From a0052f0176bd079e6a94baec59fea2ec5a8d651e Mon Sep 17 00:00:00 2001 From: friendica Date: Thu, 1 Jan 2015 22:18:27 -0800 Subject: htmlpurifier update - compatibility issue with language library autoloader --- library/HTMLPurifier/Strategy/MakeWellFormed.php | 379 +++++++++++++++-------- 1 file changed, 252 insertions(+), 127 deletions(-) (limited to 'library/HTMLPurifier/Strategy/MakeWellFormed.php') diff --git a/library/HTMLPurifier/Strategy/MakeWellFormed.php b/library/HTMLPurifier/Strategy/MakeWellFormed.php index c73658400..e389e0011 100644 --- a/library/HTMLPurifier/Strategy/MakeWellFormed.php +++ b/library/HTMLPurifier/Strategy/MakeWellFormed.php @@ -2,66 +2,97 @@ /** * Takes tokens makes them well-formed (balance end tags, etc.) + * + * Specification of the armor attributes this strategy uses: + * + * - MakeWellFormed_TagClosedError: This armor field is used to + * suppress tag closed errors for certain tokens [TagClosedSuppress], + * in particular, if a tag was generated automatically by HTML + * Purifier, we may rely on our infrastructure to close it for us + * and shouldn't report an error to the user [TagClosedAuto]. */ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy { /** * Array stream of tokens being processed. + * @type HTMLPurifier_Token[] */ protected $tokens; /** - * Current index in $tokens. + * Current token. + * @type HTMLPurifier_Token */ - protected $t; + protected $token; + + /** + * Zipper managing the true state. + * @type HTMLPurifier_Zipper + */ + protected $zipper; /** * Current nesting of elements. + * @type array */ protected $stack; /** * Injectors active in this stream processing. + * @type HTMLPurifier_Injector[] */ protected $injectors; /** * Current instance of HTMLPurifier_Config. + * @type HTMLPurifier_Config */ protected $config; /** * Current instance of HTMLPurifier_Context. + * @type HTMLPurifier_Context */ protected $context; - public function execute($tokens, $config, $context) { - + /** + * @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); - $t = false; // token index $i = false; // injector index - $token = false; // the current token - $reprocess = false; // whether or not to reprocess the same token + list($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->t =& $t; - $this->tokens =& $tokens; - $this->config = $config; + $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('InputIndex', $t); - $context->register('InputTokens', $tokens); - $context->register('CurrentToken', $token); + $context->register('InputZipper', $zipper); + $context->register('CurrentToken', $token); // -- begin INJECTOR -- @@ -73,9 +104,13 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy unset($injectors['Custom']); // special case foreach ($injectors as $injector => $b) { // XXX: Fix with a legitimate lookup table of enabled filters - if (strpos($injector, '.') !== false) continue; + if (strpos($injector, '.') !== false) { + continue; + } $injector = "HTMLPurifier_Injector_$injector"; - if (!$b) continue; + if (!$b) { + continue; + } $this->injectors[] = new $injector; } foreach ($def_injectors as $injector) { @@ -83,7 +118,9 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy $this->injectors[] = $injector; } foreach ($custom_injectors as $injector) { - if (!$injector) continue; + if (!$injector) { + continue; + } if (is_string($injector)) { $injector = "HTMLPurifier_Injector_$injector"; $injector = new $injector; @@ -95,14 +132,16 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy // variables for performance reasons foreach ($this->injectors as $ix => $injector) { $error = $injector->prepare($config, $context); - if (!$error) continue; + if (!$error) { + continue; + } array_splice($this->injectors, $ix, 1); // rm the injector trigger_error("Cannot enable {$injector->name} injector because $error is not allowed", E_USER_WARNING); } // -- end INJECTOR -- - // a note on punting: + // a note on reprocessing: // In order to reduce code duplication, whenever some code needs // to make HTML changes in order to make things "correct", the // new HTML gets sent through the purifier, regardless of its @@ -111,70 +150,75 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy // punt ($reprocess = true; continue;) and it does that for us. // isset is in loop because $tokens size changes during loop exec - for ( - $t = 0; - $t == 0 || isset($tokens[$t - 1]); - // only increment if we don't need to reprocess - $reprocess ? $reprocess = false : $t++ - ) { + for (;; + // only increment if we don't need to reprocess + $reprocess ? $reprocess = false : $token = $zipper->next($token)) { // check for a rewind - if (is_int($i) && $i >= 0) { + if (is_int($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_to = $this->injectors[$i]->getRewind(); - if (is_int($rewind_to) && $rewind_to < $t) { - if ($rewind_to < 0) $rewind_to = 0; - while ($t > $rewind_to) { - $t--; - $prev = $tokens[$t]; + $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($prev->skip[$i]); - $prev->rewind = $i; - if ($prev instanceof HTMLPurifier_Token_Start) array_pop($this->stack); - elseif ($prev instanceof HTMLPurifier_Token_End) $this->stack[] = $prev->start; + unset($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; + } } } $i = false; } // handle case of document end - if (!isset($tokens[$t])) { + if ($token === NULL) { // kill processing if stack is empty - if (empty($this->stack)) break; + if (empty($this->stack)) { + break; + } // peek $top_nesting = array_pop($this->stack); $this->stack[] = $top_nesting; - // send error + // send error [TagClosedSuppress] if ($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 - $tokens[] = new HTMLPurifier_Token_End($top_nesting->name); + $token = new HTMLPurifier_Token_End($top_nesting->name); // punt! $reprocess = true; continue; } - $token = $tokens[$t]; - - //echo '
'; printTokens($tokens, $t); printTokens($this->stack); + //echo '
'; 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; - $injector->handleText($token); - $this->processToken($token, $i); + if (isset($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; } @@ -193,12 +237,22 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy $ok = false; if ($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 = 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 - $this->swap(new HTMLPurifier_Token_End($token->name)); - $this->insertBefore(new HTMLPurifier_Token_Start($token->name, $token->attr)); + // 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) $reprocess = true; continue; @@ -211,55 +265,96 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy // ...unless they also have to close their parent if (!empty($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])) { - $elements = $definition->info[$parent->name]->child->getAllowedElements($config); - $autoclose = !isset($elements[$token->name]); - } else { - $autoclose = false; + $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 + // Check if an element can be wrapped by another + // element to make it valid in a context (for // example,