버그 수정 및 해커 그룹 분리 코드 추가

crawler.py

@@ -444,6 +444,29 @@ def find_skill_uses_for_validation(
     return skill_use_set
 
 
+def collapse_duplicated(data):
+    # 그룹핑 key: uid + timestamp + xyz + bef_xyz
+    def make_key(item):
+        return (
+            item["uid"],
+            item["@timestamp"],
+            tuple(item["body"]["xyz"]),
+            tuple(item["body"]["bef_xyz"])
+        )
+
+    grouped = defaultdict(list)
+    for d in data:
+        grouped[make_key(d)].append(d)
+
+    collapsed = []
+    for _, items in grouped.items():
+        base = items[0].copy()
+        base["duplicated"] = len(items)
+        collapsed.append(base)
+
+    return collapsed
+
+
 # =========================
 # 2) 저장 전용 (분석 없음)
 # =========================
@@ -469,10 +492,11 @@ def save_verified_batches_to_csv(
         for verified_hack_logs in verified_batches_iter:
             summary_data = defaultdict(lambda: defaultdict(int))        
 
-            for log in verified_hack_logs:
+            collapsed = collapse_duplicated(verified_hack_logs)
+            for log in collapsed:
                 uid = log.get('uid')
                 if uid:
-                    hack_type = len(log)
+                    hack_type = log.get('duplicated')
                     summary_data[uid][hack_type] += 1
 
             uids_to_lookup = {log['uid'] for log in verified_hack_logs if 'uid' in log}