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 - $i < 4) break;
  26                                 }
  27                         }
  28                 }
  29         }
  30
  31         public function compile() {
  32                 foreach ($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 {
  79                                 $subsum = 0;
  80                                 foreach ($entry['examples'] as $example => $subweight) {
  81                                         $sublower = $subsum;
  82                                         $subsum += $subweight;
  83                                         $examples[] = [$example, $sublower, $subsum];
  84                                 }
  85                         }
  86                         $result[] = [$key, $lower, $sum, $examples];
  87                 }
  88                 return $result;
  89         }
  90
  91         private function randomNext($tokens) {
  92                 $cnt = count($tokens);
  93                 for ($size = min($this->size, $cnt); $size > 0; --$size) {
  94                         $cmb = $this->generalize(array_slice($tokens, -$size));
  95                         if (isset($this->transitions[$cmb])) {
  96                                 $pick = $this->pick($this->transitions[$cmb]);
  97                                 if (!is_null($pick)) {
  98                                         return $this->exampleOf($pick);
  99                                 }
 100                         }
 101                 }
 102                 return '';
 103         }
 104
 105         private function pick($options) {
 106                 if (empty($options)) return null;
 107                 $max = end($options)[2];
 108                 $num = random_int(0, $max);
 109                 $min_index = 0;
 110                 $max_index = count($options) - 1;
 111                 while ($min_index < $max_index) {
 112                         $cur_index = intval(($min_index + $max_index) / 2);
 113                         $cur_low = $options[$cur_index][1];
 114                         $cur_high = $options[$cur_index][2];
 115                         if ($cur_low > $num) {
 116                                 $max_index = $cur_index;
 117                         } else if ($cur_high < $num) {
 118                                 $min_index = $cur_index + 1;
 119                         } else {
 120                                 $min_index = $cur_index;
 121                                 break;
 122                         }
 123                 }
 124                 return $options[$min_index];
 125         }
 126
 127         private function addTransition($state, $next) {
 128                 $cmb = $this->generalize($state);
 129                 if (!isset($this->transitions[$cmb])) {
 130                         $this->transitions[$cmb] = [];
 131                 }
 132                 $this->increment($this->transitions[$cmb], $next);
 133         }
 134
 135         private function increment(&$which, $token) {
 136                 $generalized = $this->generalize([$token]);
 137                 if (!isset($which[$generalized])) {
 138                         $which[$generalized] = [
 139                                 'count' => 1,
 140                                 'examples' => [],
 141                         ];
 142                         $which[$generalized]['examples'][$token] = 1;
 143                 } else {
 144                         ++$which[$generalized]['count'];
 145                         if (!isset($which[$generalized]['examples'][$token])) {
 146                                 $which[$generalized]['examples'][$token] = 1;
 147                         } else {
 148                                 ++$which[$generalized]['examples'][$token];
 149                         }
 150                 }
 151         }
 152
 153         private function tokenize($str) {
 154                 return array_values(array_filter(preg_split('/\b/u', $str), function($token) {
 155                         if (empty($token)) return false;
 156                         if (preg_match('/cheer\d+/u', strtolower($token))) return false;
 157                         return true;
 158                 }));
 159         }
 160
 161         private function generalize($tokens) {
 162                 $str = '';
 163                 foreach ($tokens as $token) {
 164                         $replaced = preg_replace('/\d+/', '0', $token);
 165                         $replaced = strtolower($replaced);
 166                         $str .= $replaced;
 167                 }
 168                 return $str;
 169         }
 170
 171         private function exampleOf($pick) {
 172                 $example = $this->pick($pick[3]);
 173                 return $example[0];
 174         }
 175
 176         private $size = 5;
 177         private $transitions = [];
 178
 179 }