fix: 缓存原子写+fsync+校验，diag 先打印再缓存（防卡住看不到诊断）

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

代码/code/bench.py

@@ -104,29 +104,61 @@ def _load_filtered(history_dir, test_csv, test_users):
     return item_dict, user_seq
 
 
-def _get_data(cur, ref, rebuild=False):
+def _cache_path(cur):
-    """取过滤后的 (item_dict, user_seq)，优先读磁盘缓存。
+    return cur / "bench_filtered_cache.pkl"
 
-    用 pickle 而非 torch.save/load：AI Studio overlay 文件系统对 torch 的
+
-    zip/mmap 读取会间歇性报 [Errno 38] Function not implemented。
+def _build_filtered(ref):
-    """
-    import pickle
-    cache = cur / "bench_filtered_cache.pkl"
     test_csv = ref / "test.csv"
     history = ref / "history"
-    if cache.exists() and not rebuild:
+    test_users = _test_user_ids(test_csv)
-        print(f"[BENCH] 读取过滤缓存：{cache}")
+    return _load_filtered(history, test_csv, test_users)
-        try:
+
+
+def _load_cache(cache):
+    import pickle
     with open(cache, "rb") as f:
         d = pickle.load(f)
     return d["item_dict"], d["user_seq"]
+
+
+def _save_cache(cache, item_dict, user_seq):
+    """原子写 + fsync + 写后校验；任何异常都不留毒文件。
+
+    用 pickle 而非 torch.save：AI Studio overlay 文件系统对 torch 的 zip/mmap
+    读取会间歇性报 [Errno 38]。pickle.dump 大对象较慢但顺序写更稳。
+    """
+    import pickle
+    tmp = str(cache) + ".tmp"
+    try:
+        with open(tmp, "wb") as f:
+            pickle.dump({"item_dict": item_dict, "user_seq": user_seq}, f,
+                        protocol=pickle.HIGHEST_PROTOCOL)
+            f.flush()
+            os.fsync(f.fileno())
+        os.replace(tmp, cache)
+        _load_cache(cache)  # 写后立即校验可读
+        print(f"[BENCH] 已缓存 -> {cache}")
+    except Exception as e:
+        print(f"[BENCH][WARN] 缓存写入失败（{e}），本次不缓存（不影响结果）")
+        for p in (tmp, str(cache)):
+            try:
+                os.remove(p)
+            except OSError:
+                pass
+
+
+def _get_data(cur, ref, rebuild=False):
+    """取过滤后的 (item_dict, user_seq)，优先读磁盘缓存。"""
+    cache = _cache_path(cur)
+    if cache.exists() and not rebuild:
+        print(f"[BENCH] 读取过滤缓存：{cache}")
+        try:
+            return _load_cache(cache)
         except Exception as e:
             print(f"[BENCH][WARN] 缓存读取失败（{e}），重新构建")
-    test_users = _test_user_ids(test_csv)
+    item_dict, user_seq = _build_filtered(ref)
-    item_dict, user_seq = _load_filtered(history, test_csv, test_users)
+    _save_cache(cache, item_dict, user_seq)
-    with open(cache, "wb") as f:
-        pickle.dump({"item_dict": item_dict, "user_seq": user_seq}, f, protocol=4)
-    print(f"[BENCH] 已缓存 -> {cache}")
     return item_dict, user_seq
 
 
@@ -211,10 +243,25 @@ def run_once(config_override=None, batch_size=50, max_batches=None,
 
 
 def run_diag(rebuild=False):
-    """诊断：测试用户序列长度分布 + sign-id 是否超界（判断上下文与 modulo 的价值）。"""
+    """诊断：测试用户序列长度分布 + sign-id 是否超界（判断上下文与 modulo 的价值）。
+
+    先打印诊断，再写缓存——避免缓存写入卡住时看不到诊断结果。
+    """
     cur = Path(__file__).parent
     ref = cur / "dataset"
-    item_dict, user_seq = _get_data(cur, ref, rebuild=rebuild)
+    cache = _cache_path(cur)
+    loaded = False
+    item_dict = user_seq = None
+    if cache.exists() and not rebuild:
+        print(f"[BENCH] 读取过滤缓存：{cache}")
+        try:
+            item_dict, user_seq = _load_cache(cache)
+            loaded = True
+        except Exception as e:
+            print(f"[BENCH][WARN] 缓存读取失败（{e}），重新构建")
+    if not loaded:
+        item_dict, user_seq = _build_filtered(ref)
+
     lens = np.array([len(v) for v in user_seq.values()]) if user_seq else np.array([0])
     print(f"[DIAG] 测试用户数={len(user_seq)}  总记录数={len(item_dict)}")
     print(f"[DIAG] 每用户序列长度 min/median/mean/max = "
@@ -235,6 +282,9 @@ def run_diag(rebuild=False):
           f"超界sign占比={over}/{tot}={(over / max(tot, 1)):.2%}  "
           f"（占比高=clamp 在污染 embedding → modulo 可能找回 AUC）")
 
+    if not loaded:
+        _save_cache(_cache_path(cur), item_dict, user_seq)
+
 
 def _parse_args():
     import argparse