def processAuthor(oldAuthor,author,maxHr,maxHrCount,currHrCount,oldHr,hr): if author!=oldAuthor: oldHr,maxHr,maxHrCount,currHrCount = processHr(oldHr,hr,maxHr,maxHrCount,currHrCount) for i in maxHr: print oldAuthor+" "+i oldAuthor=author maxHr=[] maxHrCount=0 currHrCount=0 return [oldAuthor,maxHr,maxHrCount,currHrCount,oldHr]
def reducer(): lineCount,maxHr,maxHrCount,currHrCount,oldAuthor,author = initVars() for line in sys.stdin: data = line.strip().split() if len(data)!=2: continue author,hr = data if lineCount==0: lineCount,oldAuthor,oldHr = initOldAuthor(lineCount,author,hr) oldAuthor,maxHr,maxHrCount,currHrCount,oldHr = processAuthor(oldAuthor,author,maxHr,maxHrCount,currHrCount,oldHr,hr) if currHrCount!=0: #only 0 right after seeing a new author oldHr,maxHr,maxHrCount,currHrCount = processHr(oldHr,hr,maxHr,maxHrCount,currHrCount) currHrCount+=1 processAuthor(oldAuthor,author,maxHr,maxHrCount,currHrCount)
# test new author
newOldAuthor,mostActiveHr,maxHrCount,currHrCount,oldHr = processAuthor('a1','a2',['08','09'],3,1,None,None)
if newOldAuthor=='a2' and not mostActiveHr and maxHrCount==0 and currHrCount==0:
print "processAuthor test1 pass"
else:
print "processAuthor test1 failed!"
# test same author
newOldAuthor,mostActiveHr,maxHrCount,currHrCount,oldHr = processAuthor('a1','a1',['08','09'],3,1,None,None)
if newOldAuthor=='a1' and mostActiveHr==['08','09'] and maxHrCount==3 and currHrCount==1:
print "processAuthor test2 pass"
else:
print "processAuthor test2 failed!"
# test new hour, not new max
newOldHr,maxHr,maxCount,currHrCount = processHr(8,9,[7],3,3)
if newOldHr==9 and maxHr==[7,8] and maxCount==3: #and currHrCount==0:
print "processHr test1 pass"
else:
print "processHr test1 failed! output:"
print processHr(8,9,[7],3,3)
# test new hour, new max
newOldHr,maxHr,maxCount,currHrCount = processHr(8,9,[7],3,4)
if newOldHr==9 and maxHr==[8] and maxCount==4: #and currHrCount==0:
print "processHr test2 pass"
else:
print "processHr test2 failed! output:"
print processHr(8,9,[7],3,4)
# test same hour
