Add %HTML.TargetNoreferrer, which adds rel="noreferrer" when target attribute is set
[htmlpurifier.git] / library / HTMLPurifier / HTMLModuleManager.php
blob2546c043c4685f0b6ea31b82900e4d44961569a0
1 <?php
3 class HTMLPurifier_HTMLModuleManager
6 /**
7 * @type HTMLPurifier_DoctypeRegistry
8 */
9 public $doctypes;
11 /**
12 * Instance of current doctype.
13 * @type string
15 public $doctype;
17 /**
18 * @type HTMLPurifier_AttrTypes
20 public $attrTypes;
22 /**
23 * Active instances of modules for the specified doctype are
24 * indexed, by name, in this array.
25 * @type HTMLPurifier_HTMLModule[]
27 public $modules = array();
29 /**
30 * Array of recognized HTMLPurifier_HTMLModule instances,
31 * indexed by module's class name. This array is usually lazy loaded, but a
32 * user can overload a module by pre-emptively registering it.
33 * @type HTMLPurifier_HTMLModule[]
35 public $registeredModules = array();
37 /**
38 * List of extra modules that were added by the user
39 * using addModule(). These get unconditionally merged into the current doctype, whatever
40 * it may be.
41 * @type HTMLPurifier_HTMLModule[]
43 public $userModules = array();
45 /**
46 * Associative array of element name to list of modules that have
47 * definitions for the element; this array is dynamically filled.
48 * @type array
50 public $elementLookup = array();
52 /**
53 * List of prefixes we should use for registering small names.
54 * @type array
56 public $prefixes = array('HTMLPurifier_HTMLModule_');
58 /**
59 * @type HTMLPurifier_ContentSets
61 public $contentSets;
63 /**
64 * @type HTMLPurifier_AttrCollections
66 public $attrCollections;
68 /**
69 * If set to true, unsafe elements and attributes will be allowed.
70 * @type bool
72 public $trusted = false;
74 public function __construct()
76 // editable internal objects
77 $this->attrTypes = new HTMLPurifier_AttrTypes();
78 $this->doctypes = new HTMLPurifier_DoctypeRegistry();
80 // setup basic modules
81 $common = array(
82 'CommonAttributes', 'Text', 'Hypertext', 'List',
83 'Presentation', 'Edit', 'Bdo', 'Tables', 'Image',
84 'StyleAttribute',
85 // Unsafe:
86 'Scripting', 'Object', 'Forms',
87 // Sorta legacy, but present in strict:
88 'Name',
90 $transitional = array('Legacy', 'Target', 'Iframe');
91 $xml = array('XMLCommonAttributes');
92 $non_xml = array('NonXMLCommonAttributes');
94 // setup basic doctypes
95 $this->doctypes->register(
96 'HTML 4.01 Transitional',
97 false,
98 array_merge($common, $transitional, $non_xml),
99 array('Tidy_Transitional', 'Tidy_Proprietary'),
100 array(),
101 '-//W3C//DTD HTML 4.01 Transitional//EN',
102 'http://www.w3.org/TR/html4/loose.dtd'
105 $this->doctypes->register(
106 'HTML 4.01 Strict',
107 false,
108 array_merge($common, $non_xml),
109 array('Tidy_Strict', 'Tidy_Proprietary', 'Tidy_Name'),
110 array(),
111 '-//W3C//DTD HTML 4.01//EN',
112 'http://www.w3.org/TR/html4/strict.dtd'
115 $this->doctypes->register(
116 'XHTML 1.0 Transitional',
117 true,
118 array_merge($common, $transitional, $xml, $non_xml),
119 array('Tidy_Transitional', 'Tidy_XHTML', 'Tidy_Proprietary', 'Tidy_Name'),
120 array(),
121 '-//W3C//DTD XHTML 1.0 Transitional//EN',
122 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd'
125 $this->doctypes->register(
126 'XHTML 1.0 Strict',
127 true,
128 array_merge($common, $xml, $non_xml),
129 array('Tidy_Strict', 'Tidy_XHTML', 'Tidy_Strict', 'Tidy_Proprietary', 'Tidy_Name'),
130 array(),
131 '-//W3C//DTD XHTML 1.0 Strict//EN',
132 'http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd'
135 $this->doctypes->register(
136 'XHTML 1.1',
137 true,
138 // Iframe is a real XHTML 1.1 module, despite being
139 // "transitional"!
140 array_merge($common, $xml, array('Ruby', 'Iframe')),
141 array('Tidy_Strict', 'Tidy_XHTML', 'Tidy_Proprietary', 'Tidy_Strict', 'Tidy_Name'), // Tidy_XHTML1_1
142 array(),
143 '-//W3C//DTD XHTML 1.1//EN',
144 'http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd'
150 * Registers a module to the recognized module list, useful for
151 * overloading pre-existing modules.
152 * @param $module Mixed: string module name, with or without
153 * HTMLPurifier_HTMLModule prefix, or instance of
154 * subclass of HTMLPurifier_HTMLModule.
155 * @param $overload Boolean whether or not to overload previous modules.
156 * If this is not set, and you do overload a module,
157 * HTML Purifier will complain with a warning.
158 * @note This function will not call autoload, you must instantiate
159 * (and thus invoke) autoload outside the method.
160 * @note If a string is passed as a module name, different variants
161 * will be tested in this order:
162 * - Check for HTMLPurifier_HTMLModule_$name
163 * - Check all prefixes with $name in order they were added
164 * - Check for literal object name
165 * - Throw fatal error
166 * If your object name collides with an internal class, specify
167 * your module manually. All modules must have been included
168 * externally: registerModule will not perform inclusions for you!
170 public function registerModule($module, $overload = false)
172 if (is_string($module)) {
173 // attempt to load the module
174 $original_module = $module;
175 $ok = false;
176 foreach ($this->prefixes as $prefix) {
177 $module = $prefix . $original_module;
178 if (class_exists($module)) {
179 $ok = true;
180 break;
183 if (!$ok) {
184 $module = $original_module;
185 if (!class_exists($module)) {
186 trigger_error(
187 $original_module . ' module does not exist',
188 E_USER_ERROR
190 return;
193 $module = new $module();
195 if (empty($module->name)) {
196 trigger_error('Module instance of ' . get_class($module) . ' must have name');
197 return;
199 if (!$overload && isset($this->registeredModules[$module->name])) {
200 trigger_error('Overloading ' . $module->name . ' without explicit overload parameter', E_USER_WARNING);
202 $this->registeredModules[$module->name] = $module;
206 * Adds a module to the current doctype by first registering it,
207 * and then tacking it on to the active doctype
209 public function addModule($module)
211 $this->registerModule($module);
212 if (is_object($module)) {
213 $module = $module->name;
215 $this->userModules[] = $module;
219 * Adds a class prefix that registerModule() will use to resolve a
220 * string name to a concrete class
222 public function addPrefix($prefix)
224 $this->prefixes[] = $prefix;
228 * Performs processing on modules, after being called you may
229 * use getElement() and getElements()
230 * @param HTMLPurifier_Config $config
232 public function setup($config)
234 $this->trusted = $config->get('HTML.Trusted');
236 // generate
237 $this->doctype = $this->doctypes->make($config);
238 $modules = $this->doctype->modules;
240 // take out the default modules that aren't allowed
241 $lookup = $config->get('HTML.AllowedModules');
242 $special_cases = $config->get('HTML.CoreModules');
244 if (is_array($lookup)) {
245 foreach ($modules as $k => $m) {
246 if (isset($special_cases[$m])) {
247 continue;
249 if (!isset($lookup[$m])) {
250 unset($modules[$k]);
255 // custom modules
256 if ($config->get('HTML.Proprietary')) {
257 $modules[] = 'Proprietary';
259 if ($config->get('HTML.SafeObject')) {
260 $modules[] = 'SafeObject';
262 if ($config->get('HTML.SafeEmbed')) {
263 $modules[] = 'SafeEmbed';
265 if ($config->get('HTML.SafeScripting') !== array()) {
266 $modules[] = 'SafeScripting';
268 if ($config->get('HTML.Nofollow')) {
269 $modules[] = 'Nofollow';
271 if ($config->get('HTML.TargetBlank')) {
272 $modules[] = 'TargetBlank';
274 // NB: HTML.TargetNoreferrer must be AFTER HTML.TargetBlank
275 // so that its post-attr-transform gets run afterwards.
276 if ($config->get('HTML.TargetNoreferrer')) {
277 $modules[] = 'TargetNoreferrer';
280 // merge in custom modules
281 $modules = array_merge($modules, $this->userModules);
283 foreach ($modules as $module) {
284 $this->processModule($module);
285 $this->modules[$module]->setup($config);
288 foreach ($this->doctype->tidyModules as $module) {
289 $this->processModule($module);
290 $this->modules[$module]->setup($config);
293 // prepare any injectors
294 foreach ($this->modules as $module) {
295 $n = array();
296 foreach ($module->info_injector as $injector) {
297 if (!is_object($injector)) {
298 $class = "HTMLPurifier_Injector_$injector";
299 $injector = new $class;
301 $n[$injector->name] = $injector;
303 $module->info_injector = $n;
306 // setup lookup table based on all valid modules
307 foreach ($this->modules as $module) {
308 foreach ($module->info as $name => $def) {
309 if (!isset($this->elementLookup[$name])) {
310 $this->elementLookup[$name] = array();
312 $this->elementLookup[$name][] = $module->name;
316 // note the different choice
317 $this->contentSets = new HTMLPurifier_ContentSets(
318 // content set assembly deals with all possible modules,
319 // not just ones deemed to be "safe"
320 $this->modules
322 $this->attrCollections = new HTMLPurifier_AttrCollections(
323 $this->attrTypes,
324 // there is no way to directly disable a global attribute,
325 // but using AllowedAttributes or simply not including
326 // the module in your custom doctype should be sufficient
327 $this->modules
332 * Takes a module and adds it to the active module collection,
333 * registering it if necessary.
335 public function processModule($module)
337 if (!isset($this->registeredModules[$module]) || is_object($module)) {
338 $this->registerModule($module);
340 $this->modules[$module] = $this->registeredModules[$module];
344 * Retrieves merged element definitions.
345 * @return Array of HTMLPurifier_ElementDef
347 public function getElements()
349 $elements = array();
350 foreach ($this->modules as $module) {
351 if (!$this->trusted && !$module->safe) {
352 continue;
354 foreach ($module->info as $name => $v) {
355 if (isset($elements[$name])) {
356 continue;
358 $elements[$name] = $this->getElement($name);
362 // remove dud elements, this happens when an element that
363 // appeared to be safe actually wasn't
364 foreach ($elements as $n => $v) {
365 if ($v === false) {
366 unset($elements[$n]);
370 return $elements;
375 * Retrieves a single merged element definition
376 * @param string $name Name of element
377 * @param bool $trusted Boolean trusted overriding parameter: set to true
378 * if you want the full version of an element
379 * @return HTMLPurifier_ElementDef Merged HTMLPurifier_ElementDef
380 * @note You may notice that modules are getting iterated over twice (once
381 * in getElements() and once here). This
382 * is because
384 public function getElement($name, $trusted = null)
386 if (!isset($this->elementLookup[$name])) {
387 return false;
390 // setup global state variables
391 $def = false;
392 if ($trusted === null) {
393 $trusted = $this->trusted;
396 // iterate through each module that has registered itself to this
397 // element
398 foreach ($this->elementLookup[$name] as $module_name) {
399 $module = $this->modules[$module_name];
401 // refuse to create/merge from a module that is deemed unsafe--
402 // pretend the module doesn't exist--when trusted mode is not on.
403 if (!$trusted && !$module->safe) {
404 continue;
407 // clone is used because, ideally speaking, the original
408 // definition should not be modified. Usually, this will
409 // make no difference, but for consistency's sake
410 $new_def = clone $module->info[$name];
412 if (!$def && $new_def->standalone) {
413 $def = $new_def;
414 } elseif ($def) {
415 // This will occur even if $new_def is standalone. In practice,
416 // this will usually result in a full replacement.
417 $def->mergeIn($new_def);
418 } else {
419 // :TODO:
420 // non-standalone definitions that don't have a standalone
421 // to merge into could be deferred to the end
422 // HOWEVER, it is perfectly valid for a non-standalone
423 // definition to lack a standalone definition, even
424 // after all processing: this allows us to safely
425 // specify extra attributes for elements that may not be
426 // enabled all in one place. In particular, this might
427 // be the case for trusted elements. WARNING: care must
428 // be taken that the /extra/ definitions are all safe.
429 continue;
432 // attribute value expansions
433 $this->attrCollections->performInclusions($def->attr);
434 $this->attrCollections->expandIdentifiers($def->attr, $this->attrTypes);
436 // descendants_are_inline, for ChildDef_Chameleon
437 if (is_string($def->content_model) &&
438 strpos($def->content_model, 'Inline') !== false) {
439 if ($name != 'del' && $name != 'ins') {
440 // this is for you, ins/del
441 $def->descendants_are_inline = true;
445 $this->contentSets->generateChildDef($def, $module);
448 // This can occur if there is a blank definition, but no base to
449 // mix it in with
450 if (!$def) {
451 return false;
454 // add information on required attributes
455 foreach ($def->attr as $attr_name => $attr_def) {
456 if ($attr_def->required) {
457 $def->required_attr[] = $attr_name;
460 return $def;
464 // vim: et sw=4 sts=4