updated runtime script for --All and --format functions and got rid of bugs

tools/plotting/fastnnlo_runtime.py

@@ -9,6 +9,7 @@ import matplotlib as mpl
 import matplotlib.gridspec as gridspec
 import matplotlib.lines as mlines
 import matplotlib.patches as mpatches
+import matplotlib.ticker as mticker
 from matplotlib.ticker import (FormatStrFormatter, LogFormatter, NullFormatter, ScalarFormatter, AutoMinorLocator, MultipleLocator)
 from matplotlib import cm
 # We do not want any interactive plotting! Figures are saved to files instead.
@@ -61,6 +62,8 @@ def main():
     # extract correct paths for input and outputfiles
     logfiles = get_files(args['logfiles'])
     outputpath = args['output']
+    all_plot = args['All']
+    fmt = args['format']
 
     # get all the information from logfiles as dict
     # dict contains: runtime, runtime_unit, channel, events
@@ -68,12 +71,12 @@ def main():
 
     # plot all the information
     if args['CPUtime']:
-        plot_elapsed_time(loginformation, outputpath)
+        plot_elapsed_time(loginformation, outputpath, all_plot, fmt)
     if args['Events']:
-        plot_events_per_hour(loginformation, outputpath)
+        plot_events_per_hour(loginformation, outputpath, all_plot, fmt)
     if not args['CPUtime'] and not args['Events']:
-        plot_elapsed_time(loginformation, outputpath)
-        plot_events_per_hour(loginformation, outputpath)
+        plot_elapsed_time(loginformation, outputpath, all_plot, fmt)
+        plot_events_per_hour(loginformation, outputpath, all_plot, fmt)
 
     exit(0)
 
@@ -86,9 +89,16 @@ def arguments():
     parser.add_argument('logfiles', nargs='+', type=str, help='Either input is a single string from LAW Task or it is a list')
 
     # optional arguments
-    parser.add_argument('-o', '--output', nargs=1, type=str, default='./', help='Set here the outputpath')
-    parser.add_argument('--CPUtime', dest='CPUtime', action='store_true', help='Plot only the elapsed time')
-    parser.add_argument('--Events', dest='Events', action='store_true', help='Plot only the events per hour')
+    parser.add_argument('-o', '--output', nargs=1, type=str, default='./',
+                        help='Set here the outputpath')
+    parser.add_argument('--CPUtime', dest='CPUtime', action='store_true',
+                        help='Plot only the elapsed time')
+    parser.add_argument('--Events', dest='Events', action='store_true',
+                        help='Plot only the events per hour')
+    parser.add_argument('--All', dest='All', action='store_true',
+                        help='Set plotoptions according to plot for all channels')
+    parser.add_argument('--format', required=False, nargs=1, type=str, default='png',
+                        help='Comma-separated list of plot formats to use: eps, pdf, png, svg. If nothing is chosen, png is used.')
 
     return vars(parser.parse_args())
 
@@ -111,13 +121,24 @@ def get_loginformation(files):
 
     for file in files:
         event = False
+        run_time_temp = []
+
         with open(file) as origin:
             for line in origin:
                 # extract elapsed time with time unit
-                if 'Elapsed time' in line:
-                    line = line.split()
-                    run_time.append(float(line[3]))
-                    unit = line[4]
+                if 'Time elapsed' in line:
+                    line = line.split(':')
+                    hours = float(line[1])
+                    minutes = float(line[2])
+                    seconds = float(line[3])
+
+                    if hours == True:
+                        run_time_temp.append(hours + minutes/60 + seconds/360)
+                        unit = 'hours'
+                    else:
+                        run_time_temp.append(minutes + seconds/60)
+                        unit = 'minutes'
+
                 # extract channel name
                 if 'Tablename' in line and not channel:
                     line = line.split()
@@ -129,30 +150,32 @@ def get_loginformation(files):
                     number_events.append(float(line[4][10:]))
                     event = True
 
-    # found bug in logfiles where elapsed time is negativ
-    # temporary fix
+        run_time.append(run_time_temp[-1])
+
+
     run_time = np.array(run_time)
     number_events = np.array(number_events)
-    run_time_arr = run_time[run_time > 0]
-    number_events_arr = number_events[run_time > 0]
-
 
     information = {
-        'runtime': run_time_arr,
+        'runtime': run_time,
         'runtime_unit': unit,
         'channel': channel,
-        'events': number_events_arr
+        'events': number_events
     }
 
     return information
 
 
-def plot_elapsed_time(informationdict, out_path):
+def plot_elapsed_time(informationdict, out_path, plot_all=False, format='png'):
 
     time = informationdict['runtime']
     unit = informationdict['runtime_unit']
     channel = informationdict['channel']
 
+    if plot_all:
+        channel = channel.split('.')
+        channel = channel[0]
+
     # get relevant values
     mean = np.mean(time)
     std = np.std(time)
@@ -163,7 +186,7 @@ def plot_elapsed_time(informationdict, out_path):
 
     # set saving location
     filename = out_path[0] + ('' if out_path[0][-1] == '/' else '/')
-    filename += channel + '.Hist_Elapsed_time.png'
+    filename += channel + '.Hist_Elapsed_time.' + format
 
     # set figure
     fig = plt.figure(figsize=(16, 12))
@@ -172,14 +195,15 @@ def plot_elapsed_time(informationdict, out_path):
     # plot histogram
     n, batches, _ = ax.hist(time, bins=20, color='deepskyblue', edgecolor='black', label='Total CPU time: {0:0.0f} hours'.format(CPUtime))
 
-    # plot mean and median
-    ax.vlines(mean, 0, max(n), colors='red', linestyles='dashed', label=r'Mean: {0:0.1f}$\pm${2:0.1f} {1}'.format(mean, unit, std))
-    ax.vlines(median, 0, max(n), colors='green', linestyles='dashed', label=r'Median: {0:0.1f}$\pm${2:0.1f} {1}'.format(median, unit, iqd))
+    if not plot_all:
+        # plot mean and median
+        ax.vlines(mean, 0, max(n), colors='red', linestyles='dashed', label=r'Mean: {0:0.1f}$\pm${2:0.1f} {1}'.format(mean, unit, std))
+        ax.vlines(median, 0, max(n), colors='green', linestyles='dashed', label=r'Median: {0:0.1f}$\pm${2:0.1f} {1}'.format(median, unit, iqd))
 
     # finish and save figure
     ax.set_title('Elapsed time of ' + channel + ' production', fontsize=20)
-    ax.set_xlabel('CPU time [' + unit + ']', horizontalalignment='right', x=1.0, verticalalignment='top', y=1.0, fontsize=20)
-    ax.set_ylabel('frequency', horizontalalignment='right', x=1.0, verticalalignment='top', y=1.0, fontsize=20)
+    ax.set_xlabel('CPU time [' + unit + ']', horizontalalignment='right', x=1.0, verticalalignment='top', y=1.0, fontsize=20, labelpad=15)
+    ax.set_ylabel('frequency', horizontalalignment='right', x=1.0, verticalalignment='top', y=1.0, fontsize=20, labelpad=15)
     ax.set_yscale('log')
     ax.tick_params(axis='both', which='major', labelsize=20)
 
@@ -189,7 +213,7 @@ def plot_elapsed_time(informationdict, out_path):
 
     fig.savefig(filename)
 
-def plot_events_per_hour(informationdict, out_path):
+def plot_events_per_hour(informationdict, out_path, plot_all=False, format='png'):
 
     time = informationdict['runtime']
     unit = informationdict['runtime_unit']
@@ -201,6 +225,10 @@ def plot_events_per_hour(informationdict, out_path):
     else:
         eph = events/(time/60)
 
+    if plot_all:
+        channel = channel.split('.')
+        channel = channel[0]
+
     # get relevant values
     mean = np.mean(eph)
     std = np.std(eph)
@@ -212,7 +240,7 @@ def plot_events_per_hour(informationdict, out_path):
 
     # set saving location
     filename = out_path[0] + ('' if out_path[0][-1] == '/' else '/')
-    filename += channel + '.Hist_Events_per_hour.png'
+    filename += channel + '.Hist_Events_per_hour.' + format
 
     # set figure
     fig = plt.figure(figsize=(16, 12))
@@ -221,14 +249,20 @@ def plot_events_per_hour(informationdict, out_path):
     # plot histogram
     n, batches, _ = ax.hist(eph, bins=20, color='deepskyblue', edgecolor='black', label='Total CPU time: {0:0.0f} hours'.format(CPUtime))
 
-    # plot mean and median
-    ax.vlines(mean, 0, max(n), colors='red', linestyles='dashed', label=r'Mean: {0:0.1e}$\pm${1:0.1e} events/hour'.format(mean, std))
-    ax.vlines(median, 0, max(n), colors='green', linestyles='dashed', label=r'Median: {0:0.2e}$\pm${1:0.2e} events/hour'.format(median, iqd))
+    if not plot_all:
+        # scientific format
+        f = mticker.ScalarFormatter(useOffset=False, useMathText=True)
+        g = lambda x,pos : "${}$".format(f._formatSciNotation('%0.2e' % x))
+        sci = mticker.FuncFormatter(g)
+
+        # plot mean and median
+        ax.vlines(mean, 0, max(n), colors='red', linestyles='dashed', label=r'Mean: {0}$\pm${1} events/hour'.format(sci(mean), sci(std)))
+        ax.vlines(median, 0, max(n), colors='green', linestyles='dashed', label=r'Median: {0}$\pm${1} events/hour'.format(sci(median), sci(iqd)))
 
     # finish and save figure
     ax.set_title('Events per hour of ' + channel + ' production', fontsize=20)
-    ax.set_xlabel('events/hour', horizontalalignment='right', x=1.0, verticalalignment='top', y=1.0, fontsize=20)
-    ax.set_ylabel('frequency', horizontalalignment='right', x=1.0, verticalalignment='top', y=1.0, fontsize=20)
+    ax.set_xlabel('events/hour', horizontalalignment='right', x=1.0, verticalalignment='top', y=1.0, fontsize=20, labelpad=15)
+    ax.set_ylabel('frequency', horizontalalignment='right', x=1.0, verticalalignment='top', y=1.0, fontsize=20, labelpad=15)
     ax.set_yscale('log')
     ax.tick_params(axis='both', which='major', labelsize=20)