#!python
from sys import argv
lines = open(argv[1]).readline()
lines = lines.split("\r")

f = open("process1.txt","w")
f.write("id,xc,st,dh,ul,numRaces,cat,catnum,raffles,seriesPts,distOfRace,distFromHome,family,racesPerMonth,age"+"\r\n")

catconv = { 'BeginnerMale'      :0,
            'BeginnerFemale'    :1,
            'SportMale'         :2,
            'SportFemale'       :3,
            'ExpertMale'        :4,
            'ExpertFemale'      :5,
            'Semipro(maleonly)' :6,
            'ProMale'           :7,
            'ProFemale'         :8  }
            
# category averages
Lnumraces = [] 
Lraffles = []
Lseriespts = []
Ldistofrace = []
Ldistfromhome = []
Lfamily = []
Lrpm = []

for L in Lnumraces, Lraffles, Lseriespts, Ldistofrace, Ldistfromhome, Lfamily, Lrpm:
    for i in range(9):
        L.append( [] )


for line in lines[1:]:
    line = line.split(",")
    
    id = line[0]
    
    if line[4] != "": xc = 1     #Cross Country
    else: xc = 0      

    if line[5] != "": st = 1    #Short Track
    else: st = 0
      
    if line[6] != "": dh = 1    #Downhill
    else: dh = 0
      
    if line[7] != "": ul = 1    #Ultra
    else: ul = 0
      
    numraces = int( "".join(line[8:14]) )
    cat = "".join(line[14:23])
    cat = cat.replace(" ","") #remove spaces
    cat = cat.replace("-","") #remove dash
    catnum = catconv[cat]
    
    raffles = int( "0"+"".join(line[32:37]) )
    seriespts = int( "0"+"".join(line[37:42]) )
    distofrace = int( "0"+"".join(line[42:47]) )
    distfromhome = int( "0"+"".join(line[47:52]) )
    family = int( "0"+"".join(line[52:57]) )
    
    months = []
    month_idxs = range(57,69)
    for month_idx in month_idxs:
      if len( line[month_idx] ) > 0:
        months.append(1) #likes this month to race
      else:
        months.append(0) #doesn't like this month to race
        
    rpm = int( "0"+"".join(line[69:74]) ) #races per month
    
    tor = []
    tor_idxs = range(74,81) #tor = TypesOfRacing
    for tor_idx in tor_idxs:
        if len( line[tor_idx] ) > 0:
            tor.append(1)
        else:
            tor.append(0)
        
    age = "".join(line[81:91])
    
    #### write summary file
    s = ""
    for elem in id,xc,st,dh,ul,numraces,cat,catnum,raffles,seriespts,distofrace,distfromhome,family,rpm,age:
        s += str(elem)+","
    # add TOR and MONTHS here
    s = s[:-1] #chop trailing comma off
    f.write(s+"\r\n")
    
    #debug
    if catnum ==8:
        print s
    
    ### race by cat
    Lnumraces[catnum].append( numraces )
    Lraffles[catnum].append( raffles )
    Lseriespts[catnum].append( seriespts )
    Ldistofrace[catnum].append( distofrace )
    Ldistfromhome[catnum].append( distfromhome )
    Lfamily[catnum].append( family )

    Lrpm[catnum].append( rpm )
    

def aL(l): return 1.*sum(l)/len(l)

print "catNum,sizeCat,numRaces,raffles,seriesPts,distOfRace,distFromHome,family,racesPerMonth"
for catnum in range(len(Lnumraces)):
    print catnum,",",len(Lnumraces[catnum]),",",
    for L in  Lnumraces, Lraffles, Lseriespts, Ldistofrace, Ldistfromhome, Lfamily, Lrpm:
        print aL( L[catnum] ),",",
    print