Gitiles
Code Review Sign In
realtimeroboticsgroup.org / RealtimeRoboticsGroup / test / 84c01035ab44cf9cd93b2efb35e0ad7532ab8d38 / . / doc / auto_escape.html
blob: 2b5dd1298bedb661f2ff44babd7dad56bf0fc813 [file] [log] [blame]
Brian Silverman70325d62015-09-20 17:00:43 -0400[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
2<html>
3<head>
4 <title>Guide to using Auto Escape</title>
5
6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <link href="designstyle.css" type="text/css" rel="stylesheet">
8 <style type="text/css">
9 ol.bluelist li {
10 color: #3366ff;
11 font-family: sans-serif;
12 }
13 ol.bluelist li p {
14 color: #000;
15 font-family: "Times Roman", times, serif;
16 }
17 ul.blacklist li {
18 color: #000;
19 font-family: "Times Roman", times, serif;
20 }
21 </style>
22</head>
23<body>
24
25<h1><a name="Guide_Auto_Escape"></a>Guide to using Auto Escape</h1>
26
27<h2>Introduction</h2>
28
29<p>Auto Escape is an optional mode of execution in the Template System
30developed to provide a better defense against cross-site scripting (XSS)
31in web applications. In this mode, the Template System assumes
32responsibility for applying the proper escaping modifiers for each
33variable in your template (and templates it may include). As such, the
34template developer no longer needs to manually apply escaping modifiers
35to each variable, a process which is repetitive and error-prone
36particularly in larger and more complex applications.</p>
37
38<p>In order for the Template System to properly determine the escaping
39modifiers to apply to variables, it activates a built-in HTML-aware
40parser during template initialization. The parser scans the template
41to determines the context in which each variable is being emitted.
42The scanning is performed at initialization-time only hence does not
43contribute to processing costs at template expansion time.</p>
44
45<p>Refer to <a href="guide.html#security">Security Considerations</a>
46for additional security information on escaping directives and their
47use according to the context in which content is being expanded.</p>
48
49<h2>Current Status</h2>
50
51<p>Auto Escape currently supports well <code>HTML</code> and
52<code>Javascript</code> templates. It also provides very basic support
53for <code>CSS</code>, <code>JSON</code> and <code>XML</code> templates
54where it applies the corresponding escaping modifier but without
55utilizing a parser. We may in the future provide parsers specific
56to these contexts and modifiers for them that are finer-grained.</p>
57
58<h2>Overview</h2>
59
60This section provides more background on Auto Escape.
61Refer to <a href="guide.html#auto_escape">How to Auto Escape</a>
62for information on how to leverage this mode in your application.
63
64<h3>Using template modifiers</h3>
65
66<p>In the simplest case, where you only want to use template-modifiers
67to achieve html-safety (that is, you only use
68<code>html_escape</code>, <code>url_query_escape</code>, and so
69forth), you can stop specifying template-modifiers in your template at
70all. The Auto Escape mode will generate the modifiers on its own and
71perform the correct escaping.</p>
72
73<p>If you need other types of modifiers, including your own custom
74modifiers, you can still specify them in your template and they will
75continue to work just the same. The Auto Escape mode will simply apply
76the appropriate escaping modifiers after your own.</p>
77
78<p>Other reasons for manually specifying variable modifiers include:</p>
79<ol>
80 <li> To indicate a variable is safe. You may add the
81 <code>:none</code> modifier as the last modifier for a given
82 variable, in which case the Auto Escape mode will leave it
83 untouched and not apply escaping directives to it. Use only
84 when you are sure the variable is trusted to be safe from XSS.
85 </li>
86
87 <li> To select a different escaping modifier from a list of
88 equivalent modifiers. Certain modifiers are equivalent from
89 an escaping point of view in that they would all ensure given
90 content is safe when escaped in a specific context. The
91 Auto Escape mode choses the most common escaping modifier for
92 that use but allows you to indicate a different choice as shown
93 below:
94
95 <table border="1" cellpadding="3" summary="Alternatives to modifiers">
96 <tr><th>TemplateContext</th><th>Primary Modifier</th>
97 <th>Accepted alternatives</th></tr>
98
99 <tr>
100 <td><code>TC_HTML</code></td>
101 <td><code>:html_escape</code></td>
102 <td><ul>
103 <li><code>:pre_escape</code></li>
104 <li><code>:html_escape_with_arg=snippet</code></li>
105 <li><code>:html_escape_with_arg=attribute</code></li>
106 <li><code>:html_escape_with_arg=url</code></li>
107 <li><code>:url_query_escape</code></li>
108 <li><code>:url_escape_with_arg=html</code></li>
109 <li><code>:img_src_url_escape_with_arg=html</code></li>
110 </ul></td>
111 </tr>
112
113 <tr>
114 <td><code>TC_HTML</code> and <code>TC_CSS</code></td>
115 <td><code>:cleanse_css</code></td>
116 <td><ul>
117 <li><code>:url_escape_with_arg=css</code></li>
118 <li><code>:img_src_url_escape_with_arg=css</code></li>
119 </ul></td>
120 </tr>
121
122 <tr>
123 <td><code>TC_XML</code></td>
124 <td><code>:xml_escape</code></td>
125 <td><ul>
126 <li><code>:html_escape</code></li>
127 <li><code>:html_escape_with_arg=attribute</code></li>
128 </ul></td>
129 </tr>
130
131 <tr>
132 <td><code>TC_JS</code></td>
133 <td><code>:javascript_escape</code></td>
134 <td><ul>
135 <li><code>:url_escape_with_arg=javascript</code></li>
136 <li><code>:img_src_url_escape_with_arg=javascript</code></li>
137 </ul></td>
138 </tr>
139
140 <tr>
141 <td><code>TC_JSON</code></td>
142 <td><code>:javascript_escape</code></td>
143 <td><ul>
144 <li><code>:json_escape</code></li>
145 </ul></td>
146 </tr>
147
148 </table></li>
149
150 <li> To add additional escaping modifiers. The Template System will
151 never remove escaping directives you explicitly specify. </li>
152</ol>
153
154<h3><A name="EscapeContext">Escaping Contexts</A>
155</h3>
156
157
158<p>The table belows indicates which modifier Auto-Escape selects to
159escape a given variable, based on the context of the variable being
160inserted and, possibly, on more specific information such as whether
161the variable is inside quotation marks. The table only applies for
162the well-supported TemplateContext values (currently
163<code>TC_HTML</code> and <code>TC_JS</code>).</p>
164
165<p>In a few cases, we do not have
166a modifier that is both safe against XSS and respecting of
167the semantics of the variable therefore we fail the template
168initialization. An error is logged indicating the cause of
169the failure.</p>
170
171<table border="1" cellpadding="3" summary="Contexts for using modifiers">
172<tr bgcolor="#4169E1">
173 <th>Context</th><th>HTML Quoted?</th>
174 <th>Examples</th><th>Action Performed</th>
175</tr>
176
177<tr bgcolor="#F0F8FF">
178 <td>Regular HTML Body and HTML comments</td><td>Any</td>
179 <td><pre>&lt;p&gt;Hello {{USER}}&lt;/p&gt;
180 </pre></td>
181 <td>Escape <code>USER</code> using
182 <code>:html_escape</code></td>
183</tr>
184
185<tr bgcolor="#FAEBD7">
186 <td>In URL attribute: Starts at pos 0</td><td>Yes</td>
187 <td><pre>&lt;a href="{{URL}}"&gt;;</pre></td>
188 <td>Escape <code>URL</code> using
189 <code>:url_escape_with_arg=html</code></td>
190</tr>
191
192<tr bgcolor="#FAEBD7">
193 <td>In URL attribute: Other</td><td>Yes</td>
194 <td><pre>&lt;a href="/foo?q={{QUERY}}"&gt;</pre></td>
195 <td>Escape <code>QUERY</code> using
196 <code>:html_escape</code></td>
197</tr>
198
199<tr bgcolor="#FAEBD7">
200 <td>In URL attribute: Starting at pos 0</td><td>No</td>
201 <td><pre>&lt;form action={{URL}}&gt;</pre></td>
202 <td><em>Fail template initialization</em></td>
203</tr>
204
205<tr bgcolor="#FAEBD7">
206 <td>In URL attribute: Other</td><td>No</td>
207 <td><pre>&lt;a href=/foo?q={{QUERY}}&gt;My Link&lt;/a&gt;</pre></td>
208 <td>Escape <code>QUERY</code> using
209 <code>:url_query_escape</code></td>
210</tr>
211
212<tr bgcolor="#F0F8FF">
213 <td>In STYLE attribute</td><td>Yes</td>
214 <td><pre>&lt;div style="color:{{COLOR}};"&gt;</pre></td>
215 <td>Escape <code>COLOR</code> using
216 <code>:cleanse_css</code></td>
217</tr>
218
219<tr bgcolor="#F0F8FF">
220 <td>In STYLE attribute</td><td>No</td>
221 <td><pre>&lt;div style=color:{{COLOR}};&gt;</pre></td>
222 <td><em>Fail template initialization</em></td>
223</tr>
224
225<tr bgcolor="#FAEBD7">
226 <td>In Javascript attribute: String literal</td><td>Yes</td>
227 <td><pre>&lt;a href="url" onclick="doFoo('{{ARG}}');"&gt;</pre></td>
228 <td>Escape <code>ARG</code> using
229 <code>:javascript_escape</code></td>
230</tr>
231
232<tr bgcolor="#FAEBD7">
233 <td>In Javascript attribute: Non-string literal</td><td>Yes</td>
234 <td><pre>&lt;a href="url" onclick="doFoo({{ARG}});"&gt;</pre></td>
235 <td>Escape <code>ARG</code> using
236 <code>:javascript_escape_with_arg=number</code></td>
237</tr>
238
239<tr bgcolor="#FAEBD7">
240 <td>In Javascript attribute: Any</td><td>No</td>
241 <td><pre>&lt;a href="url" onclick=doFoo('{{ARG}}');&gt;</pre></td>
242 <td><em>Fail template initialization.</em></td>
243</tr>
244
245<tr bgcolor="#F0F8FF">
246 <td>In all other attributes</td><td>Yes</td>
247 <td><pre>&lt;b class="{{CLASS}}"&gt;</pre></td>
248 <td>Escape <code>CLASS</code> using
249 <code>:html_escape</code></td>
250</tr>
251
252<tr bgcolor="#F0F8FF">
253 <td>In all other attributes</td><td>No</td>
254 <td><pre>&lt;table border={{BORDER}}&gt;</pre></td>
255 <td>Escape <code>BORDER</code> using
256 <code>:html_escape_with_arg=attribute</code></td>
257</tr>
258
259<tr bgcolor="#FAEBD7">
260 <td>In Javascript code: In a string literal</td><td>Any</td>
261 <td><pre>&lt;script&gt;var a = '{{VALUE}}';&lt;/script&gt;</pre></td>
262 <td>Escape <code>VALUE</code> using
263 <code>:javascript_escape</code></td>
264</tr>
265
266<tr bgcolor="#FAEBD7">
267 <td>In Javascript code: Non-string literal</td><td>Any</td>
268 <td><pre>&lt;script&gt;var a = {{VALUE}};&lt;/script&gt;</pre></td>
269 <td>Escape <code>VALUE</code> using
270 <code>:javascript_escape_with_arg=number</code></td>
271</tr>
272
273<tr bgcolor="#F0F8FF">
274 <td>In a &lt;style&gt; tag</td><td>Any</td>
275 <td><pre>&lt;style&gt;font-size={{FONTSIZE}};&lt;/style&gt;</pre></td>
276 <td>Escape <code>FONTSIZE</code> using
277 <code>:cleanse_css</code></td>
278</tr>
279
280</table>
281
282<h4>Comments:</h4>
283<ul>
284 <li>For values of URL-accepting attributes, we apply a different
285 modifier if the variable is at the beginning of the attribute
286 value or not because in the former case we consider the
287 variable to represent a complete URL and hence also check
288 that its scheme is safe (<code>HTTP</code> or
289 <code>HTTPS</code>).</li>
290 <li>We recommend that you enclose attribute values in quotes
291 in particular:
292 <ol>
293 <li>Under some conditions outlined in the table above, when
294 variables are present in attribute values that are not
295 enclosed in quotes, we may fail template initialization
296 because we do not have escaping modifiers that are safe to
297 use.</li>
298 <li>Also the determination of which escaping modifier
299 to apply for unquoted attribute values specifically
300 is likely to change in the future.</li>
301 </ol>
302</ul>
303
304<p>The Auto Escape mode is designed to be simple to use and to produce
305correct correct escaping with minimal input. It does however come with
306restrictions governing the use of this template system.</p>
307
308<h3><a name="Limitations">Limitations</a></h3>
309
310<ul>
311 <li> <p>Templates may only be included in regular HTML
312 text. Including a template within HTML comments or inside an
313 HTML tag is currently not supported. In the example below,
314 <code>HEADER</code> and <code>FOOTER</code> are properly
315 included but <code>TAG</code> and <code>COMMENT</code> are
316 not.</p>
317 <pre>
318 &lt;html&gt;&lt;head&gt;{{&gt;HEADER}}&lt;/head&gt;&lt;body&gt;
319 &lt;{{&gt;TAG}}&gt;
320 &lt;p&gt;
321 &lt;!-- {{&gt;COMMENT}} --&gt;
322 &lt;p&gt;{{&gt;FOOTER}}
323 &lt;/body&gt;&lt;/html&gt;
324 </pre>
325 <p>If the template system detects that you are including a
326 template in another HTML context, it will log a warning.</p></li>
327
328 <li> <p>Included templates may not cross HTML boundaries. They
329 cannot start with one <code>TemplateContext</code> and end in
330 another. For example, the template below crosses boundaries --
331 it starts in HTML and ends in Javascript -- and is therefore
332 not a valid included template for auto-escaping.</p>
333 <pre>
334 &lt;html&gt;
335 &lt;head&gt;
336 &lt;/head&gt;
337 &lt;body&gt;
338 Hello USER.
339 &lt;script&gt;
340 var a = 'not a nice template';
341 </pre>
342 <p>To rectify this violation, continue the template by
343 completing the javascript code and terminating it with a
344 <code>&lt;/script&gt;</code> tag.</p> </li>
345
346 <li> <p>Error reporting. We currently report errors by logging
347 a warning or an error during initial template
348 initialization and parsing if we suspect something is
349 incorrect. Please check for them, as they indicate an error
350 that may cause auto-escaping to work improperly.</p> </li>
351
352 <li> <p>Beware of double escaping. When you use the Auto Escape mode,
353 you surrender all variable escaping to the template system.
354 If you also perform your own escaping in your source code, you may
355 face situations where content was escaped multiple times.</p>
356 <p>In particular, do not use any of the legacy methods below or
357 others that perform variable escaping.</p>
358 <ul>
359 <li> <code>SetEscapedValue</code> </li>
360 <li> <code>SetEscapedFormatttedValue</code> </li>
361 <li> <code>SetEscapedValueAndShowSection</code> </li>
362 </ul>
363 </li>
364
365 <li> <p>Supported HTML contexts. HTML (<code>TC_HTML</code>) and
366 Javascript (<code>TC_JS</code>) templates are well supported.
367 Other template contexts have only basic support.
368 For these contexts, variables are escaped as follows:</p>
369 <table border="1" cellpadding="3" summary="HTML contexts for modifiers">
370 <tr><th>TemplateContext</th><th>Escaping Applied</th></tr>
371 <tr><td>TC_JSON</td>
372 <td><code>:javascript_escape</code></td></tr>
373 <tr><td>TC_XML</td>
374 <td><code>:xml_escape</code></td></tr>
375 <tr><td>TC_CSS</td>
376 <td><code>:cleanse_css</code></td></tr>
377 </table></li>
378
379 <li> <p>New restrictions on the use of <code>BI_SPACE</code> and
380 <code>BI_NEWLINE</code>. These system variables may not be
381 re-assigned to a semantically different value via
382 <code>dict->SetValue</code> or related methods, and they may
383 not have modifiers attached to them in templates. For
384 instance: it's ok to redefine <code>BI_NEWLINE</code> from
385 <code>\n</code> to <code>\r\n</code>, since html treats them
386 identically. But changing it to <code>hello, world!</code> is
387 not allowed.</p> </li>
388</ul>
389
390<h4> A new filename convention </h4>
391
392<p>The Auto Escape feature codifies simple conventions
393of template filenames: </p>
394<ol>
395 <li> CSS templates optionally include, in their filename, the
396 keywords <code>style</code> or <code>stylesheet</code> or
397 <code>css</code>. </li>
398 <li> Javascript templates optionally include, in their filename, the
399 keywords <code>js</code> or <code>javascript</code>. </li>
400</ol>
401
402<p>If we find one of these keywords in the filename but
403it does not match the template type given, we log a warning.</p>
404
405<hr>
406<ul>
407 <li> <A HREF="guide.html">User's Guide</A> </li>
408 <li> <A HREF="reference.html">Reference Manual</A> </li>
409<!--
410 <li> <A HREF="auto_escape.html">Auto Escape</A> </li>
411-->
412 <li> <A HREF="tips.html">Tips</A> </li>
413 <li> <A HREF="example.html">Example</A> </li>
414</ul>
415
416
417
418<hr>
419<address>
420Jad Boutros<br>
421</address>
422
423</body>
424</html>