address comments

Author: sfc-gh-helmeleegy

src/snowflake/snowpark/modin/plugin/compiler/snowflake_query_compiler.py

@@ -16840,7 +16840,7 @@ def output_col(
                 # Follow pandas behavior
                 return pandas_lit(np.nan)
             elif n < -1 and not pandas.isnull(pat) and len(str(pat)) > 1:
-                # Follow pandas behavior, which seems to leave the input column as is
+                # Follow pandas behavior, which based on our experiments, leaves the input column as is
                 # whenever the above condition is satisfied.
                 new_col = iff(
                     column.is_null(), pandas_lit(None), array_construct(column)
@@ -16917,7 +16917,7 @@ def output_cols(
                 final_splits = min(n + 1, max_splits)
 
             if n < -1 and not pandas.isnull(pat) and len(str(pat)) > 1:
-                # Follow pandas behavior, which seems to leave the input column as is
+                # Follow pandas behavior, which based on our experiments, leaves the input column as is
                 # whenever the above condition is satisfied.
                 final_splits = 1
 

tests/integ/modin/series/test_str_accessor.py

@@ -370,7 +370,9 @@ def test_str_replace_neg(pat, n, repl, error):
         snow_ser.str.replace(pat=pat, repl=repl, n=n)
 
 
-@pytest.mark.parametrize("pat", [None, "a", "ab", "non_occurrence_pat", "|", "%"])
+@pytest.mark.parametrize(
+    "pat", [None, "a", "ab", "abc", "non_occurrence_pat", "|", "%"]
+)
 @pytest.mark.parametrize("n", [None, np.nan, 3, 2, 1, 0, -1, -2])
 @sql_count_checker(query_count=1)
 def test_str_split_expand_false(pat, n):
@@ -383,7 +385,7 @@ def test_str_split_expand_false(pat, n):
     )
 
 
-@pytest.mark.parametrize("pat", [None, "a", "ab", "no_occurrence_pat", "|", "%"])
+@pytest.mark.parametrize("pat", [None, "a", "ab", "abc", "no_occurrence_pat", "|", "%"])
 @pytest.mark.parametrize("n", [None, np.nan, 3, 2, 1, 0, -1, -2])
 @sql_count_checker(query_count=2)
 def test_str_split_expand_true(pat, n):