Script Security Threat: Untrusted Data

The best way for scripts to accept untrusted data while preserving trust is through rigorous input validation and output encoding.

Input validation involves checking all incoming data against a strict set of rules to ensure it conforms to the expected format, type, and range. It is generally recommended to use a "whitelist" approach, allowing only known good input, rather than a "blacklist" approach, which tries to block known bad input. For instance, if a field expects an integer, reject any non-integer input. If a name field is expected, filter out any characters not typically found in names, like angle brackets (<,>) or script tags.

Output encoding, also known as contextual encoding or escaping, is crucial when displaying or processing the untrusted data. This means transforming characters that have special meaning in a particular context (e.g., HTML, JavaScript, URL) into their harmless entity equivalents before the data is rendered or used. For example, the less than symbol (<) should be encoded as < when placed into an HTML context to prevent cross-site scripting (XSS) attacks. This practice is highlighted in advisories like CERT® Advisory CA-2000–02 Malicious HTML Tags Embedded in Client Web Requests, which discusses how malicious HTML tags can be embedded in client requests.

These methods collectively help prevent common vulnerabilities like SQL injection, cross-site scripting (XSS), and command injection, as detailed in texts on computer network security, such as Guide to Computer Network Security by Joseph Migga Kizza.