1 <?php
 2 
 3 /**
 4  * CSSContentParser enables parsing & assertion running of HTML content via CSS selectors.
 5  * 
 6  * It works by converting the content to XHTML using tidy, rewriting the CSS selectors as XPath queries, and executing
 7  * those using SimpeXML.
 8  * 
 9  * It was built to facilitate testing using PHPUnit and contains a number of assert methods that will throw PHPUnit
10  * assertion exception when applicable.
11  * 
12  * Tries to use the PHP Tidy extension (http://php.net/tidy),
13  * and falls back to the "tidy" CLI tool. If none of those exists,
14  * the string is parsed directly without sanitization.
15  * 
16  * @package sapphire
17  * @subpackage core
18  */
19 class CSSContentParser extends Object {
20     protected $simpleXML = null;
21     
22     function __construct($content) {
23         if(extension_loaded('tidy')) {
24             // using the tiny php extension
25             $tidy = new Tidy();
26             $tidy->parseString(
27                 $content, 
28                 array(
29                     'output-xhtml' => true,
30                     'numeric-entities' => true,
31                 ), 
32                 'utf8'
33             );
34             $tidy->cleanRepair();
35             $tidy = str_replace('xmlns="http://www.w3.org/1999/xhtml"','',$tidy);
36             $tidy = str_replace('&#160;','',$tidy);
37 
38         } elseif(@shell_exec('which tidy')) {
39             // using tiny through cli
40             $CLI_content = escapeshellarg($content);
41             $tidy = `echo $CLI_content | tidy -n -q -utf8 -asxhtml 2> /dev/null`;
42             $tidy = str_replace('xmlns="http://www.w3.org/1999/xhtml"','',$tidy);
43             $tidy = str_replace('&#160;','',$tidy);
44         } else {
45             // no tidy library found, hence no sanitizing
46             $tidy = $content;
47         }
48         
49         $this->simpleXML = simplexml_load_string($tidy, 'SimpleXMLElement', LIBXML_NOWARNING);
50         
51         parent::__construct();
52     }
53         
54     /**
55      * Returns a number of SimpleXML elements that match the given CSS selector.
56      * Currently the selector engine only supports querying by tag, id, and classs
57      */
58     function getBySelector($selector) {
59         $xpath = $this->selector2xpath($selector);
60         return $this->simpleXML->xpath($xpath);
61     }
62     
63     /**
64      * Converts a CSS selector into an equivalent xpath expression.
65      * Currently the selector engine only supports querying by tag, id, and classs
66      */
67     function selector2xpath($selector) {
68         $parts = preg_split('/\\s+/', $selector);
69         $xpath = "";
70         foreach($parts as $part) {
71             if(preg_match('/^([A-Za-z][A-Za-z0-9]*)/', $part, $matches)) {
72                 $xpath .= "//$matches[1]";
73             } else {
74                 $xpath .= "//*";
75             }
76             $xfilters = array();
77             if(preg_match('/#([^#.\[]+)/', $part, $matches)) {
78                 $xfilters[] = "@id='$matches[1]'";
79             }
80             if(preg_match('/\.([^#.\[]+)/', $part, $matches)) {
81                 $xfilters[] = "contains(@class,'$matches[1]')";
82             }
83             if($xfilters) $xpath .= '[' . implode(',', $xfilters) . ']';
84         }
85         return $xpath;      
86     }
87 
88 }