friendica/vendor/ezyang/htmlpurifier/library/HTMLPurifier/Lexer.php

358 lines
12 KiB
PHP
Raw Normal View History

2010-09-09 03:14:17 +00:00
<?php
/**
* Forgivingly lexes HTML (SGML-style) markup into tokens.
*
* A lexer parses a string of SGML-style markup and converts them into
* corresponding tokens. It doesn't check for well-formedness, although its
* internal mechanism may make this automatic (such as the case of
* HTMLPurifier_Lexer_DOMLex). There are several implementations to choose
* from.
*
* A lexer is HTML-oriented: it might work with XML, but it's not
* recommended, as we adhere to a subset of the specification for optimization
* reasons. This might change in the future. Also, most tokenizers are not
* expected to handle DTDs or PIs.
*
* This class should not be directly instantiated, but you may use create() to
* retrieve a default copy of the lexer. Being a supertype, this class
* does not actually define any implementation, but offers commonly used
* convenience functions for subclasses.
*
* @note The unit tests will instantiate this class for testing purposes, as
* many of the utility functions require a class to be instantiated.
* This means that, even though this class is not runnable, it will
* not be declared abstract.
*
* @par
*
* @note
* We use tokens rather than create a DOM representation because DOM would:
*
* @par
* -# Require more processing and memory to create,
* -# Is not streamable, and
* -# Has the entire document structure (html and body not needed).
*
* @par
* However, DOM is helpful in that it makes it easy to move around nodes
* without a lot of lookaheads to see when a tag is closed. This is a
* limitation of the token system and some workarounds would be nice.
*/
class HTMLPurifier_Lexer
{
/**
* Whether or not this lexer implements line-number/column-number tracking.
* If it does, set to true.
*/
public $tracksLineNumbers = false;
// -- STATIC ----------------------------------------------------------
/**
* Retrieves or sets the default Lexer as a Prototype Factory.
*
* By default HTMLPurifier_Lexer_DOMLex will be returned. There are
* a few exceptions involving special features that only DirectLex
* implements.
*
* @note The behavior of this class has changed, rather than accepting
* a prototype object, it now accepts a configuration object.
* To specify your own prototype, set %Core.LexerImpl to it.
* This change in behavior de-singletonizes the lexer object.
*
2016-02-09 10:06:17 +00:00
* @param HTMLPurifier_Config $config
* @return HTMLPurifier_Lexer
* @throws HTMLPurifier_Exception
2010-09-09 03:14:17 +00:00
*/
2016-02-09 10:06:17 +00:00
public static function create($config)
{
2010-09-09 03:14:17 +00:00
if (!($config instanceof HTMLPurifier_Config)) {
$lexer = $config;
2016-02-09 10:06:17 +00:00
trigger_error(
"Passing a prototype to
HTMLPurifier_Lexer::create() is deprecated, please instead
use %Core.LexerImpl",
E_USER_WARNING
);
2010-09-09 03:14:17 +00:00
} else {
$lexer = $config->get('Core.LexerImpl');
}
$needs_tracking =
$config->get('Core.MaintainLineNumbers') ||
$config->get('Core.CollectErrors');
$inst = null;
if (is_object($lexer)) {
$inst = $lexer;
} else {
2016-02-09 10:06:17 +00:00
if (is_null($lexer)) {
do {
// auto-detection algorithm
if ($needs_tracking) {
$lexer = 'DirectLex';
break;
}
if (class_exists('DOMDocument') &&
method_exists('DOMDocument', 'loadHTML') &&
!extension_loaded('domxml')
) {
// check for DOM support, because while it's part of the
// core, it can be disabled compile time. Also, the PECL
// domxml extension overrides the default DOM, and is evil
// and nasty and we shan't bother to support it
$lexer = 'DOMLex';
} else {
$lexer = 'DirectLex';
}
} while (0);
} // do..while so we can break
2010-09-09 03:14:17 +00:00
// instantiate recognized string names
switch ($lexer) {
case 'DOMLex':
$inst = new HTMLPurifier_Lexer_DOMLex();
break;
case 'DirectLex':
$inst = new HTMLPurifier_Lexer_DirectLex();
break;
case 'PH5P':
$inst = new HTMLPurifier_Lexer_PH5P();
break;
default:
2016-02-09 10:06:17 +00:00
throw new HTMLPurifier_Exception(
"Cannot instantiate unrecognized Lexer type " .
htmlspecialchars($lexer)
);
2010-09-09 03:14:17 +00:00
}
}
2016-02-09 10:06:17 +00:00
if (!$inst) {
throw new HTMLPurifier_Exception('No lexer was instantiated');
}
2010-09-09 03:14:17 +00:00
// once PHP DOM implements native line numbers, or we
// hack out something using XSLT, remove this stipulation
if ($needs_tracking && !$inst->tracksLineNumbers) {
2016-02-09 10:06:17 +00:00
throw new HTMLPurifier_Exception(
'Cannot use lexer that does not support line numbers with ' .
'Core.MaintainLineNumbers or Core.CollectErrors (use DirectLex instead)'
);
2010-09-09 03:14:17 +00:00
}
return $inst;
}
// -- CONVENIENCE MEMBERS ---------------------------------------------
2016-02-09 10:06:17 +00:00
public function __construct()
{
2010-09-09 03:14:17 +00:00
$this->_entity_parser = new HTMLPurifier_EntityParser();
}
/**
* Most common entity to raw value conversion table for special entities.
2016-02-09 10:06:17 +00:00
* @type array
2010-09-09 03:14:17 +00:00
*/
protected $_special_entity2str =
2016-02-09 10:06:17 +00:00
array(
'&quot;' => '"',
'&amp;' => '&',
'&lt;' => '<',
'&gt;' => '>',
'&#39;' => "'",
'&#039;' => "'",
'&#x27;' => "'"
);
2010-09-09 03:14:17 +00:00
/**
* Parses special entities into the proper characters.
*
* This string will translate escaped versions of the special characters
* into the correct ones.
*
* @warning
* You should be able to treat the output of this function as
* completely parsed, but that's only because all other entities should
* have been handled previously in substituteNonSpecialEntities()
*
2016-02-09 10:06:17 +00:00
* @param string $string String character data to be parsed.
* @return string Parsed character data.
2010-09-09 03:14:17 +00:00
*/
2016-02-09 10:06:17 +00:00
public function parseData($string)
{
2010-09-09 03:14:17 +00:00
// following functions require at least one character
2016-02-09 10:06:17 +00:00
if ($string === '') {
return '';
}
2010-09-09 03:14:17 +00:00
// subtracts amps that cannot possibly be escaped
$num_amp = substr_count($string, '&') - substr_count($string, '& ') -
2016-02-09 10:06:17 +00:00
($string[strlen($string) - 1] === '&' ? 1 : 0);
2010-09-09 03:14:17 +00:00
2016-02-09 10:06:17 +00:00
if (!$num_amp) {
return $string;
} // abort if no entities
2010-09-09 03:14:17 +00:00
$num_esc_amp = substr_count($string, '&amp;');
$string = strtr($string, $this->_special_entity2str);
// code duplication for sake of optimization, see above
$num_amp_2 = substr_count($string, '&') - substr_count($string, '& ') -
2016-02-09 10:06:17 +00:00
($string[strlen($string) - 1] === '&' ? 1 : 0);
2010-09-09 03:14:17 +00:00
2016-02-09 10:06:17 +00:00
if ($num_amp_2 <= $num_esc_amp) {
return $string;
}
2010-09-09 03:14:17 +00:00
// hmm... now we have some uncommon entities. Use the callback.
$string = $this->_entity_parser->substituteSpecialEntities($string);
return $string;
}
/**
* Lexes an HTML string into tokens.
* @param $string String HTML.
2016-02-09 10:06:17 +00:00
* @param HTMLPurifier_Config $config
* @param HTMLPurifier_Context $context
* @return HTMLPurifier_Token[] array representation of HTML.
2010-09-09 03:14:17 +00:00
*/
2016-02-09 10:06:17 +00:00
public function tokenizeHTML($string, $config, $context)
{
2010-09-09 03:14:17 +00:00
trigger_error('Call to abstract class', E_USER_ERROR);
}
/**
* Translates CDATA sections into regular sections (through escaping).
2016-02-09 10:06:17 +00:00
* @param string $string HTML string to process.
* @return string HTML with CDATA sections escaped.
2010-09-09 03:14:17 +00:00
*/
2016-02-09 10:06:17 +00:00
protected static function escapeCDATA($string)
{
2010-09-09 03:14:17 +00:00
return preg_replace_callback(
'/<!\[CDATA\[(.+?)\]\]>/s',
array('HTMLPurifier_Lexer', 'CDATACallback'),
$string
);
}
/**
* Special CDATA case that is especially convoluted for <script>
2016-02-09 10:06:17 +00:00
* @param string $string HTML string to process.
* @return string HTML with CDATA sections escaped.
2010-09-09 03:14:17 +00:00
*/
2016-02-09 10:06:17 +00:00
protected static function escapeCommentedCDATA($string)
{
2010-09-09 03:14:17 +00:00
return preg_replace_callback(
'#<!--//--><!\[CDATA\[//><!--(.+?)//--><!\]\]>#s',
array('HTMLPurifier_Lexer', 'CDATACallback'),
$string
);
}
2016-02-09 10:06:17 +00:00
/**
* Special Internet Explorer conditional comments should be removed.
* @param string $string HTML string to process.
* @return string HTML with conditional comments removed.
*/
protected static function removeIEConditional($string)
{
return preg_replace(
'#<!--\[if [^>]+\]>.*?<!\[endif\]-->#si', // probably should generalize for all strings
'',
$string
);
}
2010-09-09 03:14:17 +00:00
/**
* Callback function for escapeCDATA() that does the work.
*
* @warning Though this is public in order to let the callback happen,
* calling it directly is not recommended.
2016-02-09 10:06:17 +00:00
* @param array $matches PCRE matches array, with index 0 the entire match
2010-09-09 03:14:17 +00:00
* and 1 the inside of the CDATA section.
2016-02-09 10:06:17 +00:00
* @return string Escaped internals of the CDATA section.
2010-09-09 03:14:17 +00:00
*/
2016-02-09 10:06:17 +00:00
protected static function CDATACallback($matches)
{
2010-09-09 03:14:17 +00:00
// not exactly sure why the character set is needed, but whatever
return htmlspecialchars($matches[1], ENT_COMPAT, 'UTF-8');
}
/**
* Takes a piece of HTML and normalizes it by converting entities, fixing
* encoding, extracting bits, and other good stuff.
2016-02-09 10:06:17 +00:00
* @param string $html HTML.
* @param HTMLPurifier_Config $config
* @param HTMLPurifier_Context $context
* @return string
2010-09-09 03:14:17 +00:00
* @todo Consider making protected
*/
2016-02-09 10:06:17 +00:00
public function normalize($html, $config, $context)
{
2010-09-09 03:14:17 +00:00
// normalize newlines to \n
2016-02-09 10:06:17 +00:00
if ($config->get('Core.NormalizeNewlines')) {
$html = str_replace("\r\n", "\n", $html);
$html = str_replace("\r", "\n", $html);
}
2010-09-09 03:14:17 +00:00
if ($config->get('HTML.Trusted')) {
// escape convoluted CDATA
$html = $this->escapeCommentedCDATA($html);
}
// escape CDATA
$html = $this->escapeCDATA($html);
2016-02-09 10:06:17 +00:00
$html = $this->removeIEConditional($html);
2010-09-09 03:14:17 +00:00
// 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);
2016-02-09 10:06:17 +00:00
// if processing instructions are to removed, remove them now
if ($config->get('Core.RemoveProcessingInstructions')) {
$html = preg_replace('#<\?.+?\?>#s', '', $html);
}
2010-09-09 03:14:17 +00:00
return $html;
}
/**
* Takes a string of HTML (fragment or document) and returns the content
* @todo Consider making protected
*/
2016-02-09 10:06:17 +00:00
public function extractBody($html)
{
2010-09-09 03:14:17 +00:00
$matches = array();
$result = preg_match('!<body[^>]*>(.*)</body>!is', $html, $matches);
if ($result) {
return $matches[1];
} else {
return $html;
}
}
}
// vim: et sw=4 sts=4