Theory of Operation

Zend\Escaper provides methods for escaping output data, dependent on the context in which the data will be used. Each method is based on peer-reviewed rules and is in compliance with the current OWASP recommendations.

The escaping follows a well known and fixed set of encoding rules for each key HTML context, which are defined by OWASP. These rules cannot be impacted or negated by browser quirks or edge-case HTML parsing unless the browser suffers a catastrophic bug in it’s HTML parser or Javascript interpreter - both of these are unlikely.

The contexts in which Zend\Escaper should be used are HTML Body, HTML Attribute, Javascript, CSS and URL/URI contexts.

Every escaper method will take the data to be escaped, make sure it is utf-8 encoded data, or try to convert it to utf-8, do the context-based escaping, encode the escaped data back to it’s original encoding and return the data to the caller.

The actual escaping of the data differs between each method, they all have their own set of rules according to which the escaping is done. An example will allow us to clearly demonstrate the difference, and how the same characters are being escaped differently between contexts:

$escaper = new Zend\Escaper\Escaper('utf-8');

// <script>alert("zf2")</script>
echo $escaper->escapeHtml('<script>alert("zf2")</script>');
// &lt;script&gt;alert&#x28;&quot;zf2&quot;&#x29;&lt;&#x2F;script&gt;
echo $escaper->escapeHtmlAttr('<script>alert("zf2")</script>');
// \x3Cscript\x3Ealert\x28\x22zf2\x22\x29\x3C\x2Fscript\x3E
echo $escaper->escapeJs('<script>alert("zf2")</script>');
// \3C script\3E alert\28 \22 zf2\22 \29 \3C \2F script\3E
echo $escaper->escapeCss('<script>alert("zf2")</script>');
// %3Cscript%3Ealert%28%22zf2%22%29%3C%2Fscript%3E
echo $escaper->escapeUrl('<script>alert("zf2")</script>');

More detailed examples will be given in later chapters.

The Problem with Inconsistent Functionality

At present, programmers orient towards the following PHP functions for each common HTML context:

  • HTML Body: htmlspecialchars() or htmlentities()
  • HTML Attribute: htmlspecialchars() or htmlentities()
  • Javascript: addslashes() or json_encode()
  • CSS: n/a
  • URL/URI: rawurlencode() or urlencode()

In practice, these decisions appear to depend more on what PHP offers, and if it can be interpreted as offering sufficient escaping safety, than it does on what is recommended in reality to defend against XSS. While these functions can prevent some forms of XSS, they do not cover all use cases or risks and are therefore insufficient defenses.

Using htmlspecialchars() in a perfectly valid HTML5 unquoted attribute value, for example, is completely useless since the value can be terminated by a space (among other things) which is never escaped. Thus, in this instance, we have a conflict between a widely used HTML escaper and a modern HTML specification, with no specific function available to cover this use case. While it’s tempting to blame users, or the HTML specification authors, escaping just needs to deal with whatever HTML and browsers allow.

Using addslashes(), custom backslash escaping or json_encode() will typically ignore HTML special characters such as ampersands which may be used to inject entities into Javascript. Under the right circumstances, browser will convert these entities into their literal equivalents before interpreting Javascript thus allowing attackers to inject arbitrary code.

Inconsistencies with valid HTML, insecure default parameters, lack of character encoding awareness, and misrepresentations of what functions are capable of by some programmers - these all make escaping in PHP an unnecessarily convoluted quest.

To circumvent the lack of escaping methods in PHP, Zend\Escaper addresses the need to apply context-specific escaping in web applications. It implements methods that specifically target XSS and offers programmers a tool to secure their applications without misusing other inadequate methods, or using, most likely incomplete, home-grown solutions.

Why Contextual Escaping?

To understand why multiple standardised escaping methods are needed, here’s a couple of quick points (by no means a complete set!):

HTML escaping of unquoted HTML attribute values still allows XSS

This is probably the best known way to defeat htmlspecialchars() when used on attribute values since any space (or character interpreted as a space - there are a lot) lets you inject new attributes whose content can’t be neutralised by HTML escaping. The solution (where this is possible) is additional escaping as defined by the OWASP ESAPI codecs. The point here can be extended further - escaping only works if a programmer or designer know what they’re doing. In many contexts, there are additional practices and gotchas that need to be carefully monitored since escaping sometimes needs a little extra help to protect against XSS - even if that means ensuring all attribute values are properly double quoted despite this not being required for valid HTML.

HTML escaping of CSS, Javascript or URIs is often reversed when passed to non-HTML interpreters by the browser

HTML escaping is just that - it’s designed to escape a string for HTML (i.e. prevent tag or attribute insertion) but not alter the underlying meaning of the content whether it be Text, Javascript, CSS or URIs. For that purpose a fully HTML escaped version of any other context may still have its unescaped form extracted before it’s interpreted or executed. For this reason we need separate escapers for Javascript, CSS and URIs and those writing templates must know which escaper to apply to which context. Of course this means you need to be able to identify the correct context before selecting the right escaper!

DOM based XSS requires a defence using at least two levels of different escaping in many cases

DOM based XSS has become increasingly common as Javascript has taken off in popularity for large scale client side coding. A simple example is Javascript defined in a template which inserts a new piece of HTML text into the DOM. If the string is only HTML escaped, it may still contain Javascript that will execute in that context. If the string is only Javascript escaped, it may contain HTML markup (new tags and attributes) which will be injected into the DOM and parsed once the inserting Javascript executes. Damned either way? The solution is to escape twice - first escape the string for HTML (make it safe for DOM insertion), and then for Javascript (make it safe for the current Javascript context). Nested contexts are a common means of bypassing naive escaping habits (e.g. you can inject Javascript into a CSS expression within a HTML Attribute).

PHP has no known anti-XSS escape functions (only those kidnapped from their original purposes)

A simple example, widely used, is when you see json_encode() used to escape Javascript, or worse, some kind of mutant addslashes() implementation. These were never designed to eliminate XSS yet PHP programmers use them as such. For example, json_encode() does not escape the ampersand or semi-colon characters by default. That means you can easily inject HTML entities which could then be decoded before the Javascript is evaluated in a HTML document. This lets you break out of strings, add new JS statements, close tags, etc. In other words, using json_encode() is insufficient and naive. The same, arguably, could be said for htmlspecialchars() which has its own well known limitations that make a singular reliance on it a questionable practice.