Reverse Engineering to Achieve Maintainable WWW Sites

1
Reverse Engineering to Achieve Maintainable WWW
Sites

• Cornelia Boldyreff and Richard Kewish
• RISE
• Department of Computer Science
• University of Durham
• cornelia.boldyreff_at_durham.ac.uk
• www.dur.ac.uk/cornelia.boldyreff

2
Overview

• Background
• Maintenance Problems Associated with the WWW
• Reverse Engineering of Web Sites
• Results Obtained
• So what and where next...

3
Background

• Web is a huge information base and storage medium
• Rigorous development and maintenance principles
  have not always been applied to web sites
• Speed and breadth of its growth
• Its inherent complexity
• Support where available is focused on development
  rather than maintenance

4
Maintenance Problems with the WWW

• Large web sites exhibit many of the problems of
  large scale software systems - as well
  incorporating legacy systems, the web has its own
  legacy aspects.
• Specific problems with web sites
• broken links
• incorrect or out-of-date data
• inconsistent information
• inconsistent style

5
Root Causes of Problems

• Forced Duplication of files and/or data
• HTML lacks an include directive - designer must
  choose between linking pages or copying data
• The file structure of web sites
• Directories can be used to modularise a site,
  but this is difficult to achieve as a site grows
  and files become interlinked and distributed over
  servers.
• HTML blurs the distinction between data and code

6
Reverse Engineering of Web Sites

• Databases can be used in conjunction with the Web
  
• to incorporate existing company data
• to overcome some of the problems cited, e.g. for
  link storage
• Considerable redevelopment of existing pages may
  be necessary
• Database maintenance and web site maintenance
  must be co-ordinated

7
Why and How - RE

• Older sites are likely to require RE to ensure
  replicated elements within a site are identified,
  abstracted, and moved to a DB or subsumed into a
  scripting language program.
• The approach taken here involves parsing and
  analysing web page content and style, a
  rationalised copy of each unique HTML data
  element is stored in a repository either directly
  or by reference.

8
System Overview

• 1 - HTML File Parser and Parser Tree Analyser
• 2 - Head Analyser
• 3 - Style Sheet Analyser
• 4 - Body Analyser
• 5- Data Storage and Dynamic Page Generation

1
3
2
4
5
9
Implementation Details

• Various DB tables are used - large items stored
  by references to files
• Extra data added - date of last up-date and
  person responsible
• Any changes to DB are recorded in audit files

• Log files detailing web page parsing, analysis
  and storage are produced along with an overall
  analysis summary
• Web page is dynamically recreated from a new web
  page file - its tags contain embedded script
  statements

10
Some Results Obtained
11
So what and where next...

• There is duplication in web sites and it can be
  identified and removed
• A more rational re-structuring and storage of web
  site contents is achievable
• In principle this can facilitate future
  maintenance - but more studies are needed

• Integration with a popular development could lead
  to more widespread usage and allow more in-depth
  evaluation
• More sophisticated clone analysis can be
  considered building on this experimental system

12
Key Points

• Web developers and maintainers can learn from
  Software Engineering.
• Techniques from Reverse Engineering can be
  applied to web sites.
• Better abstractions and structuring principles
  are needed for web site development in the future
  - but there will remain a role for RE as
  described here.