| blob | 64d941894e77ffda794a24eb112a7eb53ea1c477 |
12 <head>
16 <meta name="keywords" content="HTMLPurifier, HTML Purifier, HTML, filter, filtering, HTML_Safe, PEAR, comparison, kses, striptags, SafeHTMLChecker" />
17 </head>
18 <body>
25 <p>With the advent of
27 gone from passive consumer to active producer of content on the World Wide
31 put the user in control.</p>
33 <p>Give the user too much control, however, and you set yourself up
34 for <a href="http://en.wikipedia.org/wiki/Cross-site_scripting"><abbr>XSS</abbr></a> attacks. For this reason,
36 has proven to be both a blessing and a curse, and the software that processes
37 it must strike a fine balance between security and usability. How do
38 we prevent users from injecting JavaScript or inserting malformed
45 developer has come across this problem before, and many have tried
46 (albeit unsuccessfully) to solve this problem. We will analyze existing
47 libraries to demonstrate how they are ineffective and, of course,
49 standards-compliance.</p>
51 <p>I will take no quarter and pull no punches: as of the time of writing,
53 for richly formatted documents. But, nonetheless, there is a necessary
54 disclaimer:</p>
57 This comparison document was written by the author of HTML Purifier,
59 mean that it is biased: I have made every attempt to be <strong>factual and
60 fair</strong>, and I hope that you will agree, by the time you finish reading
62 filter out there today.
63 </p>
74 <thead>
75 <tr>
87 </tr>
88 </thead>
90 <tbody>
92 <tr>
104 </tr>
106 <tr>
118 </tr>
120 <tr>
132 </tr>
134 <tr>
146 </tr>
148 <tr>
160 </tr>
162 <tr>
174 </tr>
176 </tbody>
178 </table>
179 </div>
181 <p><a href="#Tidy">HTML Tidy</a> is omitted from this list because it is not an <abbr>HTML</abbr>
182 filter.</p>
188 A clever person solves a problem.
189 A wise person avoids it.
190 </div>
192 </blockquote>
194 <p>Before we jump into the weird and not-so-wonderful world
196 markup libraries. While libraries of this type really shouldn't be
198 they are the number one method of taking user input and processing it into
199 something more than plain old text. These libraries forgo
205 such markup languages (although it should be noted that
206 Wikitext and Markdown can allow
208 The benefits (to those who use it, anyway) are clear: simplicity and
209 security.
210 </p>
213 <thead>
214 <tr>
217 </tr>
218 </thead>
219 <tbody>
220 <tr>
223 </tr>
224 <tr>
227 </tr>
228 <tr>
231 </tr>
232 <tr>
235 </tr>
236 <tr>
238 <td><tt><b>B</b> <i>i</i> <a href="http://www.example.com/">link</a></tt></td>
239 </tr>
240 <tr>
243 </tr>
244 </tbody>
245 </table>
250 There are many variants of Wikitext currently extant.</li>
251 <li>Strictly speaking, the Markdown syntax is not equivalent: bold text
254 however, map those two semantic tags to the associated styling, so
255 many users assume that it really is italics (and use it improperly for,
256 say, book titles.)</li>
257 </ol>
262 source code is often criticized for being difficult to read. For example,
263 compare:</p>
265 <pre>
266 * Item 1
267 * Item 2
268 </pre>
272 <pre>
277 </pre>
279 <p>Which would you prefer to edit? The answer seems obvious, but be careful
280 not to fall into the fallacy of <a
284 editor, which blows earlier choices out of the water in terms
285 of usability.</p>
287 <p>Note that rich text editors and alternate markup syntaxes are not
288 mutually exclusive, but, when push comes to shove, it's easier
290 markup language. And in the cases when it is done, you usually end up with
291 a live preview, not a true rich text editor.</p>
296 editors aren't all that great.</q> There are many good arguments
297 against these editors, and <a
299 people have written essays</a> devoted to
301 In addition to the usual arguments against said editors, the web poses
302 another limitation: no JavaScript means no
303 editor, and no editor means... (gasp) manually typing in code.</p>
305 <p>Even the most dogmatic purist, however, should recognize that for all
307 There are steps you can take to mitigate the associated drawbacks of
308 these editors.</p>
310 <p>It is often asserted that
313 this is the case with any markup language that allows the smallest
316 A good way to reduce this trouble is to simply eliminate the
317 dialogue boxes that allow users to change colors or fonts (which
318 usually have no legitimate use) and adopt a
320 allowing users to select contextually correct formatting styles
321 for segments of text.</p>
322 </blockquote>
324 <p>Simplicity is also a double-edged sword. The moment any remotely
325 complex markup is needed, these lightweight markup languages fail to
326 produce. Sure you can make '''this text bold''' with Wikitext, but that
330 filters in that their whitelist is too restrictive (besides the fact that
331 their table markup is extraordinarily complex).</p>
337 the angled brackets with square brackets and omitting the occasional parameter
338 name? How much more un-original can you get? Somehow, I don't think BBCode
339 was meant to readable. <a
342 <blockquote>
343 BBCode was devised and put to use in order to provide a safer, easier
344 and more limited way of allowing users to format their messages.
346 which could be used to break/imitate parts of the layout, or run
347 JavaScript. Some implementations of BBCode have suffered problems related
349 security that was intended to be given by BBCode.
350 </blockquote>
354 <blockquote>
355 BBCode came to life when developers where too
357 and decided to invent their own markup language. As with all products of
358 laziness, the result is completely inconsistent, unstandardized, and
359 widely adopted.
360 </blockquote>
362 <p>Well, developers, the whole point of HTML Purifier is that I do the
363 work so you can just execute the ridiculously simple
369 <p>These alternative markup languages have their shiny points, and HTML
370 Purifier is not meant to replace them. However, a major reason for
372 using these languages?</p>
376 <p>Dave Raggett's
380 The premise is simple, the execution effective. Tidy is, in short, a great
383 <p>It is not, however, a filter. I am often surprised when people ask
394 </blockquote>
403 <p>This is not to say that I haven't seen Tidy be used in this sort of
405 output before shuttling it off to the browser. The developers, nevertheless,
406 agree that this is only a band-aid solution, and that the real way
407 to fix it is to fix the parser. Tidy's great, but in terms of security,
408 it's not suitable for untrusted sources.</p>
412 <p>I've ordered my analyses according to how bad a library is. The worst
413 is first, and then we move up the spectrum. I will point out the most
414 flagrant problems with the libraries, but note that I will omit more
417 points through. The ideal solution, however, must do all these things.</p>
419 <p>Note that besides striptags,
421 attacks. None of them (save Safe HTML Checker) fare very well
422 in the standards-compliance department though.</p>
432 </table>
436 the classic solution for attempting to clean up
439 The fact that it doesn't validate attributes at all means that anyone can
442 <p>While
443 this can be bandaided with a series of regular expressions that strip out
445 quirky browser behavior), striptags() is fundamentally flawed and should not be
446 used.
447 </p>
451 <p>Though its title may not imply it,
453 is a souped up version of striptags() with the ability to inspect
454 attributes. (Don't mind the hastily tacked on query escaping function).</p>
467 </table>
469 <p>PHP Input Filter implements an
471 performs very basic checks on whether or not tags and attributes have
472 been defined in the whitelist as well as some
474 the user to define what they'll permit.</p>
476 <p>With absolutely no checking of well-formedness, it is trivially easy
477 to trick the filter into leaving unclosed tags lying around. While to some
479 basic sanity checks like this must be implemented, otherwise a user
480 can mangle a website's layout.</p>
482 <p>More troubles: Woe to
485 layout not to be badly mutilated. To top things off,
486 the filter doesn't even preserve data properly: attributes have all
487 spaces stripped out of them. Stay away, stay away!</p>
493 It should be noted that this is the same library as
495 branding (and a different version number).</p>
508 </table>
510 <p>HTML_Safe's mechanism of action involves parsing
513 validation and filtering as the handlers are called. HTML_Safe does a lot
516 is fundamentally flawed: blacklists.</p>
518 <p>This library maintains arrays of dangerous tags, attributes and
521 intelligently disabled by default in favor of a protocol whitelist.)
522 What this means is that HTML_Safe has no qualms of accepting input
524 the tags in those arrays. Scratch standards-compliance (and that was
525 without even considering proper nesting).</p>
527 <p>For now, HTML_Safe might be safe from
529 In the future, however, one of the infinitely many tags that HTML_Safe lets
530 through might just possibly be given special functionality by browser vendors.
532 solution puts you at a perpetual arms race against crackers who are constantly
533 discovering new and inventive ways to abuse tags and attributes that you
534 didn't blacklist.</p>
554 </table>
556 <p>To be truthful, I didn't do as comprehensive a code survey for kses
557 as I did for some of the other libraries. Out of
558 all the classes I've reviewed so far, kses was definitely the hardest to
559 understand.</p>
561 <p>kses's modus operandi is splitting up html with a monster regexp
563 suffers from the same problems as Input Filter: no well-formedness
564 checks leading to rampant runaway tags (and no standards-compliance).
565 WordPress, the primary user of kses today, had to implement their
566 own custom tag-balancing code to fix this problem: don't use this
567 library without some equivalent!</p>
569 <p>Its whitelist syntax, however, is the most complex of all these libraries,
570 so I'm going to take some time to argue why this particular implementation
571 is bad. The author of this library was thoughtful enough to provide some
572 basic constraint checks on attributes like maxlen and maxval. Now, barring
573 the fact that there simply aren't enough checks, and the fact that they are
574 all lumped together in one function, we now must wonder whether or not
575 the user will go through the trouble of specifying the maximum length
576 of a title attribute.</p>
578 <p>I have my opinions about inherent human laziness, but perhaps WordPress's
579 default filterset is the most telling example:</p>
581 <pre>
582 $allowedposttags = array (
583 /* formatted and trimmed */
584 'hr' => array (
585 'align' => array (),
586 'noshade' => array (),
587 'size' => array (),
588 'width' => array ()
589 )
590 );
591 </pre>
593 <p>Hmm... do I see a blatant lack of attribute constraints? Conclusion:
594 if the user can get away with not doing work, they will! The biggest
596 the whitelist. The whitelist is just as important as the code that uses
601 <p>
603 HTML Checker</a> is (to my knowledge) the first attempt to make a filter
606 search result must have done something right.</p>
619 </table>
621 <p>Indeed, it is quite a well-written piece of code. It demonstrates
622 knowledge of inline versus block elements, thus almost nearly getting
623 nesting correct (the only exception is an unimplemented omitted SGML
626 <p>Unfortunately, part of the reason why it works so well is that it's
627 extremely restrictive. No styling, no tables, very few attributes.
628 Perfectly appropriate for blog comments, but then again, there's always
629 BBCode. This probably means that Safe HTML Checker has a different
630 goal than HTML Purifier.</p>
633 is also quite strict. Accidentally missed a < sign? The parser will
634 complain with the cryptic message:
636 is not well-formed</q>.
637 The solution is not as simple as just switching to a more permissive
638 parser: Safe HTML Checker relies on the fact that the parser will have
639 matched up the tags for them.</p>
654 </table>
665 </table>
667 <p>This is not to say that HTML Purifier doesn't have problems of its own.
668 It's a fairly nascent library (that doesn't mean its buggy though), it's big
669 (while the others usually fit in one file, this one requires a huge
671 features.</a> But even in its current state,
672 HTML Purifier is far better than the other libraries.</p>
676 </div>
677 </body>
678 </html>