This site is the archived OWASP Foundation Wiki and is no longer accepting Account Requests.
To view the new OWASP Foundation website, please visit https://owasp.org

SpoC 007 - OWASP The Anti-Samy Project

From OWASP
Revision as of 15:57, 21 September 2007 by Esheridan (talk | contribs)

Jump to: navigation, search

Back to SpoC 007 Selection page


AoC Candidate: Arshan Dabirsiaghi

Project coordinator: Jeff Williams

Project Progress: 60% Complete, Progress Page

Arshan Dabirsiaghi - OWASP The Anti-Samy Project

Executive Summary

My name is Arshan Dabirsiaghi and I am a 25 year old security consultant. I want to open the door for web sites to allow users to supply their own HTML without exposing them to cross-site scripting attacks.

Background

B.S., M.S. in Computer Science (focus on Information Security)

2.5 years security engineer/consultant experience

4 years of web and systems development

8 years of security hobbying

Challenge

Many sites today would enjoy the ability to allow users to provide their own HTML in order to customize their page layout and general user experience. Because of the concerns regarding XSS, it is generally thought of as 'too dangerous' to allow them to input any HTML at all. Sites like MySpace who have been brave enough to provide this functionality have no standardized, proven solution to validate user HTML. In many cases, it's easier to disallow everything that looks HTML or to output encode all user input. Not coincidentally, those sites have been the targets of complex attacks. Many sites today don't offer this type of functionality because of concerns for XSS and dangerous HTML.

Objectives

The first goal is to conduct a survey of existing browsers and compare their respective behavior regarding JavaScript (both well-formed and not) inside HTML with regards to W3C specifications

The second goal is to create a software library (versions in both .Net and J2EE) that can accomplish the following goals:

  • provide an input validation utility that can detect non-obvious JavaScript inside user-provided HTML
  • provide a filtering utility that can take blocks of HTML and strip any JavaScript inside of it while retaining all formatting-related code
  • provide these capabilities even when dealing with realistically dirty HTML
  • build on the mountain of research available for parsing broken HTML
  • provide feedback information to the user to help them tune their source to fall within acceptable values
  • provide these capabilities in an API that's simple and portable
  • utilize an XML engine file that can be used in various language implementations (.Net/J2EE/PHP)

I envision the project requiring 3 man-months, with a few milestones to be established at reasonable intervals, such as:

  • 3 months out: Begin browser/W3C survey.
  • 2 months out: Finish survey and begin development of API.
  • 1 month out: Complete initial API in both Java and .Net
  • Near release: Perform intense QA on API, fix any remaining bugs.

Long Term Vision

My goal is to create a peer-reviewed and eventually time-tested solution for detecting and filtering JavaScript from user input. By doing so, we can enable web applications to provide users with richer, more interactive experiences without sacrificing security. This tool will not only provide an input validation component to sites who are currently facing this challenge, but it will also act as an enabler to sites who wished to embrace the new age of user-generated content but were unable to do so.

Back to SpoC 007 Selection page