Add scripts to compute the cyclomatic increase

config.sh

@@ -9,3 +9,4 @@ function_idx_dir=$workdir/deduplicated-functions-idx-no-goto
 revng_json_dir=$workdir/revng-json
 ida_json_dir=$workdir/ida-json-normalized
 revng_metrics_dir=$workdir/revng-metrics
+cyclomatic_metrics_dir=$workdir/cyclomatic-metrics

scripting/computation/compute-cyclomatic-total.py

+#!/usr/bin/env python3
+
+import argparse
+import csv
+import json
+import numpy
+import os
+import pandas
+import sys
+
+from scipy.stats.mstats import gmean
+
+def main():
+  parser = argparse.ArgumentParser(description='My nice tool.')
+  parser.add_argument('binaries', metavar='BINARIESFILE', help='File containing the names of the binaries')
+  parser.add_argument('arch', metavar='ARCH', help='Architecture')
+  parser.add_argument('cyclomatic_metrics_dir', metavar='CYCLOMATICMETRICSDIR', help='Architecture')
+  args = parser.parse_args()
+  arch = args.arch
+  binaries_file = args.binaries
+  cyclomatic_metrics_dir = args.cyclomatic_metrics_dir
+
+  source_cyclomatic_total = 0
+  revng_cyclomatic_total = 0
+  ida_cyclomatic_total = 0
+  ghidra_cyclomatic_total = 0
+
+  with open(binaries_file, 'r') as binaries:
+    binaries_list = [line.rstrip('\n') for line in binaries]
+
+  for binary_name in binaries_list:
+
+    # Open the files containing the metrics.
+    metrics_csv_file = cyclomatic_metrics_dir + '/' + binary_name + '.csv'
+    if os.path.isfile(metrics_csv_file):
+      with open(metrics_csv_file, 'r') as metrics_csv:
+        df = pandas.read_csv(metrics_csv_file)
+        for index, row in df.iterrows():
+
+          # Load the metrics.
+          source_current_cyclomatic = int(row['source_cyclomatic'])
+          revng_current_cyclomatic = int(row['revng_cyclomatic'])
+          ida_current_cyclomatic = int(row['ida_cyclomatic'])
+          ghidra_current_cyclomatic = int(row['ghidra_cyclomatic'])
+
+          # For some functions, we couldn't get the original cyclomatic
+          # complexity for the source.
+          if source_current_cyclomatic != -1:
+
+            # Unfortunately, due to the `-2` term in the cyclomatic complexity
+            # computation, some terms goes to 0. A 0 at the denominator is not
+            # good, round it to 1.
+            if source_current_cyclomatic == 0:
+              source_current_cyclomatic = 1
+            if revng_current_cyclomatic == 0:
+              revng_current_cyclomatic = 1
+            if ida_current_cyclomatic == 0:
+              ida_current_cyclomatic = 1
+            if ghidra_current_cyclomatic == 0:
+              ghidra_current_cyclomatic = 1
+
+            # Total.
+            source_cyclomatic_total += source_current_cyclomatic
+            revng_cyclomatic_total += revng_current_cyclomatic
+            ida_cyclomatic_total += ida_current_cyclomatic
+            ghidra_cyclomatic_total += ghidra_current_cyclomatic
+
+  # Print some final information about the cyclomatic complexity.
+  print("Total source cyclomatic")
+  print(source_cyclomatic_total)
+  print("Total revng cyclomatic")
+  print(revng_cyclomatic_total)
+  print("Total ida cyclomatic")
+  print(ida_cyclomatic_total)
+  print("Total ghidra cyclomatic")
+  print(ghidra_cyclomatic_total)
+
+  print("revng/source")
+  print(revng_cyclomatic_total/source_cyclomatic_total)
+
+  print("ida/source")
+  print(ida_cyclomatic_total/source_cyclomatic_total)
+
+  print("ghidra/source")
+  print(ghidra_cyclomatic_total/source_cyclomatic_total)
+
+if __name__ == "__main__":
+  main()

scripting/computation/compute-cyclomatic-total.sh

+#!/bin/bash
+
+if [ $# -eq 0 ]; then
+  echo "No arguments supplied"
+  exit 1
+fi
+
+arch=$1
+
+# Import the config
+. ./config.sh
+
+$computation_script_dir/compute-cyclomatic-total.py $binaries_file $arch $cyclomatic_metrics_dir