better classification

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

diff --git a/app/Models/ChatLog.php b/app/Models/ChatLog.php
index 9e50addc46d50f8d8197b10e58cd7016842ba6fa..063e8036485cdf5cc31ef793215db38f63a90b4c 100644 (file)
--- a/app/Models/ChatLog.php
+++ b/app/Models/ChatLog.php
@@ -2,16 +2,138 @@
 
 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 LanguageDetection\Language;
 
 class ChatLog extends Model {
 
        use HasFactory;
 
+       public function channel() {
+               return $this->belongsTo(Channel::class);
+       }
+
+       public function user() {
+               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();
+               $this->evaluateChannel();
+
+               if (is_null($this->nick)) {
+                       $this->type = 'system';
+                       return;
+               }
+               if (in_array($this->nick, ['horstiebot', 'localhorsttv'])) {
+                       $this->type = 'self';
+                       return;
+               }
+
+               if ($this->command == 'PRIVMSG') {
+                       if (static::isKnownBot($this->nick)) {
+                               $this->type = 'bot';
+                       } else if (substr($this->params[0], 0, 1) == '#') {
+                               $this->type = 'chat';
+                       } else {
+                               $this->type = 'dm';
+                       }
+                       $this->text_content = $this->params[1];
+                       $this->detectLanguage();
+                       $tokenized = $this->tokenize();
+                       if ($tokenized->isSpammy()) {
+                               $this->banned = true;
+                       }
+                       $this->classification = $tokenized->classify();
+                       return;
+               }
+
+               throw new \Exception('unidentified message');
+       }
+
+       public static function isKnownBot($nick) {
+               return in_array(strtolower($nick), [
+                       'birrellthesquirrel',
+                       'funtoon',
+                       'nidbot2000',
+                       'nightbot',
+                       'pokemoncommunitygame',
+                       'speedgaming',
+                       'starbase47',
+                       'streamelements',
+                       'wizebot',
+                       'zockerstuebchen',
+               ]);
+       }
+
+       protected function evaluateUser() {
+       }
+
+       protected function evaluateChannel() {
+               if (empty($this->params)) {
+                       return;
+               }
+               $cname = $this->params[0];
+               if (substr($cname, 0, 1) != '#') {
+                       $cname = '#'.$cname;
+               }
+               $channel = Channel::firstWhere('twitch_chat', '=', $cname);
+               if (!is_null($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() {
+               $languages = ['de', 'en', 'es', 'fr'];
+               if (!is_null($this->channel)) {
+                       $languages = array_values($this->channel->languages);
+                       if (!in_array('en', $languages)) {
+                               $languages[] = 'en';
+                       }
+               }
+               $detector = (new Language($languages))->detect($this->getTextWithoutEmotes());
+               $scores = $detector->close();
+               $lang = strval($detector);
+               //var_dump($scores, $lang, $this->text_content);
+               if (!empty($lang) && $scores[$lang] > 0.4) {
+                       $this->detected_language = $lang;
+               }
+       }
+
        protected $casts = [
+               'banned' => 'boolean',
                'params' => 'array',
                'tags' => 'array',
+               'user_id' => 'string',
        ];
 
        protected $fillable = [