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

540 lines
20 KiB
PHP
Raw Normal View History

2010-09-09 03:14:17 +00:00
<?php
/**
* Our in-house implementation of a parser.
*
* A pure PHP parser, DirectLex has absolutely no dependencies, making
* it a reasonably good default for PHP4. Written with efficiency in mind,
* it can be four times faster than HTMLPurifier_Lexer_PEARSax3, although it
* pales in comparison to HTMLPurifier_Lexer_DOMLex.
*
* @todo Reread XML spec and document differences.
*/
class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
{
2016-02-09 10:06:17 +00:00
/**
* @type bool
*/
2010-09-09 03:14:17 +00:00
public $tracksLineNumbers = true;
/**
* Whitespace characters for str(c)spn.
2016-02-09 10:06:17 +00:00
* @type string
2010-09-09 03:14:17 +00:00
*/
protected $_whitespace = "\x20\x09\x0D\x0A";
/**
* Callback function for script CDATA fudge
2016-02-09 10:06:17 +00:00
* @param array $matches, in form of array(opening tag, contents, closing tag)
* @return string
2010-09-09 03:14:17 +00:00
*/
2016-02-09 10:06:17 +00:00
protected function scriptCallback($matches)
{
2010-09-09 03:14:17 +00:00
return $matches[1] . htmlspecialchars($matches[2], ENT_COMPAT, 'UTF-8') . $matches[3];
}
2016-02-09 10:06:17 +00:00
/**
* @param String $html
* @param HTMLPurifier_Config $config
* @param HTMLPurifier_Context $context
* @return array|HTMLPurifier_Token[]
*/
public function tokenizeHTML($html, $config, $context)
{
2010-09-09 03:14:17 +00:00
// special normalization for script tags without any armor
// our "armor" heurstic is a < sign any number of whitespaces after
// the first script tag
if ($config->get('HTML.Trusted')) {
2016-02-09 10:06:17 +00:00
$html = preg_replace_callback(
'#(<script[^>]*>)(\s*[^<].+?)(</script>)#si',
array($this, 'scriptCallback'),
$html
);
2010-09-09 03:14:17 +00:00
}
$html = $this->normalize($html, $config, $context);
$cursor = 0; // our location in the text
$inside_tag = false; // whether or not we're parsing the inside of a tag
$array = array(); // result array
// This is also treated to mean maintain *column* numbers too
$maintain_line_numbers = $config->get('Core.MaintainLineNumbers');
if ($maintain_line_numbers === null) {
// automatically determine line numbering by checking
// if error collection is on
$maintain_line_numbers = $config->get('Core.CollectErrors');
}
if ($maintain_line_numbers) {
$current_line = 1;
2016-02-09 10:06:17 +00:00
$current_col = 0;
2010-09-09 03:14:17 +00:00
$length = strlen($html);
} else {
$current_line = false;
2016-02-09 10:06:17 +00:00
$current_col = false;
2010-09-09 03:14:17 +00:00
$length = false;
}
$context->register('CurrentLine', $current_line);
2016-02-09 10:06:17 +00:00
$context->register('CurrentCol', $current_col);
2010-09-09 03:14:17 +00:00
$nl = "\n";
// how often to manually recalculate. This will ALWAYS be right,
// but it's pretty wasteful. Set to 0 to turn off
$synchronize_interval = $config->get('Core.DirectLexLineNumberSyncInterval');
$e = false;
if ($config->get('Core.CollectErrors')) {
$e =& $context->get('ErrorCollector');
}
// for testing synchronization
$loops = 0;
2016-02-09 10:06:17 +00:00
while (++$loops) {
2010-09-09 03:14:17 +00:00
// $cursor is either at the start of a token, or inside of
// a tag (i.e. there was a < immediately before it), as indicated
// by $inside_tag
if ($maintain_line_numbers) {
// $rcursor, however, is always at the start of a token.
2016-02-09 10:06:17 +00:00
$rcursor = $cursor - (int)$inside_tag;
2010-09-09 03:14:17 +00:00
// Column number is cheap, so we calculate it every round.
// We're interested at the *end* of the newline string, so
// we need to add strlen($nl) == 1 to $nl_pos before subtracting it
// from our "rcursor" position.
$nl_pos = strrpos($html, $nl, $rcursor - $length);
$current_col = $rcursor - (is_bool($nl_pos) ? 0 : $nl_pos + 1);
// recalculate lines
2016-02-09 10:06:17 +00:00
if ($synchronize_interval && // synchronization is on
$cursor > 0 && // cursor is further than zero
$loops % $synchronize_interval === 0) { // time to synchronize!
2010-09-09 03:14:17 +00:00
$current_line = 1 + $this->substrCount($html, $nl, 0, $cursor);
}
}
$position_next_lt = strpos($html, '<', $cursor);
$position_next_gt = strpos($html, '>', $cursor);
// triggers on "<b>asdf</b>" but not "asdf <b></b>"
// special case to set up context
if ($position_next_lt === $cursor) {
$inside_tag = true;
$cursor++;
}
if (!$inside_tag && $position_next_lt !== false) {
// We are not inside tag and there still is another tag to parse
$token = new
2016-02-09 10:06:17 +00:00
HTMLPurifier_Token_Text(
$this->parseData(
substr(
$html,
$cursor,
$position_next_lt - $cursor
2010-09-09 03:14:17 +00:00
)
2016-02-09 10:06:17 +00:00
)
);
2010-09-09 03:14:17 +00:00
if ($maintain_line_numbers) {
$token->rawPosition($current_line, $current_col);
$current_line += $this->substrCount($html, $nl, $cursor, $position_next_lt - $cursor);
}
$array[] = $token;
2016-02-09 10:06:17 +00:00
$cursor = $position_next_lt + 1;
2010-09-09 03:14:17 +00:00
$inside_tag = true;
continue;
} elseif (!$inside_tag) {
// We are not inside tag but there are no more tags
// If we're already at the end, break
2016-02-09 10:06:17 +00:00
if ($cursor === strlen($html)) {
break;
}
2010-09-09 03:14:17 +00:00
// Create Text of rest of string
$token = new
2016-02-09 10:06:17 +00:00
HTMLPurifier_Token_Text(
$this->parseData(
substr(
$html,
$cursor
2010-09-09 03:14:17 +00:00
)
2016-02-09 10:06:17 +00:00
)
);
if ($maintain_line_numbers) {
$token->rawPosition($current_line, $current_col);
}
2010-09-09 03:14:17 +00:00
$array[] = $token;
break;
} elseif ($inside_tag && $position_next_gt !== false) {
// We are in tag and it is well formed
// Grab the internals of the tag
$strlen_segment = $position_next_gt - $cursor;
if ($strlen_segment < 1) {
// there's nothing to process!
$token = new HTMLPurifier_Token_Text('<');
$cursor++;
continue;
}
$segment = substr($html, $cursor, $strlen_segment);
if ($segment === false) {
// somehow, we attempted to access beyond the end of
// the string, defense-in-depth, reported by Nate Abele
break;
}
// Check if it's a comment
2016-02-09 10:06:17 +00:00
if (substr($segment, 0, 3) === '!--') {
2010-09-09 03:14:17 +00:00
// re-determine segment length, looking for -->
$position_comment_end = strpos($html, '-->', $cursor);
if ($position_comment_end === false) {
// uh oh, we have a comment that extends to
// infinity. Can't be helped: set comment
// end position to end of string
2016-02-09 10:06:17 +00:00
if ($e) {
$e->send(E_WARNING, 'Lexer: Unclosed comment');
}
2010-09-09 03:14:17 +00:00
$position_comment_end = strlen($html);
$end = true;
} else {
$end = false;
}
$strlen_segment = $position_comment_end - $cursor;
$segment = substr($html, $cursor, $strlen_segment);
$token = new
2016-02-09 10:06:17 +00:00
HTMLPurifier_Token_Comment(
substr(
$segment,
3,
$strlen_segment - 3
)
);
2010-09-09 03:14:17 +00:00
if ($maintain_line_numbers) {
$token->rawPosition($current_line, $current_col);
$current_line += $this->substrCount($html, $nl, $cursor, $strlen_segment);
}
$array[] = $token;
$cursor = $end ? $position_comment_end : $position_comment_end + 3;
$inside_tag = false;
continue;
}
// Check if it's an end tag
2016-02-09 10:06:17 +00:00
$is_end_tag = (strpos($segment, '/') === 0);
2010-09-09 03:14:17 +00:00
if ($is_end_tag) {
$type = substr($segment, 1);
$token = new HTMLPurifier_Token_End($type);
if ($maintain_line_numbers) {
$token->rawPosition($current_line, $current_col);
$current_line += $this->substrCount($html, $nl, $cursor, $position_next_gt - $cursor);
}
$array[] = $token;
$inside_tag = false;
$cursor = $position_next_gt + 1;
continue;
}
// Check leading character is alnum, if not, we may
// have accidently grabbed an emoticon. Translate into
// text and go our merry way
if (!ctype_alpha($segment[0])) {
// XML: $segment[0] !== '_' && $segment[0] !== ':'
2016-02-09 10:06:17 +00:00
if ($e) {
$e->send(E_NOTICE, 'Lexer: Unescaped lt');
}
2010-09-09 03:14:17 +00:00
$token = new HTMLPurifier_Token_Text('<');
if ($maintain_line_numbers) {
$token->rawPosition($current_line, $current_col);
$current_line += $this->substrCount($html, $nl, $cursor, $position_next_gt - $cursor);
}
$array[] = $token;
$inside_tag = false;
continue;
}
// Check if it is explicitly self closing, if so, remove
// trailing slash. Remember, we could have a tag like <br>, so
// any later token processing scripts must convert improperly
// classified EmptyTags from StartTags.
2016-02-09 10:06:17 +00:00
$is_self_closing = (strrpos($segment, '/') === $strlen_segment - 1);
2010-09-09 03:14:17 +00:00
if ($is_self_closing) {
$strlen_segment--;
$segment = substr($segment, 0, $strlen_segment);
}
// Check if there are any attributes
$position_first_space = strcspn($segment, $this->_whitespace);
if ($position_first_space >= $strlen_segment) {
if ($is_self_closing) {
$token = new HTMLPurifier_Token_Empty($segment);
} else {
$token = new HTMLPurifier_Token_Start($segment);
}
if ($maintain_line_numbers) {
$token->rawPosition($current_line, $current_col);
$current_line += $this->substrCount($html, $nl, $cursor, $position_next_gt - $cursor);
}
$array[] = $token;
$inside_tag = false;
$cursor = $position_next_gt + 1;
continue;
}
// Grab out all the data
$type = substr($segment, 0, $position_first_space);
$attribute_string =
trim(
substr(
2016-02-09 10:06:17 +00:00
$segment,
$position_first_space
2010-09-09 03:14:17 +00:00
)
);
if ($attribute_string) {
$attr = $this->parseAttributeString(
2016-02-09 10:06:17 +00:00
$attribute_string,
$config,
$context
);
2010-09-09 03:14:17 +00:00
} else {
$attr = array();
}
if ($is_self_closing) {
$token = new HTMLPurifier_Token_Empty($type, $attr);
} else {
$token = new HTMLPurifier_Token_Start($type, $attr);
}
if ($maintain_line_numbers) {
$token->rawPosition($current_line, $current_col);
$current_line += $this->substrCount($html, $nl, $cursor, $position_next_gt - $cursor);
}
$array[] = $token;
$cursor = $position_next_gt + 1;
$inside_tag = false;
continue;
} else {
// inside tag, but there's no ending > sign
2016-02-09 10:06:17 +00:00
if ($e) {
$e->send(E_WARNING, 'Lexer: Missing gt');
}
2010-09-09 03:14:17 +00:00
$token = new
2016-02-09 10:06:17 +00:00
HTMLPurifier_Token_Text(
'<' .
$this->parseData(
substr($html, $cursor)
)
);
if ($maintain_line_numbers) {
$token->rawPosition($current_line, $current_col);
}
2010-09-09 03:14:17 +00:00
// no cursor scroll? Hmm...
$array[] = $token;
break;
}
break;
}
$context->destroy('CurrentLine');
$context->destroy('CurrentCol');
return $array;
}
/**
* PHP 5.0.x compatible substr_count that implements offset and length
2016-02-09 10:06:17 +00:00
* @param string $haystack
* @param string $needle
* @param int $offset
* @param int $length
* @return int
2010-09-09 03:14:17 +00:00
*/
2016-02-09 10:06:17 +00:00
protected function substrCount($haystack, $needle, $offset, $length)
{
2010-09-09 03:14:17 +00:00
static $oldVersion;
if ($oldVersion === null) {
$oldVersion = version_compare(PHP_VERSION, '5.1', '<');
}
if ($oldVersion) {
$haystack = substr($haystack, $offset, $length);
return substr_count($haystack, $needle);
} else {
return substr_count($haystack, $needle, $offset, $length);
}
}
/**
* Takes the inside of an HTML tag and makes an assoc array of attributes.
*
2016-02-09 10:06:17 +00:00
* @param string $string Inside of tag excluding name.
* @param HTMLPurifier_Config $config
* @param HTMLPurifier_Context $context
* @return array Assoc array of attributes.
2010-09-09 03:14:17 +00:00
*/
2016-02-09 10:06:17 +00:00
public function parseAttributeString($string, $config, $context)
{
$string = (string)$string; // quick typecast
2010-09-09 03:14:17 +00:00
2016-02-09 10:06:17 +00:00
if ($string == '') {
return array();
} // no attributes
2010-09-09 03:14:17 +00:00
$e = false;
if ($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 (!$key) {
2016-02-09 10:06:17 +00:00
if ($e) {
$e->send(E_ERROR, 'Lexer: Missing attribute key');
}
2010-09-09 03:14:17 +00:00
return array();
}
2016-02-09 10:06:17 +00:00
if (!$quoted_value) {
return array($key => '');
}
2010-09-09 03:14:17 +00:00
$first_char = @$quoted_value[0];
2016-02-09 10:06:17 +00:00
$last_char = @$quoted_value[strlen($quoted_value) - 1];
2010-09-09 03:14:17 +00:00
$same_quote = ($first_char == $last_char);
$open_quote = ($first_char == '"' || $first_char == "'");
2016-02-09 10:06:17 +00:00
if ($same_quote && $open_quote) {
2010-09-09 03:14:17 +00:00
// well behaved
$value = substr($quoted_value, 1, strlen($quoted_value) - 2);
} else {
// not well behaved
if ($open_quote) {
2016-02-09 10:06:17 +00:00
if ($e) {
$e->send(E_ERROR, 'Lexer: Missing end quote');
}
2010-09-09 03:14:17 +00:00
$value = substr($quoted_value, 1);
} else {
$value = $quoted_value;
}
}
2016-02-09 10:06:17 +00:00
if ($value === false) {
$value = '';
}
2010-09-09 03:14:17 +00:00
return array($key => $this->parseData($value));
}
// setup loop environment
2016-02-09 10:06:17 +00:00
$array = array(); // return assoc array of attributes
2010-09-09 03:14:17 +00:00
$cursor = 0; // current position in string (moves forward)
2016-02-09 10:06:17 +00:00
$size = strlen($string); // size of the string (stays the same)
2010-09-09 03:14:17 +00:00
// if we have unquoted attributes, the parser expects a terminating
// space, so let's guarantee that there's always a terminating space.
$string .= ' ';
2016-02-09 10:06:17 +00:00
$old_cursor = -1;
while ($cursor < $size) {
if ($old_cursor >= $cursor) {
throw new Exception("Infinite loop detected");
2010-09-09 03:14:17 +00:00
}
2016-02-09 10:06:17 +00:00
$old_cursor = $cursor;
2010-09-09 03:14:17 +00:00
$cursor += ($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) {
2016-02-09 10:06:17 +00:00
if ($e) {
$e->send(E_ERROR, 'Lexer: Missing attribute key');
}
$cursor += 1 + strcspn($string, $this->_whitespace, $cursor + 1); // prevent infinite loop
2010-09-09 03:14:17 +00:00
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 - $value_begin);
2016-02-09 10:06:17 +00:00
if ($value === false) {
$value = '';
}
2010-09-09 03:14:17 +00:00
$array[$key] = $this->parseData($value);
$cursor++;
} else {
// boolattr
if ($key !== '') {
$array[$key] = $key;
} else {
// purely theoretical
2016-02-09 10:06:17 +00:00
if ($e) {
$e->send(E_ERROR, 'Lexer: Missing attribute key');
}
2010-09-09 03:14:17 +00:00
}
}
}
return $array;
}
}
// vim: et sw=4 sts=4