danielebarchiesi@6: <?php
danielebarchiesi@6: /**
danielebarchiesi@6:  * Does cutting and matching stuff with a name string.
danielebarchiesi@6:  * Note that the string has to be UTF8-encoded.
danielebarchiesi@6:  *
danielebarchiesi@6:  */
danielebarchiesi@6: class HumanNameParser_Name {
danielebarchiesi@6:   private $str;
danielebarchiesi@6: 
danielebarchiesi@6:   function __construct($str)
danielebarchiesi@6:   {
danielebarchiesi@6:     $this->setStr($str);
danielebarchiesi@6:   }
danielebarchiesi@6: 
danielebarchiesi@6:   /**
danielebarchiesi@6:    * Checks encoding, normalizes whitespace/punctuation, and sets the name string.
danielebarchiesi@6:    *
danielebarchiesi@6:    * @param String $str a utf8-encoding string.
danielebarchiesi@6:    * @return Bool True on success
danielebarchiesi@6:    */
danielebarchiesi@6:   public function setStr($str)
danielebarchiesi@6:   {
danielebarchiesi@6:     if (!drupal_validate_utf8($str)){
danielebarchiesi@6:       throw new Exception("Name is not encoded in UTF-8");
danielebarchiesi@6:     }
danielebarchiesi@6:     $this->str = $str;
danielebarchiesi@6:     $this->norm();
danielebarchiesi@6:     return true;
danielebarchiesi@6:   }
danielebarchiesi@6: 
danielebarchiesi@6:   public function getStr()
danielebarchiesi@6:   {
danielebarchiesi@6:     return $this->str;
danielebarchiesi@6:   }
danielebarchiesi@6: 
danielebarchiesi@6: 
danielebarchiesi@6:   /**
danielebarchiesi@6:    * Uses a regex to chop off and return part of the namestring
danielebarchiesi@6:    * There are two parts: first, it returns the matched substring,
danielebarchiesi@6:    * and then it removes that substring from $this->str and normalizes.
danielebarchiesi@6:    *
danielebarchiesi@6:    * @param string $regex matches the part of the namestring to chop off
danielebarchiesi@6:    * @param integer $submatchIndex	which of the parenthesized submatches to use
danielebarchiesi@6:    * @param string	$regexFlags	optional regex flags
danielebarchiesi@6:    * @return string	the part of the namestring that got chopped off
danielebarchiesi@6:    */
danielebarchiesi@6:   public function chopWithRegex($regex, $submatchIndex = 0, $regexFlags = '')
danielebarchiesi@6:   {
danielebarchiesi@6:     $regex = $regex . "ui" . $regexFlags; // unicode + case-insensitive
danielebarchiesi@6:     preg_match($regex, $this->str, $m);
danielebarchiesi@6:     $subset = (isset($m[$submatchIndex])) ? $m[$submatchIndex] : '';
danielebarchiesi@6: 
danielebarchiesi@6:     if ($subset){
danielebarchiesi@6:       $this->str = preg_replace($regex, ' ', $this->str, -1, $numReplacements);
danielebarchiesi@6:       if ($numReplacements > 1){
danielebarchiesi@6:         throw new Exception("The regex being used to find the name: '$this->str' has multiple matches.");
danielebarchiesi@6:       }
danielebarchiesi@6:       $this->norm();
danielebarchiesi@6:       return $subset;
danielebarchiesi@6:     }
danielebarchiesi@6:     else {
danielebarchiesi@6:       return '';
danielebarchiesi@6:     }
danielebarchiesi@6:   }
danielebarchiesi@6: 
danielebarchiesi@6:   /*
danielebarchiesi@6:    * Flips the front and back parts of a name with one another.
danielebarchiesi@6:   * Front and back are determined by a specified character somewhere in the
danielebarchiesi@6:   * middle of the string.
danielebarchiesi@6:   *
danielebarchiesi@6:   * @param	String $flipAroundChar	the character(s) demarcating the two halves you want to flip.
danielebarchiesi@6:   * @return Bool True on success.
danielebarchiesi@6:   */
danielebarchiesi@6:   public function flip($flipAroundChar)
danielebarchiesi@6:   {
danielebarchiesi@6:     $substrings = preg_split("/$flipAroundChar/u", $this->str);
danielebarchiesi@6:     if (count($substrings) == 2){
danielebarchiesi@6:       $this->str = $substrings[1] . " " . $substrings[0];
danielebarchiesi@6:       $this->norm();
danielebarchiesi@6:     }
danielebarchiesi@6:     else if (count($substrings) > 2) {
danielebarchiesi@6:       throw new Exception("Can't flip around multiple '$flipAroundChar' characters in: '$this->str'.");
danielebarchiesi@6:     }
danielebarchiesi@6:     return true; // if there's 1 or 0 $flipAroundChar found
danielebarchiesi@6:   }
danielebarchiesi@6: 
danielebarchiesi@6:   /**
danielebarchiesi@6:    * Removes extra whitespace and punctuation from $this->str
danielebarchiesi@6:    * Strips whitespace chars from ends, strips redundant whitespace, converts whitespace chars to " ".
danielebarchiesi@6:    *
danielebarchiesi@6:    * @return Bool True on success
danielebarchiesi@6:    */
danielebarchiesi@6:   private function norm()
danielebarchiesi@6:   {
danielebarchiesi@6:     $this->str = preg_replace( "#^\s*#u", "", $this->str );
danielebarchiesi@6:     $this->str = preg_replace( "#\s*$#u", "", $this->str );
danielebarchiesi@6:     $this->str = preg_replace( "#\s+#u", " ", $this->str );
danielebarchiesi@6:     $this->str = preg_replace( "#,$#u", " ", $this->str );
danielebarchiesi@6:     return true;
danielebarchiesi@6:   }
danielebarchiesi@6: }
danielebarchiesi@6: ?>