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 (usually XHTML) 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         // generate the DOM
 115         $this->setupXMLCatalog();
 116         $dom = $this->createDOM($text);
 117
 118         $this->analyzeInternalSubset($dom);
 119
 120         // validate the document to force the entities to be resolved,
 121         // we don't actually care about the errors
 122         set_error_handler(array($this, 'muteErrorHandler'));
 123         $dom->validate();
 124         restore_error_handler();
 125
 126         $this->analyzeXIncludes($dom);
 127         $dom->xinclude();
 128
 129         // run DOM filters
 130         foreach ($this->DOMFilters as $filter) {
 131             $filter->setup($dom);
 132             $filter->process($dom, $page, $this);
 133         }
 134
 135         // translate back to text
 136         $text = $dom->saveXML();
 137
 138         // remove all non-default namespace declarations, may change,
 139         // but for now embedded XML namespaces are not cross-browser friendly
 140         $text = preg_replace('/ xmlns:.+?=".+?"/', '', $text);
 141         // scrub out custom DTD additions
 142         $text = preg_replace('/(<!DOCTYPE[^>]*?) ?\[[^\]]+\]/', '\1', $text);
 143         foreach ($this->postTextFilters as $filter) {
 144             $text = $filter->process($text, $page, $this);
 145         }
 146
 147         // okay, now finally do validation, and let the errors get
 148         // spit out if there are some collect parse errors
 149         set_error_handler(array($this, 'validationErrorHandler'));
 150         $dom->loadXML($text);
 151         $status = $dom->validate();
 152         restore_error_handler();
 153         if (!$status || !empty($this->errors)) {
 154             $this->buildErrors($dom);
 155             $text = $dom->saveXML();
 156         }
 157
 158         return $text;
 159     }
 160
 161     /**
 162      * Temporary error handler to use when validating a document
 163      */
 164     public function validationErrorHandler($n, $text) {
 165         $this->errors[] = $text;
 166     }
 167
 168     /**
 169      * Handler that mutes all errors
 170      */
 171     public function muteErrorHandler($n, $t) {}
 172
 173     /**
 174      * Sets up an XML catalog to speed up entity resolution
 175      */
 176     public function setupXMLCatalog() {
 177         $catalog = str_replace(array(' ', '\\'), array('%20', '/'),
 178             dirname(__FILE__)) . '/../catalog/catalog.xml';
 179         if ($catalog[1] == ':') $catalog = substr($catalog, 2); // remove drive
 180         putenv('XML_CATALOG_FILES=' . $catalog);
 181     }
 182
 183     /**
 184      * Creates a reasonable well default configured DOM
 185      * @param string $xml XML to load DOM with
 186      */
 187     public function createDOM($text = false) {
 188         $dom = new DOMDocument();
 189         $dom->preserveWhiteSpace = false;
 190         $dom->formatOutput = true;
 191         $dom->resolveExternals = true;
 192
 193         // todo: somehow, collect information on which entity files
 194         // are being added to the document, and add to xc-deps.
 195         $dom->substituteEntities = true; // allows for custom entities too!
 196
 197         if ($text !== false) $dom->loadXML($text);
 198
 199         return $dom;
 200     }
 201
 202     /**
 203      * Analyzes the internal subset of a DOM, registering any file
 204      * entity definitions as dependencies
 205      */
 206     public function analyzeInternalSubset($dom) {
 207         if (empty($dom->doctype)) return;
 208         $internal_subset = $dom->doctype->internalSubset;
 209         if ($internal_subset) {
 210             // there are some entities that need to be registered to
 211             // the dependency list. Match ones that declare SYSTEM
 212             // '<!ENTITY' S '%' S  Name  S  PEDef  S? '>'
 213             preg_match_all(
 214               '/<!ENTITY\s+%\s+[^\s]+\s+(?:PUBLIC.+?)?SYSTEM\s+(?:"([^"]*)"|\'([^\']*)\')\s*>/s',
 215               $internal_subset,
 216               $matches
 217             );
 218             foreach ($matches[1] as $filename) {
 219                 // $filename will always be relative to web root, so
 220                 // no munging necessary
 221                 $this->addDependency($filename);
 222             }
 223         }
 224     }
 225
 226     /**
 227      * Analyzes a documents XIncludes and registers necessary dependencies.
 228      * Make sure you call this before calling $dom->xinclude
 229      * @param DOMDocument $dom to process
 230      * @todo Factor into a DOMFilter
 231      * @todo Handle arbitrary nestings of includes
 232      */
 233     public function analyzeXIncludes($dom) {
 234         $xpath = new DOMXPath($dom);
 235         $xpath->registerNamespace('xi', $ns = 'http://www.w3.org/2001/XInclude');
 236         $nodes = $xpath->query('//xi:include');
 237         foreach ($nodes as $node) {
 238             if (! $node instanceof DOMElement) continue;
 239             if (! $filename = $node->getAttribute('href')) continue;
 240             $this->addDependency($filename);
 241             $sub_dom = new DOMDocument();
 242             $sub_dom->load($filename);
 243             $this->analyzeInternalSubset($sub_dom);
 244         }
 245     }
 246
 247     /**
 248      * Adds validation errors to the output document as a message
 249      */
 250     public function buildErrors($dom) {
 251         $body = $dom->getElementsByTagName('body')->item(0);
 252         if (!$body) {
 253             $dom->appendChild($html = $dom->createElement('html'));
 254             $html->appendChild($body = $dom->createElement('body'));
 255         }
 256         $warning = $dom->createElement('div');
 257         $warning->setAttribute('class', 'warning');
 258         $warning->appendChild($dom->createElement('h2', 'Warning: Errors'));
 259         $warning->appendChild($dom->createElement('p', 'This document has validation errors:'));
 260         $list = $dom->createElement('ul');
 261         foreach ($this->errors as $error) {
 262             // strip-tags removes HTML tags to make the plaintext output
 263             // more friendly, IS NOT for security reasons
 264             $list->appendChild($dom->createElement('li', strip_tags($error)));
 265         }
 266         $warning->appendChild($list);
 267         $body->insertBefore($warning, $body->childNodes->item(0));
 268     }
 269
 270 }
 271
 272 ?>