3 * Name: Language Filter
5 * Description: Filters out postings in languages not spoken by the users
6 * Author: Tobias Diekershoff <https://f.diekershoff.de/u/tobias>
11 use Friendica\Content\Text\BBCode;
12 use Friendica\Core\Hook;
13 use Friendica\Core\L10n;
14 use Friendica\Core\PConfig;
15 use Friendica\Core\Renderer;
18 /* Define the hooks we want to use
19 * that is, we have settings, we need to save the settings and we want
20 * to modify the content of a posting when friendica prepares it.
23 function langfilter_install()
25 Hook::register('prepare_body_content_filter', 'addon/langfilter/langfilter.php', 'langfilter_prepare_body_content_filter', 10);
26 Hook::register('addon_settings', 'addon/langfilter/langfilter.php', 'langfilter_addon_settings');
27 Hook::register('addon_settings_post', 'addon/langfilter/langfilter.php', 'langfilter_addon_settings_post');
30 function langfilter_uninstall()
32 Hook::unregister('prepare_body_content_filter', 'addon/langfilter/langfilter.php', 'langfilter_prepare_body_content_filter');
33 Hook::unregister('prepare_body', 'addon/langfilter/langfilter.php', 'langfilter_prepare_body');
34 Hook::unregister('addon_settings', 'addon/langfilter/langfilter.php', 'langfilter_addon_settings');
35 Hook::unregister('addon_settings_post', 'addon/langfilter/langfilter.php', 'langfilter_addon_settings_post');
39 * 1st check if somebody logged in is calling
40 * 2nd get the current settings
41 * 3rd parse a SMARTY3 template, replacing some translateable strings for the form
44 function langfilter_addon_settings(App $a, &$s)
50 $enable_checked = (intval(PConfig::get(local_user(), 'langfilter', 'disable')) ? '' : ' checked="checked" ');
51 $languages = PConfig::get(local_user(), 'langfilter', 'languages');
52 $minconfidence = PConfig::get(local_user(), 'langfilter', 'minconfidence') * 100;
53 $minlength = PConfig::get(local_user(), 'langfilter', 'minlength');
55 $t = Renderer::getMarkupTemplate("settings.tpl", "addon/langfilter/");
56 $s .= Renderer::replaceMacros($t, [
57 '$title' => L10n::t("Language Filter"),
58 '$intro' => L10n::t('This addon tries to identify the language posts are writen in. If it does not match any language specifed below, posts will be hidden by collapsing them.'),
59 '$enabled' => ['langfilter_enable', L10n::t('Use the language filter'), $enable_checked, ''],
60 '$languages' => ['langfilter_languages', L10n::t('Able to read'), $languages, L10n::t('List of abbreviations (iso2 codes) for languages you speak, comma separated. For example "de,it".')],
61 '$minconfidence' => ['langfilter_minconfidence', L10n::t('Minimum confidence in language detection'), $minconfidence, L10n::t('Minimum confidence in language detection being correct, from 0 to 100. Posts will not be filtered when the confidence of language detection is below this percent value.')],
62 '$minlength' => ['langfilter_minlength', L10n::t('Minimum length of message body'), $minlength, L10n::t('Minimum number of characters in message body for filter to be used. Posts shorter than this will not be filtered. Note: Language detection is unreliable for short content (<200 characters).')],
63 '$submit' => L10n::t('Save Settings'),
70 * 1st check it's a logged in user calling
71 * 2nd check the langfilter form is to be saved
72 * 3rd save the settings to the DB for later usage
75 function langfilter_addon_settings_post(App $a, &$b)
81 if (!empty($_POST['langfilter-settings-submit'])) {
82 PConfig::set(local_user(), 'langfilter', 'languages', trim($_POST['langfilter_languages']));
83 $enable = (!empty($_POST['langfilter_enable']) ? intval($_POST['langfilter_enable']) : 0);
84 $disable = 1 - $enable;
85 PConfig::set(local_user(), 'langfilter', 'disable', $disable);
86 $minconfidence = 0 + $_POST['langfilter_minconfidence'];
87 if (!$minconfidence) {
89 } elseif ($minconfidence < 0) {
91 } elseif ($minconfidence > 100) {
94 PConfig::set(local_user(), 'langfilter', 'minconfidence', $minconfidence / 100.0);
96 $minlength = 0 + $_POST['langfilter_minlength'];
99 } elseif ($minlength < 0) {
102 PConfig::set(local_user(), 'langfilter', 'minlength', $minlength);
104 info(L10n::t('Language Filter Settings saved.') . EOL);
108 /* Actually filter postings by their language
109 * 1st check if the user wants to filter postings
110 * 2nd get the user settings which languages shall be not filtered out
111 * 3rd extract the language of a posting
112 * 4th if the determined language does not fit to the spoken languages
113 * of the user, then collapse the posting, but provide a link to
117 function langfilter_prepare_body_content_filter(App $a, &$hook_data)
119 $logged_user = local_user();
124 // Never filter own messages
125 // TODO: find a better way to extract this
126 $logged_user_profile = DI::baseUrl()->get() . '/profile/' . $a->user['nickname'];
127 if ($logged_user_profile == $hook_data['item']['author-link']) {
131 // Don't filter if language filter is disabled
132 if (PConfig::get($logged_user, 'langfilter', 'disable')) {
136 $naked_body = BBCode::toPlaintext($hook_data['item']['body'], false);
138 // Don't filter if body lenght is below minimum
139 $minlen = PConfig::get(local_user(), 'langfilter', 'minlength', 32);
144 if (strlen($naked_body) < $minlen) {
148 $read_languages_string = PConfig::get(local_user(), 'langfilter', 'languages');
149 $minconfidence = PConfig::get(local_user(), 'langfilter', 'minconfidence');
151 // Don't filter if no spoken languages are configured
152 if (!$read_languages_string) {
155 $read_languages_array = explode(',', $read_languages_string);
157 // Extract the language of the post
158 if (!empty($hook_data['item']['language'])) {
159 $languages = json_decode($hook_data['item']['language'], true);
160 if (!is_array($languages)) {
164 foreach ($languages as $iso2 => $confidence) {
172 $lang = Text_LanguageDetect_ISO639::code2ToName($iso2);
174 $opts = $hook_data['item']['postopts'];
176 // no options associated to post
180 if (!preg_match('/\blang=([^;]*);([^:]*)/', $opts, $matches)) {
181 // no lang options associated to post
186 $confidence = $matches[2];
188 $iso2 = Text_LanguageDetect_ISO639::nameToCode2($lang);
191 // Do not filter if language detection confidence is too low
192 if ($minconfidence && $confidence < $minconfidence) {
200 if (!in_array($iso2, $read_languages_array)) {
201 $hook_data['filter_reasons'][] = L10n::t('Filtered language: %s', ucfirst($lang));