cov.py

import os
import operator
import sys
from coverage import coverage


passfile = [ "XMLData/passing/" + name for name in os.listdir("XMLData/passing/") ]
failfile = [ "XMLData/failing/" + name for name in os.listdir("XMLData/failing/") ]

def cover2set(filelist):
    sets = {}
    for name in filelist :
        filename = []
        cov = coverage()
        cov.start()

        import __main__
        sys.path[0] = os.path.dirname(sys.argv[0])
        sys.argv = ("xpcmd.py",name)
        try :
            execfile (sys.argv[0], __main__.__dict__)
        except :
            print "fail"

        cov.stop()
        cov.save()
        #get morf name
        for codeunit in cov._get_file_reporters() :
            filename.append(codeunit.name)

        for morf in filename :
            module,execstat,miss,readable_miss = cov.analysis(morf)
            runline = list(set(execstat).difference(set(miss)))
            if morf in sets :
                for i in runline :
                    if i in sets[morf] :
                        sets[morf][i] += 1
                    else:
                        sets[morf][i] = 1
            else :
                temp = {}
                for i in runline :
                    temp[i] = 1
                sets[morf] = temp
        cov.erase()
    return sets

def tarantula(ns,nf,ncf,ncs,nuf):
    susp = (float(ncf)/nf)/((float(ncf)/nf) + (float(ncs)/ns))
    return susp

def ochiai(ns,nf,ncf,ncs,nuf):
    susp = float(ncf)/((nf*(ncf+ncs))**(0.5))
    return susp

def ochiaiop(ns,nf,ncf,ncs,nuf):
    susp = abs(ncf - float(ncs)/(ns+1))
    return susp

def dstar(ns,nf,ncf,ncs,nuf):
    star = 2
    susp = float(ncf**star)/(nuf+ncs)
    return susp

def numbercover(sets,filename,line):
    try:
        result = sets[filename][line]
    except KeyError:
        return 0
    return result

def calrelevance(ns,nf,passset,failset,filelist,metric):
    coverrank = {}
    for filename in filelist :
        try :
            lines = list(set(passset[filename]).union(set(failset[filename])))
            temp = {}
            for line in lines :
                ncf = numbercover(failset,filename,line)
                ncs = numbercover(passset,filename,line)
                nuf = ns - ncs
                nus = nf - ncf
                temp[line] = metric(ns,nf,ncf,ncs,nuf)
            temp = sorted(temp.items(),key=operator.itemgetter(1))
            coverrank[filename] = temp[::-1]
        except KeyError:
            pass
    return coverrank

def output(filelist,number,resultset):
    for filename in filelist :
        print "filename : %s " % filename
        for i in range(number):
            try :
                line = resultset[filename][i][0]
                susp = resultset[filename][i][1]
                print "line:%d susp:%f" % (line,susp)
            except KeyError:
                pass

passset = cover2set(passfile)
failset = cover2set(failfile)
filelist = list(set(passset).union(set(failset)))

ns = len(passfile)
nf = len(failfile)

tarantula_result = calrelevance(ns,nf,passset,failset,filelist,tarantula)
ochiai_result = calrelevance(ns,nf,passset,failset,filelist,ochiai)
ochiaiop_result = calrelevance(ns,nf,passset,failset,filelist,ochiaiop)
dstar_result = calrelevance(ns,nf,passset,failset,filelist,dstar)

print
print "top 3 of tarantula"
output(filelist,3,tarantula_result)
print
print "top 3 of ochiai"
output(filelist,3,ochiai_result)
print
print "top 3 of ochiaiop"
output(filelist,3,ochiaiop_result)
print
print "top 3 of dstar"
output(filelist,3,dstar_result)