0abSNetali        global $ID;
4883832d0abSNetali
4893c636ad3SNetali        // only translate if the current page does not exist
4903c636ad3SNetali        if ($INFO['exists'] and !$allow_existing) return false;
4913c636ad3SNetali
4923c636ad3SNetali        // permission check
4933c636ad3SNetali        $perm = auth_quickaclcheck($ID);
4943c636ad3SNetali        if (($INFO['exists'] and $perm < AUTH_EDIT) or (!$INFO['exists'] and $perm < AUTH_CREATE)) return false;
4953c636ad3SNetali
4963832d0abSNetali        // skip blacklisted namespaces and pages
4973832d0abSNetali        if ($this->getConf('blacklist_regex')) {
4983832d0abSNetali            if (preg_match('/' . $this->getConf('blacklist_regex') . '/', $ID) === 1) return false;
4993832d0abSNetali        }
5003832d0abSNetali
5013832d0abSNetali        $split_id = explode(':', $ID);
5023832d0abSNetali        $lang_ns = array_shift($split_id);
5033832d0abSNetali        // only translate if the current page is in a language namespace
5043832d0abSNetali        if (!array_key_exists($lang_ns, $this->langs)) return false;
5053832d0abSNetali
5063832d0abSNetali        $org_id = implode(':', $split_id);
50799da9a08SNetali
50899da9a08SNetali        // if default lang is in ns: add default ns in front of org id
50999da9a08SNetali        if ($this->getConf('default_lang_in_ns')) {
510ff327fe6SNetali            $org_id = $this->get_default_lang() . ':' . $org_id;
51199da9a08SNetali        }
51299da9a08SNetali
513b33135efSNetali        // no translations for the glossary namespace
514b33135efSNetali        $glossary_ns = $this->get_glossary_ns();
515b33135efSNetali        if ($glossary_ns and substr($org_id, 0, strlen($glossary_ns)) == $glossary_ns) return false;
516b33135efSNetali
5173832d0abSNetali        // check if the original page exists
5183832d0abSNetali        if (!page_exists($org_id)) return false;
5193832d0abSNetali
5203832d0abSNetali        return true;
5213832d0abSNetali    }
5223832d0abSNetali
523153e4498SNetali    private function check_do_push_translate(): bool {
524153e4498SNetali        global $ID;
52599da9a08SNetali        global $INFO;
52699da9a08SNetali
52799da9a08SNetali        if (!$INFO['exists']) return false;
52899da9a08SNetali
529a3a51507SNetali        // only allow push translation if the user can edit this page
530a3a51507SNetali        $perm = auth_quickaclcheck($ID);
531a3a51507SNetali        if ($perm < AUTH_EDIT) return false;
532a3a51507SNetali
53399da9a08SNetali        // if default language is in namespace: only allow push translation from that namespace
53499da9a08SNetali        if($this->getConf('default_lang_in_ns')) {
53599da9a08SNetali            $split_id = explode(':', $ID);
53699da9a08SNetali            $lang_ns = array_shift($split_id);
53799da9a08SNetali
538ff327fe6SNetali            if ($lang_ns !== $this->get_default_lang()) return false;
53999da9a08SNetali        }
540153e4498SNetali
541b33135efSNetali        // no translations for the glossary namespace
542b33135efSNetali        if ($this->check_in_glossary_ns()) return false;
543b33135efSNetali
544153e4498SNetali        $push_langs = $this->get_push_langs();
545153e4498SNetali        // push_langs empty --> push_translate disabled --> abort
546153e4498SNetali        if (empty($push_langs)) return false;
547153e4498SNetali
548153e4498SNetali        // skip blacklisted namespaces and pages
549153e4498SNetali        if ($this->getConf('blacklist_regex')) {
550153e4498SNetali            // blacklist regex match --> abort
551153e4498SNetali            if (preg_match('/' . $this->getConf('blacklist_regex') . '/', $ID) === 1) return false;
552153e4498SNetali        }
553153e4498SNetali
554153e4498SNetali        return true;
555153e4498SNetali    }
556153e4498SNetali
557b33135efSNetali    private function create_glossary($src, $target, $entries): string {
558b33135efSNetali        if (!trim($this->getConf('api_key'))) {
559b33135efSNetali            msg($this->getLang('msg_bad_key'), -1);
560b33135efSNetali            return '';
561b33135efSNetali        }
562b33135efSNetali
563b33135efSNetali        if ($this->getConf('api') == 'free') {
564b33135efSNetali            $url = 'https://api-free.deepl.com/v2/glossaries';
565b33135efSNetali        } else {
566b33135efSNetali            $url = 'https://api.deepl.com/v2/glossaries';
567b33135efSNetali        }
568b33135efSNetali
569b33135efSNetali        $data = array(
570b33135efSNetali            'name' => 'DokuWiki-Autotranslate-' . $src . '_' . $target,
571b33135efSNetali            'source_lang' => $src,
572b33135efSNetali            'target_lang' => $target,
573b33135efSNetali            'entries' => $entries,
574b33135efSNetali            'entries_format' => 'tsv'
575b33135efSNetali        );
576b33135efSNetali
577b33135efSNetali        $http = new DokuHTTPClient();
578b33135efSNetali
579b33135efSNetali        $http->headers = array('Authorization' => 'DeepL-Auth-Key ' . $this->getConf('api_key'));
580b33135efSNetali
581b33135efSNetali        $raw_response = $http->post($url, $data);
582b33135efSNetali
583b33135efSNetali        if ($http->status >= 400) {
584b33135efSNetali            // add error messages
585b33135efSNetali            switch ($http->status) {
586b33135efSNetali                case 403:
587b33135efSNetali                    msg($this->getLang('msg_bad_key'), -1);
588b33135efSNetali                    break;
589b33135efSNetali                case 400:
590b33135efSNetali                    msg($this->getLang('msg_glossary_content_invalid'), -1);
591b33135efSNetali                    break;
592b33135efSNetali                default:
593b33135efSNetali                    msg($this->getLang('msg_glossary_create_fail'), -1);
594b33135efSNetali                    break;
595b33135efSNetali            }
596b33135efSNetali
597b33135efSNetali            // if any error occurred return an empty string
598b33135efSNetali            return '';
599b33135efSNetali        }
600b33135efSNetali
601b33135efSNetali        $json_response = json_decode($raw_response, true);
602b33135efSNetali
603b33135efSNetali        return $json_response['glossary_id'];
604b33135efSNetali    }
605b33135efSNetali
606b33135efSNetali    private function delete_glossary($glossary_id): bool {
607b33135efSNetali        if (!trim($this->getConf('api_key'))) {
608b33135efSNetali            msg($this->getLang('msg_bad_key'), -1);
609b33135efSNetali            return false;
610b33135efSNetali        }
611b33135efSNetali
612b33135efSNetali        if ($this->getConf('api') == 'free') {
613b33135efSNetali            $url = 'https://api-free.deepl.com/v2/glossaries';
614b33135efSNetali        } else {
615b33135efSNetali            $url = 'https://api.deepl.com/v2/glossaries';
616b33135efSNetali        }
617b33135efSNetali
618b33135efSNetali        $url .= '/' . $glossary_id;
619b33135efSNetali
620b33135efSNetali        $http = new DokuHTTPClient();
621b33135efSNetali
622b33135efSNetali        $http->headers = array('Authorization' => 'DeepL-Auth-Key ' . $this->getConf('api_key'));
623b33135efSNetali
624b33135efSNetali        $http->sendRequest($url, '', 'DELETE');
625b33135efSNetali
626b33135efSNetali        if ($http->status >= 400) {
627b33135efSNetali            // add error messages
628b33135efSNetali            switch ($http->status) {
629b33135efSNetali                case 403:
630b33135efSNetali                    msg($this->getLang('msg_bad_key'), -1);
631b33135efSNetali                    break;
632b33135efSNetali                default:
633b33135efSNetali                    msg($this->getLang('msg_glossary_delete_fail'), -1);
634b33135efSNetali                    break;
635b33135efSNetali            }
636b33135efSNetali
637b33135efSNetali            // if any error occurred return false
638b33135efSNetali            return false;
639b33135efSNetali        }
640b33135efSNetali
641b33135efSNetali        return true;
642b33135efSNetali    }
643b33135efSNetali
6440180404cSNetali    private function deepl_translate($text, $target_lang, $org_ns): string {
645b33135efSNetali        if (!trim($this->getConf('api_key'))) {
646b33135efSNetali            msg($this->getLang('msg_translation_fail_bad_key'), -1);
647b33135efSNetali            return '';
648b33135efSNetali        }
6493832d0abSNetali
6500180404cSNetali        $text = $this->patch_links($text, $target_lang, $org_ns);
6510180404cSNetali
6523832d0abSNetali        $text = $this->insert_ignore_tags($text);
6533832d0abSNetali
654b33135efSNetali        $data = array(
655b33135efSNetali            'source_lang' => strtoupper(substr($this->get_default_lang(), 0, 2)), // cut of things like "-informal"
6560180404cSNetali            'target_lang' => $this->langs[$target_lang],
6573832d0abSNetali            'tag_handling' => 'xml',
6580180404cSNetali            'ignore_tags' => 'ignore',
6593832d0abSNetali            'text' => $text
660b33135efSNetali        );
661b33135efSNetali
662b33135efSNetali        // check if glossaries are enabled
663b33135efSNetali        if ($this->get_glossary_ns()) {
664b33135efSNetali            $src = substr($this->get_default_lang(), 0, 2);
665b33135efSNetali            $target = substr($target_lang, 0, 2);
666b33135efSNetali            $glossary_id = $this->get_glossary_id($src, $target);
667b33135efSNetali            if ($glossary_id) {
668b33135efSNetali                // use glossary if it is defined
669b33135efSNetali                $data['glossary_id'] = $glossary_id;
670b33135efSNetali            }
671b33135efSNetali        }
6723832d0abSNetali
6733832d0abSNetali        if ($this->getConf('api') == 'free') {
67481931e50SNetali            $url = 'https://api-free.deepl.com/v2/translate';
6753832d0abSNetali        } else {
67681931e50SNetali            $url = 'https://api.deepl.com/v2/translate';
6773832d0abSNetali        }
6783832d0abSNetali
67981931e50SNetali        $http = new DokuHTTPClient();
680b33135efSNetali
681b33135efSNetali        $http->headers = array('Authorization' => 'DeepL-Auth-Key ' . $this->getConf('api_key'));
682b33135efSNetali
68381931e50SNetali        $raw_response = $http->post($url, $data);
6843832d0abSNetali
6855f8ab21dSNetali        if ($http->status >= 400) {
6865f8ab21dSNetali            // add error messages
6875f8ab21dSNetali            switch ($http->status) {
6885f8ab21dSNetali                case 403:
6895f8ab21dSNetali                    msg($this->getLang('msg_translation_fail_bad_key'), -1);
6905f8ab21dSNetali                    break;
691b33135efSNetali                case 404:
692b33135efSNetali                    msg($this->getLang('msg_translation_fail_invalid_glossary'), -1);
693b33135efSNetali                    break;
6945f8ab21dSNetali                case 456:
6955f8ab21dSNetali                    msg($this->getLang('msg_translation_fail_quota_exceeded'), -1);
6965f8ab21dSNetali                    break;
6975f8ab21dSNetali                default:
6985f8ab21dSNetali                    msg($this->getLang('msg_translation_fail'), -1);
6995f8ab21dSNetali                    break;
7005f8ab21dSNetali            }
7015f8ab21dSNetali
7023832d0abSNetali            // if any error occurred return an empty string
7035f8ab21dSNetali            return '';
7045f8ab21dSNetali        }
7053832d0abSNetali
7063832d0abSNetali        $json_response = json_decode($raw_response, true);
7073832d0abSNetali        $translated_text = $json_response['translations'][0]['text'];
7083832d0abSNetali
7093832d0abSNetali        $translated_text = $this->remove_ignore_tags($translated_text);
7103832d0abSNetali
7113832d0abSNetali        return $translated_text;
7123832d0abSNetali    }
7133832d0abSNetali
714153e4498SNetali    private function get_push_langs(): array {
715153e4498SNetali        $push_langs = trim($this->getConf('push_langs'));
716153e4498SNetali
717153e4498SNetali        if ($push_langs === '') return array();
718153e4498SNetali
719153e4498SNetali        return explode(' ', $push_langs);
720153e4498SNetali    }
721153e4498SNetali
7220180404cSNetali    private function patch_links($text, $target_lang, $ns): string {
7230180404cSNetali        /*
7240180404cSNetali         * 1. Find links in [[ aa:bb ]] or [[ aa:bb | cc ]]
7250180404cSNetali         * 2. Extract aa:bb
7260180404cSNetali         * 3. Check if lang:aa:bb exists
7270180404cSNetali         * 3.1. --> Yes --> replace
7280180404cSNetali         * 3.2. --> No --> leave it as it is
7290180404cSNetali         */
7303832d0abSNetali
7310180404cSNetali
7320180404cSNetali        /*
7330180404cSNetali         * LINKS
7340180404cSNetali         */
7350180404cSNetali
7364b84d3cfSNetali        preg_match_all('/\[\[([\s\S]*?)(#[\s\S]*?)?((\|)([\s\S]*?))?]]/', $text, $matches, PREG_SET_ORDER);
7370180404cSNetali
7380180404cSNetali        foreach ($matches as $match) {
7390180404cSNetali
7400180404cSNetali            // external link --> skip
741a3a51507SNetali            if (strpos($match[1], '://') !== false) continue;
7420180404cSNetali
74384cda41fSNetali            // skip interwiki links
74484cda41fSNetali            if (strpos($match[1], '>') !== false) continue;
74584cda41fSNetali
7462a12605eSnetali            // skip mail addresses
7472a12605eSnetali            if (strpos($match[1], '@') !== false) continue;
7482a12605eSnetali
74984cda41fSNetali            // skip windows share links
75084cda41fSNetali            if (strpos($match[1], '\\\\') !== false) continue;
75184cda41fSNetali
75284cda41fSNetali            $resolved_id = trim($match[1]);
7530180404cSNetali
7540180404cSNetali            resolve_pageid($ns, $resolved_id, $exists);
7550180404cSNetali
75653f3766cSNetali            $resolved_id_full = $resolved_id;
7570180404cSNetali
7586663bcb5SNetali            // if the link already points to a target in a language namespace drop it and add the new language namespace
7596663bcb5SNetali            $split_id = explode(':', $resolved_id);
7606663bcb5SNetali            $lang_ns = array_shift($split_id);
7616663bcb5SNetali            if (array_key_exists($lang_ns, $this->langs)) {
7626663bcb5SNetali                $resolved_id = implode(':', $split_id);
7636663bcb5SNetali            }
7646663bcb5SNetali
7650180404cSNetali            $lang_id = $target_lang . ':' . $resolved_id;
7660180404cSNetali
7670180404cSNetali            if (!page_exists($lang_id)) {
76853f3766cSNetali                // Page in target lang does not exist --> replace with absolute ID in case it was a relative ID
76953f3766cSNetali                $new_link = '[[' . $resolved_id_full . $match[2] . $match[3] . ']]';
77053f3766cSNetali            } else {
77153f3766cSNetali                // Page in target lang exists --> replace link
7724b84d3cfSNetali                $new_link = '[[' . $lang_id . $match[2] . $match[3] . ']]';
77353f3766cSNetali            }
7740180404cSNetali
7750180404cSNetali            $text = str_replace($match[0], $new_link, $text);
7760180404cSNetali
7770180404cSNetali        }
7780180404cSNetali
7790180404cSNetali        /*
7800180404cSNetali         * MEDIA
7810180404cSNetali         */
7820180404cSNetali
78384cda41fSNetali        preg_match_all('/\{\{(([\s\S]*?)(\?[\s\S]*?)?)(\|([\s\S]*?))?}}/', $text, $matches, PREG_SET_ORDER);
7840180404cSNetali
7850180404cSNetali        foreach ($matches as $match) {
7860180404cSNetali
7870180404cSNetali            // external image --> skip
788a3a51507SNetali            if (strpos($match[1], '://') !== false) continue;
789a3a51507SNetali
790a3a51507SNetali            // skip things like {{tag>...}}
791a3a51507SNetali            if (strpos($match[1], '>') !== false) continue;
7920180404cSNetali
79384cda41fSNetali            // keep alignment
79484cda41fSNetali            $align_left = "";
79584cda41fSNetali            $align_right = "";
79684cda41fSNetali
79784cda41fSNetali            // align left --> space in front of ID
79884cda41fSNetali            if (substr($match[1], 0, 1) == " ") $align_left = " ";
79984cda41fSNetali            // align right --> space behind id
80084cda41fSNetali            if (substr($match[1], -1) == " ") $align_right = " ";
80184cda41fSNetali
80284cda41fSNetali            $resolved_id = trim($match[2]);
80384cda41fSNetali            $params = trim($match[3]);
8040180404cSNetali
8050180404cSNetali            resolve_mediaid($ns, $resolved_id, $exists);
8060180404cSNetali
80753f3766cSNetali            $resolved_id_full = $resolved_id;
8080180404cSNetali
8096663bcb5SNetali            // if the link already points to a target in a language namespace drop it and add the new language namespace
8106663bcb5SNetali            $split_id = explode(':', $resolved_id);
8116663bcb5SNetali            $lang_ns = array_shift($split_id);
8126663bcb5SNetali            if (array_key_exists($lang_ns, $this->langs)) {
8136663bcb5SNetali                $resolved_id = implode(':', $split_id);
8146663bcb5SNetali            }
8156663bcb5SNetali
8160180404cSNetali            $lang_id = $target_lang . ':' . $resolved_id;
8170180404cSNetali
8180180404cSNetali            $lang_id_fn = mediaFN($lang_id);
8190180404cSNetali
8200180404cSNetali            if (!file_exists($lang_id_fn)) {
82153f3766cSNetali                // media in target lang does not exist --> replace with absolute ID in case it was a relative ID
82284cda41fSNetali                $new_link = '{{' . $align_left . $resolved_id_full . $params . $align_right . $match[4] . '}}';
82353f3766cSNetali            } else {
82453f3766cSNetali                // media in target lang exists --> replace it
82584cda41fSNetali                $new_link = '{{' . $align_left . $lang_id . $params . $align_right . $match[4] . '}}';
82653f3766cSNetali            }
8270180404cSNetali
8280180404cSNetali            $text = str_replace($match[0], $new_link, $text);
8290180404cSNetali
8300180404cSNetali        }
8310180404cSNetali
8320180404cSNetali        return $text;
8330180404cSNetali    }
8340180404cSNetali
8350180404cSNetali    private function insert_ignore_tags($text): string {
8360180404cSNetali        // ignore every other xml-like tags (the tags themselves, not their content), otherwise deepl would break the formatting
8370180404cSNetali        $text = preg_replace('/<[\s\S]+?>/', '<ignore>${0}</ignore>', $text);
8380180404cSNetali
8391cd781c4SNetali        // prevent deepl from breaking headings
8401cd781c4SNetali        $text = preg_replace('/={1,6}/', '<ignore>${0}</ignore>', $text);
8411cd781c4SNetali
84243d62a6bSnetali        // prevent deepl from messing with nocache-instructions
84343d62a6bSnetali        $text = str_replace("~~NOCACHE~~", "<ignore>~~NOCACHE~~</ignore>", $text);
84443d62a6bSnetali
845a3a51507SNetali        // fix for plugins like tag or template
846a3a51507SNetali        $text = preg_replace('/\{\{[\s\w]+?>[\s\S]*?}}/', '<ignore>${0}</ignore>', $text);
8470180404cSNetali
8483b1ff295SNetali        // ignore links in wikitext (outside of dokuwiki-links)
8493b1ff295SNetali        $text = preg_replace('/\S+:\/\/\S+/', '<ignore>${0}</ignore>', $text);
8503b1ff295SNetali
8510180404cSNetali        // ignore link/media ids but translate the text (if existing)
8524b84d3cfSNetali        $text = preg_replace('/\[\[([\s\S]*?)(#[\s\S]*?)?((\|)([\s\S]*?))?]]/', '<ignore>[[${1}${2}${4}</ignore>${5}<ignore>]]</ignore>', $text);
8530180404cSNetali        $text = preg_replace('/\{\{([\s\S]*?)(\?[\s\S]*?)?((\|)([\s\S]*?))?}}/', '<ignore>{{${1}${2}${4}</ignore>${5}<ignore>}}</ignore>', $text);
8540180404cSNetali
8553b1ff295SNetali        // prevent deepl from messing with tables
8563b1ff295SNetali        $text = str_replace("  ^  ", "<ignore>  ^  </ignore>", $text);
8577c99a9b0Sextrasec        $text = str_replace("  ^ ", "<ignore>  ^ </ignore>", $text);
8587c99a9b0Sextrasec        $text = str_replace(" ^  ", "<ignore> ^  </ignore>", $text);
8597c99a9b0Sextrasec        $text = str_replace("^  ", "<ignore>^  </ignore>", $text);
8607c99a9b0Sextrasec        $text = str_replace("  ^", "<ignore>  ^</ignore>", $text);
8617c99a9b0Sextrasec        $text = str_replace("^", "<ignore>^</ignore>", $text);
8627c99a9b0Sextrasec        $text = str_replace("  |  ", "<ignore>  |  </ignore>", $text);
8637c99a9b0Sextrasec        $text = str_replace("  | ", "<ignore>  | </ignore>", $text);
8647c99a9b0Sextrasec        $text = str_replace(" |  ", "<ignore> |  </ignore>", $text);
8657c99a9b0Sextrasec        $text = str_replace("|  ", "<ignore>|  </ignore>", $text);
8667c99a9b0Sextrasec        $text = str_replace("  |", "<ignore>  |</ignore>", $text);
8673b1ff295SNetali        $text = str_replace("|", "<ignore>|</ignore>", $text);
8683b1ff295SNetali
8690180404cSNetali        // prevent deepl from doing strange things with dokuwiki syntax
8706301ba6eSNetali        // if a full line is formatted, we have to double-ignore for some reason
8716301ba6eSNetali        $text = str_replace("''", "<ignore><ignore>''</ignore></ignore>", $text);
8726301ba6eSNetali        $text = str_replace("//", "<ignore><ignore>//</ignore></ignore>", $text);
8736301ba6eSNetali        $text = str_replace("**", "<ignore><ignore>**</ignore></ignore>", $text);
8746301ba6eSNetali        $text = str_replace("__", "<ignore><ignore>__</ignore></ignore>", $text);
8756301ba6eSNetali        $text = str_replace("\\\\", "<ignore><ignore>\\\\</ignore></ignore>", $text);
8760180404cSNetali
87713221d46SNetali        // prevent deepl from messing with smileys
87813221d46SNetali        $smileys = array_keys(getSmileys());
87913221d46SNetali        foreach ($smileys as $smiley) {
88013221d46SNetali            $text = str_replace($smiley, "<ignore>" . $smiley . "</ignore>", $text);
88113221d46SNetali        }
88213221d46SNetali
8830180404cSNetali        // ignore code tags
8840180404cSNetali        $text = preg_replace('/(<php[\s\S]*?>[\s\S]*?<\/php>)/', '<ignore>${1}</ignore>', $text);
885057940f7SNetali        $text = preg_replace('/(<file[\s\S]*?>[\s\S]*?<\/file>)/', '<ignore>${1}</ignore>', $text);
886057940f7SNetali        $text = preg_replace('/(<code[\s\S]*?>[\s\S]*?<\/code>)/', '<ignore>${1}</ignore>', $text);
887057940f7SNetali
8880180404cSNetali        // ignore the expressions from the ignore list
8893832d0abSNetali        $ignored_expressions = explode(':', $this->getConf('ignored_expressions'));
8903832d0abSNetali
8913832d0abSNetali        foreach ($ignored_expressions as $expression) {
8923832d0abSNetali            $text = str_replace($expression, '<ignore>' . $expression . '</ignore>', $text);
8933832d0abSNetali        }
8943832d0abSNetali
8953832d0abSNetali        return $text;
8963832d0abSNetali    }
8973832d0abSNetali
8983832d0abSNetali    private function remove_ignore_tags($text): string {
8993832d0abSNetali        $ignored_expressions = explode(':', $this->getConf('ignored_expressions'));
9003832d0abSNetali
9013832d0abSNetali        foreach ($ignored_expressions as $expression) {
9023832d0abSNetali            $text = str_replace('<ignore>' . $expression . '</ignore>', $expression, $text);
9033832d0abSNetali        }
9043832d0abSNetali
90543d62a6bSnetali        // prevent deepl from messing with nocache-instructions
90643d62a6bSnetali        $text = str_replace("<ignore>~~NOCACHE~~</ignore>", "~~NOCACHE~~", $text);
90743d62a6bSnetali
9083b1ff295SNetali        // prevent deepl from messing with tables
9093b1ff295SNetali        $text = str_replace("<ignore>^</ignore>", "^", $text);
9107c99a9b0Sextrasec        $text = str_replace("<ignore>^  </ignore>", "^  ", $text);
9117c99a9b0Sextrasec        $text = str_replace("<ignore>  ^</ignore>", "  ^", $text);
9127c99a9b0Sextrasec        $text = str_replace("<ignore> ^  </ignore>", " ^  ", $text);
9137c99a9b0Sextrasec        $text = str_replace("<ignore>  ^ </ignore>", "  ^ ", $text);
9147c99a9b0Sextrasec        $text = str_replace("<ignore>  ^  </ignore>", "  ^  ", $text);
9157c99a9b0Sextrasec        $text = str_replace("<ignore>|</ignore>", "|", $text);
9167c99a9b0Sextrasec        $text = str_replace("<ignore>|  </ignore>", "|  ", $text);
9177c99a9b0Sextrasec        $text = str_replace("<ignore>  |</ignore>", "  |", $text);
9187c99a9b0Sextrasec        $text = str_replace("<ignore> |  </ignore>", " |  ", $text);
9197c99a9b0Sextrasec        $text = str_replace("<ignore>  | </ignore>", "  | ", $text);
9203b1ff295SNetali        $text = str_replace("<ignore>  |  </ignore>", "  |  ", $text);
9213b1ff295SNetali
9226301ba6eSNetali        $text = str_replace("<ignore><ignore>''</ignore></ignore>", "''", $text);
9236301ba6eSNetali        $text = str_replace("<ignore><ignore>//</ignore></ignore>", "//", $text);
9246301ba6eSNetali        $text = str_replace("<ignore><ignore>**</ignore></ignore>", "**", $text);
9256301ba6eSNetali        $text = str_replace("<ignore><ignore>__</ignore></ignore>", "__", $text);
9266301ba6eSNetali        $text = str_replace("<ignore><ignore>\\\\</ignore></ignore>", "\\\\", $text);
9270180404cSNetali
9283b1ff295SNetali        // ignore links in wikitext (outside of dokuwiki-links)
9293b1ff295SNetali        $text = preg_replace('/<ignore>(\S+:\/\/\S+)<\/ignore>/', '${1}', $text);
9303b1ff295SNetali
9314b84d3cfSNetali        $text = preg_replace('/<ignore>\[\[([\s\S]*?)(\|)?(<\/ignore>)([\s\S]*?)?<ignore>]]<\/ignore>/', '[[${1}${2}${4}]]', $text);
9324b84d3cfSNetali        $text = preg_replace('/<ignore>\{\{([\s\S]*?)(\|)?(<\/ignore>)([\s\S]*?)?<ignore>}}<\/ignore>/', '{{${1}${2}${4}}}', $text);
9334b84d3cfSNetali
93413221d46SNetali        // prevent deepl from messing with smileys
93513221d46SNetali        $smileys = array_keys(getSmileys());
93613221d46SNetali        foreach ($smileys as $smiley) {
93713221d46SNetali            $text = str_replace("<ignore>" . $smiley . "</ignore>", $smiley, $text);
93813221d46SNetali        }
93913221d46SNetali
9400180404cSNetali        $text = preg_replace('/<ignore>(<php[\s\S]*?>[\s\S]*?<\/php>)<\/ignore>/', '${1}', $text);
9410180404cSNetali        $text = preg_replace('/<ignore>(<file[\s\S]*?>[\s\S]*?<\/file>)<\/ignore>/', '${1}', $text);
9420180404cSNetali        $text = preg_replace('/<ignore>(<code[\s\S]*?>[\s\S]*?<\/code>)<\/ignore>/', '${1}', $text);
9430180404cSNetali
944a3a51507SNetali        // fix for plugins like tag or template
945a3a51507SNetali        $text = preg_replace('/<ignore>(\{\{[\s\w]+?>[\s\S]*?}})<\/ignore>/', '${1}', $text);
9460180404cSNetali
9471cd781c4SNetali        // prevent deepl from breaking headings
9481cd781c4SNetali        $text = preg_replace('/<ignore>(={1,6})<\/ignore>/','${1}', $text);
9491cd781c4SNetali
9501cd781c4SNetali        // ignore every other xml-like tags (the tags themselves, not their content), otherwise deepl would break the formatting
9511cd781c4SNetali        $text = preg_replace('/<ignore>(<[\s\S]+?>)<\/ignore>/', '${1}', $text);
9521cd781c4SNetali
9530180404cSNetali        // restore < and > for example from arrows (-->) in wikitext
9540180404cSNetali        $text = str_replace('&gt;', '>', $text);
9550180404cSNetali        $text = str_replace('&lt;', '<', $text);
9560180404cSNetali
9573b1ff295SNetali        // restore & in wikitext
9583b1ff295SNetali        $text = str_replace('&amp;', '&', $text);
9593b1ff295SNetali
9603832d0abSNetali        return $text;
9613832d0abSNetali    }
9623832d0abSNetali}
9633832d0abSNetali
964