XHTMLCompiler/FilterManager.php

   1 <?php
   2
   3 /**
   4  * Manages various filters in a document, and performs text processing
   5  */
   6 class XHTMLCompiler_FilterManager
   7 {
   8
   9     protected $preTextFilters = array();
  10     protected $postTextFilters = array();
  11     protected $DOMFilters = array();
  12     protected $xcAttr = array();
  13
  14     protected $errors = array();
  15     protected $deps = array();
  16
  17     /**
  18      * Adds a pre-processing text filter to the queue.
  19      * @note Filters added here are run before the document is
  20      *       parsed into a DOM. Suggested use is for transforming
  21      *       non-XML style specialized markup.
  22      * @param $filter XHTMLCompiler_TextFilter
  23      */
  24     public function addPreTextFilter($filter) {
  25         $filter = $this->loadFilter($filter, 'TextFilter');
  26         $n = $filter->getName();
  27         if (isset($this->preTextFilters[$n])) {
  28             throw new Exception('Cannot overload pre text filter ' .
  29                 $filter->getName());
  30         }
  31         return $this->preTextFilters[$n] = $filter;
  32     }
  33
  34     /**
  35      * Adds a post-processing text filter to the queue.
  36      * @note Filters added here are run after the document has been
  37      *       parsed into a DOM and then serialized back. Suggested use
  38      *       is for fixing cosmetic issues with the source.
  39      * @warning Anything done on this step will not have its
  40      *          well-formedness corrected, so be careful.
  41      * @param $filter XHTMLCompiler_TextFilter
  42      */
  43     public function addPostTextFilter($filter) {
  44         $filter = $this->loadFilter($filter, 'TextFilter');
  45         $n = $filter->getName();
  46         if (isset($this->postTextFilters[$n])) {
  47             throw new Exception('Cannot overload post text filter ' .
  48                 $filter->getName());
  49         }
  50         return $this->postTextFilters[$n] = $filter;
  51     }
  52
  53     /**
  54      * Adds a DOM-processing filter to the queue
  55      * @param $filter XHTMLCompiler_DOMFilter
  56      */
  57     public function addDOMFilter($filter) {
  58         $filter = $this->loadFilter($filter, 'DOMFilter');
  59         $n = $filter->getName();
  60         if (isset($this->DOMFilters[$n])) {
  61             throw new Exception('Cannot overload DOM filter ' .
  62                 $filter->getName());
  63         }
  64         $attributes = $filter->getXCAttributesDefined();
  65         foreach ($attributes as $attribute) {
  66             if (isset($this->xcAttr[$attribute])) {
  67                 throw new Exception('Duplicate attribute definition for '.
  68                     'xc:' . $attribute);
  69             }
  70             $this->xcAttr[$attribute] = true;
  71         }
  72         return $this->DOMFilters[$n] = $filter;
  73     }
  74
  75     /**
  76      * If filter is string, load the filter based on a few guesses
  77      * @param $filter String or object filter
  78      */
  79     protected function loadFilter($filter, $subclass) {
  80         if (is_string($filter)) {
  81             $class = "XHTMLCompiler_{$subclass}_$filter";
  82             if (class_exists($class)) {
  83                 $filter = new $class;
  84             } elseif (class_exists($filter)) {
  85                 $filter = new $filter;
  86             } else {
  87                 require "$subclass/$filter.php";
  88                 $filter = new $class;
  89             }
  90         }
  91         return $filter;
  92     }
  93
  94     /** Returns the dependency array accumulated from the filter run */
  95     public function getDeps() {return $this->deps;}
  96
  97     /** Adds a file to the dependency list */
  98     public function addDependency($filename) {
  99         $this->deps[$filename] = filemtime($filename);
 100     }
 101
 102     /**
 103      * Accepts a page's text and processes it.
 104      * @param $text String text to be processed
 105      * @param $page XHTMLCompiler_Page representing currently processed page
 106      */
 107     public function process($text, $page) {
 108
 109         // do pre-text processing
 110         foreach ($this->preTextFilters as $filter) {
 111             $text = $filter->process($text, $page, $this);
 112         }
 113
 114         // setup XML catalog to improve speed
 115         $catalog = str_replace(array(' ', '\\'), array('%20', '/'),
 116             dirname(__FILE__)) . '/../catalog/catalog.xml';
 117         if ($catalog[1] == ':') $catalog = substr($catalog, 2); // remove drive
 118         putenv('XML_CATALOG_FILES=' . $catalog);
 119
 120         // configure DOMDocument
 121         $dom = new DOMDocument();
 122         $dom->preserveWhiteSpace = false;
 123         $dom->formatOutput = true;
 124         $dom->resolveExternals = true;
 125
 126         // todo: somehow, collect information on which entity files
 127         // are being added to the document, and add to xc-deps.
 128         $dom->substituteEntities = true; // allows for custom entities too!
 129
 130         $dom->loadXML($text);
 131
 132         $internal_subset = $dom->doctype->internalSubset;
 133         if ($internal_subset) {
 134             // there are some entities that need to be registered to
 135             // the dependency list. Match ones that declare SYSTEM
 136             // '<!ENTITY' S '%' S  Name  S  PEDef  S? '>'
 137             preg_match_all(
 138               '/<!ENTITY\s+%\s+[^\s]+\s+(?:PUBLIC.+?)?SYSTEM\s+(?:"([^"]*)"|\'([^\']*)\')\s*>/s',
 139               $internal_subset,
 140               $matches
 141             );
 142             foreach ($matches[1] as $filename) {
 143                 // $filename will always be relative to web root, so
 144                 // no munging necessary
 145                 $this->addDependency($filename);
 146             }
 147         }
 148
 149         set_error_handler(array($this, 'muteErrorHandler'));
 150         $dom->validate();
 151         restore_error_handler();
 152
 153         $dom->encoding = 'UTF-8'; // override document encoding
 154
 155         // XInclude
 156         // todo:
 157         // * factor into a DOMFilter
 158         // * add xincludes to the dependency list
 159         $xpath = new DOMXPath($dom);
 160         $xpath->registerNamespace('xi', $ns = 'http://www.w3.org/2001/XInclude');
 161         $nodes = $xpath->query('//xi:include');
 162         foreach ($nodes as $node) {
 163             if (! $node instanceof DOMElement) continue;
 164             if (! $filename = $node->getAttribute('href')) continue;
 165             // doesn't handle second-level includes
 166             $this->addDependency($filename);
 167         }
 168         // perform includes (we might need to loop to handle nested includes)
 169         $dom->xinclude();
 170
 171         // run DOM filters
 172         foreach ($this->DOMFilters as $filter) {
 173             $filter->setup($dom);
 174             $filter->process($dom, $page, $this);
 175         }
 176
 177         // translate back to text
 178         $text = $dom->saveXML();
 179
 180         // remove all non-default namespace declarations
 181         $text = preg_replace('/ xmlns:.+?=".+?"/', '', $text);
 182         foreach ($this->postTextFilters as $filter) {
 183             $text = $filter->process($text, $page, $this);
 184         }
 185
 186         // okay, now finally do validation, and let the errors get
 187         // spit out if there are some
 188         // collect parse errors
 189         set_error_handler(array($this, 'validationErrorHandler'));
 190         $dom->loadXML($text);
 191         $status = $dom->validate();
 192         restore_error_handler();
 193
 194         if (!$status || !empty($this->errors)) {
 195             $body = $dom->getElementsByTagName('body')->item(0);
 196             if (!$body) {
 197                 $dom->appendChild($html = $dom->createElement('html'));
 198                 $html->appendChild($body = $dom->createElement('body'));
 199             }
 200             $warning = $dom->createElement('div');
 201             $warning->setAttribute('class', 'warning');
 202             $warning->appendChild($dom->createElement('h2', 'Warning: Errors'));
 203             $warning->appendChild($dom->createElement('p', 'This document has validation errors:'));
 204             $list = $dom->createElement('ul');
 205             foreach ($this->errors as $error) {
 206                 // strip-tags removes HTML tags to make the plaintext output
 207                 // more friendly, IS NOT for security reasons
 208                 $list->appendChild($dom->createElement('li', strip_tags($error)));
 209             }
 210             $warning->appendChild($list);
 211             $body->insertBefore($warning, $body->childNodes->item(0));
 212             $text = $dom->saveXML();
 213         }
 214
 215         // scrub out XML declaration for Internet Explorer
 216         // disabled, we're going to try serving application/xhtml+xml
 217         // to browsers that support it
 218         /*$text = str_replace('<?xml version="1.0" encoding="UTF-8"?>'."\n", '', $text);*/
 219
 220         // scrub out custom DTD additions, they should have been
 221         // resolved already
 222         $text = preg_replace('/(<!DOCTYPE[^>]*?) ?\[[^\]]+\]/', '\1', $text);
 223
 224         return $text;
 225     }
 226
 227     /**
 228      * Temporary error handler to use when validating a document
 229      */
 230     public function validationErrorHandler($n, $text) {
 231         $this->errors[] = $text;
 232     }
 233
 234     /**
 235      * Handler that mutes all errors
 236      */
 237     public function muteErrorHandler($n, $t) {}
 238
 239 }
 240
 241 ?>