From 771f1761f0abec996838c0ccc71cec0219bad71a Mon Sep 17 00:00:00 2001
From: Daniel Karbach <daniel.karbach@localhorst.tv>
Date: Tue, 7 May 2024 16:54:58 +0200
Subject: [PATCH] chat adlib prototype

---
 app/Console/Commands/ChatlibDatabase.php |  52 ++++++++
 app/Models/ChatLib.php                   | 163 +++++++++++++++++++++++
 2 files changed, 215 insertions(+)
 create mode 100644 app/Console/Commands/ChatlibDatabase.php
 create mode 100644 app/Models/ChatLib.php

diff --git a/app/Console/Commands/ChatlibDatabase.php b/app/Console/Commands/ChatlibDatabase.php
new file mode 100644
index 0000000..13c881a
--- /dev/null
+++ b/app/Console/Commands/ChatlibDatabase.php
@@ -0,0 +1,52 @@
+<?php
+
+namespace App\Console\Commands;
+
+use App\Models\ChatLib;
+use App\Models\ChatLog;
+use Illuminate\Console\Command;
+
+class ChatlibDatabase extends Command {
+
+	/**
+	 * The name and signature of the console command.
+	 *
+	 * @var string
+	 */
+	protected $signature = 'chatlib:database';
+
+	/**
+	 * The console command description.
+	 *
+	 * @var string
+	 */
+	protected $description = 'Updates the ChatLib database';
+
+	/**
+	 * Execute the console command.
+	 *
+	 * @return int
+	 */
+	public function handle() {
+		$db = new ChatLib();
+
+		ChatLog::where('banned', '=', false)
+			->whereNotNull('evaluated_at')
+			->chunk(500, function ($msgs) use ($db) {
+				foreach ($msgs as $msg) {
+					$db->addMessage($msg);
+				}
+			});
+
+		$db->compile();
+
+		for ($i = 0; $i < 50; ++$i) {
+			$this->line($db->generate());
+		}
+
+		return 0;
+	}
+
+}
+
+?>
diff --git a/app/Models/ChatLib.php b/app/Models/ChatLib.php
new file mode 100644
index 0000000..d6d4f56
--- /dev/null
+++ b/app/Models/ChatLib.php
@@ -0,0 +1,163 @@
+<?php
+
+namespace App\Models;
+
+class ChatLib {
+
+	public function addMessage($msg) {
+		$tokens = array_values(array_filter(preg_split('/\b/', $msg->text_content)));
+		if (empty($tokens)) return;
+		$tokens [] = '';
+		foreach ($tokens as $num => $token) {
+			if ($num === 0) {
+				$this->addStart($token);
+			} else if ($num === 1) {
+				$this->addOne($tokens[0], $token);
+			} else if ($num === 2) {
+				$this->addOne($tokens[1], $token);
+				$this->addTwo($tokens[0], $tokens[1], $token);
+			} else {
+				$this->addOne($tokens[$num - 1], $token);
+				$this->addTwo($tokens[$num - 2], $tokens[$num - 1], $token);
+				$this->addThree($tokens[$num - 3], $tokens[$num - 2], $tokens[$num - 1], $token);
+			}
+		}
+	}
+
+	public function compile() {
+		$this->start = $this->index($this->start);
+		foreach ($this->one as $key => $value) {
+			$this->one[$key] = $this->index($this->one[$key]);
+		}
+		foreach ($this->two as $key => $value) {
+			$this->two[$key] = $this->index($this->two[$key]);
+		}
+		foreach ($this->three as $key => $value) {
+			$this->three[$key] = $this->index($this->three[$key]);
+		}
+	}
+
+	public function generate($limit = 50) {
+		$tokens = [];
+		$start = $this->randomStart();
+		$tokens[] = $start;
+		$generated = $start;
+		while (strlen($generated) < $limit) {
+			$next = $this->randomNext($tokens);
+			if (empty($next)) break;
+			$tokens[] = $next;
+			$generated .= $next;
+		}
+		return $generated;
+	}
+
+	private function index($arr) {
+		$result = [];
+		$sum = 0;
+		asort($arr);
+		foreach ($arr as $key => $weight) {
+			$lower = $sum;
+			$sum += $weight;
+			$result[] = [$key, $lower, $sum];
+		}
+		return $result;
+	}
+
+	private function randomStart() {
+		return $this->pick($this->start);
+	}
+
+	private function randomNext($tokens) {
+		$cnt = count($tokens);
+		if ($cnt >= 3) {
+			$cmb = $tokens[$cnt - 3].$tokens[$cnt - 2].$tokens[$cnt - 1];
+			if (isset($this->three[$cmb])) {
+				return $this->pick($this->three[$cmb]);
+			}
+		}
+		if ($cnt >= 2) {
+			$cmb = $tokens[$cnt - 2].$tokens[$cnt - 1];
+			if (isset($this->two[$cmb])) {
+				return $this->pick($this->two[$cmb]);
+			}
+		}
+		if ($cnt >= 1) {
+			$cmb = $tokens[$cnt - 1];
+			if (isset($this->one[$cmb])) {
+				return $this->pick($this->one[$cmb]);
+			}
+		}
+		return '';
+	}
+
+	private function pick($options) {
+		$max = end($options)[2];
+		$num = random_int(0, $max);
+		$min_index = 0;
+		$max_index = count($options) - 1;
+		while ($min_index < $max_index) {
+			$cur_index = intval(($min_index + $max_index) / 2);
+			$cur_low = $options[$cur_index][1];
+			$cur_high = $options[$cur_index][2];
+			if ($cur_low > $num) {
+				$max_index = $cur_index;
+			} else if ($cur_high < $num) {
+				$min_index = $cur_index + 1;
+			} else {
+				$min_index = $cur_index;
+				break;
+			}
+		}
+		return $options[$min_index][0];
+	}
+
+	private function addStart($token) {
+		if (empty($token)) return;
+		if (!isset($this->start[$token])) {
+			$this->start[$token] = 1;
+		} else {
+			++$this->start[$token];
+		}
+	}
+
+	private function addOne($one, $token) {
+		if (!isset($this->one[$one])) {
+			$this->one[$one] = [];
+		}
+		if (!isset($this->one[$one][$token])) {
+			$this->one[$one][$token] = 1;
+		} else {
+			++$this->one[$one][$token];
+		}
+	}
+
+	private function addTwo($one, $two, $token) {
+		$cmb = $one.$two;
+		if (!isset($this->two[$cmb])) {
+			$this->two[$cmb] = [];
+		}
+		if (!isset($this->two[$cmb][$token])) {
+			$this->two[$cmb][$token] = 1;
+		} else {
+			++$this->two[$cmb][$token];
+		}
+	}
+
+	private function addThree($one, $two, $three, $token) {
+		$cmb = $one.$two.$three;
+		if (!isset($this->three[$cmb])) {
+			$this->three[$cmb] = [];
+		}
+		if (!isset($this->three[$cmb][$token])) {
+			$this->three[$cmb][$token] = 1;
+		} else {
+			++$this->three[$cmb][$token];
+		}
+	}
+
+	private $start = [];
+	private $one = [];
+	private $two = [];
+	private $three = [];
+
+}
-- 
2.39.2