This project involves implementing the construction algorithm for the FDT
(Frame Document Tree), a hierarchial structure representation between the
frames and pages in a web program, and gathering
and evaluating statistics on the characteristics of FDT's for real
web applications. The open ended part of this project is the generalization
to the various languages available for web programming.
First you have to understand the concept of a web application , and
what we are trying to do. Sections 1 and 2 of  will give you a
pretty good idea of what we are trying to do. You also have to understand
the purpose of testing and why testing is done. A good reading would be
Then you have to learn the languages in which web applications have been
written, since we are trying to model them, based on the code. Understanding
characterisitics of the language that the web applciations are written
in will help you follow if the implementation that you are coming up with
is doing the correct required thing. We have looked at web applications
a lot. You can learn it as you go along. Try to familiarize yourself
The syntax and the semantics of the languages you can learn as you go along.
The implementation that you will develop is of the FDT. Once you have
a good understading of the languages used and what we are trying
to do, it is time to jump into the algorithm that we have come up with.
Your goal is to follow the algorithm to come up with a structure that looks
similar to what we have. We will discuss design decisions for the actual
implementation, but the overall process involves 2 steps
Generalize the FDT to other web programming languages. Run code from web
applications that are written in various web programming language through
your implementation and see if the FDT is generated as desired.
Extend the implementation to also generate statistics on characteristics
of FDT's (eg. #nodes, #edges of different types).
You have to parse the web application to get the fields that you want.
For eg: in the FDT, you are looking for a keyword <frameset>. You have
to be ableto read in the web page and pick out occurences of these terms.
These terms denote the nodes and edges of your graph, as you will understand
once youknow what the FDT does. A language that recognizes tokens and makes
it easy to read them is Perl.
Once you have all your nodes and edges, you will code in a language (C,
C++, Java ?? ) to generate a data structure and also print the FDT as shown
in the paper.
Language(C, C++, Java??)
Modelling Web applications for static analysis. Sreedevi Sampath,Amie Souter,
Laura McGlade, Margarita Golod, Lori Pollock.
Characterisitics of web applications, Sreedevi Sampath.
Modeling web applications with UML, Jim Conallen
Modeling web applications architecture with UML, Jim Conallen.
Untangling the woven web: testing web based software. Gary McGraw, David
Website testing. Edward Miller
Testing of object-oriented software, excerpt from Amie's dissertation.
Book: Java script. Gosselin.
Object-based data flow testing of web applications. Chien Hung Liu, David
C. Kung, Pei Hsia, Chih-Tung Hsu. IEEE 2000.
Structural Testing of Web Applications. Chien Hung Liu, David C. Kung,
Pei Hsia, Chih-Tung Hsu. IEEE 2000.
Building a tool for analysis and testing of web applications. Problems
and Solutions. Filippo Ricca and Paolo Tonella.
This project will focus on design and implementation of an initial prototype
of the Web Program Frame Graph (WPFG) - program representation for
By now if you have implemented the FDT, you are probably very familiar
with the concept of web applications and testing web applications. You
are also probably familiar with the main entities of the FDT that you have
The next step is to design and build an initial prototype implementation
of the WPFG. For this, you will have to read the section of  related
to buliding the WPFG. The WPFG is more complex than the FDT, in that it
contains many more things to look out for when building it. The WPFG captures
the control flow of web applications. There are various types of edges
and nodes, that have to be represented in your implementation.
Before we go into the details of implementing the construction algorithm
for the the WPFG, you need to have an idea of the basics of a compiler/interpreter
and how it works. (Reading: )
Lets see how we go about recognizing the features of a web application
that have to be captured and modelled for implementing the WPFG. We need
control flow within the web application to be modelled.That is the essence
of the WPFG.
and html. A good place to start searching would be at the developers site
I found is at http://www.mozilla.org/js/spidermonkey/. It might be
a reference point to start looking or to guide you in what you have to look for
when searching elsewhere.
Investigate and design implementation of the WPFG with local cfg's (control
flow graphs) for each unit. Can we use Perl or do we need a parser? Find
code is what we are looking for, so an open source development web site
would be a good place to start looking).
Design implementation of the local cfg's
Building data structures.
Parser and making changes to parser to generate desired information for
Book: Compilers: Principles, Techniques and Tools. Aho, Sethi and Ullman.
Rest are the same as for Project 1