add a diff-script

5 files changed, 92 insertions, 0 deletions

diff --git a/classification/tools/create_diff.py b/classification/tools/create_diff.py
new file mode 100755
index 000000000..799c26006
--- /dev/null
+++ b/classification/tools/create_diff.py
@@ -0,0 +1,50 @@
+from os import path, walk
+from pathlib import Path
+from argparse import ArgumentParser
+
+root_directory = "../../results/classifier/"
+
+parser = ArgumentParser(prog='create_diff.py')
+parser.add_argument('old')
+parser.add_argument('new')
+args = parser.parse_args()
+
+def find_changes(old_directory, new_directory):
+    old_files = {}
+    for root, dirs, files in walk(old_directory):
+        for file in files:
+            relative_path = Path(root).relative_to(old_directory)
+            old_files[file] = str(relative_path)
+
+    new_files = {}
+    for root, dirs, files in walk(new_directory):
+        for file in files:
+            relative_path = Path(root).relative_to(new_directory)
+            new_files[file] = str(relative_path)
+
+    changed_files = []
+    for file in old_files:
+        if file in new_files and old_files[file] != new_files[file]:
+            changed_files.append({
+                'name': file,
+                'old': old_files[file],
+                'new': new_files[file]
+            })
+
+    return changed_files
+
+def output_diff(changed_files):
+    with open(path.join(root_directory, args.new, f"{args.old}-{args.new}"), "w") as file:
+        file.write(f"{len(changed_files)} changes:\n")
+        for change in changed_files:
+            file.write(f"{change['name']}: {change['old']} -> {change['new']}\n")
+
+def main():
+    path_old = path.join(root_directory, args.old)
+    path_new = path.join(root_directory, args.new)
+
+    result = find_changes(path_old, path_new)
+    output_diff(result)
+
+if __name__ == "__main__":
+    main()
diff --git a/results/classifier/002/001-002 b/results/classifier/002/001-002
new file mode 100644
index 000000000..a1ebf9599
--- /dev/null
+++ b/results/classifier/002/001-002
@@ -0,0 +1,5 @@
+4 changes:
+42226390: instruction -> boot
+51610399: instruction -> boot
+60339453: other -> boot
+67821138: other -> boot
diff --git a/results/classifier/003/002-003 b/results/classifier/003/002-003
new file mode 100644
index 000000000..dabc9fa9a
--- /dev/null
+++ b/results/classifier/003/002-003
@@ -0,0 +1,11 @@
+10 changes:
+55961334: instruction -> KVM
+62179944: instruction -> network
+33802194: instruction -> KVM
+05479587: semantic -> network
+04472277: other -> KVM
+42613410: other -> KVM
+43643137: other -> KVM
+71456293: mistranslation -> KVM
+80615920: mistranslation -> KVM
+26430026: mistranslation -> KVM
diff --git a/results/classifier/004/003-004 b/results/classifier/004/003-004
new file mode 100644
index 000000000..6564ca9ef
--- /dev/null
+++ b/results/classifier/004/003-004
@@ -0,0 +1,19 @@
+18 changes:
+24190340: instruction -> device
+11357571: instruction -> vnc
+73660729: instruction -> graphic
+30680944: semantic -> graphic
+80570214: semantic -> vnc
+28596630: semantic -> device
+99674399: other -> device
+57195159: other -> device
+48245039: other -> assembly
+57231878: other -> device
+14488057: other -> device
+42226390: boot -> device
+67821138: boot -> device
+55961334: KVM -> graphic
+42613410: KVM -> vnc
+33802194: KVM -> vnc
+24930826: mistranslation -> device
+22219210: mistranslation -> graphic
diff --git a/results/classifier/005/004-005 b/results/classifier/005/004-005
new file mode 100644
index 000000000..c81de0e2b
--- /dev/null
+++ b/results/classifier/005/004-005
@@ -0,0 +1,7 @@
+6 changes:
+26095107: instruction -> device
+53568181: semantic -> graphic
+96782458: semantic -> other
+46572227: semantic -> graphic
+48245039: assembly -> device
+36568044: mistranslation -> device