app/Models/ChatLib.php

   1 <?php
   2
   3 namespace App\Models;
   4
   5 use Illuminate\Support\Facades\Storage;
   6
   7 class ChatLib {
   8
   9         public function __construct($size = 7) {
  10                 $this->size = $size;
  11
  12                 $converted = [];
  13                 foreach ($this->categories as $category => $patterns) {
  14                         $converted_patterns = [];
  15                         foreach ($patterns as $pattern) {
  16                                 $converted_patterns[] = '/\b'.$pattern.'\b/u';
  17                         }
  18                         $converted['%'.strtoupper($category).'%'] = $converted_patterns;
  19                 }
  20                 $this->categories = $converted;
  21         }
  22
  23         public function addMessage(ChatLog $msg) {
  24                 $this->addText($msg->text_content);
  25         }
  26
  27         public function addText($text) {
  28                 $tokens = $this->tokenize($text);
  29                 if (empty($tokens)) return;
  30                 $tokens[] = '';
  31                 foreach ($tokens as $num => $token) {
  32                         if ($num === 0) {
  33                                 $this->addTransition([], $token);
  34                         } else {
  35                                 $start = max(0, $num - $this->size - 1);
  36                                 $end = $num;
  37                                 for ($i = $start; $i < $end; ++$i) {
  38                                         $this->addTransition(array_slice($tokens, $i, $end - $i), $token);
  39                                         if ($end - $i < 5) break;
  40                                 }
  41                         }
  42                 }
  43         }
  44
  45         public function compile() {
  46                 foreach ($this->transitions as $key => $value) {
  47                         $this->transitions[$key] = $this->index($this->transitions[$key]);
  48                         if (empty($this->transitions[$key])) {
  49                                 unset($this->transitions[$key]);
  50                         }
  51                 }
  52         }
  53
  54         public function generate($limit = 100) {
  55                 $tokens = [''];
  56                 $generated = '';
  57                 while (strlen($generated) < $limit) {
  58                         $next = $this->randomNext($tokens);
  59                         if ($next === '') break;
  60                         $tokens[] = $next;
  61                         $generated .= $next;
  62                 }
  63                 return $generated;
  64         }
  65
  66         public function saveAs($name) {
  67                 $data = [
  68                         'size' => $this->size,
  69                         'transitions' => $this->transitions,
  70                 ];
  71                 Storage::disk('chatlib')->put($name.'.json', json_encode($data));
  72         }
  73
  74         public function loadFrom($name) {
  75                 $data = json_decode(Storage::disk('chatlib')->get($name.'.json'), true);
  76                 $this->size = $data['size'];
  77                 $this->transitions = $data['transitions'];
  78         }
  79
  80         private function index($arr) {
  81                 $result = [];
  82                 $sum = 0;
  83                 foreach ($arr as $key => $entry) {
  84                         $weight = $entry[0];
  85                         if ($weight == 1) continue;
  86                         $lower = $sum;
  87                         $sum += $weight;
  88                         $examples = [];
  89                         if ($key === ' ') {
  90                                 $examples = [[' ', 0, 1]];
  91                         } else {
  92                                 $subsum = 0;
  93                                 foreach ($entry[1] as $example => $subweight) {
  94                                         $sublower = $subsum;
  95                                         $subsum += $subweight;
  96                                         $examples[] = [$example, $sublower, $subsum];
  97                                 }
  98                         }
  99                         $result[] = [$key, $lower, $sum, $examples];
 100                 }
 101                 return $result;
 102         }
 103
 104         private function randomNext($tokens) {
 105                 $cnt = count($tokens);
 106                 for ($size = min($this->size, $cnt); $size > 0; --$size) {
 107                         $cmb = $this->generalize(array_slice($tokens, -$size));
 108                         if (isset($this->transitions[$cmb])) {
 109                                 $pick = $this->pick($this->transitions[$cmb]);
 110                                 if (!is_null($pick)) {
 111                                         return $this->exampleOf($pick);
 112                                 }
 113                         }
 114                 }
 115                 return '';
 116         }
 117
 118         private function pick($options) {
 119                 if (empty($options)) return null;
 120                 $max = end($options)[2] - 1;
 121                 $num = random_int(0, $max);
 122                 return static::search($options, $num);
 123         }
 124
 125         public static function search($options, $num) {
 126                 $min_index = 0;
 127                 $max_index = count($options) - 1;
 128                 while ($min_index < $max_index) {
 129                         $cur_index = intval(($min_index + $max_index) / 2);
 130                         $cur_low = $options[$cur_index][1];
 131                         $cur_high = $options[$cur_index][2] - 1;
 132                         if ($cur_low > $num) {
 133                                 $max_index = $cur_index;
 134                         } else if ($cur_high < $num) {
 135                                 $min_index = $cur_index + 1;
 136                         } else {
 137                                 $min_index = $cur_index;
 138                                 break;
 139                         }
 140                 }
 141                 return $options[$min_index];
 142         }
 143
 144         private function addTransition($state, $next) {
 145                 $cmb = $this->generalize($state);
 146                 if (!isset($this->transitions[$cmb])) {
 147                         $this->transitions[$cmb] = [];
 148                 }
 149                 $this->increment($this->transitions[$cmb], $next);
 150         }
 151
 152         private function increment(&$which, $token) {
 153                 $generalized = $this->generalize([$token]);
 154                 if (!isset($which[$generalized])) {
 155                         $which[$generalized] = [
 156                                 1,
 157                                 [],
 158                         ];
 159                         $which[$generalized][1][$token] = 1;
 160                 } else {
 161                         ++$which[$generalized][0];
 162                         if (!isset($which[$generalized][1][$token])) {
 163                                 $which[$generalized][1][$token] = 1;
 164                         } else {
 165                                 ++$which[$generalized][1][$token];
 166                         }
 167                 }
 168         }
 169
 170         private function tokenize($str) {
 171                 return array_values(array_filter(preg_split('/\b/u', $str), function($token) {
 172                         if ($token === '') return false;
 173                         if (preg_match('/cheer\d+/u', strtolower($token))) return false;
 174                         return true;
 175                 }));
 176         }
 177
 178         private function generalize($tokens) {
 179                 $str = '';
 180                 foreach ($tokens as $token) {
 181                         $replaced = preg_replace('/\d+/u', '0', $token);
 182                         $replaced = preg_replace('/\s+/u', ' ', $replaced);
 183                         $replaced = preg_replace('/(.)\1{2,}/u', '$1$1', $replaced);
 184                         $replaced = strtolower($replaced);
 185                         foreach ($this->aliases as $canonical => $variants) {
 186                                 if (in_array($replaced, $variants)) {
 187                                         $replaced = $canonical;
 188                                         break;
 189                                 }
 190                                 if ($replaced === $canonical) {
 191                                         break;
 192                                 }
 193                         }
 194                         $str .= $replaced;
 195                 }
 196                 foreach ($this->categories as $category => $patterns) {
 197                         $str = preg_replace($patterns, $category, $str);
 198                 }
 199                 return $str;
 200         }
 201
 202         private function exampleOf($pick) {
 203                 $example = $this->pick($pick[3]);
 204                 return $example[0];
 205         }
 206
 207         private $size = 7;
 208         private $transitions = [];
 209
 210         private $aliases = [
 211                 'chest' => ['kiste'],
 212                 'einen' => ['n', 'nen'],
 213                 'musik' => ['mukke'],
 214                 'schade' => ['schad', 'schaade'],
 215         ];
 216
 217         private $categories = [
 218                 'fail' => [
 219                         'failfish',
 220                         'holysm0notlikethis',
 221                         'notlikethis',
 222                         'tetobridge0',
 223                         'vinter0clown',
 224                 ],
 225
 226                 'hype' => [
 227                         'dergoaparty',
 228                         'dinodance',
 229                         'elemen0party',
 230                         'muftaahype',
 231                         'luckwuhype',
 232                         'olliwahype',
 233                         'osora0umbrihype',
 234                         'partyhat',
 235                         'peepocheer',
 236                         'rei0hype',
 237                         'sakayahype',
 238                         'tetotroete',
 239                         'ticknaboargeil0',
 240                         'ticknahype0',
 241                 ],
 242
 243                 'kappa' => [
 244                         'kappa(claus|hd)?',
 245                 ],
 246
 247                 'jam' => [
 248                         '(cat|dog|rat)jam',
 249                         'kanash0jam',
 250                         'rei0jamers',
 251                         'samusdance',
 252                 ],
 253
 254                 'lol' => [
 255                         ':d',
 256                         'boothi0lul',
 257                         'kekw',
 258                         'lol',
 259                         'lul',
 260                         'rei0lul',
 261                         'samusgrin',
 262                         'ticknaauslachen',
 263                         'xd',
 264                 ],
 265
 266                 'love' => [
 267                         '<3',
 268                         'duden0love',
 269                         'exec0love',
 270                         'krawal0heart',
 271                         'lodanzhug',
 272                         'luckwulove',
 273                         'luvsign',
 274                         'muftaal',
 275                         'osora0love',
 276                         'peepoexcitedhug',
 277                         'spirit0love',
 278                         'svenkalove',
 279                         'ticknaherz',
 280                 ],
 281
 282                 'name' => [
 283                         'baba',
 284                         'baka',
 285                         'bobe?r',
 286                         'brog(i|or)',
 287                         'cfate',
 288                         'danny',
 289                         'danzi+',
 290                         'daruck',
 291                         'dennsen',
 292                         'dimez',
 293                         'divi',
 294                         'dud(en|i+)',
 295                         'ele',
 296                         'eri(ror)?',
 297                         '(name)?faker',
 298                         'fetti+',
 299                         'gamma(chuu)?',
 300                         'goat(buster|ie?|y)?',
 301                         'hitsu(yan)?',
 302                         'holy',
 303                         'jem',
 304                         'kala(marino)?',
 305                         'kromb',
 306                         'koval',
 307                         'kum(i|o|p)',
 308                         'lanux',
 309                         'len(esha|chen)',
 310                         'leya+',
 311                         'magno',
 312                         'malmo',
 313                         'markam',
 314                         'micha',
 315                         'mimsy',
 316                         'muf(fy|taay)',
 317                         'murd(elizer|i+)',
 318                         'nami',
 319                         'nula',
 320                         'onio',
 321                         'paulinche',
 322                         'phaaze',
 323                         'ralen',
 324                         'ramond',
 325                         'ray(vis)?',
 326                         'schulzer',
 327                         'skunk(ner)?',
 328                         'skipsy',
 329                         'soli+',
 330                         'sven(ka+)?',
 331                         'tantalus',
 332                         'teto',
 333                         'thalanee?',
 334                         'tick(i+|naldo|y+)',
 335                         'tofu',
 336                         'tr[i0]x+',
 337                         'vin(nie?|ny|ter)',
 338                         'xall',
 339                         'yasi',
 340                 ],
 341
 342                 'pog' => [
 343                         'bumble0Pog',
 344                         'komodohype',
 345                         'pog',
 346                         'pogchamp',
 347                         'poggers',
 348                         'satono0pog',
 349                 ],
 350
 351                 'run' => [
 352                         'dennsenboots',
 353                         'lodanzrun',
 354                         'ticknaldosprint',
 355                         'vinter0run',
 356                 ],
 357
 358                 'wave' => [
 359                         'dennsenhi',
 360                         'dergoawave',
 361                         'heyguys',
 362                         'holysm0heyguys',
 363                         'muftaahey',
 364                         'rei0wave',
 365                         'sayuri0wave',
 366                         'shindi0wave',
 367                         'svenkawave',
 368                         'wuschlwave',
 369                 ],
 370
 371                 'zb' => [
 372                         'aga(hnim)?',
 373                         'armos( knights)?',
 374                         'arrghus',
 375                         'blind',
 376                         'ganon(dorf)?',
 377                         'helma',
 378                         'kholdstare',
 379                         'lanmo(las)?',
 380                         'moldorm',
 381                         'mothula',
 382                         'mott[ei]',
 383                         'trinexx',
 384                         'vit(reous|ty)',
 385                 ],
 386
 387                 'zd' => [
 388                         'eastern',
 389                         'desert( palace)?',
 390                         'gt',
 391                         'hera',
 392                         'ice ?(palace)?',
 393                         '(misery )?mire',
 394                         'pod',
 395                         'skull ?woods',
 396                         'swamp',
 397                         'thieve\'?s\'? ?town',
 398                         'tr',
 399                         'tt',
 400                 ],
 401
 402                 'zi' => [
 403                         '(big|small|retro|generic) ?keys?',
 404                         'b[oö]gen',
 405                         'bombos',
 406                         'boots',
 407                         'bottle',
 408                         'bows?',
 409                         'bugnet',
 410                         'byrna',
 411                         'cape',
 412                         'ether',
 413                         'flasche',
 414                         'flippers',
 415                         'fl[uö]te',
 416                         'frod',
 417                         '(gloves?|mitts|handschuhe?)',
 418                         '(half|quarter) ?magic',
 419                         'hammer',
 420                         'hookshot',
 421                         '(ice|fire) ?rod',
 422                         'lampe?',
 423                         'laser ?bridge',
 424                         'mearl',
 425                         'mirror',
 426                         'moon ?pearl',
 427                         'mushroom',
 428                         'ocarina',
 429                         'pilz',
 430                         'powder',
 431                         'puder',
 432                         'quake',
 433                         '(red|blue) ?cane',
 434                         '(red|green|blue) ?(goo|potion)',
 435                         '(red|green|blue|baby) ?mail',
 436                         '(red|blue|bu|boo|good|bad|both)merang',
 437                         'schaufel',
 438                         '(gro(ss|ß)er? |kleiner? )?schlüssel',
 439                         'schwert',
 440                         'shovel',
 441                         'silvers',
 442                         'somaria',
 443                         'spiegel',
 444                         'sword',
 445                 ],
 446
 447                 'zl' => [
 448                         'big chest',
 449                         'bumper( cave)?( ledge)?',
 450                         '(hyrule)? ?castle ?(tower)?',
 451                         'catfish',
 452                         'cave 0?',
 453                         'chest ?game',
 454                         'cutscene ?chest',
 455                         'damm',
 456                         'desert( ledge)?',
 457                         'dig(ging)? ?game',
 458                         '((back|front) of )?escape',
 459                         'gyl',
 460                         'hobo',
 461                         'hook ?(shot) cave',
 462                         'lava ?chest',
 463                         '(light|dark) ?world',
 464                         'lss',
 465                         'magic bat',
 466                         '(dark )?(death )?mountain',
 467                         'ped(estal)?',
 468                         'pyramid( fairy)?( ledge)?',
 469                         'red bomb',
 470                         'sahasrahla',
 471                         'sasha',
 472                         'sick kid',
 473                         'stumpy',
 474                         'tile ?room',
 475                         'torch',
 476                         'zora( ledge)?',
 477                 ],
 478         ];
 479
 480 }