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($msg) {
  10                 $tokens = $this->tokenize($msg->text_content);
  11                 if (empty($tokens)) return;
  12                 $tokens[] = '';
  13                 foreach ($tokens as $num => $token) {
  14                         if ($num === 0) {
  15                                 $this->addTransition([], $token);
  16                         } else {
  17                                 $start = max(0, $num - $this->size);
  18                                 $end = $num;
  19                                 for ($i = $start; $i < $end; ++$i) {
  20                                         $this->addTransition(array_slice($tokens, $i, $end - $i), $token);
  21                                         if ($end - $i < 3) break;
  22                                 }
  23                         }
  24                 }
  25         }
  26
  27         public function compile() {
  28                 foreach ($this->transitions as $key => $value) {
  29                         $this->transitions[$key] = $this->index($this->transitions[$key]);
  30                         if (empty($this->transitions[$key])) {
  31                                 unset($this->transitions[$key]);
  32                         }
  33                 }
  34         }
  35
  36         public function generate($limit = 100) {
  37                 $tokens = [];
  38                 $generated = '';
  39                 while (strlen($generated) < $limit) {
  40                         $next = $this->randomNext($tokens);
  41                         if (empty($next)) break;
  42                         $tokens[] = $next;
  43                         $generated .= $next;
  44                 }
  45                 return $generated;
  46         }
  47
  48         public function saveAs($name) {
  49                 $data = [
  50                         'size' => $this->size,
  51                         'transitions' => $this->transitions,
  52                 ];
  53                 Storage::disk('chatlib')->put($name.'.json', json_encode($data));
  54         }
  55
  56         public function loadFrom($name) {
  57                 $data = json_decode(Storage::disk('chatlib')->get($name.'.json'), true);
  58                 $this->size = $data['size'];
  59                 $this->transitions = $data['transitions'];
  60         }
  61
  62         private function index($arr) {
  63                 $result = [];
  64                 $sum = 0;
  65                 foreach ($arr as $key => $entry) {
  66                         $weight = $entry['count'];
  67                         if ($weight == 1) continue;
  68                         $lower = $sum;
  69                         $sum += $weight;
  70                         $examples = [];
  71                         if (is_array(end($entry['examples']))) {
  72                                 // already processed
  73                                 $examples = $entry['examples'];
  74                         } else {
  75                                 $subsum = 0;
  76                                 foreach ($entry['examples'] as $example => $subweight) {
  77                                         $sublower = $subsum;
  78                                         $subsum += $subweight * $subweight;
  79                                         $examples[] = [$example, $sublower, $subsum];
  80                                 }
  81                         }
  82                         $result[] = [$key, $lower, $sum, $examples];
  83                 }
  84                 return $result;
  85         }
  86
  87         private function randomNext($tokens) {
  88                 $cnt = count($tokens);
  89                 for ($size = min($this->size, $cnt); $size >= 0; --$size) {
  90                         $cmb = $this->generalize(array_slice($tokens, $cnt - $size, $size));
  91                         if (isset($this->transitions[$cmb])) {
  92                                 $pick = $this->pick($this->transitions[$cmb]);
  93                                 if (!is_null($pick)) {
  94                                         return $this->exampleOf($pick);
  95                                 }
  96                         }
  97                 }
  98                 return '';
  99         }
 100
 101         private function pick($options) {
 102                 if (empty($options)) return null;
 103                 $max = end($options)[2];
 104                 $num = random_int(0, $max);
 105                 $min_index = 0;
 106                 $max_index = count($options) - 1;
 107                 while ($min_index < $max_index) {
 108                         $cur_index = intval(($min_index + $max_index) / 2);
 109                         $cur_low = $options[$cur_index][1];
 110                         $cur_high = $options[$cur_index][2];
 111                         if ($cur_low > $num) {
 112                                 $max_index = $cur_index;
 113                         } else if ($cur_high < $num) {
 114                                 $min_index = $cur_index + 1;
 115                         } else {
 116                                 $min_index = $cur_index;
 117                                 break;
 118                         }
 119                 }
 120                 return $options[$min_index];
 121         }
 122
 123         private function addTransition($state, $next) {
 124                 $cmb = $this->generalize($state);
 125                 if (!isset($this->transitions[$cmb])) {
 126                         $this->transitions[$cmb] = [];
 127                 }
 128                 $this->increment($this->transitions[$cmb], $next);
 129         }
 130
 131         private function increment(&$which, $token) {
 132                 $generalized = $this->generalize([$token]);
 133                 if (!isset($which[$generalized])) {
 134                         $which[$generalized] = [
 135                                 'count' => 1,
 136                                 'examples' => [],
 137                         ];
 138                         $which[$generalized]['examples'][$token] = 1;
 139                 } else {
 140                         ++$which[$generalized]['count'];
 141                         if (!isset($which[$generalized]['examples'][$token])) {
 142                                 $which[$generalized]['examples'][$token] = 1;
 143                         } else {
 144                                 ++$which[$generalized]['examples'][$token];
 145                         }
 146                 }
 147         }
 148
 149         private function tokenize($str) {
 150                 return array_values(array_filter(preg_split('/\b/u', $str), function($token) {
 151                         if (empty($token)) return false;
 152                         if (preg_match('/cheer\d+/u', strtolower($token))) return false;
 153                         return true;
 154                 }));
 155         }
 156
 157         private function generalize($tokens) {
 158                 $str = '';
 159                 foreach ($tokens as $token) {
 160                         $replaced = preg_replace('/\d+/', '0', $token);
 161                         $replaced = strtolower($replaced);
 162                         $str .= empty($replaced) ? $token : $replaced;
 163                 }
 164                 return $str;
 165         }
 166
 167         private function exampleOf($pick) {
 168                 $example = $this->pick($pick[3]);
 169                 return $example[0];
 170         }
 171
 172         private $size = 5;
 173         private $transitions = [];
 174
 175 }