1 <?php

     2 /**

     3  * Zend Framework

     4  *

     5  * LICENSE

     6  *

     7  * This source file is subject to the new BSD license that is bundled

     8  * with this package in the file LICENSE.txt.

     9  * It is also available through the world-wide-web at this URL:

    10  * http://framework.zend.com/license/new-bsd

    11  * If you did not receive a copy of the license and are unable to

    12  * obtain it through the world-wide-web, please send an email

    13  * to license@zend.com so we can send you a copy immediately.

    14  *

    15  * @category   Zend

    16  * @package    Zend_Dom

    17  * @copyright  Copyright (c) 2005-2010 Zend Technologies USA Inc. (http://www.zend.com)

    18  * @license    http://framework.zend.com/license/new-bsd     New BSD License

    19  * @version    $Id: Query.php 23062 2010-10-08 14:05:45Z matthew $

    20  */

    21 

    22 /**

    23  * @see Zend_Dom_Query_Css2Xpath

    24  */

    25 require_once 'Zend/Dom/Query/Css2Xpath.php';

    26 

    27 /**

    28  * @see Zend_Dom_Query_Result

    29  */

    30 require_once 'Zend/Dom/Query/Result.php';

    31 

    32 /**

    33  * Query DOM structures based on CSS selectors and/or XPath

    34  *

    35  * @package    Zend_Dom

    36  * @subpackage Query

    37  * @copyright  Copyright (c) 2005-2010 Zend Technologies USA Inc. (http://www.zend.com)

    38  * @license    http://framework.zend.com/license/new-bsd     New BSD License

    39  */

    40 class Zend_Dom_Query

    41 {

    42     /**#@+

    43      * Document types

    44      */

    45     const DOC_XML   = 'docXml';

    46     const DOC_HTML  = 'docHtml';

    47     const DOC_XHTML = 'docXhtml';

    48     /**#@-*/

    49 

    50     /**

    51      * @var string

    52      */

    53     protected $_document;

    54 

    55     /**

    56      * DOMDocument errors, if any

    57      * @var false|array

    58      */

    59     protected $_documentErrors = false;

    60 

    61     /**

    62      * Document type

    63      * @var string

    64      */

    65     protected $_docType;

    66 

    67     /**

    68      * Document encoding

    69      * @var null|string

    70      */

    71     protected $_encoding;

    72 

    73     /**

    74      * XPath namespaces

    75      * @var array

    76      */

    77     protected $_xpathNamespaces = array();

    78 

    79     /**

    80      * Constructor

    81      *

    82      * @param  null|string $document

    83      * @return void

    84      */

    85     public function __construct($document = null, $encoding = null)

    86     {

    87         $this->setEncoding($encoding);

    88         $this->setDocument($document);

    89     }

    90 

    91     /**

    92      * Set document encoding

    93      * 

    94      * @param  string $encoding 

    95      * @return Zend_Dom_Query

    96      */

    97     public function setEncoding($encoding)

    98     {

    99         $this->_encoding = (null === $encoding) ? null : (string) $encoding;

   100         return $this;

   101     }

   102 

   103     /**

   104      * Get document encoding

   105      * 

   106      * @return null|string

   107      */

   108     public function getEncoding()

   109     {

   110         return $this->_encoding;

   111     }

   112 

   113     /**

   114      * Set document to query

   115      *

   116      * @param  string $document

   117      * @param  null|string $encoding Document encoding

   118      * @return Zend_Dom_Query

   119      */

   120     public function setDocument($document, $encoding = null)

   121     {

   122         if (0 === strlen($document)) {

   123             return $this;

   124         }

   125         // breaking XML declaration to make syntax highlighting work

   126         if ('<' . '?xml' == substr(trim($document), 0, 5)) {

   127             return $this->setDocumentXml($document, $encoding);

   128         }

   129         if (strstr($document, 'DTD XHTML')) {

   130             return $this->setDocumentXhtml($document, $encoding);

   131         }

   132         return $this->setDocumentHtml($document, $encoding);

   133     }

   134 

   135     /**

   136      * Register HTML document

   137      *

   138      * @param  string $document

   139      * @param  null|string $encoding Document encoding

   140      * @return Zend_Dom_Query

   141      */

   142     public function setDocumentHtml($document, $encoding = null)

   143     {

   144         $this->_document = (string) $document;

   145         $this->_docType  = self::DOC_HTML;

   146         if (null !== $encoding) {

   147             $this->setEncoding($encoding);

   148         }

   149         return $this;

   150     }

   151 

   152     /**

   153      * Register XHTML document

   154      *

   155      * @param  string $document

   156      * @param  null|string $encoding Document encoding

   157      * @return Zend_Dom_Query

   158      */

   159     public function setDocumentXhtml($document, $encoding = null)

   160     {

   161         $this->_document = (string) $document;

   162         $this->_docType  = self::DOC_XHTML;

   163         if (null !== $encoding) {

   164             $this->setEncoding($encoding);

   165         }

   166         return $this;

   167     }

   168 

   169     /**

   170      * Register XML document

   171      *

   172      * @param  string $document

   173      * @param  null|string $encoding Document encoding

   174      * @return Zend_Dom_Query

   175      */

   176     public function setDocumentXml($document, $encoding = null)

   177     {

   178         $this->_document = (string) $document;

   179         $this->_docType  = self::DOC_XML;

   180         if (null !== $encoding) {

   181             $this->setEncoding($encoding);

   182         }

   183         return $this;

   184     }

   185 

   186     /**

   187      * Retrieve current document

   188      *

   189      * @return string

   190      */

   191     public function getDocument()

   192     {

   193         return $this->_document;

   194     }

   195 

   196     /**

   197      * Get document type

   198      *

   199      * @return string

   200      */

   201     public function getDocumentType()

   202     {

   203         return $this->_docType;

   204     }

   205 

   206     /**

   207      * Get any DOMDocument errors found

   208      * 

   209      * @return false|array

   210      */

   211     public function getDocumentErrors()

   212     {

   213         return $this->_documentErrors;

   214     }

   215 

   216     /**

   217      * Perform a CSS selector query

   218      *

   219      * @param  string $query

   220      * @return Zend_Dom_Query_Result

   221      */

   222     public function query($query)

   223     {

   224         $xpathQuery = Zend_Dom_Query_Css2Xpath::transform($query);

   225         return $this->queryXpath($xpathQuery, $query);

   226     }

   227 

   228     /**

   229      * Perform an XPath query

   230      *

   231      * @param  string|array $xpathQuery

   232      * @param  string $query CSS selector query

   233      * @return Zend_Dom_Query_Result

   234      */

   235     public function queryXpath($xpathQuery, $query = null)

   236     {

   237         if (null === ($document = $this->getDocument())) {

   238             require_once 'Zend/Dom/Exception.php';

   239             throw new Zend_Dom_Exception('Cannot query; no document registered');

   240         }

   241 

   242         $encoding = $this->getEncoding();

   243         libxml_use_internal_errors(true);

   244         if (null === $encoding) {

   245             $domDoc = new DOMDocument('1.0');

   246         } else {

   247             $domDoc = new DOMDocument('1.0', $encoding);

   248         }

   249         $type   = $this->getDocumentType();

   250         switch ($type) {

   251             case self::DOC_XML:

   252                 $success = $domDoc->loadXML($document);

   253                 break;

   254             case self::DOC_HTML:

   255             case self::DOC_XHTML:

   256             default:

   257                 $success = $domDoc->loadHTML($document);

   258                 break;

   259         }

   260         $errors = libxml_get_errors();

   261         if (!empty($errors)) {

   262             $this->_documentErrors = $errors;

   263             libxml_clear_errors();

   264         }

   265         libxml_use_internal_errors(false);

   266 

   267         if (!$success) {

   268             require_once 'Zend/Dom/Exception.php';

   269             throw new Zend_Dom_Exception(sprintf('Error parsing document (type == %s)', $type));

   270         }

   271 

   272         $nodeList   = $this->_getNodeList($domDoc, $xpathQuery);

   273         return new Zend_Dom_Query_Result($query, $xpathQuery, $domDoc, $nodeList);

   274     }

   275 

   276     /**

   277      * Register XPath namespaces

   278      *

   279      * @param   array $xpathNamespaces

   280      * @return  void

   281      */

   282     public function registerXpathNamespaces($xpathNamespaces)

   283     {

   284         $this->_xpathNamespaces = $xpathNamespaces;

   285     }

   286 

   287     /**

   288      * Prepare node list

   289      *

   290      * @param  DOMDocument $document

   291      * @param  string|array $xpathQuery

   292      * @return array

   293      */

   294     protected function _getNodeList($document, $xpathQuery)

   295     {

   296         $xpath      = new DOMXPath($document);

   297         foreach ($this->_xpathNamespaces as $prefix => $namespaceUri) {

   298             $xpath->registerNamespace($prefix, $namespaceUri);

   299         }

   300         $xpathQuery = (string) $xpathQuery;

   301         if (preg_match_all('|\[contains\((@[a-z0-9_-]+),\s?\' |i', $xpathQuery, $matches)) {

   302             foreach ($matches[1] as $attribute) {

   303                 $queryString = '//*[' . $attribute . ']';

   304                 $attributeName = substr($attribute, 1);

   305                 $nodes = $xpath->query($queryString);

   306                 foreach ($nodes as $node) {

   307                     $attr = $node->attributes->getNamedItem($attributeName);

   308                     $attr->value = ' ' . $attr->value . ' ';

   309                 }

   310             }

   311         }

   312         return $xpath->query($xpathQuery);

   313     }

   314 }