fix check for spam

[alttp.git] / app / Models / ChatLog.php

diff --git a/app/Models/ChatLog.php b/app/Models/ChatLog.php
index 6f72c352b42ff356123ad3e1bdb29b8d8407e487..943ce6e659e846effb274ba467c23ecfbd1470ff 100644 (file)
--- a/app/Models/ChatLog.php
+++ b/app/Models/ChatLog.php
@@ -4,6 +4,9 @@ namespace App\Models;
 
 use Illuminate\Database\Eloquent\Factories\HasFactory;
 use Illuminate\Database\Eloquent\Model;
+use Illuminate\Support\Arr;
+use Illuminate\Support\Str;
+use LanguageDetector\LanguageDetector;
 
 class ChatLog extends Model {
 
@@ -31,7 +34,7 @@ class ChatLog extends Model {
                }
 
                if ($this->command == 'PRIVMSG') {
-                       if ($this->isKnownBot()) {
+                       if (static::isKnownBot($this->nick)) {
                                $this->type = 'bot';
                        } else if (substr($this->params[0], 0, 1) == '#') {
                                $this->type = 'chat';
@@ -39,26 +42,65 @@ class ChatLog extends Model {
                                $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($this->nick), [
+       public static function isKnownBot($nick) {
+               return in_array(strtolower($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';
+               }
+               if (Str::startsWith($rawText, 'o7') || Str::endsWith($rawText, 'o7') || Str::contains($rawText, 'salut')) {
+                       return 'o7';
+               }
+               return 'unclassified';
+       }
+
        protected function evaluateUser() {
        }
 
@@ -75,40 +117,61 @@ class ChatLog extends Model {
                $this->channel()->associate($channel);
        }
 
-       protected function scanForSpam() {
-               if (substr($this->text_content, 0, 1) == '!') {
-                       return true;
+       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 = $lang;
                }
-               if (strpos($this->text_content, '$') !== false) {
+       }
+
+       public static function spammyText($text) {
+               if (substr($text, 0, 1) == '!') {
                        return true;
                }
-               if (strpos($this->text_content, '€') !== false) {
+               if (strpos($text, '$') !== false) {
                        return true;
                }
-               if (strpos($this->text_content, '@') !== false) {
+               if (strpos($text, '€') !== false) {
                        return true;
                }
-               if (strpos($this->text_content, '://') !== false) {
+               if (strpos($text, '@') !== false) {
                        return true;
                }
-               if (is_numeric($this->text_content)) {
+               if (strpos($text, '://') !== false) {
                        return true;
                }
-               if (strpos($this->text_content, 'followers') !== false) {
+               if (strpos($text, 'followers') !== false) {
                        return true;
                }
-               if (strpos($this->text_content, 'promotion') !== false) {
+               if (strpos($text, 'promotion') !== false) {
                        return true;
                }
-               if (strpos($this->text_content, 'viewers') !== false) {
+               if (strpos($text, 'viewers') !== false) {
                        return true;
                }
-               if (strpos($this->text_content, 'view ers') !== false) {
+               if (strpos($text, 'view ers') !== false) {
                        return true;
                }
                return false;
        }
 
+       protected function scanForSpam() {
+               if (is_numeric($this->text_content)) {
+                       return true;
+               }
+               return static::spammyText($this->text_content);
+       }
+
        protected $casts = [
                'banned' => 'boolean',
                'params' => 'array',