| blob | 7074b1d54cff3bbe80ec6333dbecb40ce29297a7 |
11 <head>
15 <meta name="keywords" content="HTMLPurifier, HTML Purifier, HTML, filter, filtering, HTML_Safe, PEAR, comparison, kses, striptags, SafeHTMLChecker" />
16 </head>
17 <body>
24 <p>With the advent of
26 gone from passive consumer to active producer of content on the World Wide
30 put the user in control.</p>
32 <p>Give the user too much control, however, and you set yourself up
33 for <a href="http://en.wikipedia.org/wiki/Cross-site_scripting"><abbr>XSS</abbr></a> attacks. For this reason,
35 has proven to be both a blessing and a curse, and the software that processes
36 it must strike a fine balance between security and usability. How do
37 we prevent users from injecting JavaScript or inserting malformed
44 developer has come across this problem before, and many have tried
45 (albeit unsuccessfully) to solve this problem. We will analyze existing
46 libraries to demonstrate how they are ineffective and, of course,
48 standards-compliance.</p>
50 <p>I will take no quarter and pull no punches: as of the time of writing,
52 for richly formatted documents. But, nonetheless, there is a necessary
53 disclaimer:</p>
56 This comparison document was written by the author of HTML Purifier,
58 mean that it is biased: I have made every attempt to be <strong>factual and
59 fair</strong>, and I hope that you will agree, by the time you finish reading
61 filter out there today.
62 </p>
73 <thead>
74 <tr>
86 </tr>
87 </thead>
89 <tbody>
91 <tr>
103 </tr>
105 <tr>
117 </tr>
119 <tr>
131 </tr>
133 <tr>
145 </tr>
147 <tr>
159 </tr>
161 <tr>
173 </tr>
175 </tbody>
177 </table>
178 </div>
180 <p><a href="#Tidy">HTML Tidy</a> is omitted from this list because it is not an <abbr>HTML</abbr>
181 filter.</p>
187 A clever person solves a problem.
188 A wise person avoids it.
189 </div>
191 </blockquote>
193 <p>Before we jump into the weird and not-so-wonderful world
195 markup libraries. While libraries of this type really shouldn't be
197 they are the number one method of taking user input and processing it into
198 something more than plain old text. These libraries forgo
204 such markup languages (although it should be noted that
205 Wikitext and Markdown can allow
207 The benefits (to those who use it, anyway) are clear: simplicity and
208 security.
209 </p>
212 <thead>
213 <tr>
216 </tr>
217 </thead>
218 <tbody>
219 <tr>
222 </tr>
223 <tr>
226 </tr>
227 <tr>
230 </tr>
231 <tr>
234 </tr>
235 <tr>
237 <td><tt><b>B</b> <i>i</i> <a href="http://www.example.com/">link</a></tt></td>
238 </tr>
239 <tr>
242 </tr>
243 </tbody>
244 </table>
249 There are many variants of Wikitext currently extant.</li>
250 <li>Strictly speaking, the Markdown syntax is not equivalent: bold text
253 however, map those two semantic tags to the associated styling, so
254 many users assume that it really is italics (and use it improperly for,
255 say, book titles.)</li>
256 </ol>
261 source code is often criticized for being difficult to read. For example,
262 compare:</p>
264 <pre>
265 * Item 1
266 * Item 2
267 </pre>
271 <pre>
276 </pre>
278 <p>Which would you prefer to edit? The answer seems obvious, but be careful
279 not to fall into the fallacy of <a
283 editor, which blows earlier choices out of the water in terms
284 of usability.</p>
286 <p>Note that rich text editors and alternate markup syntaxes are not
287 mutually exclusive, but, when push comes to shove, it's easier
289 markup language. And in the cases when it is done, you usually end up with
290 a live preview, not a true rich text editor.</p>
295 editors aren't all that great.</q> There are many good arguments
296 against these editors, and <a
298 people have written essays</a> devoted to
300 In addition to the usual arguments against said editors, the web poses
301 another limitation: no JavaScript means no
302 editor, and no editor means... (gasp) manually typing in code.</p>
304 <p>Even the most dogmatic purist, however, should recognize that for all
306 There are steps you can take to mitigate the associated drawbacks of
307 these editors.</p>
309 <p>It is often asserted that
312 this is the case with any markup language that allows the smallest
315 A good way to reduce this trouble is to simply eliminate the
316 dialogue boxes that allow users to change colors or fonts (which
317 usually have no legitimate use) and adopt a
319 allowing users to select contextually correct formatting styles
320 for segments of text.</p>
321 </blockquote>
323 <p>Simplicity is also a double-edged sword. The moment any remotely
324 complex markup is needed, these lightweight markup languages fail to
325 produce. Sure you can make '''this text bold''' with Wikitext, but that
329 filters in that their whitelist is too restrictive (besides the fact that
330 their table markup is extraordinarily complex).</p>
336 the angled brackets with square brackets and omitting the occasional parameter
337 name? How much more un-original can you get? Somehow, I don't think BBCode
338 was meant to readable. <a
341 <blockquote>
342 BBCode was devised and put to use in order to provide a safer, easier
343 and more limited way of allowing users to format their messages.
345 which could be used to break/imitate parts of the layout, or run
346 JavaScript. Some implementations of BBCode have suffered problems related
348 security that was intended to be given by BBCode.
349 </blockquote>
353 <blockquote>
354 BBCode came to life when developers where too
356 and decided to invent their own markup language. As with all products of
357 laziness, the result is completely inconsistent, unstandardized, and
358 widely adopted.
359 </blockquote>
361 <p>Well, developers, the whole point of HTML Purifier is that I do the
362 work so you can just execute the ridiculously simple
368 <p>These alternative markup languages have their shiny points, and HTML
369 Purifier is not meant to replace them. However, a major reason for
371 using these languages?</p>
375 <p>Dave Raggett's
379 The premise is simple, the execution effective. Tidy is, in short, a great
382 <p>It is not, however, a filter. I am often surprised when people ask
393 </blockquote>
402 <p>This is not to say that I haven't seen Tidy be used in this sort of
404 output before shuttling it off to the browser. The developers, nevertheless,
405 agree that this is only a band-aid solution, and that the real way
406 to fix it is to fix the parser. Tidy's great, but in terms of security,
407 it's not suitable for untrusted sources.</p>
411 <p>I've ordered my analyses according to how bad a library is. The worst
412 is first, and then we move up the spectrum. I will point out the most
413 flagrant problems with the libraries, but note that I will omit more
416 points through. The ideal solution, however, must do all these things.</p>
418 <p>Note that besides striptags,
420 attacks. None of them (save Safe HTML Checker) fare very well
421 in the standards-compliance department though.</p>
431 </table>
435 the classic solution for attempting to clean up
438 The fact that it doesn't validate attributes at all means that anyone can
441 <p>While
442 this can be bandaided with a series of regular expressions that strip out
444 quirky browser behavior), striptags() is fundamentally flawed and should not be
445 used.
446 </p>
450 <p>Though its title may not imply it,
452 is a souped up version of striptags() with the ability to inspect
453 attributes. (Don't mind the hastily tacked on query escaping function).</p>
466 </table>
468 <p>PHP Input Filter implements an
470 performs very basic checks on whether or not tags and attributes have
471 been defined in the whitelist as well as some
473 the user to define what they'll permit.</p>
475 <p>With absolutely no checking of well-formedness, it is trivially easy
476 to trick the filter into leaving unclosed tags lying around. While to some
478 basic sanity checks like this must be implemented, otherwise a user
479 can mangle a website's layout.</p>
481 <p>More troubles: Woe to
484 layout not to be badly mutilated. To top things off,
485 the filter doesn't even preserve data properly: attributes have all
486 spaces stripped out of them. Stay away, stay away!</p>
492 It should be noted that this is the same library as
494 branding (and a different version number).</p>
507 </table>
509 <p>HTML_Safe's mechanism of action involves parsing
512 validation and filtering as the handlers are called. HTML_Safe does a lot
515 is fundamentally flawed: blacklists.</p>
517 <p>This library maintains arrays of dangerous tags, attributes and
520 intelligently disabled by default in favor of a protocol whitelist.)
521 What this means is that HTML_Safe has no qualms of accepting input
523 the tags in those arrays. Scratch standards-compliance (and that was
524 without even considering proper nesting).</p>
526 <p>For now, HTML_Safe might be safe from
528 In the future, however, one of the infinitely many tags that HTML_Safe lets
529 through might just possibly be given special functionality by browser vendors.
531 solution puts you at a perpetual arms race against crackers who are constantly
532 discovering new and inventive ways to abuse tags and attributes that you
533 didn't blacklist.</p>
553 </table>
555 <p>To be truthful, I didn't do as comprehensive a code survey for kses
556 as I did for some of the other libraries. Out of
557 all the classes I've reviewed so far, kses was definitely the hardest to
558 understand.</p>
560 <p>kses's modus operandi is splitting up html with a monster regexp
562 suffers from the same problems as Input Filter: no well-formedness
563 checks leading to rampant runaway tags (and no standards-compliance).
564 WordPress, the primary user of kses today, had to implement their
565 own custom tag-balancing code to fix this problem: don't use this
566 library without some equivalent!</p>
568 <p>Its whitelist syntax, however, is the most complex of all these libraries,
569 so I'm going to take some time to argue why this particular implementation
570 is bad. The author of this library was thoughtful enough to provide some
571 basic constraint checks on attributes like maxlen and maxval. Now, barring
572 the fact that there simply aren't enough checks, and the fact that they are
573 all lumped together in one function, we now must wonder whether or not
574 the user will go through the trouble of specifying the maximum length
575 of a title attribute.</p>
577 <p>I have my opinions about inherent human laziness, but perhaps WordPress's
578 default filterset is the most telling example:</p>
580 <pre>
581 $allowedposttags = array (
582 /* formatted and trimmed */
583 'hr' => array (
584 'align' => array (),
585 'noshade' => array (),
586 'size' => array (),
587 'width' => array ()
588 )
589 );
590 </pre>
592 <p>Hmm... do I see a blatant lack of attribute constraints? Conclusion:
593 if the user can get away with not doing work, they will! The biggest
595 the whitelist. The whitelist is just as important as the code that uses
600 <p>
602 HTML Checker</a> is (to my knowledge) the first attempt to make a filter
605 search result must have done something right.</p>
618 </table>
620 <p>Indeed, it is quite a well-written piece of code. It demonstrates
621 knowledge of inline versus block elements, thus almost nearly getting
622 nesting correct (the only exception is an unimplemented omitted SGML
625 <p>Unfortunately, part of the reason why it works so well is that it's
626 extremely restrictive. No styling, no tables, very few attributes.
627 Perfectly appropriate for blog comments, but then again, there's always
628 BBCode. This probably means that Safe HTML Checker has a different
629 goal than HTML Purifier.</p>
632 is also quite strict. Accidentally missed a < sign? The parser will
633 complain with the cryptic message:
635 is not well-formed</q>.
636 The solution is not as simple as just switching to a more permissive
637 parser: Safe HTML Checker relies on the fact that the parser will have
638 matched up the tags for them.</p>
653 </table>
664 </table>
666 <p>This is not to say that HTML Purifier doesn't have problems of its own.
667 It's a fairly nascent library (that doesn't mean its buggy though), it's big
668 (while the others usually fit in one file, this one requires a huge
670 features.</a> But even in its current state,
671 HTML Purifier is far better than the other libraries.</p>
675 </div>
676 </body>
677 </html>