X-Git-Url: https://git.localhorst.tv/?a=blobdiff_plain;f=app%2FModels%2FChatLib.php;h=c7a19b6998d7acafc09e4fefb2b708ba8087f02f;hb=1a863c3ecfa04eacf7df3c5bce71c12244e4625b;hp=c89ef941be5f4c2802eb89c92cc2be6dd7bcfd9d;hpb=64f5be008ef28c812cb7ed87149f4eb86f6021eb;p=alttp.git

diff --git a/app/Models/ChatLib.php b/app/Models/ChatLib.php
index c89ef94..c7a19b6 100644
--- a/app/Models/ChatLib.php
+++ b/app/Models/ChatLib.php
@@ -6,6 +6,20 @@ use Illuminate\Support\Facades\Storage;
 
 class ChatLib {
 
+	public function __construct($size = 6) {
+		$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;
+		}
+		$this->categories = $converted;
+	}
+
 	public function addMessage(ChatLog $msg) {
 		$this->addText($msg->text_content);
 	}
@@ -22,19 +36,30 @@ class ChatLib {
 				$end = $num;
 				for ($i = $start; $i < $end; ++$i) {
 					$this->addTransition(array_slice($tokens, $i, $end - $i), $token);
-					if ($end - $i < 4) break;
+					if ($end - $i < 5) break;
 				}
 			}
+			$this->addExample(array_slice($tokens, 0, $num), $token);
 		}
 	}
 
 	public function compile() {
-		foreach ($this->transitions as $key => $value) {
-			$this->transitions[$key] = $this->index($this->transitions[$key]);
+		foreach ($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]);
+			}
+		}
 	}
 
 	public function generate($limit = 100) {
@@ -53,6 +78,7 @@ class ChatLib {
 		$data = [
 			'size' => $this->size,
 			'transitions' => $this->transitions,
+			'examples' => $this->examples,
 		];
 		Storage::disk('chatlib')->put($name.'.json', json_encode($data));
 	}
@@ -61,29 +87,17 @@ 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($arr) {
+	private function index($arr, $min_weight = 2) {
 		$result = [];
 		$sum = 0;
-		foreach ($arr as $key => $entry) {
-			$weight = $entry['count'];
-			if ($weight == 1) continue;
+		foreach ($arr as $key => $weight) {
+			if ($weight < $min_weight) continue;
 			$lower = $sum;
 			$sum += $weight;
-			$examples = [];
-			if (is_array(end($entry['examples']))) {
-				// already processed
-				$examples = $entry['examples'];
-			} else {
-				$subsum = 0;
-				foreach ($entry['examples'] as $example => $subweight) {
-					$sublower = $subsum;
-					$subsum += $subweight;
-					$examples[] = [$example, $sublower, $subsum];
-				}
-			}
-			$result[] = [$key, $lower, $sum, $examples];
+			$result[] = [$key, $lower, $sum];
 		}
 		return $result;
 	}
@@ -95,7 +109,7 @@ class ChatLib {
 			if (isset($this->transitions[$cmb])) {
 				$pick = $this->pick($this->transitions[$cmb]);
 				if (!is_null($pick)) {
-					return $this->exampleOf($pick);
+					return $this->exampleOf($pick, $tokens);
 				}
 			}
 		}
@@ -104,14 +118,18 @@ class ChatLib {
 
 	private function pick($options) {
 		if (empty($options)) return null;
-		$max = end($options)[2];
+		$max = end($options)[2] - 1;
 		$num = random_int(0, $max);
+		return static::search($options, $num);
+	}
+
+	public static function search($options, $num) {
 		$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];
+			$cur_high = $options[$cur_index][2] - 1;
 			if ($cur_low > $num) {
 				$max_index = $cur_index;
 			} else if ($cur_high < $num) {
@@ -125,34 +143,33 @@ class ChatLib {
 	}
 
 	private function addTransition($state, $next) {
-		$cmb = $this->generalize($state);
-		if (!isset($this->transitions[$cmb])) {
-			$this->transitions[$cmb] = [];
+		$ctx = $this->generalize($state);
+		$cmb = $this->generalize([$next]);
+		if (!isset($this->transitions[$ctx])) {
+			$this->transitions[$ctx] = [];
+		}
+		if (!isset($this->transitions[$ctx][$cmb])) {
+			$this->transitions[$ctx][$cmb] = 1;
+		} else {
+			++$this->transitions[$ctx][$cmb];
 		}
-		$this->increment($this->transitions[$cmb], $next);
 	}
 
-	private function increment(&$which, $token) {
-		$generalized = $this->generalize([$token]);
-		if (!isset($which[$generalized])) {
-			$which[$generalized] = [
-				'count' => 1,
-				'examples' => [],
-			];
-			$which[$generalized]['examples'][$token] = 1;
+	private function addExample($context, $token) {
+		$cmb = $this->generalize([$token]);
+		if (!isset($this->examples[$cmb])) {
+			$this->examples[$cmb] = [];
+		}
+		if (!isset($this->examples[$cmb][$token])) {
+			$this->examples[$cmb][$token] = 1;
 		} else {
-			++$which[$generalized]['count'];
-			if (!isset($which[$generalized]['examples'][$token])) {
-				$which[$generalized]['examples'][$token] = 1;
-			} else {
-				++$which[$generalized]['examples'][$token];
-			}
+			++$this->examples[$cmb][$token];
 		}
 	}
 
 	private function tokenize($str) {
 		return array_values(array_filter(preg_split('/\b/u', $str), function($token) {
-			if (empty($token)) return false;
+			if ($token === '') return false;
 			if (preg_match('/cheer\d+/u', strtolower($token))) return false;
 			return true;
 		}));
@@ -161,19 +178,311 @@ class ChatLib {
 	private function generalize($tokens) {
 		$str = '';
 		foreach ($tokens as $token) {
-			$replaced = preg_replace('/\d+/', '0', $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 exampleOf($pick) {
-		$example = $this->pick($pick[3]);
-		return $example[0];
+	private function exampleOf($pick, $context) {
+		if (!isset($this->examples[$pick[0]])) {
+			return $pick[0];
+		}
+		if (isset($this->examples[$pick[0]])) {
+			$example = $this->pick($this->examples[$pick[0]]);
+			return $example[0];
+		}
+		return $pick[0];
 	}
 
-	private $size = 5;
+	private $size;
 	private $transitions = [];
+	private $examples = [];
+
+	private $aliases = [
+		'chest' => ['kiste'],
+		'einen' => ['n', 'nen'],
+		'musik' => ['mukke'],
+		'schade' => ['schad', 'schaade'],
+	];
+
+	private $categories = [
+		'fail' => [
+			'failfish',
+			'holysm0notlikethis',
+			'notlikethis',
+			'tetobridge0',
+			'vinter0clown',
+		],
+
+		'hype' => [
+			'dergoaparty',
+			'dinodance',
+			'elemen0party',
+			'muftaahype',
+			'luckwuhype',
+			'olliwahype',
+			'osora0umbrihype',
+			'partyhat',
+			'peepocheer',
+			'rei0hype',
+			'sakayahype',
+			'tetotroete',
+			'ticknaboargeil0',
+			'ticknahype0',
+		],
+
+		'kappa' => [
+			'kappa(claus|hd)?',
+		],
+
+		'jam' => [
+			'(cat|dog|rat)jam',
+			'kanash0jam',
+			'rei0jamers',
+			'samusdance',
+		],
+
+		'lol' => [
+			':d',
+			'boothi0lul',
+			'kekw',
+			'lol',
+			'lul',
+			'rei0lul',
+			'samusgrin',
+			'ticknaauslachen',
+			'xd',
+		],
+
+		'love' => [
+			'<3',
+			'duden0love',
+			'exec0love',
+			'krawal0heart',
+			'lodanzhug',
+			'luckwulove',
+			'luvsign',
+			'muftaal',
+			'osora0love',
+			'peepoexcitedhug',
+			'spirit0love',
+			'svenkalove',
+			'ticknaherz',
+		],
+
+		'name' => [
+			'baba',
+			'baka',
+			'bobe?r',
+			'brog(i|or)',
+			'cfate',
+			'danny',
+			'danzi+',
+			'daruck',
+			'dennsen',
+			'dimez',
+			'divi',
+			'dud(en|i+)',
+			'ele',
+			'eri(ror)?',
+			'(name)?faker',
+			'fetti+',
+			'gamma(chuu)?',
+			'goat(buster|ie?|y)?',
+			'hitsu(yan)?',
+			'holy',
+			'jem',
+			'kala(marino)?',
+			'kromb',
+			'koval',
+			'kum(i|o|p)',
+			'lanux',
+			'len(esha|chen)',
+			'leya+',
+			'magno',
+			'malmo',
+			'markam',
+			'micha',
+			'mimsy',
+			'muf(fy|taay)',
+			'murd(elizer|i+)',
+			'nami',
+			'nula',
+			'onio',
+			'paulinche',
+			'phaaze',
+			'ralen',
+			'ramond',
+			'ray(vis)?',
+			'schulzer',
+			'skunk(ner)?',
+			'skipsy',
+			'soli+',
+			'sven(ka+)?',
+			'tantalus',
+			'teto',
+			'thalanee?',
+			'tick(i+|naldo|y+)',
+			'tofu',
+			'tr[i0]x+',
+			'vin(nie?|ny|ter)',
+			'xall',
+			'yasi',
+		],
+
+		'pog' => [
+			'bumble0Pog',
+			'komodohype',
+			'pog',
+			'pogchamp',
+			'poggers',
+			'satono0pog',
+		],
+
+		'run' => [
+			'dennsenboots',
+			'lodanzrun',
+			'ticknaldosprint',
+			'vinter0run',
+		],
+
+		'wave' => [
+			'dennsenhi',
+			'dergoawave',
+			'falcnwavehi',
+			'heyguys',
+			'holysm0heyguys',
+			'muftaahey',
+			'rei0wave',
+			'sayuri0wave',
+			'shindi0wave',
+			'svenkawave',
+			'wuschlwave',
+		],
+
+		'zb' => [
+			'aga(hnim)?',
+			'armos( knights)?',
+			'arrghus',
+			'blind',
+			'ganon(dorf)?',
+			'helma',
+			'kholdstare',
+			'lanmo(las)?',
+			'moldorm',
+			'mothula',
+			'mott[ei]',
+			'trinexx',
+			'vit(reous|ty)',
+		],
+
+		'zd' => [
+			'eastern',
+			'desert( palace)?',
+			'gt',
+			'hera',
+			'ice ?(palace)?',
+			'(misery )?mire',
+			'pod',
+			'skull ?woods',
+			'swamp',
+			'thieve\'?s\'? ?town',
+			'tr',
+			'tt',
+		],
+
+		'zi' => [
+			'(big|small|retro|generic) ?keys?',
+			'b[oö]gen',
+			'bombos',
+			'boots',
+			'bottle',
+			'bows?',
+			'bugnet',
+			'byrna',
+			'cape',
+			'ether',
+			'flasche',
+			'flippers',
+			'fl[uö]te',
+			'frod',
+			'(gloves?|mitts|handschuhe?)',
+			'(half|quarter) ?magic',
+			'hammer',
+			'hookshot',
+			'(ice|fire) ?rod',
+			'lampe?',
+			'laser ?bridge',
+			'mearl',
+			'mirror',
+			'moon ?pearl',
+			'mushroom',
+			'ocarina',
+			'pilz',
+			'powder',
+			'puder',
+			'quake',
+			'(red|blue) ?cane',
+			'(red|green|blue) ?(goo|potion)',
+			'(red|green|blue|baby) ?mail',
+			'(red|blue|bu|boo|good|bad|both)merang',
+			'schaufel',
+			'(gro(ss|ß)er? |kleiner? )?schlüssel',
+			'schwert',
+			'shovel',
+			'silvers',
+			'somaria',
+			'spiegel',
+			'sword',
+		],
+
+		'zl' => [
+			'big chest',
+			'bumper( cave)?( ledge)?',
+			'(hyrule)? ?castle ?(tower)?',
+			'catfish',
+			'cave 0?',
+			'chest ?game',
+			'cutscene ?chest',
+			'damm',
+			'desert( ledge)?',
+			'dig(ging)? ?game',
+			'((back|front) of )?escape',
+			'gyl',
+			'hobo',
+			'hook ?(shot) cave',
+			'lava ?chest',
+			'(light|dark) ?world',
+			'lss',
+			'magic bat',
+			'(dark )?(death )?mountain',
+			'ped(estal)?',
+			'pyramid( fairy)?( ledge)?',
+			'red bomb',
+			'sahasrahla',
+			'sasha',
+			'sick kid',
+			'stumpy',
+			'tile ?room',
+			'torch',
+			'zora( ledge)?',
+		],
+	];
 
 }