diff --git a/docs/superpowers/plans/2026-07-03-sql-problem-display.md b/docs/superpowers/plans/2026-07-03-sql-problem-display.md
index f394f80..cafa560 100644
--- a/docs/superpowers/plans/2026-07-03-sql-problem-display.md
+++ b/docs/superpowers/plans/2026-07-03-sql-problem-display.md
@@ -33,6 +33,7 @@
   - `TableDump = {"name": str, "columns": [{"name": str, "type": str}], "rows": [[原始值]], "total_rows": int, "truncated": bool}`
   - `ExpectedQuery = {"columns": [str], "rows": [[原始值]], "total_rows": int, "truncated": bool}`（query 模式）
   - `ExpectedModify = {"changed_tables": [TableDump, ...]}`（modify 模式，只含执行标准答案后内容有变化的表）
+  - 被标准答案 DROP 的表以 `{"dropped": true, "rows": [], "total_rows": 0}` 条目出现（columns 取初始结构）；标准答案未改任何表时抛 SQLCaseError
 
 - [ ] **Step 1: 在 `sql_runner.py` 末尾追加展示数据生成代码**
 
diff --git a/docs/superpowers/specs/2026-07-03-sql-problem-display-design.md b/docs/superpowers/specs/2026-07-03-sql-problem-display-design.md
index 3463ffe..1764d05 100644
--- a/docs/superpowers/specs/2026-07-03-sql-problem-display-design.md
+++ b/docs/superpowers/specs/2026-07-03-sql-problem-display-design.md
@@ -44,6 +44,8 @@ SQL 题目前沿用传统题的展示结构：「输入/输出」两段富文本
 
 - query 模式：`expected` 为 `{columns, rows}` 结果集
 - modify 模式：`expected` 为 `{"changed_tables": [{name, columns, rows, total_rows, truncated}]}`，只含有变化的表
+- modify 模式中被标准答案 DROP 的表以 `{"name", "columns"(原结构), "rows": [], "total_rows": 0, "truncated": false, "dropped": true}` 条目出现，前端渲染为「XX 表已被删除」
+- modify 模式若标准答案未改动任何表，保存时报错（视为出题配置问题）
 
 ### 生成逻辑
 
diff --git a/judge/sql_runner.py b/judge/sql_runner.py
index 14ad7be..21f30f8 100644
--- a/judge/sql_runner.py
+++ b/judge/sql_runner.py
@@ -324,6 +324,8 @@ def build_display(init_sql, ref_sql, mode, *, memory_limit_mb=64):
             _execute_trusted(conn, ref_sql, time.monotonic() + trusted_limit_s, "标准答案执行失败")
             after = _dump_tables(conn)
             changed = {name for name in set(before) | set(after) if before.get(name) != after.get(name)}
+            if not changed:
+                raise SQLCaseError(JudgeStatus.SYSTEM_ERROR, "标准答案未修改任何表数据，请检查题目配置")
             changed_tables = _dump_display_tables(conn, only=changed)
             # 被标准答案 DROP 的表已不在库中，用初始展示数据补齐条目（前端据 dropped 提示“表已删除”）
             existing = {t["name"] for t in changed_tables}
diff --git a/problem/serializers.py b/problem/serializers.py
index e3ff3f9..51e4682 100644
--- a/problem/serializers.py
+++ b/problem/serializers.py
@@ -63,7 +63,7 @@ class CreateOrEditProblemSerializer(serializers.Serializer):
     input_description = serializers.CharField(allow_blank=True)
     output_description = serializers.CharField(allow_blank=True)
     samples = serializers.ListField(child=CreateSampleSerializer(), allow_empty=True)
-    test_case_id = serializers.CharField(max_length=32)
+    test_case_id = serializers.RegexField(regex=r"^[a-zA-Z0-9]+$", max_length=32)
     test_case_score = serializers.ListField(child=CreateTestCaseScoreSerializer(), allow_empty=True)
     time_limit = serializers.IntegerField(min_value=1, max_value=1000 * 60)
     memory_limit = serializers.IntegerField(min_value=1, max_value=1024)
diff --git a/problem/views/admin.py b/problem/views/admin.py
index 691428a..226297b 100644
--- a/problem/views/admin.py
+++ b/problem/views/admin.py
@@ -238,7 +238,7 @@ class ProblemBase(APIView):
             if not keys:
                 return "题目没有任何测试点"
             input_name = info["test_cases"][keys[0]]["input_name"]
-        except (KeyError, AttributeError):
+        except (KeyError, AttributeError, TypeError):
             return "测试点信息损坏，请重新上传测试点"
         try:
             with open(os.path.join(test_case_dir, input_name), encoding="utf-8") as f: