ѧ
JOURNAL OF SOFTWARE
2000Vol.11No.4P.557-562




һģھԵĶȡ
ع챦
ժҪӳ-͹ϵͼʾͺӳ֮Ĺϵ,ģھԵļ׼򣬲ɾӳģھ۶ȵӰ.ڴ˻ϣ˻ģھԵĶȡ㷨.
ؼʡ,ȡ,ھ,ܶ,ص.
ͼšTP311
An Object-Extracting Approach Using Module Cohesion
ZHOU Yu-mingXU Bao-wen
Department of Computer Science and EngineeringSoutheast UniversityNanjing 210096
AbstractIn this paper, an St (subprogram-type) graph is introduced to represent the relation between subprograms and types in programs. Several module cohesion metrics are discussed, and the effects of adding a subprogram to or deleting a subprogram from a module are analyzed based on module cohesion. An object-extracting algorithm is proposed.
Key wordsObject, object-extracting, cohesion, tightness, overlap.
ĵ,,ά.ִĿԻǷ,򾡹Եʱûʹļ,ټЩûõı̽ṹ,άʮ.,ϵͳгȡӳ֮Ĺϵ,ܵػԭϵͳԱͷеġ,תϵͳ,ϵͳĿɿԡάԺͿ.
ĹؼͨԳΪϸ,ȡصݺӳ,ϳʵĶ.ⷽвٹ14,LiuWidleĻȫֱͻ͵ķ1ݡPanoseĻڽķ3,Щֻǽݺӳͻȫֱ򵥵ķ,ûпģ,ȡĶпʵеĶ󲻷.CanforaLiuWidleķĻͼڲȵķ2,÷ʵϻȫֱ,˼븴ڶȡеòͬʵ.Ϊ˷Щ,һֻģھԵĶȡ.
1ӳ-͹ϵͼ
֪,ӳ̺ͺ,̳β,βͷֵ,ʱ,ӳڻ漰ȫֱĴ,ӳβֵ͡ͺȫֱͽ,ӳ֮ЩͶ.ڶʶ,͵ǲһ,ͱȼ͵ô.Ϊ̻͵ĸӳ̶,͸Ӷȵĸ.t1t2,t1t2һɷ.tĸӳ̶Ct,lгɷ͵ĸӳ̶ȺͳɷtǶײlй(tڱĲ0),䶨£

vΪ͵ĸӶ,tΪָָ,ComplexΪָ͵ĸӳ̶.fʾtĸӳ̶гɷt1,t2,...,tnĸӶCt1,l1,Ct2,l+1,...,Ctn,l+1ĺ,Զ.,Ada,tΪ¼ʱ,fȡCti,l+1(tiΪ¼ĳɷ); tΪʱ,fȡdim*2*Ct,l+1(Ϊtĸάd,ԪؾĿû̫ϵ.tΪԪ,2ԭһάĸӶȻֻһɷֵļ¼,ǳ2Ϻ).ڴ˻¶壺
(ӳ-͹ϵͼ). Ssubpʾϵͳеӳ,Ssubpʾϵͳеͼ.ͼG=(N,E)N=SsubpStype,E=(p,t)pSsubptSsubptΪp漰ͣ,һ(p,t)ϵȨֵΪCt,0,ͼGΪӳ-͹ϵͼ.
 ͼGеһӳp,漰ͼΪSp-t(p),Sp-t(p)=ttStype(p,t)E;һͽtйصӳΪSt-p(t),St-p(t)=ppSsubp(p,t)E.
,ϵͳӳ֮Ĺϵӳ-͹ϵͼ,򵥵Ķʶ𷽷ӳ-ͼҳһͼ,ÿͼ͹һ.ַȡ,пʶĶûʵ.Ҫ.һǽ߼ڲͬӳװͬһ.,ջPush-Stack (S: in out STACK-TYPE;Elem: in ELEM-TYPE)ӳ,жEnter-Queue (Q: in out QUEUE-TYPE; Elem: in ELEM-TYPE)ӳ,ӳ-͹ϵͼ,ӳ漰ͬELEM-TYPE,ķǾͷװͬһ.һǽ߼ڲͬԷװͬһ.磬ӳInit (S: in out STACK-TYPE; Q: in out QUEUE-TYPE)ջͶжĳʼ,ջͶӳ-͹ϵͼӳInitͬһͼ,ķͽջͶзװͬһ.Щ³ȡĶȻûʵ.ģھȡ,ԱʵĶĳ.Ϊ,ģھԼԷ.
2ģھԷ
ģͼԼЩ͵ӳ,ģӳ֮漰ͬʹڹϵ.ģھԿģڲԪ֮ϵĽ̶̻ܳ,ϵͳгȡģӦ̸ھۡϵԭ,ȡģӦнϸߵĶ,ϵеԪزӦֲڶģ,Ӧֵͬһģ.
ӳ-͹ϵͼG=(N,E),MTint(G)ʾͼGӳͬ漰,
MTint(G)=pSsubpSp-t(p).
MTint(G)ģӳ֮漰Ĺͬͼ.,Ϊ,ãSʾκļSеԪظ.Ϊʾģھ۶,ģܶȡصȺȵĸ.
ģGĽܶTightness(G)ģӳͬ漰MTint(G)Ը֮ģ漰StypeԸ֮͵ıֵʾ.ģGһ߽ܶζģGڸӳ漰ĹͬͱȽ϶漰͵ĸӶȱȽϸ,Щӳ,ģھ۳̶Ƚϸ.༴

Ȼ,MTint(G)=Stype(ģӳ漰Ͷͬ)ʱ,ģĽ϶ȴﵽֵ1.
ģGصOverlap(G)ʾƽһӳ漰УMTint(G)ռĸԱֵ,һӳжԹִƽ̶.صȸ߱ӳģĴй,ĳ淴ӳӳĸ߶ȹϵ,Ҳģиھ۵ı.ɴˣ

CanforaĶȡ,Ϊͼеӳӳڲȵĸ,ȻݸӳڲȶԱͼӦĺϲƬɾ,õͼ,ÿͼΪһѡ.Ķȡӳıй,ĸӳ̶޹,ȻһЩ.ʹõӳ-͹ϵͼ,ҿǵӳ֮Ĺ̶ȲǴ͸й,͵ĸӳ̶,ˣCanforaӳڲ¶.
ǽͼڲȶΪ㶼ͼڵбϵȨֵ֮һͼڵбϵȨֵ֮͵ıֵ.,ϵȨֵøñһ˵ͶĸӶȼ.ӳpڲָ漰Sp-t(p)ӼӳͼSp-t(p)ɵͼڲ.,ǵõͼڲIC(p)ʽ:

P(p)=piSp-t(ppi)Sp-t(p).
ģĽܶȡصȺӳģھԵķӳ,ǵֵԽ,ģھ۶Խ.Щֶ,һڵ㷨Ϳӳ-͹ϵͼҳһЩͼ:ͽӳɵͼ,нϸߵھ,һ̶ϿԱ⽫߼ڲͬӳװͬһڣ߼ڲͬԷװͬһ.
3ȡ
3.1Ӻɾӳģھ۶ȵӰ
ӳ-͹ϵͼ,ȡͼҪкϲͷ,ʱжģɾӳģھ۶ȵӰ.Ϊ,ȶԷģӺɾӳھ۶ɵӰ.ӻɾӳΪp,漰ΪSp-t(p),ıģΪG.
3.1.1 ģһӳģھ۶ȵӰ
ģGһӳpʱ,ģܶȡصȼMTintĶֱƵ.
1. 

2. 

1,ģͬ漰͵ĸӶֵӳ漰͸Ӷ֮͵ıȴڵԭģĽܶʱ,ģĽܶȲŲС.2ģصȵı仯ȡӳ漰ģеĹ͵ĸӶ֮漰͸Ӷ֮͵ıֵԭģص֮Ĺϵ.ģȶ,ȻSp-t(p)Sp-t(p)=ʱ,Ȳ䣻Sp-t(p)Sp-t(p)ʱ,ȱ,ȱС.
ķ֪ӳ漰Ͱԭģ漰Ĺʱ,ӳ漰Ĺ͵Ը漰͸֮ȴԭģĽܶȻص,ģھ۶Ȼᱣֲ..
1. MTint(G)Sp-t(p)ʱ,
 
 Tightness(G).
ر,MTint(G)=Sp-t(p)ʱ,Overlap(G)Overlap(G),Tightness(G)Tightness(G). 
3.1.2 ģɾһӳģھ۶ȵӰ
ɾӳӳĹ෴,ģɾһӳpʱ,ģܶȡصȼMTintĶֱƵ.
3. 

4. 

ӳPȵı仯ڱɾӸP'漰P漰֮Ĺϵ.Sp-t(p')Sp-t(p)ʱ,Ȳ;Sp-t(p')Sp-t(p),ȱС,,Ȼ.
3֪,ģܶȵı仯ȡڱɾӳɵģͬ漰͸Ӷȵ漰͵ĸӶȵıֵԭģ֮ܶĹϵ.4˵,ģصȿҲܼС,ȡڱɾӳ漰ģеĹ͵ĸӶ֮漰͸Ӷ֮͵ıֵԭģص֮ĺϵ.֪,ɾӳp',ģ漰Ĺûıʱ,ɾӳp'漰ԭģ鹫͵Ը漰͸֮СԭģĽܶȻصʱ,ģھ۶Ȼᱣֲ..
2. MTint(G)=MTint(G)ʱ,
 
 
һѡģھ۶ȵĳ,Եطģھ۶Ⱥģĸıھ۶ȵӰ,ӶΪѡģĳȡṩ˺ϲѵԭ.,ڴ˻۶ȡ㷨.
3.2ȡ㷨
ͼ֮һ(coincidental connection)ͼ(spurious connection)ϵʽ2.һָһӳʵֶܶȡݽṹ,ÿ߼һ.ʱɽӳָɶӳ,ÿӳʵһֹ,߼һ.磬ӳInitջͶеĳʼ,ʱɽInitָͬӳ,ֱʵֶջͶеĳʼ.ӳʵضĹܶȡͬݽṹ.,ӳCopyջеԪظƵ,ʱָû,ڴнӳȥ.ӳĳͼз,ͿжϳӳһӻǼ.
ӳ-ͼ,ÿͿɵóӦһͼ,Щͼھ۶.ڳȡģҪнϸߵھ۶,ֻЩھ۶ȴֵͼ.Щͼӳ㼯ĽΪ,ӦӻɾӳЩͼھ۶ȵӰ̶жӳ򻮹һͼ.ͼͽ㼯ĽΪ,ӦжǷӦЩͼϲ.ȡ㷨:
(1) ӳ-͹ϵͼ,ÿt,ӳSt-p(t)ͼpSt-p(t)Sp-t(p)ɵͼΪMt,Tightness(Mt)Overlap(Mt)(ھ۶ȵķ֪,ֵԽ,ʾͼھ۶Խ),MtĽܶȻصȴһֵStep,MtOverStepб.
(2) ͼMt1,Mt2OverStep,ͼͽĽаt1,t2ʱ,¼:
 Mt1,Mt2ϲΪһµͼMt,ͼھ۶ȴֵStep,ͬһϵеҪ,˽ϲͼMtOverStepб.ӳ-͹ϵͼ,µ͹ϵͼ,ͽt1,t2ϲΪһͽt,ԭͼ,дӳָt1,t2Ļͼ϶ָͽt.
 Mt1,Mt2ϲΪһµͼMt,ͼھ۶СֵStep.ʱ,ĳӳͬʱȡt1,t2ͼMt1,Mt2֮ļ,ӳMt1Mt2ͬʱȥĳӳɵһ,ָɼ߼ܶӳ.
(3) ظ(2),ֱMt1,Mt2OverStep,ͼͽĽвͬʱt1,t2Ϊֹ.
(4) ͼMt1,Mt2OverStep,ӳĽΪSet.Ϊÿӳ߼ֻһ,ˣSetеӳӦָͬͼ.Mt1,Mt2ȥSetӳ㼰ıߺͼֱΪMt1,Mt2.Setеӳp,pMt1Mt2ھ۶СStep,p.3֮һ:
 pMt1ھ۶ӶMt2ھ۶ȼС;
 pMt1ھ۶ӵĳ̶ȴڻMt2ھ۶ӵĳ̶;
 pMt1ھ۶ȼСĳ̶ȴڻMt2ھ۶ȼСĳ̶ȣ
pӦM1.ıͼOverStepб.
(5) ظ(4)ֱMt1,Mt2OverStep,ͼӳĽΪΪֹ.
(6) OverStepбеÿͼͽΪ,ӳΪװһ.
㷨еֵStepͨԴģھ۶ͳƷõ,ҲΪһ,Թ۲ͷͬھ۶ȲζϵͳгȡĶӰ.
3.3̳йϵĳȡ
Ϊڳȡļ̳,棬ǷʽһЩ.һãli:Bi1..niݱʾ,liǶķ,Bili.A:BʾAB,̺ǣA:Ba:A,a:B.̺ĺһ͵͵ʵҲǸ͵ʵ.ӦĶ֮͹ϵ,̺֮ϵ,һ̳һԺͷ.̳Եĳȡ̾ҳӦ͵͹ϵĹ.͵͹ϵĲͬӦڶ̳йϵĲͬ.һǿĶ͵Ķ:͹ϵBi:Bi,i1..n,Уli:Bi1..n+miݣ:li:Bi1..ni.
ȡļ̳ԷΪ漸һڵ㷨ڳгȡ,ʽرʾ;Ȼͨ͵ҳͼ͹ϵ,ȡļ̳йϵ;󽫶ͼ̳йϵתΪӦļ̳йϵ,Ӷ̳иԺͷ,תӶԺͷҪʵɾ.
4
ӴгȡεĶ,ϾǴԭ,ԱͷеġһĲ,ҳδʵʩԱҪĶ.ͨ򹤳̵ķȡеġ,ԭϵͳ˼,ĿάԺʹĿ.LiuWidleĻȫֱͻ͵ķԼڽķûпģ,ֻǽݺӳͻȫֱ򵥵ؼԷ,пܽͬһķԷɢģ,ӶȡĶпʵеĶ󲻷CanforaķĳЩµóĽܲ.ĴģھԳ,̶ȸߵӳϳɶ,ӶȡĶڲнϸߵھ۶.ѽϡAda򹤳ά֧żо⿪һԭϵͳ.Ĺ,Ada83ʶ𲢳ȡֶ,תΪAda95,ҪĲ־Ƿ񵽱ת5.ڽĹУڶȡʵܼӽ֪ʶĴ.
عϴѧѧ빤ϵϾ210096
챦ģϴѧѧ빤ϵϾ210096
ο
1Liu S S, Wilde N. Identifying objects in a conventional procedural language: an example of data design recovery. In: Proceedings of the IEEE Conference on Software Maintenance. San Diego, CA: IEEE Computer Society Press, 1990. 266271
2Canfora G, Cimitile A, Munro M et al. A reverse engineering method for identifying reusable abstract data types. In: Proceedings of the 1st IEEE Working Conference on Reverse Engineering. Baltimore, MD: IEEE Computer Society Press, 1993. 7382
3Livadas Panose, Johnson Theodore. A new approach to finding objects in programs. Journal of Software Maintenance: Research and Practice, 1994,6:249260
4Pedrycz Witold, Waletky James. Fuzzy clustering in software reusability. Software-Practice and Experience, 1997,27(3):245270
5Li Bang-qing, Xu Bao-wen, Yu Hui-ming. Transforming Ada severing tasks into protected objects. In: Proceedings of the ACM SIGAda Annual International Conference. Washington, DC: ACM Press, 1998. 240245


