fix bug when 'check_count' == 0

Author: anmyachev

pandas/core/tools/datetimes.py

@@ -54,18 +54,21 @@ def should_cache(arg, check_count: int, unique_share: float):
     ----------
     arg: listlike, tuple, 1-d array, Series
     check_count: int
-        0 < check_count <= len(arg)
+        0 <= check_count <= len(arg)
     unique_share: float
         0 < unique_share < 1
 
     Returns
     -------
     do_caching: bool
     """
-    assert 0 < check_count <= len(arg), ('check_count must be in next bounds: '
-                                         '(0; len(arg)]')
+    assert 0 <= check_count <= len(arg), ('check_count must be in next bounds:'
+                                          ' [0; len(arg)]')
     assert 0 < unique_share < 1, 'unique_share must be in next bounds: (0; 1)'
 
+    if check_count == 0:
+        return False
+
     do_caching = True
 
     unique_elements = unique(arg[:check_count])

pandas/tests/indexes/datetimes/test_tools.py

@@ -2044,7 +2044,7 @@ def test_should_cache(listlike, do_caching):
 
 
 @pytest.mark.parametrize('check_count,unique_share, err_message', [
-    (11, 0.5, r'check_count must be in next bounds: \(0; len\(arg\)]'),
+    (11, 0.5, r'check_count must be in next bounds: [0; len\(arg\)]'),
     (10, 2, r'unique_share must be in next bounds: \(0; 1\)')
 ])
 def test_should_cache_errors(check_count, unique_share, err_message):