Utility to aggregate time series (typically, a log file) in CSV format

csvtsarg.py

import pendulum
import argparse
import csv

parser = argparse.ArgumentParser(description='Aggregate time series data')

parser.add_argument('filename', type=str,
                    help='a CSV file to process')
parser.add_argument('--column', type=str, default='time',
                    help='in which column is the datetime (default: time)')
parser.add_argument('--period', type=str, default='minute',
                    help='time period: second, hour, .. (default: minute)')
parser.add_argument('--sum', type=str, default=None,
                    help='optional column to sum-up (default: +1)')
parser.add_argument('--name', type=str, default='sum',
                    help='name of sum-up column (default: sum)')

args = parser.parse_args()

last_time = None
last_row = None
last_sum = 0

def to_num(val):
  if '.' in val:
    return float(val)
  return int(val)

def print_last_row():
  if last_row is None:
    return
  # Output with previous row counter values
  last_row[args.column] = last_time
  last_row[args.name] = last_sum
  print(','.join([str(v) for v in last_row.values()]))

with open(args.filename, newline='') as csvfile:
  doctorwho = csv.DictReader(csvfile)
  for row in doctorwho:
    # Get the date time column
    t = row[args.column]
    # Parse date and go back to start of selected period
    pt = pendulum.parse(t)
    ptt = pt.start_of(args.period)
    # First time through, output the header
    if last_time is None:
      last_time = ptt
      last_row = row
      print(','.join(row.keys()) + ',' + args.name)
    elif last_time == ptt:  
      # Append to row summation
      if args.sum is not None:
        last_sum = last_sum + to_num(row[args.sum])
      else:
        last_sum = last_sum + 1
      last_row = row
    else:
      print_last_row()
      last_time = ptt
      last_row = None
      last_sum = 0
  # Print the last row's data
  print_last_row()