token classification test

[alttp.git] / app / Console / Commands / ChatlibDatabase.php

diff --git a/app/Console/Commands/ChatlibDatabase.php b/app/Console/Commands/ChatlibDatabase.php
index bfe0135c4b8e00e684a9f2974faf862f447296fd..b701ad144b3d0aed3abc26cb481813e4347b0b4e 100644 (file)
--- a/app/Console/Commands/ChatlibDatabase.php
+++ b/app/Console/Commands/ChatlibDatabase.php
@@ -28,13 +28,18 @@ class ChatlibDatabase extends Command {
         * @return int
         */
        public function handle() {
+               $count = 0;
+
                $de = new ChatLib();
                $en = new ChatLib();
 
                ChatLog::where('type', '=', 'chat')
                        ->where('banned', '=', false)
                        ->whereNotNull('evaluated_at')
-                       ->chunk(5000, function ($msgs) use ($de, $en) {
+                       ->where('created_at', '<', now()->sub(7, 'day'))
+                       ->whereNotIn('classification', ['gg', 'gl', 'number', 'o7'])
+                       ->whereRaw('LENGTH(`text_content`) > 12')
+                       ->chunk(5000, function ($msgs) use (&$count, $de, $en) {
                                foreach ($msgs as $msg) {
                                        if ($msg->detected_language === 'de') {
                                                $de->addMessage($msg);
@@ -44,20 +49,16 @@ class ChatlibDatabase extends Command {
                                                $de->addMessage($msg);
                                                $en->addMessage($msg);
                                        }
+                                       ++$count;
                                }
+                               $this->line($count);
                        });
 
                $de->compile();
-               $en->compile();
+               $de->saveAs('de');
 
-               $this->line('');
-               for ($i = 0; $i < 50; ++$i) {
-                       $this->line($de->generate());
-               }
-               $this->line('');
-               for ($i = 0; $i < 50; ++$i) {
-                       $this->line($en->generate());
-               }
+               $en->compile();
+               $en->saveAs('en');
 
                return 0;
        }