X-Git-Url: https://git.localhorst.tv/?a=blobdiff_plain;f=app%2FModels%2FChatLib.php;h=d6a7ffcbbe7906bef5ae3e25b40cdf0e080cad71;hb=1dea58cb6fa9cf28966e75c1e1af87f67e6c0fd1;hp=c7a19b6998d7acafc09e4fefb2b708ba8087f02f;hpb=5df7f3ed455c56370b3c4b690b0ca47b183fa829;p=alttp.git

diff --git a/app/Models/ChatLib.php b/app/Models/ChatLib.php
index c7a19b6..d6a7ffc 100644
--- a/app/Models/ChatLib.php
+++ b/app/Models/ChatLib.php
@@ -6,79 +6,65 @@ use Illuminate\Support\Facades\Storage;
 
 class ChatLib {
 
-	public function __construct($size = 6) {
+	public function __construct($size = 3) {
 		$this->size = $size;
-
 		$converted = [];
 		foreach ($this->categories as $category => $patterns) {
 			$converted_patterns = [];
 			foreach ($patterns as $pattern) {
 				$converted_patterns[] = '/\b'.$pattern.'\b/u';
 			}
-			$converted['%'.strtoupper($category).'%'] = $converted_patterns;
+			$converted[strtoupper($category)] = $converted_patterns;
 		}
 		$this->categories = $converted;
 	}
 
-	public function addMessage(ChatLog $msg) {
-		$this->addText($msg->text_content);
+	public function addMessage(ChatLog $msg, ChatLog $previous = null) {
+		if ($msg->isReply()) {
+			$this->addText($msg->text_content, $msg->getReplyParent());
+		} else if (!is_null($previous)) {
+			$this->addText($msg->text_content, $previous->text_content);
+		} else {
+			$this->addText($msg->text_content);
+		}
 	}
 
-	public function addText($text) {
+	public function addText($text, $context = '') {
 		$tokens = $this->tokenize($text);
-		if (empty($tokens)) return;
-		$tokens[] = '';
-		foreach ($tokens as $num => $token) {
-			if ($num === 0) {
-				$this->addTransition([], $token);
-			} else {
-				$start = max(0, $num - $this->size - 1);
-				$end = $num;
-				for ($i = $start; $i < $end; ++$i) {
-					$this->addTransition(array_slice($tokens, $i, $end - $i), $token);
-					if ($end - $i < 5) break;
-				}
+		for ($i = 0; $i < count($tokens) - $this->size; ++$i) {
+			$this->addTransition(array_slice($tokens, $i, $this->size), $tokens[$i + $this->size]);
+		}
+		if (!empty($context)) {
+			$tokens = $this->tokenizeWithContext($text, $context);
+			$size = min($this->size - 1, count($tokens) - $this->size);
+			for ($i = 0; $i < $size; ++$i) {
+				$this->addTransition(array_slice($tokens, $i, $this->size), $tokens[$i + $this->size]);
 			}
-			$this->addExample(array_slice($tokens, 0, $num), $token);
 		}
 	}
 
 	public function compile() {
 		foreach ($this->transitions as $key => $values) {
-			$this->transitions[$key] = $this->index($values, 2);
-			if (empty($this->transitions[$key])) {
-				unset($this->transitions[$key]);
-			}
-		}
-		foreach ($this->examples as $key => $values) {
-			if (in_array($key, ['', ' '])) {
-				unset($this->examples[$key]);
-				continue;
-			}
-			$this->examples[$key] = $this->index($values, 1);
-			if (empty($this->examples[$key]) || (count($this->examples[$key]) === 1 && $this->examples[$key][0][0] === $key)) {
-				unset($this->examples[$key]);
-			}
+			$this->transitions[$key] = $this->index($values);
 		}
 	}
 
-	public function generate($limit = 100) {
-		$tokens = [''];
-		$generated = '';
-		while (strlen($generated) < $limit) {
-			$next = $this->randomNext($tokens);
-			if ($next === '') break;
-			$tokens[] = $next;
-			$generated .= $next;
+	public function generate($context = null) {
+		if (!is_null($context)) {
+			$tokens = $this->tokenizeWithContext('', $context);
+			$generated = $this->loop($tokens);
+			if (!empty($generated)) {
+				return $generated;
+			}
 		}
-		return $generated;
+		$tokens = $this->tokenize('');
+		return $this->loop($tokens);
 	}
 
 	public function saveAs($name) {
 		$data = [
 			'size' => $this->size,
 			'transitions' => $this->transitions,
-			'examples' => $this->examples,
 		];
 		Storage::disk('chatlib')->put($name.'.json', json_encode($data));
 	}
@@ -87,14 +73,12 @@ class ChatLib {
 		$data = json_decode(Storage::disk('chatlib')->get($name.'.json'), true);
 		$this->size = $data['size'];
 		$this->transitions = $data['transitions'];
-		$this->examples = $data['examples'];
 	}
 
-	private function index($arr, $min_weight = 2) {
+	private function index($arr) {
 		$result = [];
 		$sum = 0;
 		foreach ($arr as $key => $weight) {
-			if ($weight < $min_weight) continue;
 			$lower = $sum;
 			$sum += $weight;
 			$result[] = [$key, $lower, $sum];
@@ -102,18 +86,20 @@ class ChatLib {
 		return $result;
 	}
 
-	private function randomNext($tokens) {
-		$cnt = count($tokens);
-		for ($size = min($this->size, $cnt); $size > 0; --$size) {
-			$cmb = $this->generalize(array_slice($tokens, -$size));
-			if (isset($this->transitions[$cmb])) {
-				$pick = $this->pick($this->transitions[$cmb]);
-				if (!is_null($pick)) {
-					return $this->exampleOf($pick, $tokens);
-				}
-			}
+	private function loop($tokens) {
+		while (count($tokens) < 50) {
+			$next = $this->randomNext($tokens);
+			if ($next === ' ') break;
+			$tokens[] = $next;
 		}
-		return '';
+		return $this->untokenize($tokens);
+	}
+
+	private function randomNext($tokens) {
+		$key = $this->makeKey($tokens);
+		if (!isset($this->transitions[$key])) return ' ';
+		$pick = $this->pick($this->transitions[$key]);
+		return $pick[0];
 	}
 
 	private function pick($options) {
@@ -142,84 +128,62 @@ class ChatLib {
 		return $options[$min_index];
 	}
 
-	private function addTransition($state, $next) {
-		$ctx = $this->generalize($state);
-		$cmb = $this->generalize([$next]);
-		if (!isset($this->transitions[$ctx])) {
-			$this->transitions[$ctx] = [];
+	private function addTransition($tokens, $next) {
+		$key = $this->makeKey($tokens);
+		if (!isset($this->transitions[$key])) {
+			$this->transitions[$key] = [];
 		}
-		if (!isset($this->transitions[$ctx][$cmb])) {
-			$this->transitions[$ctx][$cmb] = 1;
+		if (!isset($this->transitions[$key][$next])) {
+			$this->transitions[$key][$next] = 1;
 		} else {
-			++$this->transitions[$ctx][$cmb];
+			++$this->transitions[$key][$next];
 		}
 	}
 
-	private function addExample($context, $token) {
-		$cmb = $this->generalize([$token]);
-		if (!isset($this->examples[$cmb])) {
-			$this->examples[$cmb] = [];
-		}
-		if (!isset($this->examples[$cmb][$token])) {
-			$this->examples[$cmb][$token] = 1;
-		} else {
-			++$this->examples[$cmb][$token];
+	private function splitText($text) {
+		if (trim($text) === '') return [];
+		return preg_split('/\s+/u', $text);
+	}
+
+	private function makeKey($tokens) {
+		$key = $this->joinText(array_slice($tokens, $this->size * -1));
+		$key = mb_strtolower($key);
+		$key = str_replace(['.', ',', ':', ';', '!', '?', '^', '+', '-', '"', "'", '(', ')', '[', ']'], '', $key);
+		$key = preg_replace('/\d+/u', '0', $key);
+		foreach ($this->categories as $category => $patterns) {
+			$key = preg_replace($patterns, $category, $key);
 		}
+		return $key;
 	}
 
-	private function tokenize($str) {
-		return array_values(array_filter(preg_split('/\b/u', $str), function($token) {
-			if ($token === '') return false;
-			if (preg_match('/cheer\d+/u', strtolower($token))) return false;
-			return true;
-		}));
+	private function joinText($tokens) {
+		return implode(' ', $tokens);
 	}
 
-	private function generalize($tokens) {
-		$str = '';
-		foreach ($tokens as $token) {
-			$replaced = preg_replace('/\d+/u', '0', $token);
-			$replaced = preg_replace('/\s+/u', ' ', $replaced);
-			$replaced = preg_replace('/(.)\1{2,}/u', '$1$1', $replaced);
-			$replaced = strtolower($replaced);
-			foreach ($this->aliases as $canonical => $variants) {
-				if (in_array($replaced, $variants)) {
-					$replaced = $canonical;
-					break;
-				}
-				if ($replaced === $canonical) {
-					break;
-				}
-			}
-			$str .= $replaced;
-		}
-		foreach ($this->categories as $category => $patterns) {
-			$str = preg_replace($patterns, $category, $str);
-		}
-		return $str;
+	private function untokenize($tokens) {
+		return $this->joinText(array_slice($tokens, $this->size));
 	}
 
-	private function exampleOf($pick, $context) {
-		if (!isset($this->examples[$pick[0]])) {
-			return $pick[0];
+	private function tokenize($text) {
+		$tokens = $this->splitText($text);
+		$combined = array_merge(array_fill(0, $this->size, ' '), $tokens);
+		if (!empty($tokens)) {
+			$combined[] = ' ';
 		}
-		if (isset($this->examples[$pick[0]])) {
-			$example = $this->pick($this->examples[$pick[0]]);
-			return $example[0];
+		return $combined;
+	}
+
+	private function tokenizeWithContext($text, $context) {
+		$combined = $this->tokenize($text);
+		$context_tokens = array_slice($this->splitText($context), $this->size * -1 + 1);
+		for ($i = 0; $i < count($context_tokens); ++$i) {
+			$combined[$this->size - $i - 2] = $context_tokens[count($context_tokens) - $i - 1];
 		}
-		return $pick[0];
+		return $combined;
 	}
 
 	private $size;
 	private $transitions = [];
-	private $examples = [];
-
-	private $aliases = [
-		'chest' => ['kiste'],
-		'einen' => ['n', 'nen'],
-		'musik' => ['mukke'],
-		'schade' => ['schad', 'schaade'],
-	];
 
 	private $categories = [
 		'fail' => [