belongsTo(Channel::class); } public function user() { return $this->belongsTo(User::class); } public function evaluate() { $this->evaluateUser(); $this->evaluateChannel(); if (is_null($this->nick)) { $this->type = 'system'; return; } if (in_array($this->nick, ['horstiebot', 'localhorsttv'])) { $this->type = 'self'; return; } if ($this->command == 'PRIVMSG') { if ($this->isKnownBot()) { $this->type = 'bot'; } else if (substr($this->params[0], 0, 1) == '#') { $this->type = 'chat'; } else { $this->type = 'dm'; } $this->text_content = $this->params[1]; $this->detectLanguage(); if ($this->scanForSpam()) { $this->banned = true; } $this->classification = static::classify($this->text_content); return; } throw new \Exception('unidentified message'); } public function isKnownBot() { return in_array(strtolower($this->nick), [ 'birrellthesquirrel', 'funtoon', 'nidbot2000', 'nightbot', 'pokemoncommunitygame', 'speedgaming', 'starbase47', 'streamelements', 'wizebot', 'zockerstuebchen', ]); } public static function classify($text) { if (empty($text)) { return 'unclassified'; } if (is_numeric(trim($text))) { return 'number'; } $rawText = strtolower(preg_replace('/[^\w]/', '', $text)); $tokenizedText = preg_split('/\s+/', strtolower(trim($text))); if (Str::startsWith($rawText, 'gg') || Str::endsWith($rawText, 'gg')) { return 'gg'; } if (Str::contains($rawText, ['glgl', 'glhf', 'hfgl'])) { return 'gl'; } if (Str::contains($rawText, ['haha', 'hehe', 'hihi', 'kekw', 'lol', 'lul', 'xd'])) { return 'lol'; } if (Str::startsWith($rawText, ['ahoi', 'hallo', 'hello', 'hi', 'huhu']) || Str::endsWith($rawText, ['hi', 'wave'])) { return 'hi'; } if (Str::contains($rawText, ['pog', 'wow'])) { return 'pog'; } if (Str::contains($rawText, ['hype'])) { return 'hype'; } return 'unclassified'; } protected function evaluateUser() { } protected function evaluateChannel() { if (empty($this->params)) { $this->channel()->associate(null); return; } $cname = $this->params[0]; if (substr($cname, 0, 1) != '#') { $cname = '#'.$cname; } $channel = Channel::firstWhere('twitch_chat', '=', $cname); $this->channel()->associate($channel); } protected function detectLanguage() { $languages = ['de', 'en', 'es', 'fr']; if (!is_null($this->channel)) { $languages = array_values($this->channel->languages); if (!in_array('en', $languages)) { $languages[] = 'en'; } } $detector = LanguageDetector::detect($this->text_content, $languages); $scores = $detector->getScores(); $lang = strval($detector->getLanguage()); //var_dump($scores, $lang, $this->text_content); if (is_array($scores) && isset($scores[$lang]) && $scores[$lang] > 0.35) { $this->detected_language = $lang; } } protected function scanForSpam() { if (substr($this->text_content, 0, 1) == '!') { return true; } if (strpos($this->text_content, '$') !== false) { return true; } if (strpos($this->text_content, '€') !== false) { return true; } if (strpos($this->text_content, '@') !== false) { return true; } if (strpos($this->text_content, '://') !== false) { return true; } if (is_numeric($this->text_content)) { return true; } if (strpos($this->text_content, 'followers') !== false) { return true; } if (strpos($this->text_content, 'promotion') !== false) { return true; } if (strpos($this->text_content, 'viewers') !== false) { return true; } if (strpos($this->text_content, 'view ers') !== false) { return true; } return false; } protected $casts = [ 'banned' => 'boolean', 'params' => 'array', 'tags' => 'array', 'user_id' => 'string', ]; protected $fillable = [ 'command', 'nick', 'params', 'tags', ]; }