Esempio n. 1
0
 /**
  * Static method __cutDAG
  *
  * @param string $sentence # input sentence
  * @param array  $options  # other options
  *
  * @return array $words
  */
 public static function __cutDAG($sentence, $options = array())
 {
     $defaults = array('mode' => 'default');
     $options = array_merge($defaults, $options);
     $words = array();
     $N = mb_strlen($sentence, 'UTF-8');
     $DAG = Jieba::getDAG($sentence);
     Jieba::calc($sentence, $DAG);
     $x = 0;
     $buf = '';
     while ($x < $N) {
         $current_route_keys = array_keys(Jieba::$route[$x]);
         $y = $current_route_keys[0] + 1;
         $l_word = mb_substr($sentence, $x, $y - $x, 'UTF-8');
         if ($y - $x == 1) {
             $buf = $buf . $l_word;
         } else {
             if (mb_strlen($buf, 'UTF-8') > 0) {
                 if (mb_strlen($buf, 'UTF-8') == 1) {
                     if (isset(self::$word_tag[$buf])) {
                         $buf_tag = self::$word_tag[$buf];
                     } else {
                         $buf_tag = "x";
                     }
                     array_push($words, array("word" => $buf, "tag" => $buf_tag));
                     $buf = '';
                 } else {
                     $regognized = self::__cutDetail($buf);
                     foreach ($regognized as $key => $word) {
                         array_push($words, $word);
                     }
                     $buf = '';
                 }
             }
             if (isset(self::$word_tag[$l_word])) {
                 $buf_tag = self::$word_tag[$l_word];
             } else {
                 $buf_tag = "x";
             }
             array_push($words, array("word" => $l_word, "tag" => $buf_tag));
         }
         $x = $y;
     }
     if (mb_strlen($buf, 'UTF-8') > 0) {
         if (mb_strlen($buf, 'UTF-8') == 1) {
             if (isset(self::$word_tag[$buf])) {
                 $buf_tag = self::$word_tag[$buf];
             } else {
                 $buf_tag = "x";
             }
             array_push($words, array("word" => $buf, "tag" => $buf_tag));
         } else {
             $regognized = self::__cutDetail($buf);
             foreach ($regognized as $key => $word) {
                 array_push($words, $word);
             }
         }
     }
     return $words;
 }