diff --git a/README.md b/README.md
index ad15a59..ff87b6d 100644
--- a/README.md
+++ b/README.md
@@ -15,7 +15,7 @@ https://arxiv.org/abs/2001.10275. To have a look at that one could
 just run the data from their Extended Data Table 1, in the following
 way, to get a plot like that in period.png
 
-> python getper.py -i CHIME_seconds -p1 86400.0 -p2 8640000.0 -maxdiff 86400000.0 -pstep 8640.0 -days 1
+> python getper.py CHIME_seconds -p1 86400.0 -p2 8640000.0 -maxdiff 86400000.0 -pstep 8640.0 -days
 
 run with a -h to see what the options are.
 
diff --git a/getper.py b/getper.py
index 64c3f9e..f6894cf 100644
--- a/getper.py
+++ b/getper.py
@@ -1,85 +1,208 @@
-#!/usr/bin/python
+#
+# Determine periodicity by arrival time differencing.
+#
+# 2020 Fabian Jankowski: slightly reworked the code and formatting. It
+# works with python 2 and 3 now.
+#
 
-# Load some packages
-import numpy as np
-import scipy as sp
-from scipy.stats import skew
-import matplotlib.pylab as plt
 import argparse
+import os.path
 import sys
 
-# Parse command line arguments
-parser = argparse.ArgumentParser()
-parser.add_argument('-i', dest='input_file', help='set the input file name (default: file)', default="file")
-parser.add_argument('-p1', type=float, dest='p1', help='set the lowest period (default: 0.1 seconds)', default=0.100)
-parser.add_argument('-p2', type=float, dest='p2', help='set the lowest period (default: 10.0 seconds)', default=10.0)
-parser.add_argument('-days', dest='days', help='plot output in days rather than seconds (default: seconds)', default=False)
-parser.add_argument('-maxdiff', type=float, dest='maxdiff', help='maximum time difference to use (default: 3600.0 seconds)', default=3600.0)
-parser.add_argument('-mjd', dest='mjd', help='flag input times as MJD rather than seconds (default: false)', default=False)
-parser.add_argument('-pstep', type=float, dest='pstep', help='trial period step size in seconds (default: 0.001)', default=0.001)
-parser.add_argument('-tol', type=float, dest='tol', help='percentage tolerance in phase to count as a match (default: 10.0)', default=10.0)
-parser.add_argument('--version', action='version', version='%(prog)s 0.0.3')
-args = parser.parse_args()
-
-## Read in data
-# Read in barycentred times
-input_file = args.input_file
-print "Reading input from file:",input_file
-times = np.genfromtxt(input_file)
-times.sort()
-ntimes = np.size(times)
-print "Total number of pulse times:",ntimes
-print "Total number of unique time differences:",ntimes*(ntimes-1)/2
-# If input times in MJD, convert to seconds
-mjd = args.mjd
-if mjd:
-    times = (times-int(times[0]))*86400.0
-
-## Get the time differences
-maxdiff = args.maxdiff
-#pstep = args.pstep*0.001
-pstep = args.pstep
-diffs = np.zeros(ntimes*ntimes)
-k = 0
-for i in range(0,ntimes):
-    for j in range(0,ntimes):
-        diff = times[j] - times[i]
-        if ( np.abs(diff) <= maxdiff):
-            if ( diff < 0.0 ):
-                diffs[k] = - diff
-            else:
-                diffs[k] = diff
-            k = k+1
-diffs.sort()
-diffs_unique = np.unique(diffs)
-print "Total number of unique time differences less than ",maxdiff," seconds:",np.size(diffs_unique)
-print diffs_unique
-
-## Search period range
-p1 = args.p1
-p2 = args.p2
-#pstep = 0.001
-nsteps = int((p2 - p1)/pstep)
-print "Searching from",p1,"to",p2,"in",nsteps,"steps"
-
-# Number of matches within 10 percent
-phase_tol = args.tol*0.01
-periods = np.zeros(nsteps)
-matches = np.zeros(nsteps)
-for i in range(0,nsteps):
-    periods[i] = p1+i*pstep
-    for j in range(0,np.size(diffs_unique)):
-        phase = (diffs_unique[j]/periods[i]) - int(diffs_unique[j]/periods[i])
-        if ( (phase < 0.5*phase_tol) or (phase > (1.0-0.5*phase_tol))):
-            matches[i] = matches[i] + 1
-
-## Plot the matches
-plt.ylabel("Number of matches")
-days=args.days
-if args.days:
-    plt.xlabel("Trial Period (days)")
-    plt.plot(periods/86400.0,matches)
-else:
-    plt.xlabel("Trial Period (s)")
-    plt.plot(periods,matches)
-plt.show()
+import matplotlib.pylab as plt
+import numpy as np
+
+
+def parse_args():
+    """
+    Parse the commandline arguments.
+
+    Returns
+    -------
+    args: populated namespace
+        The commandline arguments.
+    """
+
+    parser = argparse.ArgumentParser(
+        description='Determine period by arrival time differencing.',
+        formatter_class=argparse.ArgumentDefaultsHelpFormatter
+    )
+
+    parser.add_argument(
+        'filename',
+        help='input file name'
+    )
+
+    parser.add_argument(
+        '-p1',
+        type=float,
+        dest='p1',
+        help='set the lowest trial period in seconds',
+        default=0.100
+    )
+
+    parser.add_argument(
+        '-p2',
+        type=float,
+        dest='p2',
+        help='set the highest trial period in seconds',
+        default=10.0
+    )
+
+    parser.add_argument(
+        '-days',
+        dest='days',
+        action='store_true',
+        help='plot output in days rather than seconds',
+        default=False
+    )
+
+    parser.add_argument(
+        '-maxdiff',
+        type=float,
+        dest='maxdiff',
+        help='maximum time difference in seconds to use',
+        default=3600.0
+    )
+
+    parser.add_argument(
+        '-mjd',
+        dest='mjd',
+        action='store_true',
+        help='flag input times as MJD rather than seconds',
+        default=False
+    )
+
+    parser.add_argument(
+        '-pstep',
+        type=float,
+        dest='pstep',
+        help='trial period step size in seconds',
+        default=0.001
+    )
+
+    parser.add_argument(
+        '-tol',
+        type=float,
+        dest='tol',
+        help='percentage tolerance in phase to count as a match',
+        default=10.0
+    )
+
+    parser.add_argument(
+        '--version',
+        action='version',
+        version='0.0.3'
+    )
+
+    args = parser.parse_args()
+
+    return args
+
+
+def plot_matches(periods, matches, args):
+    """
+    Plot the matches versus trial period.
+
+    Parameters
+    ----------
+    periods: ~np.array
+        The trial periods.
+    matches: ~np.array
+        The number of matches corresponding to the trial periods.
+    args: populated namespace
+        The commandline arguments.
+    """
+
+    fig = plt.figure()
+    ax = fig.add_subplot(111)
+
+    if args.days:
+        ax.plot(periods / 86400.0, matches)
+        ax.set_xlabel("Trial Period (days)")
+    else:
+        ax.plot(periods, matches)
+        ax.set_xlabel("Trial Period (s)")
+
+    ax.grid()
+    ax.set_ylabel("Number of matches")
+
+
+#
+# MAIN
+#
+
+def main():
+    args = parse_args()
+
+    ## Read in data
+    # Read in barycentred times
+    if not os.path.isfile(args.filename):
+        print('The file does not exist: {0}'.format(args.filename))
+        sys.exit(1)
+
+    print("Reading input from file: {}".format(args.filename))
+    times = np.genfromtxt(args.filename)
+    times = np.sort(times)
+    ntimes = np.size(times)
+    print("Total number of pulse times: {0}".format(ntimes))
+    print("Total number of unique time differences: {0}".format(ntimes * (ntimes - 1) / 2.0))
+
+    # If input times in MJD, convert to seconds
+    if args.mjd:
+        times = (times - int(np.floor(times[0]))) * 86400.0
+
+    ## Get the time differences
+    maxdiff = args.maxdiff
+    diffs = np.zeros(ntimes * ntimes)
+    k = 0
+    for i in range(0, ntimes):
+        for j in range(0, ntimes):
+            diff = times[j] - times[i]
+            if np.abs(diff) <= maxdiff:
+                if diff < 0.0:
+                    diffs[k] = - diff
+                else:
+                    diffs[k] = diff
+                k = k+1
+
+    diffs = np.sort(diffs)
+    diffs_unique = np.unique(diffs)
+    print("Total number of unique time differences less than {0} seconds: {1}".format(
+        maxdiff,
+        np.size(diffs_unique)
+        )
+    )
+    print(diffs_unique)
+
+    ## Search period range
+    p1 = args.p1
+    p2 = args.p2
+    pstep = args.pstep
+    nsteps = int((p2 - p1) / pstep)
+    print("Searching from {0} to {1} in {2} steps".format(
+        p1,
+        p2,
+        nsteps
+        )
+    )
+
+    # Number of matches within X percent
+    phase_tol = args.tol * 0.01
+    periods = np.zeros(nsteps)
+    matches = np.zeros(nsteps)
+
+    for i in range(0, nsteps):
+        periods[i] = p1 + i * pstep
+        for j in range(0, np.size(diffs_unique)):
+            phase = (diffs_unique[j] / periods[i]) - int(diffs_unique[j] / periods[i])
+            if ( (phase < 0.5 * phase_tol) or (phase > (1.0 - 0.5 * phase_tol))):
+                matches[i] = matches[i] + 1
+
+    plot_matches(periods, matches, args)
+
+    plt.show()
+
+
+if __name__ == '__main__':
+    main()