Add tool to generate zoneup opcodes if they are offset

Also only accept confident matches if the score is above 0

generate_diff_offset.py

@@ -0,0 +1,43 @@
+import click
+import json
+
+
+@click.command()
+@click.argument("diff_file", type=click.File("r"))
+@click.argument("offset", type=int)
+def generate_diff_offset(diff_file, offset):
+    """
+    Given a diff file, produces another diff file where the matches
+    are offset by OFFSET.
+
+    Example:
+    python generate_diff_offset.py -- diff.json -1 > offset.diff.json
+    """
+    diff_json = json.load(diff_file)
+    olds = []
+    news = []
+    for pair in diff_json:
+        if "old" not in pair or "new" not in pair:
+            continue
+        olds.append(pair["old"])
+        news.append(pair["new"])
+
+    if offset < 0:
+        news = [[] for _ in range(-offset)] + news
+    if offset > 0:
+        olds = [[] for _ in range(offset)] + olds
+
+    opcodes_object = []
+    for old, new in zip(olds, news):
+        opcodes_object.append(
+            {
+                "old": old,
+                "new": new,
+            }
+        )
+
+    print(json.dumps(opcodes_object, indent=2))
+
+
+if __name__ == "__main__":
+    generate_diff_offset()

sanity_check.py

@@ -35,13 +35,22 @@ def sanity_check(vtable_diff, minor_patch_diff):
     for old_opcode, new_opcodes in diff2.items():
         if old_opcode not in diff1:
             if len(new_opcodes) < 50:
-                print(f"Missing old opcode in vtable diff: {old_opcode}")
+                print(f"Missing old opcode in vtable diff: {hex(old_opcode)}")
                 all_good = False
             continue
 
         other_new_opcodes = diff1[old_opcode]
-        if list(other_new_opcodes)[0] not in new_opcodes:
-            print(f"vtable diff mismatch for case {old_opcode} => {other_new_opcodes}")
+        vtable_new_opcode = None
+        if len(other_new_opcodes) > 0:
+            vtable_new_opcode = list(other_new_opcodes)[0]
+        elif len(new_opcodes) == 0:
+            continue
+
+        if vtable_new_opcode not in new_opcodes:
+            new_text = (
+                hex(vtable_new_opcode) if vtable_new_opcode is not None else "None"
+            )
+            print(f"vtable diff mismatch for case {hex(old_opcode)} => {new_text}")
             all_good = False
 
     if all_good:

vtable_alignment.py

@@ -135,6 +135,7 @@ class Similarity:
                 rev_top_matches = rev_scores[new_op]
                 if (
                     rev_top_matches[0][0] == old_op
+                    and rev_top_matches[0][1] > 0
                     and rev_top_matches[0][1] - rev_top_matches[1][1] >= threshold
                 ):
                     matches.append((old_op, new_op))
@@ -203,7 +204,7 @@ def vtable_alignment(old_exe, new_exe, similarity_json_file):
         filter(lambda x: x[0] in old_seq_set and x[1] in new_seq_set, matches)
     )
 
-    eprint(f"Found {len(matches)} confident matches")
+    eprint(f'Found {len(matches)} "confident" matches')
 
     eprint("Running initial alignment...")
     alignment, score = needleman_wunsch(old_seq, new_seq, similarity, -1)