Comment for bug that needs to get fixed.
[htmlpurifier.git] / docs / enduser-youtube.html
blob87a36b9aaa7910522fa222121d8b9fab265b4b57
1 <?xml version="1.0" encoding="UTF-8"?>
2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
3 "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
4 <html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en"><head>
5 <meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
6 <meta name="description" content="Explains how to safely allow the embedding of flash from trusted sites in HTML Purifier." />
7 <link rel="stylesheet" type="text/css" href="./style.css" />
9 <title>Embedding YouTube Videos - HTML Purifier</title>
11 </head><body>
13 <h1 class="subtitled">Embedding YouTube Videos</h1>
14 <div class="subtitle">...as well as other dangerous active content</div>
16 <div id="filing">Filed under End-User</div>
17 <div id="index">Return to the <a href="index.html">index</a>.</div>
18 <div id="home"><a href="http://htmlpurifier.org/">HTML Purifier</a> End-User Documentation</div>
20 <p>Clients like their YouTube videos. It gives them a warm fuzzy feeling when
21 they see a neat little embedded video player on their websites that can play
22 the latest clips from their documentary &quot;Fido and the Bones of Spring&quot;.
23 All joking aside, the ability to embed YouTube videos or other active
24 content in their pages is something that a lot of people like.</p>
26 <p>This is a <em>bad</em> idea. The moment you embed anything untrusted,
27 you will definitely be slammed by a manner of nasties that can be
28 embedded in things from your run of the mill Flash movie to
29 <a href="http://blog.spywareguide.com/2006/12/myspace_phish_attack_leads_use.html">Quicktime movies</a>.
30 Even <code>img</code> tags, which HTML Purifier allows by default, can be
31 dangerous. Be distrustful of anything that tells a browser to load content
32 from another website automatically.</p>
34 <p>Luckily for us, however, whitelisting saves the day. Sure, letting users
35 include any old random flash file could be dangerous, but if it's
36 from a specific website, it probably is okay. If no amount of pleading will
37 convince the people upstairs that they should just settle with just linking
38 to their movies, you may find this technique very useful.</p>
40 <h2>Looking in</h2>
42 <p>Below is custom code that allows users to embed
43 YouTube videos. This is not favoritism: this trick can easily be adapted for
44 other forms of embeddable content.</p>
46 <p>Usually, websites like YouTube give us boilerplate code that you can insert
47 into your documents. YouTube's code goes like this:</p>
49 <pre>
50 &lt;object width=&quot;425&quot; height=&quot;350&quot;&gt;
51 &lt;param name=&quot;movie&quot; value=&quot;http://www.youtube.com/v/AyPzM5WK8ys&quot; /&gt;
52 &lt;param name=&quot;wmode&quot; value=&quot;transparent&quot; /&gt;
53 &lt;embed src=&quot;http://www.youtube.com/v/AyPzM5WK8ys&quot;
54 type=&quot;application/x-shockwave-flash&quot;
55 wmode=&quot;transparent&quot; width=&quot;425&quot; height=&quot;350&quot; /&gt;
56 &lt;/object&gt;
57 </pre>
59 <p>There are two things to note about this code:</p>
61 <ol>
62 <li><code>&lt;embed&gt;</code> is not recognized by W3C, so if you want
63 standards-compliant code, you'll have to get rid of it.</li>
64 <li>The code is exactly the same for all instances, except for the
65 identifier <tt>AyPzM5WK8ys</tt> which tells us which movie file
66 to retrieve.</li>
67 </ol>
69 <p>What point 2 means is that if we have code like <code>&lt;span
70 class=&quot;youtube-embed&quot;&gt;AyPzM5WK8ys&lt;/span&gt;</code> your
71 application can reconstruct the full object from this small snippet that
72 passes through HTML Purifier <em>unharmed</em>.
73 <a href="http://repo.or.cz/w/htmlpurifier.git?a=blob;hb=HEAD;f=library/HTMLPurifier/Filter/YouTube.php">Show me the code!</a></p>
75 <p>And the corresponding usage:</p>
77 <pre>&lt;?php
78 $config-&gt;set('Filter.YouTube', true);
79 ?&gt;</pre>
81 <p>There is a bit going in the two code snippets, so let's explain.</p>
83 <ol>
84 <li>This is a Filter object, which intercepts the HTML that is
85 coming into and out of the purifier. You can add as many
86 filter objects as you like. <code>preFilter()</code>
87 processes the code before it gets purified, and <code>postFilter()</code>
88 processes the code afterwards. So, we'll use <code>preFilter()</code> to
89 replace the object tag with a <code>span</code>, and <code>postFilter()</code>
90 to restore it.</li>
91 <li>The first preg_replace call replaces any YouTube code users may have
92 embedded into the benign span tag. Span is used because it is inline,
93 and objects are inline too. We are very careful to be extremely
94 restrictive on what goes inside the span tag, as if an errant code
95 gets in there it could get messy.</li>
96 <li>The HTML is then purified as usual.</li>
97 <li>Then, another preg_replace replaces the span tag with a fully fledged
98 object. Note that the embed is removed, and, in its place, a data
99 attribute was added to the object. This makes the tag standards
100 compliant! It also breaks Internet Explorer, so we add in a bit of
101 conditional comments with the old embed code to make it work again.
102 It's all quite convoluted but works.</li>
103 </ol>
105 <h2>Warning</h2>
107 <p>There are a number of possible problems with the code above, depending
108 on how you look at it.</p>
110 <h3>Cannot change width and height</h3>
112 <p>The width and height of the final YouTube movie cannot be adjusted. This
113 is because I am lazy. If you really insist on letting users change the size
114 of the movie, what you need to do is package up the attributes inside the
115 span tag (along with the movie ID). It gets complicated though: a malicious
116 user can specify an outrageously large height and width and attempt to crash
117 the user's operating system/browser. You need to either cap it by limiting
118 the amount of digits allowed in the regex or using a callback to check the
119 number.</p>
121 <h3>Trusts media's host's security</h3>
123 <p>By allowing this code onto our website, we are trusting that YouTube has
124 tech-savvy enough people not to allow their users to inject malicious
125 code into the Flash files. An exploit on YouTube means an exploit on your
126 site. Even though YouTube is run by the reputable Google, it
127 <a href="http://ha.ckers.org/blog/20061213/google-xss-vuln/">doesn't</a>
128 mean they are
129 <a href="http://ha.ckers.org/blog/20061208/xss-in-googles-orkut/">invulnerable.</a>
130 You're putting a certain measure of the job on an external provider (just as
131 you have by entrusting your user input to HTML Purifier), and
132 it is important that you are cognizant of the risk.</p>
134 <h3>Poorly written adaptations compromise security</h3>
136 <p>This should go without saying, but if you're going to adapt this code
137 for Google Video or the like, make sure you do it <em>right</em>. It's
138 extremely easy to allow a character too many in <code>postFilter()</code> and
139 suddenly you're introducing XSS into HTML Purifier's XSS free output. HTML
140 Purifier may be well written, but it cannot guard against vulnerabilities
141 introduced after it has finished.</p>
143 <h2>Help out!</h2>
145 <p>If you write a filter for your favorite video destination (or anything
146 like that, for that matter), send it over and it might get included
147 with the core!</p>
149 </body>
150 </html>
152 <!-- vim: et sw=4 sts=4