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