| blob | fae010f48bc286f736d0a612ece42ff637b83073 |
8 <head>
12 <meta name="keywords" content="HTMLPurifier, HTML Purifier, HTML, filter, filtering, HTML_Safe, PEAR, comparison, kses, striptags, SafeHTMLChecker" />
13 </head>
14 <body>
21 <p>With the advent of
23 gone from passive consumer to active producer of content on the World Wide
27 put the user in control.</p>
29 <p>Give the user too much control, however, and you set yourself up
30 for <a href="http://en.wikipedia.org/wiki/Cross-site_scripting"><abbr>XSS</abbr></a> attacks. For this reason,
32 has proven to be both a blessing and a curse, and the software that processes
33 it must strike a fine balance between security and usability. How do
34 we prevent users from injecting JavaScript or inserting malformed
41 developer has come across this problem before, and many have tried
42 (albeit unsuccessfully) to solve this problem. We will analyze existing
43 libraries to demonstrate how they are ineffective and, of course,
45 standards-compliance.</p>
47 <p>I will take no quarter and pull no punches: as of the time of writing,
49 for richly formatted documents. But, nonetheless, there is a necessary
50 disclaimer:</p>
53 This comparison document was written by the author of HTML Purifier,
55 mean that it is biased: I have made every attempt to be <strong>factual and
56 fair</strong>, and I hope that you will agree, by the time you finish reading
58 filter out there today.
59 </p>
70 <thead>
71 <tr>
83 </tr>
84 </thead>
86 <tbody>
88 <tr>
100 </tr>
102 <tr>
114 </tr>
116 <tr>
128 </tr>
130 <tr>
142 </tr>
144 <tr>
156 </tr>
158 <tr>
170 </tr>
172 </tbody>
174 </table>
175 </div>
177 <p><a href="#Tidy">HTML Tidy</a> is omitted from this list because it is not an <abbr>HTML</abbr>
178 filter.</p>
184 A clever person solves a problem.
185 A wise person avoids it.
186 </div>
188 </blockquote>
190 <p>Before we jump into the weird and not-so-wonderful world
192 markup libraries. While libraries of this type really shouldn't be
194 they are the number one method of taking user input and processing it into
195 something more than plain old text. These libraries forgo
201 such markup languages (although it should be noted that
202 Wikitext and Markdown can allow
204 The benefits (to those who use it, anyway) are clear: simplicity and
205 security.
206 </p>
209 <thead>
210 <tr>
213 </tr>
214 </thead>
215 <tbody>
216 <tr>
219 </tr>
220 <tr>
223 </tr>
224 <tr>
227 </tr>
228 <tr>
231 </tr>
232 <tr>
234 <td><tt><b>B</b> <i>i</i> <a href="http://www.example.com/">link</a></tt></td>
235 </tr>
236 <tr>
239 </tr>
240 </tbody>
241 </table>
246 There are many variants of Wikitext currently extant.</li>
247 <li>Strictly speaking, the Markdown syntax is not equivalent: bold text
250 however, map those two semantic tags to the associated styling, so
251 many users assume that it really is italics (and use it improperly for,
252 say, book titles.)</li>
253 </ol>
258 source code is often criticized for being difficult to read. For example,
259 compare:</p>
261 <pre>
262 * Item 1
263 * Item 2
264 </pre>
268 <pre>
273 </pre>
275 <p>Which would you prefer to edit? The answer seems obvious, but be careful
276 not to fall into the fallacy of <a
280 editor, which blows earlier choices out of the water in terms
281 of usability.</p>
283 <p>Note that rich text editors and alternate markup syntaxes are not
284 mutually exclusive, but, when push comes to shove, it's easier
286 markup language. And in the cases when it is done, you usually end up with
287 a live preview, not a true rich text editor.</p>
292 editors aren't all that great.</q> There are many good arguments
293 against these editors, and <a
295 people have written essays</a> devoted to
297 In addition to the usual arguments against said editors, the web poses
298 another limitation: no JavaScript means no
299 editor, and no editor means... (gasp) manually typing in code.</p>
301 <p>Even the most dogmatic purist, however, should recognize that for all
303 There are steps you can take to mitigate the associated drawbacks of
304 these editors.</p>
306 <p>It is often asserted that
309 this is the case with any markup language that allows the smallest
312 A good way to reduce this trouble is to simply eliminate the
313 dialogue boxes that allow users to change colors or fonts (which
314 usually have no legitimate use) and adopt a
316 allowing users to select contextually correct formatting styles
317 for segments of text.</p>
318 </blockquote>
320 <p>Simplicity is also a double-edged sword. The moment any remotely
321 complex markup is needed, these lightweight markup languages fail to
322 produce. Sure you can make '''this text bold''' with Wikitext, but that
326 filters in that their whitelist is too restrictive (besides the fact that
327 their table markup is extraordinarily complex).</p>
333 the angled brackets with square brackets and omitting the occasional parameter
334 name? How much more un-original can you get? Somehow, I don't think BBCode
335 was meant to readable. <a
338 <blockquote>
339 BBCode was devised and put to use in order to provide a safer, easier
340 and more limited way of allowing users to format their messages.
342 which could be used to break/imitate parts of the layout, or run
343 JavaScript. Some implementations of BBCode have suffered problems related
345 security that was intended to be given by BBCode.
346 </blockquote>
350 <blockquote>
351 BBCode came to life when developers where too
353 and decided to invent their own markup language. As with all products of
354 laziness, the result is completely inconsistent, unstandardized, and
355 widely adopted.
356 </blockquote>
358 <p>Well, developers, the whole point of HTML Purifier is that I do the
359 work so you can just execute the ridiculously simple
365 <p>These alternative markup languages have their shiny points, and HTML
366 Purifier is not meant to replace them. However, a major reason for
368 using these languages?</p>
372 <p>Dave Raggett's
376 The premise is simple, the execution effective. Tidy is, in short, a great
379 <p>It is not, however, a filter. I am often surprised when people ask
390 </blockquote>
399 <p>This is not to say that I haven't seen Tidy be used in this sort of
401 output before shuttling it off to the browser. The developers, nevertheless,
402 agree that this is only a band-aid solution, and that the real way
403 to fix it is to fix the parser. Tidy's great, but in terms of security,
404 it's not suitable for untrusted sources.</p>
408 <p>I've ordered my analyses according to how bad a library is. The worst
409 is first, and then we move up the spectrum. I will point out the most
410 flagrant problems with the libraries, but note that I will omit more
413 points through. The ideal solution, however, must do all these things.</p>
415 <p>Note that besides striptags,
417 attacks. None of them (save Safe HTML Checker) fare very well
418 in the standards-compliance department though.</p>
428 </table>
432 the classic solution for attempting to clean up
435 The fact that it doesn't validate attributes at all means that anyone can
438 <p>While
439 this can be bandaided with a series of regular expressions that strip out
441 quirky browser behavior), striptags() is fundamentally flawed and should not be
442 used.
443 </p>
447 <p>Though its title may not imply it,
449 is a souped up version of striptags() with the ability to inspect
450 attributes. (Don't mind the hastily tacked on query escaping function).</p>
463 </table>
465 <p>PHP Input Filter implements an
467 performs very basic checks on whether or not tags and attributes have
468 been defined in the whitelist as well as some
470 the user to define what they'll permit.</p>
472 <p>With absolutely no checking of well-formedness, it is trivially easy
473 to trick the filter into leaving unclosed tags lying around. While to some
475 basic sanity checks like this must be implemented, otherwise a user
476 can mangle a website's layout.</p>
478 <p>More troubles: Woe to
481 layout not to be badly mutilated. To top things off,
482 the filter doesn't even preserve data properly: attributes have all
483 spaces stripped out of them. Stay away, stay away!</p>
489 It should be noted that this is the same library as
491 branding (and a different version number).</p>
504 </table>
506 <p>HTML_Safe's mechanism of action involves parsing
509 validation and filtering as the handlers are called. HTML_Safe does a lot
512 is fundamentally flawed: blacklists.</p>
514 <p>This library maintains arrays of dangerous tags, attributes and
517 intelligently disabled by default in favor of a protocol whitelist.)
518 What this means is that HTML_Safe has no qualms of accepting input
520 the tags in those arrays. Scratch standards-compliance (and that was
521 without even considering proper nesting).</p>
523 <p>For now, HTML_Safe might be safe from
525 In the future, however, one of the infinitely many tags that HTML_Safe lets
526 through might just possibly be given special functionality by browser vendors.
528 solution puts you at a perpetual arms race against crackers who are constantly
529 discovering new and inventive ways to abuse tags and attributes that you
530 didn't blacklist.</p>
550 </table>
552 <p>To be truthful, I didn't do as comprehensive a code survey for kses
553 as I did for some of the other libraries. Out of
554 all the classes I've reviewed so far, kses was definitely the hardest to
555 understand.</p>
557 <p>kses's modus operandi is splitting up html with a monster regexp
559 suffers from the same problems as Input Filter: no well-formedness
560 checks leading to rampant runaway tags (and no standards-compliance).
561 WordPress, the primary user of kses today, had to implement their
562 own custom tag-balancing code to fix this problem: don't use this
563 library without some equivalent!</p>
565 <p>Its whitelist syntax, however, is the most complex of all these libraries,
566 so I'm going to take some time to argue why this particular implementation
567 is bad. The author of this library was thoughtful enough to provide some
568 basic constraint checks on attributes like maxlen and maxval. Now, barring
569 the fact that there simply aren't enough checks, and the fact that they are
570 all lumped together in one function, we now must wonder whether or not
571 the user will go through the trouble of specifying the maximum length
572 of a title attribute.</p>
574 <p>I have my opinions about inherent human laziness, but perhaps WordPress's
575 default filterset is the most telling example:</p>
577 <pre>
578 $allowedposttags = array (
579 /* formatted and trimmed */
580 'hr' => array (
581 'align' => array (),
582 'noshade' => array (),
583 'size' => array (),
584 'width' => array ()
585 )
586 );
587 </pre>
589 <p>Hmm... do I see a blatant lack of attribute constraints? Conclusion:
590 if the user can get away with not doing work, they will! The biggest
592 the whitelist. The whitelist is just as important as the code that uses
597 <p>
599 HTML Checker</a> is (to my knowledge) the first attempt to make a filter
602 search result must have done something right.</p>
615 </table>
617 <p>Indeed, it is quite a well-written piece of code. It demonstrates
618 knowledge of inline versus block elements, thus almost nearly getting
619 nesting correct (the only exception is an unimplemented omitted SGML
622 <p>Unfortunately, part of the reason why it works so well is that it's
623 extremely restrictive. No styling, no tables, very few attributes.
624 Perfectly appropriate for blog comments, but then again, there's always
625 BBCode. This probably means that Safe HTML Checker has a different
626 goal than HTML Purifier.</p>
629 is also quite strict. Accidentally missed a < sign? The parser will
630 complain with the cryptic message:
632 is not well-formed</q>.
633 The solution is not as simple as just switching to a more permissive
634 parser: Safe HTML Checker relies on the fact that the parser will have
635 matched up the tags for them.</p>
650 </table>
661 </table>
663 <p>This is not to say that HTML Purifier doesn't have problems of its own.
664 It's a fairly nascent library (that doesn't mean its buggy though), it's big
665 (while the others usually fit in one file, this one requires a huge
667 features.</a> But even in its current state,
668 HTML Purifier is far better than the other libraries.</p>
672 </div>
673 </body>
674 </html>