Add more cache conditions

Author: mroeschke

asv_bench/benchmarks/timeseries.py

@@ -356,6 +356,8 @@ def setup(self):
 
         self.s = Series((['19MAY11', '19MAY11:00:00:00'] * 100000))
         self.s2 = self.s.str.replace(':\\S+$', '')
+        self.numeric_data = Series([range(100000)])
+        self.datetime_data = [dt.datetime(2010, 1, 1)] * 100000
 
     def time_format_YYYYMMDD(self):
         to_datetime(self.stringsD, format='%Y%m%d')
@@ -381,6 +383,12 @@ def time_format_exact(self):
     def time_format_no_exact(self):
         to_datetime(self.s, format='%d%b%y', exact=False)
 
+    def time_cache_numeric_data(self):
+        to_datetime(self.numeric_data)
+
+    def time_cache_datetime_data(self):
+        to_datetime(self.datetime_data)
+
 
 class Offsets(object):
     goal_time = 0.2

pandas/core/tools/datetimes.py

@@ -16,7 +16,8 @@
     is_float,
     is_list_like,
     is_scalar,
-    is_numeric_dtype)
+    is_numeric_dtype,
+    is_string_dtype)
 from pandas.core.dtypes.generic import (
     ABCIndexClass, ABCSeries,
     ABCDataFrame, ABCDateOffset)
@@ -521,14 +522,19 @@ def _convert_listlike(arg, box, format, name=None, tz=tz):
         arg = arg + offset
 
     convert_cache = None
-    if cache and is_list_like(arg) and not isinstance(arg, DatetimeIndex):
+    if cache and is_list_like(arg):
+        # Create a cache only if there are more than 10k values and the user
+        # passes in datestrings
+        min_cache_threshold = 10**5
+        if len(arg) >= min_cache_threshold and is_string_dtype(arg):
         # unique currently cannot determine dates that are out of bounds
-        # use the cache only if the data is a string and there are more than 10**5 values
-        unique_dates = algorithms.unique(arg)
-        if len(unique_dates) != len(arg):
-            from pandas import Series
-            cache_data = _convert_listlike(unique_dates, True, format)
-            convert_cache = Series(cache_data, index=unique_dates)
+        # recurison errors with datetime
+            unique_dates = algorithms.unique(arg)
+            # Essentially they need to all be the same value
+            if len(unique_dates) == 1:
+                from pandas import Series
+                cache_data = _convert_listlike(unique_dates, True, format)
+                convert_cache = Series(cache_data, index=unique_dates)
 
     if isinstance(arg, tslib.Timestamp):
         result = arg

pandas/tests/indexes/datetimes/test_tools.py

@@ -370,6 +370,38 @@ def test_datetime_invalid_datatype(self):
             pd.to_datetime(bool)
         with pytest.raises(TypeError):
             pd.to_datetime(pd.to_datetime)
+    
+    @pytest.mark.parametrize("utc", [True, None])
+    @pytest.mark.parametrize("format", ['%Y%m%d %H:%M:%S', None])
+    @pytest.mark.parametrize("box", [True, False])
+    @pytest.mark.parametrize("constructor", [list, tuple, np.array, pd.Index])
+    def test_to_datetime_cache(self, utc, format, box, constructor):
+        date = '20130101 00:00:00'
+        test_dates = [date] * 10**5
+        data = constructor(test_dates)
+        result = pd.to_datetime(data, utc=utc, format=format, box=box)
+        expected = pd.to_datetime(data, utc=utc, format=format, box=box,
+                                  cache=False)
+        if box:
+            tm.assert_index_equal(result, expected)
+        else:
+            tm.assert_numpy_array_equal(result, expected)
+
+    @pytest.mark.parametrize("utc", [True, None])
+    @pytest.mark.parametrize("format", ['%Y%m%d %H:%M:%S', None])
+    def test_to_datetime_cache_series(self, utc, format):
+        date = '20130101 00:00:00'
+        test_dates = [date] * 10**5
+        data = pd.Series(test_dates)
+        result = pd.to_datetime(data, utc=utc, format=format, cache=True)
+        expected = pd.to_datetime(data, utc=utc, format=format)
+        tm.assert_series_equal(result, expected)
+
+    def test_to_datetime_cache_scalar(self):
+        date = '20130101 00:00:00'
+        result = pd.to_datetime(date, cache=True)
+        expected = pd.Timestamp('20130101 00:00:00')
+        assert result == expected
 
 
 class TestToDatetimeUnit(object):