interface/modules/zend_modules/library/Zend/Dom/Css2Xpath.php

   1 <?php
   2 /**
   3  * Zend Framework (http://framework.zend.com/)
   4  *
   5  * @link      http://github.com/zendframework/zf2 for the canonical source repository
   6  * @copyright Copyright (c) 2005-2013 Zend Technologies USA Inc. (http://www.zend.com)
   7  * @license   http://framework.zend.com/license/new-bsd New BSD License
   8  */
   9
  10 namespace Zend\Dom;
  11
  12 /**
  13  * Transform CSS selectors to XPath
  14  */
  15 class Css2Xpath
  16 {
  17     /**
  18      * Transform CSS expression to XPath
  19      *
  20      * @param  string $path
  21      * @return string
  22      */
  23     public static function transform($path)
  24     {
  25         $path = (string) $path;
  26         if (strstr($path, ',')) {
  27             $paths       = explode(',', $path);
  28             $expressions = array();
  29             foreach ($paths as $path) {
  30                 $xpath = self::transform(trim($path));
  31                 if (is_string($xpath)) {
  32                     $expressions[] = $xpath;
  33                 } elseif (is_array($xpath)) {
  34                     $expressions = array_merge($expressions, $xpath);
  35                 }
  36             }
  37             return implode('|', $expressions);
  38         }
  39
  40         $paths    = array('//');
  41         $path     = preg_replace('|\s+>\s+|', '>', $path);
  42         $segments = preg_split('/\s+/', $path);
  43         foreach ($segments as $key => $segment) {
  44             $pathSegment = static::_tokenize($segment);
  45             if (0 == $key) {
  46                 if (0 === strpos($pathSegment, '[contains(')) {
  47                     $paths[0] .= '*' . ltrim($pathSegment, '*');
  48                 } else {
  49                     $paths[0] .= $pathSegment;
  50                 }
  51                 continue;
  52             }
  53             if (0 === strpos($pathSegment, '[contains(')) {
  54                 foreach ($paths as $pathKey => $xpath) {
  55                     $paths[$pathKey] .= '//*' . ltrim($pathSegment, '*');
  56                     $paths[]      = $xpath . $pathSegment;
  57                 }
  58             } else {
  59                 foreach ($paths as $pathKey => $xpath) {
  60                     $paths[$pathKey] .= '//' . $pathSegment;
  61                 }
  62             }
  63         }
  64
  65         if (1 == count($paths)) {
  66             return $paths[0];
  67         }
  68         return implode('|', $paths);
  69     }
  70
  71     /**
  72      * Tokenize CSS expressions to XPath
  73      *
  74      * @param  string $expression
  75      * @return string
  76      */
  77     protected static function _tokenize($expression)
  78     {
  79         // Child selectors
  80         $expression = str_replace('>', '/', $expression);
  81
  82         // IDs
  83         $expression = preg_replace('|#([a-z][a-z0-9_-]*)|i', '[@id=\'$1\']', $expression);
  84         $expression = preg_replace('|(?<![a-z0-9_-])(\[@id=)|i', '*$1', $expression);
  85
  86         // arbitrary attribute strict equality
  87         $expression = preg_replace_callback(
  88             '|\[@?([a-z0-9_-]+)=[\'"]([^\'"]+)[\'"]\]|i',
  89             function ($matches) {
  90                 return '[@' . strtolower($matches[1]) . "='" . $matches[2] . "']";
  91             },
  92             $expression
  93         );
  94
  95         // arbitrary attribute contains full word
  96         $expression = preg_replace_callback(
  97             '|\[([a-z0-9_-]+)~=[\'"]([^\'"]+)[\'"]\]|i',
  98             function ($matches) {
  99                 return "[contains(concat(' ', normalize-space(@" . strtolower($matches[1]) . "), ' '), ' "
 100                      . $matches[2] . " ')]";
 101             },
 102             $expression
 103         );
 104
 105         // arbitrary attribute contains specified content
 106         $expression = preg_replace_callback(
 107             '|\[([a-z0-9_-]+)\*=[\'"]([^\'"]+)[\'"]\]|i',
 108             function ($matches) {
 109                 return "[contains(@" . strtolower($matches[1]) . ", '"
 110                      . $matches[2] . "')]";
 111             },
 112             $expression
 113         );
 114
 115         // Classes
 116         if(false === strpos($expression, "[@")) {
 117             $expression = preg_replace(
 118                 '|\.([a-z][a-z0-9_-]*)|i',
 119                 "[contains(concat(' ', normalize-space(@class), ' '), ' \$1 ')]",
 120                 $expression
 121             );
 122         }
 123
 124         /** ZF-9764 -- remove double asterisk */
 125         $expression = str_replace('**', '*', $expression);
 126
 127         return $expression;
 128     }
 129 }