re_similar, re_struct_audit; next_functions favours raw functions with a readable near-duplicate

harbor/queue.py

@@ -103,6 +103,15 @@ class FunctionQueue:
                 return {int(row["address"], 16): row for row in csv.DictReader(f)}
         return {int(row["address"], 16): row for row in json.loads(Path(path).read_text(encoding="utf-8"))}
 
+    def similar_readable(self):
+        """Raw functions with a readable near-duplicate (from the [reverse] similar_clusters file)."""
+        path = self.config.get("similar_clusters")
+        if not path or not Path(path).exists():
+            return frozenset()
+        data = json.loads(Path(path).read_text(encoding="utf-8"))
+        return frozenset(int(m["address"], 16) for c in data.get("clusters", []) for m in c.get("members", [])
+                         if m.get("best_readable") and m.get("status") != "readable")
+
     # ------------------------------------------------------------------ queue
     def score(self, addr, facts, named, rejected, similar_readable=frozenset()):
         d = facts[addr]