From 6854cb3f4d8219cf1829e32122eb2502a916eae9 Mon Sep 17 00:00:00 2001 From: Andreas Baumann Date: Sat, 1 Feb 2020 09:05:48 +0100 Subject: initial checkin --- .../library/HTMLPurifier/Strategy/Composite.php | 30 + .../library/HTMLPurifier/Strategy/Core.php | 17 + .../library/HTMLPurifier/Strategy/FixNesting.php | 181 ++++++ .../HTMLPurifier/Strategy/MakeWellFormed.php | 659 +++++++++++++++++++++ .../Strategy/RemoveForeignElements.php | 207 +++++++ .../HTMLPurifier/Strategy/ValidateAttributes.php | 45 ++ 6 files changed, 1139 insertions(+) create mode 100644 vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/Composite.php create mode 100644 vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/Core.php create mode 100644 vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/FixNesting.php create mode 100644 vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/MakeWellFormed.php create mode 100644 vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/RemoveForeignElements.php create mode 100644 vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/ValidateAttributes.php (limited to 'vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy') diff --git a/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/Composite.php b/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/Composite.php new file mode 100644 index 0000000..d7d35ce --- /dev/null +++ b/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/Composite.php @@ -0,0 +1,30 @@ +strategies as $strategy) { + $tokens = $strategy->execute($tokens, $config, $context); + } + return $tokens; + } +} + +// vim: et sw=4 sts=4 diff --git a/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/Core.php b/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/Core.php new file mode 100644 index 0000000..4414c17 --- /dev/null +++ b/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/Core.php @@ -0,0 +1,17 @@ +strategies[] = new HTMLPurifier_Strategy_RemoveForeignElements(); + $this->strategies[] = new HTMLPurifier_Strategy_MakeWellFormed(); + $this->strategies[] = new HTMLPurifier_Strategy_FixNesting(); + $this->strategies[] = new HTMLPurifier_Strategy_ValidateAttributes(); + } +} + +// vim: et sw=4 sts=4 diff --git a/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/FixNesting.php b/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/FixNesting.php new file mode 100644 index 0000000..6fa673d --- /dev/null +++ b/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/FixNesting.php @@ -0,0 +1,181 @@ +getHTMLDefinition(); + + $excludes_enabled = !$config->get('Core.DisableExcludes'); + + // setup the context variable 'IsInline', for chameleon processing + // is 'false' when we are not inline, 'true' when it must always + // be inline, and an integer when it is inline for a certain + // branch of the document tree + $is_inline = $definition->info_parent_def->descendants_are_inline; + $context->register('IsInline', $is_inline); + + // setup error collector + $e =& $context->get('ErrorCollector', true); + + //####################################################################// + // Loop initialization + + // stack that contains all elements that are excluded + // it is organized by parent elements, similar to $stack, + // but it is only populated when an element with exclusions is + // processed, i.e. there won't be empty exclusions. + $exclude_stack = array($definition->info_parent_def->excludes); + + // variable that contains the start token while we are processing + // nodes. This enables error reporting to do its job + $node = $top_node; + // dummy token + list($token, $d) = $node->toTokenPair(); + $context->register('CurrentNode', $node); + $context->register('CurrentToken', $token); + + //####################################################################// + // Loop + + // We need to implement a post-order traversal iteratively, to + // avoid running into stack space limits. This is pretty tricky + // to reason about, so we just manually stack-ify the recursive + // variant: + // + // function f($node) { + // foreach ($node->children as $child) { + // f($child); + // } + // validate($node); + // } + // + // Thus, we will represent a stack frame as array($node, + // $is_inline, stack of children) + // e.g. array_reverse($node->children) - already processed + // children. + + $parent_def = $definition->info_parent_def; + $stack = array( + array($top_node, + $parent_def->descendants_are_inline, + $parent_def->excludes, // exclusions + 0) + ); + + while (!empty($stack)) { + list($node, $is_inline, $excludes, $ix) = array_pop($stack); + // recursive call + $go = false; + $def = empty($stack) ? $definition->info_parent_def : $definition->info[$node->name]; + while (isset($node->children[$ix])) { + $child = $node->children[$ix++]; + if ($child instanceof HTMLPurifier_Node_Element) { + $go = true; + $stack[] = array($node, $is_inline, $excludes, $ix); + $stack[] = array($child, + // ToDo: I don't think it matters if it's def or + // child_def, but double check this... + $is_inline || $def->descendants_are_inline, + empty($def->excludes) ? $excludes + : array_merge($excludes, $def->excludes), + 0); + break; + } + }; + if ($go) continue; + list($token, $d) = $node->toTokenPair(); + // base case + if ($excludes_enabled && isset($excludes[$node->name])) { + $node->dead = true; + if ($e) $e->send(E_ERROR, 'Strategy_FixNesting: Node excluded'); + } else { + // XXX I suppose it would be slightly more efficient to + // avoid the allocation here and have children + // strategies handle it + $children = array(); + foreach ($node->children as $child) { + if (!$child->dead) $children[] = $child; + } + $result = $def->child->validateChildren($children, $config, $context); + if ($result === true) { + // nop + $node->children = $children; + } elseif ($result === false) { + $node->dead = true; + if ($e) $e->send(E_ERROR, 'Strategy_FixNesting: Node removed'); + } else { + $node->children = $result; + if ($e) { + // XXX This will miss mutations of internal nodes. Perhaps defer to the child validators + if (empty($result) && !empty($children)) { + $e->send(E_ERROR, 'Strategy_FixNesting: Node contents removed'); + } else if ($result != $children) { + $e->send(E_WARNING, 'Strategy_FixNesting: Node reorganized'); + } + } + } + } + } + + //####################################################################// + // Post-processing + + // remove context variables + $context->destroy('IsInline'); + $context->destroy('CurrentNode'); + $context->destroy('CurrentToken'); + + //####################################################################// + // Return + + return HTMLPurifier_Arborize::flatten($node, $config, $context); + } +} + +// vim: et sw=4 sts=4 diff --git a/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/MakeWellFormed.php b/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/MakeWellFormed.php new file mode 100644 index 0000000..a6eb09e --- /dev/null +++ b/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Strategy/MakeWellFormed.php @@ -0,0 +1,659 @@ +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 (;; + // only increment if we don't need to reprocess + $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. See Note [Injector skips] + 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! + $reprocess = true; + continue; + } + + //echo '
'; printZipper($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[$i])) { + // See Note [Injector skips] + continue; + } + if ($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,