Category:Encoding

Description
Encoding, closely related to Escaping is a powerful mechanism to help protect against many types of attack, especially injection attacks and Cross-site Scripting (XSS). Essentially, encoding involves translating special characters into some equivalent that is no longer significant in the target interpreter. So, for example, using HTML entity encoding before sending untrusted data into a browser will protect against many forms of Cross-site Scripting (XSS).

Considerations:


 * What interpreter?
 * To encode properly, you need to know what interpreters the data might end up in. For example, if the data is going into a SQL interpreter, you should consider encoding based on syntax of the SQL engine you are using.


 * What characters? Complete?
 * You want to make sure that you encode all the characters that might cause a problem, so the best approach is to use a positive encoding scheme, where all characters except a minimal known good set are encoded.


 * What encoding scheme?
 * There are dozens of ways to encode characters and many interpreters allow multiple forms of a single significant character. For a browser, HTML entity encoding is a good way to prevent script injection, but URL encoding or Unicode encoding (%xx) will not prevent scripts from running. Be sure to use the appropriate encoding scheme for the target interpreter.


 * Double encoding and decoding?
 * Be careful not to double encode your data. In some cases, doubly encoding data can inadvertently introduce special characters in the final output. Also, be aware that some processors may automatically undo your encoding. There is some evidence that XML processors are decoding HTML entity encoding, thus reintroducing potential XSS problems.

See the Category:OWASP Encoding Project for more information.

Examples
ASP.NET Example

ASP.NET pages have a "ValidateRequest" property which is set to true by default. This prevents XSS-type attacks in submitted form fields - ASP.NET will throw an exception if it detects script-type "unsafe" content in a request. Unfortunately this mechanism will be triggered in response to certain characters you may actually want to receive in a request, such as > and <.

To get around this, disable page validation for the page and use Server.HTMLEncode (and its URL equivalent URLEncode for GET operations) to encode user input: -

PHP Example

PHP offers two identical built in functions to perform entity encoding: htmlentities and htmlspecialchars. By default both functions require only the string to encode as a parameter, however optionally additional parameters can be used to govern how apostrophes and double quotes are encoded (if the parameter is not provided the function will convert double quotes and leave apostrophes alone), which character set to output the results as (ISO ISO-8859-1 is the default if not provided), and whether or not to double encode the results. If the double encode parameter is not provided, or set to true, existing html entities will be encoded a second time.

Java Example

Unlike ASP.Net and PHP, Java does not have built in entity encoding functionality. There is, however, an example function to perform entity encoding provided by OWASP: How to perform HTML entity encoding in Java. As a word of warning for those writing internationalized websites, the implementation is very "basic ASCII" centric in terms of encoding characters, and will encode every character not in the a-z,A-Z,0-9 range even if the page character set would normally support the characters un-encoded without issue. This can dramatically increase the size of page being served. An alternative to avoid this issue is to use the encoding function provided by the widely distributed Apache Commons Lang library. To use this library first import org.apache.commons.lang.StringEscapeUtils. Additionally the input must be unescaped prior to encoding, hence the nested function call in the sample code below.

Related Countermeasures

 * Input Validation