make_stats.py

import math
import os.path
from os import path
 
class ReadStats:
 
    def __init__(self):
        self.generations = []
        self.times = []
        self.evals = []
        self.mins = []
        self.maxs = []
        self.avgs = []
        self.stds = []
 
    def readLogFile(self, fileName):
        try:
            logFile = open(fileName, 'r')
        except IOError:
            print("Couldn't open the file ", fileName)
            quit()
 
        generation = []
        eval = []
        time = []
        max = []
        min = []
        avg = []
        std = []
        inGeneration = False
        readMax = readMin = readAvg = readStd = False
        readEval = readTime = False
 
        for line in logFile:
            items = [word for word in line.split()]
            if(len(items) == 0):
                continue
            if(items[0] == "Generation:"):
                generation.append(int(items[1]))
                inGeneration = True
            if(not readEval and items[0] == "Evaluations:"):
                eval.append(int(items[1]))
                readEval = True
            if(not readTime and items[0] == "Elapsed"):
                time.append(int(items[2]))
                readTime = True
            if(not readMax and items[0] == "max:"):
                max.append(float(items[1]))
                readMax = True
            if(not readMin and items[0] == "min:"):
                min.append(float(items[1]))
                readMin = True
            if(not readAvg and items[0] == "avg:"):
                avg.append(float(items[1]))
                readAvg = True
            if(not readStd and items[0] == "stdev:"):
                std.append(float(items[1]))
                readStd = True
            if(inGeneration and readMax and readMin and readAvg and readStd):
                inGeneration = readMax = readMin = readAvg = readStd = False
                readEval = readTime = False
 
        logFile.close()
        self.generations.append(generation)
        self.times.append(time)
        self.evals.append(eval)
        self.mins.append(min)
        self.maxs.append(max)
        self.avgs.append(avg)
        self.stds.append(std)
#       print("eval:", eval)
#       print("time:", time)
#       print("max:", max)
#       print("min:", min)
#       print(len(self.evals))
 
        
    def getPartialLog(self, percentage):
        stats = []
        for i in range(len(self.generations)):
            logNo = int((len(self.generations[i]) - 1) * percentage);
            stat = []
            stat.append(self.mins[i][logNo])
            stat.append(self.maxs[i][logNo]);
            stat.append(self.avgs[i][logNo]);
            stat.append(self.generations[i][logNo]);
            stat.append(self.evals[i][logNo]);
            stat.append(self.times[i][logNo]);
            stats.append(stat);
        return stats;
 
 
    def getStats(self):
        stats = []
        for i in range(len(self.generations)):
            stat = []
            stat.append(min(self.mins[i]))
            stat.append(max(self.maxs[i]));
            stat.append(self.avgs[i][-1]);
            stat.append(self.stds[i][-1]);
            stat.append(self.evals[i][-1]);
            stat.append(self.times[i][-1]);
            stat.append(self.generations[i][-1]);
            stats.append(stat);
        return stats;
 
 
    def getMean(self, values):
        sum = float(0)
        for i in range(len(values)):
            sum += values[i]
        return sum / len(values)
 
 
    def getStdDev(self, values):
        if len(values) == 0:
            return 0
        numerator = float(0)
        average = self.getMean(values)
        for i in range(len(values)):
            numerator += ((values[i] - average) * (values[i] - average));
        denominator = (len(values) - 1)
        return math.sqrt(numerator / denominator)
 
 
    def getMin(self, values):
        min = values[0]
        for i in range(len(values)):
            if min > values[i]:
                min = values[i]
        return min
 
 
    def getMax(self, values):
        max = values[0]
        for i in range(len(values)):
            if max < values[i]:
                max = values[i]
        return max
 
 
    def getMedian(self, values):
        values.sort()
        if len(values) % 2 == 0:
            return (values[int(len(values) / 2) - 1] + values[int(len(values) / 2)]) / 2
        else:
            return values[int(len(values) / 2)]
 
 
def printStr(FloatNumber, Precision):
    return "%0.*f" % (Precision, FloatNumber)
 
 
 
 
# main
if __name__ == "__main__":
 
    # default values
    logBase = "log_";
    logExt = ".txt";
    outFile = "stats.txt";
    logStartNum = 1;
    logEndNum = 0;
    nDigits = 2;
 
    line = input(" -- Log filename base (enter - \"" + logBase + "\") : ")
    if line != "":
        logBase = line
 
    line = input(" -- Starting log number (enter - " + str(logStartNum) + ") : ")
    if line != "":
        logStartNum = int(line)
 
    line = input(" -- End log number (enter - read all): ")
    if line != "":
        logEndNum = int(line)
 
    line = input(" -- Log filename num of digits (enter - 2) : ")
    if line != "":
        nDigits = int(line)
 
    line = input(" -- Extension (enter - \"" + logExt + "\") : ")
    if line != "":
        logExt = line
 
    line = input(" -- Output file (enter - \"" + outFile + "\") : ")
    if line != "":
        outFile = line
 
 
    logNum = logStartNum
    rs = ReadStats()
 
    while True:
        fileName = ""
        fileName += logBase
        fileName += str(logNum).zfill(nDigits) 
        fileName += logExt
 
        print("Log filename: ", fileName)
        opened = False
        if path.exists(fileName): 
            print("Found!")
            opened = True
            rs.readLogFile(fileName)
        else:
            print("File not found!")
 
        logNum += 1
        cont = False
        if logEndNum >= logNum or opened == True:
            cont = True
        if not cont:
            break
 
    stats = rs.getStats()
    nRuns = len(stats)
    #print(stats)
 
    file = open(outFile, "w")
    file.write("runId\tfit_min\tfit_max\tfit_avg\tfit_std\t#evals\ttime\tgen\n")
    for sample in range(nRuns):
        output = str(sample + 1) + "\t"
        for value in stats[sample]:
            output += str(value) + "\t"
        output += "\n"
        file.write(output)
    file.close()