X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=langfilter%2Flangfilter.php;h=aeb53eebde126892f39a99deca64d2ea7905bade;hb=943187d02ff8420735b076a31cdd97f554bbcb79;hp=d4fe942d0b23c1fe19b44461d9aaa5fe8cc3d052;hpb=5360f08f4295f2306d815a7659fbdff94fda01a3;p=friendica-addons.git diff --git a/langfilter/langfilter.php b/langfilter/langfilter.php index d4fe942d..aeb53eeb 100644 --- a/langfilter/langfilter.php +++ b/langfilter/langfilter.php @@ -7,21 +7,22 @@ * License: MIT */ -require_once('library/langdet/Text/LanguageDetect.php'); +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() { - register_hook('prepare_body', 'addon/langfilter/langfilter.php', 'langfilter_prepare_body', 10); - register_hook('plugin_settings', 'addon/langfilter/langfilter.php', 'langfilter_addon_settings'); - register_hook('plugin_settings_post', 'addon/langfilter/langfilter.php', 'langfilter_addon_settings_post'); -} -function langfilter_uninstall() { - unregister_hook('prepare_body', 'addon/langfilter/langfilter.php', 'langfilter_prepare_body'); - unregister_hook('plugin_settings', 'addon/langfilter/langfilter.php', 'langfilter_addon_settings'); - unregister_hook('plugin_settings_post', 'addon/langfilter/langfilter.php', 'langfilter_addon_settings_post'); + +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 @@ -29,58 +30,63 @@ function langfilter_uninstall() { * 2nd get the current settings * 3rd parse a SMARTY3 template, replacing some translateable strings for the form */ -function langfilter_addon_settings(&$a,&$s) { - if(! local_user()) + +function langfilter_addon_settings(App $a, &$s) +{ + if (!local_user()) { return; + } - $enable_checked = (intval(get_pconfig(local_user(),'langfilter','disable')) ? '' : ' checked="checked" '); - $languages = get_pconfig(local_user(),'langfilter','languages'); - $minconfidence = get_pconfig(local_user(),'langfilter','minconfidence')*100; - $minlength = get_pconfig(local_user(),'langfilter','minlength'); - if(! $languages) - $languages = 'en,de,fr,it,es'; - - $t = get_markup_template("settings.tpl", "addon/langfilter/" ); - $s .= replace_macros ($t, array( - '$title' => t("Language Filter"), - '$intro' => t ('This addon tries to identify the language of a postings. If it does not match any language spoken by you (see below) the posting will be collapsed. Remember detecting the language is not perfect, especially with short postings.'), - '$enabled' => array('langfilter_enable', t('Use the language filter'), $enable_checked, ''), - '$languages' => array('langfilter_languages', t('I speak'), $languages, t('List of abbreviations (iso2 codes) for languages you speak, comma separated. For example "de,it".') ), - '$minconfidence' => array('langfilter_minconfidence', t('Minimum confidence in language detection'), $minconfidence, t('Minimum confidence in language detection being correct, from 0 to 100. Posts will not be filtered when the confidence of language detection is below this percent value.') ), - '$minlength' => array('langfilter_minlength', t('Minimum length of message body'), $minlength, t('Minimum length of message body for language filter to be used. Posts shorter than this number of characters will not be filtered.') ), - '$submit' => t('Save Settings'), - )); + $enabled = DI::pConfig()->get(local_user(), 'langfilter', 'enable', + !DI::pConfig()->get(local_user(), 'langfilter', 'disable')); + + $enable_checked = $enabled ? ' checked="checked"' : ''; + $languages = DI::pConfig()->get(local_user(), 'langfilter', 'languages'); + $minconfidence = DI::pConfig()->get(local_user(), 'langfilter', 'minconfidence', 0) * 100; + $minlength = DI::pConfig()->get(local_user(), 'langfilter', 'minlength' , 32); + + $t = Renderer::getMarkupTemplate("settings.tpl", "addon/langfilter/"); + $s .= Renderer::replaceMacros($t, [ + '$title' => DI::l10n()->t("Language Filter"), + '$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'), $enable_checked, ''], + '$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'), $minlength, DI::l10n()->t('Minimum number of characters in message body for filter to be used. Posts shorter than this will not be filtered. Note: Language detection is unreliable for short content (<200 characters).')], + '$submit' => DI::l10n()->t('Save Settings'), + ]); return; } + /* 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(&$a,&$b) { - if(! local_user()) + +function langfilter_addon_settings_post(App $a, &$b) +{ + if (!local_user()) { return; + } - if($_POST['langfilter-settings-submit']) { - set_pconfig(local_user(),'langfilter','languages',trim($_POST['langfilter_languages'])); - $enable = ((x($_POST,'langfilter_enable')) ? intval($_POST['langfilter_enable']) : 0); - $disable = 1-$enable; - set_pconfig(local_user(),'langfilter','disable', $disable); - $minconfidence = 0+$_POST['langfilter_minconfidence']; - if ( ! $minconfidence ) $minconfidence = 0; - else if ( $minconfidence < 0 ) $minconfidence = 0; - else if ( $minconfidence > 100 ) $minconfidence = 100; - set_pconfig(local_user(),'langfilter','minconfidence', $minconfidence/100.0); - - $minlength = 0+$_POST['langfilter_minlength']; - if ( ! $minlength ) $minlength = 32; - else if ( $minlength < 0 ) $minlength = 32; - set_pconfig(local_user(),'langfilter','minlength', $minlength); - - info( t('Language Filter Settings saved.') . EOL); + if (!empty($_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; + } + + DI::pConfig()->set(local_user(), 'langfilter', 'enable' , $enable); + DI::pConfig()->set(local_user(), 'langfilter', 'languages' , $languages); + DI::pConfig()->set(local_user(), 'langfilter', 'minconfidence', $minconfidence); + DI::pConfig()->set(local_user(), 'langfilter', 'minlength' , $minlength); } } + /* Actually filter postings by their language * 1st check if the user wants to filter postings * 2nd get the user settings which languages shall be not filtered out @@ -89,51 +95,99 @@ function langfilter_addon_settings_post(&$a,&$b) { * of the user, then collapse the posting, but provide a link to * expand it again. */ -function langfilter_prepare_body(&$a,&$b) { - - $logged_user = local_user(); - if ( ! $logged_user ) return; - # Never filter own messages - # TODO: find a better way to extract this - $logged_user_profile = $a->config['system']['url'] . '/profile/' . $a->user['nickname']; - if ( $logged_user_profile == $b['item']['author-link'] ) return; +function langfilter_prepare_body_content_filter(App $a, &$hook_data) +{ + $logged_user = local_user(); + if (!$logged_user) { + return; + } - # Don't filter if language filter is disabled - if( get_pconfig($logged_user,'langfilter','disable') ) return; + // Never filter own messages + // TODO: find a better way to extract this + $logged_user_profile = DI::baseUrl()->get() . '/profile/' . $a->getLoggedInUserNickname(); + if ($logged_user_profile == $hook_data['item']['author-link']) { + return; + } - # Don't filter if body lenght is below minimum - $minlen = get_pconfig(local_user(),'langfilter','minlength'); - if ( ! $minlen ) $minlen = 32; - if ( strlen($b['item']['body']) < $minlen ) return; + // Don't filter if language filter is disabled + if (!DI::pConfig()->get($logged_user, 'langfilter', 'enable', + !DI::pConfig()->get($logged_user, 'langfilter', 'disable')) + ) { + return; + } - $spoken_config = get_pconfig(local_user(),'langfilter','languages'); - $minconfidence = get_pconfig(local_user(),'langfilter','minconfidence'); + if (!empty($hook_data['item']['rendered-html'])) { + $naked_body = strip_tags($hook_data['item']['rendered-html']); + } else { + $naked_body = BBCode::toPlaintext($hook_data['item']['body'], false); + } - # Don't filter if no spoken languages are configured - if ( ! $spoken_config ) return; - $spoken_languages = explode(',', $spoken_config); + // Don't filter if body lenght is below minimum + $minlen = DI::pConfig()->get(local_user(), 'langfilter', 'minlength', 32); + if (!$minlen) { + $minlen = 32; + } - # Extract the language of the post - $opts = $b['item']['postopts']; - if ( ! $opts ) return; # no options associated to post - if ( ! preg_match('/\blang=([^;]*);([^:]*)/', $opts, $matches ) ) - return; # no lang options associated to post + if (strlen($naked_body) < $minlen) { + return; + } - $lang = $matches[1]; - $confidence = $matches[2]; + $read_languages_string = DI::pConfig()->get(local_user(), 'langfilter', 'languages'); + $minconfidence = DI::pConfig()->get(local_user(), 'langfilter', 'minconfidence'); - # Do not filter if language detection confidence is too low - if ( $minconfidence && $confidence < $minconfidence ) return; + // Don't filter if no spoken languages are configured + if (!$read_languages_string) { + return; + } + $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($languages)) { + return; + } + + foreach ($languages as $iso2 => $confidence) { + break; + } + + if (empty($iso2)) { + return; + } + + $lang = $iso639->languageByCode1($iso2); + } 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); + } - $iso2 = Text_LanguageDetect_ISO639::nameToCode2($lang); + // Do not filter if language detection confidence is too low + if ($minconfidence && $confidence < $minconfidence) { + return; + } - if ( ! $iso2 ) return; - $spoken = in_array($iso2, $spoken_languages); + if (!$iso2) { + return; + } - if( ! $spoken ) { - $rnd = random_string(8); - $b['html'] = ''; - } + if (!in_array($iso2, $read_languages_array)) { + $hook_data['filter_reasons'][] = DI::l10n()->t('Filtered language: %s', ucfirst($lang)); + } } -?>