Recovering a Business ObjectModel from Web Applications    U. De Carlini, A.R. Fasolino, P. Tramontana             Diparti...
Web Applications (WA):             problems and open issuesWorld Wide Web is evolvingWA are realized with a limited qualit...
Web Applications (WA):           problems and open issues⇒ Maintenance,  Reengineering and Migration are very critical tas...
Web Application Model                                                        0..n                    redirect             ...
Business Level UML DiagramsBusiness level UML diagrams describethe relevant conceptual components(business objects) from t...
Recovering a Business Level Class DiagramA Reverse Engineering process in three main   steps:1) Identification of candidat...
1) Identification of candidate classes Searching for groups of logically related data making up the state of objects.⇒Look...
Identification of relevant groups of data Groups of data in input/output operations    data involved in HTML forms, HTML t...
Synonyms and Homonyms Analysis  Synonyms are identifiers with different names but the  same meaning  Homonyms are identifi...
An automatic procedure to propose a set of            candidate classes    the identified data groups are arranged in a li...
2) Associating methods to candidate                 classes Possible functional units to be considered should include web ...
Coupling between clusters and objects Measures of coupling between clusters and objects are computed based on the accesses...
Associating pages to candidate classes ⇒ If a cluster accesses exclusively one object, it will be  assigned as a method of...
3) Identification of relationships           between classesRelationships are found in two cases:   If two or more candida...
Common attributes If two or more classes Ci have a common attribute:  ⇒  the attribute is assigned to one of these    clas...
Clusters accessing more than one class⇒ Ifa cluster is assigned as a method to a candidate class C0, but it accesses attri...
A case studya WA designed to support the activities ofundergraduated courses offered by aComputer Science DepartmentDevelo...
Identification of groups of dataAfter static analysis of the WA:  128 references to data groups (485 data  items)After Syn...
Data Groups                 Data groups                                                           #An excerpt   C1 Student...
Candidate classes                             Candidate classes and corresponding attributes8 Candidate                   ...
Associating methods to candidate             classesAutomatic clustering approach was applied  44 valid clusters were reco...
Business Level UML Class Diagram                                       Student                                    Name    ...
ConclusionsA reverse engineering approaches for recovering,from the code of a Web application, business aBusiness Object M...
Future WorkThe definition of criteria for a further automation ofthe model reconstruction will be addressed, as wellas the...
Upcoming SlideShare
Loading in …5
×

Recovering a Business Object Model from Web Applications

379 views
298 views

Published on

In this paper a reverse engineering approach for reconstructing UML diagrams at business level of the application domain of a Web Application is presented. In particular the approach allows the reconstruction of the UML class diagram providing an object-oriented conceptual model of the application domain, sequence diagrams modeling the interactions among the identified business objects and use case diagrams modeling the user functionalities provided by the Web Application. Heuristic criteria exploiting source code analysis are used for recovering the diagrams. Tools for implementing these criteria have been produced, and experiments for validating them have been carried out with the support of case studies. Experimental results showed the feasibility and the effectiveness of the proposed approach.

Published in: Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
379
On SlideShare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
2
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Recovering a Business Object Model from Web Applications

  1. 1. Recovering a Business ObjectModel from Web Applications U. De Carlini, A.R. Fasolino, P. Tramontana Dipartimento di Informatica e Sistemistica University of Naples Federico II, Italy G.A. Di Lucca RCOST – Research Centre on Software Technology University of Sannio, Benevento, Italy 1
  2. 2. Web Applications (WA): problems and open issuesWorld Wide Web is evolvingWA are realized with a limited quality due to thepressure of a short time-to-market and anextremely high competition- WA developed without a disciplined process- Poor, inadequate, incomplete documentation- Disordered architecture 2
  3. 3. Web Applications (WA): problems and open issues⇒ Maintenance, Reengineering and Migration are very critical tasks for WA⇒ ObjectOriented Business Level Models are a fundamental starting point for these tasks 3
  4. 4. Web Application Model 0..n redirect link Web Page 0..nClient Module 0..n 0..n 0..n include include 0..n 0..n Client Page Server Page 0..1 build 0..1 0..n 0..n 0..n submit Web Object 0..n Client Page Load in Frame with Frame 0..n 4
  5. 5. Business Level UML DiagramsBusiness level UML diagrams describethe relevant conceptual components(business objects) from the domain ofthe problem addressed by the WA andtheir mutual relationships andinteractions 5
  6. 6. Recovering a Business Level Class DiagramA Reverse Engineering process in three main steps:1) Identification of candidate classes and their attributes;2) Association of methods to candidate classes;3) Identification of relationships between candidate classes. 6
  7. 7. 1) Identification of candidate classes Searching for groups of logically related data making up the state of objects.⇒Looking for language mechanisms that allow grouping of related data implementing a relevant concept, either from the domain of the application or from domain of the solution 7
  8. 8. Identification of relevant groups of data Groups of data in input/output operations data involved in HTML forms, HTML tables, …. Groups of data in database/files read/write operations data involved in Recordsets, Arrays, Collections, ... Groups of data passed throw distinct page data involved in Querystrings 8
  9. 9. Synonyms and Homonyms Analysis Synonyms are identifiers with different names but the same meaning Homonyms are identifiers with the same name but different meanings⇒Synonym identifiers must be assigned with the same unique name.⇒Homonym identifiers must be associated with distinct names.During synonyms & homonyms analysis, a meaningful name is assigned to each data item 9
  10. 10. An automatic procedure to propose a set of candidate classes the identified data groups are arranged in a list and sorted in descending order with the number of references made to each data group, and in ascending order with the cardinality of each group; this sorted list is called OrdList; the first group in OrdList is considered as a candidate object and moved into a new list of candidate objects, CAND; the OrdList is sequentially visited and each group is considered: if a group comprises at least a new data item not yet included in any other group in the CAND list, it will be inserted in CAND; the OrdList is examined until it includes at least a group, or until the union set of all the data items of the candidate objects in CAND and the union set of all the data items of the groups in Glist are equal; if a group h from the OrdList includes all the data items making up one or more groups Ci in CAND, only the k data items in h that are not yet included in any group of CAND are added to the Ci groups whose elements are all included in h 10
  11. 11. 2) Associating methods to candidate classes Possible functional units to be considered should include web pages, functions, script blocks, depending on the requested degree of granularity. The automatic clustering approach proposed in [Di Lucca et al.-IWPC 2002], is used to group Web pages of a WA into meaningful (highly cohesive) and independent (loosely coupled) clusters⇒Our approach proposes to consider these clusters of related Web pages as potential methods to be associated with the candidate objects of the WA 11
  12. 12. Coupling between clusters and objects Measures of coupling between clusters and objects are computed based on the accesses of each page in the cluster to the candidate object. A page accesses a candidate object when it includes instructions that define or use the value of some object attribute 12
  13. 13. Associating pages to candidate classes ⇒ If a cluster accesses exclusively one object, it will be assigned as a method of that object. ⇒ If a cluster accesses more objects, it will be assigned to the candidate class it accesses prevalently ⇒ Clusters that do not make access to any objects will be considered as coordinating modules controlling the executions of other methods 13
  14. 14. 3) Identification of relationships between classesRelationships are found in two cases: If two or more candidate classes have common attributes; If a cluster accesses more than one candidate class 14
  15. 15. Common attributes If two or more classes Ci have a common attribute: ⇒ the attribute is assigned to one of these classes (Cj) and deleted from other classes Ck; ⇒ A relationship will be defined between the class Cj and each class Ck⇒These relationships will be depicted as UML association relationships 15
  16. 16. Clusters accessing more than one class⇒ Ifa cluster is assigned as a method to a candidate class C0, but it accesses attributes of other classes Ci, a relationship will be defined between class C0 and each class Ci⇒Also this kind of relationships will be depicted as UML association relationships 16
  17. 17. A case studya WA designed to support the activities ofundergraduated courses offered by aComputer Science DepartmentDeveloped with ASP, Javascript, HTMLlanguages75 server pages, 23 client pages, 1 utilitymodule (7648 LOCs) 17
  18. 18. Identification of groups of dataAfter static analysis of the WA: 128 references to data groups (485 data items)After Synonyms and Homonyms Analysis: 43 different data groups (26 different data items) 18
  19. 19. Data Groups Data groups #An excerpt C1 Student name, Student surname, Student code, Student email, 14 Student phone number, Student passwordof the 43 C2 Teacher name, Teacher surname, Teacher email, Teacher phone 11different number, Teacher password, Teacher code C3 Exam date, Exam time, Exam classroom 10Data C4 Student name, Student surname, Student code, Student email, 8Groups Student phone number C5 Tutoring date, Tutoring start time, Tutoring end time, Course code, 7obtained Course nameafter C6 Student code 6Synonyms C7 Course code, Course name, Course academic year C8 Course code 5 4and C9 Course code, Course name 4Homonyms C10 Teacher name, Teacher surname, Teacher email, Teacher phone number, Teacher password 4Analysis C11 Course code, Course academic year 3 19
  20. 20. Candidate classes Candidate classes and corresponding attributes8 Candidate Student (Student name, Student surname, StudentClasses code, Student email, Student phone number,obtained after Student password)the execution of Teacher (Teacher name, Teacher surname, Teacherthe algorithm email, Teacher phone number, Teacher password, Teacher code)Produce Exam Session (Exam date, Exam time, Exam classroom)Candidate Tutoring (Tutoring date, Tutoring start time,Objects. Tutoring end time, Course code, Course name) Course (Course code, Course name, Course academic year) Tutoring Request (Student name, Student surname, Student code, Tutoring request date) News (Course code, News text, News number, News date, Teacher code) Exam Reservation (Student code, Student name, Student surname, Course code, Exam date, Exam reservation date) 20
  21. 21. Associating methods to candidate classesAutomatic clustering approach was applied 44 valid clusters were recovered 16 clusters did not reference any data group, so they were considered as coordinator modules 28 clusters were assigned as methods of the candidate classes 21
  22. 22. Business Level UML Class Diagram Student Name Surname Exam Reservation E-mail Date Password Code Phone number Tutoring Date Start time End t ime Course Exam Tutoring request Academic year DateDate Code Time Name Classroom Teacher Name News Surname Number E-mail Date Phone number Password Text Code 22
  23. 23. ConclusionsA reverse engineering approaches for recovering,from the code of a Web application, business aBusiness Object Model has been presentedSome experiments were carried out to assess theeffectiveness of the proposed approaches.Encouraging results as to the adequacy of therecovered models were obtained.A limited human effort required for reconstructingthe models was recorded too. 23
  24. 24. Future WorkThe definition of criteria for a further automation ofthe model reconstruction will be addressed, as wellas the investigation on possible approaches foridentifying UML aggregations, compositions, orgeneralization-specialization relationships betweenclasses will be carried out.A wider experimentation involving more complexWeb Applications, implemented with differenttechnologies, will be moreover carried out, in orderto extend the validity of the proposed approaches 24

×