[Summary view]

   1  <?php
   2  /**
   3   * Copyright 2010-2017 Horde LLC (http://www.horde.org/)
   4   *
   5   * @author   Chuck Hagenbuch <chuck@horde.org>
   6   * @license  http://www.horde.org/licenses/bsd BSD
   7   * @category Horde
   8   * @package  Support
   9   */
  10  
  11  /**
  12   * @author   Chuck Hagenbuch <chuck@horde.org>
  13   * @license  http://www.horde.org/licenses/bsd BSD
  14   * @category Horde
  15   * @package  Support
  16   */
  17  class Horde_Support_Numerizer_Locale_Base
  18  {
  19      public $DIRECT_NUMS = array(
  20          'eleven' => '11',
  21          'twelve' => '12',
  22          'thirteen' => '13',
  23          'fourteen' => '14',
  24          'fifteen' => '15',
  25          'sixteen' => '16',
  26          'seventeen' => '17',
  27          'eighteen' => '18',
  28          'nineteen' => '19',
  29          'ninteen' => '19',      // Common mis-spelling
  30          'zero' => '0',
  31          'one' => '1',
  32          'two' => '2',
  33          'three' => '3',
  34          'four(\W|$)' => '4$1',  // The weird regex is so that it matches four but not fourty
  35          'five' => '5',
  36          'six(\W|$)' => '6$1',
  37          'seven(\W|$)' => '7$1',
  38          'eight(\W|$)' => '8$1',
  39          'nine(\W|$)' => '9$1',
  40          'ten' => '10',
  41          '\ba[\b^$]' => '1',     // doesn't make sense for an 'a' at the end to be a 1
  42      );
  43  
  44      public $TEN_PREFIXES = array(
  45          'twenty' => 20,
  46          'thirty' => 30,
  47          'forty' => 40,
  48          'fourty' => 40, // Common mis-spelling
  49          'fifty' => 50,
  50          'sixty' => 60,
  51          'seventy' => 70,
  52          'eighty' => 80,
  53          'ninety' => 90,
  54          'ninty' => 90, // Common mis-spelling
  55      );
  56  
  57      public $BIG_PREFIXES = array(
  58          'hundred' => 100,
  59          'thousand' => 1000,
  60          'million' => 1000000,
  61          'billion' => 1000000000,
  62          'trillion' => 1000000000000,
  63      );
  64  
  65      public function numerize($string)
  66      {
  67          // preprocess
  68          $string = $this->_splitHyphenatedWords($string);
  69          $string = $this->_hideAHalf($string);
  70  
  71          $string = $this->_directReplacements($string);
  72          $string = $this->_replaceTenPrefixes($string);
  73          $string = $this->_replaceBigPrefixes($string);
  74          $string = $this->_fractionalAddition($string);
  75  
  76          return $string;
  77      }
  78  
  79      /**
  80       * will mutilate hyphenated-words but shouldn't matter for date extraction
  81       */
  82      protected function _splitHyphenatedWords($string)
  83      {
  84          return preg_replace('/ +|([^\d])-([^d])/', '$1 $2', $string);
  85      }
  86  
  87      /**
  88       * take the 'a' out so it doesn't turn into a 1, save the half for the end
  89       */
  90      protected function _hideAHalf($string)
  91      {
  92          return str_replace('a half', 'haAlf', $string);
  93      }
  94  
  95      /**
  96       * easy/direct replacements
  97       */
  98      protected function _directReplacements($string)
  99      {
 100          foreach ($this->DIRECT_NUMS as $dn => $dn_replacement) {
 101              $string = preg_replace("/$dn/i", $dn_replacement, $string);
 102          }
 103          return $string;
 104      }
 105  
 106      /**
 107       * ten, twenty, etc.
 108       */
 109      protected function _replaceTenPrefixes($string)
 110      {
 111          foreach ($this->TEN_PREFIXES as $tp => $tp_replacement) {
 112              $string = preg_replace_callback(
 113                  "/(?:$tp)( *\d(?=[^\d]|\$))*/i",
 114                  function ($m) use ($tp_replacement) {
 115                      return $tp_replacement + (isset($m[1]) ? (int)$m[1] : 0);
 116                  },
 117                  $string
 118              );
 119          }
 120          return $string;
 121      }
 122  
 123      /**
 124       * hundreds, thousands, millions, etc.
 125       */
 126      protected function _replaceBigPrefixes($string)
 127      {
 128          foreach ($this->BIG_PREFIXES as $bp => $bp_replacement) {
 129              $string = preg_replace_callback(
 130                  '/(\d*) *' . $bp . '/i',
 131                  function ($m) use ($bp_replacement) {
 132                      return $bp_replacement * (int)$m[1];
 133                  },
 134                  $string
 135              );
 136              $string = $this->_andition($string);
 137          }
 138          return $string;
 139      }
 140  
 141      protected function _andition($string)
 142      {
 143          while (true) {
 144              if (preg_match('/(\d+)( | and )(\d+)(?=[^\w]|$)/i', $string, $sc, PREG_OFFSET_CAPTURE)) {
 145                  if (preg_match('/and/', $sc[2][0]) || (strlen($sc[1][0]) > strlen($sc[3][0]))) {
 146                      $string = substr($string, 0, $sc[1][1]) . ((int)$sc[1][0] + (int)$sc[3][0]) . substr($string, $sc[3][1] + strlen($sc[3][0]));
 147                      continue;
 148                  }
 149              }
 150              break;
 151          }
 152          return $string;
 153      }
 154  
 155      protected function _fractionalAddition($string)
 156      {
 157          return preg_replace_callback(
 158              '/(\d+)(?: | and |-)*haAlf/i',
 159              function ($m) {
 160                  return (string)((float)$m[1] + 0.5);
 161              },
 162              $string
 163          );
 164      }
 165  
 166  }