Skip to content

Commit a1fdb20

Browse files
alecplalecpl
committed
Extend rcube_washtml with SVG support
1 parent 0bde1c2 commit a1fdb20

File tree

2 files changed

+120
-32
lines changed

2 files changed

+120
-32
lines changed

‎program/lib/Roundcube/rcube_washtml.php‎

Lines changed: 104 additions & 24 deletions
Original file line numberDiff line numberDiff line change
@@ -97,7 +97,20 @@ class rcube_washtml
9797
'tbody', 'td', 'tfoot', 'th', 'thead', 'tr', 'tt', 'u', 'ul', 'var', 'wbr', 'img',
9898
'video', 'source',
9999
// form elements
100-
'button', 'input', 'textarea', 'select', 'option', 'optgroup'
100+
'button', 'input', 'textarea', 'select', 'option', 'optgroup',
101+
// SVG
102+
'svg', 'altglyph', 'altglyphdef', 'altglyphitem', 'animate', 'animatecolor',
103+
'animatemotion', 'animatetransform', 'circle', 'clippath', 'defs', 'desc',
104+
'ellipse', 'font', 'g', 'glyph', 'glyphref', 'hkern', 'image', 'line',
105+
'lineargradient', 'marker', 'mask', 'mpath', 'path', 'pattern',
106+
'polygon', 'polyline', 'radialgradient', 'rect', 'set', 'stop', 'switch', 'symbol',
107+
'text', 'textpath', 'tref', 'tspan', 'use', 'view', 'vkern', 'filter',
108+
// SVG Filters
109+
'feblend', 'fecolormatrix', 'fecomponenttransfer', 'fecomposite',
110+
'feconvolvematrix', 'fediffuselighting', 'fedisplacementmap',
111+
'feflood', 'fefunca', 'fefuncb', 'fefuncg', 'fefuncr', 'fegaussianblur',
112+
'feimage', 'femerge', 'femergenode', 'femorphology', 'feoffset',
113+
'fespecularlighting', 'fetile', 'feturbulence',
101114
);
102115

103116
/* Ignore these HTML tags and their content */
@@ -111,14 +124,53 @@ class rcube_washtml
111124
'abbr', 'char', 'charoff', 'clear', 'compact', 'coords', 'vspace', 'hspace',
112125
'cellborder', 'size', 'lang', 'dir', 'usemap', 'shape', 'media',
113126
// attributes of form elements
114-
'type', 'rows', 'cols', 'disabled', 'readonly', 'checked', 'multiple', 'value'
127+
'type', 'rows', 'cols', 'disabled', 'readonly', 'checked', 'multiple', 'value',
128+
// SVG
129+
'accent-height', 'accumulate', 'additivive', 'alignment-baseline',
130+
'ascent', 'attributename', 'attributetype', 'azimuth', 'basefrequency', 'baseprofile',
131+
'baseline-shift', 'begin', 'bias', 'by', 'clip', 'clip-path', 'clip-rule',
132+
'color', 'color-interpolation', 'color-interpolation-filters', 'color-profile',
133+
'color-rendering', 'cx', 'cy', 'd', 'dx', 'dy', 'diffuseconstant', 'direction',
134+
'display', 'divisor', 'dur', 'edgemode', 'elevation', 'end', 'fill', 'fill-opacity',
135+
'fill-rule', 'filter', 'flood-color', 'flood-opacity', 'font-family', 'font-size',
136+
'font-size-adjust', 'font-stretch', 'font-style', 'font-variant', 'font-weight',
137+
'fx', 'fy', 'g1', 'g2', 'glyph-name', 'glyphref', 'gradientunits', 'gradienttransform',
138+
'image-rendering', 'in', 'in2', 'k', 'k1', 'k2', 'k3', 'k4', 'kerning', 'keypoints',
139+
'keysplines', 'keytimes', 'lengthadjust', 'letter-spacing', 'kernelmatrix',
140+
'kernelunitlength', 'lighting-color', 'local', 'marker-end', 'marker-mid',
141+
'marker-start', 'markerheight', 'markerunits', 'markerwidth', 'maskcontentunits',
142+
'maskunits', 'max', 'mask', 'mode', 'min', 'numoctaves', 'offset', 'operator',
143+
'opacity', 'order', 'orient', 'orientation', 'origin', 'overflow', 'paint-order',
144+
'path', 'pathlength', 'patterncontentunits', 'patterntransform', 'patternunits',
145+
'points', 'preservealpha', 'r', 'rx', 'ry', 'radius', 'refx', 'refy', 'repeatcount',
146+
'repeatdur', 'restart', 'rotate', 'scale', 'seed', 'shape-rendering', 'specularconstant',
147+
'specularexponent', 'spreadmethod', 'stddeviation', 'stitchtiles', 'stop-color',
148+
'stop-opacity', 'stroke-dasharray', 'stroke-dashoffset', 'stroke-linecap',
149+
'stroke-linejoin', 'stroke-miterlimit', 'stroke-opacity', 'stroke', 'stroke-width',
150+
'surfacescale', 'targetx', 'targety', 'transform', 'text-anchor', 'text-decoration',
151+
'text-rendering', 'textlength', 'u1', 'u2', 'unicode', 'values', 'viewbox',
152+
'visibility', 'vert-adv-y', 'version', 'vert-origin-x', 'vert-origin-y', 'word-spacing',
153+
'wrap', 'writing-mode', 'xchannelselector', 'ychannelselector', 'x', 'x1', 'x2',
154+
'xmlns', 'y', 'y1', 'y2', 'z', 'zoomandpan',
155+
// XML
156+
'xml:id', 'xlink:title'
115157
);
116158

117159
/* Elements which could be empty and be returned in short form (<tag />) */
118160
static $void_elements = array('area', 'base', 'br', 'col', 'command', 'embed', 'hr',
119-
'img', 'input', 'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr'
161+
'img', 'input', 'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr',
162+
// SVG
163+
'altglyph', 'altglyphdef', 'altglyphitem', 'animate', 'animatecolor',
164+
'animatemotion', 'animatetransform', 'circle', 'clippath', 'defs', 'desc',
165+
'ellipse', 'font', 'g', 'glyph', 'glyphref', 'hkern', 'image', 'line',
166+
'lineargradient', 'marker', 'mask', 'mpath', 'path', 'pattern',
167+
'polygon', 'polyline', 'radialgradient', 'rect', 'set', 'stop', 'switch', 'symbol',
168+
'text', 'textpath', 'tref', 'tspan', 'use', 'view', 'vkern', 'filter',
120169
);
121170

171+
/* Attributes that may contain insecure content */
172+
static $insecure_attribs = array('href', 'to', 'from');
173+
122174
/* State for linked objects in HTML */
123175
public $extlinks = false;
124176

@@ -149,10 +201,11 @@ class rcube_washtml
149201
*/
150202
public function __construct($p = array())
151203
{
152-
$this->_html_elements = array_flip((array)$p['html_elements']) + array_flip(self::$html_elements) ;
153-
$this->_html_attribs = array_flip((array)$p['html_attribs']) + array_flip(self::$html_attribs);
154-
$this->_ignore_elements = array_flip((array)$p['ignore_elements']) + array_flip(self::$ignore_elements);
155-
$this->_void_elements = array_flip((array)$p['void_elements']) + array_flip(self::$void_elements);
204+
$this->_html_elements = array_flip((array)$p['html_elements']) + array_flip(self::$html_elements);
205+
$this->_html_attribs = array_flip((array)$p['html_attribs']) + array_flip(self::$html_attribs);
206+
$this->_insecure_attribs = array_flip((array)$p['insecure_attribs']) + array_flip(self::$insecure_attribs);
207+
$this->_ignore_elements = array_flip((array)$p['ignore_elements']) + array_flip(self::$ignore_elements);
208+
$this->_void_elements = array_flip((array)$p['void_elements']) + array_flip(self::$void_elements);
156209

157210
unset($p['html_elements'], $p['html_attribs'], $p['ignore_elements'], $p['void_elements']);
158211

@@ -232,25 +285,26 @@ private function wash_style($style)
232285
*/
233286
private function wash_attribs($node)
234287
{
235-
$t = '';
288+
$t = '';
236289
$washed = '';
237290

238-
foreach ($node->attributes as $key => $plop) {
239-
$key = strtolower($key);
240-
$value = $node->getAttribute($key);
291+
foreach ($node->attributes as $name => $attr) {
292+
$key = strtolower($name);
293+
$value = $attr->nodeValue;
241294

242295
if (isset($this->_html_attribs[$key]) ||
243-
($key == 'href' && ($value = trim($value))
296+
(isset($this->_insecure_attribs[$key])
297+
&& ($value = trim($value))
244298
&& !preg_match('!^(javascript|vbscript|data:text)!i', $value)
245299
&& preg_match('!^([a-z][a-z0-9.+-]+:|//|#).+!i', $value))
246300
) {
247-
$t .= ' ' . $key . '="' . htmlspecialchars($value, ENT_QUOTES) . '"';
301+
$t .= ' ' . $attr->nodeName . '="' . htmlspecialchars($value, ENT_QUOTES) . '"';
248302
}
249303
else if ($key == 'style' && ($style = $this->wash_style($value))) {
250304
// replace double quotes to prevent syntax error and XSS issues (#1490227)
251305
$t .= ' style="' . str_replace('"', '&quot;', $style) . '"';
252306
}
253-
else if ($key == 'background'
307+
else if ($key == 'background' || $key == 'href'
254308
|| ($key == 'src' && preg_match('/^(img|source)$/i', $node->tagName))
255309
|| ($key == 'poster' && strtolower($node->tagName) == 'video')
256310
) {
@@ -275,7 +329,7 @@ private function wash_attribs($node)
275329
}
276330
}
277331
else {
278-
$washed .= ($washed ? ' ' : '') . $key;
332+
$washed .= ($washed ? ' ' : '') . $attr->nodeName;
279333
}
280334
}
281335

@@ -322,8 +376,25 @@ private function dumpHtml($node, $level = 20)
322376
}
323377
else if (isset($this->_html_elements[$tagName])) {
324378
$content = $this->dumpHtml($node, $level);
325-
$dump .= '<' . $tagName . $this->wash_attribs($node) .
326-
($content === '' && isset($this->_void_elements[$tagName]) ? ' />' : ">$content</$tagName>");
379+
$dump .= '<' . $tagName;
380+
381+
if ($tagName == 'svg') {
382+
$xpath = new DOMXPath($node->ownerDocument);
383+
foreach ($xpath->query('namespace::*') as $ns) {
384+
if ($ns->nodeName != 'xmlns:xml') {
385+
$dump .= ' ' . $ns->nodeName . '="' . $ns->nodeValue . '"';
386+
}
387+
}
388+
}
389+
390+
$dump .= $this->wash_attribs($node);
391+
392+
if ($content === '' && isset($this->_void_elements[$tagName])) {
393+
$dump .= ' />';
394+
}
395+
else {
396+
$dump .= ">$content</$tagName>";
397+
}
327398
}
328399
else if (isset($this->_ignore_elements[$tagName])) {
329400
$dump .= '<!-- ' . htmlspecialchars($tagName, ENT_QUOTES) . ' not allowed -->';
@@ -375,14 +446,18 @@ public function wash($html)
375446
// Detect max nesting level (for dumpHTML) (#1489110)
376447
$this->max_nesting_level = (int) @ini_get('xdebug.max_nesting_level');
377448

449+
// SVG need to be parsed as XML
450+
$xml = stripos($html, '<svg') !== false || stripos($html, '<?xml') !== false;
451+
$method = $xml ? 'loadXML' : 'loadHTML';
452+
$options = 0;
453+
378454
// Use optimizations if supported
379455
if (PHP_VERSION_ID >= 50400) {
380-
@$node->loadHTML($html, LIBXML_PARSEHUGE | LIBXML_COMPACT);
381-
}
382-
else {
383-
@$node->loadHTML($html);
456+
$options = LIBXML_PARSEHUGE | LIBXML_COMPACT | LIBXML_NONET;
384457
}
385458

459+
@$node->{$method}($html, $options);
460+
386461
return $this->dumpHtml($node);
387462
}
388463

@@ -399,6 +474,8 @@ public function get_config($prop)
399474
*/
400475
private function cleanup($html)
401476
{
477+
$html = trim($html);
478+
402479
// special replacements (not properly handled by washtml class)
403480
$html_search = array(
404481
// space(s) between <NOBR>
@@ -420,17 +497,19 @@ private function cleanup($html)
420497
'',
421498
'<html>',
422499
);
500+
423501
$html = preg_replace($html_search, $html_replace, trim($html));
424502

425-
//-> Replace all of those weird MS Word quotes and other high characters
503+
// Replace all of those weird MS Word quotes and other high characters
426504
$badwordchars = array(
427505
"\xe2\x80\x98", // left single quote
428506
"\xe2\x80\x99", // right single quote
429507
"\xe2\x80\x9c", // left double quote
430508
"\xe2\x80\x9d", // right double quote
431509
"\xe2\x80\x94", // em dash
432-
"\xe2\x80\xa6" // elipses
510+
"\xe2\x80\xa6" // elipses
433511
);
512+
434513
$fixedwordchars = array(
435514
"'",
436515
"'",
@@ -439,6 +518,7 @@ private function cleanup($html)
439518
'&mdash;',
440519
'...'
441520
);
521+
442522
$html = str_replace($badwordchars, $fixedwordchars, $html);
443523

444524
// PCRE errors handling (#1486856), should we use something like for every preg_* use?
@@ -484,7 +564,7 @@ public static function html_tag_callback($matches)
484564
$tagname = $matches[2];
485565
$tagname = preg_replace(array(
486566
'/:.*$/', // Microsoft's Smart Tags <st1:xxxx>
487-
'/[^a-z0-9_\[\]\!-]/i', // forbidden characters
567+
'/[^a-z0-9_\[\]\!?-]/i', // forbidden characters
488568
), '', $tagname);
489569

490570
// fix invalid closing tags - remove any attributes (#1489446)

‎program/steps/mail/get.inc‎

Lines changed: 16 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -515,12 +515,20 @@ function rcmail_message_part_output($body, $part, $mimetype, $download)
515515
*/
516516
function rcmail_svg_filter($body)
517517
{
518-
$dom = new DOMDocument;
519-
$dom->loadXML($body);
520-
521-
foreach ($dom->getElementsByTagName('script') as $node) {
522-
$node->parentNode->removeChild($node);
523-
}
524-
525-
return $dom->saveXML() ?: '';
518+
// clean SVG with washhtml
519+
$wash_opts = array(
520+
'show_washed' => false,
521+
'allow_remote' => false,
522+
'charset' => RCUBE_CHARSET,
523+
'html_elements' => array('title'),
524+
// 'blocked_src' => 'program/resources/blocked.gif',
525+
);
526+
527+
// initialize HTML washer
528+
$washer = new rcube_washtml($wash_opts);
529+
530+
// allow CSS styles, will be sanitized by rcmail_washtml_callback()
531+
$washer->add_callback('style', 'rcmail_washtml_callback');
532+
533+
return $washer->wash($body);
526534
}

0 commit comments

Comments
 (0)