!9544 Fix a possible seg fault in GenerateCRC caused by std::regex

From: @lixiachen
Reviewed-by: @jonyguo,@heleiwang
Signed-off-by: @heleiwang

mindspore/ccsrc/minddata/dataset/engine/datasetops/dataset_op.cc

@@ -453,8 +453,8 @@ uint32_t DatasetOp::GenerateCRC(const std::shared_ptr<DatasetOp> &op) {
   ss_str = std::regex_replace(ss_str, std::regex("device_id.*\n"), "");
 
   // Filter out the operator id field
-  ss_str = std::regex_replace(ss_str, std::regex(".*Parent.*\n"), "");
-  ss_str = std::regex_replace(ss_str, std::regex(".*Child.*\n"), "");
+  ss_str = std::regex_replace(ss_str, std::regex(" *Parent.*\n"), "");
+  ss_str = std::regex_replace(ss_str, std::regex(" *Child.*\n"), "");
   ss_str = std::regex_replace(ss_str, std::regex(R"(\(\s*\d+?\))"), "");
 
   // Doesn't matter whether there is any parent node above CacheOp or not.

tests/ut/python/cachetests/cachetest_py.sh

@@ -324,6 +324,9 @@ HandleRcExit $? 0 0
 PytestCmd "test_cache_nomap.py" "test_cache_nomap_get_repeat_count"
 HandleRcExit $? 0 0
 
+PytestCmd "test_cache_nomap.py" "test_cache_nomap_long_file_list"
+HandleRcExit $? 0 0
+
 for i in $(seq 1 3)
 do
    test_name="test_cache_nomap_multiple_cache${i}"

tests/ut/python/dataset/test_cache_map.py

@@ -1828,9 +1828,7 @@ def test_cache_map_cifar3():
 
        cache
          |
-     Map(resize)
-         |
-      Cifar100
+      Cifar10
     """
 
     logger.info("Test cache map cifar3")
@@ -1841,9 +1839,7 @@ def test_cache_map_cifar3():
 
     some_cache = ds.DatasetCache(session_id=session_id, size=1, spilling=False)
 
-    ds1 = ds.Cifar10Dataset(CIFAR10_DATA_DIR)
-    resize_op = c_vision.Resize((224, 224))
-    ds1 = ds1.map(input_columns=["image"], operations=resize_op, cache=some_cache)
+    ds1 = ds.Cifar10Dataset(CIFAR10_DATA_DIR, cache=some_cache)
 
     num_epoch = 2
     iter1 = ds1.create_dict_iterator(num_epochs=num_epoch)

tests/ut/python/dataset/test_cache_nomap.py

@@ -1775,6 +1775,7 @@ def test_cache_nomap_textfile2():
          |
      TextFile
     """
+
     def my_tokenizer(line):
         words = line.split()
         if not words:
@@ -1884,6 +1885,34 @@ def test_cache_nomap_get_repeat_count():
         num_iter += 1
     assert num_iter == 12
 
+
+@pytest.mark.skipif(os.environ.get('RUN_CACHE_TEST') != 'TRUE', reason="Require to bring up cache server")
+def test_cache_nomap_long_file_list():
+    """
+    Test cache after TFRecord with a long list of files as arguments
+
+        Cache
+          |
+      TFRecord
+    """
+
+    logger.info("Test cache nomap long file list")
+    if "SESSION_ID" in os.environ:
+        session_id = int(os.environ['SESSION_ID'])
+    else:
+        raise RuntimeError("Testcase requires SESSION_ID environment variable")
+
+    some_cache = ds.DatasetCache(session_id=session_id, size=1, spilling=False)
+
+    ds1 = ds.TFRecordDataset([DATA_DIR[0] for _ in range(0, 1000)], SCHEMA_DIR, columns_list=["image"],
+                             cache=some_cache)
+
+    with pytest.raises(RuntimeError) as e:
+        sum([1 for _ in ds1])
+    assert "Out of memory" in str(e.value)
+    logger.info("test_cache_nomap_long_file_list Ended.\n")
+
+
 if __name__ == '__main__':
     test_cache_nomap_basic1()
     test_cache_nomap_basic2()