/** * @param HTMLPurifier_URI $uri * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return bool */ public function filter(&$uri, $config, $context) { // check if filter not applicable if (!$config->get('HTML.SafeIframe')) { return true; } // check if the filter should actually trigger if (!$context->get('EmbeddedURI', true)) { return true; } $token = $context->get('CurrentToken', true); if (!($token && $token->name == 'iframe')) { return true; } // check if we actually have some whitelists enabled if ($this->regexp === null) { return false; } // actually check the whitelists if (!preg_match($this->regexp, $uri->toString())) { return false; } // Make sure that if we're an HTTPS site, the iframe is also HTTPS if (is_https() && $uri->scheme == 'http') { // Convert it to a protocol-relative URL $uri->scheme = null; } return $uri; }
/** * @param HTMLPurifier_Context $context */ public function __construct($context) { $this->locale =& $context->get('Locale'); $this->context = $context; $this->_current =& $this->_stacks[0]; $this->errors =& $this->_stacks[0]; }
/** * @param HTMLPurifier_URI $uri * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return bool */ public function filter(&$uri, $config, $context) { if (!$context->get('EmbeddedURI', true)) { return true; } return parent::filter($uri, $config, $context); }
/** * @param HTMLPurifier_Node[] $children * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return bool */ public function validateChildren($children, $config, $context) { if ($context->get('IsInline') === false) { return $this->block->validateChildren($children, $config, $context); } else { return $this->inline->validateChildren($children, $config, $context); } }
public function testNull() { $context = new HTMLPurifier_Context(); $var = NULL; $context->register('var', $var); $this->assertNull($context->get('var')); $context->destroy('var'); }
/** * Checks if CurrentToken is set and equal to $this->element * @param string $string * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return bool|string */ public function validate($string, $config, $context) { $token = $context->get('CurrentToken', true); if ($token && $token->name == $this->element) { return false; } return $this->def->validate($string, $config, $context); }
/** * @param string $string * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return bool|string */ public function validate($string, $config, $context) { $token = $context->get('CurrentToken', true); if (!$token || $token->name !== $this->tag) { return $this->withoutTag->validate($string, $config, $context); } else { return $this->withTag->validate($string, $config, $context); } }
/** * @param string $id * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return bool|string */ public function validate($id, $config, $context) { if (!$this->selector && !$config->get('Attr.EnableID')) { return false; } $id = trim($id); // trim it first if ($id === '') { return false; } $prefix = $config->get('Attr.IDPrefix'); if ($prefix !== '') { $prefix .= $config->get('Attr.IDPrefixLocal'); // prevent re-appending the prefix if (strpos($id, $prefix) !== 0) { $id = $prefix . $id; } } elseif ($config->get('Attr.IDPrefixLocal') !== '') { trigger_error('%Attr.IDPrefixLocal cannot be used unless ' . '%Attr.IDPrefix is set', E_USER_WARNING); } if (!$this->selector) { $id_accumulator =& $context->get('IDAccumulator'); if (isset($id_accumulator->ids[$id])) { return false; } } // we purposely avoid using regex, hopefully this is faster if ($config->get('Attr.ID.HTML5') === true) { if (preg_match('/[\\t\\n\\x0b\\x0c ]/', $id)) { return false; } } else { if (ctype_alpha($id)) { // OK } else { if (!ctype_alpha(@$id[0])) { return false; } // primitive style of regexps, I suppose $trim = trim($id, 'A..Za..z0..9:-._'); if ($trim !== '') { return false; } } } $regexp = $config->get('Attr.IDBlacklistRegexp'); if ($regexp && preg_match($regexp, $id)) { return false; } if (!$this->selector) { $id_accumulator->add($id); } // if no change was made to the ID, return the result // else, return the new id if stripping whitespace made it // valid, or return false. return $id; }
/** * @param HTMLPurifier_URI $uri * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return bool */ public function filter(&$uri, $config, $context) { // check if filter not applicable if (!$config->get('HTML.SafeIframe')) { return true; } // check if the filter should actually trigger if (!$context->get('EmbeddedURI', true)) { return true; } $token = $context->get('CurrentToken', true); if (!($token && $token->name == 'iframe')) { return true; } // check if we actually have some whitelists enabled if ($this->regexp === null) { return false; } // actually check the whitelists return preg_match($this->regexp, $uri->toString()); }
/** * filter * * @param HTMLPurifier_URI $uri * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return boolean */ public function filter(&$uri, $config, $context) { $result = TRUE; $token = $context->get('CurrentToken', true); if (Tinebase_Core::isLogLevel(Zend_Log::TRACE)) { Tinebase_Core::getLogger()->trace(__METHOD__ . '::' . __LINE__ . ' URI: ' . var_export($uri, TRUE) . ' ' . ' TOKEN: ' . var_export($token, TRUE)); } if ($uri->host) { $result = $this->_checkExternalUrl($uri, $token); } return $result; }
/** * @param HTMLPurifier_URI $uri * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * * @return bool */ public function filter(&$uri, $config, $context) { // skip non-resource URIs if (!$context->get('EmbeddedURI', true)) { return true; } //if(empty($this->allowed)) return false; if (!empty($uri->scheme) && strtolower($uri->scheme) != 'http' && strtolower($uri->scheme) != 'https') { // do not touch non-HTTP URLs return true; } // relative URLs permitted since email templates use it // if(empty($uri->host)) return false; // allow URLs with no query if (empty($uri->query)) { return true; } // allow URLs for known good hosts foreach ($this->allowed as $allow) { // must be equal to our domain or subdomain of our domain if ($uri->host == $allow || substr($uri->host, -(strlen($allow) + 1)) == ".{$allow}") { return true; } } // Here we try to block URLs that may be used for nasty XSRF stuff by // referring back to Sugar URLs // allow URLs that don't start with /? or /index.php? if (!empty($uri->path) && $uri->path != '/') { $lpath = strtolower($uri->path); if (substr($lpath, -10) != '/index.php' && $lpath != 'index.php') { return true; } } $query_items = []; parse_str(from_html($uri->query), $query_items); // weird query, probably harmless if (empty($query_items)) { return true; } // suspiciously like SugarCRM query, reject if (!empty($query_items['module']) && !empty($query_items['action'])) { return false; } // looks like non-download entry point - allow only specific entry points if (!empty($query_items['entryPoint']) && !in_array($query_items['entryPoint'], ['download', 'image', 'getImage'])) { return false; } return true; }
/** * Validates the attributes of a token, mutating it as necessary. * that has valid tokens * @param HTMLPurifier_Token $token Token to validate. * @param HTMLPurifier_Config $config Instance of HTMLPurifier_Config * @param HTMLPurifier_Context $context Instance of HTMLPurifier_Context */ public function validateToken($token, $config, $context) { $definition = $config->getHTMLDefinition(); $e =& $context->get('ErrorCollector', true); // initialize IDAccumulator if necessary $ok =& $context->get('IDAccumulator', true); if (!$ok) { $id_accumulator = HTMLPurifier_IDAccumulator::build($config, $context); $context->register('IDAccumulator', $id_accumulator); } // initialize CurrentToken if necessary $current_token =& $context->get('CurrentToken', true); if (!$current_token) { $context->register('CurrentToken', $token); } if (!$token instanceof HTMLPurifier_Token_Start && !$token instanceof HTMLPurifier_Token_Empty) { return; } // create alias to global definition array, see also $defs // DEFINITION CALL $d_defs = $definition->info_global_attr; // don't update token until the very end, to ensure an atomic update $attr = $token->attr; // do global transformations (pre) // nothing currently utilizes this foreach ($definition->info_attr_transform_pre as $transform) { $attr = $transform->transform($o = $attr, $config, $context); if ($e) { if ($attr != $o) { $e->send(E_NOTICE, 'AttrValidator: Attributes transformed', $o, $attr); } } } // do local transformations only applicable to this element (pre) // ex. <p align="right"> to <p style="text-align:right;"> foreach ($definition->info[$token->name]->attr_transform_pre as $transform) { $attr = $transform->transform($o = $attr, $config, $context); if ($e) { if ($attr != $o) { $e->send(E_NOTICE, 'AttrValidator: Attributes transformed', $o, $attr); } } } // create alias to this element's attribute definition array, see // also $d_defs (global attribute definition array) // DEFINITION CALL $defs = $definition->info[$token->name]->attr; $attr_key = false; $context->register('CurrentAttr', $attr_key); // iterate through all the attribute keypairs // Watch out for name collisions: $key has previously been used foreach ($attr as $attr_key => $value) { // call the definition if (isset($defs[$attr_key])) { // there is a local definition defined if ($defs[$attr_key] === false) { // We've explicitly been told not to allow this element. // This is usually when there's a global definition // that must be overridden. // Theoretically speaking, we could have a // AttrDef_DenyAll, but this is faster! $result = false; } else { // validate according to the element's definition $result = $defs[$attr_key]->validate($value, $config, $context); } } elseif (isset($d_defs[$attr_key])) { // there is a global definition defined, validate according // to the global definition $result = $d_defs[$attr_key]->validate($value, $config, $context); } else { // system never heard of the attribute? DELETE! $result = false; } // put the results into effect if ($result === false || $result === null) { // this is a generic error message that should replaced // with more specific ones when possible if ($e) { $e->send(E_ERROR, 'AttrValidator: Attribute removed'); } // remove the attribute unset($attr[$attr_key]); } elseif (is_string($result)) { // generally, if a substitution is happening, there // was some sort of implicit correction going on. We'll // delegate it to the attribute classes to say exactly what. // simple substitution $attr[$attr_key] = $result; } else { // nothing happens } // we'd also want slightly more complicated substitution // involving an array as the return value, // although we're not sure how colliding attributes would // resolve (certain ones would be completely overriden, // others would prepend themselves). } $context->destroy('CurrentAttr'); // post transforms // global (error reporting untested) foreach ($definition->info_attr_transform_post as $transform) { $attr = $transform->transform($o = $attr, $config, $context); if ($e) { if ($attr != $o) { $e->send(E_NOTICE, 'AttrValidator: Attributes transformed', $o, $attr); } } } // local (error reporting untested) foreach ($definition->info[$token->name]->attr_transform_post as $transform) { $attr = $transform->transform($o = $attr, $config, $context); if ($e) { if ($attr != $o) { $e->send(E_NOTICE, 'AttrValidator: Attributes transformed', $o, $attr); } } } $token->attr = $attr; // destroy CurrentToken if we made it ourselves if (!$current_token) { $context->destroy('CurrentToken'); } }
/** * Prepares the injector by giving it the config and context objects: * this allows references to important variables to be made within * the injector. This function also checks if the HTML environment * will work with the Injector (see checkNeeded()). * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return bool|string Boolean false if success, string of missing needed element/attribute if failure */ public function prepare($config, $context) { $this->htmlDefinition = $config->getHTMLDefinition(); // Even though this might fail, some unit tests ignore this and // still test checkNeeded, so be careful. Maybe get rid of that // dependency. $result = $this->checkNeeded($config); if ($result !== false) { return $result; } $this->currentNesting =& $context->get('CurrentNesting'); $this->currentToken =& $context->get('CurrentToken'); $this->inputZipper =& $context->get('InputZipper'); return false; }
/** * @param HTMLPurifier_Token[] $tokens * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return HTMLPurifier_Token[] * @throws HTMLPurifier_Exception */ public function execute($tokens, $config, $context) { $definition = $config->getHTMLDefinition(); // local variables $generator = new HTMLPurifier_Generator($config, $context); $escape_invalid_tags = $config->get('Core.EscapeInvalidTags'); // used for autoclose early abortion $global_parent_allowed_elements = $definition->info_parent_def->child->getAllowedElements($config); $e = $context->get('ErrorCollector', true); $i = false; // injector index list($zipper, $token) = HTMLPurifier_Zipper::fromArray($tokens); if ($token === NULL) { return array(); } $reprocess = false; // whether or not to reprocess the same token $stack = array(); // member variables $this->stack =& $stack; $this->tokens =& $tokens; $this->token =& $token; $this->zipper =& $zipper; $this->config = $config; $this->context = $context; // context variables $context->register('CurrentNesting', $stack); $context->register('InputZipper', $zipper); $context->register('CurrentToken', $token); // -- begin INJECTOR -- $this->injectors = array(); $injectors = $config->getBatch('AutoFormat'); $def_injectors = $definition->info_injector; $custom_injectors = $injectors['Custom']; unset($injectors['Custom']); // special case foreach ($injectors as $injector => $b) { // XXX: Fix with a legitimate lookup table of enabled filters if (strpos($injector, '.') !== false) { continue; } $injector = "HTMLPurifier_Injector_{$injector}"; if (!$b) { continue; } $this->injectors[] = new $injector(); } foreach ($def_injectors as $injector) { // assumed to be objects $this->injectors[] = $injector; } foreach ($custom_injectors as $injector) { if (!$injector) { continue; } if (is_string($injector)) { $injector = "HTMLPurifier_Injector_{$injector}"; $injector = new $injector(); } $this->injectors[] = $injector; } // give the injectors references to the definition and context // variables for performance reasons foreach ($this->injectors as $ix => $injector) { $error = $injector->prepare($config, $context); if (!$error) { continue; } array_splice($this->injectors, $ix, 1); // rm the injector trigger_error("Cannot enable {$injector->name} injector because {$error} is not allowed", E_USER_WARNING); } // -- end INJECTOR -- // a note on reprocessing: // In order to reduce code duplication, whenever some code needs // to make HTML changes in order to make things "correct", the // new HTML gets sent through the purifier, regardless of its // status. This means that if we add a start token, because it // was totally necessary, we don't have to update nesting; we just // punt ($reprocess = true; continue;) and it does that for us. // isset is in loop because $tokens size changes during loop exec for (;; $reprocess ? $reprocess = false : ($token = $zipper->next($token))) { // check for a rewind if (is_int($i)) { // possibility: disable rewinding if the current token has a // rewind set on it already. This would offer protection from // infinite loop, but might hinder some advanced rewinding. $rewind_offset = $this->injectors[$i]->getRewindOffset(); if (is_int($rewind_offset)) { for ($j = 0; $j < $rewind_offset; $j++) { if (empty($zipper->front)) { break; } $token = $zipper->prev($token); // indicate that other injectors should not process this token, // but we need to reprocess it unset($token->skip[$i]); $token->rewind = $i; if ($token instanceof HTMLPurifier_Token_Start) { array_pop($this->stack); } elseif ($token instanceof HTMLPurifier_Token_End) { $this->stack[] = $token->start; } } } $i = false; } // handle case of document end if ($token === NULL) { // kill processing if stack is empty if (empty($this->stack)) { break; } // peek $top_nesting = array_pop($this->stack); $this->stack[] = $top_nesting; // send error [TagClosedSuppress] if ($e && !isset($top_nesting->armor['MakeWellFormed_TagClosedError'])) { $e->send(E_NOTICE, 'Strategy_MakeWellFormed: Tag closed by document end', $top_nesting); } // append, don't splice, since this is the end $token = new HTMLPurifier_Token_End($top_nesting->name); // punt! $reprocess = true; continue; } //echo '<br>'; printZipper($zipper, $token);//printTokens($this->stack); //flush(); // quick-check: if it's not a tag, no need to process if (empty($token->is_tag)) { if ($token instanceof HTMLPurifier_Token_Text) { foreach ($this->injectors as $i => $injector) { if (isset($token->skip[$i])) { continue; } if ($token->rewind !== null && $token->rewind !== $i) { continue; } // XXX fuckup $r = $token; $injector->handleText($r); $token = $this->processToken($r, $i); $reprocess = true; break; } } // another possibility is a comment continue; } if (isset($definition->info[$token->name])) { $type = $definition->info[$token->name]->child->type; } else { $type = false; // Type is unknown, treat accordingly } // quick tag checks: anything that's *not* an end tag $ok = false; if ($type === 'empty' && $token instanceof HTMLPurifier_Token_Start) { // claims to be a start tag but is empty $token = new HTMLPurifier_Token_Empty($token->name, $token->attr, $token->line, $token->col, $token->armor); $ok = true; } elseif ($type && $type !== 'empty' && $token instanceof HTMLPurifier_Token_Empty) { // claims to be empty but really is a start tag // NB: this assignment is required $old_token = $token; $token = new HTMLPurifier_Token_End($token->name); $token = $this->insertBefore(new HTMLPurifier_Token_Start($old_token->name, $old_token->attr, $old_token->line, $old_token->col, $old_token->armor)); // punt (since we had to modify the input stream in a non-trivial way) $reprocess = true; continue; } elseif ($token instanceof HTMLPurifier_Token_Empty) { // real empty token $ok = true; } elseif ($token instanceof HTMLPurifier_Token_Start) { // start tag // ...unless they also have to close their parent if (!empty($this->stack)) { // Performance note: you might think that it's rather // inefficient, recalculating the autoclose information // for every tag that a token closes (since when we // do an autoclose, we push a new token into the // stream and then /process/ that, before // re-processing this token.) But this is // necessary, because an injector can make an // arbitrary transformations to the autoclosing // tokens we introduce, so things may have changed // in the meantime. Also, doing the inefficient thing is // "easy" to reason about (for certain perverse definitions // of "easy") $parent = array_pop($this->stack); $this->stack[] = $parent; $parent_def = null; $parent_elements = null; $autoclose = false; if (isset($definition->info[$parent->name])) { $parent_def = $definition->info[$parent->name]; $parent_elements = $parent_def->child->getAllowedElements($config); $autoclose = !isset($parent_elements[$token->name]); } if ($autoclose && $definition->info[$token->name]->wrap) { // Check if an element can be wrapped by another // element to make it valid in a context (for // example, <ul><ul> needs a <li> in between) $wrapname = $definition->info[$token->name]->wrap; $wrapdef = $definition->info[$wrapname]; $elements = $wrapdef->child->getAllowedElements($config); if (isset($elements[$token->name]) && isset($parent_elements[$wrapname])) { $newtoken = new HTMLPurifier_Token_Start($wrapname); $token = $this->insertBefore($newtoken); $reprocess = true; continue; } } $carryover = false; if ($autoclose && $parent_def->formatting) { $carryover = true; } if ($autoclose) { // check if this autoclose is doomed to fail // (this rechecks $parent, which his harmless) $autoclose_ok = isset($global_parent_allowed_elements[$token->name]); if (!$autoclose_ok) { foreach ($this->stack as $ancestor) { $elements = $definition->info[$ancestor->name]->child->getAllowedElements($config); if (isset($elements[$token->name])) { $autoclose_ok = true; break; } if ($definition->info[$token->name]->wrap) { $wrapname = $definition->info[$token->name]->wrap; $wrapdef = $definition->info[$wrapname]; $wrap_elements = $wrapdef->child->getAllowedElements($config); if (isset($wrap_elements[$token->name]) && isset($elements[$wrapname])) { $autoclose_ok = true; break; } } } } if ($autoclose_ok) { // errors need to be updated $new_token = new HTMLPurifier_Token_End($parent->name); $new_token->start = $parent; // [TagClosedSuppress] if ($e && !isset($parent->armor['MakeWellFormed_TagClosedError'])) { if (!$carryover) { $e->send(E_NOTICE, 'Strategy_MakeWellFormed: Tag auto closed', $parent); } else { $e->send(E_NOTICE, 'Strategy_MakeWellFormed: Tag carryover', $parent); } } if ($carryover) { $element = clone $parent; // [TagClosedAuto] $element->armor['MakeWellFormed_TagClosedError'] = true; $element->carryover = true; $token = $this->processToken(array($new_token, $token, $element)); } else { $token = $this->insertBefore($new_token); } } else { $token = $this->remove(); } $reprocess = true; continue; } } $ok = true; } if ($ok) { foreach ($this->injectors as $i => $injector) { if (isset($token->skip[$i])) { continue; } if ($token->rewind !== null && $token->rewind !== $i) { continue; } $r = $token; $injector->handleElement($r); $token = $this->processToken($r, $i); $reprocess = true; break; } if (!$reprocess) { // ah, nothing interesting happened; do normal processing if ($token instanceof HTMLPurifier_Token_Start) { $this->stack[] = $token; } elseif ($token instanceof HTMLPurifier_Token_End) { throw new HTMLPurifier_Exception('Improper handling of end tag in start code; possible error in MakeWellFormed'); } } continue; } // sanity check: we should be dealing with a closing tag if (!$token instanceof HTMLPurifier_Token_End) { throw new HTMLPurifier_Exception('Unaccounted for tag token in input stream, bug in HTML Purifier'); } // make sure that we have something open if (empty($this->stack)) { if ($escape_invalid_tags) { if ($e) { $e->send(E_WARNING, 'Strategy_MakeWellFormed: Unnecessary end tag to text'); } $token = new HTMLPurifier_Token_Text($generator->generateFromToken($token)); } else { if ($e) { $e->send(E_WARNING, 'Strategy_MakeWellFormed: Unnecessary end tag removed'); } $token = $this->remove(); } $reprocess = true; continue; } // first, check for the simplest case: everything closes neatly. // Eventually, everything passes through here; if there are problems // we modify the input stream accordingly and then punt, so that // the tokens get processed again. $current_parent = array_pop($this->stack); if ($current_parent->name == $token->name) { $token->start = $current_parent; foreach ($this->injectors as $i => $injector) { if (isset($token->skip[$i])) { continue; } if ($token->rewind !== null && $token->rewind !== $i) { continue; } $r = $token; $injector->handleEnd($r); $token = $this->processToken($r, $i); $this->stack[] = $current_parent; $reprocess = true; break; } continue; } // okay, so we're trying to close the wrong tag // undo the pop previous pop $this->stack[] = $current_parent; // scroll back the entire nest, trying to find our tag. // (feature could be to specify how far you'd like to go) $size = count($this->stack); // -2 because -1 is the last element, but we already checked that $skipped_tags = false; for ($j = $size - 2; $j >= 0; $j--) { if ($this->stack[$j]->name == $token->name) { $skipped_tags = array_slice($this->stack, $j); break; } } // we didn't find the tag, so remove if ($skipped_tags === false) { if ($escape_invalid_tags) { if ($e) { $e->send(E_WARNING, 'Strategy_MakeWellFormed: Stray end tag to text'); } $token = new HTMLPurifier_Token_Text($generator->generateFromToken($token)); } else { if ($e) { $e->send(E_WARNING, 'Strategy_MakeWellFormed: Stray end tag removed'); } $token = $this->remove(); } $reprocess = true; continue; } // do errors, in REVERSE $j order: a,b,c with </a></b></c> $c = count($skipped_tags); if ($e) { for ($j = $c - 1; $j > 0; $j--) { // notice we exclude $j == 0, i.e. the current ending tag, from // the errors... [TagClosedSuppress] if (!isset($skipped_tags[$j]->armor['MakeWellFormed_TagClosedError'])) { $e->send(E_NOTICE, 'Strategy_MakeWellFormed: Tag closed by element end', $skipped_tags[$j]); } } } // insert tags, in FORWARD $j order: c,b,a with </a></b></c> $replace = array($token); for ($j = 1; $j < $c; $j++) { // ...as well as from the insertions $new_token = new HTMLPurifier_Token_End($skipped_tags[$j]->name); $new_token->start = $skipped_tags[$j]; array_unshift($replace, $new_token); if (isset($definition->info[$new_token->name]) && $definition->info[$new_token->name]->formatting) { // [TagClosedAuto] $element = clone $skipped_tags[$j]; $element->carryover = true; $element->armor['MakeWellFormed_TagClosedError'] = true; $replace[] = $element; } } $token = $this->processToken($replace); $reprocess = true; continue; } $context->destroy('CurrentToken'); $context->destroy('CurrentNesting'); $context->destroy('InputZipper'); unset($this->injectors, $this->stack, $this->tokens); return $zipper->toArray($token); }
/** * @param HTMLPurifier_Token[] $tokens * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return array|HTMLPurifier_Token[] */ public function execute($tokens, $config, $context) { $definition = $config->getHTMLDefinition(); $generator = new HTMLPurifier_Generator($config, $context); $result = array(); $escape_invalid_tags = $config->get('Core.EscapeInvalidTags'); $remove_invalid_img = $config->get('Core.RemoveInvalidImg'); // currently only used to determine if comments should be kept $trusted = $config->get('HTML.Trusted'); $comment_lookup = $config->get('HTML.AllowedComments'); $comment_regexp = $config->get('HTML.AllowedCommentsRegexp'); $check_comments = $comment_lookup !== array() || $comment_regexp !== null; $remove_script_contents = $config->get('Core.RemoveScriptContents'); $hidden_elements = $config->get('Core.HiddenElements'); // remove script contents compatibility if ($remove_script_contents === true) { $hidden_elements['script'] = true; } elseif ($remove_script_contents === false && isset($hidden_elements['script'])) { unset($hidden_elements['script']); } $attr_validator = new HTMLPurifier_AttrValidator(); // removes tokens until it reaches a closing tag with its value $remove_until = false; // converts comments into text tokens when this is equal to a tag name $textify_comments = false; $token = false; $context->register('CurrentToken', $token); $e = false; if ($config->get('Core.CollectErrors')) { $e =& $context->get('ErrorCollector'); } foreach ($tokens as $token) { if ($remove_until) { if (empty($token->is_tag) || $token->name !== $remove_until) { continue; } } if (!empty($token->is_tag)) { // DEFINITION CALL // before any processing, try to transform the element if (isset($definition->info_tag_transform[$token->name])) { $original_name = $token->name; // there is a transformation for this tag // DEFINITION CALL $token = $definition->info_tag_transform[$token->name]->transform($token, $config, $context); if ($e) { $e->send(E_NOTICE, 'Strategy_RemoveForeignElements: Tag transform', $original_name); } } if (isset($definition->info[$token->name])) { // mostly everything's good, but // we need to make sure required attributes are in order if (($token instanceof HTMLPurifier_Token_Start || $token instanceof HTMLPurifier_Token_Empty) && $definition->info[$token->name]->required_attr && ($token->name != 'img' || $remove_invalid_img)) { $attr_validator->validateToken($token, $config, $context); $ok = true; foreach ($definition->info[$token->name]->required_attr as $name) { if (!isset($token->attr[$name])) { $ok = false; break; } } if (!$ok) { if ($e) { $e->send(E_ERROR, 'Strategy_RemoveForeignElements: Missing required attribute', $name); } continue; } $token->armor['ValidateAttributes'] = true; } if (isset($hidden_elements[$token->name]) && $token instanceof HTMLPurifier_Token_Start) { $textify_comments = $token->name; } elseif ($token->name === $textify_comments && $token instanceof HTMLPurifier_Token_End) { $textify_comments = false; } } elseif ($escape_invalid_tags) { // invalid tag, generate HTML representation and insert in if ($e) { $e->send(E_WARNING, 'Strategy_RemoveForeignElements: Foreign element to text'); } $token = new HTMLPurifier_Token_Text($generator->generateFromToken($token)); } else { // check if we need to destroy all of the tag's children // CAN BE GENERICIZED if (isset($hidden_elements[$token->name])) { if ($token instanceof HTMLPurifier_Token_Start) { $remove_until = $token->name; } elseif ($token instanceof HTMLPurifier_Token_Empty) { // do nothing: we're still looking } else { $remove_until = false; } if ($e) { $e->send(E_ERROR, 'Strategy_RemoveForeignElements: Foreign meta element removed'); } } else { if ($e) { $e->send(E_ERROR, 'Strategy_RemoveForeignElements: Foreign element removed'); } } continue; } } elseif ($token instanceof HTMLPurifier_Token_Comment) { // textify comments in script tags when they are allowed if ($textify_comments !== false) { $data = $token->data; $token = new HTMLPurifier_Token_Text($data); } elseif ($trusted || $check_comments) { // always cleanup comments $trailing_hyphen = false; if ($e) { // perform check whether or not there's a trailing hyphen if (substr($token->data, -1) == '-') { $trailing_hyphen = true; } } $token->data = rtrim($token->data, '-'); $found_double_hyphen = false; while (strpos($token->data, '--') !== false) { $found_double_hyphen = true; $token->data = str_replace('--', '-', $token->data); } if ($trusted || !empty($comment_lookup[trim($token->data)]) || $comment_regexp !== null && preg_match($comment_regexp, trim($token->data))) { // OK good if ($e) { if ($trailing_hyphen) { $e->send(E_NOTICE, 'Strategy_RemoveForeignElements: Trailing hyphen in comment removed'); } if ($found_double_hyphen) { $e->send(E_NOTICE, 'Strategy_RemoveForeignElements: Hyphens in comment collapsed'); } } } else { if ($e) { $e->send(E_NOTICE, 'Strategy_RemoveForeignElements: Comment removed'); } continue; } } else { // strip comments if ($e) { $e->send(E_NOTICE, 'Strategy_RemoveForeignElements: Comment removed'); } continue; } } elseif ($token instanceof HTMLPurifier_Token_Text) { } else { continue; } $result[] = $token; } if ($remove_until && $e) { // we removed tokens until the end, throw error $e->send(E_ERROR, 'Strategy_RemoveForeignElements: Token removed to end', $remove_until); } $context->destroy('CurrentToken'); return $result; }
/** * Takes a piece of HTML and normalizes it by converting entities, fixing * encoding, extracting bits, and other good stuff. * @param string $html HTML. * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return string * @todo Consider making protected */ public function normalize($html, $config, $context) { // normalize newlines to \n if ($config->get('Core.NormalizeNewlines')) { $html = str_replace("\r\n", "\n", $html); $html = str_replace("\r", "\n", $html); } if ($config->get('HTML.Trusted')) { // escape convoluted CDATA $html = $this->escapeCommentedCDATA($html); } // escape CDATA $html = $this->escapeCDATA($html); $html = $this->removeIEConditional($html); // extract body from document if applicable if ($config->get('Core.ConvertDocumentToFragment')) { $e = false; if ($config->get('Core.CollectErrors')) { $e =& $context->get('ErrorCollector'); } $new_html = $this->extractBody($html); if ($e && $new_html != $html) { $e->send(E_WARNING, 'Lexer: Extracted body'); } $html = $new_html; } // expand entities that aren't the big five $html = $this->_entity_parser->substituteNonSpecialEntities($html); // clean into wellformed UTF-8 string for an SGML context: this has // to be done after entity expansion because the entities sometimes // represent non-SGML characters (horror, horror!) $html = HTMLPurifier_Encoder::cleanUTF8($html); // if processing instructions are to removed, remove them now if ($config->get('Core.RemoveProcessingInstructions')) { $html = preg_replace('#<\\?.+?\\?>#s', '', $html); } return $html; }
/** * @param HTMLPurifier_URI $uri * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context */ protected function makeReplace($uri, $config, $context) { $string = $uri->toString(); // always available $this->replace['%s'] = $string; $this->replace['%r'] = $context->get('EmbeddedURI', true); $token = $context->get('CurrentToken', true); $this->replace['%n'] = $token ? $token->name : null; $this->replace['%m'] = $context->get('CurrentAttr', true); $this->replace['%p'] = $context->get('CurrentCSSProperty', true); // not always available if ($this->secretKey) { $this->replace['%t'] = hash_hmac("sha256", $string, $this->secretKey); } }
/** * @param HTMLPurifier_URI $uri * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return bool */ public function filter(&$uri, $config, $context) { return !$context->get('EmbeddedURI', true); }
/** * Takes the inside of an HTML tag and makes an assoc array of attributes. * * @param string $string Inside of tag excluding name. * @param HTMLPurifier_Config $config * @param HTMLPurifier_Context $context * @return array Assoc array of attributes. */ public function parseAttributeString($string, $config, $context) { $string = (string) $string; // quick typecast if ($string == '') { return array(); } // no attributes $e = false; if ($config->get('Core.CollectErrors')) { $e =& $context->get('ErrorCollector'); } // let's see if we can abort as quickly as possible // one equal sign, no spaces => one attribute $num_equal = substr_count($string, '='); $has_space = strpos($string, ' '); if ($num_equal === 0 && !$has_space) { // bool attribute return array($string => $string); } elseif ($num_equal === 1 && !$has_space) { // only one attribute list($key, $quoted_value) = explode('=', $string); $quoted_value = trim($quoted_value); if (!$key) { if ($e) { $e->send(E_ERROR, 'Lexer: Missing attribute key'); } return array(); } if (!$quoted_value) { return array($key => ''); } $first_char = @$quoted_value[0]; $last_char = @$quoted_value[strlen($quoted_value) - 1]; $same_quote = $first_char == $last_char; $open_quote = $first_char == '"' || $first_char == "'"; if ($same_quote && $open_quote) { // well behaved $value = substr($quoted_value, 1, strlen($quoted_value) - 2); } else { // not well behaved if ($open_quote) { if ($e) { $e->send(E_ERROR, 'Lexer: Missing end quote'); } $value = substr($quoted_value, 1); } else { $value = $quoted_value; } } if ($value === false) { $value = ''; } return array($key => $this->parseData($value)); } // setup loop environment $array = array(); // return assoc array of attributes $cursor = 0; // current position in string (moves forward) $size = strlen($string); // size of the string (stays the same) // if we have unquoted attributes, the parser expects a terminating // space, so let's guarantee that there's always a terminating space. $string .= ' '; $old_cursor = -1; while ($cursor < $size) { if ($old_cursor >= $cursor) { throw new Exception("Infinite loop detected"); } $old_cursor = $cursor; $cursor += $value = strspn($string, $this->_whitespace, $cursor); // grab the key $key_begin = $cursor; //we're currently at the start of the key // scroll past all characters that are the key (not whitespace or =) $cursor += strcspn($string, $this->_whitespace . '=', $cursor); $key_end = $cursor; // now at the end of the key $key = substr($string, $key_begin, $key_end - $key_begin); if (!$key) { if ($e) { $e->send(E_ERROR, 'Lexer: Missing attribute key'); } $cursor += 1 + strcspn($string, $this->_whitespace, $cursor + 1); // prevent infinite loop continue; // empty key } // scroll past all whitespace $cursor += strspn($string, $this->_whitespace, $cursor); if ($cursor >= $size) { $array[$key] = $key; break; } // if the next character is an equal sign, we've got a regular // pair, otherwise, it's a bool attribute $first_char = @$string[$cursor]; if ($first_char == '=') { // key="value" $cursor++; $cursor += strspn($string, $this->_whitespace, $cursor); if ($cursor === false) { $array[$key] = ''; break; } // we might be in front of a quote right now $char = @$string[$cursor]; if ($char == '"' || $char == "'") { // it's quoted, end bound is $char $cursor++; $value_begin = $cursor; $cursor = strpos($string, $char, $cursor); $value_end = $cursor; } else { // it's not quoted, end bound is whitespace $value_begin = $cursor; $cursor += strcspn($string, $this->_whitespace, $cursor); $value_end = $cursor; } // we reached a premature end if ($cursor === false) { $cursor = $size; $value_end = $cursor; } $value = substr($string, $value_begin, $value_end - $value_begin); if ($value === false) { $value = ''; } $array[$key] = $this->parseData($value); $cursor++; } else { // boolattr if ($key !== '') { $array[$key] = $key; } else { // purely theoretical if ($e) { $e->send(E_ERROR, 'Lexer: Missing attribute key'); } } } } return $array; }