Python 3 min read

Automating the bench

Every insight so far comes together here: a script that reads a serial capture, splits each line into fields and writes a CSV you can chart in any spreadsheet. With argparse and pathlib it runs on any capture with one command.

The serial terminal saves capture.log. log2csv.py reads it, splits each line into time, temperature and voltage, and writes capture.csv, which opens in any spreadsheet for charting. — Capture once, parse with one script, chart in any spreadsheet.

Parse, then write

The firmware prints one line per reading: a timestamp, a temperature and the battery voltage. The csv module handles quoting and separators for you. Here the log is a string and the CSV goes to an in-memory buffer, so you can run it as it is:

to_csv.pyPython

import csvimport iolog = """00:01 T=23.45 V=3.3100:02 T=23.51 V=3.3000:03 T=23.62 V=3.12"""out = io.StringIO()writer = csv.writer(out, lineterminator="\n")writer.writerow(["time", "temp_c", "vbat"])for line in log.splitlines():    time, temp, volts = line.split()    writer.writerow([time, temp[2:], volts[2:]])print(out.getvalue(), end="")

Output

time,temp_c,vbat
00:01,23.45,3.31
00:02,23.51,3.30
00:03,23.62,3.12

A tool you can run on any capture

Turn it into a command-line tool. argparse reads the file name from the command line and writes --help for you; pathlib builds the output name from the input name. Run it as python log2csv.py capture.log.

log2csv.pyPython

import argparseimport csvfrom pathlib import Pathparser = argparse.ArgumentParser(description="Serial log to CSV")parser.add_argument("log", type=Path, help="capture from the serial terminal")args = parser.parse_args()target = args.log.with_suffix(".csv")with args.log.open() as src, target.open("w", newline="") as dst:    writer = csv.writer(dst)    writer.writerow(["time", "temp_c", "vbat"])    for line in src:        if line.strip():            time, temp, volts = line.split()            writer.writerow([time, temp[2:], volts[2:]])print("wrote", target)

Because the script names its output after its input, you can run it over a whole folder of captures, or call it after every automated test run. The pattern scales: read, parse, write, and keep each step small enough to try on a string first, as in the first example.