// Reduce
func (this *FileWorker) Reduce(key interface{}, values []interface{}) (kv mr.KeyValue) {
const threhold = 0
var occurence = stats.StatsSum(mr.ConvertAnySliceToFloat(values))
if occurence > threhold {
kv = mr.NewKeyValue()
kv[NIL_KEY] = occurence
}
return
}
func (this *UniWorker) Reduce(key interface{}, values []interface{}) (kv mr.KeyValue) {
// here we don't care about the key
// we only care about values
aggregate := stats.StatsSum(mr.ConvertAnySliceToFloat(values))
if aggregate > 100 {
kv = mr.NewKeyValue()
kv[NIL_KEY] = aggregate
}
return
}
// The key is already sorted
func (this *KxiWorker) Reduce(key interface{}, values []interface{}) (kv mr.KeyValue) {
kv = mr.NewKeyValue()
switch key.(mr.GroupKey).Group() {
case GROUP_URL_SERV:
vals := mr.ConvertAnySliceToFloat(values)
kv[TIME_ALL] = stats.StatsSum(vals)
kv[TIME_MAX] = stats.StatsMax(vals)
kv[TIME_TOP] = stats.StatsSumTopN(vals, topsum)
kv[TIME_AVG] = stats.StatsMean(vals)
kv[TIME_STD] = stats.StatsSampleStandardDeviationCoefficient(vals)
kv[CALL_ALL] = float64(stats.StatsCount(vals))
case GROUP_KXI:
vals := mr.ConvertAnySliceToFloat(values)
kv[TIME_ALL] = stats.StatsSum(vals)
kv[TIME_MIN] = stats.StatsMin(vals)
kv[TIME_TOP] = stats.StatsSumTopN(vals, topsum)
kv[TIME_MAX] = stats.StatsMax(vals)
kv[TIME_AVG] = stats.StatsMean(vals)
kv[TIME_STD] = stats.StatsSampleStandardDeviationCoefficient(vals)
kv[CALL_ALL] = float64(stats.StatsCount(vals))
case GROUP_URL_RID:
vals := mr.ConvertAnySliceToFloat(values)
kv[CALL_ALL] = float64(stats.StatsCount(vals))
kv[TIME_ALL] = stats.StatsSum(vals)
case GROUP_URL_SQL:
vals := mr.ConvertAnySliceToFloat(values)
kv[CALL_ALL] = float64(stats.StatsCount(vals))
kv[TIME_MAX] = stats.StatsMax(vals)
kv[TIME_AVG] = stats.StatsMean(vals)
case GROUP_URL:
vals := mr.ConvertAnySliceToString(values) // rids of this url
c := stats.NewCounter(vals)
kv[REQ_ALL] = float64(len(c))
}
return
}