MDL-9151 HTML Purifier cleaning support - enable switch is in experimental section

MDL-9435 Reviewved url cleaning in redirect()
This commit is contained in:
skodak
2007-04-18 21:52:03 +00:00
parent c85607f0be
commit e0ac8448c7
129 changed files with 10389 additions and 20 deletions
@@ -0,0 +1,30 @@
<?php
require_once 'HTMLPurifier/Strategy.php';
require_once 'HTMLPurifier/Config.php';
/**
* Composite strategy that runs multiple strategies on tokens.
*/
class HTMLPurifier_Strategy_Composite extends HTMLPurifier_Strategy
{
/**
* List of strategies to run tokens through.
*/
var $strategies = array();
function HTMLPurifier_Strategy_Composite() {
trigger_error('Attempt to instantiate abstract object', E_USER_ERROR);
}
function execute($tokens, $config, &$context) {
foreach ($this->strategies as $strategy) {
$tokens = $strategy->execute($tokens, $config, $context);
}
return $tokens;
}
}
?>
@@ -0,0 +1,25 @@
<?php
require_once 'HTMLPurifier/Strategy/Composite.php';
require_once 'HTMLPurifier/Strategy/RemoveForeignElements.php';
require_once 'HTMLPurifier/Strategy/MakeWellFormed.php';
require_once 'HTMLPurifier/Strategy/FixNesting.php';
require_once 'HTMLPurifier/Strategy/ValidateAttributes.php';
/**
* Core strategy composed of the big four strategies.
*/
class HTMLPurifier_Strategy_Core extends HTMLPurifier_Strategy_Composite
{
function HTMLPurifier_Strategy_Core() {
$this->strategies[] = new HTMLPurifier_Strategy_RemoveForeignElements();
$this->strategies[] = new HTMLPurifier_Strategy_MakeWellFormed();
$this->strategies[] = new HTMLPurifier_Strategy_FixNesting();
$this->strategies[] = new HTMLPurifier_Strategy_ValidateAttributes();
}
}
?>
@@ -0,0 +1,292 @@
<?php
require_once 'HTMLPurifier/Strategy.php';
require_once 'HTMLPurifier/HTMLDefinition.php';
/**
* Takes a well formed list of tokens and fixes their nesting.
*
* HTML elements dictate which elements are allowed to be their children,
* for example, you can't have a p tag in a span tag. Other elements have
* much more rigorous definitions: tables, for instance, require a specific
* order for their elements. There are also constraints not expressible by
* document type definitions, such as the chameleon nature of ins/del
* tags and global child exclusions.
*
* The first major objective of this strategy is to iterate through all the
* nodes (not tokens) of the list of tokens and determine whether or not
* their children conform to the element's definition. If they do not, the
* child definition may optionally supply an amended list of elements that
* is valid or require that the entire node be deleted (and the previous
* node rescanned).
*
* The second objective is to ensure that explicitly excluded elements of
* an element do not appear in its children. Code that accomplishes this
* task is pervasive through the strategy, though the two are distinct tasks
* and could, theoretically, be seperated (although it's not recommended).
*
* @note Whether or not unrecognized children are silently dropped or
* translated into text depends on the child definitions.
*
* @todo Enable nodes to be bubbled out of the structure.
*/
class HTMLPurifier_Strategy_FixNesting extends HTMLPurifier_Strategy
{
function execute($tokens, $config, &$context) {
//####################################################################//
// Pre-processing
// get a copy of the HTML definition
$definition = $config->getHTMLDefinition();
// insert implicit "parent" node, will be removed at end.
// ! we might want to move this to configuration
// DEFINITION CALL
$parent_name = $definition->info_parent;
array_unshift($tokens, new HTMLPurifier_Token_Start($parent_name));
$tokens[] = new HTMLPurifier_Token_End($parent_name);
// setup the context variables
$is_inline = false; // reference var that we alter
$context->register('IsInline', $is_inline);
//####################################################################//
// Loop initialization
// stack that contains the indexes of all parents,
// $stack[count($stack)-1] being the current parent
$stack = array();
// stack that contains all elements that are excluded
// same structure as $stack, but it is only populated when an element
// with exclusions is processed, i.e. there won't be empty exclusions.
$exclude_stack = array();
//####################################################################//
// Loop
// iterate through all start nodes. Determining the start node
// is complicated so it has been omitted from the loop construct
for ($i = 0, $size = count($tokens) ; $i < $size; ) {
//################################################################//
// Gather information on children
// child token accumulator
$child_tokens = array();
// scroll to the end of this node, report number, and collect
// all children
for ($j = $i, $depth = 0; ; $j++) {
if ($tokens[$j]->type == 'start') {
$depth++;
// skip token assignment on first iteration, this is the
// token we currently are on
if ($depth == 1) continue;
} elseif ($tokens[$j]->type == 'end') {
$depth--;
// skip token assignment on last iteration, this is the
// end token of the token we're currently on
if ($depth == 0) break;
}
$child_tokens[] = $tokens[$j];
}
// $i is index of start token
// $j is index of end token
//################################################################//
// Gather information on parent
// calculate parent information
if ($count = count($stack)) {
$parent_index = $stack[$count-1];
$parent_name = $tokens[$parent_index]->name;
if ($parent_index == 0) {
$parent_def = $definition->info_parent_def;
} else {
$parent_def = $definition->info[$parent_name];
}
} else {
// unknown info, it won't be used anyway
$parent_index = $parent_name = $parent_def = null;
}
// calculate context
if ($is_inline === false) {
// check if conditions make it inline
if (!empty($parent_def) && $parent_def->descendants_are_inline) {
$is_inline = $count - 1;
}
} else {
// check if we're out of inline
if ($count === $is_inline) {
$is_inline = false;
}
}
//################################################################//
// Determine whether element is explicitly excluded SGML-style
// determine whether or not element is excluded by checking all
// parent exclusions. The array should not be very large, two
// elements at most.
$excluded = false;
if (!empty($exclude_stack)) {
foreach ($exclude_stack as $lookup) {
if (isset($lookup[$tokens[$i]->name])) {
$excluded = true;
// no need to continue processing
break;
}
}
}
//################################################################//
// Perform child validation
if ($excluded) {
// there is an exclusion, remove the entire node
$result = false;
$excludes = array(); // not used, but good to initialize anyway
} else {
// DEFINITION CALL
if ($i === 0) {
// special processing for the first node
$def = $definition->info_parent_def;
} else {
$def = $definition->info[$tokens[$i]->name];
}
if (!empty($def->child)) {
// have DTD child def validate children
$result = $def->child->validateChildren(
$child_tokens, $config, $context);
} else {
// weird, no child definition, get rid of everything
$result = false;
}
// determine whether or not this element has any exclusions
$excludes = $def->excludes;
}
// $result is now a bool or array
//################################################################//
// Process result by interpreting $result
if ($result === true) {
// leave the node as is
// register start token as a parental node start
$stack[] = $i;
// register exclusions if there are any
if (!empty($excludes)) $exclude_stack[] = $excludes;
// move cursor to next possible start node
$i++;
} elseif($result === false) {
// remove entire node
// calculate length of inner tokens and current tokens
$length = $j - $i + 1;
// perform removal
array_splice($tokens, $i, $length);
// update size
$size -= $length;
// there is no start token to register,
// current node is now the next possible start node
// unless it turns out that we need to do a double-check
if (!$parent_def->child->allow_empty) {
// we need to do a double-check
$i = $parent_index;
array_pop($stack);
}
// PROJECTED OPTIMIZATION: Process all children elements before
// reprocessing parent node.
} else {
// replace node with $result
// calculate length of inner tokens
$length = $j - $i - 1;
// perform replacement
array_splice($tokens, $i + 1, $length, $result);
// update size
$size -= $length;
$size += count($result);
// register start token as a parental node start
$stack[] = $i;
// register exclusions if there are any
if (!empty($excludes)) $exclude_stack[] = $excludes;
// move cursor to next possible start node
$i++;
}
//################################################################//
// Scroll to next start node
// We assume, at this point, that $i is the index of the token
// that is the first possible new start point for a node.
// Test if the token indeed is a start tag, if not, move forward
// and test again.
$size = count($tokens);
while ($i < $size and $tokens[$i]->type != 'start') {
if ($tokens[$i]->type == 'end') {
// pop a token index off the stack if we ended a node
array_pop($stack);
// pop an exclusion lookup off exclusion stack if
// we ended node and that node had exclusions
if ($i == 0 || $i == $size - 1) {
// use specialized var if it's the super-parent
$s_excludes = $definition->info_parent_def->excludes;
} else {
$s_excludes = $definition->info[$tokens[$i]->name]->excludes;
}
if ($s_excludes) {
array_pop($exclude_stack);
}
}
$i++;
}
}
//####################################################################//
// Post-processing
// remove implicit parent tokens at the beginning and end
array_shift($tokens);
array_pop($tokens);
// remove context variables
$context->destroy('IsInline');
//####################################################################//
// Return
return $tokens;
}
}
?>
@@ -0,0 +1,158 @@
<?php
require_once 'HTMLPurifier/Strategy.php';
require_once 'HTMLPurifier/HTMLDefinition.php';
require_once 'HTMLPurifier/Generator.php';
/**
* Takes tokens makes them well-formed (balance end tags, etc.)
*/
class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy
{
function execute($tokens, $config, &$context) {
$definition = $config->getHTMLDefinition();
$generator = new HTMLPurifier_Generator();
$result = array();
$current_nesting = array();
$escape_invalid_tags = $config->get('Core', 'EscapeInvalidTags');
foreach ($tokens as $token) {
if (empty( $token->is_tag )) {
$result[] = $token;
continue;
}
// DEFINITION CALL
$info = $definition->info[$token->name]->child;
// test if it claims to be a start tag but is empty
if ($info->type == 'empty' &&
$token->type == 'start' ) {
$result[] = new HTMLPurifier_Token_Empty($token->name,
$token->attr);
continue;
}
// test if it claims to be empty but really is a start tag
if ($info->type != 'empty' &&
$token->type == 'empty' ) {
$result[] = new HTMLPurifier_Token_Start($token->name,
$token->attr);
$result[] = new HTMLPurifier_Token_End($token->name);
continue;
}
// automatically insert empty tags
if ($token->type == 'empty') {
$result[] = $token;
continue;
}
// we give start tags precedence, so automatically accept unless...
// it's one of those special cases
if ($token->type == 'start') {
// if there's a parent, check for special case
if (!empty($current_nesting)) {
$parent = array_pop($current_nesting);
$parent_name = $parent->name;
$parent_info = $definition->info[$parent_name];
if (isset($parent_info->auto_close[$token->name])) {
$result[] = new HTMLPurifier_Token_End($parent_name);
$result[] = $token;
$current_nesting[] = $token;
continue;
}
$current_nesting[] = $parent; // undo the pop
}
$result[] = $token;
$current_nesting[] = $token;
continue;
}
// sanity check
if ($token->type != 'end') continue;
// okay, we're dealing with a closing tag
// make sure that we have something open
if (empty($current_nesting)) {
if ($escape_invalid_tags) {
$result[] = new HTMLPurifier_Token_Text(
$generator->generateFromToken($token, $config, $context)
);
}
continue;
}
// first, check for the simplest case: everything closes neatly
// current_nesting is modified
$current_parent = array_pop($current_nesting);
if ($current_parent->name == $token->name) {
$result[] = $token;
continue;
}
// undo the array_pop
$current_nesting[] = $current_parent;
// okay, so we're trying to close the wrong tag
// scroll back the entire nest, trying to find our tag
// feature could be to specify how far you'd like to go
$size = count($current_nesting);
// -2 because -1 is the last element, but we already checked that
$skipped_tags = false;
for ($i = $size - 2; $i >= 0; $i--) {
if ($current_nesting[$i]->name == $token->name) {
// current nesting is modified
$skipped_tags = array_splice($current_nesting, $i);
break;
}
}
// we still didn't find the tag, so translate to text
if ($skipped_tags === false) {
if ($escape_invalid_tags) {
$result[] = new HTMLPurifier_Token_Text(
$generator->generateFromToken($token, $config, $context)
);
}
continue;
}
// okay, we found it, close all the skipped tags
// note that skipped tags contains the element we need closed
$size = count($skipped_tags);
for ($i = $size - 1; $i >= 0; $i--) {
$result[] = new HTMLPurifier_Token_End($skipped_tags[$i]->name);
}
// done!
}
// we're at the end now, fix all still unclosed tags
if (!empty($current_nesting)) {
$size = count($current_nesting);
for ($i = $size - 1; $i >= 0; $i--) {
$result[] =
new HTMLPurifier_Token_End($current_nesting[$i]->name);
}
}
return $result;
}
}
?>
@@ -0,0 +1,86 @@
<?php
require_once 'HTMLPurifier/Strategy.php';
require_once 'HTMLPurifier/HTMLDefinition.php';
require_once 'HTMLPurifier/Generator.php';
require_once 'HTMLPurifier/TagTransform.php';
HTMLPurifier_ConfigSchema::define(
'Core', 'RemoveInvalidImg', true, 'bool',
'This directive enables pre-emptive URI checking in <code>img</code> '.
'tags, as the attribute validation strategy is not authorized to '.
'remove elements from the document. This directive has been available '.
'since 1.3.0, revert to pre-1.3.0 behavior by setting to false.'
);
/**
* Removes all unrecognized tags from the list of tokens.
*
* This strategy iterates through all the tokens and removes unrecognized
* tokens. If a token is not recognized but a TagTransform is defined for
* that element, the element will be transformed accordingly.
*/
class HTMLPurifier_Strategy_RemoveForeignElements extends HTMLPurifier_Strategy
{
function execute($tokens, $config, &$context) {
$definition = $config->getHTMLDefinition();
$generator = new HTMLPurifier_Generator();
$result = array();
$escape_invalid_tags = $config->get('Core', 'EscapeInvalidTags');
foreach($tokens as $token) {
if (!empty( $token->is_tag )) {
// DEFINITION CALL
if (isset($definition->info[$token->name])) {
// leave untouched, except for a few special cases:
// hard-coded image special case, pre-emptively drop
// if not available. Probably not abstract-able
if ( $token->name == 'img' ) {
if (!isset($token->attr['src'])) {
continue;
}
if (!isset($definition->info['img']->attr['src'])) {
continue;
}
$token->attr['src'] =
$definition->
info['img']->
attr['src']->
validate($token->attr['src'],
$config, $context);
if ($token->attr['src'] === false) continue;
}
} elseif (
isset($definition->info_tag_transform[$token->name])
) {
// there is a transformation for this tag
// DEFINITION CALL
$token = $definition->
info_tag_transform[$token->name]->
transform($token, $config, $context);
} elseif ($escape_invalid_tags) {
// invalid tag, generate HTML and insert in
$token = new HTMLPurifier_Token_Text(
$generator->generateFromToken($token, $config, $context)
);
} else {
continue;
}
} elseif ($token->type == 'comment') {
// strip comments
continue;
} elseif ($token->type == 'text') {
} else {
continue;
}
$result[] = $token;
}
return $result;
}
}
?>
@@ -0,0 +1,129 @@
<?php
require_once 'HTMLPurifier/Strategy.php';
require_once 'HTMLPurifier/HTMLDefinition.php';
require_once 'HTMLPurifier/IDAccumulator.php';
HTMLPurifier_ConfigSchema::define(
'Attr', 'IDBlacklist', array(), 'list',
'Array of IDs not allowed in the document.');
/**
* Validate all attributes in the tokens.
*/
class HTMLPurifier_Strategy_ValidateAttributes extends HTMLPurifier_Strategy
{
function execute($tokens, $config, &$context) {
$definition = $config->getHTMLDefinition();
// setup id_accumulator context
$id_accumulator = new HTMLPurifier_IDAccumulator();
$id_accumulator->load($config->get('Attr', 'IDBlacklist'));
$context->register('IDAccumulator', $id_accumulator);
// create alias to global definition array, see also $defs
// DEFINITION CALL
$d_defs = $definition->info_global_attr;
foreach ($tokens as $key => $token) {
// only process tokens that have attributes,
// namely start and empty tags
if ($token->type !== 'start' && $token->type !== 'empty') continue;
// copy out attributes for easy manipulation
$attr = $token->attr;
// do global transformations (pre)
// nothing currently utilizes this
foreach ($definition->info_attr_transform_pre as $transform) {
$attr = $transform->transform($attr, $config, $context);
}
// do local transformations only applicable to this element (pre)
// ex. <p align="right"> to <p style="text-align:right;">
foreach ($definition->info[$token->name]->attr_transform_pre
as $transform
) {
$attr = $transform->transform($attr, $config, $context);
}
// 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;
// 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 || $result === null) {
// remove the attribute
unset($attr[$attr_key]);
} elseif (is_string($result)) {
// simple substitution
$attr[$attr_key] = $result;
}
// we'd also want slightly more complicated substitution
// involving an array as the return value,
// although we're not sure how colliding attributes would
// resolve (certain ones would be completely overriden,
// others would prepend themselves).
}
// post transforms
// ex. <x lang="fr"> to <x lang="fr" xml:lang="fr">
foreach ($definition->info_attr_transform_post as $transform) {
$attr = $transform->transform($attr, $config, $context);
}
// ex. <bdo> to <bdo dir="ltr">
foreach ($definition->info[$token->name]->attr_transform_post as $transform) {
$attr = $transform->transform($attr, $config, $context);
}
// commit changes
// could interfere with flyweight implementation
$tokens[$key]->attr = $attr;
}
$context->destroy('IDAccumulator');
return $tokens;
}
}
?>