average_multiple_fio.py

#!/usr/bin/python

import sys
import csv
debug=False

stats={}
files=sys.argv

for csvFile in files:
        fh = open(csvFile)
        csv_fh = csv.reader(fh)
        rowCount=0
        for row in csv_fh:
                if len(row)>0:
                        if rowCount==0:
                                headers={}
                                for i in range(3,len(row)):
                                        headers[row[i]]=i
                        else:
                                for col in headers.keys():
                                        if debug: print str(rowCount)+" "+row[0]
                                        if row[0] not in stats:
                                                if debug: print "Creating stats["+row[0]+"]"
                                                stats[row[0]]={}
                                        if row[1] not in stats[row[0]]:
                                                if debug: print "Creating stats["+row[0]+"]["+row[1]+"]"
                                                stats[row[0]][row[1]]={};
                                        if row[2] not in stats[row[0]][row[1]]:
                                                if debug: print "Creating stats["+row[0]+"]["+row[1]+"]["+row[2]+"]"
                                                stats[row[0]][row[1]][row[2]]={}
                                        if col not in stats[row[0]][row[1]][row[2]]:
                                                if debug: print "Creating stats["+row[0]+"]["+row[1]+"]["+row[2]+"]["+col+"]"
                                                stats[row[0]][row[1]][row[2]][col]=[]
                                        stats[row[0]][row[1]][row[2]][col].append(float(row[headers[col]]))
                rowCount+=1

headerLine="host,osd"
outLineCount=0
for host in stats.keys():
        for osd in stats[host].keys():
                outline=host+","+osd
                for opsize in stats[host][osd]:
                        headerLine+=","+opsize+" "+",{opsize} ".format(opsize=opsize).join(x for x in headers.keys())
                        for col in headers.keys():
                                avg=float(sum(stats[host][osd][opsize][col]))/float(len(stats[host][osd][opsize][col]))
                                outline+=",{0:0.2f}".format(avg)
                if outLineCount==0:
                        print headerLine
                print outline
                outLineCount+=1