X-Git-Url: https://git.localhorst.tv/?a=blobdiff_plain;f=app%2FModels%2FChatLib.php;h=3f1530b396b9b752652918c45d9f0a0fdea3d596;hb=26d47ca368d8e7c2690cec49f6ae2ad509a0428d;hp=8134d136b61122c22847193e0402e28d1a0ed281;hpb=f18af7cfb219ab9c07635ea8bbae80f2a9cee78e;p=alttp.git

diff --git a/app/Models/ChatLib.php b/app/Models/ChatLib.php
index 8134d13..3f1530b 100644
--- a/app/Models/ChatLib.php
+++ b/app/Models/ChatLib.php
@@ -2,82 +2,67 @@
 
 namespace App\Models;
 
+use Illuminate\Support\Facades\Storage;
+
 class ChatLib {
 
-	public function addMessage($msg) {
-		$tokens = array_values(array_filter(preg_split('/\s+/u', $msg->text_content)));
+	public function addMessage(ChatLog $msg) {
+		$this->addText($msg->text_content);
+	}
+
+	public function addText($text) {
+		$tokens = $this->tokenize($text);
 		if (empty($tokens)) return;
-		$tokens [] = '';
+		$tokens[] = '';
 		foreach ($tokens as $num => $token) {
 			if ($num === 0) {
-				$this->addStart($token);
-			}
-			if ($num > 0) {
-				$this->addOne($tokens[$num - 1], $token);
-			}
-			if ($num > 1) {
-				$this->addTwo($tokens[$num - 2], $tokens[$num - 1], $token);
-			}
-			if ($num > 2) {
-				$this->addThree($tokens[$num - 3], $tokens[$num - 2], $tokens[$num - 1], $token);
-			}
-			if ($num > 3) {
-				$this->addFour($tokens[$num - 4], $tokens[$num - 3], $tokens[$num - 2], $tokens[$num - 1], $token);
-			}
-			if ($num > 4) {
-				$this->addFive($tokens[$num - 5], $tokens[$num - 4], $tokens[$num - 3], $tokens[$num - 2], $tokens[$num - 1], $token);
+				$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;
+				}
 			}
 		}
 	}
 
 	public function compile() {
-		$this->start = $this->index($this->start);
-		foreach ($this->one as $key => $value) {
-			$this->one[$key] = $this->index($this->one[$key]);
-			if (empty($this->one[$key])) {
-				unset($this->one[$key]);
-			}
-		}
-		foreach ($this->two as $key => $value) {
-			$this->two[$key] = $this->index($this->two[$key]);
-			if (empty($this->two[$key])) {
-				unset($this->two[$key]);
-			}
-		}
-		foreach ($this->three as $key => $value) {
-			$this->three[$key] = $this->index($this->three[$key]);
-			if (empty($this->three[$key])) {
-				unset($this->three[$key]);
-			}
-		}
-		foreach ($this->four as $key => $value) {
-			$this->four[$key] = $this->index($this->four[$key]);
-			if (empty($this->four[$key])) {
-				unset($this->four[$key]);
-			}
-		}
-		foreach ($this->five as $key => $value) {
-			$this->five[$key] = $this->index($this->five[$key]);
-			if (empty($this->five[$key])) {
-				unset($this->five[$key]);
+		foreach ($this->transitions as $key => $value) {
+			$this->transitions[$key] = $this->index($this->transitions[$key]);
+			if (empty($this->transitions[$key])) {
+				unset($this->transitions[$key]);
 			}
 		}
 	}
 
-	public function generate($limit = 75) {
-		$tokens = [];
-		$start = $this->randomStart();
-		$tokens[] = $start;
-		$generated = $start;
+	public function generate($limit = 100) {
+		$tokens = [''];
+		$generated = '';
 		while (strlen($generated) < $limit) {
 			$next = $this->randomNext($tokens);
-			if (empty($next)) break;
+			if ($next === '') break;
 			$tokens[] = $next;
-			$generated .= ' '.$next;
+			$generated .= $next;
 		}
 		return $generated;
 	}
 
+	public function saveAs($name) {
+		$data = [
+			'size' => $this->size,
+			'transitions' => $this->transitions,
+		];
+		Storage::disk('chatlib')->put($name.'.json', json_encode($data));
+	}
+
+	public function loadFrom($name) {
+		$data = json_decode(Storage::disk('chatlib')->get($name.'.json'), true);
+		$this->size = $data['size'];
+		$this->transitions = $data['transitions'];
+	}
+
 	private function index($arr) {
 		$result = [];
 		$sum = 0;
@@ -85,16 +70,18 @@ class ChatLib {
 			$weight = $entry['count'];
 			if ($weight == 1) continue;
 			$lower = $sum;
-			$sum += intval(pow($weight, 1.4));
+			$sum += $weight;
 			$examples = [];
 			if (is_array(end($entry['examples']))) {
 				// already processed
 				$examples = $entry['examples'];
+			} else if ($key === ' ') {
+				$examples = [[' ', 0, 1]];
 			} else {
 				$subsum = 0;
 				foreach ($entry['examples'] as $example => $subweight) {
 					$sublower = $subsum;
-					$subsum += $subweight * $subweight;
+					$subsum += $subweight;
 					$examples[] = [$example, $sublower, $subsum];
 				}
 			}
@@ -103,79 +90,18 @@ class ChatLib {
 		return $result;
 	}
 
-	private function randomStart() {
-		$pick = $this->pick($this->start);
-		if (is_null($pick)) return '';
-		return $this->exampleOf($pick);
-	}
-
 	private function randomNext($tokens) {
 		$cnt = count($tokens);
-		$picks = [];
-		if ($cnt >= 5) {
-			$cmb = $this->generalize(array_slice($tokens, $cnt - 5, 5));
-			if (isset($this->five[$cmb])) {
-				$pick = $this->pick($this->five[$cmb]);
-				if (!is_null($pick)) {
-					$picks[$pick[0]] = [
-						'count' => 10,
-						'examples' => $pick[3],
-					];
-				}
-			}
-		}
-		if ($cnt >= 4) {
-			$cmb = $this->generalize(array_slice($tokens, $cnt - 4, 4));
-			if (isset($this->four[$cmb])) {
-				$pick = $this->pick($this->four[$cmb]);
+		for ($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)) {
-					$picks[$pick[0]] = [
-						'count' => 12,
-						'examples' => $pick[3],
-					];
+					return $this->exampleOf($pick);
 				}
 			}
 		}
-		if ($cnt >= 3) {
-			$cmb = $this->generalize(array_slice($tokens, $cnt - 3, 3));
-			if (isset($this->three[$cmb])) {
-				$pick = $this->pick($this->three[$cmb]);
-				if (!is_null($pick)) {
-					$picks[$pick[0]] = [
-						'count' => 14,
-						'examples' => $pick[3],
-					];
-				}
-			}
-		}
-		if ($cnt >= 2) {
-			$cmb = $this->generalize(array_slice($tokens, $cnt - 2, 2));
-			if (isset($this->two[$cmb])) {
-				$pick = $this->pick($this->two[$cmb]);
-				if (!is_null($pick)) {
-					$picks[$pick[0]] = [
-						'count' => 4,
-						'examples' => $pick[3],
-					];
-				}
-			}
-		}
-		if ($cnt >= 1) {
-			$cmb = $this->generalize(array_slice($tokens, $cnt - 1, 1));
-			if (isset($this->one[$cmb])) {
-				$pick = $this->pick($this->one[$cmb]);
-				if (!is_null($pick)) {
-					$picks[$pick[0]] = [
-						'count' => 2,
-						'examples' => $pick[3],
-					];
-				}
-			}
-		}
-		if (empty($picks)) return '';
-		$picks = $this->index($picks);
-		$pick = $this->pick($picks);
-		return $this->exampleOf($pick);
+		return '';
 	}
 
 	private function pick($options) {
@@ -200,49 +126,12 @@ class ChatLib {
 		return $options[$min_index];
 	}
 
-	private function addStart($token) {
-		if (empty($token)) return;
-		$this->increment($this->start, $token);
-	}
-
-	private function addOne($one, $token) {
-		$cmb = $this->generalize([$one]);
-		if (!isset($this->one[$cmb])) {
-			$this->one[$cmb] = [];
+	private function addTransition($state, $next) {
+		$cmb = $this->generalize($state);
+		if (!isset($this->transitions[$cmb])) {
+			$this->transitions[$cmb] = [];
 		}
-		$this->increment($this->one[$cmb], $token);
-	}
-
-	private function addTwo($one, $two, $token) {
-		$cmb = $this->generalize([$one, $two]);
-		if (!isset($this->two[$cmb])) {
-			$this->two[$cmb] = [];
-		}
-		$this->increment($this->two[$cmb], $token);
-	}
-
-	private function addThree($one, $two, $three, $token) {
-		$cmb = $this->generalize([$one, $two, $three]);
-		if (!isset($this->three[$cmb])) {
-			$this->three[$cmb] = [];
-		}
-		$this->increment($this->three[$cmb], $token);
-	}
-
-	private function addFour($one, $two, $three, $four, $token) {
-		$cmb = $this->generalize([$one, $two, $three, $four]);
-		if (!isset($this->four[$cmb])) {
-			$this->four[$cmb] = [];
-		}
-		$this->increment($this->four[$cmb], $token);
-	}
-
-	private function addFive($one, $two, $three, $four, $five, $token) {
-		$cmb = $this->generalize([$one, $two, $three, $four, $five]);
-		if (!isset($this->five[$cmb])) {
-			$this->five[$cmb] = [];
-		}
-		$this->increment($this->five[$cmb], $token);
+		$this->increment($this->transitions[$cmb], $next);
 	}
 
 	private function increment(&$which, $token) {
@@ -263,13 +152,22 @@ class ChatLib {
 		}
 	}
 
+	private function tokenize($str) {
+		return array_values(array_filter(preg_split('/\b/u', $str), function($token) {
+			if (empty($token)) return false;
+			if (preg_match('/cheer\d+/u', strtolower($token))) return false;
+			return true;
+		}));
+	}
+
 	private function generalize($tokens) {
 		$str = '';
 		foreach ($tokens as $token) {
-			$replaced = preg_replace('/\W/u', '', $token);
-			$replaced = preg_replace('/\d+/', '0', $replaced);
-			$replaced = strtolower(trim($replaced));
-			$str .= empty($replaced) ? $token : $replaced;
+			$replaced = preg_replace('/\d+/', '0', $token);
+			$replaced = preg_replace('/\s+/', ' ', $token);
+			$replaced = preg_replace('/(.)\1{2,}/', '$1', $token);
+			$replaced = strtolower($replaced);
+			$str .= $replaced;
 		}
 		return $str;
 	}
@@ -279,11 +177,7 @@ class ChatLib {
 		return $example[0];
 	}
 
-	private $start = [];
-	private $one = [];
-	private $two = [];
-	private $three = [];
-	private $four = [];
-	private $five = [];
+	private $size = 7;
+	private $transitions = [];
 
 }