friendica-addons/langfilter/langfilter.php

198 lines
7.0 KiB
PHP
Raw Permalink Normal View History

<?php
/*
* Name: Language Filter
* Version: 0.1
* Description: Filters out postings in languages not spoken by the users
* Author: Tobias Diekershoff <https://f.diekershoff.de/u/tobias>
* License: MIT
*/
use Friendica\App;
use Friendica\Content\Text\BBCode;
use Friendica\Core\Hook;
use Friendica\Core\Renderer;
use Friendica\DI;
/* Define the hooks we want to use
* that is, we have settings, we need to save the settings and we want
* to modify the content of a posting when friendica prepares it.
*/
function langfilter_install()
{
Hook::register('prepare_body_content_filter', 'addon/langfilter/langfilter.php', 'langfilter_prepare_body_content_filter', 10);
Hook::register('addon_settings', 'addon/langfilter/langfilter.php', 'langfilter_addon_settings');
Hook::register('addon_settings_post', 'addon/langfilter/langfilter.php', 'langfilter_addon_settings_post');
}
/* The settings
* 1st check if somebody logged in is calling
* 2nd get the current settings
* 3rd parse a SMARTY3 template, replacing some translateable strings for the form
*/
function langfilter_addon_settings(array &$data)
{
2022-10-20 21:51:49 +00:00
if (!DI::userSession()->getLocalUserId()) {
return;
}
2022-10-20 21:51:49 +00:00
$enabled = DI::pConfig()->get(DI::userSession()->getLocalUserId(), 'langfilter', 'enable',
!DI::pConfig()->get(DI::userSession()->getLocalUserId(), 'langfilter', 'disable'));
2022-10-20 21:51:49 +00:00
$languages = DI::pConfig()->get(DI::userSession()->getLocalUserId(), 'langfilter', 'languages');
$minconfidence = DI::pConfig()->get(DI::userSession()->getLocalUserId(), 'langfilter', 'minconfidence', 0) * 100;
$minlength = DI::pConfig()->get(DI::userSession()->getLocalUserId(), 'langfilter', 'minlength', 32);
$t = Renderer::getMarkupTemplate('settings.tpl', 'addon/langfilter/');
$html = Renderer::replaceMacros($t, [
2021-03-24 22:33:42 +00:00
'$intro' => DI::l10n()->t('This addon tries to identify the language posts are written in. If it does not match any language specified below, posts will be hidden by collapsing them.'),
'$enabled' => ['langfilter_enable', DI::l10n()->t('Use the language filter'), $enabled],
'$languages' => ['langfilter_languages', DI::l10n()->t('Able to read'), $languages, DI::l10n()->t('List of abbreviations (ISO 639-1 codes) for languages you speak, comma separated. For example "de,it".')],
'$minconfidence' => ['langfilter_minconfidence', DI::l10n()->t('Minimum confidence in language detection'), $minconfidence, DI::l10n()->t('Minimum confidence in language detection being correct, from 0 to 100. Posts will not be filtered when the confidence of language detection is below this percent value.')],
'$minlength' => ['langfilter_minlength', DI::l10n()->t('Minimum length of message body'), $minlength, DI::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).')],
2018-01-15 13:15:33 +00:00
]);
$data = [
'addon' => 'langfilter',
'title' => DI::l10n()->t('Language Filter'),
'html' => $html,
'submit' => ['langfilter-settings-submit' => DI::l10n()->t('Save Settings')],
];
}
/* Save the settings
* 1st check it's a logged in user calling
* 2nd check the langfilter form is to be saved
* 3rd save the settings to the DB for later usage
*/
function langfilter_addon_settings_post(array &$b)
{
2022-10-20 21:51:49 +00:00
if (!DI::userSession()->getLocalUserId()) {
return;
}
if (!empty($_POST['langfilter-settings-submit'])) {
$enable = intval($_POST['langfilter_enable'] ?? 0);
$languages = trim($_POST['langfilter_languages'] ?? '');
$minconfidence = max(0, min(100, intval($_POST['langfilter_minconfidence'] ?? 0))) / 100;
$minlength = intval($_POST['langfilter_minlength'] ?? 32);
if ($minlength <= 0) {
$minlength = 32;
}
2022-10-20 21:51:49 +00:00
DI::pConfig()->set(DI::userSession()->getLocalUserId(), 'langfilter', 'enable' , $enable);
DI::pConfig()->set(DI::userSession()->getLocalUserId(), 'langfilter', 'languages' , $languages);
DI::pConfig()->set(DI::userSession()->getLocalUserId(), 'langfilter', 'minconfidence', $minconfidence);
DI::pConfig()->set(DI::userSession()->getLocalUserId(), 'langfilter', 'minlength' , $minlength);
}
}
/* Actually filter postings by their language
* 1st check if the user wants to filter postings
* 2nd get the user settings which languages shall be not filtered out
* 3rd extract the language of a posting
* 4th if the determined language does not fit to the spoken languages
* of the user, then collapse the posting, but provide a link to
* expand it again.
*/
function langfilter_prepare_body_content_filter(&$hook_data)
{
2022-10-20 21:51:49 +00:00
$logged_user = DI::userSession()->getLocalUserId();
if (!$logged_user) {
return;
}
2015-09-09 17:41:38 +00:00
// Never filter own messages
// TODO: find a better way to extract this
$logged_user_profile = DI::baseUrl() . '/profile/' . DI::userSession()->getLocalUserNickname();
2018-04-01 06:31:37 +00:00
if ($logged_user_profile == $hook_data['item']['author-link']) {
return;
}
// Don't filter if language filter is disabled
if (!DI::pConfig()->get($logged_user, 'langfilter', 'enable',
!DI::pConfig()->get($logged_user, 'langfilter', 'disable'))
) {
return;
}
2015-09-10 07:54:26 +00:00
$naked_body = strip_tags(
$hook_data['item']['rendered-html']
??''?: // Equivalent of !empty()
BBCode::convertForUriId($hook_data['item']['uri-id'], $hook_data['item']['body'], BBCode::ACTIVITYPUB)
);
$naked_body = preg_replace('#\s+#', ' ', trim($naked_body));
// Don't filter if body lenght is below minimum
2022-10-20 21:51:49 +00:00
$minlen = DI::pConfig()->get(DI::userSession()->getLocalUserId(), 'langfilter', 'minlength', 32);
if (!$minlen) {
$minlen = 32;
}
if (strlen($naked_body) < $minlen) {
return;
}
2022-10-20 21:51:49 +00:00
$read_languages_string = DI::pConfig()->get(DI::userSession()->getLocalUserId(), 'langfilter', 'languages');
$minconfidence = DI::pConfig()->get(DI::userSession()->getLocalUserId(), 'langfilter', 'minconfidence');
// Don't filter if no spoken languages are configured
2018-04-01 06:31:37 +00:00
if (!$read_languages_string) {
return;
2018-04-01 06:31:37 +00:00
}
$read_languages_array = explode(',', $read_languages_string);
$iso639 = new Matriphe\ISO639\ISO639;
// Extract the language of the post
if (!empty($hook_data['item']['language'])) {
$languages = json_decode($hook_data['item']['language'], true);
if (!is_array($languages)) {
return;
}
2018-04-01 06:31:37 +00:00
foreach ($languages as $iso2 => $confidence) {
break;
}
if (empty($iso2)) {
return;
}
$lang = $iso639->languageByCode1(substr($iso2, 0, 2));
} else {
$opts = $hook_data['item']['postopts'];
if (!$opts) {
// no options associated to post
return;
}
if (!preg_match('/\blang=([^;]*);([^:]*)/', $opts, $matches)) {
// no lang options associated to post
return;
}
$lang = $matches[1];
$confidence = $matches[2];
$iso2 = $iso639->code1ByLanguage($lang);
}
// Do not filter if language detection confidence is too low
if ($minconfidence && $confidence < $minconfidence) {
return;
}
if (!$iso2) {
return;
}
2018-04-01 06:31:37 +00:00
if (!in_array($iso2, $read_languages_array)) {
$hook_data['filter_reasons'][] = DI::l10n()->t('Filtered language: %s', ucfirst($lang));
}
}