XHTMLCompiler/DOMFilter/Acronymizer.php

   1 <?php
   2
   3 /**
   4  * Based on a list of known acronyms, populates of the title attribute
   5  * of acronym elements in documents.
   6  */
   7 class XHTMLCompiler_DOMFilter_Acronymizer extends XHTMLCompiler_DOMFilter
   8 {
   9
  10     protected $name = 'Acronymizer';
  11
  12     /**
  13      * Array of recognized acronyms.
  14      * @todo Make a public API for this, allow multiple acronym sets
  15      *       and different precedences for them.
  16      */
  17     protected $acronyms = array(
  18         // markup languages and related technologies
  19         'SGML' => 'Standard Generalized Markup Language',
  20         'HTML' => 'HyperText Markup Language',
  21         'XHTML' => 'Extensible HyperText Markup Language',
  22         'XML' => 'Extensible Markup Language',
  23         'RSS' => 'Really Simple Syndication',
  24         'DTD' => 'Document Type Definition',
  25         'CSS' => 'Cascading Style Sheets',
  26         // programming/apis
  27         'PHP' => 'PHP: HyperText Preprocessor',
  28         'CMS' => 'Content Management System',
  29         'API' => 'Application Programming Interface',
  30         'DOM' => 'Document Object Module',
  31         'SAX' => 'Simple API for XML',
  32         // web-app security
  33         'XSS' => 'Cross-Site Scripting',
  34         // organizations/groups
  35         'W3C' => 'World Wide Web Consortium',
  36         'RFC' => 'Request for Comment',
  37         'PEAR' => 'PHP Extension and Application Repository',
  38         'PECL' => 'PHP Extension Community Library',
  39         // paradigms
  40         'WYSIWYG' => 'What You See Is What You Get',
  41         'WYSIWYM' => 'What You See Is What You Mean',
  42         // character encodings
  43         'UTF-8' => '8-bit Unicode Transformation Format',
  44         'ASCII' => 'American Standard Code for Information Interchange',
  45         // other
  46         'INI' => 'Initialization',
  47         'UI' => 'User Interface',
  48         'SHA-1' => 'Secure Hash Algorithm 1',
  49         'CPU' => 'Central Processing Unit',
  50         'LGPL' => 'Lesser GNU Public License',
  51         'URI' => 'Uniform Resource Identifier',
  52     );
  53
  54     public function process(DOMDocument $dom, $page) {
  55         $nodes = $this->query("//html:acronym[not(@title)]");
  56         foreach ($nodes as $node) {
  57             $acronym = $node->textContent;
  58             if (!isset($this->acronyms[$acronym])) {
  59                 trigger_error(htmlspecialchars($acronym) . ' is not a recognized acronym (missing title attribute)');
  60                 continue;
  61             }
  62             $node->setAttribute('title', $this->acronyms[$acronym]);
  63         }
  64     }
  65
  66 }
  67
  68 ?>