BUG: Do not attempt to cache unhashable values in to_datetime (pandas-dev#39756) (pandas-dev#40414)

Author: maroth96

doc/source/whatsnew/v1.3.0.rst

@@ -592,6 +592,7 @@ Reshaping
 - Bug in :meth:`DataFrame.pivot_table` returning a ``MultiIndex`` for a single value when operating on and empty ``DataFrame`` (:issue:`13483`)
 - Allow :class:`Index` to be passed to the :func:`numpy.all` function (:issue:`40180`)
 - Bug in :meth:`DataFrame.stack` not preserving ``CategoricalDtype`` in a ``MultiIndex`` (:issue:`36991`)
+- Bug in :func:`to_datetime` raising error when input sequence contains unhashable items (:issue:`39756`)
 
 Sparse
 ^^^^^^

pandas/core/tools/datetimes.py

@@ -147,7 +147,11 @@ def should_cache(
 
     assert 0 < unique_share < 1, "unique_share must be in next bounds: (0; 1)"
 
-    unique_elements = set(islice(arg, check_count))
+    try:
+        # We can't cache if the items are not hashable.
+        unique_elements = set(islice(arg, check_count))
+    except TypeError:
+        return False
     if len(unique_elements) > check_count * unique_share:
         do_caching = False
     return do_caching

pandas/tests/tools/test_to_datetime.py

@@ -1651,6 +1651,12 @@ def test_to_datetime_unprocessable_input(self, cache):
         with pytest.raises(TypeError, match=msg):
             to_datetime([1, "1"], errors="raise", cache=cache)
 
+    @pytest.mark.parametrize("cache", [True, False])
+    def test_to_datetime_unhashable_input(self, cache):
+        series = Series([["a"]] * 100)
+        result = to_datetime(series, errors="ignore", cache=cache)
+        tm.assert_series_equal(series, result)
+
     def test_to_datetime_other_datetime64_units(self):
         # 5/25/2012
         scalar = np.int64(1337904000000000).view("M8[us]")