A script to calculate total time from a set of (start. end) tuples

timecalc.py

buf = """
15/10/1997  15/10/2001
03/03/1997  30/06/2000
01/10/1996  30/06/1997
15/11/2011  14/10/2012
15/11/2008  14/11/2011
10/10/2007  10/11/2008
01/12/2005  01/10/2007
01/11/2001  01/11/2005
"""

import datetime
import time

def intersect(periods):
  """
  Perform iterative interval intersection from the input list of intervals.
  """

  result = []
  last_period = None
  for period in periods:
    if not last_period:
      last_period = period
    else:
      if last_period['end'] < period['start']:
        result.append(last_period)
        last_period = period
      else:
        if last_period['end'] < period['end']:
          last_period['end'] = period['end']
        else:
          print "period included"

  result.append(last_period)
  return result

def summum(periods):
  """
  Calculate the sum (in years, months, days) from a list of periods.
  Note that it does carry over if e.g. the number of months exceeds 12
  when summing up.
  """
  result = {'years': 0, 'months':0, 'days':0}

  for period in periods:
    e = period['end']
    s = period['start']
    years = e.year - s.year
    months = e.month - s.month
    days = e.day - s.day

    # Get the days to be positive by substracting months
    #
    while days < 0:
      days += 30
      months -= 1

    # Get the years to be positive by substracting years
    #
    while months < 0:
      months += 12
      years -= 1

    result['years'] += years
    result['months'] += months
    result['days'] += days

    while result['days'] > 30:
      result['months'] += 1
      result['days'] -= 30

    while result['months'] > 12:
      result['years'] += 1
      result['months'] -= 12

  return result

def main():
  lines = [val.strip() for val in buf.split('\n')]

  periods = []

  # Load the lines in datetime.date objects.
  #
  for l in lines:
    if not l:
      continue

    start, end = l.split()
    print "Parsing:", start, end
    start = time.strptime(start,"%d/%m/%Y")
    end = time.strptime(end,"%d/%m/%Y")

    start = datetime.date(start.tm_year, start.tm_mon, start.tm_mday)
    end = datetime.date(end.tm_year, end.tm_mon, end.tm_mday)
    periods.append({'start': start, 'end': end})

  # Sort the time intervals based on their start values
  #
  periods = sorted(periods, key=lambda period: period['start'])

  # Calcualte intersections
  #
  new_periods = intersect(periods)

  # Print the result
  #
  print summum(new_periods)


if __name__ == "__main__":
  main()