Driver script for parametrically calculating Python scripts

What is this

This is a driver script created to measure the error of my own Timer library. This is a driver tool for calculating a certain python script under different conditions. I didn't know how to express it, but I imagined WebDriver and expressed it as a driver. It's like a driver script to drive a Python script.

Driver script

python

from timer import Timer

def main():
    seconds = 5
    interval_set = [0.01,0.025,0.05,0.1,0.15,0.2,0.25,0.5,0.75,1,2,2.5,3,3.5,4,4.5]

    num = len(interval_set) + 1

    for case in range(1,num):
        interval = interval_set[case-1]
        timer = Timer(seconds, interval)
        timer.test(seconds, interval, case)

if __name__ == '__main__':
    main()

When this is executed, test in timer.py is executed for each value specified in ʻinterval_setand the result is recorded. The first half oftimer.py` is reprinted, but it is as follows.

python

import time
import csv
import platform

class Timer:
    def __init__(self, seconds, interval):
        self.interval = interval
        self.seconds = seconds
        self.n = int((seconds - (seconds % interval))/interval)

    def timer(self, seconds,interval):
        time_start = time.perf_counter()
        time.sleep(self.interval*self.n)
        time_stop = time.perf_counter()
        self.remain = seconds - (time_stop - time_start)
        if self.remain > 0: #Is it here if you add correction?
            time.sleep(self.remain)
        else:
            pass
        self.real = time.perf_counter() - time_start
        self.error = self.real - seconds
        return False

    def test(self,seconds, interval, case):
        filename = f"./error_data{case}"+"_"+str(platform.system())+".csv" #Formatted string literal
        with open(filename, 'w') as f: #w is new
            f.write('seconds(sec),interval(sec),real(sec),error(msec)\n')
            f.close()
        for i in range(1000): #Setting of measurement times
            while self.timer(seconds, interval):
                pass
            else:
                print('loop'+str(case)+' '+str(i+1)+'Time'+'interval=' + str(interval) + 'sec Error:'+str(self.error*1000)+'msec')
                save_data = [self.seconds,self.interval,self.real, self.error*1000]
                with open(filename, 'a', newline='') as f:
                    writer = csv.writer(f, lineterminator='\r\n')
                    writer.writerow(save_data)

In this example, the error is measured about 1000 times for every 16 interval values. It's quite a waiting time, so it's almost like training to execute it once.

This kind of work is where computers come into play.

By executing main.py, the result will be recorded in ʻerror_data1_Darwin.csv`. I'm running it on a Mac, so it's Darwin. If you run it on Raspberry Pi, it will be Linux.

$ python main.py

Visualization

If you display a large number of graphs one by one, your heart will be broken, so the following script is used to graph them.

python

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

num_figure = 9    #Number of csv files you want to draw

fig = plt.figure(figsize=(10.0, 15.0))


for i in range(1, num_figure+1):
    df = pd.read_csv('./data/error_data{}_Darwin.csv'.format(i))    #read csv file
    df_error = df['error(sec)']   #Read Error column
    df_param = df['interval(sec)'] #Read Interval column
    df_sec = df['seconds(sec)'] #Read Seconds column
    
    ax = fig.add_subplot(5, 2, i)  #It draws in order with a 5 row 2 column arrangement.
    sns.distplot(df_error, kde=False, rug=False, bins=100) 
#    plt.hist(df_error,bins=100, alpha=0.4, histtype='stepfilled', color='g')
    ax.set_title("Case{}".format(i)+' ('+str(df_sec[1])+'sec Timer, Int.='+str(df_param[1])+'sec)')  #Give the graph a title
    ax.set_ylim(0,25) #Set y-axis range
    ax.set_xlim(-0.001,0.011) #Set x-axis range

plt.tight_layout() #Adjust graph spacing
plt.savefig('Error.png')
plt.show()

The graph will be drawn in no time. It will be reprinted just by changing the arrangement of the graph.

I would be grateful if you could comment, such as "This is better."