perfer-report

#!/usr/bin/python3

import pprint as pprint
import time as time
import csv as csv
import sys as sys
import perfer as perfer

# Locally defined (maybe later to be put into module, with
# some surrounding functionality?
end="Command finish time (calculated)"

numberFields = (
        perfer.wall, 
        perfer.userCpu, 
        perfer.sysCpu, 
        perfer.mem, 
        perfer.cacheio, 
        perfer.diskio, 
        perfer.swapout, 
        perfer.blockin, 
        perfer.blockout, 
        perfer.vcsw, 
        perfer.icsw, 
        perfer.start 
    )



#####
##### SUBROUTINES / OUTPUT / FORMATTING
#####


def perferReader(inputFile):

    reader = csv.DictReader(inputFile)

    dataPoints = []

    for line in reader:
        for number in numberFields:
            # NOTE: Some of these are actually integers...
            line[number] = float(line[number]) 

        line[end] = line[perfer.start] + line[perfer.wall] # Calcuated, for convenience(?)

        dataPoints.append(line)

    return dataPoints


def topNaggregateMeasurements(data, variable, aggregateOn=perfer.cmd, N=5):
    # After aggregating based on the specified variable, then:
    # sort by the 'variable' being tested, to report the highest N values in that
    # column.
    aggCol = 0
    dataCol = 1
    valueCol = 0
    countCol = 1
    
    aggregate = {}

    for datapoint in data:
        agg = datapoint[aggregateOn]
        thisAgg = aggregate.get(agg, (0.0, 0))
        # Add the value for this datapoint to previous sum, starting at 0.0.
        # Increment the counter for # of occurrences aggregated into this sum
        aggregate[agg] = ( datapoint.get(variable) + thisAgg[valueCol], thisAgg[countCol] + 1)

    output = list(aggregate.items())
    output.sort(key=lambda datapoint: datapoint[dataCol][valueCol])

    print("Programs using the most %s" % (variable))
    print("total\t\tcount\t(avg)\taggregate")

    # return output[(-1*(N)):]

    for topAgg in output[(-1*N):]:
        print("%f:\t%d\t(%f)\t%s" % (topAgg[dataCol][valueCol], topAgg[dataCol][countCol], topAgg[dataCol][valueCol] / topAgg[dataCol][countCol], topAgg[aggCol]))



def topNmeasurements(data, variable, N=5):
    # sort by the 'variable' being tested, to report the highest N values in that
    # column.
    
    data.sort(key=lambda datapoint: datapoint[variable])
    return data[(-1*N):]


if __name__ == "__main__":

    reportOn = (perfer.userCpu, perfer.sysCpu, perfer.mem, perfer.blockin, perfer.blockout,)

    data = perferReader(sys.stdin)

    print("Individual processes using the most CPU")
    measures = topNmeasurements(data, perfer.userCpu)
    for measure in measures:
        print("%f:\t%s" % (measure[perfer.userCpu], measure[perfer.cmd]))
    print()
    for report in reportOn:
        print()
        topNaggregateMeasurements(data, report)