better classification

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

diff --git a/app/Models/ChatLog.php b/app/Models/ChatLog.php
index f918269073a0fa733e7ea9765a0a57d514b2bca2..063e8036485cdf5cc31ef793215db38f63a90b4c 100644 (file)
--- a/app/Models/ChatLog.php
+++ b/app/Models/ChatLog.php
@@ -2,11 +2,12 @@
 
 namespace App\Models;
 
+use App\TwitchBot\TokenizedMessage;
 use Illuminate\Database\Eloquent\Factories\HasFactory;
 use Illuminate\Database\Eloquent\Model;
 use Illuminate\Support\Arr;
 use Illuminate\Support\Str;
-use LanguageDetector\LanguageDetector;
+use LanguageDetection\Language;
 
 class ChatLog extends Model {
 
@@ -20,6 +21,28 @@ class ChatLog extends Model {
                return $this->belongsTo(User::class);
        }
 
+       public function tokenize() {
+               return TokenizedMessage::fromLog($this);
+       }
+
+       public function getTextWithoutEmotes() {
+               $text = $this->text_content;
+               if (isset($this->tags['emotes']) && !empty($this->tags['emotes'])) {
+                       $emotes = explode('/', $this->tags['emotes']);
+                       foreach ($emotes as $emote) {
+                               $set = explode(':', $emote);
+                               $positions = explode(',', $set[1]);
+                               foreach ($positions as $position) {
+                                       $coords = explode('-', $position);
+                                       for ($i = intval($coords[0]); $i <= intval($coords[1]); ++$i) {
+                                               $text[$i] = ' ';
+                                       }
+                               }
+                       }
+               }
+               return trim(preg_replace('/\s+/', ' ', $text));
+       }
+
        public function evaluate() {
                $this->evaluateUser();
                $this->evaluateChannel();
@@ -34,7 +57,7 @@ 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';
@@ -43,66 +66,37 @@ class ChatLog extends Model {
                        }
                        $this->text_content = $this->params[1];
                        $this->detectLanguage();
-                       if ($this->scanForSpam()) {
+                       $tokenized = $this->tokenize();
+                       if ($tokenized->isSpammy()) {
                                $this->banned = true;
                        }
-                       $this->classification = static::classify($this->text_content);
+                       $this->classification = $tokenized->classify();
                        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';
-               }
-               return 'unclassified';
-       }
-
        protected function evaluateUser() {
        }
 
        protected function evaluateChannel() {
                if (empty($this->params)) {
-                       $this->channel()->associate(null);
                        return;
                }
                $cname = $this->params[0];
@@ -110,7 +104,12 @@ class ChatLog extends Model {
                        $cname = '#'.$cname;
                }
                $channel = Channel::firstWhere('twitch_chat', '=', $cname);
-               $this->channel()->associate($channel);
+               if (!is_null($channel)) {
+                       $this->channel()->associate($channel);
+                       if (empty($this->twitch_category) && now()->sub(15, 'minute')->isBefore($this->created_at)) {
+                               $this->twitch_category = $channel->twitch_category;
+                       }
+               }
        }
 
        protected function detectLanguage() {
@@ -121,49 +120,15 @@ class ChatLog extends Model {
                                $languages[] = 'en';
                        }
                }
-               $detector = LanguageDetector::detect($this->text_content, $languages);
-               $scores = $detector->getScores();
-               $lang = strval($detector->getLanguage());
+               $detector = (new Language($languages))->detect($this->getTextWithoutEmotes());
+               $scores = $detector->close();
+               $lang = strval($detector);
                //var_dump($scores, $lang, $this->text_content);
-               if (is_array($scores) && isset($scores[$lang]) && $scores[$lang] > 0.35) {
+               if (!empty($lang) && $scores[$lang] > 0.4) {
                        $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',