Have parse_logs compute stats from multiple runs

docker/parse_logs

@@ -9,6 +9,10 @@ my $mode = '';
 my %res = ();
 my $offlinebytes = 0;
 my $netsetup = '';
+my %online_s_data = ();
+my %online_kib_data = ();
+my %total_s_data = ();
+my %total_kib_data = ();
 
 while(<>) {
     if (/Network setup: (.*)/) {
@@ -41,26 +45,105 @@ while(<>) {
         $res{$size}->{'totbytes'} += $totbytes;
     }
     if (/===== End /) {
+        my $label = "CircuitORAM read $size $netsetup$iters";
         my $online = &max3($res{$size}->{'COnline Time'},
             $res{$size}->{'DOnline Time'},
             $res{$size}->{'EOnline Time'});
         my $total = &max3($res{$size}->{'CETE'},
             $res{$size}->{'DETE'},
             $res{$size}->{'EETE'});
-        print "CircuitORAMOnln read $size $netsetup$iters $online s\n";
-        print "CircuitORAMTotl read $size $netsetup$iters $total s\n";
+        &accum_data(\%online_s_data, $label, $online);
+        &accum_data(\%total_s_data, $label, $total);
         # Note that the java code reports bandwidth *per iteration*
         my $onlinekib = $res{$size}->{'onlbytes'} * $iters / 1024;
         my $totalkib = $res{$size}->{'totbytes'} * $iters / 1024;
-        print "CircuitORAMOnln read $size $netsetup$iters $onlinekib KiB\n";
-        print "CircuitORAMTotl read $size $netsetup$iters $totalkib KiB\n";
+        &accum_data(\%online_kib_data, $label, $onlinekib);
+        &accum_data(\%total_kib_data, $label, $totalkib);
         undef $res{$size};
     }
 }
 
+# Convert the data (in the form [n, sum, sum_squares]) to statistics (in
+# the form [mean, variance])
+my %online_s_stats = ();
+my %online_kib_stats = ();
+my %total_s_stats = ();
+my %total_kib_stats = ();
+&statsify(\%online_s_stats, \%online_s_data);
+&statsify(\%online_kib_stats, \%online_kib_data);
+&statsify(\%total_s_stats, \%total_s_data);
+&statsify(\%total_kib_stats, \%total_kib_data);
+
+# Output the data
+&output_stats(\%online_s_stats, "Onln", "s");
+&output_stats(\%online_kib_stats, "Onln", "KiB");
+&output_stats(\%total_s_stats, "Totl", "s");
+&output_stats(\%total_kib_stats, "Totl", "KiB");
+
+# Subroutines
+
 sub max3 {
     my $m = $_[0];
     $m = $_[1] if $_[1] > $m;
     $m = $_[2] if $_[2] > $m;
     $m;
 }
+
+# Pass:
+# - a reference to a dictionary
+# - the key into that dictionary
+# - the new data point
+# Data is stored in the dictionary as a triple (n, sum, sum_squares)
+sub accum_data {
+    my ($dict, $key, $data) = @_;
+    $dict->{$key} = [0, 0, 0] unless defined $dict->{$key};
+    $dict->{$key}->[0] += 1;
+    $dict->{$key}->[1] += $data;
+    $dict->{$key}->[2] += ($data * $data);
+}
+
+# Convert data (in the form [n, sum, sum_squares]) to statistics (in
+# the form [mean, variance])
+sub statsify {
+    my ($sdict, $ddict) = @_;
+    my $key;
+    foreach $key (keys %$ddict) {
+        my $data = $ddict->{$key};
+        my $n = $data->[0];
+        my $sum = $data->[1];
+        my $sumsq = $data->[2];
+        if ($n == 0) {
+            $sdict->{$key} = [undef, undef];
+        } elsif ($n == 1) {
+            $sdict->{$key} = [$sum, undef];
+        } else {
+            $sdict->{$key} = [$sum/$n, ($sumsq - ($sum*$sum/$n))/($n-1)];
+        }
+    }
+}
+
+# Turn a stat array [mean, variance] into a string to display
+sub statstr {
+    my $data = $_[0];
+    if (defined $data->[1]) {
+        my $mean = $data->[0];
+        my $stddev = $data->[1] > 0 ? sqrt($data->[1]) : 0;
+        return "$mean ± $stddev";
+    } elsif (defined $data->[0]) {
+        return $data->[0];
+    } else {
+        return "none"
+    }
+}
+
+# Output the stats in the given dictionary. Append $phase to the
+# protocol name, and add $units to the end.
+sub output_stats {
+    my ($dict, $phase, $units) = @_;
+    my $label;
+    foreach $label (sort keys %$dict) {
+        my $printlabel = $label;
+        $printlabel =~ s/CircuitORAM/CircuitORAM$phase/;
+        print $printlabel, " ", &statstr($dict->{$label}), " $units\n";
+    }
+}