feat(let): add strftime function

Author: mesejo

python/xorq/backends/let/datafusion/compiler.py

@@ -93,7 +93,6 @@ class DataFusionCompiler(SQLGlotCompiler):
         ops.CountDistinctStar,
         ops.DateDelta,
         ops.RowID,
-        ops.Strftime,
         ops.TimeDelta,
         ops.TimestampDelta,
     )
@@ -664,5 +663,8 @@ def visit_ArrayAny(self, op, *, arg):
             ),
         )
 
+    def visit_Strftime(self, op, *, arg, format_str):
+        return self.f.temporal_strftime(arg, format_str)
+
 
 compiler = DataFusionCompiler()

python/xorq/backends/let/datafusion/udfs.py

@@ -129,3 +129,15 @@ def regex_split(s: str, pattern: str) -> list[str]:
         )
     pattern = patterns[0].as_py()
     return pc.split_pattern_regex(s, pattern)
+
+
+def temporal_strftime(array: dt.Timestamp(scale=9), pattern: str) -> dt.string:
+    patterns = pattern.unique()
+
+    if len(patterns) != 1:
+        raise com.XorqError(
+            "Only a single scalar pattern is supported for DataFusion strftime"
+        )
+
+    pattern = patterns[0].as_py()
+    return pc.strftime(array, pattern)

python/xorq/backends/let/tests/test_temporal.py

@@ -158,6 +158,24 @@ def test_timestamp_extract_week_of_year(alltypes, alltypes_df):
     assert_series_equal(result, expected)
 
 
+@pytest.mark.parametrize(
+    ("expr_fn", "pandas_pattern"),
+    [
+        param(
+            lambda t: t.timestamp_col.strftime("%Y%m%d").name("formatted"),
+            "%Y%m%d",
+            id="literal_format_str",
+        ),
+    ],
+)
+def test_strftime(alltypes, alltypes_df, expr_fn, pandas_pattern):
+    expr = expr_fn(alltypes)
+    expected = alltypes_df.timestamp_col.dt.strftime(pandas_pattern).rename("formatted")
+
+    result = expr.execute()
+    assert_series_equal(result, expected)
+
+
 PANDAS_UNITS = {
     "m": "Min",
     "ms": "L",