Transcript Reverse Engineering

Reverse Engineering
Valeriya Perelman 12/10/04
Outline






Motivation
Terminology
Related work
Approach discussion
Challenges
References
Motivation




Improving comprehensibility of software
systems
Modifying software while keeping design and
source code synchronized
Software re-use
Reengineering software on the design level
Terminology
Re-engineering
Requirements
Forward
engineering
Analysis
Design
Implementation
Reverse
engineering
Related Work
-- Commercial CASE Tools

UML-Based Static Reverse Engineering is supported by:
–
–
–
–
Rational Rose
Together
Poseidon
CodeLogic (Can generate control-flow and sequence diagrams per method
given by user)
Pros:


Fully automatic process of RE
Highly scalable
Cons:



Dynamic aspect of design is not supported in the RE
process(except for CodeLogic)
Not all the relations are identified
Multiplicity is not supported
Related Work
-- Research Prototypes

UML-Based
–
IDEA (Redocumentation of Java programs using UML class
diagrams)
–
FUJABA (generating both dynamic and static aspects,
requires specific code conventions):
Example for an Association Comment
/**
* <h2>Associations</h2>
*
* <pre>
*
cardLeft
cardRight
* LeftClass ------------------------------ RightClass
*
leftRole
rightRole
* </pre>
*/
Related Work
-- Research Prototypes – cont.

Others
–
Rigi






Has its own format (RSF-Rigi Standard Format)
Semi-automatic
Implemented for C/C++
Hierarchical clustering of related artifacts into subsystems
Supports model visualization
The RE based on certain modularity principles (such as
data abstraction, low coupling among subsystems e.t.c.)
Related Work
-- Design Pattern-Based Design
Recovery Techniques

Searching for minimal key structures:
–


Searching for class structures (Pat, IDEA)
Searching based on fuzzy logic:
–

DP++, KT, SPOOL
Technical Report tr-ri-01-222, University of Paderborn
Searching based on metrics (Object oriented, structural
and procedural):
–
Wizzard for C++ by Kim, H., Boldyreff, C.
Approach Discussion
-- Input/Output


Input: Java source code (widely used, OOP,
built in design patterns, interfaces)
Output: OPM design model (equal relation to
Objects and Processes, hierarchical, giving
simple way to express both dynamic and static
aspects of the system)
Approach Discussion
-- Techniques





Static Analysis
Identifying Objects vs. Processes (every class
is object except for…)
Filtering simple getters/setters
Detecting GOF patterns
Improving analysis to the OPM generated code
using its’ javadoc.
Approach Discussion
-- Simple Example
public class Hello {
int main(){
System.out.println(“Hello
World!”);
return 1;
}
}
Approach Discussion
-- Challenges




Weakly typed containers (Sets, Hash tables,
e.t.c.)
Scalability problem (number of patterns vs.
processing, over-control flow analysis)
Language Independency
Re-Tour engineering
References

Links:
–
http://www.rigi.csc.uvic.ca/UVicRevTut/F4rev.html

–
http://www.iste.uni-stuttgart.de/ps/reengineering

–
Reengineering Bibliography
http://www.gupro.de/GXL/Introduction/intro.html


Overview on Reverse Engineering + list of RE Tools
Graph eXchange Language format
Papers:
–
–
Design Pattern Recovery in Architectures for Supporting
Product Line Development and Application
Philippow, Streitferdt & Riebisch
Pattern-Based Design Recovery of Java Software
Seemann & Wolf von Gudenberg