htmlfilter.php File Reference

Go to the source code of this file.

Functions
	tln_tagprint ($tagname, $attary, $tagtype)
	This function returns the final tag out of the tag name, an array of attributes, and the type of the tag. More...
	tln_casenormalize (&$val)
	A small helper function to use with array_walk. More...
	tln_skipspace ($body, $offset)
	This function skips any whitespace from the current position within a string and to the next non-whitespace value. More...
	tln_findnxstr ($body, $offset, $needle)
	This function looks for the next character within a string. More...
	tln_findnxreg ($body, $offset, $reg)
	This function takes a PCRE-style regexp and tries to match it within the string. More...
	tln_getnxtag ($body, $offset)
	This function looks for the next tag. More...
	tln_deent (&$attvalue, $regex, $hex=false)
	Translates entities into literal values so they can be checked. More...
	tln_defang (&$attvalue)
	This function checks attribute values for entity-encoded values and returns them translated into 8-bit strings so we can run checks on them. More...
	tln_unspace (&$attvalue)
	Kill any tabs, newlines, or carriage returns. More...
	tln_fixatts ( $tagname, $attary, $rm_attnames, $bad_attvals, $add_attr_to_tag, $trans_image_path, $block_external_images)
	This function runs various checks against the attributes. More...
	tln_fixurl ($attname, &$attvalue, $trans_image_path, $block_external_images)
	tln_fixstyle ($body, $pos, $trans_image_path, $block_external_images)
	tln_body2div ($attary, $trans_image_path)
	tln_sanitize ( $body, $tag_list, $rm_tags_with_content, $self_closing_tags, $force_tag_closing, $rm_attnames, $bad_attvals, $add_attr_to_tag, $trans_image_path, $block_external_images)
	HTMLFilter ($body, $trans_image_path, $block_external_images=false)

Function Documentation

HTMLFilter()

HTMLFilter	(		$body,
			$trans_image_path,
			$block_external_images = false
	)

Definition at line 1013 of file htmlfilter.php.

{
 
    $tag_list = array(
        false,
        "object",
        "meta",
        "html",
        "head",
        "base",
        "link",
        "frame",
        "iframe",
        "plaintext",
        "marquee"
    );
 
    $rm_tags_with_content = array(
        "script",
        "applet",
        "embed",
        "title",
        "frameset",
        "xmp",
        "xml"
    );
 
    $self_closing_tags =  array(
        "img",
        "br",
        "hr",
        "input",
        "outbind"
    );
 
    $force_tag_closing = true;
 
    $rm_attnames = array(
        "/.*/" =>
            array(
                // "/target/i",
                "/^on.*/i",
                "/^dynsrc/i",
                "/^data.*/i",
                "/^lowsrc.*/i"
            )
    );
 
    $bad_attvals = array(
        "/.*/" =>
        array(
            "/^src|background/i" =>
            array(
                array(
                    '/^([\'"])\s*\S+script\s*:.*([\'"])/si',
                    '/^([\'"])\s*mocha\s*:*.*([\'"])/si',
                    '/^([\'"])\s*about\s*:.*([\'"])/si'
                ),
                array(
                    "\\1$trans_image_path\\2",
                    "\\1$trans_image_path\\2",
                    "\\1$trans_image_path\\2"
                )
            ),
            "/^href|action/i" =>
            array(
                array(
                    '/^([\'"])\s*\S+script\s*:.*([\'"])/si',
                    '/^([\'"])\s*mocha\s*:*.*([\'"])/si',
                    '/^([\'"])\s*about\s*:.*([\'"])/si'
                ),
                array(
                    "\\1#\\1",
                    "\\1#\\1",
                    "\\1#\\1"
                )
            ),
            "/^style/i" =>
            array(
                array(
                    "/\/\*.*\*\//",
                    "/expression/i",
                    "/binding/i",
                    "/behaviou*r/i",
                    "/include-source/i",
                    '/position\s*:/i',
                    '/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i',
                    '/url\s*\‍(\s*([\'"])\s*\S+script\s*:.*([\'"])\s*\‍)/si',
                    '/url\s*\‍(\s*([\'"])\s*mocha\s*:.*([\'"])\s*\‍)/si',
                    '/url\s*\‍(\s*([\'"])\s*about\s*:.*([\'"])\s*\‍)/si',
                    '/(.*)\s*:\s*url\s*\‍(\s*([\'"]*)\s*\S+script\s*:.*([\'"]*)\s*\‍)/si'
                ),
                array(
                    "",
                    "idiocy",
                    "idiocy",
                    "idiocy",
                    "idiocy",
                    "idiocy",
                    "url",
                    "url(\\1#\\1)",
                    "url(\\1#\\1)",
                    "url(\\1#\\1)",
                    "\\1:url(\\2#\\3)"
                )
            )
        )
    );
 
    if ($block_external_images) {
        array_push(
            $bad_attvals{'/.*/'}{'/^src|background/i'}[0],
            '/^([\'\"])\s*https*:.*([\'\"])/si'
        );
        array_push(
            $bad_attvals{'/.*/'}{'/^src|background/i'}[1],
            "\\1$trans_image_path\\1"
        );
        array_push(
            $bad_attvals{'/.*/'}{'/^style/i'}[0],
            '/url\‍(([\'\"])\s*https*:.*([\'\"])\‍)/si'
        );
        array_push(
            $bad_attvals{'/.*/'}{'/^style/i'}[1],
            "url(\\1$trans_image_path\\1)"
        );
    }
 
    $add_attr_to_tag = array(
        "/^a$/i" =>
            array('target' => '"_blank"')
    );
 
    $trusted = tln_sanitize(
        $body,
        $tag_list,
        $rm_tags_with_content,
        $self_closing_tags,
        $force_tag_closing,
        $rm_attnames,
        $bad_attvals,
        $add_attr_to_tag,
        $trans_image_path,
        $block_external_images
    );
    return $trusted;
}

References tln_sanitize().

Here is the call graph for this function:

tln_body2div()

tln_body2div	(		$attary,
			$trans_image_path
	)

Definition at line 791 of file htmlfilter.php.

{
    $divattary = array('class' => "'bodyclass'");
    $text = '#000000';
    $has_bgc_stl = $has_txt_stl = false;
    $styledef = '';
    if (is_array($attary) && sizeof($attary) > 0){
        foreach ($attary as $attname=>$attvalue){
            $quotchar = substr($attvalue, 0, 1);
            $attvalue = str_replace($quotchar, "", $attvalue);
            switch ($attname){
                case 'background':
                    $styledef .= "background-image: url('$trans_image_path'); ";
                    break;
                case 'bgcolor':
                    $has_bgc_stl = true;
                    $styledef .= "background-color: $attvalue; ";
                    break;
                case 'text':
                    $has_txt_stl = true;
                    $styledef .= "color: $attvalue; ";
                    break;
            }
        }
        // Outlook defines a white bgcolor and no text color. This can lead to
        // white text on a white bg with certain themes.
        if ($has_bgc_stl && !$has_txt_stl) {
            $styledef .= "color: $text; ";
        }
        if (strlen($styledef) > 0){
            $divattary{"style"} = "\"$styledef\"";
        }
    }
    return $divattary;
}

References $text.

Referenced by tln_sanitize().

Here is the caller graph for this function:

tln_casenormalize()

tln_casenormalize

(

&

$val

)

A small helper function to use with array_walk.

Modifies a by-ref value and makes it lowercase.

Parameters

string

$val

a value passed by-ref.

Returns

void since it modifies a by-ref value.

Definition at line 69 of file htmlfilter.php.

{
    $val = strtolower($val);
}

tln_deent()

tln_deent	(	&	$attvalue,
			$regex,
			$hex = false
	)

Translates entities into literal values so they can be checked.

Parameters

string	$attvalue	the by-ref value to check.
string	$regex	the regular expression to check against.
boolean	$hex	whether the entities are hexadecimal.

Returns

boolean True or False depending on whether there were matches.

Definition at line 439 of file htmlfilter.php.

{
    preg_match_all($regex, $attvalue, $matches);
    if (is_array($matches) && sizeof($matches[0]) > 0) {
        $repl = array();
        for ($i = 0; $i < sizeof($matches[0]); $i++) {
            $numval = $matches[1][$i];
            if ($hex) {
                $numval = hexdec($numval);
            }
            $repl{$matches[0][$i]} = chr($numval);
        }
        $attvalue = strtr($attvalue, $repl);
        return true;
    } else {
        return false;
    }
}

Referenced by tln_defang().

Here is the caller graph for this function:

tln_defang()

tln_defang

(

&

$attvalue

)

This function checks attribute values for entity-encoded values and returns them translated into 8-bit strings so we can run checks on them.

Parameters

string

$attvalue

A string to run entity check against.

Skip this if there aren't ampersands or backslashes.

Definition at line 465 of file htmlfilter.php.

{
    if (strpos($attvalue, '&') === false
        && strpos($attvalue, '\\') === false
    ) {
        return;
    }
    do {
        $m = false;
        $m = $m || tln_deent($attvalue, '/\&#0*(\d+);*/s');
        $m = $m || tln_deent($attvalue, '/\&#x0*((\d|[a-f])+);*/si', true);
        $m = $m || tln_deent($attvalue, '/\\\\(\d+)/s', true);
    } while ($m == true);
    $attvalue = stripslashes($attvalue);
}

References tln_deent().

Referenced by tln_fixatts(), and tln_fixstyle().

Here is the call graph for this function:

Here is the caller graph for this function:

tln_findnxreg()

tln_findnxreg	(		$body,
			$offset,
			$reg
	)

This function takes a PCRE-style regexp and tries to match it within the string.

Parameters

string	$body	The string to look for needle in.
integer	$offset	Start looking from here.
string	$reg	A PCRE-style regex to match.

Returns

array|boolean Returns a false if no matches found, or an array with the following members:

• integer with the location of the match within $body
• string with whatever content between offset and the match
• string with whatever it is we matched

Definition at line 127 of file htmlfilter.php.

{
    $matches = array();
    $retarr = array();
    $preg_rule = '%^(.*?)(' . $reg . ')%s';
    preg_match($preg_rule, substr($body, $offset), $matches);
    if (!isset($matches[0]) || !$matches[0]) {
        $retarr = false;
    } else {
        $retarr[0] = $offset + strlen($matches[1]);
        $retarr[1] = $matches[1];
        $retarr[2] = $matches[2];
    }
    return $retarr;
}

Referenced by tln_getnxtag().

Here is the caller graph for this function:

tln_findnxstr()

tln_findnxstr	(		$body,
			$offset,
			$needle
	)

This function looks for the next character within a string.

It's really just a glorified "strpos", except it catches the failures nicely.

Parameters

string	$body	The string to look for needle in.
integer	$offset	Start looking from this position.
string	$needle	The character/string to look for.

Returns

integer location of the next occurrence of the needle, or strlen($body) if needle wasn't found.

Definition at line 105 of file htmlfilter.php.

{
    $pos = strpos($body, $needle, $offset);
    if ($pos === false) {
        $pos = strlen($body);
    }
    return $pos;
}

Referenced by tln_getnxtag().

Here is the caller graph for this function:

tln_fixatts()

tln_fixatts	(		$tagname,
			$attary,
			$rm_attnames,
			$bad_attvals,
			$add_attr_to_tag,
			$trans_image_path,
			$block_external_images
	)

This function runs various checks against the attributes.

Parameters

string	$tagname	String with the name of the tag.
array	$attary	Array with all tag attributes.
array	$rm_attnames	See description for tln_sanitize
array	$bad_attvals	See description for tln_sanitize
array	$add_attr_to_tag	See description for tln_sanitize
string	$trans_image_path
boolean	$block_external_images

Returns

array with modified attributes.

See if this attribute should be removed.

Remove any backslashes, entities, or extraneous whitespace.

Now let's run checks on the attvalues. I don't expect anyone to comprehend this. If you do, get in touch with me so I can drive to where you live and shake your hand personally. :)

There are two arrays in valary. First is matches. Second one is replacements

See if we need to append any attributes to this tag.

Definition at line 514 of file htmlfilter.php.

  {
    while (list($attname, $attvalue) = each($attary)) {
        foreach ($rm_attnames as $matchtag => $matchattrs) {
            if (preg_match($matchtag, $tagname)) {
                foreach ($matchattrs as $matchattr) {
                    if (preg_match($matchattr, $attname)) {
                        unset($attary{$attname});
                        continue;
                    }
                }
            }
        }
        $oldattvalue = $attvalue;
        tln_defang($attvalue);
        if ($attname == 'style' && $attvalue !== $oldattvalue) {
            $attvalue = "idiocy";
            $attary{$attname} = $attvalue;
        }
        tln_unspace($attvalue);
 
        foreach ($bad_attvals as $matchtag => $matchattrs) {
            if (preg_match($matchtag, $tagname)) {
                foreach ($matchattrs as $matchattr => $valary) {
                    if (preg_match($matchattr, $attname)) {
                        list($valmatch, $valrepl) = $valary;
                        $newvalue = preg_replace($valmatch, $valrepl, $attvalue);
                        if ($newvalue != $attvalue) {
                            $attary{$attname} = $newvalue;
                            $attvalue = $newvalue;
                        }
                    }
                }
            }
        }
        if ($attname == 'style') {
            if (preg_match('/[\0-\37\200-\377]+/', $attvalue)) {
                $attary{$attname} = '"disallowed character"';
            }
            preg_match_all("/url\s*\‍((.+)\‍)/si", $attvalue, $aMatch);
            if (count($aMatch)) {
                foreach($aMatch[1] as $sMatch) {
                    $urlvalue = $sMatch;
                    tln_fixurl($attname, $urlvalue, $trans_image_path, $block_external_images);
                    $attary{$attname} = str_replace($sMatch, $urlvalue, $attvalue);
                }
            }
        }
     }
    foreach ($add_attr_to_tag as $matchtag => $addattary) {
        if (preg_match($matchtag, $tagname)) {
            $attary = array_merge($attary, $addattary);
        }
    }
    return $attary;
}

References tln_defang(), tln_fixurl(), and tln_unspace().

Referenced by tln_sanitize().

Here is the call graph for this function:

Here is the caller graph for this function:

tln_fixstyle()

tln_fixstyle	(		$body,
			$pos,
			$trans_image_path,
			$block_external_images
	)

First look for general BODY style declaration, which would be like so: body {background: blah-blah} and change it to .bodyclass so we can just assign it to a

Fix url('blah') declarations.

Remove any backslashes, entities, and extraneous whitespace.

Definition at line 666 of file htmlfilter.php.

{
    // workaround for </style> in between comments
    $content = '';
    $sToken = '';
    $bSucces = false;
    $bEndTag = false;
    for ($i=$pos,$iCount=strlen($body);$i<$iCount;++$i) {
        $char = $body{$i};
        switch ($char) {
            case '<':
                $sToken = $char;
                break;
            case '/':
                 if ($sToken == '<') {
                    $sToken .= $char;
                    $bEndTag = true;
                 } else {
                    $content .= $char;
                 }
                 break;
            case '>':
                 if ($bEndTag) {
                    $sToken .= $char;
                    if (preg_match('/<\/\s*style\s*>/i',$sToken,$aMatch)) {
                        $newpos = $i + 1;
                        $bSucces = true;
                        break 2;
                    } else {
                        $content .= $sToken;
                    }
                    $bEndTag = false;
                 } else {
                    $content .= $char;
                 }
                 break;
            case '!':
                if ($sToken == '<') {
                    // possible comment
                    if (isset($body{$i+2}) && substr($body,$i,3) == '!--') {
                        $i = strpos($body,'-->',$i+3);
                        if ($i === false) { // no end comment
                            $i = strlen($body);
                        }
                        $sToken = '';
                    }
                } else {
                    $content .= $char;
                }
                break;
            default:
                if ($bEndTag) {
                    $sToken .= $char;
                } else {
                    $content .= $char;
                }
                break;
        }
    }
    if ($bSucces == FALSE){
        return array(FALSE, strlen($body));
    }
 
 
 
    $content = preg_replace("|body(\s*\{.*?\})|si", ".bodyclass\\1", $content);
 
    //   $content = preg_replace("|url\s*\‍(\s*([\'\"])\s*\S+script\s*:.*?([\'\"])\s*\‍)|si",
    //                           "url(\\1$trans_image_path\\2)", $content);
 
    // first check for 8bit sequences and disallowed control characters
    if (preg_match('/[\16-\37\200-\377]+/',$content)) {
        $content = '<!-- style block removed by html filter due to presence of 8bit characters -->';
        return array($content, $newpos);
    }
 
    // remove @import line
    $content = preg_replace("/^\s*(@import.*)$/mi","\n<!-- @import rules forbidden -->\n",$content);
 
    $content = preg_replace("/(\\\\)?u(\\\\)?r(\\\\)?l(\\\\)?/i", 'url', $content);
    preg_match_all("/url\s*\‍((.+)\‍)/si",$content,$aMatch);
    if (count($aMatch)) {
        $aValue = $aReplace = array();
        foreach($aMatch[1] as $sMatch) {
            // url value
            $urlvalue = $sMatch;
            tln_fixurl('style',$urlvalue, $trans_image_path, $block_external_images);
            $aValue[] = $sMatch;
            $aReplace[] = $urlvalue;
        }
        $content = str_replace($aValue,$aReplace,$content);
    }
 
    $contentTemp = $content;
    tln_defang($contentTemp);
    tln_unspace($contentTemp);
 
    $match   = array('/\/\*.*\*\//',
                    '/expression/i',
                    '/behaviou*r/i',
                    '/binding/i',
                    '/include-source/i',
                    '/javascript/i',
                    '/script/i',
                    '/position/i');
    $replace = array('','idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', 'idiocy', '');
    $contentNew = preg_replace($match, $replace, $contentTemp);
    if ($contentNew !== $contentTemp) {
        $content = $contentNew;
    }
    return array($content, $newpos);
}

References tln_defang(), tln_fixurl(), and tln_unspace().

Referenced by tln_sanitize().

Here is the call graph for this function:

Here is the caller graph for this function:

tln_fixurl()

tln_fixurl	(		$attname,
		&	$attvalue,
			$trans_image_path,
			$block_external_images
	)

Replace empty src tags with the blank image. src is only used for frames, images, and image inputs. Doing a replace should not affect them working as should be, however it will stop IE from being kicked off when src for img tags are not set

Definition at line 598 of file htmlfilter.php.

{
    $sQuote = '"';
    $attvalue = trim($attvalue);
    if ($attvalue && ($attvalue[0] =='"'|| $attvalue[0] == "'")) {
        // remove the double quotes
        $sQuote = $attvalue[0];
        $attvalue = trim(substr($attvalue,1,-1));
    }
 
    if ($attvalue == '') {
        $attvalue = $sQuote . $trans_image_path . $sQuote;
    } else {
        // first, disallow 8 bit characters and control characters
        if (preg_match('/[\0-\37\200-\377]+/',$attvalue)) {
            switch ($attname) {
                case 'href':
                    $attvalue = $sQuote . 'http://invalid-stuff-detected.example.com' . $sQuote;
                    break;
                default:
                    $attvalue = $sQuote . $trans_image_path . $sQuote;
                    break;
            }
        } else {
            $aUrl = parse_url($attvalue);
            if (isset($aUrl['scheme'])) {
                switch(strtolower($aUrl['scheme'])) {
                    case 'mailto':
                    case 'http':
                    case 'https':
                    case 'ftp':
                        if ($attname != 'href') {
                            if ($block_external_images == true) {
                                $attvalue = $sQuote . $trans_image_path . $sQuote;
                            } else {
                                if (!isset($aUrl['path'])) {
                                    $attvalue = $sQuote . $trans_image_path . $sQuote;
                                }
                            }
                        } else {
                            $attvalue = $sQuote . $attvalue . $sQuote;
                        }
                        break;
                    case 'outbind':
                        $attvalue = $sQuote . $attvalue . $sQuote;
                        break;
                    case 'cid':
                        $attvalue = $sQuote . $attvalue . $sQuote;
                        break;
                    default:
                        $attvalue = $sQuote . $trans_image_path . $sQuote;
                        break;
                }
            } else {
                if (!isset($aUrl['path']) || $aUrl['path'] != $trans_image_path) {
                    $$attvalue = $sQuote . $trans_image_path . $sQuote;
                }
            }
        }
    }
}

Referenced by tln_fixatts(), and tln_fixstyle().

Here is the caller graph for this function:

tln_getnxtag()

tln_getnxtag	(		$body,
			$offset
	)

This function looks for the next tag.

Parameters

string	$body	String where to look for the next tag.
integer	$offset	Start looking from here.

Returns

array|boolean false if no more tags exist in the body, or an array with the following members:

• string with the name of the tag
• array with attributes and their values
• integer with tag type (1, 2, or 3)
• integer where the tag starts (starting "<")
• integer where the tag ends (ending ">") first three members will be false, if the tag is invalid.

We are here: blah blah <tag attribute="value"> ------—^

There are 3 kinds of tags:

1. Opening tag, e.g.: aClosing tag, e.g.:
2. XHTML-style content-less tag, e.g.:

A comment or an SGML declaration.

Assume tagtype 1 for now. If it's type 3, we'll switch values later.

Look for next [\W-_], which will indicate the end of the tag name.

$match can be either of these: '>' indicating the end of the tag entirely. '\s' indicating the end of the tag name. '/' indicating that this is type-3 xhtml tag.

Whatever else we find there indicates an invalid tag.

This is an xhtml-style tag with a closing / at the end, like so: . Check if it's followed by the closing bracket. If not, then this tag is invalid

Check if it's whitespace

This is an invalid tag! Look for the next closing ">".

At this point we're here: <tagname attribute='blah'> ----—^

At this point we loop in order to find all attributes.

Non-closed tag.

See if we arrived at a ">" or "/>", which means that we reached the end of the tag.

Yep. So we did.

There are several types of attributes, with optional [:space:] between members. Type 1: attrname[:space:]=[:space:]'CDATA' Type 2: attrname[:space:]=[:space:]"CDATA" Type 3: attr[:space:]=[:space:]CDATA Type 4: attrname

We leave types 1 and 2 the same, type 3 we check for '"' and convert to "&quot" if needed, then wrap in double quotes. Type 4 we convert into: attrname="yes".

Looks like body ended before the end of tag.

We arrived at the end of attribute name. Several things possible here: '>' means the end of the tag and this is attribute type 4 '/' if followed by '>' means the same thing as above '\s' means a lot of things – look what it's followed by. anything else means the attribute is invalid.

This is an xhtml-style tag with a closing / at the end, like so: . Check if it's followed by the closing bracket. If not, then this tag is invalid

Skip whitespace and see what we arrive at.

Two things are valid here: '=' means this is attribute type 1 2 or 3. \w means this was attribute type 4. anything else we ignore and re-loop. End of tag and invalid stuff will be caught by our checks at the beginning of the loop.

Here are 3 possibilities: "'" attribute type 1 '"' attribute type 2 everything else is the content of tag type 3

These are hateful. Look for \s, or >.

If it's ">" it will be caught at the top.

That was attribute type 4.

An illegal character. Find next '>' and return.

The fact that we got here indicates that the tag end was never found. Return invalid tag indication so it gets stripped.

Definition at line 157 of file htmlfilter.php.

{
    if ($offset > strlen($body)) {
        return false;
    }
    $lt = tln_findnxstr($body, $offset, '<');
    if ($lt == strlen($body)) {
        return false;
    }
    $pos = tln_skipspace($body, $lt + 1);
    if ($pos >= strlen($body)) {
        return array(false, false, false, $lt, strlen($body));
    }
    switch (substr($body, $pos, 1)) {
    case '/':
        $tagtype = 2;
        $pos++;
        break;
    case '!':
            if (substr($body, $pos + 1, 2) == '--') {
            $gt = strpos($body, '-->', $pos);
            if ($gt === false) {
                $gt = strlen($body);
            } else {
                $gt += 2;
            }
            return array(false, false, false, $lt, $gt);
        } else {
            $gt = tln_findnxstr($body, $pos, '>');
            return array(false, false, false, $lt, $gt);
        }
        break;
    default:
        $tagtype = 1;
        break;
    }
 
    $regary = tln_findnxreg($body, $pos, '[^\w\-_]');
    if ($regary == false) {
        return array(false, false, false, $lt, strlen($body));
    }
    list($pos, $tagname, $match) = $regary;
    $tagname = strtolower($tagname);
 
    switch ($match) {
    case '/':
        if (substr($body, $pos, 2) == '/>') {
            $pos++;
            $tagtype = 3;
        } else {
            $gt = tln_findnxstr($body, $pos, '>');
            $retary = array(false, false, false, $lt, $gt);
            return $retary;
        }
            //intentional fall-through
    case '>':
        return array($tagname, false, $tagtype, $lt, $pos);
        break;
    default:
        if (!preg_match('/\s/', $match)) {
            $gt = tln_findnxstr($body, $lt, '>');
            return array(false, false, false, $lt, $gt);
        }
        break;
    }
 
    $attary = array();
 
    while ($pos <= strlen($body)) {
        $pos = tln_skipspace($body, $pos);
        if ($pos == strlen($body)) {
            return array(false, false, false, $lt, $pos);
        }
        $matches = array();
        if (preg_match('%^(\s*)(>|/>)%s', substr($body, $pos), $matches)) {
            $pos += strlen($matches[1]);
            if ($matches[2] == '/>') {
                $tagtype = 3;
                $pos++;
            }
            return array($tagname, $attary, $tagtype, $lt, $pos);
        }
 
        $regary = tln_findnxreg($body, $pos, '[^\w\-_]');
        if ($regary == false) {
            return array(false, false, false, $lt, strlen($body));
        }
        list($pos, $attname, $match) = $regary;
        $attname = strtolower($attname);
        switch ($match) {
        case '/':
            if (substr($body, $pos, 2) == '/>') {
                $pos++;
                $tagtype = 3;
            } else {
                $gt = tln_findnxstr($body, $pos, '>');
                $retary = array(false, false, false, $lt, $gt);
                return $retary;
            }
                //intentional fall-through
        case '>':
            $attary{$attname} = '"yes"';
            return array($tagname, $attary, $tagtype, $lt, $pos);
            break;
        default:
            $pos = tln_skipspace($body, $pos);
            $char = substr($body, $pos, 1);
            if ($char == '=') {
                $pos++;
                $pos = tln_skipspace($body, $pos);
                $quot = substr($body, $pos, 1);
                if ($quot == '\'') {
                        $regary = tln_findnxreg($body, $pos + 1, '\'');
                    if ($regary == false) {
                        return array(false, false, false, $lt, strlen($body));
                    }
                    list($pos, $attval, $match) = $regary;
                    $pos++;
                    $attary{$attname} = '\'' . $attval . '\'';
                } elseif ($quot == '"') {
                    $regary = tln_findnxreg($body, $pos + 1, '\"');
                    if ($regary == false) {
                        return array(false, false, false, $lt, strlen($body));
                    }
                    list($pos, $attval, $match) = $regary;
                    $pos++;
                            $attary{$attname} = '"' . $attval . '"';
                } else {
                    $regary = tln_findnxreg($body, $pos, '[\s>]');
                    if ($regary == false) {
                        return array(false, false, false, $lt, strlen($body));
                    }
                    list($pos, $attval, $match) = $regary;
                    $attval = preg_replace('/\"/s', '&quot;', $attval);
                    $attary{$attname} = '"' . $attval . '"';
                }
            } elseif (preg_match('|[\w/>]|', $char)) {
                $attary{$attname} = '"yes"';
            } else {
                $gt = tln_findnxstr($body, $pos, '>');
                return array(false, false, false, $lt, $gt);
            }
            break;
        }
    }
    return array(false, false, false, $lt, strlen($body));
}

References tln_findnxreg(), tln_findnxstr(), and tln_skipspace().

Referenced by tln_sanitize().

Here is the call graph for this function:

Here is the caller graph for this function:

tln_sanitize()

tln_sanitize	(		$body,
			$tag_list,
			$rm_tags_with_content,
			$self_closing_tags,
			$force_tag_closing,
			$rm_attnames,
			$bad_attvals,
			$add_attr_to_tag,
			$trans_image_path,
			$block_external_images
	)

Parameters

string	$body	The HTML you wish to filter
array	$tag_list	see description above
array	$rm_tags_with_content	see description above
array	$self_closing_tags	see description above
boolean	$force_tag_closing	see description above
array	$rm_attnames	see description above
array	$bad_attvals	see description above
array	$add_attr_to_tag	see description above
string	$trans_image_path
boolean	$block_external_images

Returns

string Sanitized html safe to show on your pages.

Normalize rm_tags and rm_tags_with_content.

See if tag_list is of tags to remove or tags to allow. false means remove these tags true means allow these tags

Take care of netscape's stupid javascript entities like &{alert('boo')};

Take care of <style>

Got to the end of tag we needed to remove.

$rm_tags_with_content

See if this is a self-closing type and change tagtype appropriately.

See if we should skip this tag and any content inside it.

Convert body into div.

This is where we run other checks.

Definition at line 842 of file htmlfilter.php.

  {
    $rm_tags = array_shift($tag_list);
    @array_walk($tag_list, 'tln_casenormalize');
    @array_walk($rm_tags_with_content, 'tln_casenormalize');
    @array_walk($self_closing_tags, 'tln_casenormalize');
    $curpos = 0;
    $open_tags = array();
    $trusted = "<!-- begin tln_sanitized html -->\n";
    $skip_content = false;
    $body = preg_replace('/&(\{.*?\};)/si', '&amp;\\1', $body);
    while (($curtag = tln_getnxtag($body, $curpos)) != false) {
        list($tagname, $attary, $tagtype, $lt, $gt) = $curtag;
        $free_content = substr($body, $curpos, $lt-$curpos);
        if ($tagname == "style" && $tagtype == 1){
            list($free_content, $curpos) =
                tln_fixstyle($body, $gt+1, $trans_image_path, $block_external_images);
            if ($free_content != FALSE){
                if ( !empty($attary) ) {
                    $attary = tln_fixatts($tagname,
                                         $attary,
                                         $rm_attnames,
                                         $bad_attvals,
                                         $add_attr_to_tag,
                                         $trans_image_path,
                                         $block_external_images
                                         );
                }
                $trusted .= tln_tagprint($tagname, $attary, $tagtype);
                $trusted .= $free_content;
                $trusted .= tln_tagprint($tagname, null, 2);
            }
            continue;
        }
        if ($skip_content == false){
            $trusted .= $free_content;
        }
        if ($tagname != false) {
            if ($tagtype == 2) {
                if ($skip_content == $tagname) {
                    $tagname = false;
                    $skip_content = false;
                } else {
                    if ($skip_content == false) {
                        if ($tagname == "body") {
                            $tagname = "div";
                        }
                        if (isset($open_tags{$tagname}) &&
                            $open_tags{$tagname} > 0
                        ) {
                            $open_tags{$tagname}--;
                        } else {
                            $tagname = false;
                        }
                    }
                }
            } else {
                if ($skip_content == false) {
                    if ($tagtype == 1
                        && in_array($tagname, $self_closing_tags)
                    ) {
                        $tagtype = 3;
                    }
                    if ($tagtype == 1
                        && in_array($tagname, $rm_tags_with_content)
                    ) {
                        $skip_content = $tagname;
                    } else {
                        if (($rm_tags == false
                             && in_array($tagname, $tag_list)) ||
                            ($rm_tags == true
                                && !in_array($tagname, $tag_list))
                        ) {
                            $tagname = false;
                        } else {
                            if ($tagname == "body"){
                                $tagname = "div";
                                $attary = tln_body2div($attary, $trans_image_path);
                            }
                            if ($tagtype == 1) {
                                if (isset($open_tags{$tagname})) {
                                    $open_tags{$tagname}++;
                                } else {
                                    $open_tags{$tagname} = 1;
                                }
                            }
                            if (is_array($attary) && sizeof($attary) > 0) {
                                $attary = tln_fixatts(
                                    $tagname,
                                    $attary,
                                    $rm_attnames,
                                    $bad_attvals,
                                    $add_attr_to_tag,
                                    $trans_image_path,
                                    $block_external_images
                                );
                            }
                        }
                    }
                }
            }
            if ($tagname != false && $skip_content == false) {
                $trusted .= tln_tagprint($tagname, $attary, $tagtype);
            }
        }
        $curpos = $gt + 1;
    }
    $trusted .= substr($body, $curpos, strlen($body) - $curpos);
    if ($force_tag_closing == true) {
        foreach ($open_tags as $tagname => $opentimes) {
            while ($opentimes > 0) {
                $trusted .= '</' . $tagname . '>';
                $opentimes--;
            }
        }
        $trusted .= "\n";
    }
    $trusted .= "<!-- end tln_sanitized html -->\n";
    return $trusted;
}

References tln_body2div(), tln_fixatts(), tln_fixstyle(), tln_getnxtag(), and tln_tagprint().

Referenced by HTMLFilter().

Here is the call graph for this function:

Here is the caller graph for this function:

tln_skipspace()

tln_skipspace	(		$body,
			$offset
	)

This function skips any whitespace from the current position within a string and to the next non-whitespace value.

Parameters

string	$body	the string
integer	$offset	the offset within the string where we should start looking for the next non-whitespace character.

Returns

integer the location within the $body where the next non-whitespace char is located.

Definition at line 84 of file htmlfilter.php.

{
    preg_match('/^(\s*)/s', substr($body, $offset), $matches);
    if (sizeof($matches[1])) {
        $count = strlen($matches[1]);
        $offset += $count;
    }
    return $offset;
}

Referenced by tln_getnxtag().

Here is the caller graph for this function:

tln_tagprint()

tln_tagprint	(		$tagname,
			$attary,
			$tagtype
	)

This function returns the final tag out of the tag name, an array of attributes, and the type of the tag.

htmlfilter.inc

This set of functions allows you to filter html in order to remove any malicious tags from it. Useful in cases when you need to filter user input for any cross-site-scripting attempts.

Copyright (C) 2002-2004 by Duke University

This library is free software; you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation; either version 2.1 of the License, or (at your option) any later version.

This library is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details.

You should have received a copy of the GNU Lesser General Public License along with this library; if not, write to the Free Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA

@Author Konstantin Riabitsev icon@.nosp@m.linu.nosp@m.x.duk.nosp@m.e.ed.nosp@m.u @Author Jim Jagielski <jim@j.nosp@m.aguN.nosp@m.ET.co.nosp@m.m / jimja.nosp@m.g@gm.nosp@m.ail.c.nosp@m.om> @Version 1.1 ($Date$) This function is called by tln_sanitize internally.

Parameters

string	$tagname	the name of the tag.
array	$attary	the array of attributes and their values
integer	$tagtype	The type of the tag (see in comments).

Returns

string A string with the final tag representation.

Definition at line 41 of file htmlfilter.php.

{
    if ($tagtype == 2) {
        $fulltag = '</' . $tagname . '>';
    } else {
        $fulltag = '<' . $tagname;
        if (is_array($attary) && sizeof($attary)) {
            $atts = array();
            while (list($attname, $attvalue) = each($attary)) {
                array_push($atts, "$attname=$attvalue");
            }
            $fulltag .= ' ' . join(' ', $atts);
        }
        if ($tagtype == 3) {
            $fulltag .= ' /';
        }
        $fulltag .= '>';
    }
    return $fulltag;
}

Referenced by tln_sanitize().

Here is the caller graph for this function:

tln_unspace()

tln_unspace

(

&

$attvalue

)

Kill any tabs, newlines, or carriage returns.

Our friends the makers of the browser with 95% market value decided that it'd be funny to make "java[tab]script" be just as good as "javascript".

Parameters

string

$attvalue

The attribute value before extraneous spaces removed.

Definition at line 491 of file htmlfilter.php.

{
    if (strcspn($attvalue, "\t\r\n\0 ") != strlen($attvalue)) {
        $attvalue = str_replace(
            array("\t", "\r", "\n", "\0", " "),
            array('', '', '', '', ''),
            $attvalue
        );
    }
}

Referenced by tln_fixatts(), and tln_fixstyle().

Here is the caller graph for this function: